Scribd Logo
Upload

Welcome to Scribd!

  • A Synopsis of Elementary Results in Pure Mathematics

    Uploaded by

    khanke magey
    2K views1,037 pages

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Report this Document

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Flag for inappropriate content
    2K views1,037 pages

    A Synopsis of Elementary Results in Pure Mathematics

    Uploaded by

    khanke magey

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Flag for inappropriate content
    of 1037

    This is a reproduction of a library book that was digitized

    by Google as part of an ongoing effort to preserve the


    information in books and make it universally accessible.

    https://books.google.com
    A SYNOPSIS
    OF

    ELEMENTARY RESULTS
    IN

    PURE MATHEMATICS :
    CONTAINING

    PROPOSITIONS, FORMULÆ , AND METHODS OF ANALYSIS ,


    WITH

    ABRIDGED DEMONSTRATIONS.

    SUPPLEMENTED BY AN INDEX TO THE PAPERS ON PURE MATHEMATICS WHICH ARE TO


    BE FOUND IN THE PRINCIPAL JOURNALS AND TRANSACTIONS OF LFARNED SOCIETIES,
    BOTH ENGLISH AND FOREIGN , OF THE PRESENT CENTURY,

    BY

    G . S. CARR, M .A.

    OF TEC
    UNIVERSITY

    LONDON :
    FRANCIS HODGSON , 89 FARRINGDON STREET, E.C.
    CAMBRIDGE : MACMILLAN & BOWES.
    1886.
    (All rights reservell )
    ♡A41
    C3
    Engin ’ ering
    Libar

    LONDON :
    PRINTED BY C . F . HODGSON AND SON ,
    GOUGH SQUARE, FLEET STREET.
    76606
    o s
    Er

    VIVI CITY

    PREFACE TO PART I.

    The work, of which the part now issued is a first instal


    ment, has been compiled from notes made at various periods
    of the last fourteen years, and chiefly during the engagements
    of teaching. Many of the abbreviated methods and mnemonic
    rules are in the form in which I originally wrote them for my
    pupils.
    The general object of the compilation is, as the title
    indicates, to present within a moderate compass the funda
    mentaltheorems, formulæ , and processes in the chief branches
    of pure and applied mathematics .
    The work is intended , in the first place, to follow and
    supplement the use of the ordinary text-books, and it is
    arranged with the view of assisting the student in the task of
    revision of book-work . To this end I have, in many cases ,
    merely indicated the salient points of a demonstration , or
    merely referred to the theoremsby which the proposition is
    proved . I am convinced that it is more beneficial to the
    student to recall demonstrations with such aids, than to read
    and re-read them . Let them be read once ,but recalled often .
    The difference in the effect upon the mind between reading a
    mathematical demonstration, and originating one wholly or
    PREFACE .

    partly, is very great. It may be compared to the difference


    between the pleasure experienced , and interest aroused,when
    in the one case a traveller is passively conducted through the
    roads of a novel and unexplored country, and in the other
    case he discovers the roads for himself with the assistance of
    a map .
    In the second place, I venture to hope that the work ,
    when completed, may prove useful to advanced students as
    an aide-mémoire and book of reference . The boundary of
    mathematical science forms, year by year, an ever widening
    circle, and the advantage of having at hand some condensed
    statement of results becomes more and more evident.
    To the original investigator occupied with abstruse re
    searches in some one of the many branches of mathematics, a
    work which gathers together synoptically the leading propo
    sitions in all, may not therefore prove unacceptable . Abler
    hands than mine undoubtedly , might have undertaken the task
    ofmaking such a digest ; but abler hands might also, perhaps,
    bemore usefully employed , — and with this reflection I have the
    less hesitation in commencing the work myself. The design
    which I have indicated is somewhat comprehensive , and in
    relation to it the present essay may be regarded as tentative.
    The degree of success which it may meet with , and the
    suggestions or criticisms which it may call forth , will doubt
    195s have their effect on the subsequent portions of the work .
    With respect to the abridgment of the demonstrations, I
    may remark , that while some diffuseness of explanation is not
    only allowable but very desirable in an initiatory treatise,
    conciseness is oneof the chief requirements in a work intended
    PREFACE .

    for the purposes of revision and reference only . In order ,


    however , not to sacrifice clearness to conciseness, much more
    labour has been expended upon this part of the subject matter
    of the book than will at first sight be at all evident. The only
    palpable result being a compression of the text, the result is
    so far a negative one. The amount of compression attained
    is illustrated in the last section of the present part, in which
    more than the number of propositions usually given in
    treatises on Geometrical Conics are contained , together with
    the figures and demonstrations, in the space of twenty -four
    pages.
    The foregoing remarks have a general application to the
    work as a whole . With the view , however , of making the
    earlier sections more acceptable to beginners , it will be found
    that, in those sections, important principles have sometimes
    been more fully elucidated and more illustrated by examples,
    than the plan of the work would admit of in subsequent
    divisions.
    A feature to which attention may be directed is the uni.
    form system of reference adopted throughout all the sections.
    With the object of facilitating such reference,the articles have
    been numbered progressively from the commencement in
    large Clarendon figures ; the breaks which will occasionally
    be found in these numbers having been purposely made,in order
    to leave room for the insertion of additionalmatter, if it should
    be required in a future edition, without disturbing the original
    numbers and references . With the sameobject,demonstrations
    and examples have been made subordinateto enunciations and
    formulæ , the former being printed in small, the latter in bold
    PREFACE .

    type. By these aids,the interdependence of propositions is


    more readily shown,and it becomes easy to trace the connexion
    between theorems in different branches of mathematics, with
    out the loss of time which would be incurred in turning to
    separate treatises on the subjects . The advantage thus gained
    will, however , become more apparent as the work proceeds.
    The Algebra section was printed some years ago, and does
    not quite correspond with the succeeding ones in some of
    the particulars named above. Under the pressure of other
    occupations, this section moreover was not properly revised
    before going to press. On that account the table of errata
    will be found to apply almost exclusively to errors in that
    section ; but I trust that the list is exhaustive . Great pains
    have been taken to secure the accuracy of the rest of the
    volume. Any intimation of errors will be gladly received .
    I have now to acknowledge someof the sources from which
    the present part has been compiled . In the Algebra, Theory
    of Equations, and Trigonometry sections, I am largely in
    debted to Todhunter's well-known treatises , the accuracy and
    completeness of which it would be superfluous in me to dwell
    upon.
    In the section entitled Elementary Geometry, I have added
    to simpler propositions a selection of theorems from Town
    send's Modern Geometry and Salmon's Conic Sections.
    In Geometrical Conics, the line of demonstration followed
    agrees ,in themain ,with that adopted in Drew 's treatise on the
    subject. I am inclined to think that the method of that
    author cannot be much improved . It is true that some im
    portant properties of the ellipse , which are arrived at in
    PREFACE , vii

    Drew 's Conic Sections through certain intermediate proposi


    tions, can be deduced at once from the circle by themethod of
    orthogonal projection . But the intermediate propositions can
    not on that account be dispensed with , for they are of value in
    themselves. Moreover, the method of projection applied to
    the hyperbola is not so successful ; because a property which
    has first to be proved true in the case of the equilateral
    hyperbola , mightas will be proved at once for the general case .
    I have introduced the method of projection but sparingly,
    always giving preference to a demonstration which admits of
    being applied in the same identical form to the ellipse and to
    the hyperbola. The remarkable analogy subsisting between
    the two curves is thus kept prominently before the reader .
    The account of the C. G . S. system of units given in the
    preliminary section, has been compiled from a valuable con
    tribution on the subject by Professor Everett , of Belfast ,
    published by the Physical Society of London .* This abstract,
    and the tables of physical constants,might perhaps have found
    a more appropriate place in an after part of thework . I have,
    however, introduced them at the commencement, from a sense
    of the great importance of the reform in the selection of units
    of measurement which is embodied in the C . G . S . system ,
    and from a belief that the student cannot be too early
    familiarized with the same.
    The Factor Table which follows is , to its limited extent, a
    reprint of Burckhardt's “ Tables des diviseurs,” published in

    * “ Illustrations of the Centimetre-Gramme-Second System of Units."


    London : Taylor and Francis. 1875.
    viii PREFACE .

    1814 – 17 , which give the least divisors of all numbers from 1


    to 3 ,036 ,000. In a certain sense, it may be said that this is
    the only sort of purely mathematicaltable which is absolutely
    indispensable , because the information which it gives cannot
    be supplied by any process of direct calculation . The loga
    rithm of a number, for instance, may be computed by a
    formula . Not so its prime factors. These can only be
    arrived at through the tentative process of successive divisions
    by the prime numbers, an operation of a most deterrent kind
    when the subject of it is a high integer.
    A table similar to and in continuation of Burckhardt's has
    recently been constructed for the fourth million by J. W . L .
    Glaisher, F .R . S ., who I believe is also now engaged in com
    pleting the fifth and sixth millions. The factors for the seventh ,
    eighth, and ninth millions were calculated previously by Dase
    and Rosenberg, and published in 1862–65, and the tenth
    million is said to exist in manuscript. The history of the
    formation of these tables is both instructive and interesting.*
    As, however, such tables are necessarily expensive to pur
    chase,and not very accessible in any other way tothe majority
    of persons, it seemed to me that a small portion of them
    would form a useful accompaniment to the present volume.
    I have, accordingly , introduced the first eleven pages of Burckh
    ardt's tables, which give the least factors of the first 100,000
    integers nearly. Each double page of the table here printed is

    * See “ Factor Table for the Fourth Million ." By James Glaisher, F.R . S .
    London : Taylor and Francis. 1880. Also Camb. Phil. Soc. Proc., Vol. III.,
    Pt. IV., and Nature,No. 542, p. 462.
    PREFACE .

    an exact reproduction , in all but the type, of a single quarto


    page of Burckhardt's great work.
    It may be noticed here that Prof. Lebesque constructed
    a table to about this extent, on the plan of omitting the
    multiples of seven, and thus reducing the size of the table
    by about one-sixth .* But a small calculation is required in
    using the table which counterbalances the advantage so gained .
    The values of the Gamma-Function , pages 30 and 31, have
    been taken from Legendre's table in his “ Exercices de Calcul
    Intégral," Tome I. The table belongs to Part II. of this
    Volume, hut it is placed here for the convenience of having
    all the numerical tables of Volume I. in the same section.
    In addition to the authors already named , the following
    treatises have been consulted - Algebras, by Wood , Bourdon ,
    and Lefebure de Fourcy ; Snowball's Trigonometry ; Salmon's
    Higher Algebra ; the GeometricalExercises in Potts's Euclid ;
    and Geometrical Conics by Taylor, Jackson , and Renshaw .
    Articles 260, 431, 569, and very nearly all the examples ,
    are original. The latter have been framed with great care , in
    order that they might illustrate the propositions as completely
    S2S COL

    as possible.
    G . S. C .
    HADLEY, MIDDLESEX ;
    May 23, 1880.

    * “ Tables diverses pour la décomposition des nombres en leurs facteurs


    premiers.” Par V. A . Lebesque. Paris. 1864.
    ERRATA.

    for - a²62 read + a b3.


    Line 1, 93
    a.
    com

    » numerators 1, 1, 1 , 1, a, as.
    denominator r - 1 » ^ - 1.
    taken taken m at a time.
    , (196 ) (360).
    »
    2934

    ( - 1) ( - 1) 23.
    » 3,6,7, „ 6x 3.x.
    34 .
    204, 459 102, 306.
    » 9 459 9n .
    10.9.8
    1. 2 . 3 7 .8.9 .10 .
    COM

    „ (2 + 1)* (q + 1)(*) Notation of (96).


    » 20, -, in numerator
    .COELOE

    „ (163) (164).
    (x + y + z) 2 (x + y + z) .
    » (1) square of (1).
    x = 1 2: = - 1.
    , (2 : - 4x + 8) on left side (r? –4x + 8)3.
    (234) Dele.
    (29) (28).
    (267) (266).
    » [11 2 11.
    , 13, , P+ 2
    » (P - 1)
    p + 1.
    (p - 1.
    » X= 1 a = b.
    „ n-1 n + 1.
    Hr, n - 1) 1 (n, r - 1).
    SOCOT

    H (r + 1, n - 1) H (», r).
    » P
    P.P. P.3, last line but one Q.Q. 23.
    ( )
    3528
    ( + )".
    10281.
    n -3 n - 1.
    .

    applying Descartes'rule Dele .


    517,
    5, 544,
    „ a3 23.
    Transpose F and f.
    ,, 551, 1, B B.
    » a -n » Q - K.
    ,, 704, „ (11, 12) (9, 10 , 1).
    » 729, „ (940) (960).
    Article 112 should be as follows:
    1 1 + 2 / 3 + 2 1 + 2./ 3 + 2 (1 + 2,/ 3 + 17/ 2) (11 - 4 3)
    3
    1 + 2 3 - 2 (1 + 273): - 2 11 + 4 3
    S
    E
    S
    * ‫؟‬ . 1.

    TABLE OF CONTENTS.

    PART 1.
    SECTION 1.— MATHEMATICAL TABLES. Page
    INTRODUCTION. THE C . G . S. System Of UNITS
    Notation and Definitions of Units ... ... ... ... 1
    Physical Constants and Formulæ ... ... 2
    TABLE I. - English Measures and Equivalents in C . G . S. Units
    II .- Pressure of Aqueous Vapour at different temperatures
    III. — Wave lengths and Wave frequency for the principali
    lines of the Spectrum ... ... .. . ... ...
    IV . — THE PRINCIPAL METALS — Their Densities ; Coeffi
    cients of Elasticity, Rigidity , and Tenacity ; Expan
    sion by Heat ; Specific Heat ; Conductivity ; Rate
    of conduction of Sound ; Electro-magnetic Specific
    Resistance ... .. . . .. . .. .. . . ..
    V . — THE PLANETS — Their Dimensions, Masses, Densities,
    and Elements of Orbits ... ... ... ...
    VI. - Powers and Logarithms of and e ... ...
    VII. - Square and Cube Roots of the Integers 1 to 30
    VIII .- Common and Hyperbolic Logarithms of
    Prime numbers from 1 to 109... ...
    IX . - FACTOR TABLE
    Explanation of the Table ...
    The Least Factors of all numbers from 1 to 99000 ...
    X . - VALUES OF THE GAMMA -FUNCTION

    SECTION II. - ALGEBRA. Nt of


    Article
    FACTORS
    Newton 's Rule for expanding a Binomial. ..
    MULTIPLICATION AND DIVISION
    INDICES
    Highest COMMON FACTOR
    S
    xii CONTENT .

    No. of
    Article
    LOWEST COMMON MULTIPLE ... ... 33
    EVOLUTION
    Square Root and Cube Root
    Useful Transformations ...

    : : : : : :
    QUADRATIC EQUATIONS . .. .. . ..
    THEORY OF QUADRATIC EXPRESSIONS ...
    EQUATIONS IN ONE UNKNOWN QUANTITY. - EXAMPLES
    Maxima and Minima by a Quadratic Equation
    SIMULTANEOUS EQUATIONS AND EXAMPLES ...

    : : : : :
    RATIO AND PROPORTION

    : : : : : : : : : : :
    The k Theorem

    : : :
    Duplicate and Triplicate Ratios
    Compound Ratios
    VARIATION . .. ... .. .
    ARITHMETICAL PROGRESSION ...

    : : : : : : : : : : : : : :
    GEOMETRICAL PROGRESSION

    :
    HARMONICAL PROGRESSION
    PERMUTATIONS AND COMBINATIONS

    : : : : : : : : : : : :
    SURDB ... ... ... ...
    Simplification of Va + ✓b and Va + 121
    Simplification of " a + vb ... 124
    BINOMIAL THEOREM 125
    MultiNOMIAL THEOREM 137
    LOGARITHMS .. .. ... 142
    EXPONENTIAL THEOREM ... .. 149
    CONTINUED FRACTIONS AND CONVERGENTS 160
    General Theory of same ... ... 167
    To convert a Series into a Continued Fraction ... 182
    A Continued Fraction with Recurring Quotients 186
    INDETERMINATE EQUATIONS ... ... ... ...
    TO REDUCE A QUADRATIC SURD TO A CONTINUED FRACTION ... 195
    To form high Convergents rapidly .. 197
    : : :

    General Theory ... ... ... ... ... ... 199


    EQUATIONS --
    Special casesin the Solution of Simultaneous Equations... 211
    Method by Indeterminate Multipliers ... ini .. . 213
    Miscellaneous Equations and Solutions ... ... 214
    On Symmetrical Expressions ... 219
    : : : : : : :

    IMAGINARY EXPRESSIONS 223


    METHOD OF INDETERMINATE COEFFICIENTS 232
    METHOD OF Proof by INDUCTION ... 233
    PARTIAL FRACTIONS. - Four CASES ... 235
    CONVERGENCY AND DIVERGENCY OF SERIES ... 239
    General Theorem of g (x ) ... 243
    CONTENTS . xiii
    No . of
    Article
    EXPANSION OF A FRACTION ... ... ... ... ... 248
    RECURRING SERIES .. . . .. .. . .. . ...... . .. 251
    The General Term ...
    Case of Quadratic Factor with Imaginary Roots... 258
    Lagrange's Rule 263
    SOMMATION OF SERIES BY THE METHOD OF DIFFERENCES 264
    Interpolation of a term 267
    DIRECT FACTORIAL SERIES 268

    : : : :
    INVERSE FACTORIAL SERIES ... 270
    SUMMATION BY PARTIAL FRACTIONS ... 272
    COMPOSITE FACTORIAL SERIES

    :
    MISCELLANEOUS SERIES
    Sams of the Powers of the Natural Numbers 276
    Sum of a + (a + d) r + (a + 2d ) 7 + & o. 279

    : : : : : : : : : :
    :
    Sum of n - n (n - 1) ' + & c. ... ...
    POLYGONAL NUMBERS ...
    FIGURATE NUMBERS
    : :
    HYPERGEOMETRICAL SERIES 29 ]
    Proof that en is incommensurable 295
    INTEREST 296
    : : : :

    UITIES 302
    PROBABILITIES.. . ... ... 309
    INEQUALITIES ... 330
    Arithmetic Mean > Geometric Mean 332
    Arithmetic Mean of mth powers > mth power of A . 334
    SCALES OF NOTATION ... . 342
    Theorem concerning Sam or Difference of Digits ... 347
    THEORY OF NUMRERS 349
    Highest Power of a Prime p contained in | m ... 365
    Fermat's Theorem ... ... 369
    Wilson 's Theorem .. . 371
    Divisors of a Number 374
    S, divisible by 2n + 1 380

    SECTION III.— THEORY OF EQUATIONS.


    FACTORS OF AN EQUATION ... ... 400
    To compute f (a ) numerically ... 403
    Discrimination of Roots ... ... 409
    DESCARTES' RULE OF Signs ... ... 416
    : : : : : :

    THE DERIVED FUNCTIONS OF f (a ) ... ... 424


    • To remove an assigned term
    To transform an equation ... 430
    Equal ROOTS OF AN EQUATION 432
    Practical Rule ... 445
    :
    xiy CONTENTS .
    No. of
    Article
    LIMITS OF THE Roots 448
    Newton 's Method ... 452

    : : : : : : : : : :
    Rollo 's Theorem ... ... 454
    Newton 's METHOD OF DIVISORS ... 459
    RECIPROCAL EQUATIONS 466
    BINOMIAL EQUATIONS ... , 472
    Solution of " + 1 = 0 by De Moivre's Theorem ... 480
    · Cubic EQUATIONS ... ... 483
    Cardan's Method ... .. . 484

    :
    Trigonometrical Method 489

    :
    BIQUADRATIC EQUATIONS - .
    Descartes' Solution .. . 492
    Ferrari' s Solution ... 496

    :
    Euler's Solution 499
    COMMENSURABLE ROOTS ... : : 502
    INCOMMENSURABLE Roots
    Sturm 's Theorem 506
    : : : :

    Fourier's Theorem . .. 518


    Lagrange's Method of Approximation 525
    Newton 's Method of Approximation 527
    Fourier's Limitation to the same 528
    Newton ' s Rule for the Limits of the Roots 530
    Sylvester's Theorem ... 532
    Horner's Method ... 533
    SYMMETRICAL FUNCTIONS OF THE Roots OF AN EQUATION
    Sums of the Powers of the Roots .... 534
    Symmetrical Functions not Powers of the Roots... 538
    The Equation whose Roots are the Squares of the
    Differences of the Roots of a given Equation ... 541
    Sum of the mth Powers of the Roots of a Quadratic
    Equation ... ... 545
    Approximation to the Root of an Equation through the
    Sums of the Powers of the Roots 548
    EXPANSION OF AN IMPLICIT FUNCTION OF x ... ... 551
    DETERMINANTS
    Definitions ... ... ... ... 554
    General Theory . 556
    To raise the Order of a Determinant 564
    Analysis of a Determinant... ... 568
    Synthesis of a Determinant 569
    Product of two Determinants of the nth Order 570
    :

    Symmetrical Determinants .. .. 574


    Reciprocal Determinants ... .. 575
    : : :

    Partial and Complementary Determinants


    CONTENTS . XV
    No. of
    Article .
    Theorem of a Partial Reciprocal Determinant ... 577
    Product of Differences of n Quantities ... . ... 578
    Product of Squares of Differences of same . 579
    Rational Algebraic Fraction expressed as a Determinant 581
    ELIMINATION
    Solution of Linear Equations ... ... ... 582
    Orthogonal Transformation 584
    Theorem of the n — 2th Power of a Determinant... 585
    Bezout's Method of Elimination ... 586
    Sylvester's Dialytic Method 587
    Method by Symmetrical Functions 588
    ELIMINATION BY HIGHEST COMMON FACTOR ... 593

    SECTION IV .- PLANE TRIGONOMETRY.


    ANGULAR MEASUREMENT ... ... ... 600
    TRIGONOMETRICAL RATIOS ... ... 606
    Formulæ involving one Angle .. 613
    Formulæ involving two Angles and Multiple Angles ... 627
    Formulæ involving three Angles ... 674
    Ratios of 45°, 60°, 15°, 18°, & c. ... ... ... 690
    PROPERTIES OF THE TRIANGLE 700
    The s Formulæ for sin LA, & c. 704
    The Triangle and Circle ... 709
    SOLUTION OF TRIANGLES — .
    Right-angled Triangles ... 718
    Scalene Triangles. — Three cases ... 720
    Examples on the same ... .. 859
    Quadrilateral in a Circle ... ... 733
    Bisector of the Side of a Triangle ... 738
    Bisector of the Angle of a Triangle 742
    Perpendicular on the Base of a Triangle ... 744
    REGULAR POLYGON AND CIRCLE ... ... 746
    SUBSIDIARY ANGLES ... 749
    LIMITS OF RATIOS ... .. 753
    De Moivre's THEOREM ... , 756
    Expansion of cos no, & c. in powers of sin 0 and cos 0 ... 758
    Expansion of sin 0 and cos 0 in powers of 0 .. . ... 764
    Expansion of cos” O and sin " o in cosines or sines of
    multiples of 0 ... ... ... ...
    Expansion of cos no and sin no in powers of sin ...
    Expansion of cosno and sin no in powers of cos 0 .. .
    Expansion of cos no in descending powers of cos @ ...
    Sin u + c sin (a + B ) + & c., and similar series ...
    xvi CONTENTS .
    No, of
    Article .
    Gregory's Series for 0 in powers of tan 0...
    0 ... 791
    Formulæ for the calculation of a ... ... 792

    ; : : : : :
    Proof that r is incommensurable ... , 795
    Sin x = n sin (x + a ). - Series for a 796
    Sum of sines or cosines of Angles in A . P . 800
    Expansion of the sine and cosine in Factors 807
    Sin ng and cos no expanded in Factors 808
    Sin 0 and cos 0 in Factors involving 0 ... ... 815
    e* — 2 cos 0 + e -s expanded in Factors ... 817

    : : : : :
    De Moivre's Property of the Circle ... 819
    Cotes's Properties ... . ... 821
    ADDITIONAL FORMULA . .. .. . .. . 823
    Properties of a Right-angled Triangle 832
    Properties of any Triangle... ... ... 835
    Area of a Triangle ... ... 838
    Relations between a Triangle and the Inscribed ,
    Escribed, and Circumscribed Circles ... 841
    Other Relations between the Sides and Angles ofa Triangle 850
    Examples of the Solution of Triangles ... ... ... 859

    SECTION V. - SPHERICAL TRIGONOMETRY.


    INTRODUCTORY THEOREMS —
    Definitions 870
    Polar Triangle
    :

    RIGHT-ANGLED TRIANGLES
    Napier's Rules ... ... ... 881
    OBLIQUE-ANGLED TRIANGLES.
    Formulæ for cos a and cos A ... 882
    884
    The S Formulæ for sin {A, sin ja, & c.
    sin A sin B _ sino
    :

    894
    sin a sin b sinc . 895
    cos b cos C = cot a sin b - cot A sin C
    Napier's Formulæ ... ... 896
    : :

    Gauss's Formula ... 897


    SPHERICAL TRIANGLE AND Circle
    Inscribed and Escribed Circles 898
    Circumscribed Circles 900
    :

    SPHERICAL AREAS ,
    SphericalExcess ... 902
    Area of Spherical Polygon 903
    : : : :

    Cagnoli's Theorem ... 904


    Lhuillier's Theorem 905
    CONTENTS. xvii
    No. of
    Article .
    POLYHEDRONS 906
    The five Regular Solids ... 907
    The Angle between Adjacent Faces 909
    Radii of Inscribed and Circumscribed Spheres ... 910

    SECTION VI. - ELEMENTARY GEOMETRY.


    MISCELLANEOUS PROPOSITIONS
    Reflection of a point at a single surface ... . 920
    do. do. at successive surfaces . 921
    Relations between the sides of a triangle, the segments
    of the base, and the line drawn from the vertex ... 922
    Equilateral triangle ABC ; PA² + PB + PC ? ... ... 923
    Sum of squares of sides of a quadrilateral ... ... 924
    Locus of a point whose distances from given lines or
    points are in a given ratio ... 926
    To divide a triangle in a given ratio , 930
    Sides of triangle in given ratio. Locus of vertex 932
    Harmonic division of base . ... 933
    Triangle with Inscribed and Circumscribed circles 935
    THE PROBLEMS OF THE TANGENCIES 937
    Tangents and chord of contact. By = u ? ... , 948
    To find any sub-multiple of a line ... . .. .. . 950
    Triangle and three concurrent lines; Three cases 951
    Inscribed and Escribed circles ; 8, 8 - a, & c. 953
    NINE-Point CIRCLE. .. ... ... ... ... ... .. . 954
    CONSTRUCTION OF TRIANGLES . 960
    Locus of a point from which the tangents to two circles
    have a constant ratio ... 963
    COLLINEAR AND CONCURRENT SYSTEMS .. 967
    Triangle of constant species circumscribed or inscribed
    to a triangle ... ... .. 977
    RADICAL Axis
    Of two Circles 984
    Of three Circles ... 997
    INVERSION
    Inversion of a point 1000
    do. circle 1009
    do. right line 1012
    POLE AND POLAR ... 1016
    COAXAL CIRCLES ... 1021
    CENTRES AND Axes Of SIMILITUDE
    Homologous and Anti-homologous points 1037
    do. do. chords ... 1038
    C
    xviii CONTENTS .
    No. of
    Article.
    Constant product of anti-similitude 1043
    Circle of similitude 1045
    Axes of similitude of three circles 1046
    Gergonne's Theorem ... 1049
    ANHARMONIC RATIO AND PENCIL ... 1052
    HOMOGRAPHIC SYSTEMS OF Points 1058
    INVOLUTION .. . . .. ... . . 1066
    PROJECTION ... . 1075
    On Perspective Drawing ... 1083
    Orthogonal Projection ... 1087
    Projectionsof the Sphere ... 1090

    :
    ADDITIONAL THEOREMS
    Squares of distances of P from equidistant points on a
    circle ... ... . .. .. . .. . . .. . .. 1094
    Squares of perpendiculars on radii, & c. ... ... ... 1095
    Polygon with inscribed and circumscribed circles. Sum
    of perpendiculars on sides, & c.... ... ... ... 1099

    * SECTION VII. - GEOMETRICAL CONICS.


    SECTIONS OF THE CONE
    Defining property of Conic PS = ePM .. . 1151
    Fundamental Equation ... ... ... 1156
    Projection from Circle and Rectangular Hyperbola 1158
    Joint PROPERTIES OF THE ELLIPSE AND IIYPERBOLA
    Definitions ... ... 1160
    CS : CA : CX 1162
    PS + PS = AA ... 1163
    CS = AC F BC? ... 1164
    SZ bisects LQSR ... 1166
    If PZ be a tangent PSZ is a right angle 1167
    Tangent makes equal angles with focal distances 1168
    Tangents of focal chord meet in directrix 1169
    CN . CT = AC ... ... 1170
    GS : PS = e . .. 1171
    NG : NC == BC2 : AC 1172
    Auxiliary Circle ... 1173
    PN : QN = BC : AC 1174
    PN ? : AN . NA' = BC? : . 1176
    Cn . Ct = BC .... 1177
    SY.SY' = 1°C .. 1178
    PE = AC ... ... ... 1179
    To draw two tangents ... 1180
    Tangents subtend equal angles at the focus 1181
    CONTENTS. xix
    No. of
    Article .
    ASYMPTOTIC PROPERTIES OF THE HYPERBOLA
    RN ? – PN ? = BC : 1183
    PR . Pr = BC .. 1184

    : : : : : : :
    : : : :
    CE = AC ... . 1186
    PD is parallel to the Asymptote 1187

    :
    QR = qr ... 1188
    PL = Pl_ and QV = V ... 1189
    QR . Qr = PL = RV - Q V2 1191

    :
    4PH . PK = CS ... ... 1192
    JOINT PROPERTIES OF ELLIPSE AND HYPERBOLA RESUMED . Con
    JUGATE DIAMETERS
    QV ? : PV . VP' = CD : CP 1193
    PF . CD = AC . BC 1194

    : : : : : : : : : : : : :
    PF . PG = BC and PF . PG = AC ? 1195
    PG . PG = CD2 ... 1197
    Diameter bisects parallel chords ... 1198
    Supplemental chords 1201
    Diameters are mutually conjugate 1202
    CV . CT = CP ... . .. ... 1203
    CN = dR, CR = pN ... ... 1205
    CN ? + C'R² = AC , DR + PN = BC? 1207
    CP + CD = AC' + BC ... 1211
    PS . PS = CD 1213
    OQ . Oq : 0Q'. Oq = CD : CD"? ... 1214
    SR : QL = e ... 1216
    Director Circle .. 1217
    Properties of Parabola deduced from the Ellipse 1219
    : :

    THE PARABOLA
    Defining property PS = PM .. 1220
    Latus Rectum = 4AS ... 1222
    If PZ be a tangent, PSZ is a right angle 1223
    Tangent bisects < SPM and SZIL 1224
    ST = SP = SG ... ... ... ... 1225
    :

    Tangents of a focal chord intersect at right angles in


    directrix ... 1226
    AN = AT ... 1227
    :

    NG = 2 AS ...
    : : : : : :

    1228
    : : : :

    PN ? = 4AS . AN .. 1229
    SA : SY : SP 1231
    : : : : : : :

    To draw two tangents 1232


    SQ : SO : SQ 1233
    ZOSQ = 0SQ and QOQ = iQSQ 1234
    DIAMETERS .. . .. . . .. 1235
    The diameter bisects parallel chords 1238
    :
    XX CONTENTS .
    No. of
    Article .
    QV: = 4PS. PV ... 1239
    OQ . 04 : 00 : 00 = PS : P'S ... 1242
    Parabola two-thirds of circumscribing parallelogram 1244
    METHODS OF Drawing A Conic ... 1245
    To find the axes and centre 1252
    To construct a conic from the conjugate diameters 1253
    CIRCLE OF CURVATURE ... ... ... .. 1254
    e, ?CD• PVVulult.t. **...
    Chord of curvature" a=turQV 1258
    Semi-chordsof curvature, CD CD CD 1259
    ull, CP PF AC ***
    In Parabola , Focal chord of curvature = 4SP ... 1260
    do. Radius of curvature = 2SP = SY 1261
    Common chords of a circle and conics are equally
    clined to the axis ... 1263
    To find the centre of curvature ... 1265
    MISCELLANEOUS THEOREMS . .. ... ... ... 1267
    INDEX TO PROPOSITIONS OF EUCLID
    REFERRED TO IN THIS WORK.

    The references to Euclid are made in Roman and Arabic numerals ; e.g. (VI. 19).

    BOOK I.
    I. 4. - Triangles are equal and similar if two sides and the included
    angle of each are equal each to each.
    I. 5. — The angles at the base of an isosceles triangle are equal.
    I. 6 . - The converse of 5 .
    I. 8 . - Triangles are equal and similar if the three sides of each are
    equal each to each .
    I. 16 . — The exterior angle of a triangle is greater than the interior
    and opposite .
    I. 20 . — Two sides of a triangle are greater than the third .
    I. 26 . - Triangles are equal and similar if two angles and one corres
    ponding side of each are equal each to each.
    I. 27 . - Two straight lines are parallel if they make equal alternate
    angles with a third line.
    I. 29. - The converse of 27.
    I. 32. — The exterior angle of a triangle is equal to the two interior
    and opposite ; and the three angles of a triangle are equal
    to two right angles.
    Cor. 1. — The interior angles of a polygon of n sides
    = (n - 2 ) a .
    Cor. 2. - The exterior angles = 27 .
    I. 35 to 38 . - Parallelograms or triangles upon the same or equal
    bases and between the same parallels are equal.
    I. 43.— The complements of the parallelograms about the diameter
    of a parallelogram are equal.
    I. 47. — The square on thehypotenuse of a right-angled triangle is
    equal to the squares on the other sides .
    I. 48. - The converse of 47.
    xxii INDEX TO PROPOSITIONS OF EUCLID

    BOOK II.
    II. 4. - If a, b are the two parts of a right line,(a + b)' = a + 2ab + b?.
    If a right line be bisected, and also divided , internally or
    externally, into two unequal segments, then
    II. 5 and 6 . - The rectangle of the unequal segments is equal to the
    difference of the sqnares on half* the line, and on the line
    between the points of section ; or (a + b ) (a - b) = a* — 5 .
    II. 9 and 10. - The squares on the sameunequalsegments are together
    double the squares on the other parts ; or
    (a + b)? + (a − b) = 2a’ + 269.
    II. 11. — To divide a right line into two parts so that the rectangle
    of the whole line and one part may be equal to the square on
    the other part.
    II. 12 and 13. — The square on the base of a triangle is equal to the
    sum of the squares on the two sides plus or minus (as the
    vertical angle is obtuse or acute), twice the rectangle under
    either of those sides, and the projection of the other upon it ;
    or a = 1 ? + – 2bc cos A (702).

    BOOK III.
    III. 3. - If a diameter of a circle bisects a chord ,it is perpendicular to
    it : and conversely .
    III. 20. - The angle at the centre of a circle is twice the angle at the
    circumference on the same arc.
    III. 21. - Angles in the same segment of a circle are equal.
    III. 22. — The opposite angles of a quadrilateral inscribed in a circle
    are together equal to two right angles .
    III. 31. — The angle in a semicircle is a right angle.
    III. 32. — The angle between a tangent and a chord from the point of
    contact is equal to the angle in the alternate segment.
    III . 33 and 34. — To describe or to cutvff' a segment of a circle which
    shall contain a given angle.
    III. 35 and 36. — The rectangle of the segments of any chord of a
    circle drawn through an internal or external point is equal
    to the square of the semi-chord perpendicular to the
    diameter through the internal point, or to the square of the
    tangent from the external point.
    III. 37. — The converse of 36. If the rectangle be equal to the square,
    the line which meets the circle touches it.
    REFERRED TO IN THIS WORK . xxiii

    BOOK IV .
    IV . 2.— To inscribe a triangle of given form in a circle.
    · 3. — To describe the same about a circle .
    4 . - To inscribe a circle in a triangle.
    IV. 5. — To describe a circle about a triangle.
    IV . 10. – To construct two-fifths of a right angle .
    IV . 11. – To construct a regular pentagon.

    :!

    BOOK VI.
    VI. 11.–
    VI. Triangles and
    . - Triangles their bases. deofof athetriasame
    onaltoparallelograms ngle cualtitude
    ts are

    VI. 2. - á right line parallel to the side of a triangle cuts the other
    sides proportionally ; and conversely.
    VI. 3 and A . — The bisector of the interior or exterior vertical angle of
    a triangle divides the base into segments proportional to
    the sides .
    VI. 4.- Equiangular triangles have their sides proportional homo
    logously .
    VI. 5 . — The converse of 4 .
    VI. 6.— Two triangles are equiangular if they have two angles equal,
    and the sides about them proportional.
    VI. 7 . - Two triangles are equiangular if they have two angles equal
    and the sides about two other angles proportional, provided
    that the third angles are both greater than, both less than,
    or both equal to a right angle.
    VI. 8. - A right-angled triangle is divided by the perpendicular from
    the
    the right angle upon the hypotenuse into triangles similar
    to itself.
    VI. 14 and 15 . - Equal parallelograms, or triangles which have two
    angles equal, have the sides about those angles reciprocally
    proportional; and conversely, if the sides are in this pro
    portion, the figures are equal.
    VI. 19. — Similar triangles are in the duplicate ratio of their homo
    logous sides.
    VI. 20, - Likewise similar polygons.
    VI. 23. - Equiangular parallelograms are in the ratio compounded of
    the ratios of their sides.
    V1. B . – The rectangle of the sides of a triangle is equal to the square
    of the bisector of the vertical angle plus the rectangle of
    the segments of the base.
    xxiv INDEX TO PROPOSITIONS OF EUCLID.

    VI. C . - The rectangle of the sides of a triangle is equal to the rect


    angle under the perpendicular from the vertex on the
    base and the diameter of the circumscribing circle .
    VI. D . - Ptolemy's Theorem . The rectangle of the diagonals of a
    quadrilateral inscribed in a circle is equal to both the
    rectangles under the opposite sides.

    BOOK XI.
    XI. 4 . - A right line perpendicular to two others at their point of
    intersection is perpendicular to their plane.
    XI. 5.— The converse of 4. If the first line is also perpendicular to a
    fourth at the same point, that fourth line and the other
    two are in the same plane.
    XI. 6 .— Right lines perpendicular to the same plane are parallel.
    XI. 8. - If one of two parallel lines is perpendicular to a plane, the
    other is also.
    XI. 20. — Any two of three plane angles containing a solid angle are
    greater than the third .
    XI. 21. — The plane angles of any solid angle are together less than
    four right angles .
    TABLE OF CONTENTS .

    PART II.
    SECTION VIII. - DIFFERENTIAL CALCULUS. No. of
    Article ,
    INTRODUCTION ... 1400
    Successive differentiation 1405
    Infinitesimals. Differentials ... : 1407
    DIFFERENTIATION .
    Methods. .. . .. . . .. . 1411-21
    Successive DIFFERENTIATION
    :

    Leibnitz 's theorem ... 1460


    Derivatives of the nth order (see Index ) 1461- 71
    PARTIAL DIFFERENTIATION . .. . .. ... .. . 1480
    : : : :

    THEORY OF OPERATIONS 1483


    Distributive, Commutative, and Index laws... 1488
    EXPANSION OF EXPLICIT FUNCTIONS
    Taylor's and Maclaurin's theorems ... ... ... 1500, 1507
    Symbolic forms of the same ... . 1520 – 3
    f (a + h, y + k ), & c. ... 1512 -4
    Methods of expansion by indeterminate coefficients.
    Four rules. .. ... ... ... ... 1527 – 34
    Method by Maclaurin 's theorem ... 1524
    Arbogast's method of expanding ♡ ( ) 1536
    Bernoulli's numbers ... 1539
    Expansions of q (x + h ) - ° (w). Stirling and Boole 1546- 7
    EXPANSIONS OF IMPLICIT FUNCTIONS
    Lagrange's, Laplace's, and Burmann's theorems, 1552, 1556 -63
    Cayley's series for – 1555
    Abel's series for + a) 1572
    INDETERMINATE FORMS ... .. . 1580
    : : : :

    JACOBIANS ... . .. .... 1600


    Modulus of transformation 1604
    d
    xxvi CONTENTS .
    No. of
    Article,
    QUANTICS ... ... ... ... ... ... ... 1620
    Euler's theorem 1621
    Eliminant,Discriminant, Invariant,Covariant,Hessian 1626 – 30
    Theorems concerning discriminants ... 1635 –45
    Notation A = bc - f ?, & c . ... 1642
    Invariants . .. ... .. 1648 –52
    Cogredients and Emanents 1653 –5
    Implicit FUNCTIONS

    :
    One independent variable ... 1700
    Two independent variables ... · 1725
    n independent variables 1737

    :
    CHANGE OF THE INDEPENDENT VARIABLE 1760
    Linear transformation ... ... 1794

    : : : : :
    Orthogonal transformation ... 1799
    Contragredient and Contravariant 1813
    Notation %. = p, & c., q, r, s, t... 1815
    MAXIMA AND MINIMA —
    One independent variable ... , 1830
    Two independent variables ... 1841
    :

    Three or more independent variables... 1852


    Discriminating cubic ... ... ... 1849
    :

    Method of undetermined multipliers ... 1862


    Continuous maxima and minima ... 1866
    : :

    SECTION IX . - INTEGRAL CALCULUS.


    INTRODUCTION ... 1900
    Multiple Integrals 1905
    METHODS OF INTEGRATION -
    By Substitution , Parts, Division, Rationalization,
    Partial fractions, Infinite series ... 1908- 19
    STANDARD INTEGRALS ... ... ... 1921
    VARIOUS INDEFINITE INTEGRALS
    Circular functions ... ... ... 1954
    Exponential and logarithmic functions 1998
    Algebraic functions .. ... ... 2007
    Integration by rationalization ... ... 2110
    Integrals reducible to Elliptic integrals 2121- 47
    Elliptic integrals approximated to ... 2127
    SUCCESSIVE INTEGRATION 2148
    HYPERBOLIC FUNCTIONS cosh x, sinh 2, tanbæ 2180
    Inverse relations ... ... 2210
    Geometricalmeaning of tanh S ... 2213
    Logarithm of an imaginary quantity ... 2214
    CONTENTS . xxvii
    No. of
    Article .
    DEFINITE INTEGRALS
    Summation of series ... ... ... ... 2230
    THEOREMS RESPECTING LIMITS OF INTEGRATION 2233
    METHODS OF EVALUATING DEFINITE INTEGRALS (Eight rules ) 2245
    DIFFERENTIATION UNDER THE SIGN OF INTEGRATION ... 2253
    Integration by this method ... ... 2258
    Change in the order of integration ... 2261
    A PPROXIMATE INTEGRATION BY BERNOULLI'S SERIES 2262
    THE INTEGRALS B (1,m ) and r (n ) ... 2280
    log 1 (1 + n ) in a converging series ... 2294
    Numerical calculation of r (2 ) 2317
    INTEGRATION OF ALGEBRAIC FORMS ... 2341
    INTEGRATION OF LOGARITHMIC AND EXPONENTIAL FORMS 2391
    INTEGRATION OF CIRCULAR FORMS . .. 2451
    INTEGRATION OF CIRCULAR LOGARITHMIC AND EXPONENTIAL FORMS 2571
    MISCELLANEOUS THEOREMS
    Frullani's, Poisson 's, Abel's,Kummer's,and Cauchy's
    formulæ ... ... ... 2700 - 13
    FINITE VARIATION OF A PARAMETER ... " 2714
    Fourier's formula 2726
    THE FUNCTION + (x ) ... 2743
    Summation of series by the function y ( ) ... 2757
    ( x) as a definite integral independent of y (1) ... 2766
    NUMERICAL CALCULATION OF LOG ' (a ) 2771
    CHANGE OF THE VARIABLES IN A DEFINITE MULTIPLE INTEGRAL 2774
    MULTIPLE INTEGRALS
    EXPANSIONS OF FUNCTIONS IN CONVERGING SERIES
    Derivatives of the nth order ... 2852
    Miscellaneous expansions 2911
    Legendre's function X , 2936
    EXPANSION OF FUNCTIONS IN TRIGONOMETRICAL SERIES 2955
    APPROXIMATE INTEGRATION .... 2991
    Methods by Simpson, Cotes, and Gauss 2992- 7

    SECTION X .- CALCULUS OF VARIATIONS.


    Functions. OF ONE. INDEPENDENT VARIABLE 3028
    Particular cases... 3033
    Other exceptional cases 3045
    FUNCTIONS OF TWO DEPENDENT VARIABLES 3051
    Relative maxima and minima... 3069
    Geometrical applications ... 3070
    xxviii CONTENTS .
    No , of
    Article.
    FUNCTIONS OF TWO INDEPENDENT VARIABLES ... 3075
    Geometrical applicatious 3078
    APPENDIX
    On the general object of the Calculus of Variations... 3084
    Successive variation ... ... ... ... 3087
    Immediate integrability .. 3090

    SECTION XI. - DIFFERENTIAL EQUATIONS.


    GENERATION OF DIFFERENTIAL EQUATIONS 3150
    DEFINITIONS AND RULES.. . 3158
    SINGULAR SOLUTIONS .. 3168
    FIRST ORDER LINEAR EQUATIONS 3184
    Integrating factor for Mdx + Ndy = 0 3192
    Riccati's Equation 3214
    FIRST ORDER NON -LINEAR EQUATIONS... 3221
    Solution by factors 3222
    Solution by differentiation ... 3236
    HIGHER ORDER LINEAR EQUATIONS ... 3237
    Linear Equations with Constant Coefficients 3238
    HIGHER ORDER NON-LINEAR EQUATIONS 3251
    Depression of Order by Unity... ... 3262
    Exact DIFFERENTIAL EQUATIONS ... 3270
    MISCELLANEOUS METHODS ... ... 3276
    Approximate solution of Differential Equations
    Taylor's theorem ... ... ... 3289
    SINGULAR SOLUTIONS OF Higher Order EQUATIONS ... 3301
    EQUATIONS WITH MORE THAN TWO VARIABLES ... 3320
    SIMULTANEOUS EQUATIONS WITH ONE INDEPENDENT VARIABLE 3340
    PARTIAL DIFFERENTIAL EQUATIONS ... ... 3380
    Linear first order P . D . Equations ... 3381
    Non-linear first order P. D . Equations 3399
    Non -linear first order P . D . Equations with more
    than two independent variables 3409
    SECOND ORDER P . D . EQUATIONS 3420
    LAW OF RECIPROCITY ... 3446
    SYMBOLIC METHODS . 3470
    SOLUTION OF LINEAR DIFFERENTIAL EQUATIONS BY SERIES 3604
    SOLUTION BY DEFINITE INTEGRALS ... ... 3617
    P . D . EQUATIONS WITH MORE THAN TWO INDEPENDENT VARIABLES 3629
    DIFFERENTIAL RESOLVENTS OF ALGEBRAIC EQUATIONS 3631
    CONTENTS . xxix
    No. of
    Article .

    SECTION XII. – CALCULUS OF FINITE


    DIFFERENCES.
    FORMULÆ FOR FIRST AND nth Differences ... 3706
    Expansion by factorials 3730
    Generating functions ... .. 3732
    The operations E , A , and dr ... 3735
    Herschel's theorem ... ... 3757
    A theorem conjugate to Maclaurin 's 3759
    INTERPOLATION ... ... 3762
    Lagrange's interpolation formula 3768
    MECHANICAL QUADRATURE ... 3772
    Cotes's and Gauss's formule 3777
    Laplace's formula 3778
    SUMMATION OF SERIES ... 3781
    APPROXIMATE SUMMATION 3820

    SECTION XIII.— PLANE COORDINATE GEOMETRY.


    SYSTEMS OF COORDINATES
    Cartesian, Polar, Trilinear, Areal, Tangential, and
    Intercept Coordinates ... 4001 - 28

    ANALYTICAL CONICS IN CARTESIAN COORDINATES.


    LENGTHS AND AREAS ... . .. 4032
    4048
    : : : : : : : : : : : : : : :

    TRANSFORMATION OF COORDINATES
    The Right LINE . . ... 4052
    Equations of two or more right lines... 4110
    GENERAL METHODS ... 4114
    Poles and Polars 4124
    TAE CIRCLE ... " 4136
    Co -axal circles ... 4161
    TAE PARABOLA ... ... 4200
    THE ELLIPSE AND HYPERBOLA ... 4250
    Right line and ellipse ... ... 4310
    Polar equations of the conic ... 4336
    Conjugate diameters ... ... 4346
    Determination of various angles 4375
    THE HYPERBOLA REFERRED TO ITS ASYMPTOTES 4387
    The rectangular hyperbola . . 4392
    XXX CONTENTS .
    No. of
    Article .
    THE GENERAL EQUATION 4400
    The ellipse and hyperbola 4402
    Invariants of the conic 4417
    The parabola ... ... 4430
    Method without transformation of the axes ... 4445
    Rules for the analysis of the general equation 4464
    Right line and conic with the general equation 4487
    Intercept equation of a conic ... 4498
    SIMILAR CONICS ... ... ... . 4522
    CIRCLE OF CURVATURE
    Contact of Conics .. . ... 4527
    ConFOCAL Conics 4550
    ANALYTICAL CONICS IN TRILINEAR COORDINATES.
    The Right LINE 4601
    Equations of particular lines and coordinate ratios
    of particular points in the trigon 4628
    ANHARMONIC Ratio ... . .. 4648
    The complete quadrilateral ... 4652
    The GENERAL EQUATION OF A Conic ... 4656
    : : :

    Director-Circle ... ... ... 4693


    PARTICULAR CONICS ... ... 4697
    Conic circumscribing the trigon 4724
    Inscribed conic of the trigon ... 4739
    : : : : : : : :

    Inscribed circle of the trigon ... 4747


    General equation of the circle... 4751
    Nine-point circle . ..... ... 4754
    Triplicate-ratio circle ... ... 47546
    Seven-point circle ... 47540
    CONIC AND SELF-CONJUGATE TRIANGLE... 4755
    On lines passing through imaginary points ... 4761
    Carnor's, PASCAL'S, AND BRIANCHON'S THEOREMS
    BNLUS ... 4778– 83
    THE CONIC REFERRED TO TWO TANGENTS AND THE CHORD OF
    CONTACT - .
    Related conics ... .... 4803
    ANHARMONIC Pencils Of Conics ... ... 4809
    CONSTRUCTION OF CONICS .. .. ... 4822
    Newton's method of generating a conic 4829
    Maclaurin's method of generating a conic 4830
    THE METHOD OF RECIPROCAL POLARS... 4844
    TANGENTIAL COORDINATES 4870
    Abridged notation 4907
    CONTENTS . xxxi
    No. of
    Article.
    ON THE INTERSECTION OF Two Conics —
    Geometrical meaning of ✓ ( - 1). . .. . ... 4916
    THE METHOD OF PROJECTION ... ... 4921
    INVARIANTS AND COVARIANTS 4936
    To find the foci of the general conic 5008

    THEORY OF PLANE CURVES.


    Tangent AND NORMAL ..." 5100
    RADIUS OF CURVATURE AND EVOLUTE ... 5134
    INVERSE PROBLEM AND INTRINSIC EQUATION 5160

    ; : : :
    ASYMPTOTES . .. ... .. . 5167
    Asymptotic curves .. . 5172
    SINGULARITIES OF Corves---
    Concavity and Convexity 5174
    Points of inflexion, multiple points, & c. 5175 -87
    : : : : : :

    CONTACT OF CURVES ... 5188


    ENVELOPES ... 5192
    INTEGRALS OF CURVES AND AREAS 5196
    INVERSE Curves... 5212
    : : :

    PEDAL CURVES ... ... 5220


    ROULETTES 5229
    Area, length, and radius of curvature ... 5230 - 5
    : :

    The envelope of a carried curve 5239


    Instantaneous centre ... 5243
    Holditch 's theorem 5244
    : : : : :

    Trajectories ... 5246


    Curves of pursuit 5247
    Caustics... .. 5248
    Quetelet's theorem 5249
    TRANSCENDENTAL AND OTHER CURVES
    The cycloid ... 5250
    The companion to the cycloid ... 5258
    : :

    Prolate and cartate cycloids ... 5260


    Epitrochoids and hypotrochoids 5262
    Epicycloids and hypocycloids ... 5266
    The Catenary ... 5273
    The Tractrix ... 5279
    : :

    The Syntractrix . .. 5282


    The Logarithmic Curve 5284
    The Equiangular Spiral 5288
    : : : :

    The Spiral of Archimedes 5296


    The Hyperbolic or Reciprocal Spiral... 5302
    xxxii CONTENTS .
    No. of
    Article .
    The Involute of the Circle ... 5306
    The Cissoid

    : : : : : :
    ... ... 5309
    The Cassinian or Oval of Cassini 5313
    The Lemniscate 5317
    The Conchoid ... . 5320
    The Limaçon ... 5327
    The Versiera (or Witch of Agnesi) 5335
    The Quadratrix ... 5338
    The Cartesian Oval ...

    : : : : : : : : : : : : :
    5341
    The semi-cubical parabola 5359
    The folium of Descartes 5360
    LINKAGES AND LINKWORK : ... 5400
    Kempe's five-bar linkage. Eight cases ...5401- 5417
    Reversor, Multiplicator, and Translator 5407
    Pearcellier's linkage ... ... ... 5410
    The six -bar invertor ... ... ... 5419
    The eight-bar double invertor . .. 5420
    The Quadruplane or Versor Invertor 5422
    The Pentograph or Proportionator ... 5423
    The Isoklinostat or Angle-divider . .. 5425
    A linkage for drawing an Ellipse ... " ... 5426
    A linkage for drawing a Limaçon, and also a bicir
    ' cular quartic . ... ... 5427
    A linkage for solving a cubic equation 5429
    On three-bar motion in a plane : 5430
    The Mechanical Integrator ... 5450
    The Planimeter 5452

    SECTION XIV . - SOLID COORDINATE GEOMETRY .


    Systems of COORDINATES 5501
    The Right LINE 5507
    : : : : : : :

    THE PLANE 5545


    TRANSFORMATION OF COORDINATES 5574
    THE SPHERE ... ... ... 5582
    The Radical Plane 5585
    : : :

    Poles of similitude 5587


    CYLINDRICAL AND CONICAL SURFACES ... 5590
    Circular Sections 5596
    ELLIPSOID , HYPERBOLOID , AND PARABOLOID ... ...5599 -5621
    :
    CONTENTS . xxxiii
    No. of
    Article .
    CENTRAL QUADRIC SURFACE
    Tangentand diametral planes. .. 5626
    Eccentric values of the coordinates 5638
    CONFOCAL QUADRICS ... 5656
    Reciprocal and Enveloping Cones 5664
    THE GENERAL EQUATION OF A QUADRIC 5673
    RECIPROCAL POLARS 5704
    THEORY OF Tortuous Curves... ... 5721
    The Helix ... ... 5756
    GENERAL THEORY OF SURFACES —
    General equation of a surface ... 5780
    Tangent line and cone at a singular point 5783
    The Indicatrix Conic ... 5795
    Euler 's and Meunier's theorems 5806 – 9
    Curvature of a surface. .. 5826
    Osculating plane of a line of curvature 5835
    GEODESICS 5837- 48
    INVARIANTS .. . 5856
    INTEGRALS FOR VOLUMES AND SURFACES 5871
    Guldin 's rules ... ... 5879
    CENTRE OF Mass ... 5884
    MOMENTS AND PRODUCTS OF INERTIA ... 5903
    Momental ellipsoids ... ... 5925- 40
    Momental ellipse ... 5953
    Integrals for moments of inertia 5978
    PERIMETERS, AREAS, VOLUMES, CENTRES OF MASS, AND MOMENTS
    OF INERTIA OF VARIOUS FIGURES —
    Rectangular lamina and Right Solid ... 6015
    The Circle ... 6019
    : : : : :

    The Right Cone 6043


    Frustum of Cylinder 6048
    The Sphere ... 6050
    The Parabola ... 6067
    The Ellipse ... 6083
    Fagnani's, Griffith's, and Lambert's theorems ...6088 –6114
    :

    The Hyperbola ... 6115


    The Paraboloid ... ... 6126
    :

    The Ellipsoid ... 6142


    Prolate and Oblate Spheroids... 6152 – 65
    : :
    PREFACE TO PART II.

    APOLOGIES for the non- completion of this volume at an


    earlier period are due to friends and enquirers . The labour
    involved in its production, and the pressure of other duties ,
    must form the author's excuse.
    In the compilation of Sections VIII. to XIV ., the
    following works have been made use of:
    Treatises on the Differential and IntegralCalculus, by Bertrand,
    Hymer, Todhunter, Williamson , and Gregory's Examples
    on the samesubjects ; Salmon 's Lessons on Higher Algebra .
    Treatises on the Calculus of Variations, by Jellett and Tod
    hunter ; Boole' s Differential Equations and Supplement ;
    Carmichael's Calculus of Operations ; Boole' s Calculus of
    Finite Differences, edited by Moulton.
    Salmon's Conic Sections ; Ferrers's Trilinear Coordinates ;
    Kempe on Linkages ( Proc . of Roy . Soc ., Vol. 23 ) ; Frost
    and Wolstenholme's Solid Geometry ; Salmon 's Geometry
    of Three Dimensions.
    Wolstenholme's Problems.

    The Index which concludes the work, and which, it is


    hoped , will supply a felt want , deals with 890 volumes of
    32 serial publications : of these publications, thirteen belong
    to Great Britain , one to New South Wales , two to America ,
    four to France, five to Germany, three to Italy, two to
    erman

    Russia , and two to Sweden .


    As the volumes only date from the year 1800, the
    xxxvi PREFACE .

    important contributions of Euler to the “ Transactions of


    the St. Petersburg Academy,” in the last century, are
    excluded. It was, however, unnecessary to include them ,
    because a very complete classified index to Euler 's papers,
    as well as to those of David Bernoulli, Fuss, and others in
    the same Transactions, already exists.
    The titles of this Index, and of the works of Euler
    therein referred to, are here appended , for the convenience
    of those who may wish to refer to the volumes.
    Tableau général des publications de l'Académie Impériale de
    St. Pétersbourg depuis sa fondation . 1872. [B. M .C .:* R .R .
    2050, e.]
    I. Commentarii
    6
    Academiæ Scientiarum Imperialis Petropolitanæ .
    170 15
    1726 – 1746 ; 14 vols. [ B . M . C . : 431, f .]
    II. Novi Commentarii A . S. I. P . 1747– 75 , 1750 –77 ; 21 vols.
    [ B . M . C . : 431, f. 15 – 17, g. 1 - 16 , . 1 , 2 .]
    III. Acta A . S . I. P . 1778 – 86 ; 12 vols. [ B . M . C : 431, h .3 – 8 ; or
    T. 0 . 8 , a . 11.]
    IV . Nova Acta A . S. I. P . 1787 – 1806 ; 15 vols. [B . M . C . : 431,
    h . 9 – 15 , i. 1 - 8 ; or T . 0 . 8, a .23.]
    V . Leonhardi Euler Opera minora collecta, vel Commentationes Arith
    meticæ collectæ ; 2 vols . 1849. [ B . M . . : 8534, ee .]
    VI. Opera posthuma mathematica et physica ; 2 vols. 1862. [B . M .C .:
    8534, f.]
    VII. Opuscula analytica ; 1783– 5 ; 2 vols. [ B . M . C .: 50,i. 15 .]
    Analysis infinitorum . [ B . M . C . : 529,6 . 11. ]

    G . S. C.
    ENDSLEIGH GARDENS,
    London, N .W ., 1886.

    * British Museum Catalogue.


    ERRATA CONTINUED FROM PAGE X.
    (Corrections which are important are marked with an asterisk.)
    como
    Page 1, Line 7, for volume read weight.
    dos
    „ gramme-million gramme-six.
    añoſó

    ,, 1.4971499 , .4971499 .
    „, .6679358 1.1447299.
    „ 25 2 15 .
    259, , a? + B2 (a ” + 89)" .
    276 , „ 3n2+ n - 1 3n2 + 3n - 1.
    Toamna

    291, 1, 8
    292, „ . 3 & 4, , a .

    322, » 45 and 13 35 and 10.


    361, , 7, 3528
    459,
    470, 91,
    9, ,
    „
    -6
    Xm
    , - 16 .
    489, 9, „ -
    655, „, a number of rows two columns.
    693, 12, ,, R and R R, and R .
    604, » one-sixtieth one-ninetieth .
    713, 2, „ II. III.
    897, 5 , cos fc sin c .
    922 ii., , b2 + 2c » 262 + c%.
    949, » D » C.
    1076, dele “ The projections . .. are parallel.”
    1158, , last, „, 1201 , 1217.
    1178, „ 4, ,, PS PS'.
    1241, , 1, y parallel conjugate.
    1413, „ 3, , - dudv + du dv .
    , 1491, 3, -
    ., 1849, 1, +
    » 1903, footnote, » J (2 ) » f(x).
    1925, supply dx.
    1954-6, supply x.
    2030-2, erroreous, because l in (1427) is necessarily an integer .
    2035, yax » a.
    2140, , 1, applies to the whole denominator.
    2136, „, last, „ 2294 , 2293.
    klas+i

    2354,
    2392, X - 1.
    I alev

    2465,
    3237, , 1, · (n − 1) x (n — 1).
    3751, supply ur.
    4678, dele 2 in the second term .
    10taon

    4680, supply the factor 4 on the left.


    4692, dele 2 in the second term .
    4903, supply the factor 4 on the left.
    5154, , 3155 5155 .
    5330 , , 2, » m b.
    and refer to Fig . 129, Art . 5332 on the cardioid is wanting.
    CRITIRSITZ

    MATHEMATICAL TABLES.

    INTRODUCTION .
    The Centimetre-Gramme- Second system of units .
    NOTATION . — The decimalmeasures of length are the kilo
    metre, hectometre , decametre , metre, decimetre, centimetre,
    millimetre. The same prefixes are used with the litre and erscita
    gramme for measures of capacity and volume, weight
    Also , 10 metres is denominated a metre-seven ; 10 - 7 metres,
    a seventh -metre ; 1015 grammes, a gramme-fifteen ; and so on .
    A gramme-million is also called a megagramme ; and a
    millionth -gramme, a microgramme; and similarly with other
    measures.
    DEFINITIONS. — The C . G . S . system of units refers all phy
    sicalmeasurements to the Centimetre (cm . ) , theGramme (gm .) ,
    and the Second (sec.) as the units of length , mass, and time.
    The quadrant of a meridian is approximately a metre
    seven . More exactly , one metre = 3 .2808694 feet = 39.370432
    inches.
    The Gramme is the Unit of mass, and the weight of a
    gramme is the Unit of weight, being approximately the weight
    of a cubic centimetre of water ; more exactly , 1 gm . =
    15.432349 grs.
    The Litre is a cubic decimetre : but one cubic centimetre
    is the C . G . S . Unit of volume.
    1 litre = .035317 cubic feet = .2200967 gallons.
    The Dyne (dn .) is the Unit of force , and is the force which ,
    in one second generates in a gramme of matter a velocity of
    one centimetre per second.
    The Erg is the Unit of work and energy , and is the work
    done by a dyne in the distance of one centimetre.
    The absolute Unit of atmospheric pressure is onemegadyne
    per square centimetre = 74: 964 cm ., or 29.514 in . of mercurial
    column at 0° at London , where g = 981:17 dynes.
    Elasticity of Volume = k , is the pressure per unit area
    upon a body divided by the cubic dilatation.
    B
    MATHEMATICAL TABLES .

    Rigidity = n , is the shearing stress divided by the angle


    of the shear.
    Young's Modulus = M , is the longitudinal stress divided
    by the elongation produced , = Ink - (3% + n ).
    Tenacity is the tensile strength of the substance in dynes
    per square centimetre.
    The Gramme-degree is the Unit of heat, and is the amount
    of heat required to raise by 1°C . the temperature of 1 gramme
    of water at or near 0°.
    Thermal capacity of a body is the increment of heat
    divided by the increment of temperature. When the incre
    ments are small, this is the thermal capacity at the given
    temperature .
    Specific heat is the thermal capacity of unit mass of the
    body at the given temperature.
    The Electrostatic unit is the quantity of electricity which
    repels an equal quantity at the distance of 1 centimetre with
    the force of 1 dyne.
    The Electromagnetic unit of quantity = 3 x 1010 electro
    static units approximately .
    The Unit of potential is the potential of unit quantity at
    unit distance .
    The Ohm is the common electromagnetic unit of resistance,
    and is approximately = 109 C . G . S . units.
    The Volt is the unit of electromotive force, and is = 108
    C . G . S . units of potential.
    The Weber is the unit of current, being the current due to
    an electromotive force of 1 Volt , with a resistance of 1 Ohm .
    It is = to C . G . S. unit.
    Resistance of a Wire = Specific resistance x Length - Section .
    Physical constants and Formulæ .
    In the latitude of London, g = 3:2: 19084 feet per second .
    = 981: 17 centimetres per second.
    In latitude i , at a height h above the sea level,
    g = ( 980 .6056 — 25028 cos 21 – 000003 h ) centimetres per second .
    Seconds pendulum = ( 99-3562 - -2536 cos 21 — .0000003 h ) centimetres.
    THE EARTH . - Semi-polar axis, 20,854895 feet* = 6 :35411 x 108 centims.
    Mean semi-equatorial diameter, 20,926202 „ * = 6 37824 x 108
    Quadrant ofmeridian , 39:377786 x 10 inches* = 1 :000196 x 10?metres.
    Volume, 1:08279 cubic centimetre-nines.
    Mass (with a density 53) = Six gramme.twenty-sevens nearly .
    * These dimensions are taken from Clarke's " Geodesy ," 1880.
    MATHEMATICAL TABLES.

    Velocity in orbit = 2933000 centims, per sec. Obliquity, 23° 27' 16". *
    Angular velocity of rotation = 1 : 13713.
    Precession, 50"-26.* Progression of Apse, 11":25. Eccentricity, e = .01679.
    Centrifugal force of rotation at the equator, 3.3912 dynes per gramme.
    Force of attraction upon moon , .2701. Force of sun 's attraction , •5839.
    Ratio of g to centrifugal force of rotation , girwa -= 289.
    Sun 's horizontal parallax, 8":7 to 9'.* Aberration, 20" :11 to 20":79. *
    Semi-diameter at earth 's mean distance, 16 ' l" :82 .*
    Approximate mean distance, 92,000000 miles, or 1.48 centimetre -thirteens.t
    Tropical year, 365-242216 days, or 31,550927 seconds.
    Sidereal year, 365.256374 , 31,558150 ,
    Anomalistic year, 365-259544 days. Sidereal day, 86164 seconds.
    THE MOON . - Mass = Earth 's mass x :011364 = 6 .98 102 grammes.
    Horizontal parallax. From 53' 56" to 61' 24 ".*
    Sidereal revolution , 27d. 7h . 43m . 11:46s. Luvarmonth, 29d . 12b . 44m . 2.87s.
    Greatest distance from the earth , 251700 miles, or 4 :05 centimetre-tens,
    Least 225600 , 3 .63
    Inclination of Orbit, 5° 9'. Annual regression of Nodes, 19° 20'.
    RULE. — ( The Year + 1 ) : 19. The remainder is the Golden Number.
    ( TheGolden Number - 1) 11 : 30 . The remainder is the Epact.
    GRAVITATION . - Attraction between masses ) min '
    m , m ' at a distance ! Sa 1 x 1:57:3 107 uyuer
    The mass which at unit distance (1 cm .) attracts an equal mass with unit
    force ( 1 dn.) is = ✓ (1:543 x 107) gm . = 3928 gm .
    WATER .— Density at 0°C., unity; at 4°, 1 000013 (Kupffer).
    Volume elasticity at 15°, 2. 2.2 x 1010.
    Compression for 1 megadyne per sq. cm ., 4:51 x 10 -6 (Amaury and
    Descamps ).
    The heat required to raise the temperature of a mass of water from 0° to
    tº is proportional to t + .00002 + 0000003 % (Regnault) .
    GASES. - Expansion for 1° C ., .003665 = 1 : 273.
    Specific heat at constant pressure – 1 :108 .
    Specific heat at constant volume
    Density of dry air at 0° with Bar. at 76 cm . = .0012932 gm . per cb. cm .
    (Regnault ) .
    At unit pres. (a megadyne) Density = .0012759.
    Density at press. p = px 1.2759 x 10 -9.
    Density of saturated steam at tº, with p taken - •79360987
    from Table II., is approximately 5 (1 + .003604) 109°
    SOUND. — Velocity = v (elasticity ofmedium - density ).
    Velocity in dry air at t = 33240 ✓ (1 + :00360t) centimetres per second.
    Velocity in water at 0° = 143000
    LIGHT. — Velocity in a medium of absolute refrangibility u
    = 3 :004 x 100 = u (Cornu ) .
    If P be the pressure in dynes per sq. cm ., and t the temperature,
    url = 2903 x 10 - 13P : ( 1 + :00366t) ( Biot & Arago ).

    . These data are from the “ Nautical Almanack ” for 1883.


    + Transit of Venus, 1871, “ Astrum . Sic. Notices," Vols. 37, 38.
    MATHEMATICAL TABLES.

    TABLE I.
    Various Measures and their Equivalents in C . G . S. units .
    Dimensions. Pressure.
    1 inch = 2:5400 cm 1 gm . persq .cm . = 981 dynespersq.cm .
    1 foot = 30:4797 1 lb. per sq. foot = 479
    1 mile = 160933 1 lb. per sq. in . = 68971
    1 nautical do. = 185230 , 76 centimetres)
    1 sq. inch = 6:4516 sq. cm . of mercury ) = 1,014,000 ,
    at 0° C .
    1 sq. foot = 929:01 ,, lbs . per sq. in . = 70-307 = .
    1 sq. yard = 8361. 13 ., ( 014223
    1 sg. mile = 2:59 x 1010,, gms. per sq. cm .
    1 cb. inch = 16.387 cb . cm . Force of Gravity
    " y ..
    1 cb . foot = 28316 upon 1 gramme = 981 dynes
    1 cb. yard = 764535 , 1 grain = 63:56777
    1 gallon = 4541 1 oz . = 2.7811 x 10 - ,,
    = 277:274 cb . in . or the vo , 1 lb . = 4 :4497 x 105 ,
    lume of 10 lbs. of water , 1 cwt. = 4 .9837 x 107
    at 62° Fab ., Bar. 30 in . 1 ton = 9.9674 X 108 ,
    Mass. Work (g = 981).
    1 grain = .06479895 gm . 1 gramme-centimetre = 981 ergs.
    1 ounce = 28: 3495 1 kilogram -metre = 981 x 105
    1 pound = 453-5926 , 1 foot-grain = 1.937 x 103 ",
    1 ton = 1,0160479 1 foot-pound = 1:356 x 107 ,
    1 kilogramme = 2. 20462125 lbs. 1 foot-ton = 3:04 x 1010 ,
    1 pound Avoir. = 7000 grains 1 'horse power' p . sec. = 7:46 x 100 ,
    1 pound Troy = 5760 , Heat.
    Velocity . 1 gramme-degree C . = 42 x 109 ergs.
    1 mile per hour = 44:704 cm . per sec. 1 pound -degree = 191 x 108
    1 kilometre . = 27 .777 92 1 pound -degree Fah . = 106 x 108 ,,

    TABLE II . TABLE III.


    Pressure of Aqueous Vapour in Values for the principal Lines of
    dynes per square centim . the Spectrum in air at 160°C .
    with Bar. 76 cm .
    Temp. Pressure. Temp. Pressure.
    - 20° 12:36 | 40° 73200 Wave-length No.of vibrations
    - 15° 1866 50° 122600 in centims. per second .
    2790 600 198500 7 .604 x 10 -5 3.950 x 101
    - 50 4150 800 472900 6 .867 4 :373
    6133 1000 1014000 6 :56201 , | 4 :577
    8710 1200 1988000 D (mean 5 .89212 , 5.097
    100 12:2:20 140° 3626000 5 . 26913 , 5 .700
    150 16930 160° 6210000 4 :86072 ,, 6 -179
    20° 23190 180° 10060000 4 :30725 6.973
    25° 31400 | 200° 15600000 3 . 96801 , 7.569
    30° 42050 3.93300 , 7.636
    IVinnits
    .CGSuTABLE
    Finansion Linear
    S| pecific
    Relative
    .MElect
    ofagn
    Rate
    Density 's
    Young Rigidity Elasticity f
    ?|oVolume
    bel
    Heat
    Expansion Cond
    Speciucti
    fic on
    Water Modulus of
    volume Tena
    . city tween Conduc |R1ofesistance
    Sound
    =1. n. k. pbetween
    degree
    I0 er
    &100C
    .10 00 ty
    t,1J.C&ivi 0°C
    p.asec
    cm
    inter
    Platinum 21 158
    9
    9000875
    1.0%x1|2:60335381
    00027
    Gold .26
    19 01483
    .01· 00045 4
    |,1.71000 2081
    Mercury .596
    13 x10-5012
    42 000180 ·0330 96190
    Lead :3012
    x1.011
    559 02861
    .01x12208
    800086 3
    1.2180 19850

    111
    Silver :47
    10 .012:6·0973 1
    0196
    00061
    0557 1521
    Copper 1.28834
    43 .6x14:401784 41 40175
    898
    0949
    .4·03.71000054 1615
    ,drawn
    Brass .0841 75 71 6.3 6 .01 0193
    00054 3.53576
    Iron
    ,cast 1.37235 49 5:32 , 90. 64 1.0· 01111
    00033 4:32
    wIron
    , rought 961.7 77 63 6.7 9 1:456 1.0•1098
    6.001258
    58 00040 5.06 9827
    Steel 2:17.839
    49 8:19 1.841 .
    379 001260
    ·000037 5.22
    ,cast
    Tin 7.29 ·03:1700063 ·00227 304 13360
    cZinc
    , ast 7:19 1:5 7 927
    0294
    00088
    -·0363 5690
    ,flint
    Glass 42
    0·62.9,03 42: 0 410
    : 15 1.0·•1770
    00015
    0081 ,54: 3

    V.
    TABLE
    Greatest
    distance Least Sidereal Inclination Diameter
    .
    Sun
    from distance Revolution Orbit
    of of
    Time in .
    Mass .
    Density
    mEarth
    's ean .
    Sun
    from Days
    in. .
    Ecliptic
    to .Rotation Miles
    .
    MATHEMATICAL TABLES.

    =l.
    distance
    .h .m 8.
    Son 600 0 888000 354936 2.0 5
    Mercury 0.46669 . 69
    93078707508 24 5 3000 0:118 2:01
    Venus .70 2826 32224
    -70. 301
    1840 23 21 7700 .80 83 .90 7
    Earth 1.01678 8322
    .90365
    -256 7926 1.000 1:00
    Mars 1.66578 15686
    -9.380
    18160 4500 0132 0:72
    Jupiter 5.45378 54332
    1.49885
    518240 92000 .034
    338 -20 4
    Saturn .07328
    10 2.210759
    -90920
    044228 75000 .0101
    64 0:13
    Uranus 020
    . 7612 18
    46
    .230686
    :88916 21 30 36000 :789
    14 0:15
    Neptune -29888
    30 -729
    1460126
    . 227506
    6 35000 . 48
    624 2.0 7
    MATIEMATICAL TABLES.

    TABLE VI. - Functions of a and e.


    # = 3:1415926 3183099 = 2:71828183
    7 ? = 9.8696044 •1013212 = 7 38905611
    71 % = 31.0062761 •0322515 = 0. 3678794
    a a = 1.7724539 2009 = 639.6619772 = 0 .1353353
    at

    .. ..
    r logo = 1:4971499 180° = = 570.2957795 logo e = 0 .43429448
    Cr
    log , a = 06679358 = 206264" :8 log, 10 = 2:30258509
    ttt1704a
    TABLE VII. TABLE VIII,
    No. Square root. / Cube root. log10 N . 1 log, N .
    1:4142136 | 1.2599210 •3010300 .69314718
    1.7320508 | 1 :4422496 -4771213 | 1.09861229
    2:0000000 1:5874011 .6989700 1.60943791
    2: 2360680 1.7099759 .8450980 1.94591015
    2:4494897 | 1:8171206 1.0413927 2: 39789527
    2.6457513 1.9129312 1.1139434 2:56494936
    2:8284271 2:0000000 1 .2304489 2 -83321334
    3 .0000000 | 2:0800837 1. 2787536 2 . 94443898
    3: 1622777 | 2: 1544347 1.3617278 3:13549422
    3:3166248 2: 2239801 1 :4623980 3 :36729583
    3:4641016 2: 2894286 1 :4913617 3:43398720
    13 3:6055513 2: 3513347 1.5682017 | 3.61091791
    14 3:7416574 2:4101422 1.6127839 3-71357207
    15 3.8729833 2:4662121 1.6334685 | 376120012
    4 :0000000 2:5198421 1.6720979 3 .85014760
    4 :1231056 2:5712816 1.7242759 | 3 :97029191
    4 . 2426407 2.6207414 1.7708520 4 :07753744
    4 :3588989 2:6684016 1.7853298 | 4 : 11087386
    4 :4721360 2:7144177 1.8260748 4 .20469262
    4 :5825757 2:7589243 71 | 1:8512583 | 4 .26267988
    4 .6904158 2 :8020393 1.8633229 4 .29045944
    4 .7958315 2:8438670 1.8976271 4 :36944785
    4 :8989795 2.8847991 1.9190781 4 :41884061
    5 .0000000 2:9240177 89 1. 9493900 | 4 :48863637
    5 .0990195 2.9624960 97 1.9867717 | 4 :57471098
    5 :1961524 3:0000000 101 2:0043214
    103 2:0128372
    4:61512052
    5.2915026 3:0365889 4:63472899
    5 :3851648 3:0723168 107 2.0293838 4 :67282883
    5:4772256 | 3:1072325 109 .2:0374265
    . 4:69134788

    NOTE. — The authorities for Table IV . are as follows: - Columns 2, 3, and


    4 (Mercury excepted), Everett's experiments (Pbil. Trans., 1867) ; g is here
    taken = 981:4. The densities in these cases are those of the specimens
    employed . Cols. 5 and 7, Rankine. Col. 6 , Watt's Dict. of Chemistry .
    Col. 8, Dulong and Petit. Col. 10, Wertheim . Col. 11, Matthiessen .
    Table V . is abridged from Loomis's Astronomy.
    The values in Table III. are Angström ’s.
    BURCKHARDT' S FACTOR TABLES.
    FOR ALL NUMBERS FROM 1 to 99000.

    EXPLANATION . — The tables give the least divisor of every


    number from 1 up to 99000 : but numbers divisible by 2, 3,
    or 5 are not printed . All the digits of the number whose
    divisor is sought, excepting the units and tens, will be found
    in one of the three rows of larger figures. The two remaining
    digits will be found in the left-hand column. The least divisor
    will then be found in the column of the first named digits, and
    in the row of the units and tens.
    If the number be prime, a cipher is printed in the place of
    its least divisor.
    The numbers in the first left -hand column are not conse
    cutive. Those are omitted which have 2 , 3 , or 5 for a divisor.
    Since 22 . 3 . 5% = 300 , it follows that this column of numbers
    will re-appear in the same order after each multiple of 300 is
    reached .
    MODE OF USING THE TABLES. — If the number whose prime
    factors are required is divisible by 2 or 5 , the fact is evident
    upon inspection , and the division must be effected . The
    quotient then becomes the number whose factors are required .
    If this number , being within the range of the tables, is yet
    not given , it is divisible by 3 . Dividing by 3 , we refer to the
    tables again for the new quotient and its least factor,and so on .
    EXAMPLES. – Required the prime factors of 310155.
    Dividing by 5 , the quotient is 62031. This number is within the range
    of the tables . But it is not found printed . Therefore 3 is a divisor of it.
    Dividing by 3 , the quotient is 20677. The table gives 23 for the least factor
    of 20677. Dividing by 23, the quotient is 899.
    The table gives 29 for the least factor of 899. Dividing by 29, the quo .
    tient is 31, a prime number. Therefore 310155 = 3. 5. 23 .29. 31.
    Again , required the divisors of 92881. The table gives 293 for the least
    divisor. Dividing by it, the quotient is 317. Referring to the tables for 317,
    a cipher is found in the place of the least divisor, and this signifies that 317
    is a prime number .
    Therefore 92881 = 293 x 317, the product of two primes.
    It may be remarked that, to have resolved 92881 into these factors with .
    out the aid of the tables by the method of Art. 360, would have involved
    fifty -nine fruitless trial divisions by prime numbers.
    ‫‪1184187‬‬
    ‫‪54157‬‬
    ‫‪935178‬‬

    ‫‪191‬‬
    ‫‪160‬‬
    ‫‪;6166‬‬
    ‫‪87|172‬‬

    ‫‪_0/13‬‬
    ‫‪_-29‬‬

    ‫‪이‬‬
    ‫‪2[193‬‬
    ‫‪19‬‬

    ‫|‬

    ‫‪711‬‬
    ‫| |‬

    ‫‪13‬‬

    ‫‪11‬‬
    ‫‪0‬ܙ‬

    ‫‪41‬‬
    ‫‪0/‬‬
    ‫‪71‬ܐ‬
    ‫‪111‬‬
    ‫‪171‬‬

    ‫‪59‬‬

    ‫‪/41‬‬
    ‫‪11‬‬

    ‫‪11‬‬
    ‫‪61‬‬
    ‫‪13‬‬

    ‫ܠܐܠܐ‬
    ‫‪0‬‬
    ‫‪.‬‬

    ‫‪291‬‬
    ‫‪01‬‬
    ‫ܙ‬

    ‫‪17‬‬
    ‫‪/31‬‬
    ‫‪/‬‬
    ‫ܟܢܤܩ ܚܗܒ‬
    ‫܂‬ ‫ܢܘ ܘ ܕܗ ܩ ܗ ܚܗ ܝ ܚ ܘ ܝܘ ܘ ܫ ܘ ܘ‬ ‫‪3‬‬
    ‫‪/297‬‬

    ‫‪_059‬‬
    ‫‪_37‬‬
    ‫‪|_5473‬‬

    ‫ܗ ܩܘܓܘ‬ ‫ܝ ܢ ܗ ܗ ܟ ܕ ܗ ܩ ܗܧܪܘ ‪-‬‬ ‫‪e be Je‬‬ ‫܂‬


    ‫‪|48‬ܘ ‪42 145‬‬

    ‫ܘ ‪ -‬ܘ ܗ ܗ ܚ ܕܘ ܕ ܘ ܕ ܗ ܚ‬
    ‫ܚ ܝ ܢ ܗ ܗ ܗ ܗ ܚ ܗ ܢܕ | ܗ‬ ‫ܢܢܟܗ‬
    ‫ܗ‬ ‫‪5 E‬‬ ‫ܗ ܘ ܢܗ ܩ ܝ ܢܗ ܗ‬
    ‫‪|37_43211354120896‬‬
    ‫‪43‬‬

    ‫‪ - €‬ܘ ܗ ܢܕ ܩ ܗ ܗ ܗ ܗ ܗ ܢܕ‬
    ‫ܘܪ ܗ‬ ‫ܙܝ‬
    ‫ܙ ܚ ܗܢܕ ܝ ܩ ܧ ܝ ܘ‬
    ‫ܘ ܘ ܢ ܚ ܝ ܘ ܘ ܨ ܘܨܢܘ ܘ ܘ ܚ ܘ ܘ ܘ ܘ ܝ ܘ ܨ ܘ ܘ | ܚ‬
    ‫‪47j‬‬
    ‫‪_0)nul‬‬

    ‫ܩ ‪ -‬ܘ ܘ ܘ ܝ ܘ ܘ ܢܕ ܝܕܗ ܘ ܘ ܢ ܚ ܘ ܘ ܨ ܘ ܘ ܟ ܘ ܘ‬
    ‫‪411‬‬

    ‫‪.171‬‬

    ‫‪01‬‬
    ‫‪0‬ܠ‬

    ‫‪7‬‬
    ‫‪11‬܂‬

    ‫‪11‬‬
    ‫܂‬‫‪3‬ܐ‬

    ‫‪237‬‬
    ‫‪171‬‬

    ‫‪_0‬‬
    ‫‪111‬‬
    ‫‪410‬‬
    ‫‪191‬‬

    ‫‪111‬‬
    ‫‪00‬‬

    ‫‪07‬‬
    ‫‪11‬‬

    ‫‪13‬‬
    ‫‪/13‬‬

    ‫‪41‬‬

    ‫‪31‬‬

    ‫‪1‬‬
    ‫‪/o_[10 9‬‬
    ‫‪_ 7‬‬
    ‫‪77‬‬
    ‫‪0/19‬‬‫‪17‬‬
    ‫‪0]1lu‬‬
    ‫‪|1060925‬‬

    ‫ܘܘ ܝ ܘ ܚ ܫ ܘ ܘ ܘ‬
    ‫‪-‬ܚܫ ܚ‬
    ‫‪0710‬‬
    ‫‪이‬‬
    ‫‪11‬‬
    ‫‪00|03‬‬

    ‫|‬
    ‫ܐܐ‬
    ‫|‬
    งง๑๐ ๐ ๐ ๐๐ง • = ๐ ๘ ๐ ๐๐๐๐๐ ผงะง ๐ ๐๐ ๐ ๐ ๐ ๐ ๘ ๐๐ ง ๐ ๐ ๐
    ๑๐๑ ง ๑๐๐๐ & ๖ง ๑๑งง• • - งง ๐ ๐ ๐๔ง• • • •
    02
    0| 5

    | 8 5ง๐๐๐ ๐ ๕ = ๐๐ ๐e ง ๖ง๑๐๐ ๐ มี 4ง = 6 ๑๐ผง๐ ๐ ๐


    7

    E๐๕๐ง ๑ ๒๐ง 6 0 ง ง = ๐ง ๐ ๐ - ๐๐
    19
    |
    |7
    103

    0
    019

    อง๑๐ ๐ ๐ ๒๒๐๐๕ ๑ง 24 66 ๑๐ง & ๑๐ง ๑ ๑๐.๐ะอง ๑๑ งง 599


    |13
    이

    0
    10
    6
    123|3 7

    !
    ol 23

    ๑๔ง อ้ะะ งง ๐ ๐ ๐ ๐ ๐ งe
    งง๑๕๘๐ อง : ง ง ๑๐๐ - ๑๕๐ ๐ = ง่ ๑
    )

    อ้ง๑๐๖ง ๑๐ งงะ ๑๐๐


    29

    .
    |4321417415963820

    ๑๑๐๑๖ง งดี • งงงะ ง๑๐ ๐ ๐ะงง๑


    14
    I II
    o

    •E1e e eyeseง องงE๐ ๐ ๐ ๐ ๐


    53 37| 6|
    4 o|

    ๐๐๐ง * งะ - ๑๐ง ๒๐ งอ ๕ ๐๐
    6

    ง๐๐ ง•••ง ๖งะ


    61|
    60
    |

    1|20 39

    งง• •ง๑๐๐๐๐๐
    เE

    969 = ๐ ๐ ๐ ๐ = ๐๔ง๐ เ ง ง ง ๑๐๐ง ๑๐ะ * ง ง ง งว

    6
    |

    \
    31

    0l
    23
    e - gะ๐ง : ๐ ๐ ๐ ๐ งะ • ง๑๐๐๐ งงะ •
    ของ E : ss3 4ะองง งง๑ - ๑๐
    10|2

    •ะ๑๐๐ ๐๒๐๐๐ ๐ = ง & งง ๑๐๘= =


    |67
    3113 ป

    0
    ๐ ง ๐ ๐ ๐ ง ๐ ๐ = 8 ง 4 ง:

    |0
    0

    4
    12
    ll

    10

    13
    13

    13

    -
    31
    ถึง๐๐๕ มี ๑๒๐๑๕ ๑งง ๒ ๕๑๔ง ๑๖ง 019
    eceง ๐ ๐ ๕ ๐๐ - ๑๐ ๑๐๐ ซึ่งอง
    0006 - 1

    87|96534782960
    |88346547
    111
    | 1
    147
    103
    1071
    113

    0
    11

    71
    13

    17
    1
    |171747

    4 - 38no 07807 Bogos agood 070000| 558708 0738 308 )

    이11

    10

    23

    0
    이

    |7666295
    normn 21|] 313 a B390 70007이 oonsg 70007 03 이
    )0
    53

    11|
    | 13| 11|

    moons onwoo

    11
    13
    이

    -
    |6565592

    371
    411
    731
    113

    171
    79
    -

    |11
    23
    19
    ]
    103

    moaom 07007 3880g congo 07030

    01
    |411

    13
    기
    .

    이
    | 1이10 이
    111
    183

    097 30705 009


    ]7|101

    370 0
    23

    11
    43

    이
    3

    |
    891
    이 7moo Aomoo 08070 0800m 7008

    01113
    13

    |11
    이
    148
    1

    |43933225304176
    |4447

    00072 00m00 7mooS goman a0707


    기

    | | 13]
    111
    411

    109
    101

    291

    107117
    61
    111 13

    83
    7
    .
    159

    01

    01
    |23

    1
    13
    |321091902578963014

    | 이47

    | 70080 oo78m 07080 03070 gooan


    0700m 03330 07070 03809 755070 _ oooss Nㅇㅇㅇㅇ
    |107
    11

    43
    101 53
    23

    13
    ]
    111

    3| 11

    Lon070 80000 709


    71
    19
    23

    |37

    19
    1
    19

    |
    Ol

    1031
    ] |

    147
    | |

    111
    171
    291

    177

    IN
    11
    41

    11
    10

    이
    !

    |
    |
    |09917403

    08708 oooom mono7 3079m oomono


    | 14087 00970 0030M7nmo7 08m400
    oooon 37037oogon omong graon
    5 =๑๐๕ งอะง๐ - ๑๕ ๘ ๑๐๐ ง๑๐๐
    13
    |11 0

    |99258
    | |

    n)67 7 이
    이
    29
    lll
    이 이
    0|97 23 37
    79
    131

    |01 47
    l0

    0/
    |

    ๑๐ง * * * * * ๐ ๐ = งงงง : ๐ ๐ ๐ % 8 | 9 | ๐ - ง๐ % ๑๐ อง ๐ ๐ อง
    |4i
    |23

    10
    73

    ET
    II
    43
    103

    131

    = = ๑ ผง9 อะงง ๐๕๔ง ๕๐ ง ๑๐ * | * | งออง ง ๑ะ = = ง 39


    0

    li

    IT
    13
    13

    08

    10
    107
    103
    0
    |

    ง่ ๑๒ะ๐ ๐ง ๐ ๐ ๐ ง๐๐๐ ๐ ง ง = 0 - 2
    79

    0
    |
    |
    29

    19
    10
    : 7
    0

    ๑๐๐ง = 8 9 10 ง & 6s • ง * ๐ ๐งมอง| อง= ๐ ๐ ๕ ง อ ง • = = -


    |11
    19
    19 0

    83|4i

    = ๐ ๐ ง ๐ ๐ ๐ ๐ ๔ง๑๐ง ง = ๐ผง๑๐
    13
    |0711
    0101
    53

    13
    167
    IRT

    I
    2

    1913

    • งง งง๐ ๐ ๐๐๖ง Eะอง๐ ๐๒๐๐ะ( ง


    20

    59
    43
    ICI

    0
    IL
    71|170 13

    01

    23
    "
    13
    10T

    10l
    10
    10
    |
    |

    องง ๐ ๕๐- ๐ ๐งง = ๐ ๐๐๐ง65 | 6 | & ะ ๑๐๔ง งง ๑๐ ๐ ๐ ๐ n


    4l

    13

    1
    103

    ๑๕๐๐ถึง ๑๐๑๐๐ ง อ ง ง 4 องงะ อง๐= ง ๕๐๕๕๑ ๑


    3
    0

    10
    179
    81
    03

    17
    09

    10
    ๖ง๑๐๕ ๒๐๐- ๑ ง 3
    10

    1ปี 7
    | 7|

    9|137
    3| 1
    29
    107
    23 9

    107
    |

    6I
    ๐๐๐ง ๐ % งอ งงอะ๐ ง๐๕๐ ๐๐ง =

    /0

    0
    IL

    17

    43
    /13
    0

    oll31
    ( ๑๐ถึง๕๐ ๐๐๐ งงงง -

    1| 07 0 ]0|127
    1
    I

    4
    0/
    3 |

    1u
    13
    17

    127
    10

    III
    0008I - 0006

    |765432110641579208363

    20
    19
    |918952 |109630407

    9
    34|37 |65814
    47 11 1| 27 0 111
    149 137
    71
    01 0 171 10
    55
    )|109 139 이기 11 1|131

    i
    1| 3 97
    17 | 1901
    0 1 13 15111
    1311 171 0'
    11 143 17 101
    71 151
    13 591 |19
    17 0 ]13 61 0 17
    )
    01 19 107
    131 ,|18909 ]
    31 111 69
    111
    103 131
    20 111 .
    13
    ]
    113

    | 78430 38708 370


    61
    59 |291 7 191 이
    17 0 111
    31 0 113 )
    109

    .
    m0750 80800 3807 0!

    111 11 1. 9 ] 7
    13

    .
    17 101 31 19 이 11 1101

    -
    711 231|
    0 |110 70 이1이 10 010
    101 |
    103 107 71 311
    이671 117
    .13 |이691 29 이 131 | 01
    0 |1711 |3이6151113 113]이111
    31|1231 |]1 111
    |
    17 )109 기157 111
    /19|102이기01033 17 59101119
    13 |11
    Knong 70307 49788 07mom 080700
    00857 00073 gogos 70407 on0300
    70 00wsm 70gap mgoog 007057

    |987063 108|41 |22063 41|54047 |55396 |655298


    23 이1기7 291
    177
    9710 13 22 123717이|3
    7 37 23 1| 10
    0 |203177911
    111 11

    .
    01 101 |109 13 |157
    1391 131
    기 ] 3
    11 ]
    13 .
    01
    .
    23 111 171
    63 371

    -
    |131|3131 기 109]

    00700 mooam nooo7 90780 onomage ognrm omgoo Tsa

    Orgap 33070 800


    |No70m 8798 BAE |
    Scam monam 070 |
    19| 이 430 1 | | | 17|| 이 | || 53| | | 19| 23| | 37| 113
    1011 1| 71)0기391 110
    6 67|0 11이 0 107
    이 [ 79|
    13 0 101
    6) 이31
    163
    1091
    19
    13 이
    0기31 10
    11|103 |이191
    477

    07m omgom7 D
    1911이 1 591
    이17 ]13 이이 1이9
    기| 이이 이3831 1277
    기1이 1
    19113|11
    17기 |기'이210
    3 1 기201 3

    017 00A200 mmo


    73monlign 007
    기1이이3
    이기이1|53
    )기이137

    omo Mana08 7090


    ]이|43

    700 170309 8079


    11 7기 1 | 19|| 이130]0109
    |23
    11371)1 | 19 이1 17일 이
    ]1|11 109 이기67
    31
    이|191 37

    ๐๖ง ๐ ๐ งงง* = งง 3
    00070
    ono 0007mm 08a00
    985|818 01 |21060852419 |543301568293 |6576309
    )
    109 0 43 19 37
    ]
    1310 12 1371
    113 0 이
    이 |171 3 43 19 13
    13 .
    43 11

    -
    11 19 | 이17
    139 1| 03 13
    SI 71 |111 | |
    | 13
    01 | 3 | 1 1 01 |
    113
    19 0 |177
    .19 13
    |01

    -
    47

    07351 7893. 8
    711 이111 1| 31 UI
    7 11 |107 기 11 671 |11091 90
    |191 131 191 11
    0

    7| 1 11 11 71 14
    | 11 |127 11 01
    | 17 139

    C.
    11 19 59
    11 0 191 14
    0| 1 9
    111 | 0 |0141 9
    71 23 11
    133이7 89]73|11|491
    이17 13 53
    ]
    101 0 131 이471
    |
    1391 )13 83 23

    0
    10.3||67기 17190
    47 13 이 113
    ] 19 이 11 1
    ) 9 0
    137
    |
    43 ]
    113 이| 이17 149 4| 17 19|1077 /0
    ]
    31 17 611
    1| 391 1-이이07 01 171 111 1| 3
    |107 이11
    0183 23|1191
    13|1491
    11
    |] 13
    0 7000 70060 09700 oomom oaon

    17
    791 13|10319|이|17
    18000 | 27000

    19|이61149 11 4
    1131
    77
    - Bao Buo700 58075| 7 _070m 89m07 0037googog 798moo
    - ons 003870 0700 | SLOTTAC 307mm goo07 08TB Png ?

    |Roomn 07003 70moo 03708 17300T


    cer coogenusacon- glaosoo ErgoToggo ga708 870mco

    %0mg RinRO
    9838791425 |106Do9632 |330639
    13 23 0 19 4 11 4 6ป 13
    13 โ| ป
    10 09 0 290167831|43740]317 ๆ 37 797 13
    3) 013 23 |163
    13
    11 |103
    )53 43 29 17 /0 71 71
    ]0 01 13 19 05
    | | 2 ก 0โg |113 l
    13

    0
    10 0ป
    61 89
    23139
    |015 |127
    ol47 |109
    13 11 I บ
    015 |1ปี 03 13 59
    |19 0 |03g
    19
    )0 10 31
    19 130 0 151 0 l 19 3
    ก ปี
    15 บ 23 โ57 123 9 )03
    17 15ปี
    0)ill 179
    แl
    ใย| 5 |
    13 l 41 53 109 lll
    109 0 3 59

    ๐ ๐๐๐๐ง ง๑๖ง •ง 6๑๕


    139 13 17 15 20 17
    13 Ill โ9 13

    0
    lll
    10 13 53 181
    0 12
    10 1) 1 )
    13 0 0 97 70 โ9 |11527
    3
    139 0101 11 0 ol
    of151 )2
    103 )6 13 ll 3 l 139
    0 99
    03 7| 13 23
    이 이 11| 7|13 |139 |19 | 1670 0
    ง . ถึง ๑
    ๑๐๕= ง ง ง งงะ.. งงง๑๐๐ %

    งงง . =๑๐๐๐๐๐ถึง ๑๐๐๐ ตั้ง = = =


    |98958302 อ80็47ม 9.\10
    71 37 11|0| 29|113
    7|23 ป )
    131
    131 |19 13
    บ ]
    103 ก/
    0 โป 13 137
    23 17
    |127
    0ป
    |โ119
    0
    ป
    13 0| |13 |โ5 0 31
    19 ย 7 37
    บ | 1) 3 |
    ไ09 i6
    19 0131

    0
    |61
    - ง ๕๐- 4 อง)

    = ๐ ๐ อง- ๐ ๕ อัง
    ง°= ๘ ๐ ๐ ซึ่งมีอง
    on
    รข o|
    103
    19 00
    13
    |83 0)|7 10 |o 13
    27

    ง๐ ๐ง • งดง
    이 이 18 031

    o|
    13| 101 23 o ป
    0 0 0 3
    197 58 073
    l07
    19 /9 09 1 17 9 01 163
    07 1
    0 1 0 |9010 |

    ๐ ผง ๐ ๐ ๕๐ผง = ๐ อง
    )07 1| 67 1|0 339

    *ผงะงๆ งอะ๐ งง
    ๑๕ = ๘ = =๑๖ง ๑ะ. .
    ๑๕๒๖ง• •ง งดี • ะ .
    1\ 8 |984870 220| 63 |32952 |550|4938174

    | | . -งององ๕๕
    1 31 67
    10
    ป 1o 3 101
    I 11 23) .6
    191 /0 7
    9

    157 13 9l 0|ol 47|11| 8g 7 ]


    59 |127
    109
    97 19 3 391 0

    ๐
    ป01931
    3 0
    93 /0 0167 |
    14g
    0 |
    139 0|11973 )
    53

    ๕๑งเอง
    A

    9 11 11
    13 ht
    29 0 ll 011749 AA 0 83
    4llo3 93 127
    |31
    17
    li|
    07 27 103
    149 II |103 4l 0
    /0 149 13 10113 u| จ
    /
    49 83 11 โ7 11

    ๒๕งนี้๑ อง๐๐
    49 163 19
    0 3 73 9
    7
    0

    ๕ - งงะ
    ม
    13 I 43) 107
    17 107 | | 03
    127
    107 l
    |
    li 8
    1

    167
    47]79 .77 3161
    89|23 0 0 831 2)679

    ง. ๐๐- ๑๐๐

    * งง ๐ ๐ = * ง ง อ เ ๐ ๐๐๐ง งดง
    1

    ) 0)
    11|230 163 จl |10 731 41|37
    27000 - 86000

    งงงง
    ๖ง๑๐๐๐๑๐ ผงรั้ง ๑๐ ง ๐ = มง
    ง๑๕-๒๑ จึงต๐ ง๐ ๕๕๑๕๐๐ ๒๑๑๕ งอง• •งแeee
    MATHEMATICAL TABLES.

    mohol
    TABLE VI. - Functions of a and e.
    a = 3:1415926 3183099 e = 2:71828183
    7 = 9 .8696044 •1013212 ea = 738905611

    || || || ||
    73 = 31:0062761 •0322515 e -l = 0 .3678794
    a va = 1.7724539 2009 = 639-6619772
    at

    ..
    r logoa = 1:4971499 180° = = 570.2957795 logee = 0:43429448
    Cr
    log, h = 06679358 = 206264":8 log, 10 = 2:30258509

    TABLE VII. TABLE VIII,


    Square root. | Cube root. log10 N . log , N .

    cong
    cooprine

    aver
    1:4142136 1.2599210 •3010300 69314718
    1.7320508 1:4422496 •4771213 | 1:09861229
    voe

    2:0000000 1:5874011 .6989700 1.60943791


    2:2360680 1.7099759 .8430980 1.94591015
    2:4494897 1 .8171206 1:0413927 2 :39789527
    2:6457513 1.9129312 1:1139434 2:56494936
    2:8284271 | 2:0000000 1. 2304489 2. 83321334
    o

    3:0000000 2:0800837 1. 2787536 2:94443898


    3: 1622777 2: 1544347 1.3617278 3 : 13549422
    3 :3166248 2: 2239801 1:4623980 3:36729583
    12 3 4641016 2: 2894286 1:4913617 3:43398720
    3

    3.6055513 2:3513347 1.5682017 3:61091791


    3 :7416574 2:4101422 1.6127839 3 :71357207
    3.8729833 2:4662121 1.6334685 3.76120012
    4 :0000000 2:5198421 1.6720979 3 85014760
    4 :1231056 2:5712816 1.7242759 3 . 97029191
    4 . 2426407 | 2:6207414 1.7708520 4 :07753744
    4 :3588989 2 6684016 1.7853298 4 :11087386
    4 :4721360 2:7144177 1.8260748 4 : 20469262
    4 .5825757 2:7589243 1.8512583 4: 26267988
    4 .6904158 | 2: 8020393 1.8633229 4 : 29045944
    4 :7958315 2:8438670 1.8976271 4 :36944785
    4 :8989795 2 8841991 1.9190781 4 :41884061
    5 :0000000 2 .9240177 89 1.94939001 4 :48863637
    5 .0990195 2. 9624960 97 1.9867717 4 :57471098
    5 :1961524 3:0000000 101 2:0043214 4 :61512052
    5 .2915026 3:0365889 103 2:0128372 4 .63472899
    5 :3851649 3:0723168 107 2.0293838 4 :67282883
    30 | 5:4772256 | 3:1072325 109 2:0374265 4 :69134788

    Note. — The authorities for Table IV . are as follows: - Columns 2, 3, and


    4 (Mercury excepted ), Everett's experiments (Phil. Trans., 1867) ; g is here
    taken = 981.4 . The densities in these cases are those of the specimens
    employed . Cols. 5 and 7, Rankine. Col. 6 , Watt's Dict. of Chemistry .
    Col. 8, Dulong and Petit . Col. 10 , Wertheim . Col. 11, Matthiessen.
    Table V . is abridged from Loomis's Astronomy.
    The values in Table III. are Angström 's.
    BURCKHARDT' S FACTOR TABLES.
    FOR ALL NUMBERS FROM 1 TO 99000 .

    EXPLANATION . — The tables give the least divisor of every


    number from 1 up to 99000 : but numbers divisible by 2, 3,
    or 5 are not printed . All the digits of the number whose
    divisor is sought, excepting the units and tens, will be found
    in one of the three rows of larger figures. The two remaining
    digits will be found in the left-hand column. The least divisor
    will then be found in the column of the first named digits, and
    in the row of the units and tens.
    If the number be prime, a cipher is printed in the place of
    its least divisor.
    The numbers in the first left-hand column are not conse
    cutive. Those are omitted which have 2, 3, or 5 for a divisor.
    Since 22. 3 . 52 = 300 , it follows that this column of numbers
    will re-appear in the same order after each multiple of 300 is
    reached .
    MODE OF USING THE TABLES. — If the number whose prime
    factors are required is divisible by 2 or 5 , the fact is evident
    upon inspection , and the division must be effected . The
    quotient then becomes the number whose factors are required .
    If this number, being within the range of the tables, is yet
    not given , it is divisible by 3 . Dividing by 3 , we refer to the
    tablesagain for the new quotient and its least factor, and so on .
    EXAMPLES. - Required the prime factors of 310155 .
    Dividing by 5 , the quotient is 62031. This number is within the range
    of the tables. But it is not found printed. Therefore 3 is a divisor of it .
    Dividing by 3, the quotient is 20677. The table gives 23 for the least factor
    of 20677. Dividing by 23, the quotient is 899.
    The table gives 29 for the least factor of 899. Dividing by 29, the quo.
    tient is 31, a prime number. Therefore 310155 = 3 . 5 . 23 . 29 . 31.
    Again , required the divisors of 92881. The table gives 293 for the least
    divisor. Dividing by it, the quotient is 317. Referring to the tables for 317 ,
    a cipher is found in the place of the least divisor, and this signifies that 317
    is a prime number.
    Therefore 92881 = 293 x 317, the product of two primes.
    It may be remarked that, to have resolved 92881 into these factors with
    out the aid of the tables by the method of Art. 360 , would have involved
    fifty -nine fruitless trial divisions by prime numbers.
    ‫‪6|71[ 85931184787‬‬

    ‫‪l) ll‬‬ ‫‪13‬‬


    ‫‪168‬‬
    ‫‪8178‬‬

    ‫‪)1;299‬‬
    ‫‪5/_7' 9‬‬
    ‫‪|879852‬‬
    ‫‪00‬‬
    ‫ܚ ܝܘ ܘ ܘ‬ ‫ܐܢܗ ܩܢܘ‬
    ‫‪111‬‬
    ‫‪ -‬ܘ& ܘ ܙܗ ‪ -‬ܘ| ܝ‬ ‫‪e e eavu 84‬‬

    ‫ܠܗ‬

    ‫‪_(0‬‬
    ‫ܐ?‬
    ‫‪13‬‬
    ‫‪[19‬‬

    ‫|ܙܐ ‪_|7‬‬
    ‫‪64‬‬
    ‫‪|6d5442145‬‬

    ‫ܢܘ |‬ ‫‪5‬‬ ‫ܘܽܘܗ‬


    ‫ܚܗ ܨ ܘ ܘ ܘ ܘ ܘ ܗ ܢ ‪ , -‬ܘ ܀ ܗ ܗ‬ ‫ܘܢܗ‬
    ‫ܩ ܩ ܝ ܫܘ ܘ ‪-‬‬ ‫ܘ © ܘ ܚܗ ‪ -‬ܚ ܘ ܘܘ‬
    ‫ܘܰܟܗ ܢ‬ ‫ܐ‬
    ‫ܗ ܗ ‪ -‬ܘ ܟ ܘ ‪£‬ܢܘܚ ܐܢܘ ܗܢ‬
    ‫‪4o7183‬‬

    ‫‪58‬‬
    ‫‪Bv‬‬ ‫ܘ ܘ ܘ ‪-‬ܘܘܗ‬ ‫ܘ ܝ ܘܦ ܚܚ‬
    ‫ܝܗܒ ܚ‬ ‫܂ ܘ ܘ ܚ ܩ ܕܘ ܗ ܝ‬
    ‫ܘܢܗ ܘ ܝܘܝܕܘ‬ ‫‪3‬‬
    ‫ܗ ܗ ܚ ܗ ܢܢ ܩ ܶܚ ܙ ܘ ܚ ܘ ܘ ܘ ܘ ܗ ܝ ܗ ܗ ܗ ܢ ܘ ‪-‬‬ ‫]‬

    ‫ܘ ‪ -‬ܗ ܘ ܗ ܟ ܗ ܗܢܕ ܘ ܚ | ܩ‬
    ‫‪77‬ܐ‬
    ‫‪191‬‬
    ‫‪411‬‬

    ‫‪111‬‬
    ‫‪01‬‬

    ‫‪in‬‬

    ‫‪0‬‬
    ‫‪11‬‬

    ‫‪0‬ܙ‬
    ‫ܠܠܐ‬

    ‫;‬
    ‫‪0/‬‬

    ‫‪0‬‬

    ‫‪|4.3317304‬‬

    ‫ܘܘ ܗ‬
    ‫ܢ ܚܗ ܕ ܚܩ ܩ ܝ ‪ ; - .‬ܝ ܘ ܘ ܨ ܝܘ ܘ ܗ ܘ ܘ ܢ‬ ‫‪G‬‬
    ‫‪191‬ܐ‬
    ‫‪:‬‬

    ‫‪111‬‬

    ‫‪201‬‬
    ‫‪01‬‬

    ‫‪0‬‬
    ‫‪0‬ܠ‬

    ‫‪0‬ܠ‬

    ‫‪3‬‬
    ‫‪0/‬‬
    ‫[‪0‬‬
    ‫‪111‬ܐ‬

    ‫‪)_0 1‬‬
    ‫‪11‬‬

    ‫‪s‬‬ ‫ܝܘ ܗ ܘ ܝ |‬ ‫ܘܕܘ‬


    ‫ܕܘ‬
    ‫ܚܘ‬ ‫ܕܗ܀ ܝܝ‬ ‫ܖܢܘܘ ܗ‬
    ‫ܗ‬
    ‫‪이 기.‬‬
    ‫‪110‬‬

    ‫‪331‬‬
    ‫‪11‬ܙ‬
    ‫‪71‬‬

    ‫‪01‬‬
    ‫‪11‬‬

    ‫‪13‬‬

    ‫‪11‬‬
    ‫‪/0‬‬
    ‫‪/3‬‬

    ‫܂‬

    ‫ܢܕܘܘ ܘ‬
    ‫ܗ ܘܘ ܘ ܟ‬ ‫ܩܝܘ ܢ ܘ‬
    ‫ܘܘ ܘܘ ܘܘ ܩܝ ܙܚܗ‬
    ‫‪00103‬‬

    ‫܂‬
    ‫‪106‬‬

    ‫ܘܢܕ ܘ ܟ ܘ ܘ ܘ ܘ ܝ ܘ ܘ ܝ ܘ ܘ ܘ ܘ ܙ | ܚ‬
    ‫܂‬ ‫ܘܘܘܚ‬ ‫ܛܐܘܢ ܚ ܝ ܗ |‬
    ‫ܕ ܗ ܩ ܘ ܘ ܚܘ ܘ ܗ‬
    boSsoㅇ ㅇㅇㅇㅇNood ㅇㅇㅇㅇㅇ Gsocs ㅇㅇㅇㅇㅇㅇㅇㅇㅇ ㅇㅇㅇ
    ㅇㅇboy ㅇㅇㅇㅇ % osdoo sassoovㅇㅇ osood oEscco 。 。
    Evooo od = oo oc & sE ONㅇㅇㅇ ㅇㅇㅇ Goosot Bocaso oos |
    EoBoNo Posoo doo , S - ovo Booo ㅇㅇㅇㅇ & gooo Nㅇㅇ
    ㅇㅇㅇㅇㅇㅇ vodso ㅇㅇㅇ ㅇㅇㅇㅇㅇ vaSol
    soood ovoNoSoo Sosoo58 -
    | 17

    o Neooo dogEos odoos Gooso


    00g 337 =00 703000
    ㅇㅇㅇㅇㅇ ㅇㅇ RSsoooo coSdg Esooy
    | 13

    |2306
    0
    - 이
    |

    oTeess Bosoco05, ost oooo | 8


    13
    311
    이

    3
    19
    47|
    | 23|

    So to osoS Nooo& Eoso o so 070 ooo087 8 n0730


    11|
    53|
    이 13
    이 23|

    37|
    | 11|

    00 137300 7600mg|
    이
    이

    | 70370 0
    이 11

    000 0 70mm 007309


    | | 이 29| 이 53
    10

    이 17| 이

    o
    | ooooo ados o ㅇㅇㅇㅇ|
    59
    이 1이 이
    이 이 이
    이 이- 이

    ocSosobe& EPoo Bodyo Stool


    ]11
    01
    이
    |

    11
    |
    이 23|

    191 0 29|
    이 이 기
    7| 2 | 3 |

    | osboot ㅇㅇㅇㅇ - No God s ㅇㅇㅇ |


    | 35 | 38 41 | 44 | 47 | 50 | 53 | 56 | 59

    000 ono070 070


    1/ 62

    mom aora03 784n00


    이 17|
    | 29|
    이 43|

    | | 19| 61|
    | 67| 13|

    |
    0

    67
    Loogno moorm 570mm Omano an0709

    11
    13
    101

    - 371
    11
    oooo Nood & so
    lol

    | | 11
    0

    ol
    9868|
    177 0 59
    | 1 - 9000
    |9699093 68|76565914447505384326292170508111315204
    74177
    17 13 이1| 1 3|2기4413이| 3717 |591
    01 11 53기 이
    10 110 3|0 1 이이 29이11이7791
    기이 13
    ] 이
    |013 591
    이11 37 73 67 1 기.110이 |이
    109 1 231 89
    7기 17기
    61 )
    109 기
    2 1 23
    11 0 83 2
    23 83 이
    35 1기 01
    III 19
    , 10
    131 1 1
    10
    ]
    01107
    19 103
    411 371
    이 1 ]|
    101 이 17| 13| 73
    0) 131
    311

    89 01 41 .
    19

    11
    47 31 7 37 17
    143 111 1| 7
    11 13 01 3 107 191 111

    700wo oo78m 07080 03070 900am


    72300 3870m 87aap mog7o omogo

    03708 oooom mono7 3079m oomono


    aroo Monoo 08070 0goon 700870

    1087 00970 0030 7mmon 08nmoo


    4 _ 00073 oomoo 7000s goman 107047
    91|94
    197 |432115|629037418 48

    19
    177
    |1) 13
    11 13
    891
    3

    13
    13
    7

    10

    73
    1기 이 이 691 111 00

    Ago momoo 070


    Homon 07007 33803 oongo 07030 | 403780 170mm moo

    Mona OnRn0 07이

    007 margo o3o


    8700 97008 00㎞
    |0700m omgmo 03070 0n809 711070 | 13087070387 100 |
    I_ Boa gsgyo os =| 8 ㅇㅇㅇㅇNoPood modos yoksooERROTE
    7 0 1 0 |40|69111 of109 0 11 0
    | 03 7| 19 10 7 |47 17| 3
    | 11
    | |nๆ17 6|13 00 | |7 83 |1
    19 53
    |
    1l

    9
    31/ " โg
    4ะป 1ปี 7 9
    บ3 20

    18

    11
    | 3713
    50
    23 0 89|
    041 1] 07 |110693 5|43252761593784260

    16 | ซึ่ง๐๕๐ ๐๐ง ซึ่ง รั้ง


    201 13 11 20
    |0431|401337 ll 0
    29 18 เl
    |017 103
    107
    17 70
    /50
    171 lll
    ด 0
    109

    2
    103 0
    93
    T ll 0 | |
    13 6 0 0 | 0| |29
    19
    43 ป 19 127
    ol 0 13 0
    109

    ง ๕๐๐ ง *๑๐๐งมอง | งse 6 6 6 งง 6 % =


    11 107 17| 0 1
    in
    Ig ก 01 0 43 บ
    3 0 III 10 41

    ง ๑๐ ๐ ง = ๐ ๐ ๐ ๐ ง & : 8 ง . = ๐
    0131 0 III
    ll 0 97
    01 l |10
    13
    53
    (0) 13 0|
    1 | ht 0 II ]7|0 3 13 Ill03
    ๑๖ง อองซื้อ = & ง ระง•ผง๑๐ % งง๑๐๕ ๑๕๐- ๑ ๒๑

    13 103 : 01/ 13
    ]
    0008I - 0008

    19 0| o] 107 |

    ๐อง * * * * * * ๐๐ง งงง ๑๐๐% 8


    ๐๐ ๐ = =
    •งง 6 6 : ง ง ๐ ๐งง = ๐ ๐ ๐๐ง ๐๕ งง 6 | =๐๐๔ง ยังง๐๐ ๐๐๑

    ง Eะ. อง๐๐ & ง๑๐๐๐ ผงอง = ๐ ๑ 3 | = 8ง ซึ่ง อ . -

    ง๐ -
    0

    งองงอ๐ ๐ ๔ ๕ ๖ ง ๐ ๘ = ง ๐ ๐ ๐ . ๐ ๐ ๐
    0|98806218263190745 43|653127631850497
    47 11 127 TU
    ]|13
    01151 1 |13737 11
    1기
    111 153 01 T 10|
    |18913751 ]|131이 09 91
    |
    1091 )
    139 101
    | | 01 | 11 131
    411 10 17
    10 |131 137 197 20 23
    |131 1| 11 4|151 .
    19
    111 21 9
    10 이 17 10 13 0 2. 01103 151

    s
    | 01 1311 111| 7 1 611 111 CS

    -
    0
    11 CI 01 17 101
    71 |
    29 1| 0 )
    59 2

    -
    131 .
    01 311 19 17 1013 37
    101 61 11 0
    3. 0
    41 101 83 11 0 .2 )
    19 23]
    171 ]
    13 D
    | 19
    1091 37 3 1411 17 110
    53] 23 111 201 101 411
    ]|103 111 131 53 14
    11 19
    29 |191 14 131 1391
    10
    1 01113 23
    108 |111 기 97 190
    10 1310 11 157
    131
    |131 기113 37기
    131 169 IU 13 4| 77 0 -
    10
    149 89
    471 1411 .
    13 110 22 13
    10 13 101 113 1 111
    110 10 471 13 | 0 17 0

    ioi
    171 01. ]
    31 0 131 0 .
    17 1 41
    10
    }

    111 711 기 500 | 11


    13 6 | 0
    111| 0 79 12 0 10 이113
    Li
    101
    13
    10 3
    14|기 111 1|이103 107
    111 0 1371
    671 11 17 0
    59
    .0 .29 111 0)
    31711 131 11
    1011
    177|91613 3
    7이1이 23
    109
    47 13 101 111
    | |
    171
    111이| 01 31 1이 91 11
    이1123
    |1기 103 171 131 )5111
    |9 191
    80007 05378 0ango 70307000mm|

    9870|66385 3|25829 5|470 65|68

    : 70m SanAm 00007 007mm eng030 | 3 |


    71 1
    693|97|14813 71
    63 191 29| | 73|| 13|17| 11)|
    0

    43
    000

    53 139

    | 07040
    1131977 11 131 기
    ]13 231139 13 ]
    11 10
    79 291
    23
    |13
    기 ' | 37 13
    10 591 1| 10 3
    01 | 111 |23 101 10 10 |
    103141 | 0 21 1' 3
    7 11 17 3| 11 101 11 7
    08 11 131
    113 )
    109
    이 - 70830 8800m oon
    Oscan Boom070
    | 887880 00378 000
    2LE795 0nogo goo70 890mm7517 | 8| 0000m 70987 800

    0 mmonomeo
    |
    | 이 191 이61
    기]14|903이 1기 7 |이1531 67이기 23 |01이373
    191 이139 이
    기이2091319
    71
    ]1|3743 47
    61|이4]131 이| 이 이5기 91 |79
    1391 13이 01171 이.3|15311301 이7
    이101 0
    |101
    1311 1기 30 1
    71 | 67 163
    011091
    09 131
    191
    |0
    1
    |11
    11| |14
    111 0 11271 111
    S| 3
    01 19

    .
    0 13
    109 이 11|의 137|의 이 109)
    1 111 0 8m0700 68078

    no ooo7nm osa00
    |6543210603041789256341
    109 01 42 | 171
    131 83 113
    371 0 14013 0 ON
    107 13| 127
    1101 43 13
    -

    53 | 0 6| 1

    D
    17 53기 13이
    101)139 1371
    01
    103 11
    1| 3 1 11 기
    63 11 |
    113
    0) 177
    13 )
    19 13
    11
    70

    이1기이13기 이 107 11 1) 9 1091


    111
    19 13 11 0 191
    이기1091 |01
    1271 171110
    191
    1이677 기 |171
    19 67 411
    3이 1 0 |
    73 131 7 01
    ]
    23 1| 3 0 29 )
    101149
    1| 1 89 13
    1491
    | 01 1
    7 ]
    103
    13]이 677
    |149 |
    191
    1377
    191 111
    113 2107 1071
    1391 61
    177 143
    101 19
    11 67|43 1
    |0 171

    0 706030 8000mm| 3| 79070 90mms now88 75080 307


    83 | |17 |
    111

    con goairm 178080- 8 | Escam morno onger Bourg 0gan


    01
    99|21|93911 13]|101 이 19 1491
    79 기| 이149
    | 18000 - 27000

    892 A3078 000070- 5 03700 on05 7 89870 0oon0 7300mg .3|131043 11


    |1171
    131
    | 07000 70050 09700 oomom 040370
    070m man07 0037 0030g 798300

    700 17000g 807970- 8 |ommoo BB870 070mm o0908 084705

    _ 08700 00007 00870 50000 7030am


    soonn 07003 70moo 03705 17moon

    07080 7mono 40789 009am 001875

    oo 087 0 0mmon agara Room07


    8|825 109|2 3|5430715964823
    ]
    13 23 17기 이 이 11 |
    83
    61 37 13]431011 531 611 )
    19
    113 IN 191 1127 171 0 0 13
    ON 311 4313] 기 1| 1 0 2|이 3 163 111
    11 ]53 11 |191 17 0) 01
    3 17
    ) 01
    19 67 | | 23
    611 189 1| 11 19 ]
    01113 ]11 11 ) 9
    13
    1511 127
    이471 |1577 |
    109 /7 0| 1
    01
    151 1기 03 110 50 |
    139
    0이 1 191
    19
    ) 이 )10 이131 ]13
    11 01 113

    633 9738
    |1311 1511 11
    0 131 191
    1290 103

    0
    1 1511 13 0 173
    1371 |이113 |31
    171 이111
    1511 191 |111 119 0
    0moo7 09370 30,
    111 103 1511 1
    711
    /
    13 0
    111 411
    0 581 1
    NO

    )109 13 기171 )13


    -

    1 139 18 11 151 1811


    117 171
    1731 107 |01
    111 .
    19 CR 131 131
    111
    101 기 ]기|131 131
    111
    131 10
    ! 17 161
    11 11 29
    11
    131 711 191 0 13/ 127
    139 1011 17 611 131 11
    00

    ]61
    103 111 110 591 10139|이151 | 1 13

    0 79amp morn 0700m Som*


    117 111

    BOT 0307
    1 0 이 923
    ) 13
    이 이0 기 139 191 10 이 이 131
    8109070 mmoom 70000 orage 09730

    SC

    14 883 85
    01

    00073 Escal Fana79

    oo gumgo Omoo7 0g700 Noamoo


    00 7gooo Cacom 057831
    m35 , 750008
    00m7w oon0 7800m 80gle 337307

    80Iml 192
    83|896 |01069291745 |110963 3|2851 5|492
    1
    | 1 7
    1|2911 12|913 15 01
    |163 23
    1기 0 71
    103 19 41 | 0 63 131 171

    0
    1 113 1
    19 |23 1이1기 0)
    )0 591
    11|311기
    127 137기 1| 11 13
    ]
    13
    1 이
    1671
    |167 | 311 53
    13 01 71 ]13
    2001 이 | 591 1811
    43
    177
    |이 이 이 167
    기 13|| 531| 이 131
    50700 07580 00
    19| 103| 기13이 || |1131| 831
    ]
    13
    PP030 07

    311091 53110 6이 1 기43 71|

    30073 0 ooon 73 이
    5 37|이 13)| 29|이 17)
    300m moooo 070

    ano 08370 105


    1703700 50% |

    / / :1| T T .
    007n0 070nm 031

    97on78008E }
    = = งง ๐ ๐ = ง ง้อ ๕ ๐ ๐ ๐๐ง งงะอง
    |
    79|
    23| )0
    147|67

    ๑๐๒๒๑๖ง 8 = ง ๑๐๐ง งง๑๑๑ * งดี = = ตั้งแ: ง๑๐.


    A

    ll

    ร
    19

    " 0
    157

    ๒๖งผe & งง๑๐๐ องๆ


    97
    ทน

    |3

    |
    l2
    0|
    0

    81
    157 18

    107
    14g
    All
    103
    23 0

    ๖ง๑๐๐ % ะ ๕๐ะ * * ง ๐๐๕๐ง 6 อง๐๕| งง 6 อง8มีอะ


    |600]197
    13

    5) 3

    0 19
    690
    0
    2971]0
    13

    •ง งดี 6เ• ง๑๐ ๕ = 5ง งด ง = ๐ ๐ ๐๐๐ถึง = 4


    ll

    113
    10

    1|0 1/099

    / 07
    0
    /

    01707

    23
    13
    17

    13

    0
    19
    127

    149

    0
    0

    S9
    1
    |247839

    โม 28

    |10 731
    13

    15
    13

    3 I

    ผงงอ ง 8 = องง
    0
    in
    |13
    137
    109

    11
    0131

    530
    59l
    1ป 63
    |11
    |17 379
    108

    ๕๐๐ถึง ๒๕๐๐ ง ๑๐๐ องงงง ๑๕ = = ๐๐ง๑๑


    3
    โ97
    0|13
    10| 07

    /
    |

    1
    19

    83
    io

    0
    ง %26 ๑ ๐ ๐ ๐ ๐๐ ๐ ง๐ ง = ๐ ๐ ๐ ๐ ๐ | ๑๐สิ่ง 8 * ง ๐ ๐๐ง
    |217360

    งผ๑๐ะ งงงะ ง = * * * งงอ ๑ - ๑ - ๒ อง๐ ๐งมีผe see


    167
    4

    อง๐๐๐๐๐๐ง แผงรั้ง ๑๐ลง ๑ะองe| | ]


    /
    9

    16

    4l
    0l
    )89

    43
    59

    107
    139
    149
    73|139

    องง๑๐๕ ๑๑๒ งะองคะ . .| F = ง . ๒๕๑๖ง มอง


    งะ 6 6se & งee ๐ ๐ง งองนี้ * * * |431852
    |0
    0

    9|

    0 43
    0

    l
    0
    7| 3

    0l
    .

    81
    119

    53
    13

    13
    19
    0 01

    179
    149
    45| 76930
    1|7 49
    27000 - 88000
    |7666592 |98730 |432141897152630
    103 11 43
    19
    | 이 기 |101 1
    || 13
    137 411
    193 73| 1 0 179 107 139 13
    71 791
    107 171 53

    -
    107 13
    )
    131
    1791 이
    1677
    131 7 41 11 |
    31 13 11 197
    181
    1091
    1571
    |17기
    ,101 1| 97
    13 071 19|이13 이 |13711기 17 7 0
    ]11 171 111 1391
    137 0 )17| 이 | 0
    1571 111
    171 23
    ]
    39
    10311 |
    311
    기 1011
    171 01157
    013

    00
    |13
    --

    01 371 0 개 13 101
    163
    111
    191 |14367691 109
    71 591 |기1|163311 29
    131 83
    811 73
    31 1071
    103 139 181 11 923

    9
    01 ]13 13 173 17
    0 11 113 0| 1 1 131 191 29 53 89
    291 ]
    113
    131 11 | 7
    1 179 1| 091
    109 13
    111
    .109 10 |73 )이113 01
    43 0) 53 41
    1511 01 311
    371 11 191 0
    |
    151
    111 | 61
    1811 13 103 171 23 13
    .
    0

    17 이 이 10 13 193
    0

    0370g 0000 moon7 3370 . 8 a410


    8 , 708mm 60700 87mon on870 Mong0o

    85700 3mmon 00807 8073m ooooma


    107035 0000g ana780808 70mo78

    |76466307 |985818 |322163740 39|42


    13 61|31
    1기 1| 911
    3 11 |109
    이7 89|191이|191 기171이13이97기 |이13 |121|1991 5911143
    12 1577 23
    23 01
    0 111 171 1| 63
    |1511
    11
    73| 191 611 |
    151 17
    13 23
    70m 070

    |
    171 411 11 1337
    |
    1511
    01 197 127
    .
    01 13 1791 511
    3
    |791이3931 ]1
    이 1917기 53|
    87000 8507011

    891
    61|| 11|
    5 |이 17|8 ]|
    10 0 이 0
    11| 23,이 13이
    3 . 00a7a300

    5| 00070 13030 7moom 30000 807007| 8 | 950mg omano 670


    | 171 107) 이 15 167| 이 그 13|
    70 moooo 7mon
    3
    2| 1871 1|이210103
    ]957 이7 10 ]|기0113 기 이기
    - 이 이 13 031 11|127
    183
    411 1| 31
    311 61 10 111 이기 이671911
    113이 31이17기 130177
    111 109
    .

    Rin
    131
    131
    |
    101 10
    | 3 37 181 149 0 17
    1| 9
    131 13 181 19 1131
    10 |11731
    ] 011
    163
    1411 31
    971191 101 1 111
    13
    6793|11111
    10
    11 ] 149 11 0 10 171
    31 191 71 1
    1171 0 10| 18
    01 731 13 0 LL
    ]11 1271 11 16
    10
    )2 |5311039

    10
    10
    76|16284175 |9868529 0| 4
    01 |3210910631275428

    -
    ] 13
    173 이기 53 111기|17 13
    ] 111
    11 III ]
    109 69
    131
    10 116 이기 111
    83101191이이|
    73
    10 이23 83)|13 기|11|31
    IL 193 23
    193 ]
    173
    201 1271
    137 13| 139
    .
    13
    113
    이23 7123|11 기73|00 1이 91
    |107

    -
    C
    이이 이기 13
    . 11 이19|1577 101
    131 10
    10 111 4| 2
    11 기 이 19 |1911 101
    471
    10
    131 |
    109 2|3
    1기 71311
    10
    131이 기
    11기 이17 11
    ]11 17 231
    11 17
    /43
    10 10 11 67 19
    137 0 19
    0

    0
    .
    131 139 73기 1 10] 111 이기
    ]
    101 이47 기이 61 |1323 이 13
    1391
    |189 1 1791 7
    이
    107
    173
    이 111 591 .기 ]
    43
    17기기 93
    07 0
    163 100

    100
    157 137
    13 )
    19 |1 163 19 11
    1311 411 13 17 13 | 0
    0

    101 101
    기127 111 157
    0

    13|이 1111
    /13 23 103
    71
    gon ogasmB7R35 ooogo 19789 0Ros7 moong ogan
    807 Onomom 8sosn| 9| twoo 0700m 7mmg0700 W7g에

    0F7898 |10708 nno9708073 500go ream


    311 19| 이 기1이513791 61 0 11| 73| 0 29
    111

    Soo 700m2g gooo08| 9| 7more into gogog mong moram


    36000 - 45000 )

    70 008227 090720- Seg07 60000 0mons Arger As mang


    | 00" omgora 170358| | =good agram TES07Cam70 000000
    |716657418 80 83|86 |328210131619:160918994152753
    )|13891971 51 9 71이 3 111
    171 83
    29|이591 131 | 01
    31 이8|9 17 43
    |3171
    7 - 이
    이 이11171 4이 11 이 11|이17
    109 13 미이
    기
    113 1193
    | 113 100 이
    177
    1이11 이기
    181 이 이191 이기
    6111 18 31 111 1 |이 0
    0 이|11 127 67 12| 93
    11 61431 1 13 1 137 |1105301 311
    ]이103 ]103 13 0 11 0 이 131
    | 13 19| 1|이0|171 11 29|이 13
    19 591 43 |11 기
    171677 10291
    97 1이기 1 1311
    91

    -
    173 181
    3| 1
    113 111| 01 67 |107
    .

    131 1|01 1| 07 41
    1091 107 기이111 )
    113
    13 01311 77139
    131
    | 103 이 ) 23
    |17 13| 31|이61 41737 이 |1이4327 23
    13 기. ]879191 | | |23 | 이111 이 이 이기 71
    1| 011
    191 |이
    |이13 1191
    | 1 711.191 이
    11
    0 223
    0 21 101 1011 193
    01
    ~
    기111 01 1577
    | 127
    91
    |1131
    59
    091 4기 이167
    13

    231
    611 001 1| 31 이기 11|1기
    191 111
    131 21 137 13:11143|1911기
    0 |이23129|이3 31 11 19
    190 01 153이30
    |기101 11 71 1490 0
    4[]1이|31

    7030g moreo R7308 03870


    )
    17
    6605|1734 9|88106374 1|00518 2|2063
    | | ]13|103 291 49

    | |
    83]233 1577
    23 |291 139
    193 |이13137
    13)17기 1677
    | 31 |111 113
    ]97 491
    93
    840]|113311
    87m goggo regoo onese 187001 | 이

    2|407|317기 1 73
    3 2311113 |129기009131
    이기179 117
    |3 ] .
    13 53
    08800 17087 Agoog momoo A78080 | | 08

    .01 |11기 23291 311


    |79 |1기4339
    09 127 161 )167 131 11]|5 191
    3 10]기이 017571기 3
    29
    111
    17 111 ]이19|149 0 17
    | 271
    169
    |131 141 ]
    1131

    70085 1079 7803 05070 mgsoo80887 0


    13 311 0 131
    111 ol | 이
    111 0 이17
    이 0

    0307o goon
    1이91 1110| 이
    1491 이 이37.13103 199
    1] 3 171]| 391
    11

    9 nos7 0
    0 0 531기 기971 491
    |7기131 59
    71
    이19|17 177 |1167
    ]이37
    404 1491 |[°19 °| |13
    111 10 193
    이기31
    |791 11 ] 이47 71
    591 611
    171 371
    기21103

    0
    m97a030
    E moo7a man
    90 0 000g 007
    870 0mgom man
    이기37
    101
    13471
    591
    17 89| |이29 |1131
    091
    99 211이 23이 111 171 13
    1 29
    1이1163 1911111511
    130|이 기 기이 , 9
    10 1|17811 이기19]11이기
    1311 1) 0

    ]
    181 이72|916117371 167
    1| 991511
    이23 7
    13191
    )10 11

    32 0
    |
    311 1G 23|기1991]11 167
    0191
    23|8131 |
    13
    01
    151
    1117
    19
    / 0
    13
    171
    이1기27791
    101 08
    이791111
    100 1131 SY |131 17
    01
    . .
    109 171
    111 0 11

    이
    01 1911 1511 141
    191
    |
    53 191

    700 A79035 337


    413111
    01 591
    151

    ooss= gsa3 . 5 ce,


    13 17기이 13179119 7

    7m 800037on70
    157 |

    37 3000am 9800m
    1000
    |998765841896307527 1|2 |33369
    0]163 )0( 101 127 |231 311 |1111
    001091 13 |1611
    113 기1701 43 111
    131
    1031
    191
    1071
    17
    |011 17 0
    기 371
    139 0| 1 1| 3 |7 8| 3 이1971
    311 89
    |
    111 | ]7
    0|00 131 83| 11 691 11
    13
    191 173 111 | 29231
    01 13 117
    17
    13 기
    | 19
    ]
    47기
    1291
    ]
    19 59 1271 111 103 |
    )83 131 191 223 71 0 0 199
    17 111 1391 411 1311 |
    1811 1 11
    0 01 0 11 | | 1 41 111
    191 1/ 63 |1111 220 14
    | 11
    |113 0 11 197 0 111 131 이179 기 111113
    0 |기 0 53 311 13

    -
    79 23] | 13231 |171 12 0 0
    1011 1791 17 !229
    |311 191 | 111 113 1311
    71 11
    13 1011 131 11 123
    137 157 0 171
    127 171 1 137 31
    79 111 01 .
    11 19 197

    our0
    91 13 1 11

    0
    |이2이 3 131
    531 1) 9 01
    137 17
    111 13 109
    471 101
    | 17 67 ) 7
    19
    101 431
    371 11 173
    411
    0]
    111기 기 이1|0 37833
    127 191 53
    111 10 3
    107
    43 17
    119
    |163
    |111
    101
    0
    231
    45000 - 54000

    | 01
    moo moo79 070 | 1_ simomo78m Bono7 30370 003087
    an000783030814 300714

    0 0070809mor moong myo807


    Laos-S BREBoooyo 5 .Per doe
    070 05mm, 00070078|9500700 = STAR ON , BS090 |
    10
    |
    |658525415793 01578|169716529348 |27
    기이 이 이 177 111 127 0
    |1911
    227 91 23|113
    4
    | 3 |
    13 )19
    100
    |
    53
    79 | || 01 97
    이 01 ]
    113 1기 391
    301
    1731
    |7911117191
    13 |89 0 29
    1기
    37기 23 이 13]111
    103 137
    411 313|이591 기
    | 0 1 ) 이 43 23 1631
    | 111 227
    13
    193 이 131
    111 1577 |1391 13 99
    이 111 1311
    11 127 )09 11
    43 / 이193 1791' 이 211
    이
    17기 137기 17317 | 53199 0107 67 191 111 2. 17 01163
    01 ]
    163 011 103 1731 0 13 37 0 149
    111 7.
    137 0 111

    -
    101 |377 2| 3 101-31
    139 01 491
    기11471 0
    53 이179 )59 기111
    01|
    0 1기 271 이 이 13 기 59 01

    0 10
    01 이 101| |
    311 1 131 61
    191 33 ]131 01 0 11 19
    이7 01
    177
    7311 43113
    |1311 15기
    1기 111 0 - 0 11
    13 181
    1117기 119
    61 1 0 01149 |
    111 0| 1 41 1971 11 13
    | 1 0 111 731 )이179
    19 131 01 17 11
    1
    1671
    131 | | 1| 01 11 ]
    173 | | 000 130
    I 0 11 | 37
    13 기 |111 233
    17
    | 13 /
    )
    149 10
    13 2 1677 01 131
    | '0 1| 91 11
    |1171
    11 기이1091 기83|17기0 10 7 - ]197
    13
    17 |199
    11
    1671
    )711
    1| 271 이 311 | |01
    1371
    1311 0 |171
    59
    |431 이기13 이기13기 이| 이 11
    1311

    501087 83079 080337na87 gon88


    |5630 59|6887652480713 89|8083
    186
    3| 1
    |21925
    10
    291| 31 111 |7이1137 101 | )|25299
    |15 |53 110 00/ 103 IK
    62 411 191
    ]]이17377391 1199
    이]| 033 171 1010 31 181
    11 0 |19311 37 01 107
    171
    53 0 011091 71 101 11
    23
    9 %8700mg smarAnnng moo08| 06 | 0007

    13 01 111
    431
    171 |0101
    1 0
    1 03 13 191 ]103
    43
    113 17 131 1이 01 131111 10
    ]59 |111 171 13 88
    1
    131 37 |
    2231 |
    11 01 19
    13|10110 13
    기 19123 이 71 111
    127 이7| 191 | 1
    7기113
    1이
    67 830 |241 |153203|기 23이 기
    2] 2271
    |27
    070mm 68570 이

    |41| 191 기1기


    77

    0970 mmsoo 708 39040 0070 | loonoa 0gooo 370

    BBnorm conos naong gooon 0375 | 00 oo 00am 070


    700 3700m 080

    B _ 0387 SBA7008moo 78087so %C50| 779300 0soom 037


    83007 007ao soo

    Loos sosoyo
    173 |053 131 23
    1|209791 9| 77 4|이131 31 1이기 1 기7041 17
    이 이 19| 이 | 193) )139 73 1|89267933 117|이기 0 이 이 ]이79
    |61 |
    |1]이 931
    10 21 ]103 1971

    GT
    101 311 0 |
    13

    = 97 0 0
    이 19 11 01 131 131
    01 111 )
    109 0
    1000 29 101 111 .11
    19 171 17 73 |83
    0 0 ]
    13 1571 01
    2291 103 0
    ]211 233 1| 7
    283 163 13 101 |
    13
    0 181 ]
    01211 | 677

    0 070713 0교
    157 11 411 131 01
    371 11 SO 1031
    191 111
    163) 이 13
    |1548471 |87692418590376 1:|20857401
    11 131 |0 311 191 171 01
    이1| 7기67
    0 111 10 13 1227 139
    111 | 01 53 ]
    23 ]
    13
    기이 131 이181

    | 900
    |07910고 0 93 |
    13
    01151 13 1991
    10 |기29 이 |177 |113 89 93 111 |
    101
    11
    |157 이 기 177 1971 11 137 | 0
    0 111 151 이89123 37 31 이 13 이
    239
    )
    179
    151 |1197 7

    -
    139
    111 113
    |13
    53 163 07
    173 | 311기
    13
    0 131
    1기 07
    197
    111
    7기191 기
    17기 151 13| 11| 131 1 이31
    19 03 127 .
    19
    11 10 |10 31 1이|3 이 |1111 이
    111 229
    1511
    411
    0 111 23]이60 이1기 3기 이 10 3
    )
    19
    13
    11 223
    222 11 131 163 233
    ,
    100
    |1391
    )71
    113
    /13
    233
    0 기37 이 1291
    | 93
    이 431 1071
    471
    17
    기111 |113 1 이 )
    223 43
    |
    | 1 229
    13 1개
    111 23 11| 기239
    | 070 onomor man730| | $0087 3806 8 poon 97007 383
    |

    이 1978
    1) 7 211
    971 0 0 mor Emaily 000087| 8| Comms 87080 70mmcarms orge 0 0 17 | 71
    54000 - 63000

    131 19 220 7

    SBEYOR 057310| | ESPss We=70 07mspace SEONE


    Soodas soEye ogo =s ggago yodo=| 2013. = secsc & _
    | 700 870805 oo7mg 1 0 gon omors 87080 18988 on 7a8

    ooo omg700 85837 | | 0070m 8037 8572 30303 TISE


    2 . PRO CREE| | 0070m8870375 4000m 700mcg
    3|303 |4528 6|5607934 0|10217458
    53|191 0 |이13] 111 이31 01011
    971
    ]10
    10
    251 ]|149 11 3 13 1271
    기1511 1 이 2101
    | 37이7911 | |111
    23 |
    1511
    111 17
    |1577
    110 251
    ]13
    |01 0 /0 191
    1101071

    08
    23|기1511

    00
    1491
    |01 0 | 177 111
    )
    149
    411 기 이107 177 103
    01 23|]97
    01131
    111 기기
    103 291 17 13 19 3|181 )
    109
    ]241
    173
    111 191 |
    193 13
    |477 1131 77 13 2511
    1
    011 이
    227
    17 233 131
    231
    191 29 221
    41 0 110 113
    47 17 0K
    0

    )
    109 03

    .
    )
    19 )103 43
    1791 0 01 139 )13
    163 157 12 111 01
    11 193 0 | 01 11 이
    7 347

    | 17| 이 |1기
    .233 161 17
    103| | 13
    00

    |이17 11 13 13
    137
    이 7 131A1 1311 111 19 )|109 163
    11 17
    017
    1091 이
    157 01 기 1911 |41 11 |23

    5 _ fn . ss 80700 370am 0307m 0p


    13
    | 이 | 5이6||2이831939 0 이197
    13
    131
    311 기257
    10
    291 1111 이 |17 25 0
    111
    291
    111

    이SAY 9938A8833 788


    | 0 12
    07
    |1이 313 191
    157 4|233 os =go goon & Soyo& obobs toose| )
    89 191

    HORS 807nparena maomg

    07093 ooong omom 3moo0 700973


    | | 00am , 800700980g 73807 amoon0

    |4/3&46073 49 67 los
    11
    194
    9191
    |88518 03 15|18
    891 13] 0 01
    107 12개
    131
    191
    111
    163
    | 23| 이 11 71
    611
    183
    111
    23
    31
    13 423|이3 |17
    17

    1271 이
    111
    111 이 |2391
    21 |1091
    233
    47 이
    6 _ son7 0970

    29
    41|131 |19 477
    이2391

    o= Esseoyo
    doSoyo Boy E
    am7000 0mgn mors wesoo 79100| Emong Namer 50
    | Pa07a nomom 70.
    |81318 70182 007:9000 870am 00370 || 9| 1m80 00978 01.
    |131017]0 |130 0| 199 13
    0 il |67107 ๆ
    137 23 โg
    ) 1| 59 2|0 1 13 11
    11 |
    179 68
    11 10ป 7in 0 23
    13 61 โ29 01 0| จบ
    71 13 11
    23 โgn 127 20 157
    23 10
    81 59 โ99 3| 7|033
    ll 263
    )197
    19 l 101
    30 13
    )(39 )
    181 ll 27 |0

    0
    งง • งะ
    3 19 |43 53
    ]157 13

    = = ๑๗๐ อง
    0
    โ79 027 13 17 13 23|107 9
    111| 09
    29 123 ย 17 09 0 0/ 23 42
    3 10 01 223 l

    จ
    13 0| 103 |181 139 \1 2g|
    131 1l| 13 220 97 163 |
    โ9l ย
    기 이 17 53 161
    2
    5|48813074 74ย' 9|888592 1|10639
    ol 89 59 239 73 |167
    13 23 |117139 7

    0
    07 01 lll 7 0 83 10 บ 3|18
    163 24 227
    70
    | 1| 0 137 /13 0
    โ|โ09 41| 7 03 ร 53

    2
    1911 210
    | | lll
    ol 24 25i 19
    17
    |139 147 0. 27 63 3|2 7

    - ๑้อง | | . .งง - -
    0 0
    7|119]091 59 13 0บ |95
    37 1071971
    107
    A
    0/
    이 80 3 l
    29
    07

    ง . 468 ง ง ง ==อง ๑๔ง -


    0 11 โ: |
    233 )0 1i 80 127

    ง
    A
    13101
    |193 0

    0
    ]13 83| ol 0|17
    |1263
    0 0 ! 101
    2g 41 23|20113273991 o]
    19
    |103 18 6i| l 0| l9 115 23
    2

    /43 10l 0) 01 )
    53 ll
    23 40173 9
    โ7 อั9 71 13
    l ]0 171 113 0 0
    N

    417 |0 0ป 10 63 23 01 | 29
    137 0 |1139 8ม
    5 ป 13 0 |13 All 0 Y | | 171 103 |109
    23
    13
    |15 23
    บ
    7
    2i| 59
    0 0
    191 ป
    |109
    13
    |
    10l
    Ill 53
    13
    )29|107
    |
    | 21 ]0 |)โ5 179
    ] 39 13 /
    1

    93 0 โ|137g5 01 29 3 103 101 91 19


    49 Ill ol21
    13 53
    0 10 0|149โ93 13
    เg
    | 07 01 3|13 151
    1 0 149 0ITT น| 47])37
    16 8 l 0l179 0151 olin
    149 13
    40 13 1 22
    157
    17 3ป 167
    31
    7 0 10 0
    68000 - 72000

    โi 0 150
    ๐ถึงผ๑๐ ๐ = ง ๑๒ ง๐ ๐งะ ๑๐๐๕= =

    ะอง๑๐ งงะ. . ง.
    จ
    0
    654998|3272026493057185638
    131
    891 17 71 31 이47
    1기 111 2571
    170
    13]이 ]
    113
    1811
    431
    8.31 0)
    14|107
    0393
    0107 0 0 0
    191
    31 11 0 291 NI 181 131 기| .
    23
    ]|223
    11 0 13 1277

    .
    | 01 1991 기|191 23
    |12|36313377
    113

    311 191
    111 13
    이미 |123
    0 111 |213271 281
    1511 기1263 17이이기 1011 01
    131|이 03 ]5]|277933|131 19 71 290 611 01
    23 이 1371
    |
    151 기
    471 01 )10 111

    -
    13 43 17
    251 0 0 131 III 157 109 17 | 257) 263|
    1091 0 | 13
    100 151
    131 11 |31
    1791 121 10
    0 70

    in
    1511

    -
    0 233 이19 101 261 111
    91 이113 53 2 111 971 2| 511
    0
    177 2391 31이17기 13
    ]|891 기. 83 71
    1577
    1611
    1091 |471
    1 |1101
    811 151 0 131
    31
    193 ]131
    163
    1271
    10 91 기71 0 1131
    111
    1|01 |
    13 01 9 2. 3 |111
    )
    691
    131
    8| 3 377
    31|201111 17 이 13
    01 )179 )|1131 1
    103 1711기 81
    0 1 191
    47
    1577 이 이0 101 91
    1이123 111
    1077
    43 13 |0 010
    |110
    |13
    ]163 2| 9
    191 |1911
    23
    17 13717
    277
    가이131
    07ano mo07

    13 이23 이 13 ]10 71|)191


    97791
    011
    3 43
    09291
    11

    ALPn833 97107 6080g 000mm 075030


    _ roong 397 07000 Hyoro Onsmom

    |654420735418 |987841073 108139182|093965


    13|이기131 |83 0
    79ill
    ol기of
    1이
    191
    10
    263 177 13 2111
    199
    이|13 1991
    7|23 2301]10 9139

    i
    61 13|11 기0 411 )
    19

    )
    1) 9 13 137 13 기
    11 19이 |24769 |13 10
    157 기 이211
    139 이 5| 31 191
    1677 17기 11
    223

    이 37300 이
    |131
    11 199
    431 193 5. 9
    ]
    0이103 ]17 )
    199 | 11
    113 1|기 91
    27 223 11
    F1 131
    0 1|
    29|100257 0 13 131
    1071
    231
    531
    113 1이 1001
    01 13]기1기
    191 | 01 1971 1411
    43
    | 117 기
    131 이5기2|61130727191 111
    |1311
    ]49173
    67292이1기17기3101
    7 이 977 이43
    11119|이229
    20 )10 7117기 0101311 73기
    이기1|3770171
    ] 776331 이 이1기 기|20 9391
    7m Mgmom 700 |

    이 8msan O7207 aoogo an980 07: 00 | 100703Anna gog |


    55 9mo Moons ma070 momos nmo070| Snacoonner ono |
    기1 23|1기
    411 |7127기 |기1229113110 이 이
    |172131971
    10
    691 23
    63 기10
    23
    기17191 11|16117371
    67 1| 27기 이67 311 1기 1 177 33
    0121091
    31 1

    311
    231 01 111
    131 19
    171 1| 113
    111 |163
    ]31
    11 131

    e
    8.
    ooo
    531 1391
    79
    47 2711
    241 이191
    기1139031
    |191 190 831기

    onor
    |143403734

    A0% 007090 70789 |


    1435937482601
    |8765440 |oo191
    |
    103 41|이477 1기 67
    611
    이1577 )
    10 7| 1 01 131
    2711
    기 1311
    11
    |이129|1311
    3163| 173 1091 10| 기 ]13기 11
    7711
    239 01
    11기|113 이 I |
    43 이기 29
    91
    이07531
    1 103 기이 50 61 가|]13
    |163
    To 10
    이 67 |
    4329|이
    10
    269 5.31
    1491
    111
    1710 2391 0 277 기1491
    1
    1731 291
    )13 14 137 131
    10 191 111 이
    INT

    기 0
    197 1931
    26311| 3 0
    |
    241 89
    )
    12711
    1911 131 17 3
    10
    0078
    | 기
    LIN 181
    71 231 101 17
    71
    229
    1311 101 |13
    71 | 00 1 |111103 179
    131 0 111 1731 19 11
    0 29
    101
    0111 1
    | 07 기1이6105|1913313 10
    ]
    163
    231 0 11 011
    91
    23|기이18191
    157
    ]
    113
    131 1361111967783기
    127
    732791
    기이|1291
    map 100700 sa007 || Param gmo070837 06380 705

    기1|05091 ]1이1191
    기113
    Ema Bmr870 79gEng| | 07030 70100 0a700 mggoo mong
    ]
    13 10 3 미269
    기6|1917 기1이33기1기 |16326729기60미이2끼327991|11이52049739139311
    173
    17107
    72000 - 81000

    10
    Cha oooo70 87 %EA Elasgos osmo moogy Omero Togg
    erwoongeragons | 08monOmous 0028a grow, Omison
    ESGod, EgoosSooJoa 18IgGod onto Sos_

    Sayot= cooS= doodkyoco oggs| ElsaEugsgoooyo ER,


    155sㅇㅇ으 ㅇㅇㅇ Roy= sucos odos | 8 | E a의 sgBoss og.
    |321615642815946307 9|987671852947
    11 |이21기3391 91 67111 ]
    193 |12371
    0 1399
    10 11905기|1이971
    3 13177
    이|1591299기671 1631|1277
    0 130
    1011 0 ]
    11 2391
    977 1]2|59111 11 )7283
    | 7109

    190 | So
    1091
    191
    131 1 110 이

    0
    233
    1기이
    1|911
    *
    1 이
    000
    4|이1이 3
    111)791 ] 3
    17
    -

    11 ]이21019 73 |
    163 | 171
    151 191101
    1977 11 131
    | 2기 291
    3 기
    17 43 2231
    211

    @8mmgmonono amg
    |163 )0 1이 9 0
    ,

    -
    110 31
    191 이 기1971 229 173 97
    19 111 01 223
    371 13 239

    1
    Or Smaro Oma
    기233 11 13
    11 01 )
    113 .
    59

    )
    .
    83 1 01 0 |263

    -
    |
    137 |101
    31 03

    0
    0
    103 13 이53 |1991
    13 163 고 기
    191 이
    2111 7731 101
    111 1191 13
    | |23271
    11 1 141 ]
    om0m 70aimCon
    181
    311 0 11 | | 260
    - |19 / 11 107171110

    700 07100
    191 101 111
    1731 0 ]
    223 1071
    01 311 5 257 13

    Cor =wromo
    |111 1071
    26 1| 93 111 93 ) 7
    19 01 11 01 101
    131 150
    131
    smoo

    163
    |13
    191 731
    31 ON 1031
    101
    1이 91 1137
    370mm 9 : 373sn.

    17 11 13 기
    831 990 |
    111 01 229 281
    171 ]
    79
    111 /0
    |
    177151
    1677 |219|27113 이 이 기 이 251 0 3| 1 이

    317080p 0373m87005 1878 092005


    700 o07 178 200700

    |65535296
    go03 ostm01|

    3|223296 8|053
    1
    |
    43 19 111 0
    17 0 4이| 1 191 | 069
    111 13 01 |1811 31 | | 13 231 |
    271 |2251
    27
    1131
    198

    0 43 233
    |07
    5 311
    기77 11 1| 3

    0
    2|0133 |53 |
    13

    0
    263 283
    01 이4lo171l 791 기 2313 이
    01191 101 7|1091 이
    13 101
    01이기 |이219391 |151
    0 17 111 111 113
    11281 103 13|27211133 0|191 | 01 13
    011371 1977 이 19 |이151 이
    |)13|139이 |2727771
    1117101 ]이1기3831 이 13
    23 이 |233
    19
    7 0191 이 711511
    | 17007a goono 750 ||
    13이기|109]1071
    1911
    291 0) 17 1K
    111
    41 7 0 51|1 301 014917
    | 1| 3 311 |2기1371191
    79013기
    17기0 31
    1기 10 01 1271
    191
    이 43
    191
    10
    ]11071

    ono ,
    10 1기 1 |113 01
    ]01491
    07 8

    0
    163 11 11 |2579 |13
    281
    677
    111
    291 193 0971
    1031
    71 0| 1 60 11 )
    11 |5491171
    2411
    131
    기1491 이 111
    ] 09 179
    59! 231
    171 11 191
    7어17
    기17
    19 19| 1491
    이1이01103 10
    13|1283
    11101
    89
    103
    | 13
    3
    |
    이 7
    313771401
    | 131 이 이17 |
    1331
    700 131
    | 591 1311
    01 0 157
    1377 11
    1263
    |:270 131|기 491
    431 이기 01 ]3131 137 241
    57 43|1913
    269 기 113 11 0
    -

    53ost Escoso = =____


    |1533110
    1|91 6 2111
    이67. )
    109

    onCaron monon-
    |215118 |33063 |665442630741859 |9870854371
    1911 13|011 이 1371 1431 191
    7

    )2007 1071 0 DI 2231 233 2111 43 0


    101 0 13|2| 230 11 0 가 31
    131|111
    1811 23 13 131 191 0
    |6111| 13| 0 31 7 19|3122891
    이기|11 47|3729
    531 101

    307
    271
    23| 2이기9 T!
    ) 1811
    이 110 173 |
    131
    기 137
    이기 |1
    41|1177 이 10 11
    43191
    1기 011311
    -|977
    110 기 1911
    1| 3

    .
    0
    3| 7 .기 023
    1391 1| 59
    233
    111 071 241
    ror

    1 23 7 173 11
    |1591
    ,

    113 이53|17 197

    -
    19
    ) 3이 11
    977 53
    1991
    |113 7
    269 |11 19
    13
    | 01 |113
    ]03
    7139 )10
    1
    311
    11 이199 0 11기41
    67
    17기 1
    , 99
    263 |77
    10
    1311 1153
    ] 277 1977 10
    )7 791
    111 131 7 2 ]
    103
    2511
    01 100 271
    13137 |
    103
    177 3
    |2011091
    19 09 17기 3 109
    1131 이11
    3 1기 13 23
    |10 |191 291
    153 기2| 131 37
    이477 31 지 5|133 이 179
    2107
    ]9 19이07 이13
    | 231 0 19 301 73 23 10 7 1|93 0
    1기이089
    | 81000 - 90000

    이기19 기113|이81 11191 01]043911 12이이 957 이 31|이

    Ima 70mmo 0808 | alrmoro Onospage 70SET Maram


    기
    Stemsm07 08078 - 580007 Onmyoo080p 07807 8000m

    yotago S9seo gosg= & oodo|

    aalmer09oo again B787 00s200


    roTag , 98170 moo05 700500|
    |125 |32963704 |65457630 |1769852 |8417
    73 이기
    13 이 231 13 0 1 )
    89
    11 17

    ,
    23 7 0 41 01 13
    /
    13 411
    251 1기11791

    300 ~
    000이
    기 30 1
    2이 3 01
    |1191171131 1이 7 |251 13 1911 1131 0 17
    이131 01 131
    269
    59
    1971 1기
    791
    이2291| 01 1811 17291 109
    )7기 |이 ) 5
    22 109 11 기 0 이
    1이
    7103 | 13
    1311] 31 013 257
    111
    )13 |이123233
    2391 1791 7 991 이1기 137 111 131
    이기91162|17710311 |291
    13331 이1] 03
    391
    31271 |]1192169 477 23
    67 0
    03|103
    |1기 21]이7|1111991
    11193 이17 31 111
    |1371
    431 111
    277 171 .
    139 0 13
    01 13 01

    .
    000
    oner

    | |17기| 1031 0 61
    113 13|2|763 891이기
    291 229 기173
    ]17 19 1 71 13
    111
    0 111 29 123 |283
    0

    || 23
    or

    477
    기.1317
    11 211
    163 기 33 0 1911

    03807 08079 oona8 70007


    13 ]113이
    1 307 111 29
    111 391 ]23
    11 29|271 0 191
    77 이 기0 13 131 177 기173
    111
    기 11 기1|2233
    11
    491
    |127|0 ]|161933591 01 3| 07 1491
    1531
    430 이 13

    jeno09 70580 0s70m ²700m20078 0

    na0383 0 007037
    1| 6 |1)9 2) 2 2) 5 43|433174)46|4091 |879852
    |13 177 기|012295737391
    33 1811 31 이1017 |1기437기
    283
    13 110 |
    71 17 |11 01 1 149 111 31
    |1137
    기 17 19|이11
    173 기
    2917 83기 |이89 ]|107
    이3기 1491 1 | 111 117 191 0
    2511 0 791
    이13 149 기119
    |1349 229
    131 31 2|137
    97 |
    107 0 0 2|이1123 177 기23 179
    17 17
    11 |
    131 257
    23
    이271
    131
    311
    13307 K
    59 411 181
    2271 51 91 23 791 19111977
    233 1]| 10
    37
    17|7기
    1491 |13 10675911 이 10
    soyo
    93 129 |3137
    1731 4이 )이기3|149 |
    233 UI
    1 |1기[728251
    834399131377
    231 59|13149 | 837기9 223 기이 기1이 3939111
    |2163 |3이121277
    기251
    311
    7139
    0 |이기2 417|] 이 1 ) 131 29| 101| 1313 | 091기
    이1531
    |8기179177
    3111
    18103 기
    1|149 1이 11
    227
    157| 이 31| 19

    n coon0 070
    1511
    이101)3기 677 11
    101
    |10 291
    이기1이71241 이110
    ]|16313
    191이11
    1|29171 269|이103 기 이
    3 |기2이1991기 19이 713] 1391
    41
    23 715
    231기1383
    69|15177
    477 이 61
    11

    one
    10
    11 151 181 |
    131 127 163
    ] 109
    1391 137 1011 13|1 |
    | 41 ]
    313
    111 이2411
    431 11
    13 131 101
    ]1 이 143
    | 31 IN 11

    Tomogo
    231 151
    431 01 0 311
    63 131 11 191 11 227119
    이1109
    | 11 0 151
    431 211 13 1| 11
    0)281 27 33 1511 1431
    173 080307 8087
    20
    0
    13 111 | 4이1|3 이

    800 097030 70400g

    0 875080 non430
    13 820070
    010 880²70 072890|
    Song00|

    3
    5|43210056351784029 8|1765693048617
    %1 13 기 이257 151
    131117 293 191 /
    |이101 19
    ,
    41191
    1 17 263
    1 171
    3|7m70
    13
    31 1
    - -
    Can79000m 60g pool glan

    131 1 2111
    1271 101
    01 SO 13 104 167 .이
    0 199
    171 111
    103 291 127 131 53
    61 89
    |107 0 211 11 391 991
    00sus
    330007 Baign

    13 0)17
    1011 151
    19 1 23 83 1391
    13 101
    ]|701기71 )
    109 기 ]
    13 903
    1090 131 17
    ] |0 97
    111
    4이7 11 291 이11 191 2391
    111 01
    127 190
    29
    117 411 !
    101 677 147
    171 111 13 ]23
    |191
    167 기 1 131

    04070
    1
    20080

    111 179 )0 기 이 이 111


    1/ 9 13 0 |0)193 0 271
    471 611 ]13223 11 291
    0
    3:2|9 311 97 ]13
    101 0 0 |1571
    11 |60
    191 137 131 ]
    01283 19291 이 1011
    1
    |3 11 이 127 |231411
    7msan

    13
    )223269 11 |163
    1571 2777
    이 1| 01 0 101 개
    기 111
    263 )
    19 )
    103 | 01
    -1

    1 |31이131
    191
    1911 611 11 |17 기|471 | 01
    이 71 131 183 233
    |791 13] 137 1|21|63391 61기 7

    8 73030
    0 41 113 63 29
    90000 - 99000,

    00g 303 90700 800no ooon70

    ego 80070 0700m 0033 908700


    007g onwoo7
    imrgon
    30 Log T (n).
    M | 0 | 4 | 5 | 6 | 7 | 8 | 9

    1.00 | 197497 1 95001 1 92512 | 90030 | 87555 185087 | 82627 | 80173 | 777277
    1.0119.9975287 172855 |70430 | 68011 165600 | 63196 | 60799 | 58408 | 56025 | 53648
    1.02. 51279 48916 46561| 44212 | 41870 |39535 | 37207 34886 32572 | 30265
    1.03 ) 27964 25671 |23384 21104 18831 16564| 14305 12052 | 09806 107567
    1.04 05334 |03108100889 98677 196471 194273192080 89895187716 185544
    1, 0519.9883379 | 81220 | 79068176922174783 172651170525 168406 / 66294 | 64188
    1 .06 62089 159996157910 155830| 53757 151690 | 49630 | 47577 1 45530 | 43489
    1.07 41469 39428 |37407 |35392 | 33384 | 31382 | 29387 |27398| 25415 | 23449)
    1 .08 21469 19506 | 17549 15599 | 13655 | 11717 | 09785 | 07860 05941| 04029 |
    1.09 | 102123 | 00223 198329 96442| 94561 192686 90818 188956 187100 |852500
    |1.1019.9783407
    1. 11
    81570| 79738 |77914 | 76095| 74283| 72476 |70676 68882| 67095
    65313 | 63538 | 61768 | 60005158248 1 56497154753 | 53014 151281 | 49555
    1. 12 ) 47834 | 46120 | 44411 | 42709 | 41013 1 39323 1 37638135960 | 34288 | 32622
    1 . 13 30962129308 1 27659 | 26017 124381 |22751| 21126 | 19508 | 17896 | 1628
    11.14 ] 14689 | 13094 | 11505109922108345 | 06774 | 05209 | 03650 | 02096 100549
    19 .9699007 197471 195941 194417 192898 191386 | 89879 | 88378 | 86883 | 85393
    1.16 83910 |82432 180960 | 79493 | 78033 176578| 75129 | 73686 | 72248 170816
    1.17 69390 | 67969 166554 ! 65145163742 | 62344 | 60952 | 59566 | 58185 | 56810
    1 .18 55440 | 54076 | 52718151366 | 50019 1 48677 | 47341| 46011 | 44687 | 43368
    42054 | 40746 |39444 38147| 36856 |35570 | 34290 |33016 31747 | 30483)
    1.20 29225 | 27973 | 26725 | 25484 | 24248 | 23017| 21792 | 20573 | 19358 | 18150)
    1 .21 16946 | 15748| 14556 13369 12188| 11011| 09841 | 08675| 07515 106361
    1 .22 05212 |04068 102930 01796|00669 199546| 08430 97318 | 96212 |05111
    1.2319 .9594015 192925 191840 |90760 | 89685 | 88616 | 87553 | 86494 | 85441 | 84393
    1 .24 83350 | 82313 |81280 80253 | 79232 78215 | 77204|76198 | 75197| 74201
    1 .25 73211 72226171246 70271 | 69301| 68337167377 166423 | 65474 | 64530
    1.26 63592 | 62658 | 61730 | 60806 | 59888 158975 | 58067 | 57165 | 56267 | 55374
    1.27 54487 | 53604 1 52727 151855 150988 | 50126 1 49268 | 48416 | 47570 | 46728
    1.28 45891 | 45059 44232 | 43410 42593 41782 | 40975 | 40173 | 39376 | 38585
    1 .29 37798 37016 |36239 35467 | 34700 | 33938 | 33181 | 32439 31682 | 3094 )
    30203 29470 | 28743 28021 | 27303 |26590 | 25883 | 25180 | 24482 | 23789)
    23100 | 22417121739 21065 120396 | 19732 | 19073 | 18419 | 17770 | 17125
    1.32 16485 | 15850 | 15220 14595 | 13975| 13359 | 12748| 12142| 11540 | 10944
    1.33 10353 | 09766 | 09184 08606 | 08034107 466 | 06903 1 06344 | 05791 | 05242
    1.34 04698 | 04158 103624 03094 102568 102048 | 01532 | 01021| 00514 100012
    1.35 19 .9499515 199023 198535 98052 197573 97100196630| 96166 | 95706 195251
    1 .36 | 94800 194355 193913 43477 193044 1 92617 192194 91776 | 91362 190953
    1. 37 90549 | 9014 189754 89363 | 88977 | 88595 188218 187846 | 87478 | 87115
    86756 | 86402 | 86052 85707185366 185030184698 | 84371 | 84049 | 83731
    139 . 83417 | 83108| 82803 82503 182208 | 81916 81630 | 81348| 81070 | 80797
    1 .40 80528 80263| 80003 79748 | 79497| 79250 79008 | 78770| 78537 |78308
    78084 | 77864 |77648 77437 177230 | 77027 1 76829 | 76636 | 76446 | 76261
    1 . 42) 76081 | 75905 | 75733 75565 | 75402 1 75243 | 75089 | 74039 | 74793 | 74652
    1.43 74515 174382174254 74130174010 173894 173783 173676 | 73574 |73476
    1.44 73382173292 173207173125 1730449 | 72976 172908172844 | 72784 | 72728
    1 .45 72677| 72630 72587| 72549 72514| 72484| 724549 72437| 72419| 72406
    1.46 72397172303172302 72306172404 1724161724:321724.52 1 72477 172506
    1 .47 72539 72576 | 72617 72662 |72712 | 72766 72824 172886 72952 | 73022)
    1 .48 73097| 73175 | 73258 73345 173436| 73531 |73630 |73734 73841| 73953
    1 .49 74068 74188)174312 74440 | 74572174708174348 17 4992175141 175293
    Note. This table is taken from Vol. II. of Legendre' s work , and not
    from Vol. I., as stated in the Preface : the numbers given in Vol. I. being
    inaccurate in the seventh decinal place. In Vol II. the values are given to
    twelve places of decimals. The figure here printed in the ai venth place is
    Log f (n). 31
    13 4 / 5 78
    1.50 9 .9475449 75610 |75774 759437611676292 76473 76658 76847 77040
    1.51 772377743877642 7785178064 78281 78502 78727 | 78956 79189
    1.52 79426 79667 79912180161 80414 80671 8093281196 /8146581738
    1 .53 82015 8229582580 |828688316183457 83758840628437084682
    1.54 84998 | 8531685642 8597086302866383697787321 87668 88019
    1.55 88374 88733 89096 8946389834 9020890587 90969 91355 91745
    1.56 9213992537 92938 93344 93753 94166 94583 95004 95429 95857
    1.57 96289 9672597165 97609 9805698508 98963 99422 99885 70351
    1.589.9500822 01296 01774 02255 02741 03230 0372304220 04720 05225
    1.59 05733 0624506760 0728007803 08330 08860 09395 09933 10475
    1 .60 1102011569 12122 12679 13240 13804 14372 14943 15519 16098
    1.61 16680 17267 17857 18451 19048 19650 20254 20862 21475 22091
    1.62 22710 /23333 23960 24591 25225 25863 26504 27149 2779828451
    1 .63 29107 | 297673043031097 31767 /32442 33120 33801 3448635175
    1.64 35867 3656337263 37966 38673 / 39383 40097 40815 41536 42260
    1 .65 4298943721 44456 | 45195 45938 4668447434 48187 | 48944 | 49704
    1.66 50468 51236 52007 | 52782 | 53560 54342 55127 55916 |5670857504
    1.67 58303 59106 5991360723 61536 62353 63174 63998 |64826 65656
    1.68 66491 |6732968170 69015 69864 70716 71571 72430 73293 74159
    1.69 75028 75901 76777 77657 78540 79427 80317 81211 82108 83008
    1.701 8391284820 85731 86645 87563 88484 89409 90337 91268 92203
    1.71 93141 94083 95028 9597796929 97884 98843 9980570771 71740
    1.72 9.960271203688 04667 056500663607625 08618 09614 10613 11616
    1.73 1262213632 14645 | 15661 16681 17704 18730 | 1976020793 / 21830
    1.74 22869 23912 2495926009 27062 28118 29178 /30241 3130832377
    1 .75 33451 / 34527 3560713669037776 38866139959 |41055 42155 / 43258
    1.76 44364 45473 46586 47702 4882149944 51070 52200 5333154467
    1.77 55606 56749 57894 5904360195 61350 6250963671 64836 66004
    1 .78 67176 68351 6952970710 71895 73082 7427475468 | 76665 77866
    1.79 790708027781488 82701 83918 85138 86361 875888881890051
    1 .80 91287 92526 93768 95014 96263 97515 98770 7002971291 02555
    1.81 9.970382305095 0636907646 |08927 1021111498 12788 14082 15378
    1 .82 16678 17981 19287 20596 21908 23224 2454225864 27189 28517
    1.83 29848 31182 32520 33860 35204 36551 37900 39254 40610 41969
    1.84 43331 44697 46065 47437 48812 50190 51571 52955 54342 55733
    1.85 57126 58522 59922 61325 62730 64140 65551 6696668384 69805
    1.86 7123072657 74087 |75521 7695778397 79839 81285 82734 84186
    1.87 85640 8709888559 90023 91490 92960 94433 95910 9738998871
    1.88 9.9800356 01844 03335 04830 06327 07827 09331 10837 | 12346 13859
    1.89 15374 16893 18414 19939 2146622996 245302606627606 29148
    1.90 30693 32242 33793 35348 36905 38465 40028 41595 43164 44736
    1.91 46311 47890 49471 51055 52642 54232 55825 57421 | 59020 60622
    1.92 62226 63834 65445 67058 68675 70294 71917 73542 |7517076802
    1.93 7843680073 81713 83356 85002 86651 88302 8995791614 93275
    1.94 94938 96605 98274 99946 01621 73299 5498076663 7835010039
    9 .9911732 13427 1512516826 1853020237 | 21947 2365925375
    1.96 28815 3053932266 33995 3572837464 39202 40943 42688 44435
    1.97 46185 47937 49693 51451 53213 5497756744 5851360286 62062
    1 . 98 63840 65621 67405 69192 70982 7277474570 76368 78169 79972
    1.99 8177983588 85401 87216 89034 90854 92678 94504 96333 98165
    the one nearest to the true value whether in excess or defect. This table , and
    the table of Least Factors, havo each been subjected to two complete and in
    dependent revisions before finally printing off.
    ALGEBRA .

    FACTORS .

    a ' — b = (a - b ) (a + b ).
    a ' _ b² = (a — b ) (a ’ + ab + b?).
    a ' + 63 = (a + b ) (a’ – ab + b ) .
    And generally ,
    4 a" — 6 " = (a − b) (an -l + a "-26 + ... + 6 "-1) always.
    5 a " — b" = (a + b ) (an-1 - an -2b + ... – 61-1) if n be even.
    6 a” + 6 " = (a + b ) (an-1 — a " -26 + ... + 61- 1) if n be odd.
    ny ( x + a ) ( x + b ) = x + + (a + b ) x + ab.
    8 (x + a ) (x + b ) (x + c) = x + (a + b + c) x2
    9 (a + b )' = a + 2ab + 62 + (bc + ca + ab ) x + abc.
    10 ( a , b ) = a - 2ab + b ?.
    11 ( a + b ) ) = a + 3a+b + 3ab² + 63 = a + 53 + 3ab (a + b ) .
    12 (a − b )3 = - 3a+ b + 3ab2 - 13 = a - 63 – 3ab (a − b ).
    Generally,
    (a + b ) = a ' + 7a®6 + 21a*b + 35a4b3 + 35a'b* + 21a*b* + 7ab® + .
    Newton's Rule for forming the coefficients : Multiply any
    coefficient by the index of the leading quantity , and divide by
    the number of terms to that place to obtain the coefficient of the
    term next following. Thus 21 x 5 : 3 gives 35 , the following
    coefficient in the example given above. See also (125 ).

    To square a polynomial : Add to the square of each term


    twice the product of that term and every term that follows it.
    Thus, (a + b + c + d ) ?
    = a + 2a (b + c + d ) + 62 + 26 ( c + d ) + c + 2cd + d².
    34 ALGEBRA

    13 a * a?b? + 6 * = (a’ + ab + b?) (a” – ab + 6?).


    14 a*+ b = (a'+ abV2 + b?)(a’— ab V2 + b ).
    15 (x + 1) = x + 2 +2, (x + 1) = x0+ +3(x + 1).
    16 (a + b + c)? = a’ + b + c + 2bc + 2ca + 2ab.
    17 (a + b + c) = a' + b + c + 3 (b'c + bc? + ca + ca ?
    ta'b + ab ) + 6abc.

    Observe that in an algebraical equation the sign of any


    letter may be changed throughout , and thus a new formula
    obtained , it being borne in mind that an even power of a
    negative quantity is positive. For example, by changing the
    sign of c in (16 ), we obtain
    (a + 6 — c) = a ’ + + 62 – 2bc — 2ca + 2ab.

    18 a° + 6° — c* + 2ab = (a + b)2 – c = (a + b + c) (a + b — c)
    by (1).
    19 a’– b —c* + 2bc = a' — (6 — c)2 = (a + b — c) (a − 6 + c).
    20 a + 63 + - 3abc = (a + b + c) (a ’ + b + c - bc - ca - ab ).
    21 bc + b'c + ca ' + c’a + ab? + a’b + a' + 63 + c3
    = ( a + b + c) (a ? + 62 + cº).
    22 bc + b 'c + ca’ + ca + ab? + a’b + 3abc
    = (a + b + c) (bc + ca + ab).
    23 bc? + 6 *c + caº + cʻa + ab? + a ’b + 2abc = (b + c) (c + a )(a + b )
    24 bc + bc + ca + cºatalº + ab - 2abc - dº - bº - cº
    = (b + c - a ) (c + a - b) (a + 6 — c).
    25 be? — b'c + ca’ — cºa + ab? — a’b = (6 — c) (c - a ) (a − b).
    26 26 c? + 2cʻa’ + 2a²b? – a* — 64 - C*
    = (a + b + c) (6 + c - a) (c + a — b) (a + b — c).
    27 + 2xy + 2xy + y = ( + 1) ( + y + y).
    Generally for the division of(x + y)" — (x" + y") by x2 + xy + yº
    see (545) .
    MULTIPLOATION AND DIVISION. 35

    MULTIPLICATION AND DIVISION,


    BY THE METHOD OF DETACHED COEFFICIENTS.

    28 Ex. 1 : (a* — 3a²b? + 2ab + 64) * (a ? — 2ab? — 26 ).


    1+0 -3+ 2+1
    1+0-2-2
    1+0 -3 +2+ 1
    - 2 - 0 +6 - 4 -2
    - 2 - 0 +6 - 4 - 2
    1 + 0 - 5 + 0 + 7 + 2 - 6 -- 2
    Result a? - 50%b? + 7a*b* + 2a²65 – 6ab9 – 267.

    Ex. 2: (z? — 5:2% + 7x8 + 2x2 – 6x - 2) = (x* — 3x2 + 2x + 1).


    1 + 0 - 3 + 2 + 1) 1 + 0 – 5 + 0 + 7 + 2 – 6 – 2 ( 1 + 0 – 2 – 2
    - 1- 0 + 3 – 2 - 1
    - 0 – 2 –2 +6 + 2 -6
    + 2 +0 –6 + *+ 2
    -2+0 +6 -4 -2
    +2 +0–6 +4 +2

    Result 22– 2x – 2. .

    Synthetic Division .
    Es. 3: Employing the last example, the work stands thus,
    | 1 + 0 - 5 + 0 + 7 + 2 –6 – 2
    - 0 0 + 0 +0 + 0
    + 3 + 3 + 0 –6 -6
    -2 - 2 + 0 +4 +4
    -1+0 +2 +2
    1+0- 2- 2
    Result 20 — 2x — 2. [ See also (248).
    Note that, in all operations with detached coefficients, the result must be
    written out in successive powers of the quantity which stood in its successive
    powers in the origiual expression .
    ALGEBRA .

    INDICES.

    29 Multiplication : a xa = ał+ 1 = aš, or 20% ;


    anxan = à * = a ", or "Wa***
    Division : 08 + a = at-1 = aš, or Va;
    azah =am= a***, or "Wa*-*.
    Involution : (23)} = q*** = ał, or Va.
    Evolution : Vaš = a $** = a ' , or Wa?.
    no a' = 1.

    HIGHEST COMMON FACTOR .


    30 RULE. — To find the highest common factor of two ex .
    pressions : Divide the one which is of the highest dimension by
    the other , rejecting first any factor of either expression which
    is not also a factor of the other. Operate in the same manner
    upon the remainder and the divisor, and continue the process
    until there is no remainder. The last divisor will be the
    highest common factor required .

    31 EXAMPLE. — To find the H . C . F . of


    3.25 – 10x + 15x + 8 and 2x – 2x* – 6.2% + ta ? + 13.0 + 6 .
    1 - 2 - 6 + 4 + 13 + 6 3 + 0 – 10 + 0 + 15 + 8 3
    - 3 + 6 + 18 - 12 - 39 - 18
    3 – 6 – 18 + 12 + 39 + 18 2 ) 6 + 8 - 12 - 24 - 10
    - 3 - 4 + 6 + 12 + 5
    3 + 4 - 6 – 12 – 5
    2 ) – 10 – 12 + 24 + 41 + 18 3 – 9- 9- 3
    - 5 - 6 + 12 + 22 + 9 - 5 - 15 – 15 – 5
    + 5 + 15 + 15 + 5
    - 15 – 18 + 36 + 66 + 27
    + 15 + 20 – 30 – 60 – 25
    2 )2 + 6+ 6+ 2 Result H . C . F . = a + 3x* + 3x + 1 .
    1+ 3 + 3 + 1
    EVOLUTION . 37

    32 Otherwise. — To form the H . C . F . of two or more


    algebraical expressions : Separate the expressions into their
    simplest factors . The H . C . F . will be the product of the
    factors common to all the expressions, taken in the lowest
    powers that occur.

    LOWEST COMMON MULTIPLE .

    33 The L . C . M . of two quantities is equal to their product


    divided by the H . C . F .
    34 Otherwise. — To form the L . C . M . of two or more
    algebraical expressions : Separate them into their simplest
    factors. The Ž . C . M . will be the product of all the factors
    that occur, taken in the highest powers that occur.

    EXAMPLE. — The H . C. F. of a’(6 — x)ºc?d and a’(6 — x) che is a’(6 — 2 ) c^;


    and the L . C . M . is a'(b − x )"c de.

    EVOLUTION

    To extract the Square Root of


    a’ – 3av2 a _ 3Va + 4la + 1.
    2 ' 16 T
    Arranging according to powers of a , and reducing to one denominator, the
    expression becomes 16a? — 24a: + 41a – 24ał + 16
    16

    35 Detaching the coefficients, the work is as follows :


    16 – 24 + 41 – 24 + 16 ( 4 - 3 + 4
    16
    8 - 3 - 24 + 41
    -3 24 - 9
    8 -6 +4 32 - 24 + 16
    - 32 + 24 – 16
    4a - 3a1 + 4
    Result = a - va + 1
    ALGEBRA.

    To extract the Cube Root of


    37 8xº — 36x” y + 66x*y – 63x*yvy + 33x*y* — 9x*Jy + ya.
    The terms here contain the successive powers of x and Vy; therefore,
    detaching the coefficients , the work will be as follows :
    I. II. III.
    6 - 3 ) 12 8 - 36 + 66 – 63 + 33 - 9 + 1 ( 2 - 3 + 1
    -6 ) - 18 + 9 - 8
    6 -9 +1 12 – 18 + 9 - 36 + 66 – 63 + 33 - 9 + 1
    + 9) + 36 – 54 + 27
    12 - 36 + 27 12 - 36 + 33 - 9 + 1
    6 -9 + 1 - 12 + 36 - 33 + 9 - 1
    12 - 36 + 33 - 9 + 1
    Result 2x * — 3x vy + y .
    EXPLANATION. — The cube root of 8 is 2 , the first term of the result.
    Place 3 x2 = 6 in the first column I., 3 x2 = 12 in column II., and 28 = 8
    in III., changing its sign for subtraction .
    - 36 - 12 = - 3, the second term of the result .
    Put – 3 in I.; (6 - 3) ^ ( - 3 ) gives – 18 + 9 for II.
    (12 - 18 + 9 ) 3 (changing sign ) gives 36 - 54 + 27 for III. Then add.
    Put twice ( - 3 ), the term last found, in I., and the square of it in II.
    Add the two last rows in I., and the three last in II.
    12 : 12 gives 1 , the third term of the result.
    Put 1 in col. I., (6 — 9 + 1) x1 gives 6 - 9 + 1 for col. II.
    (12 - 36 + 33 – 9 + 1 ) xl gives the same for III. Change the signs, and
    add, and the work is finished.

    The foregoing process is but a slight variation of Horner's


    rule for solving an equation of any degree. See (533).

    Transformations frequently required .


    38 If = g,then ata = ....................... [68.
    39 If x + y = al, then <a = (a + b ) .
    and x - y = bíf,
    then { x = } (a − b ).

    (x + y) + (x − y)' = 2 (2® + y ).
    (x + y) — (x − y) = 4xy .
    EQUATIONS.
    :

    (x + y)² = (x − y)? + 4xy.


    (x − y)' = (x + y) — 4xy .
    EXAMPLES .

    2 Va? — 6? + Vc? — * * _ 3Va? — 6? + V2–


    2 Va’ - 62 - V c? — 3Va – 6 : - 10% – ď?'
    Vé – aut _ Veď ....................[ 38.
    2V/~* — 633VQ2- 63
    9 (0* — * *) = 4 (cº - d ”),
    x = V5c? + 4d .

    To simplify a compound fraction, as

    a - ab + 62a + ab + 6%
    11
    a ’ - ab + 62a² + ab + 63
    moltiply the numerator and denominator by the L . C . M . of all the smaller
    denominators.
    Result (a' + ab + 6?) + (a ' — ab + 6 ?) _ a' + 6°
    (a ’ + ab + 6 ) - (a* - ab + 6 %) ab

    QUADRATIC EQUATIONS.

    45 If ax + bx + c = 0, x = =63 2a
    46 Ifax? + 2bx + c = 0 ; that is, if the coefficient of x be
    _ - 6EV 62 - ac
    an even number , x =

    47 Method of solution without the formula .


    Ex. : 2x* — 7x + 3 = 0 .
    Divide by 2, P + = 0.
    ALGEBRA.

    • 7 1712 493 _ 25
    Complete the square, x*

    lo
    I
    Take square root, 2-1 = =

    H
    * = 745= 3 or
    48 Rule for “ completing the square" of an expression like
    22 — žx : Add the square of half the coefficient of x .

    49 The solution of the foregoing equation, employing formula (45), is


    = =b + VD
    2a
    –4ac – 7 # V49 –24 – 7 # 5 = 3 or ?

    THEORY OF QUADRATIC EXPRESSIONS.

    If a , ß be the roots of the equation ax® + bx + c = 0, then


    50 ax: + bx + c = a (x — a )(x — B ).
    51 Sum of roots a + B = - .
    52 Product of roots aß =
    Condition for the existence of equal roots
    53 62 - 4ac must vanish .

    54 The solution of equations in one unknown quantity may


    sometimes be simplified by changing the quantity sought.

    Ex. ( 1 ) : 2x +_ 3x33x +- 11 '. 6x218+ x5x+ -6 1


    -- - + = 14 .................... (1).
    6x® + 5.0 – 1 , 6 (3x +0 -1)1 - 14
    3x + 1 ' 6x” + 5x - 1
    Put y = 6x? + 5.0-- 1......................... (2).
    y = 6x + 5 .0 - 1
    3x + 1
    EQUATIONS.

    thus y + 6 = 14 .
    y? – 14y + 6 = 0 .
    y having been determined from this quadratic,æ is afterwardsfound from ( 2).

    55 Ex.2: + 3 +2 + 1 = 4.
    (0+ 1) +(2 + 1) = 6.
    Pat æ + 1 = y, and solve the quadratic in y.

    56 Ex.3: **+x + V2x + x + 2 = + 1.


    2x* + x + 3 / 2x* + & + 2 = 2,
    2x + x + 2 + 3v2.x* + x + 2 = 4 .
    Put V2.0 + x + 2 = y, and solve the quadratic
    y* + 3y = 4 .

    57 Ex.4 : Vx + 3 x = *+.

    A quadratic in

    58 To find Maxima and Minima values by means of a


    Quadratic Equation .
    Ex. - Given y = 3. + 6x + 7,
    to find what value of x will make y a maximum or minimum .
    Solve the quadratic equation
    3x? + 6x + 7 - y = 0 .
    Thus x =_ =- 3 V3y - 12 [ 46 .

    In order that x may be a real quantity ,we must have 3y not less than 12 ;
    therefore 4 is a minimum value of y , and the value of a which makes y a
    minimum is – 1.
    ALGEBRA

    SIMULTANEOUS EQUATIONS.

    General solution with two unknown quantities.


    Given
    59 a, x +by = ci ? - C16 , - 0. 6 , Ca. — C ,Q1
    02x + b, y = c. l ' a , b, - a ,bi b, a , – b ,a ,

    General solution with three unknown quantities.


    60 Given ayx + by + c % = d )
    09x + by + cx = d ,
    0.3x + buy + 0: 2 = dz)
    _ d1(1) ,C3 - b3c2) + d . (b3C1 - 6 ,c ) + d3 (67C2 - b , c,)
    a , (62C3 — 63C2) + Q2 (63C, - 6 , 63) + (lz (6 , , - b2cı)'
    and symmetrical forms for y and z.
    1

    Methods of solving simultaneous equations between two


    unknown quantities x and y .

    61 I. By substitution. Find one unknown in terms of the


    other from one of the two equations, and substitute this value
    in the remaining equation . Then solve the resulting equation .
    Ex. : x + 5y = 23 ...... (1 ) ?
    7y = 28 ...... (2 )
    From (2 ), y = 4 . Substitute in ( 1) ; thus
    x + 20 = 23, x = 3.

    62 II. By the method of Multipliers.


    Ex.: 3x + 5y = 36 ...... ( 1 )
    2x – 3y = 5 ...... (2 ) S '
    Eliminate x bymultiplying eq. (1) by 2 ,and ( 2) by 3 ; thus
    6 .x + 10y = 72,
    60 – 9y = 15 ,
    19y = 57, by subtraction,
    y = 3;
    ii = 7, by substitution in eq. (2).
    EQUATIONS. 43

    63 III. By changing the quantities sought.


    Ex. 1 : 2 -y = 2 ...... ( 1) 2
    22 — y + x + y = 30 ...... (2 ) $
    Let x + y = U, æ — y = v.
    Substitute these values in ( 1) and (2 ),
    v = 2 .
    uv + u = 30 %
    :: 2u + u = 30,
    u = 10 ;
    : x + y = 10,
    2 - Y = 2.
    From which æ = 6 and y = 4 .

    64 Ex.2: 2®X +- !Y + 10 X% + )y = .......(1) .


    **+ 74° = 64.......(2)
    Substitute z for * *) in (1);
    On

    . 23 + 10 = 9 ;
    22 –9z+ 10 = 0.
    e

    From which
    •

    2 - Y
    From which
    x = 3y or šy.
    Substitute in (2) ; thus y = 2 and x = 6 ,
    y = 6 , and æ = 2v7.

    65 Ex. 3 : 3x + 5y = xy ...... (1 )
    2x + 7y = 3xy ...... (2 ) ) .
    Divide each quantity by wy ;
    3 + 5 = 1... ..(3) )
    2 + ? = 3 ...... (4 ) )
    ya
    Multiply (3 ) by 2,-and ( 1 ) by 3, and by subtraction y is eliminated.
    ALGEᏴᎡᎪ.

    66 IV. By substituting y = tx , when the equations are


    homogeneous in the termswhich contain x and y .
    Ex . 1 : : 52x* + 7xy = 5yº ...... ( 1 )
    5x – 3y = 17 ...... ( 2 )
    From ( 1), 52x + 7txa = 5+* ?... .. (3 ) 2
    and, from ( 2 ), 5x — 3tx = 17 ...... (4 ) S .
    ( 3 ) gives | 52 + 7t = 5t ,
    a quadratic equation from which t must be found, and its value substituted
    in (4 ).
    æ is thus determined ; and then y from y = tx.

    67 Ex. 2 : 2x* + xy + 3y2 = 16 ...... ( 1)


    3y - 2x = 4 ...... ( 2) ) .
    From ( 1 ), by putting y = tx ,
    (2 + 1 + 31 ) = 16 ...... (3) .
    from ( 2 ), æ (3t - 2 ) = 4 ...... (4 ) S '
    squaring , ai (982 – 12t + 4) = 16 ;
    i 91 – 12t + 4 = 2 + t + 3t",
    a quadratic equation for t.
    t being found from this, equation ( 4) will determine x ; and finally y = tx .

    RATIO AND PROPORTION .

    68 If a : 6 :: c : d ; then ad = be, and = i ;


    ath c + d a - 6 C- d a + 6 c + d
    a - 6 C - dº

    69 If ( = 8 = * = &c.;then ( = %# &#PAGO:
    General theorem .

    70 If © = á = = &c. = k say, then


    Spa " + qc" + re" + & c . ) ,
    (pb" + qd " + rf " + & c.)
    where p , q , r , & c. are any quantities whatever. Proved as
    in (71)
    RATIO AND PROPORTION. 45

    71 RULE. — To verify any equation between such proportional


    quantities : Substitute for a , c, e, foc., their equivalents kb, kd ,
    kf, & c. respectively, in the given equation .
    Ex. - If a :6 :: c : d , to show that
    va - b _ va -
    ſo - d ſc - ſd
    Put kb for a , and kd for c ; thus
    Va - h _ V kb - 6 _ 1bVk - 1 _
    vo - d Vkd -d vd / k - 1 v
    also va - Vb _ kb - Vb - Vb ( k - 1) .
    vc- vd Vkd - vd vd ( v k - 1)
    Identical results being obtained , the proposed equation must be true.

    72 If a : b : c : d : e & c., forming a continued proportion ,


    then a : c :: a’ : b , the duplicate ratio of a : 6 ,
    a : d : : a : 63, the triplicate ratio of a : b, and so on.
    Also Va : vb is the subduplicate ratio of a : b ,
    al : 37 is the sesquiplicate ratio of a : b .

    73 The fraction Ÿ is made to approach nearer to unity in


    value, by adding the same quantity to the numerator and
    denominator. Thus
    ab ++ ®ac is nearer to 1 than
    nearer
    is.
    74 DEF. — The ratio compounded of the ratios a : b and c : d
    is the ratio ac : bd .
    75 If a : 6 :: c : d , and a ' : ' :: c' : d '; then , by compound
    ing ratios, aa' : bb ' :: cc' : dd'.

    VARIATION .
    76 If a & c and b « c, then (a + b) occ and Vaba c.
    77 If aab}, then ac o bd and ca
    and cod )'
    78 If a ob, wemay assume a = mb,where m is some constant.
    46 ALGEBRA

    ARITHMETICAL PROGRESSION .

    General form of a series in A . P .


    a , atd, a + 2d , a + 3d , ...... a + (n - 1) d .
    a = first term ,
    d = common difference ,
    l = last of n terms,
    $ = sum of n terms ; then
    1 = a + (n - 1) d.
    81 : s = (a + 1)
    82 s = {2a + (n - 1)d } "
    · Proof. — By writing (79 ) in reversed order , and adding both series
    together .

    GEOMETRICAL PROGRESSION.

    General form of a series in G . P .


    a , ar, ar ?, ar ), ...... apr -1.
    a = first term ,
    go = common ratio ,
    1 = last of n terms,
    s = sum of n terms; then
    84 1 = ar -1.
    85
    srama] or at
    If r be less than 1, and n be infinite ,
    86 s = 4 ,, since 1* = 0.
    Proof. — (85) is obtained by multiplying (83) by r,and subtracting one
    series from the other.
    PERMUTATIONS AND COMBINATIONS. 47

    HARMONICAL PROGRESSION .

    87 a , b , c, d, & c. are in Harm . Prog. when the reciprocals


    a z 1 1 .&c.are in Arith.Prog.,
    88 Or when a : b :: a - 6 : 6 - c is the relation subsisting
    between any three consecutive terms.
    ab
    89 nth term of the series = (n
    7 - 1) a - ( n - 2) [87, 80 .
    90 Approximate sum of n terms of the Harm . Prog. eurata
    1 la la
    , when d is small compared with a ,
    atď a + 2d ' a + 3d &
    (a + d )" - a"
    da + d )" "
    ta
    1 a ta 2
    st
    C a
    e
    at
    Proof. — By taking instead the G . P . - G

    91 Arithmetic mean between a and b


    92 Geometric do. = Vab.
    93 Harmonic = 2ab
    do.
    = a +6
    The three means are in continued proportion .

    PERMUTATIONS AND COMBINATIONS.

    94 The number of permutations of n things taken all at a


    time = n (n - 1) (n - 2) ... 3 . 2 . 1 = n ! or n (n).
    Proof BY INDUCTION.-- Assume the formula to be true for n things.
    Now take n + 1 things. After each of these the remaining n things may be
    arranged in n ! ways, making in all n x n ! (that is (n + 1) ! ] permutations of
    n + 1 things ; therefore, & c . See also (233) for the mode of proof by
    Induction.
    48 ALGEBRA

    95 The number of permutations of n things taken r at a


    time is denoted by P (n , r).
    P (n ,r ) = n (n - 1) (n — 2) ... (n — r + 1) = nm).
    PROOF. - By (94) ; for (18 — ) things are left out of each permutation ;
    therefore P (n , r) = n ! : (n - r )! .
    Observe that r = the number of factors.

    96 The number of combinations of n things taken r at a 10

    time is denoted by C (n , r ).
    n ( n - 1 ) ( n - 2 ) ... ( n - r + 1 ) .
    Can ,
    1 . 2 . 3 . .. 1
    n !
    = C (n , n - r ).
    r ! (n - 1 ) !
    For every combination of things admits of nu! permuta
    tions ; therefore C (n , r ) = P ( n , ? ) • r !

    97 C (n , r ) is greatest when r = in or ž (n + 1), according


    as n is even or odd .

    98 The number of homogeneous products of r dimensions


    of n things is denoted by H (n , ).
    H ( . r) = n (n + 1) (n + 2) ... (n + r - 1) = (n + r - 1)(*)
    =
    1 . 2 . .. * r !

    When r is > n , this reduces to


    ta (r + 1 ) ( r + 2 ) .. . (ntr - 1
    erra 99 (x - 1 ) !

    PROOF. — H (n , r ) is equal to the number of terms in the product of the


    expansions by the Bin . Th . of the n expressions ( 1 - ax) - ?, (1 - bx)- ?,
    (1 - cx) ?, & c.
    Pat a = b = c = & c. = l. The number will be the coefficient of win
    (1 - x ) -". (128, 129 .)
    PERMUTATIONS AND COMBINATIONS. 49

    100 The number of permutations of n things taken all to


    gether, when a of them are alike, b of them alike, c alike, & c.
    n !
    a ! 6 ! c ! . .. & c .
    For, if the a things were all different, they would form a !
    permutations where there is now but one. So of b , c, & c .

    101 The number of combinations of n things r at a time,


    in which any p of them will always be found, is
    = C (n - p , r - p ) .
    For , if the p things be set on one side, wehave to add to them
    g - p things taken from the remaining n - p things in every
    possible way .

    102 THEOREM : C (n - 1, r - 1) + C (n - 1, r ) = C (n , r ).
    PROOF BY INDUCTION ; or as follows : Put one out of n
    letters aside; there are C (1 - 1 , r ) combinations of the re
    maining n - 1 letters p at a time. To complete the total
    C ( n , r ), wemust place with the excluded letter all the com
    binations of the remaining n - 1 letters P - 1 at a time.

    103 If there be one set of P things, another of Q things,


    another of R things, and so on ; the number of combinations
    formed by taking one out of each set is = PQR .., & c ., the
    product of the numbers in the several sets.
    For one of the P things will form Q combinations with
    the Q things. A second of the P things will form Q more
    combinations , and so on . In all, PQ combinations of two
    things . Similarly there will be PQR combinations of three
    things; and so on . This principle is very important.

    104 On the same principle , if p , q , r , & c . things be taken


    out of each set respectively, the number of combinations will
    be the product of the numbers of the separate combinations ;
    that is, = C (Pp) . C (QI) . C (Rr) ... & c .
    ALGEBRA .

    105 The number of combinations of n things taken m at a


    time, when p of the n things are alike, 9 of them alike, r of
    them alike, & c., will be the sum of all the combinations of
    each possible form of m dimensions, and this is equal to the
    coefficient of ac" in the expansion of
    (1 + x + x + .. . + ) (1 + x + x2 + ... + 29) ( 1 + x + x + . .. + x ^) ... .
    106 The total number of possible combinations under the
    same circumstances, when the n things are taken in all ways,
    1 , 2 , 3 ... n at a time,
    = (p + 1) (9 + 1) (r + 1) ... - 1.
    107 The number of permutations when they are taken m
    at a time in all possible ways will be equal to the product of
    m ! and the coefficient of am in the expansion of
    na 3

    {1+2+ 3 + 3 +...+ }{1+0 + * + **+...+ -

    SURDS.

    108 To reduce V2808. Decompose the number into its


    prime factors by (360) ; thus,
    2808 = V28 . 38. 13 = 6 713,
    Vu5 610 * = a * b *c' = a*b*c* 6'cí = a* b* c* Vīc“.

    109 To bring 5 V3 to an entire surd.


    5 / 3 = V54. 3 = V1875,
    z yizi = % ya z= = * x*" y®zł".

    110 To rationalise fractions having surds in their


    denominators.

    have been 99= VAN


    SURDS. 51

    111 3 _ _ 3 (9 + 780 ) = 3 (9 + 7/80),


    9 - 780 81 - 80
    since (9 – 780) (9 + V80) = 81– 80, by (1).
    112 1 +2 /3 + 2 _ 1 + 2 / 3 + 72
    1+ 2 /3 - > 2 (1 + 2 / 3) - 2 11 + 4 / 3
    ( 1 + 2 3 + 2 ) (11 - 4 / 3 )
    73

    113 $ 3 - 723 - 28
    Put 31 = x, 21 = y, and take 6 the L . C . M . of the denominators 2 and 3, then
    i amb tr*y + me%y® + x 'y8 + ry* + y
    x - Y aº - º .

    therefore
    _ 38 + 31 24 + 3 :21 + 372 + 3. 28 + 21
    31 _ 23 31 - 23
    = 399 +372 + 6 + 29/648 + 4 :33 + 4 / 2.

    114 3 + 127 . Here the result will be the same as in thelast example
    if the signs of the even terms be changed. [See 5 .

    115 A surd cannot be partly rational ; that is , va cannot


    be equal to vo + c. Proved by squaring.
    116 The product of two unlike squares is irrational ;
    V7x3 = / 21, an irrational quantity .
    117 The sum or difference of two unlike surds cannot
    produce a single surd ; that is, va + b cannot be equal
    to vc. By squaring .
    118 Ifat vm = b + vn; then a = b and m = n .
    Theorems (115 ) to (118) are proved indirectly.

    119 If Va + vb = Vx + vy,
    then va - vb = Vx — Vy.
    By squaring and by (118 ).
    A
    52 ALGEBR .

    120 To express in two terms V7 + 2 v6.


    Let ✓7 + 2 / 6 = væ + vy;
    then x + y = 7 ............... by squaring and by (118 ),
    and 2 - y = V7 — (2 / 6)* = 149 – 24 = 5, by (119) ;
    : : x = 6 and y = 1.
    Result − 6 + 1.

    General formula for the same

    121 VatVb = vi(a + Va? – 6)EV }(a – Va? — b).


    Observe that no simplification is effected unless a: - 6 is a
    perfect square.

    122 To simplify Va + vb.


    Assume Va + vb = x + vy.
    Let c = a _ b.
    Then x must be found by trial from the cubic equation
    4x8 — 3cx = a ,
    and y = 22 — C .
    No simplification is effected unless a² — is a perfect cube.

    Ex. 1 : V7 + 5 / 2 = x + vy.
    c = 149 – 50 = - 1.
    4x + 3x = 7 ; . . x = 1, y = 2 .
    Result 1 + 2 .

    V913 - 11 / 2 = ( x + . y, two different surds.


    Cubing , 9 / 3 - 11 / 2 = x / x + 3. Vy + 3y x + y vy ;
    .: 913 = (x + 3y) Vx . (118 )
    11 / 2 = ( 3x + y ) vys ?
    . : x = 3 and y = 2.
    BINOMIAL THEOREM .

    123 To simplify ✓ (12 + 4 / 3 + 4 .75 + 2./ 15).


    Assome ✓ (12 + 4 / 3 + 4 75 + 2 / 15) = væ + vy+ % .
    Square, and equate corresponding surds.
    Result 3 + 4 + 75 .

    124 To express A + B in the form of two surds, where A


    and B are one or both quadratic surds and n is odd. Take 9
    such that 9 (AP - Bº) may be a perfect nth power , say p ” , by
    (361). Take s and t the nearest integers to " q (A + B )?
    and "Vq (4 - B )”, then
    VA + B = {V8+* + 2p + Vs+ t= 2p}.
    EXAMPLE : To reduce 893+ 109 / 2 .
    Here A = 89 / 3 , B = 109 / 2 ,
    A - B* = 1 ; i : p = l and q = 1.
    Vq ( A + B )' = 9 + f ) f being a proper fraction ;
    q (A - B ) = 1 - f1 ' i 8 = 9, t = l.
    Result (V9+ 1 + 2 + V9 + 1 – 2) = 13 + 72.

    BINOMIAL THEOREM.

    125 (a + b)" =
    a* +na*-16+ n (n2+! 1) - + (n - 1)(n
    3 !
    — 2)an-s787 & c.
    126 General or (r + 1)th term ,
    m ( n - 1 ) ( n - 2) ... ( - x + 1 ) . -PIP
    r!

    127 or in ")!man-to
    if n be a positive integer .
    If b be negative, the signs of the even terms will be changed.
    54 ALGEBRA

    If u be negative the expansion reduces to


    128 (a + b )-* =
    a -* - na - -167 " (n2 +! 1) a-n-262 _ n (n + 1) (n + 2) q -*-86 + & c.
    129 General term ,
    Hon (n + 1)(n + 2)...(n + r - 1)a---rb". [See 98.
    Euler 's proof. — Let the expansion of (1 + x )", as in (125),
    be called f (n ). Then it may be proved by Induction that the
    equation f (m )xf( n ) = f (m + n ) ... ... ... ... ... ... ( 1)
    is true when m and n are integers, and therefore universally
    true ; because the form of an algebraical product is not altered
    by changing the letters involved into fractional or negative
    quantities. Hence
    f(m + n + p + & c.) = f (m )Xf(n ) xf(p), & c.
    Put m = n = p = & c. to k terms, each equal , and the
    n

    theorem is proved for a fractional index.


    Again , put — n for m in (1) ; thus, whatever n may be,
    : f ( - n ) xf (n ) = f (0 ) = 1 ,
    which proves the theorem for a negative index.
    130 For the greatest term in the expansion of (a + b)", take
    (n + 1 ) b- (n - 1) b
    yo = the integral part of or
    O atb a - 6°
    according as n is positive or negative.
    But if b be greater than a , and n negative or fractional,
    the terms increase without limit.

    EXAMPLES
    Required the 40th terr

    Here r = 39 ; a = 1; b = - 2,0; n = 42.


    By ( 127), the term will be
    1. 21.40
    2 ( 20)" = - 42.4 . 3 (240)*
    ( 3 ) by ( 96 ).
    BINOMIAL THEOREM .
    - - - -- - -

    Required the 31st term of (a - x ) -4.


    Here r = 30 ; b = - 2 ; n = - 4 .
    By (129), the term is
    50 4 .. 55 ..66 ... 30 . 31 . 32 . 33
    ( - 1 )154 80 _ 31 . 32 . 33 2.80
    a -34 ( - x ) 50 = ! by (98).
    1 . 2 . 3 ... 30 1. 2 .3 23

    131 Required thegreatest term intheexpansion of difqja when x = }}. ni

    a = (1 + x) ". Here n = 6, a = 1, b = x in the formula


    (n - 1 ) b _ 5x11 _ 921 .
    · a -6 1 - it
    therefore r = 23, by (130), and the greatest term
    . 2 .. 37 ...
    = ( - 1)2 51 .6 ... 27 14 ) " = _ 24 .1 .252 .. 26
    23 (( 17) -
    .27 (14 ) .
    3 . 4 (17) . .

    132 Find the first negative term in the expansion of (2a + 36) " .
    Wemust take r the first integer which makes n — + 1 negative ; there
    fore r > n + 1 = 1 + 1 = 63 , therefore r = 7 . The term will be
    \?. ** . 1 . 73!: . (- ) (2a )-+ (36 )" by (126 )
    _ _ 17 . 14 . 11 . 8 5 . 2 . 1 27
    7! (2x)

    133 Required the coefficient of 2004 in the expansion

    = (1+3 + *x ){1+2(89)+ 2 3(19) =......


    ...+33(85)*+ 34( )*+35(*)*+ &.....},
    the three terms last written being those which produce *** after maltiplying
    by the factor ( 1 + 3x + ) ; for we have

    * **33(**)* + 32x34(*)*+ 1x35( 2)",


    giving for the coefficient of 2% in the result
    297 ( )*+ 102(3)*+ 35( 3)* = 306( 2)*
    The coefficient of 2 " will in like manner bc In ( 3)" -?.
    A
    56 ALGEBR .

    134 To write the coefficient of 2,3m + 1 in the expansion of


    The general term is
    (2n + 1 ) ! - 22 (2n - r + 1) 1 (2n + 1 ) ! x - 4r + ?
    (2n + 1 - r ) ! r ! " car (2n ++ 11 --
    (2n r) ! r !"
    Equate 4n - 4r + 2 to 3m + 1, thus
    4n - 3m + 1
    p =
    4
    Substitute this value ofr in the general term ; therequired coefficient becomes
    (2n + 1 ) !
    [+ (4n + 3m + 3) ]! [ (4n —3m + 1)] !
    An - 3m + 1 is an
    The value of r shows that there is no term in 28m +1 unless PN
    integer.

    135 An approximate value of ( 1 + x )" , when x is small, is


    1 + nx , by (125 ), neglecting æ * and higher powers of a .

    136 Ex. - An approximation to 999 by Bin . Th . (125) is obtained from


    the first two or three terms of the expansion of
    (1000 – 1)* = 10 – 5 . 1000 -1 = 10 - gov = 93: 7 nearly.

    MULTINOMIAL THEOREM .

    The generalterm in the expansion of (a + bx + cx* + & c.)" is


    137 n (n - 1) (n — 2) ... (p + 1) apb? c*de...x9+2r+38+...,
    q ! r ! 8 ! ...
    where p + a + r + 8 + & c. = n ,
    and the number of terms p , q , 1 , & c . corresponds to the
    number of terms in the given multinomial.
    p is integral, fractional, or negative, according as n is one
    or the other.
    If n be an integer, ( 137) may be written
    n ! ja ” b ? c” de ... x9 + 2r + 88.
    138 - p! q !r ! s!
    [ Deduced from the Bin . Theor.
    MULTINOMIAL THEOREM . 57

    Ex. 1. - To write the coefficient of a®bo in the expansion of (a + b + c + d ) .


    Here put n = 10, x = 1, p = 3, q = 1, r = 5, s = 0 in (138).
    10 !
    Result Tri = 7 . 8 . 9 . 10

    Ex. 2. — To obtain the coefficient of 28in the expansion of


    (1 – 2x + 3x — 4« )*.
    Here, comparing with (137), we have a = 1, b = - 2, c = 3, d = - 4,
    p + q + r + s = 4,
    q + 2r + 38 = 8,
    ilo i 2
    0 2012
    1 2 1

    The numbers 1, 0, 1, 2 are particular values of p, q, r, s respectively,


    which satisfy the two equations given above.
    0 , 2, 0 , 2 are another set of values which also satisfy those equations;
    and the four rows of numbers constitute all the solutions. In forming these
    rows always try the bigbest possible numbers on the right first .
    Now substitute each set of values of p , q, r, s in formula (138 ) succes
    sively, as under :
    4 1'(-2)°3'(–4) = 576
    2191"(- 2)*3°(+4)* = 384
    2) '3 ' ( - 4 )' = 864
    2)° 3* ( - 4 )' = 81
    Result 1905

    Ex. 3. - Required the coefficient of 2 * in (1 + 2x — 422 — 2:29)


    Here a = 1, b = 2, c = - 4, d = - 2;n = - 1 ; and the two equations are
    p +q + r + s= - ,
    9 + 2r + 38 = 4,
    10: 1
    0 2 0
    2101
    41010
    58 ALGEBRA.

    Employing formula ( 137), the remainder of the work stands as follows:


    (- ) (- 3)1-121(+4)*(+2) = = 3
    (- ) (- 3) 1-12 (–4) (–2)° = 6
    H ( - )( - ;) (- )1- (-4)'(-2) = 15
    (- ) (- )(- )(- )1- 2"( 4)*(-2)*= 5
    Result 223
    139 The number of terms in the expansion of the multi
    nomial (a + b + ct to n terms)” is the same as the number of
    homogeneous products of n things of r dimensions. See (97)
    and (98 ).

    The greatest coefficient in the expansion of (a + b + c +


    to m terms) ", n being an integer, is
    140 n !
    ( g !) ” ( a + 1 ) (k) where qmtk = n .
    Proof . — By making the denominator in (138) as small as possible. The
    notation is explained in (96 ).

    LOGARITHMS.

    142 log, N = x signifies that a ' = N , or


    DEF. — The logarithm of a number is the power to which the
    base must be raised to produce that number.

    143 log, a = 1 , log 1 = 0 .


    144 log MN = log M + log N .
    = log M - log N .

    log (M )" = n log M .


    logVM = 1 log M . [ 142
    EXPONENTIAL THEOREM . 59

    145 logca
    logo a =_ log .b

    That is — The logarithm of a number to any base is equal to


    the logarithm of the number divided by the logarithm of the
    base , the two last named logarithms being taken to any the
    same base at pleasure .
    Proof. — Let log,a = x and log ,b = y ; then a = cm, b = c" . Eliminate c.
    crat = b ; a = 6", that is, log,a = O
    1
    146 log,a = log b Put c = a in (145).
    147 log10 N = logg . N
    lo . 10 by (145 ).

    148 a = .43429448 ...


    log, 10
    is called the modulus of the common system of logarithms;
    that is , the factor which will convert logarithms of numbers
    calculated to the base e into the corresponding logarithms to
    the base 10 . See (154).

    EXPONENTIAL THEOREM .

    149
    « = 1 + cv + + * + &c.,
    where c = (a – 1 ) – } (a – 1) + } (a − 1) — & c.
    PROOF. q * = { 1 + (a - 1 ) } . Expand this by Binomial Theorem , and
    collect the coefficients of æ ; thus c is obtained . Assume C2, C3, & c., as the
    coefficients of the succeeding powers of x, and with this assumption write
    ont the expansions of af, a ", and a + y. Form the product of the first two
    series, which product must be equivalent to the third . Therefore equate
    the coefficient of x in this product with that in the expansion of ax +y. In
    the identity so obtained , equate the coefficients of the successive powers of
    y to determine Cg, Cs, & c .
    A
    EBR
    60 ALG

    Let e be that value of a which makes c = 1 , then

    150 € = 1+ 2 + * + * + &c.
    151
    v = 1 + 1 + 3 + 37 + &c.
    = 2.718281828 ... . .. [See (295).
    Proof. — By making x = lin (150 ).
    - - - - - -

    152 Bymaking x = 1 in (149) and x = c in (150), we obtain


    a = e'; that is, c = log , a . Therefore by (149)
    154 logea = (a – 1) — ; (a – 1) + } (a - 1): — & c.

    155 log (1+ x) = x - + -+* + &c.


    156 log (1- x) = -x - - - - &c. [154
    157 · log 1 = 2{-+ + *+&c.}.

    Putm71
    m + 1 for x in (157); thus,
    sm - 1 , lim - 113 , 1m - 15 ,
    158 log m
    Im + 1 \m + 1 5 m + 1!
    - - - - -- -- -- - - - - -

    Putty for x in (157); thus,


    159 log (n + 1) –log n
    = 2 2n + 1 + :3 (2n + 1)* + 5 (2n + 1)* + &c.} .
    - - - -
    CONTINUED FRACTIONS. 61

    CONTINUED FRACTIONS AND CONVERGENTS .


    314159
    160 To find convergents to 3. 14159o =_ 100000
    Proceed as
    in the rule for H . C . F .
    7 100000 | 314159 3 The continued fraction is
    99113 300000 3 + 1
    887 14159 7 + 1
    854 887 15 + & c .
    33 5289 or , as it is more conve
    29 4435 niently written ,
    854 25 21
    3 + 7 1+
    66 - & c.
    " 7 + 15 +
    194
    165
    29
    28

    The convergents are formed as follows :


    3 7 15 1 25 1 7 4
    3 22 333 355 9208 9563 76149 314159
    1 ' 7 ' 106 ' 113' 2931' 3044' 24239' 100000*
    161 RULE . — Write the quotients in a row , and the first two
    convergents at sight (in the example 3 and 3 + * ) . Multiply
    the numerator of any convergent by the next quotient, and
    add the previous numerator. The result is the numerator of
    the next convergent. Proceed in tbe same way to determine
    the denominator. The last convergent should be the original
    fraction in its lowest terms.

    162 Formula for forming the convergents.


    If Pn-?, Pn -1, Pn are any consecutive convergents , and
    In - In - 1 In
    An - 2, 2n - 1, an the corresponding quotients ; then
    Pn = unP -1 + Pn-2 , In = un9n - 1 + 4n -2.
    62 ALGEBRA.

    The nth convergent is therefore


    _ anPn - i + Pn - 2
    an9n -1 + In -2 "
    The true value of the continued fraction will be expressed
    by
    163 a'n Pn -1t pr - 2
    F =
    anqn - it qn - 2
    in which an is the complete quotient or value of the continued
    fraction commencing with an.
    164 Pu9n-1 - Pn-19n = + 1 alternately, by (162).
    The convergents are alternately greater and less than the
    original fraction, and are always in their lowest terms.

    165 The difference between F , and the true value of the


    continued fraction is
    1
    - and >
    In qn + 1 In ( On + an+ 1)
    and this difference therefore diminishes as n increases.
    Proof. — By taking the difference . Pn - a Pn +1 + Pn ( 163)
    In a'qn +1 + qn
    Also F is nearer the true value than any other fraction
    with a less denominator .
    - -- - -
    166 F , Fm+ 1 is greater or less than F ? according as Fn is
    greater or less than Fn+1:

    General Theory of Continued Fractions.


    167 First class of continued Second class of continued
    fraction . fraction .
    V -= 6 bz bz
    ait a., + Ag + & c . 1 9 , - , – Az - & c .
    an, bi, & c. are taken as positive quantities.
    CONTINUED FRACTIONS. 63

    , & c . are termed components of the continued frac


    ai da
    tion. If the components be infinite in number, the continued
    fraction is said to be infinite.
    Let the successive convergents be denoted by
    pi be . P2 _ bi bz . P3 = bı by _ Os ; and so on.
    9 a 22 dit a , 93 at 0 , + dz ’
    - - - - - - - --- -- - - - -

    168 The law of formation of the convergents is


    For F , For V ,
    s Pr = a „Pn-1tb» Pn-2 i Pu = anPn-1 - b » Pn-2
    = Qn9n -it bn 9n- 2 ( qn = an9n -1 - 6n9n-2
    [Proved by Induction.

    The relation between the successive differences of the


    convergents is, by (168),
    169 P9n +#1–
    1 Bu
    In = 7543290-1(9:
    9n + 1 lan - 9n -:).
    -1 ?
    Take the – sign for F , and the + for V .
    - - -- - - - -

    170 Pn9n-1 - Pn -19n = ( - 1)"-16,62b3 ... bm. (168)

    171 The odd convergents for F , P2,


    91
    P3,
    93
    & c., continually
    decrease , and the even convergents , P2, Pt , & c., continually
    qa' qa '
    increase. (167)
    Every odd convergent is greater, and every even con
    vergent is less , than all following convergents. ( 169)

    172 DEF. - If the difference between consecutive conver


    gents diminishes without limit , the infinite continued fraction
    is said to be definite . If the same difference tends to a fixed
    value greater than zero , the infinite continued fraction is in
    definite ; the odd convergents tending to one value, and the
    even convergents to another.
    64 ALGEBRA

    173 Fis definite if the ratio of every quotient to the next


    component is greater than a fixed quantity .
    Proof. — Apply ( 169) successively .
    174 F is incommensurable when the components are all
    proper fractions and infinite in number.
    Proof. — Indirectly , and by ( 168).
    175 If a be never less than 1 + 1 , the convergents of V are
    all positive proper fractions, increasing in magnitude, Pn and
    In also increasing with 1 . By ( 167) and ( 168).
    176 If, in this case, V be infinite, it is also definite, being
    = l , if a always = b + 1 while b is less than 1 , ( 175 ) ; and
    being less than 1 , if a is ever greater than 6 + 1 . By ( 190).
    170 V is incommensurable when it is less than 1 , and the
    components are all proper fractions and infinite in number.
    180 If in the continued fraction V ( 167), we have an = bn + 1
    always ; then , by (168),
    Pn = b2 + 1,62 + 6 ,62b3 + ... to n terms, and qn = Pn + 1.
    - -- -
    181 If, iv the continued fraction F, un and on are constant
    and equal, say, to a and b respectively ; then P, and an are
    respectively equal to the coefficients of a " - ? in the expansions
    of da+ bx
    1 - ax - 622 l - ax - bx2°
    Proof. - P and qn are the nth terms of two recurring series. See (168 )
    and (251).

    182 To convert a Series into a Continued Fraction .

    The series +U
    * +U*, + ... + ***
    is equal to a continued fraction V ( 167) , with n + 1 com
    ponents ; the first , second , and n + 1th components being
    u’ x 0 -1X
    uit ux' Untun -10
    [Proved by Induction .
    CONTINUED FRACTIONS. 65

    183 The series


    pan
    1 + 4 + .. + ... + VV , V2 .. . Un
    V . vvi VV, V2

    is equal to a continued fraction V ( 167 ) , with n + 1 compo


    nents , the first, second , and n + 1th components being
    VV
    1 VX Vn -1
    - ]
    ......unta
    [Proved
    [Proved by Induction .
    v ' v- tx

    184 The sign of a may be changed in either of the state


    ments in (182) or (183).
    185 Also , if any of these series are convergent and infinite ,
    the continued fractions become infinite .

    186 To find the value of a continued fraction with


    recurring quotients.
    Let the continued fraction be
    r = b bin where = bn + 1 _ bn + m
    ait . .. + Anty Antit . .. tantum ty
    so that there are m recurring quotients. Form the nth con
    vergent for x , and the mth for y . Then , by substituting the
    complete quotients anty for an, and an + m ty for an +m in ( 168) ,
    two equations are obtained of the forms
    Ay + B and y = Ey + F
    Cy + ] Gy + H '

    from which, by eliminating y , a quadratic equation for de


    termining x is obtained .

    187 If

    be a continued fraction , and

    к
    66 ALGEBRA

    the corresponding first n convergents ; then 9n-1, developed


    by (168),produces the continued fraction
    i bn bn- 1 _ 63 bz
    an + an-ıt An- 2 + ... + ā , + a ,
    the quotients being the samebut in reversed order.

    INDETERMINATE EQUATIONS.

    188 Given ax + by = 0
    free from fractions, and a , ß integral values of x and y wnich
    satisfy the equation , the complete integral solution is given by
    X = a - bt
    y = B + at
    where t is any integer.
    EXAMPLE . — Given 5x + 3y = 112.
    Then x = 20 , y = 4 are values ;
    : x = 20 — 37 )
    y = 4 + 5 ).
    The values of x and y may be exhibited as under :
    t = - 2 -- 1 0 1 2 3 4 5 6 7
    x = 26 23 20 17 14 11 8 5 2 - 1
    y = -6 - 1 4 9 14 19 24 29 34 39
    For solutions in positive integers t must lie between 20 = 6 and — ;
    that is, t must be 0 , 1 , 2 , 3 , 4 , 5 , or 6 , giving 7 positive integral solutions.

    189 If the equation be


    ax — by = c
    the solutions are given by
    x = a + bt
    y = B + at.
    İNDETERMINATE EQUATIONS. 67

    EXAMPLE : 4x – 3y = 19.
    Here x = 10 , y = 7 satisfy the equation ;
    i w = 10 + 3t
    y = 7 + 4t ) furnish all the solutions.
    The simultaneous values of t, x , and y will be as follows :
    t = -5 -4 -3 -2 -1 0 1 2 3
    x = - 5 -2 1 4 7 10 13 16 19
    y = - 13 - 9 -5 -1 3 7 11 15 19
    The number of positive integral solutions is infinite, and the least positive
    integral values of x and y are given by the limiting value of t, viz .,
    10
    t> - 10 and t >
    that is, t must be – 1, 0, 1, 2, 3, or greater.
    - -- - - - -- - - - - - -- -

    190 If two values, a and B , cannot readily be found by


    inspection , as, for example , in the equation
    17x + 13y = 14900 ,
    divide by the least coefficient, and equate the remaining frac
    tions to t, an integer ; thus

    y+* + ** = 1146+ 1 ....................... (1);


    :. 4 « — 2 = 13t.
    Repeat the process ; thus
    2 - 4 = 3t+
    i . t + 2 = 4u.
    Put n = 1,
    . t+ = 2,2 ,
    13t + 2 = 7 = a ;

    and y + x + t = 1146, by (1 ),
    y = 1146 — 7 — 2 = 1137 = 1 .
    The general solution will be
    a = 7 – 13t,
    y = 1137 + 177,
    Or, changing the sign of t for convenience,
    a = 7 + 13t,
    y = 1137 – 177.
    68 ALGEBR .
    A

    Here the number of solutions in positive integers is equal to the number of

    integers lying between and 111:37 7 ;

    - and 66}}; that is,67.


    191 Otherwise . — Two values of x and y may be found in
    the following manner :
    Find the nearest converging fraction to [By ( 160) .

    This is By (164)we have


    17 x 3 – 13 x 4 = - 1.
    Multiply by 14900, and change the signs;
    i 17 ( - 44700) + 13 (59600) = 14900 ;
    which shews that we may takee Sa = - 44700
    B = 59600
    and the general solution may be written
    x = - 44700 + 131,
    y = 59600 – 170.
    This method has the disadvantage of producing high values of a and B .

    192 The values of x and y , in positive integers, which


    satisfy the equation ax + by = c, form two Arithmetic Pro
    gressions, of which b and a are respectively the common
    differences. See examples (188) and ( 189).
    193 Abbreviation of the method in ( 169).
    EXAMPLE : 11x – 18y = 63.
    Put x = 9z, and divide by 9 ; then proceed as before.

    194 To obtain integral solutions of ax + by + c % = d .


    Write the equation thus
    ax + by = d - cz.
    Put successive integers for z ,and solve for x, y in each case.
    RÉDUCTION OF A QUADRATIO SÜRD. 69

    TO REDUCE A QUADRATIC SURD TO A


    CONTINUED FRACTION .

    195 EXAMPLE: MPLE

    ✓29 = 5 + 29 - 5 = 5 + 7294 + 5°
    er
    ✓29 + 5
    co
    = 2+ 2 +
    ✓29 + 3 '
    V29 +3 = 1 + V29 –2 = 1 + 2619 +
    ✓29 + 2 = - 3 4
    1 +
    5 * 129 + 3
    +

    Ter

    / 29 + 3 29 - 5
    = 2 + = 2 + 29 + 5°
    4
    ✓29 + 5 = 10 + v29 – 5 = 10 + -
    ✓29 + 5 :
    The quotients 5 , 2 , 1, 1 , 2 , 10 are the greatest integers
    contained in the quantities in the first column. The quotients
    now recur , and the surd ✓29 is equivalent to the continued
    fraction
    1 1 1 1 1 1 1 1 1
    5 + 2 + 1 + 1 + 2 + 10 + 2 + 1 + 1 + 2 + & c.
    The convergents to ✓29, formed as in (160), will be
    5 11 16 27 70 727 1524 2251 3775 9801
    1 ' 2 ' 3 ' 5 ' 13' 135' 283 ' 418 ' 701 ' 1820*
    196 Note that the last quotient 10 is the greatest and twice
    the first, that the second is the first of the recurring ones, and
    that the recurring quotients , excluding the last, consist of
    pairs of equal terms, quotients equi-distant from the first and
    last being equal. These properties are universal. (See 204
    - 210).

    To form high convergents rapidly .


    197 Suppose m the number of recurring quotients, or any
    ALGEBRA.

    multiple of that number, and let the mth convergent to v Q be


    represented by Fm ; then the 2mh convergent is given by the

    formula F.. = } {F + 192


    } by (203) and (210).
    198 For example, in approximating to ✓29 above,there are five recorring
    quotients. Take m = 2 x 5 = 10 ; therefore, by
    s 29 )

    9801 , the 16th convergent.


    Fje = 1820
    Therefore E = ( 9801 + 29 y 1820 ? _ 192119201
    = 1820 + 29 9801 ) = 35675640
    the 20th convergent to V29 ; and the labour of calculating the intervening
    convergents is saved .

    GENERAL THEORY.

    199 The process of (174 ) may be exhibited as follows :


    VQ + C = a + 12
    vQ + c,

    vente = ust vetes

    n + 1

    Fen = a.tv
    1 1 1
    200 Then
    a mi
    VQ == at
    VQ nt+ a
    + 12 azt
    + in + t& c .
    0 , fo

    The quotients 17,19,Az, & c. are the integral parts of the frac
    tions on the left .
    REDUCTION OF A QUADRATIC SURD. 71

    201 The equations connecting the remaining quantities are


    c = 0 r = 1
    C, = ar - G = 2- 0

    Cz = a .r, - C,
    ... ... ...
    Co = An -1°m-1 - CM-1
    27 - 1 - 1
    rn = &11=- 1
    The nth convergent to VQ will be
    202 n - AnPn- 1 + Pn - 2 [ By Induction .
    In An In – 1 + In- 2
    The true value of VQ is what this becomes when we
    substitute for an the complete quotient Veren, of which an
    is only the integral part. This gives
    Q + cn) Pn - itronPn - 2
    203 (vQ + en) In -1 + rm9n-2

    By the relations (199) to (203) the following theorems are


    demonstrated :
    204 All the quantities a , r , and c are positive integers .
    205 The greatest c is c2,and cn = az.
    206 No a or ^ can be greater than 2az.
    207 If in = 1, then cn = az.
    208 For all values of n greater than 1, a - cn is < ? .
    209 The number of quotients cannot be greater than 2a .
    The last quotient is 2a1, and after that the terms repeat.
    The first complete quotient that is repeated is V TC , and
    02, , C , commence each cycle of repeated terms.
    ALGEBRA.
    - - - - -

    210 Let am , Ime Co be the last terms of the first cycle ; then
    Am - 1, ? m -1, Cm - 1 are respectively equal to (12, 1º2, C2; am - 2, 1 -2,
    Cm - 2 are equal to 113, 143, C3, and so on . [ By ( 187)

    EQUATIONS.

    Special Cases in the Solution of Simultaneous Equations.


    211 First, with two unknown quantities.
    a & + by = ct2 Giba -- cıbı 2 – Can - cây
    0.x2 + b2y = coſ " abe — ambi' y bya, — b ,ai
    If the denominators vanish , we have
    al =
    ? , and x = 0, y = 0 ;
    A2 b,
    unless at the same time the numerators vanish , for then

    a, b2ce '
    and the equations are not independent, one being produced by
    multiplying the other by some constant.
    212 Next, with three unknown quantities. See (60) for
    the equations.
    If d ,12,dz all vanish, divide each equation by z,and we
    have three equations for finding the two ratios and % , two
    only of which equations are necessary, any one being dedu
    cible from the other two if the three be consistent.

    213 To solve simultaneous equations by Indeterminate


    COU

    Multipliers.
    Ex. — Take the equations
    x + 2y + 3x + 4w = 27,
    3x + 5y + 7 % + 11 = 48 ,
    5x + 8y + 10z — 2w = 65,
    7x + 6y + 5z + 4w = 53.
    MISCELLANEOUS EQUATIONS.
    - - -

    Multiply the first by A , the second by B , the third by C,


    leaving one equation unmultiplied ; and then add the results.
    Thus (A + 3B + 50 + 7) x + (24 + 5B + 8C + 6 ) y
    + (3A + 7B + 10C + 5 ) z + (4A + B - 20 + 4 ) w
    = 27A + 48B + 650 + 53.
    To determine either of the unknowns, for instance a ,
    equate the coefficients of the other three separately to zero ,
    and from the three equations find A , B , C . T'hen
    X =
    27A + 48B + 650 + 53
    A + 3B + 50 + 7

    MISCELLANEOUS EQUATIONS AND SOLUTIONS.

    214 20 + 1 = 0 .
    Divide by x®, and throw into factors, by (2) or (3 ). See also
    (480 ).

    215 23 — 7x — 6 = 0 .
    x = - 1 is a root, by inspection ; therefore x + 1 is a factor .
    Divide by x + 1, and solve the resulting quadratic .

    216 "? + 16x = 455.


    ** + 16. = 455x = 65 x 7X ,
    **+65&P+(65) = 49.x2+65x7c+ ( 35 ), 12

    *+ 65 = 7x+ 652
    x = 7x ; i. X = 7.
    factors , if possible , such that one of them , minus the square
    of the other, equals the coefficient of x . See (483) for general
    solution of a cubic equation .
    ALGEBRA.

    217 x * — y* = 14560, x - y = 8 .
    Put x = x +v and y = % - v.
    Eliminate v, and obtain a cubic in %, which solve as in (216 ).
    218 125 — yº = 3093, X - Y = 3.
    Divide the first equation by the second , and subtract from
    the result the fourth power of x - y . Eliminate (ac? + y ?) , and
    obtain a quadratic in xy .

    219 On forming Symmetrical Expressions.


    Take, for example,the equation
    (y — c) (z — b) = a'.
    To form the remaining equations symmetrical with this,write
    the corresponding letters in vertical columns, observing the
    circular order in which a is followed by b , 6 by c, and c by a .
    So with x , y, and z . Thus the equations become
    (y — c) ( — b) = a ',
    (* — a ) (x — c) = b ,
    ( ~ — b ) (y - a ) = c .
    To solve these equations, substitute
    x = b + c + a , y = c + a + y', % = a + b + ';
    and,multiplying out, and eliminating y and z , we obtain
    – bc (b + c) - a (12 + (?)
    bc - ca - ab
    and therefore, by symmetry, the values of y and , by the
    rule just given .
    220 y' + x + yz = a ? ...... .............
    het + 2 ? + 2x = 6 .. . ... .... .. .. . .. .... (2 ) ,
    * + y' + xy = c ..................... (3 ) ;
    :: 3 (193 + * +xy)? = 21%cº + 2e*u* + 2a²b? — ~ * — * — c*... .. (4).
    IMAGINARY EXPRESSIONS. 75

    Now add ( 1 ), (2 ), and ( 3 ), and we obtain


    2 (x + y + z): – 3 (y2 + zx + xy) = a' + b% + 0 ...... (5).
    From (4 ) and (5 ), (x + y + z ) is obtained , and then (1), (2 ),
    and (3) are readily solved.

    221 xi - yz = a? ................. ...... (1),


    yº — zx = b? ... . ................. (2),
    me — xy = c ..... ..... ...........
    Multiply (2) by (3), and subtract the square of(1).
    Result « (3xyz —28— 48 —z") = b?e?— a ,
    . butone at
    2

    _00 = *v*_ co = .........(4).


    Obtain 1? by proportion as a fraction with numerator
    = * — y2 = a'.
    222 = cy + bz .. ... ..
    = ax + cx ........ ..
    % = bx + ay . ... .. .. (3).
    Eliminate a between (2) and (3 ), and substitute the value of
    æ from equation (1).
    Result

    IMAGINARY EXPRESSIONS.

    223 The following are conventions:


    That / ( - a ) is equivalent to a / ( - 1) ; that a 7 ( - 1 )
    vanishes when a vanishes ; that the symbol a 7 ( - 1) is sub
    ject to the ordinary rules of Algebra. v ( - 1) is denoted
    by i.
    ALGEBRA.

    224 If a + iß = ytid ; then a = y and ß = d.


    225 a + iß and a -- iß are conjugate expressions; their pro
    duct = a' + ß ?.
    226 The sum and product of two conjugate expressions are
    both real, but their difference is imaginary.

    227 The modulus is + ve? + ß ".


    228 If the modulus vanishes, a and ß must vanish.
    229 If two imaginary expressions are equal, their moduli
    are equal, by (224).
    230 The modulus of the product of two imaginary expres
    pres IV

    sions is equal to the product of their moduli.


    231 Also the modulus of the quotient is equal to the
    quotient of their moduli.

    METHOD OF INDETERMINATE COEFFICIENTS.

    232 If A + Be + Cx2 + ... = A' + B'r + C'x' + ... be an equa


    tion which holds for all values of x , the coefficients A , B , & c .
    not involving x , then A = A ', B = B ', C = C ', & c . ; that is,
    the coefficients of like powers of æ must be equal. Proved by
    putting x = 0 , and dividing by x alternately. See (234) for
    an example .

    233 METHOD OF PROOF BY INDUCTION.

    Ex. — To prove that


    1 + 2 + 3 + ... + n ? =_ "n (12 + 1) ( 2n + 1)

    Assume 1 + 2° + 3² + ... tn = n (n + 1 ) ( 2n + 1)
    PARTIAL FRACTIONS.

    :: 1 + 2° + 32 + ... +nº+ (n + 1)' -= " (n + 1)6 (2n + 1) + (n + 1)'


    n (n + 1) (2n + 1) + 6 (n + 1)? _ (n + 1) {n (2n + 1) + 6 (n + 1)}
    6 6

    = (n+1)(n+2)(21+ 3) ="'(u + 1)(2n + 1),


    6

    where n ' is written for n + 1 ;


    : 1 + 2 + 3 + ... tn"? = " (n + 1) (2n + 1)
    It is thus proved that if the formula be true for n it is also true for n + 1..
    Bat the formnla is true when n = 2 or 3 , as may be shewn by actual
    trial ; therefore it is true when n = 4 ; therefore also when n = 5 , and so on ;
    therefore aniversally true.

    234 Ex. — The sametheorem proved by the method of In


    determinate coefficients.
    Assume
    1 + 2° + 32 + ... + n ? = A + Bn + Cn? + Duis + & c. ;
    :: 1 + 2 + 3 + ... + n° + (n + 1) = A + B (n + 1) + ( (n + 1) + D (n + 1 )* + & c. ;
    therefore, by subtraction ,
    m2 + 2n + 1 = B + C (2n + 1) + D (3n + 31 + 1),
    writing no terms in this equation which contain higher powers of n than the
    bighest which occurs on the left-haud side, for the coefficients of such terms
    may be shewn to be separately equal to zero.
    Now equate the coefficients of like powers of n ; thus
    3D = 1, · D = ;
    2C + 3D = 2, i 0 = 5 , and A = 0;
    B + C + D = l, i
    therefore the sum of the series is equal to
    ime
    + - n ( n + 1 ) (2n + 1 )

    PARTIAL FRACTIONS.
    In the resolution of a fraction into partial fractions four
    cases present themselves, which are illustrated in the follow
    ing examples.
    78 ALG
    EBR .
    A

    235 First . - When there are no repeated factors in the de


    nominator of the given fraction .
    3x - 2
    Ex. - To resolv
    resolve (n– - 1)( 2016 - 3) into partial fractions.
    B 0
    Assume - 3x - 2
    (x - 1 ) (x - 2 ) (x - 3 ) 2 -1 2- 2 'x -3'
    :: 3x — 2 = A (2 — 2) (« — 3 ) + B (2 – 3) (2 - 1) + (x - 1) ( - 2 ).
    Since A, B , and do not contain x, and this equation is true for all values
    of x , put x = l ; then
    3 - 2 = A (1 – 2) (1 – 3), from which A =
    Similarly, if a be put = 2 ,we have
    6 – 2 = B (2 - 3) (2 - 1) ; B = -4 ;
    and, putting x = 3 ,
    9 – 2 = 0 (3 – 1) (3 – 2); 0=?
    3x - 2
    Hence
    (2 - 1)(2 -2)(x- 3) = 2(2 -1) -2 =2 + 2(2- 3):
    236 Secondly . — When there is a repeated factor.
    7203 — 10x + 6x
    Ex.— Resolve into partial fractions
    18 (x - 1) (x + 2)
    ume

    Assume you that a 10 -130+( teiten


    These forms are necessary and sufficient. Multiplying up, we have
    7.28 – 10x + 6x = A (x + 2) + B (x - 1) (x + 2) + 0 (x - 1)* (x + 2).+ D (c - 1)
    ..... (1 ).
    Make x = 1 ; i 7 – 10 + 6 = A (1 + 2) ; . . A = 1.
    Substitute this valae of A in (1 ) ; thus
    7x3 – 10x + 5.0 2 = B (x - 1)(x + 2) + (x - 1)*(x + 2) + D (x - 1)!
    Divide by 2 - 1 ; thus
    7x* — 3x + 2 = B (x + 2 ) + C (2 - 1) (x + 2 ) + D (x - 1 ) .........
    Make x = 1 again, 7 – 3 + 2 = B (1 + 2) ; :. B = 2.
    Substitute this value of B in (2), and we have
    7x* — 5x – 2 = C (x - 1 ) (x + 2) + D (x - 1) .
    Divide by 2 - 1, 7x + 2 = ( (x + 2) + D (x - 1) .........
    Put x = l a third time, 7 + 2 = C (1 + 2) ; . 0 = 3.
    PARTIAL FRACTIONS. 79

    Lastly , make x = - 2 in (3 ) ,
    – 14 + 2 = D ( - 2 - 1 ) ; i D = 4.

    Result ( +13+ 0-13x + $1* * 2


    237 Thirdly.- When there is a quadratic factor of imaginary
    roots not repeated .
    ve ( 2
    Ex. - Resolvez + 1)( to + ]) into partial fractions.
    Here we must assume
    Ax + B , Cx + D
    (2® + 1) (2* + 2 + 1) = m + 1 + x4 + x + 1
    2 + 1 and 2 + x + 1 have no real factors, and are therefore retained as
    denominators. The requisite form of the numerators is seen by adding
    a CC
    together two simple fractions, such as x- + +
    b ' x + ď
    Multiplying up, we have the equation
    1 = (Ax + B) (x2 + 2 + 1) + (Cæ + D ) (x* + 1) .......
    Let 2* + 1 = 0 ; i = - 1.
    Substitute this value of a in ( 1 ) repeatedly ; thus
    1 = (Ax + B ) x = A * * + Bx = - A + Bx ;
    or Bx - A - 1 = 0 .
    Equate coefficients to zero ; i. B = 0,
    A = - 1.
    Again, let X* + x + 1 = 0 ;
    . X * = - 2 - 1.
    Substitute this value of x repeatedly in (1); thus
    l = (Cx + D ) ( - x ) = - Ca* – Dx = Cx + C – Dx ;
    or (C - D ) x + C - 1 = 0.
    Equate coefficients to zero; thus C = 1 ,
    D = 1.
    1 2 + 1
    Hence
    (x*+ 1)(* + +1) = x +2+ 1 2 + 1
    238 Fourthly.— When there is a repeated quadratic factor
    of imaginary roots.
    40x — 103
    " (x + 1)² (2x2 — 4x + 8avg into partial fractions.
    Ex. - Resolve
    )3
    ALGEBRA.

    Assume
    40v - 103 Ax + B , Cæ + D Ex + F
    (æ + 1)*(a*— 4.c + 8 ): * (x* — 4x + 8)3 + (x* — 4x + 8): * x* – 4x + 8

    40x — 103 = {(Ax + B ) + (Cx + D )(22 —4x + 8) + (Ex + F ) (z ? — 4x + 8) } (x + 1)


    + {G + H (2 + 1 ) } (x* — 4x + 8 ): ... ................... ( 1) .
    In the first place, to determine A and B , equate xº — 4x + 8 to zero ; thus
    2 * = 4x - 8 .
    Substitute this value of a repeatedly in ( 1) , as in the previous example ,
    until the first power of w alone remains. The resulting equation is
    40x – 103 = ( 17A + 6B) x – 48A – 7B.
    Equating coefficients, we obtain two equations
    17 A + 6B = 401 from which A = 2
    48 A + 7B = 103 ! B = 1.
    Next, to determine C and D , substitute these values of A and B in ( 1 ) ;
    the equation will then be divisible by x* — 4x + 8 . Divide,and the resulting
    equation is
    0 = 2x + 13 + {Cx + D + (Ex + F ) (w ? — 4x + 8) } (x + 1)
    + {G + H (x + 1) } (x* — 4.2 + 8 ) ........... ( 2) .
    Equate x2— 4x + 8 again to zero, and proceed exactly as before,when
    finding A and B .
    Next, to determine E and F , substitute the values of Cand D , last found
    in equation (2 ) ; divide, and proceed as before.
    Lastly , G and H are determined by equating x + 1 to zero successively ,
    as in Example 2 .

    CONVERGENCY AND DIVERGENCY OF SERIES.

    239 Let a , + az + Az + & c. be a series, and aig antı any two


    consecutive terms. The following tests of convergency may
    be applied . The series will converge , if , after any fixed term
    (i.) The terms decrease and are alternately positive and
    negative.
    (ii.) Or if (an
    n+1
    is always greater than some quantity
    greater than unity
    SERIES. 81

    (iii.) Or if an is never less than the corresponding ratio


    in a known converging series.
    (iv.) Or if (nan –n ) is always greater than some quan
    lanti

    tity greater than unity. [By 244 and iii.


    (v.) Or if (non
    lan + 1
    - n - 1) log n is always greater than
    some quantity greater than unity .
    240 The conditions of divergency are obviously the converse
    of rules (i.) to (v .).

    241 The series az + ana + aza? + & c . converges, if On+1 is


    - An

    always less than some quantity p, and a less than 5.


    [By 239 (ii.)
    242 To make the sum of the last series less than an assigned an

    quantity p , make a less than P2, k being the greatest co


    efficient.

    General Theorem .
    243 If • (a ) be positive for all positive integral values of x ,
    and continually diminish as a increases , and if m be any posi
    tive integer, then the two series
    $ (1 ) + $ (2 ) + $ ( 3) + + (4 ) + ......
    $ ( 1) + mø (m ) + m ++ (mº) + mºd (mº) + ......
    are either both convergent or divergent.
    244 Application of this theorem . To ascertain whether the
    series

    series + + + +.....
    is divergent or convergent when p is greater than unity.
    82 ALGEBRA,

    --- - - -

    Taking m = 2, the second series in (243) becomes


    2 .48
    1+ + + + &c......
    a geometrical progression which converges ; therefore the
    given series converges.

    245 The series of which is the general term is


    convergent if p be greater than unity, and divergent if p be
    not greater than unity . [By (243 ), (244).
    246 The series of which the general term is
    1
    nd (n ) 1*(n ) ...... \" (n) {1*+1(n )}" ’
    where 1 (n ) signifies log n , 1 ? (n ) signifies log {log (n )} , and
    so on , is convergent if p be greater than unity , and divergent
    if p be not greater than unity . [By Induction, and by (243).

    247 The series a , + + & c . is convergent if


    nanlog (n ) logº(n ) .. ... log" (n ) {log r+1 (n ) }'
    is always finite for a value of p greater than unity ; logº(n )
    here signifying log (log n ), and so on.
    [See Todhunter's Algebra, or Boole's Finite Differences.

    EXPANSION OF A FRACTION .

    4x -- 10x2
    248 A fractional expression such as 11.- 6x + 1172 _ 6173 may
    be expanded in ascending powers of æ in three different ways.
    First, by dividing the numerator by the denominator in
    the ordinary way, or by Synthetic Division , as shewn in (28) .
    Secondly , by the method of Indeterminate Coefficients
    (232).
    Thirdly , by Partial Fractions and the Binomial Theorem .
    SERIES.

    To expand by the method of Indeterminate Coefficients ,


    proceed as follows :
    4x - 10x2
    Assume i s = A + Bx + C . ? + Dx* + Ext + & c.
    1 - 6x + 1lx — 623
    . 42 — 10x* = A + Bæ + Cx² + Dx® + Ex* + Fx + ...
    - 6Ax — 6Bx’ – 6Cx3 – 6Dxt— 6Ex — ...
    + 11Ax + 11 Bxs + 11Cx* + 110x + ...
    - 6 Axi — 6Bx* — 6C — ...
    Equate coefficients of like powers of x, thus
    A = 0,
    B - 6A = 4, .. B = 4 ;
    0 - 6B + 11A = - 10 , i C = 14 ;
    D - 6C + 11B - 6A = 0, i. D = 40 ;
    E - 6D + 110 - 6B = 0, ii E = 110 ;
    F -6E + 11D , 6C = 0, i F = 304 ;
    - .. . ... ... .. . ... .. .

    The formation of the same coefficients by synthetic division is now


    exbibited, in order that the connexion between the two processes may be
    clearly seen .
    The division of 4x - 10x? by 1 - 6x + 112 – 6.23 is as follows:
    10 + 4 - 10
    + 6 24 + 84 + 240 + 660
    - 11 - 44 – 154 — 440 - 1210
    + 6 + 24 + 84 + 240 + 660
    10 + 4 + 14 + 40 + 110 + 304 + ............
    A B C D E F
    If we stop at the term 110x4, then the undivided remainder will be
    30tx — 9704 + 660.x?, and the complete result will be
    4x + 14x² + 40.28 + 110x* + 304.rº — 970x® + 660x7
    1 - 6x + 11.7 – 623 .
    249 Here the concluding fraction may be regarded as the
    sum to infinity after four terms of the series, just as the
    original expression is considered to be the sum to infinity of
    the whole series.

    250 If the general term be required, the method of ex


    pansion by partial fractions must be adopted . See (257),
    where the general term of the foregoing series is obtained .
    84 ALGEBRA .

    RECURRING SERIES .

    ao tax + ax' + azł + & c . is a recurring series if the co


    efficients are connected by the relation
    251 An = pian - 1 + p ,An - 2 + ... + Pmln - m .
    The Scale of Relation is
    252 1 - pix - p20 - . .. - PmX " .

    The sum of n terms of the series is equal to


    253 [ The first m terms
    - pix (first m - 1 terms + the last term )
    - pax (first m - 2 terms + the last 2 terms)
    - pzi ( first m - 3 terms + the last 3 terms)

    - Pm-1.909 –1 (first term + the last m - 1 terms)


    - Pm0." (the last m terms) ] = [ 1 -px - px? — . . - PmX" ).
    254 If the series converges , and the sum to infinity is re
    quired , omit all “ the last terms ” from the formula .

    255 EXAMPLE. — Required the Scale of Relation , the general


    term , and the apparent sum to infinity, of the series
    4x + 14.2? + 40.x3 + 110x4 + 304.r — 854x6 + ... .
    Observe that six arbitrary terms given are sufficient to determine a Scale
    of Relation of the form 1 - pc - qx – rx , involving three constants p , q , r,
    for, by (251), we can write three equations to determine these constants ;
    namely, 110 = 40p + 14q + 4r ) The solution gives
    304 = 110p + 40q + 14r . p = 6 , q = - 11, r = 6.
    854 = 304p + 110q + 40r)
    Hence the Scale of Relation is 1 - 6x + 11x – 6x'.
    The sum of the series without limit will be found from (254), by putting
    p = 6 , p = - 11, P , = 6 , m = 3 .
    The first three terms = 4x + 14x2 + 40.28
    - 6x the first two terms = – 24x — 84x®
    + 11x x the first term + 44.23
    4x – 10x
    RECURRING SERIES.

    4x — 10ac?
    . S = 1 - 6x + 11.22 — 628?
    the meaning of which is that, if this fraction be expanded in ascending
    powers of x , the first six terms will be those given in the question.

    256 To obtain more termsof the series,wemay use the Scale of Relation ;
    thus the 7th term will be
    (6 x 854 – 11 x 304 + 6 x 110) x = 2440x².
    257 To find the general term , S must be decomposed into
    partial fractions ; thus, by the method of (235),
    1 2 3

    By the Binomial Theorem ( 128),


    - = 1 + 3x + 3 x + ...... + 3 "w" ,
    1 - 3x .
    2
    -- = 2 + 2+x + 2922 + ...... + 2n+!x»,
    3 = - 3 – 3x – 3x? — . .... – 3x”.
    Hence the general term involving x" is
    (3 " + 2n +1 – 3 ) xr ".
    And by this formula we can write the “ last terms " required in (253), and
    so obtain the sum of any finite number of terms of the given series. Also,
    by the same formula we can calculate the successive terms at the beginning
    of the series. In the present case this mode will be more expeditious than
    that of employing the scale of Relation .

    258 If, in decomposing S into partial fractions for the sake


    of obtaining the general term , a quadratic factor with ima
    ginary roots should occur as a denominator, the samemethod
    must be pursued , for the imaginary quantities will disappear
    in the final result. In this case, however, it is more con
    venient to employ a general formula . Suppose the fraction
    which gives rise to the imaginary roots to be
    L + Mx _ L + Mæ
    a + bx + x2 + (p — x ) (q — « ) '
    p and q being the imaginary roots of a + bx + x? = 0 .
    Suppose - pP == at
    a + iß ,
    q = a - iß, where i = V - 1.
    86 · ALGEBRA.

    If, now , the above fraction be resolved into two partial


    fractions in the ordinary way , and if these fractions be ex
    panded separately by the Binomial Theorem , and that part of
    the general term furnished by these two expansions written
    out, still retaining p and q , and if the imaginary values of p
    and q be then substituted, it will be found that the factor will
    disappear, and that the result may be enunciated as follows.

    259 The coefficient of xn -1 in the expansion of


    L + Mx
    ata
    err (a ’ + B ) _ 2ax + x2
    will be
    L
    BC L & { na"-1B - C (n, 3) an -889 + C (n,5)a"-5 86 — ...}
    M
    teli \n-i {(n - 1)an-?B - C (n - 1,3) a"-4 BS
    + C (n - 1, 5 ) a "-683 — ... }.
    260 With the aid of the known expansion of sin no in
    Trigonometry, this formula for the nth term may be reduced to
    /(L + Ma)? + MⓇB . sin (16 - 6),
    B ? (a + )"
    in which 0 = tan = tan - 1 _ MB
    " L + Ma
    If n be not greater than 100 , sin (10 - 0 ) may be obtained
    from the tables correct to about six places of decimals , and
    accordingly the nth term of the expansion may be found with
    corresponding accuracy . As an example , the 100th term in
    the expansion of 5 —11
    2x + x2 is readily found by this method
    to be 41108241 4 20
    XV .

    To determine whether a given Series is recurring or not.


    261 If certain first terms only of the series be given , a scale
    of relation may be found which shall produce a recurring
    RECURRING SERIES. 87

    series whose first terms are those given . The method is


    exemplified in (255 ). The number of unknown coefficients
    P , 9, 1, & c. to be assumed for the scale of relation must be
    equal to half the number of the given terms of the series, if
    that number be even . If the number of given terms be odd ,
    it may be made even by prefixing zero for the first term of
    the series .
    262 Since this method may, however, produce zero values
    for one or more of the last coefficients in the scale of relation ,
    it may be advisable in practice to determine a scale from the
    first two terms of the series, and if that scale does not produce
    the following terms, we may try a scale determined from the
    first four terms, and so on until the true scale is arrived at.
    If an indefinite number of terms of the series be given ,
    we may find whether it is recurring or not by a rule of
    Lagrange's.
    263 Let the series be
    S = A + Bx + Ca? + Dx8 + & c .
    Divide unity by S as far as two terms of the quotient, which
    will be of the form p + qx, and write the remainder in the form
    S'x*, S' being another indefinite series of the same form as S .
    Next, divide S by S' as far as two terms of the quotient,
    and write the remainder in the form S "x . .
    Again , divide S' by S " , and proceed as before, and repeat
    this process until there is no remainder after one of the
    divisions. The series will then be proved to be a recurring
    series, and the order of the series, that is, the degree of the
    scale of relation , will be the same as the number of divisions
    which have been effected in the process.

    EXAMPLE. — To determine whether the series 1, 3, 6 , 10, 15 , 21, 28, 36 ,


    45 , ... is recurring or not.
    Introducing x , we may write
    $ = 1 + 3x + 6x° + 10x3 + 15x * + 2125+ 282 + 36x? + 4.525... .
    Then we shall have = 1 - 3x + ... with a remainder
    -3x2 + 8x8 + 152* + 24.0” + 35x® + & c.
    Therefore S' = 3 + 8x + 15x²+ 24x + 35x* + & c .,
    S _ 112
    = 379
    88 .
    ALG
    EBR
    A

    with a remainder - (x² + 3x + 6x* + 10x + & c. ... ).

    Therefore we may take S" = 1 + 3x + 6x? + 10x + & c.


    Last = 3 - x without any remainder.
    Consequently the series is a recurring series of the third order. It is, in
    fact, the expansion of
    1 - 3x + 3.22 —

    SUMMATION OF SERIES BY THE METHOD OF


    DIFFERENCES.

    264 RULE. — Form successive series of differences until a series


    of equal differences is obtained. Let a , b , c, d , & c . be the first
    terms of the several series ; then the nth term of the given
    series is
    265 4 +(n - 1) 6 (n - 1)(x - 2) (n - 1)(n - 2)(n - 3) -
    1.2 1 .2 .3
    The sum of n terms

    266 4 n (n -1)04
    =- nunuTou - 1)(
    t n(n 16 - 2) ..
    oct & c .
    1.2 1. 2. 3
    Proved by Induction.
    EXAMPLE : a ... 1 + 5 + 15 + 35 + 70 + 126 + ...
    ... 4 + 10 + 20 + 35 + 56 + ...
    ... 6 + 10 + 15 + 21 + ...
    d ... 4 + 5 + 6 + ...
    e ... 1 + 1 + ...
    The 100th term of the first series
    = 1 + 99.4 + 99. 98 6 + 99. 98 .97 99. 98. 97. 96
    1 .2 . 2 .. 33
    11 .2 *' 1 .2 . 3 .4
    The sum of 100 terms
    L 100 , 100 .99 100 . 99.98 € , 100 .99. 98. 97 , 100.99 .98.97 . 96
    1.2 1 .2 .3 T 1. 2 .43 . T 1. 2 . 3 . 4 .5
    FACTORIAL SERIES.

    267 To interpolate a term between two terms of a series by


    the method of differences .
    Ex. - Given log 71, log 72, log 73, log 74, it is required to find log 72:54.
    Form the series of differences from the given logarithms, as in (266),
    log 71 log 72 log 73 log 74
    a ... 1.85125831 .8573325 1.86332291 .8692317
    b ... .0060742 0059904 ·0059088
    C ... – .0000838 – .0000816
    d ... — .0000022 considered to vanish .
    Log 72:54 must be regarded as an interpolated term , the number of its
    place being 2:54.
    Therefore pat 2.5 $ for n in formula ( 265).
    Result log 72.54 = 1.8605777.

    DIRECT FACTORIAL SERIES.

    268 Ex.: 5.7.9 + 7.9 .11 + 9. 11 . 13 + 11. 13 .15 + ...


    d = common difference of factors,
    m = number of factors in each term ,
    n = number of terms,
    - 1

    a = first factor of first term – d .

    nth term = (a + nd)(a + n + 1d)......(a +n + m - 1d).


    269 To find the sum of n terms.
    RULE. — From the last term with the neict highest factor take
    the first term with the next lowest factor, and divideby (m + 1 ) d .
    Proof.- By Induction .
    Thus the sum of 4 terms of the above series will be, putting d = 2, m = 3,
    n = 4, a = 3, 11. 13 . 15 . 17 - 3 . 5 . 7 . 9
    (3 + 1) 2
    ed either by Induction , or by themethod of Indeterminate Coefficients .
    NO

    N
    ALGEBRA.

    INVERSE FACTORIAL SERIES.

    270 Ex.: 5.7.9 +7.9.11+9.11.13 +11.13.16+...


    Defining d , m , n , a as before, the
    1
    nth term = TAI
    ( a + nd) (a + m + 1 d ) ... ( a + m + m – 1 d)
    271 To find the sum of n terms. RULE . — From the first
    term wanting its last factor take the last term wanting its first
    factor , and divide by (m . — 1) d .
    Thus the sum of 4 terms ofthe above series will be,putting d = 2, m = 3,
    1 1

    n = 4, a = 3, 5 .7 13 . 15
    (3 - 1) 2
    Proof. - By Induction, or by decomposing theterms,as in the following
    example.

    272 Ex.: To som the same series by decomposing the terms into partial
    fractions. Take the general term in the simple form

    ( — 2 ) r (r + 2 )
    Resolve this into the three fractions
    1 _ _ 4r1 _+ 8 (7 1+ _2) hu
    8 (r — 2) by (235).
    Substitute 7, 9 , 11, & c . successively for r , and the given series has for
    its equivalent the three series

    8 ] § + 3 + 9 + i + 13 " ............. + 2n + 3 }
    11ļ _ 2 _ 2 - 2 - 2 _ _ 2_ - _ 2 }
    817 9 11 13 . .. 2n +32n + 5
    1

    + 1} 1 + 1 + 1 + ......+ 1 + 1 + 1 }
    + 8° 9 * 11 * 13 * . ** . 7 2n + 3 + 2n + 5 * 2n + 7 )
    and thesum of n terms is seen, by inspection, to be
    1 1 1 1 1 2 1 1 1 1
    8 7 5 7 2n + 5 ' 2n + 75 4 / 5 .7 (21 + 5 ) (2n + 7) )
    a result obtained at once by the rule in (271), taking - o for the first
    term , and - for the nth or last term .
    ( 2n + 3 ) ( 21 + 5 ) ( 2n + 7 )
    FACTORIAL SERIES.

    273 Analogous series may be reduced to the types in (268)


    and (270 ) , or else the termsmay be decomposed in the manner
    shewn in (272 ).
    10 .
    Ex. : 1 + 4 + -3 . 4?. 5 ta
    1. 2 .3 . 2 . 3. 4 + 4. 5 . 6 + ... ...
    has for its general term
    3n - 2 11 5 - 4 hv ( 235).
    n ( n + 1) ( n + 2 ) n ' n + 1 n 19
    + 2 by (235 ),
    and we may proceed as in (272) to find the sum of n terms.
    The method of (272) includes the method known as “ Summation by
    Subtraction,” but it has the advantage of being more general and easier of
    application to complex series.

    COMPOSITE FACTORIAL SERIES.

    274 If the two series


    5 .6 5 .6 . 7 . 8 5 .6 . 7 . 8
    (1 -2 )*° = 1 + 5x + 1,24 +t 1.2 .3Ⓡ203 ++ 1 .2 .3 .4- "2 * + .. .
    (1 ~ ~) -8 = 1 + 3x + 3 .4 ,2 + 3 4 .5 m3 1 3. 4 .5 .6 .X * t .. .
    07 1 .27 1. 2 . 3 " 71. 2 . 3 . 4."
    bemultiplied together, and the coefficient of 2 * in the product
    be equated to the coefficient of 2 * in the expansion of (1 ) - 8,
    we obtain as the result the sum of the composite series
    5 . 6 . 7 .8x1. 2 + 4 . 5 . 6 .7x2. 3 + 3 . 4 . 5 . 6 x 3 . 4
    + 2 . 3 . 4 .5x4 . 5 + 1 . 2 . 3 .4x5. 6 - 4 . 2 . 11 ! .
    7! 4 !

    275 Generally , if the given series be


    PQ + P ,Q2 + ... + Pn-1Qn-1 ............... (1),
    where Qx = q (r + 1) (1 + 2) ... (r + q - 1),
    and P , = (n - r ) (n — 1 + 1) ... (n - r + p - 1) ;
    the sum of n - 1 terms will be
    p!q! (n + p + 9 – 1) !
    (p + q + 1)! (n = 2)!
    92 ALGEBRA

    MISCELLANEOUS SERIES.

    276 Sum of the powers of the terms of an Arithmetical


    Progression.
    1 + 2 + 3 + ... + n = n (n + 1 )
    2

    1 + 22 + 3 + tn = n (n + 1) (2n + 1 )
    6

    1 + 2 + 3 + ... + nº = n ( n + 1)
    _ n (n + 1 ) ( 2n + 1 ) ( 3nº + 3n - 1)
    1 + 2 + 3 + ... + n * =
    30
    [ By the method of Indeterminate Coefficients (23t).
    A general formula for the sum of the pth powers of
    1 . 2 . 3 ... 1 , obtained in the same way is
    S = nº+ 1 + 1n " + Ain " - 1 + ... Ar -in ,

    where A1, A2, & c.,are determined by putting p = 1, 2, 3, & c.


    successively in the equation

    2 (p + 1)!
    (p + 2)! 1 (P )! re (p - 1) ( p - 1 ) 1 (p 1 ) ... (r - p + 1)

    277 q ” + (a + d )" + (a + 20)" + ... + (a + nd)"


    = (n + 1) a " + Sima" -id + S. C (m , 2 ) a " -2 dº
    + S C (m , 3 ) an-3,13 + & c .
    Proof. — By Binomial Theorem and (276 ).

    278 Summation of a series partly Arithmetical and


    partly Geometrical .
    ESAMPLE. — To find the sum of the series 1 + 3x + 5.2 ? + to n
    terms.
    MISCELLANEOUS SERIES. 93
    93
    Let s = 1 + 3x + 5x2+ 7a® + ... + (2n - 1) xn -),
    sx = x + 3x+ + 5x®+ ... + (2n — 3) « »-) + (2n - 1) x”,
    . : by subtraction ,
    8 ( 1 - x ) = 1 + 2x + 2x* + 2x3 + ... + 2xn -1 — (2n - 1) x"
    = 1 + 222 1=2*-
    - X
    -(2n – 1).x",
    e - 1 - (2n - 1) " 23 (1 - 21 -1)
    1-2 (1 - x )'

    279 A general formula for the sum of n terms of


    a + (a + d) r + (a + 2d ) + (a + 3d) m + &c.
    is S = a - (a + n - 1d) ? ” dr (1 - pon -1)
    1 - r (1 - 7 )
    Obtained as in ( 278 ).
    RULE. — Multiply by the ratio and subtract the resulting
    series.

    280 F2 = 1+x +x +x++...+2+++ im


    281
    = 1 + 2x + 3x ' + 423 + ...
    ... +non -17 (n + 1) 8" - nxn+ 1
    (1 — x )

    282 (n - 1)x + (n — 2) x2 + (n - 3)2 + ... + 2.xn-2 + xn -1


    _ (n - 1) x - nx + x4+1 By (253)
    (1 - x )?

    283 1 +a + ” (x2 -! 1) n (n - 1)(x - 2) + & e. = 2",


    3 !

    1-1+ 1(9,7") _" (n=}}{n —2)+ &c.= 0.


    By making 4 = b in (125) . errata
    X = 1
    94 ALGEBRA

    284 The series

    1-9 ;3 +(n– 4)(n–5) _(n—5)(976)(n—7)+...


    .. . + ( n - p - 1 ) (n - p — 2 ) ... n( 2r + 1 )

    CON Sum 0

    consists of " or " I terms, and the sum is given by


    if n be of the form 6m + 3,

    if n be of the form 6m + 1 ,

    if n be of the form 6m ,

    S= ? if n be of the form 6m + 2.
    Proof. – By ( 545), putting p = x + y, q = xy, and applying (546 ).

    285 The series n* — n (n - 1)"+ " (171)


    2 ! (n — 2)"
    - " \" 3, 2 - 2 )(n -3)* + & c....
    takes the values 0, n !, żn (n + 1) !
    according as r is < n , = n , or = n + 1.
    Proof. — By expanding (ex - 1)", in two ways : first, by the Exponential
    Theorem and Multinomial ; secondly, by the Bin . Th., and each term of
    the expansion by the Exponential. Equate the coefficients of x " in the two
    results.

    Other results are obtained by putting r = n + 2 , n + 3 , & c.


    The series (285) , when divided by p !, is , in fact, equal to
    the coefficient of x ' in the expansion of
    ( 32 33
    {2 + 27 + 3 7 +...}".
    POLYGONAL NUMBERS. 95

    286 By exactly the same process wemay deduce from the


    function {e* — e - }" the result that the series
    | arần (1 – 2)” + ” ( 2 P1)
    ! ( – 4)^ _ & c.
    takes the values 0 or 2n .nl, according as r is < n or = n ;
    this series, divided by r !, being equal to the coefficient of x *
    in the expansion of

    POLYGONAL NUMBERS.

    287 The nth term of the 7th order of polygonal numbers is


    equal to the sum of n terms of an Arith . Prog . whose first
    term is unity and common difference r – 2 ; that is

    = {2 +(n - 1)(r — 2)} = n + żn (n - 1)(r ^ 2).


    288 The sum of n terms
    - n (n + 1) _ n (n - 1) (n + 1) (r — 2 )
    By resolving into two series.

    Order. nth term .

    1 1 1 1 1 1 1
    1 2 3 4 5 6 7
    in (n + 1) 1 3 6 10 15 21 28
    11 4 9 16 25 36 49
    in (3n - 1) 1 5 12 22 35 51 70
    (2n - 1) 1 6 15 28 45 66 91
    ... ... ... . .. .. . .. . ... .. . .. . .. .

    n + " (^,^ ) (r - 2) 1, r, 3 + 3 (r – 2 ), 4 + 6 (r 2), 5 + 10 (r — 2),


    6 + 15 (r — 2), & c.
    A
    96 ALGEBR

    In practice - to form , for instance, the 6th order of poly


    gonal numbers — write the first three terms by the formula ,
    and form the rest by the method of differences.
    Ex. : 1 6 15 28 45 66 91 120 ...
    5 9 13 17 21 25 29 ...
    [r — 2 = 4 ] 4 4 4 4 4 ...

    FIGURATE NUMBERS.

    289 The nth term of any order is the sum of n terms of the
    preceding order .
    The nth term of the 7th order is
    n (n + 1) ... (n + r - 2) = HH (n 1).
    (n ,, rr -- 1). [By 98.
    (r - 1) ! =

    290 The sum of n terms is


    n (n + 1 ) ... (ntr - 1 ) =
    H (n , r ) .

    Order. Figurate Numbers. nth term .

    1 1, 1, 1, 1, 1, 1
    2 1, 2 , 3 , 4 , 5, 6 12

    n (n + 1)
    1, 3, 6, 10, 15, 21
    co

    1 .2
    n (n + 1) (n + 2 )
    1, 4, 10, 20, 35, 56 1. 2 .3
    n (1 + 1) (1 + 2 ) (n + 3 )
    1, 5, 15, 35, 70, 126
    or

    1 . 2 . 3 .4
    11. 6 . 21 . 56. 126. 252 n (n + 1) (n + 2) (n + 3 ) (n + 4 )
    1 .2 . 3.4 .5
    o
    HYPERGEOMETRICAL SERIES. 97

    HYPERGEOMETRICAL SERIES.

    291 14a .Brta (a + 1) B (B + 1) m2


    71. " 1.2. y ( + 1)
    , a (a + 1 ) (a + 2 ) B (B + 1) (B + 2) 934 & c.
    1 . 2 . 3 . y (n + 1) ( + 2 )
    on
    is convergent if x is < 1,
    and divergent if x is > l ; . (239 ii.)
    and if x = 1 , the series is
    convergent if y - a - ß is positive,
    divergent if y - a - ß is negative, (239 iv.)
    and divergent if y - a - ß is zero . (239 v.)

    Let the hypergeometrical series (291) be denoted by


    F (a , ß , Y) ; then, the series being convergent, it is shewn by
    induction that
    292 F (a, ß + 1, y + 1) _ concluding with
    F (a , b , y) Ik

    1 - k , 1 - ker - 1
    1 - & c . . .. 1 - k2rm2r

    where kj, kia, kiz, & c., with zzr, are given by the formula
    (a + p - 1) (y + - ] - B ) x
    . ker - 1 =
    ( y + 2r - 2 ) (y + 2r - 1 )
    (B + r ) (ytr - a ) x
    (y + 2r - 1)(x + 2r)
    – F (a + r, B + r + 1, y + 2r + 1)
    F (a + r, + r , 7 + 2r )
    The continued fraction may be concluded at any point
    with lipiep When quis infinite, 727 = 1 and the continued
    fraction is infinite .
    98 ALGEBRA.
    -

    293 Let

    $199= 1+1. +1.2. ( +1) 1.2.3.4(3+1)(3+3 +&e.


    the result of substituting for x in (291), and making
    B = a = 0 . Then, by last, or independently by induction ,
    fly + 1 ) _ 1 P P2 _ Pm .
    MG) = i + 1 + 1 ... + 17 & c.
    with p.„ = (y + m - 1) (y + m )

    294 In this resultput y = } and for X , and we obtain by


    Exp. Th. (150),
    — p-
    + 3 + 37 &c. thesell component being
    Or the continued fraction may be formed by ordinary division
    of one series by the other.

    295 e" is incommensurable, m and n being integers. From


    the last and (174), by putting x = n

    INTEREST.

    If r be the Iuterest on £1 for 1 year,


    n the number of years,
    P the Principal ,
    A the amount in n years. Then
    - - - - -

    296 At Simple Interest A = P (1 + nr).


    297 At Compound Interest A = P (1 + r)". By (8 + ).
    INTEREST AND ANNUITIES. 99

    298 But if the payments of


    Interest be made a { A = ,
    times a year ... ... )
    1+ ")".
    If A be an amount due in n years' time, and P the present
    worth of A . Then
    299 At Simple Interest P = 1 + nr By (296).
    A
    300 At Compound Interest P = 7 tron By (297 ).

    301 Discount = A - P .

    ANNUITIES.

    302 The amount of an Annu - ) nt :n (n2


    ity of £l in n years , { =- nt - 1)
    o
    r. By (82).
    at Simple Interest ... )
    303 Present value of same = ntin
    " (n - 1 )
    1tnr

    304 Amount at Compound - ( 1 + r )" - 1 . By (85) .


    Interest ... ... ... S (1 + r ) - 1

    Present worth of same


    1 - (1 + r) - n
    By (300 ).
    (1 + r ) - 1

    305 Amount when the pay


    ments of Interest
    aremade q times per
    = By (298 ).
    are madeq
    annum
    tim ... )
    ... ..es. perS (1+ 5)*-1
    - 17

    Present value of same =


    100 ALGEBRA.

    306 Amount when the pay .


    ments of the Annuity ( 1 + r )" - 1
    are made m times per
    annum ... ... ... ) m {(1 + r) - 1}
    Present value of same = 1 - (1 + r) "
    m {(1 + r) - 1}
    \ ng
    307 Amount when the In
    terest is paid q times
    and the Annuity m = m *(1+ 1 )! _ 1?
    and the Annuity m
    times per annum .. .. )
    - 19

    Present value of same = 1- (1+ 4)***


    M

    PROBABILITIES.

    309 If all the ways in which an event can happen be m


    in number, all being equally likely to occur, and if in n of
    these m ways the event would happen under certain restrictive
    conditions ; then the probability of the restricted event hap
    pening is equal to něm .
    Thus, if the letters of the alphabet be chosen at random ,
    any letter being equally likely to be taken , the probability of
    a vowel being selected is equal to . The number of un
    restricted cases here is 26 , and the number of restricted
    ones 5 .
    310 If, however, all the m events are not equally probable .
    they may be divided into groups of equally probable cases.
    The probability of the restricted event happening in each
    group separately must be calculated, and the sum of these
    probabilities will be the total probability of the restricted
    event happening at all.
    PROBABILITIES. 101

    · EXAMPLE . — There are three bags A , B , and C .


    A contains 2 white and 3 black balls.
    . B 3
    C . , 4 · 5
    A bag is taken at random and a ball drawn from it. Required the pro
    bability of the ball being wbite .
    Here the probability of the bag A being chosen = j, and the subsequent
    probability of a white ball being drawn = š.
    Therefore the probability of a white ball being drawn from A

    Sinilarly the probability of a white ball being drawn from B

    And the probability of a white ball being drawn from C

    Therefore the total probability of a white ball being drawn

    If a be the number of ways in which an event can happen ,


    and b the number of ways in which it can fail , then the
    a
    311 Probability of the event happening = " .
    a + 6°
    b
    312 Probability of the event failing =
    Thus Certainty = 1.

    If p , p' be the respective probabilities of two independent


    events, then
    313 Probability of both happening = pp'.
    314 „ of not both happening = 1 - pp'.
    315 „ of one happening and one failing
    = p + p' — 2pp'.
    316 „ of both failing = (1 - P) (1 - v').
    102 ALGEBRA.

    If the probability of an event happening in one trial be p,


    and the probability of its failing 7, then
    317 Probability of the event happening r times in n trials
    = C (n , r) p q" -".
    318 Probability of the event failing r times in n trials
    = C (n , r ) p " - " q ". [ By induction.

    319 Probability of the event happening at least 1 times in


    n trials = the sum of the first n - r + 1 terms in the expansion
    of ( p + q)" .
    320 Probability of the event failing at least 1 times in n
    trials = the sum of the last n - p + 1 terms in the same ex
    pansion .

    321 The number of trials in which the probability of the


    same event happening amounts to p '
    log ( 1 - p ')
    log ( 1 - P )
    From the equation (1 – p)* = 1 – p .
    322 DEFINITION. — When a sum of money is to be received if
    a certain event happens, that sum multiplied into the proba
    bility of the event is termed the expectation .
    EXAMPLE . — If three coins be taken at random from a bag
    containing one sovereign , four half-crowns, and five shillings,
    the expectation will be the sum of the expectations founded
    upon each way of drawing three coins. But this is also equal
    to the average value of three coins out of the ten ; that is,
    3ths of 35 shillings, or 10s. 6d .

    323 The probability that, after y chance selections of the


    numbers 0 , 1 , 2 , 3 ... n , the sum of the numbers drawn will
    be s, is equal to the coefficient of ear in the expansion of
    ( 2" + ? + x2 + ... + .2") * = (1 + 1 )".
    PROBABILITIES. 103

    324 The probability of the existence of a certain cause of


    an observed event out of several known causes, one of which
    must have produced the event, is proportional to the a priori
    probability of the cause existing multiplied by the probability
    of the event bappening from it if it does exist.
    Thus, if the a priori probabilities of the causes be P , P ,
    ... & c., and the corresponding probabilities of the event hap
    pening from those causes Qı, Q2 ... & c ., then the probability
    of the poth cause having produced the event is
    P , Qr
    E (PQ)
    325 If P1, P , ... & c . be the a priori probabilities of a second
    event happening from the same causes respectively, then ,
    after the first event has happened , the probability of the

    second happening is (PQP')


    E (PQ ,
    For this is the sum of sucb probabilities as P ,Q Pr, which is
    " I (PQ ) , WI
    ty
    the probabili of the poth cause existing multiplied by the
    probability of the sec
    sec
    ond event happenivg from it .
    - - - -
    Ex. 1. - Suppose there are
    4 vases containing each 5 white and 6 black balls ,
    2 vases containing each 3 white and 5 black balls ,
    and 1 vase containing 2 white and 1 black ball.
    A white ballhas been drawn,and the probability that it came from the group
    of 2 vases is required .
    os
    Va
    c.oo

    Here
    P, = P= P, = }
    ‫لناها‬

    Therefore, by (324), the probability required is


    2.3
    7.8 - 99
    4 .5 2 .3 1.2
    7 . 11 * 7 .8 * 7 . 3
    Ex. 2 . - After the white ball has been drawn and replaced , a ball is
    drawn again ; required the probability of the ball being black .
    104 ALGEBRA.

    Here P=i Ps = Ps= }


    The probability , by (325 ), will be
    4 .5 .6 2 . 3 . 5 , 1 .2 . 1
    7 . 11 . 11 * 7 . 8 . 8 * 7 . 3. 3 _ 58639
    4 . 5 , 2 . 3 21. 1 12728
    7 . 11 * 7 .877. 3
    If the probability of the second ball being white is required , Q .QqQ .
    must be employed instead of PíPPž.

    326 The probability of one event at least happening out of


    a number of events whose respective probabilities are a , b , c ,
    & c., is P , - P , + P3 - P , + & c.
    where P , is the probability of 1 event happening ,
    P, , , 2 ,
    and so on . For,by (316), the probability is
    1 - ( 1 - a) (1 — b) (1 - c) ... = £a - 2ab + Eabe - , ...
    324 The probability of the occurrence of assigned events
    and no more out of n events is
    Qr - Qr+1 + Qr+2 - Qr+s + & c.,
    where Q , is the probability of the r assigned events ; Qrt , the
    probability of r + l events including the r assigned events.
    For if a , b , c ... be the probabilities of the r events, and
    a ', b , c .. . the probabilities of the excluded events, the re
    quired probability will be
    abc ... (1 - a ') (1 - 6 ) (1 - 0') ...
    = abc ... (1 - Ea' + Sa'l' — Sa'b'c' + ...).

    328 The probability ofany r eventshappening and no more


    EQr - EQr+1 + EQr+2 — & c.
    NotE. — If a = b = c = & c., then XQ, = C (n, r ) Qr, & c.
    INEQUALITIES. 105

    INEQUALITIES .

    330 az + az + ... + an lies between the greatest and least of


    bi + b2 + ... + on

    the fractions ... , the denominators being all of


    the same sign .
    Proof. — Let k be the greatest of the fractions, and an any other; then
    a > < kby. Substitute in this way for each a . Similarly if k be the least
    fraction .

    331 . atb > Vab.


    a , taz + .. . ta ? > va ,02 . .. An
    . 332 n
    ;

    or, Arithmetic mean > Geometric mean.


    Proov. - Substitute both for the greatest and least factors tbeir Arith
    metic mean . The product is thus increased in value. Repeat the process
    indefinitely . The limiting value of the G . M . is the A . M . of the quantities.

    333 " " > (449)",


    excepting when m is a positive proper fraction.
    Proof: q2" + " = (~ 7%)" {(1 +2)* + (1 —2)" },
    where x = 95
    a +.b Employ Bin. Th.
    922

    334 ( +4 +...+0%
    n >(9,1+2: .+a )",
    excepting when m is a positive proper fraction.
    106 ALGEBRA.
    -- - -

    Otherwise . — The Arithmetic mean of the mth powers is


    greater than the mth power of the Arithmetic mean , excepting
    when m is a positive proper fraction .
    Proof. -- Similar to (332 ). Substitute for the greatest and least on the
    left side, employing (333) .

    336 If a and m are positive, and x and mx less than unity ;


    then (1 + x ) -m > 1 - mx. (125, 240)

    337 If x , m , and n are positive, and n greater than m ; then ,


    by taking a small enough, we can make
    1 + nx > ( 1 + x )"
    For æ may be diminished until 1 + nx is > (1 - mx)- , and this
    is > (1 + x ) ", by last.

    338 If x be positive, log (1 + x ) < x . (150)


    If æ be positive and > 1, log (1 + x ) > x -
    e
    (155, 240)
    1lve
    If æ be positive and < 1, log > . (156)

    339 When n becomes infinite in the two expressions


    1 . 3 . 5 . .. (2n - 1 ) 3 . 5 . 7 ... (2n + 1)
    2 . 4 . 6 ... 2n 2 . 4 . 6 ... 2n
    the first vanishes, the second becomes infinite, and their
    product lies between 1 and 1.
    Shewn by adding 1 to each factor (see 73), and multi
    plying the result by the original fraction.

    340 If m be > 11, and n > a ,


    mta " cinta "
    m - a In - a .
    SCALES OF NOTATION . 107

    341 If a , b be positive quantities,


    +

    «*Vº is > (47 )***


    ( a + b + c a +b + c
    Similarly aabº cº >

    These and similar theorems may be proved by taking loga .


    rithms of each side, and employing the Expon. Th. (158), & c .
    L NEL

    SCALES OF NOTATION .

    342 If N be a whole number of n + 1 digits, and r the radix


    of the scale , N = P2P + Pn- 1709 - 1 + Pn -27011– + ... + Pir + Po,

    where Pn, Pn-1, ... p, are the digits .


    343 Similarly a radix-fraction will be expressed by
    *1 + + + & c.,
    where P1, P2, & c. are the digits.
    EXAMPLES : 3426 in the scale of 7 = 3 .78 + 4 . 7² + 2 .7 + 6 ;
    10 * in the same scale = 7 + 72 + 73 + 7

    344 Ex. — To transform 34268 from the scale of 5 to the


    scale of 11.
    RULE . — Divide successively by the new radix .
    11 : 34263
    11 1343 - t
    11| 40 - 3
    1 -9 Result 193t, in which t stands for 10.
    108 ALGEBRA.

    345 Ex. - To transform t0el from the scale of 12 to that


    of 7 , e standing for 11, and t for 10 .
    Rule . — Multiply successively by the new radix.
    •toel

    5 4657
    6 :1931
    1 :0497
    0 . 2971 Result •5610 ...

    346 Ex. - In what scale does 2t7 represent the number 475
    in the scale of ten ?
    Solve the equation 2r? + 10r + 7 = 475 . [ 178
    Result r = 13.

    347 The sum of the digits of any number divided by p - 1


    leaves the same remainder as the number itself divided by
    7 - 1 ; r being the radix of the scale. (401)
    348 The difference between the sums of the digits in the
    even and odd places divided by 1 + 1 leaves the same re
    mainder as the number itself when divided by r + 1 .

    THEORY OF NUMBERS.

    349 If a is prime to b, is in its lowest terms.


    ms.

    Proof. — Let a fraction in lower terms.


    Divide a by an, remainder a , quotient qı,
    b by bı, remainder b, quotient qui
    and so on, as in finding the H . C . F . of a and an, and of b and b , ( see 30) .
    Let an and bn be the highest common factors thus determined .
    THEORY OF NUMBERS. 109

    Then, because ő =
    use a – 2 – 9,
    . = 7–9,6 bi (70)
    and so on;thus ; = = = &......
    Therefore a and b are equimultiples of an and bn ; that is, a is not prime to b
    if any fraction exists in lower terms.

    350 If a is prime to b, and = ; then a' and bare


    equimultiples of a and b .
    PROOF.
    Proof.— Let reduced to its lowest terms be . Then l = ,and ,
    since p is now prime to q, and a prime to b , it follows, by 349, that is
    neither greater nor less than ; that is, it is equal to it. Therefore, & c.

    351 If ab is divisible by c, and a is not ; then b must be .


    ab
    Proof.–Let ab == 9;
    9; :i = %.
    But a is prime to c ; therefore, by last, b is a multiple of c.
    352 If a and b be each of them prime to c , ab is prime
    to c . . [By ( 351).

    353 If abcd . .. is divisible by a prime, one at least of the


    factors a, b, c, & c. must also be divisible by it .
    Or, if p be prime to all but one of the factors, that factor
    is divisible by p.
    (351)
    354 Therefore, if a " is divisible by p , p cannot be prime to
    a ; and if p be a prime it must divide a .

    355 Ifa is prime to b , any power of a is prime to any


    power of b .
    Also , if a , b , c , & c . are prime to each other, the product
    of any of their powers is prime to any other product of their
    powers .
    110 ALGEBRA.

    356 No expression with integral coefficients, such as


    A + BX + Cx? + ..., can represent primes only.
    Proof. - For it is divisible by x if A = 0 ; and if not, it is divisible by A ,
    when x = A .

    357 The number of primes is infinite.


    Proof. — Suppose if possible p to be the greatest prime. Then the pro
    duct of all primes up to p , plus unity , is either a prime, in which case it
    would be a greater prime than p , or it must be divisible by a prime ; but
    no prime up to p divides it , because there is a remainder 1 in each case .
    Therefore, if divisible at all, it must be by a prime greater than p . In
    either case, then , a prime greater than p exists .

    358 If a be prime to b, and the quantities a , 2a , 3a , ...


    (6 - 1) a be divided by b , the remainders will be different.
    Proof. — Assume ma — nb = m 'a - n 'b , m and n being less than b ,
    i =
    bm- m '
    -n . Then by ( 350).

    359 A number can be resolved into prime factors in one


    way only . [ By (353).
    360 To resolve 5040 into its primefactors.
    Rule.- Divide by the prime numbers successively.
    2 x 5 5040
    2 / 504
    2 / 252
    2 126
    7 63
    319
    3 Tbus 5040 = 24. 32. 5. 7 .

    361 Required the least multiplier of 4704 which will make


    the product a perfect fourth power.
    By (196), 4704 = 25. 3. 72
    Then 23 . 3 .72 x 23. 35.72 = 28. 34. 7* = 84*,
    the indices 8, 4 , 4 being the least multiples of 4 which are not less than
    5 , 1, 2 respectively .
    Thus 28. 38 .72 = 3584 is the multiplier required.
    THEORY OF NUMBERS. 111

    362 All numbers are of one of the forms 2n or 2n + 1


    2n or 2n - 1
    3n or 3n + 1
    4n or 4n + 1 or 4n + 2
    4n or 4n + 1 or 4n - 2
    5n or 5n + 1 or 5n + 2
    and so on .
    363 All square numbers are of the form 5n or 5n + 1 .
    Proof. — By squaring the forms õn, 5n + 1, 5n + 2 , which comprehend
    all numbers whatever .
    364 All cube numbers are of the form 7n or 7n31.
    And similarly for other powers.

    365 The highest power of a prime p , which is contained in


    the product m !, is the sum of the integral parts of
    m m m &
    p ' peo psi & c.
    For there are º factors in m ! which p will divide; m, which
    it will divide a second time ; and so on . The successive
    divisions are equivalent to dividing by

    pö.pa... & c. = pi +*
    XAMPLE. — The highest power of 3 which will divide 29 !. Here the
    factors 3, 6 , 9 , 12, 15, 18, 21, 24, 27 can be divided by 3 . Their number is
    S = 9 (the integral part) .
    The factors 9 , 18 , 27 can be divided a second time. Their number is
    = 3 (the integral part).
    29
    One factor, 27, is divisible a third time. 33
    = 1 (integral part).
    9 + 3 + 1 = 13 ;that is, 318 is the highest power of 3 which willdivide 29 !.

    366 The product of any r consecutive integers is divisible


    byr!.
    Proof: n (n - 1) .. (" - - + 1) is necessarily an integer, by (96 ),
    112 ALGEBRA

    367 If n be a prime, every coefficient in the expansion of


    (a + b)", except the first and last, is divisible by n . By last.
    368 If n be a prime, the coefficient of every term in the ex
    pansion of (a + b + c .. .) " , except a ", 6 " , & c ., is divisible by n .
    Proof. — By ( 367). Put ß for (b + c + ...) .

    369 Fermat's Theorem . - If p be a prime, and N primeto


    p ; then Nr- 1 - 1 is divisible by p .
    PROOF : Nº = ( 1 + 1 + ...)" = N + Mp. By ( 368).
    370 If p be any number, and if 1, a , b, c, ... ( p - 1) be all
    the numbers less than, and prime to p ; and if n be their
    number, and x any one of them ; then æ " – 1 is divisible by p .
    PROOF. — If x, ax, bx ... (p - 1 ) x be divided by p , the remainders will be
    all different and prime to p ( as in (358 ) ] ; therefore the remainders will be
    1 , a , b, c ... (p - 1) ; therefore the product
    "abc ... ( p - 1 ) = abc ... ( p - 1) + Mp.

    371 Wilson 's Theorem . If p be a prime, and only then ,


    1 + ( p - 1 ) ! is divisible by p .
    Put p - 1 for r and n in (285),and apply Fermat's Theorem
    to each term .
    372 If p be a prime = 2n + 1, then (n !)? + ( - 1)" is divisible
    by p .
    Proof. — By multiplying together equi-distant factors of ( p - 1) ! in
    Wilson's Theorem , and putting 2n + 1 for p .

    373 Let N = 1,9c” ... in prime factors ; the number of in


    tegers, including 1 , which are less than n and prime to it, is

    (1 - )(1 - 7)(1 - 1)....


    PROOF. – The number of integers prime to N contained in ap is ab — 4
    Similarly in / ?, c", & c. Take the product of these,
    THEORY OF NUMBERS. 113

    Also the number of integers less than and prime to


    (NX MX & c.) is the product of the corresponding numbers
    for N , M , & c . separately .

    374 The number of divisors of N , including 1 and N itself,


    is = ( P + 1) ( 9 + 1 ) ( r + 1 ) . ... For it is equal to the number
    of terms in the product
    (1 + a + ... tap) (1 + b + ... + 62) (1 + c + ... + c") ... & c .
    375 The number of ways of resolving N into two factors is
    half the number of its divisors (374 ) . If the number be a
    square the two equal factors must , in this case, be reckoned
    as two divisors.
    376 If the factors of each pair are to be prime to each other .
    put P , 9, ? , & c. each equal to one.

    377 The sum of the divisors of N is


    QP + 1 – 1 69+ 1 - 1 cr+ 1 _ 1
    a - 1 6 - 1 6 - 1 **
    PROOF. — By the product in (374), and by (85) .
    378 If p be a prime, then the p — 1th power of any number
    is of the form mp or mp + 1 . By Fermat' s Theorem ( 369) .
    The 12th power of any number is ofthe form 13m or 13m + 1.
    379 To find all the divisors of a number ; for instance, of 504.

    22

    28
    63 | 3 | 3 6 12 24
    21 / 3 9 18 36 72
    7 7 14 28 56 21 42
    84 168 63 126 252 504
    EXPLANATION . — Resolve 504 into its prime factors, placing them in
    column II.
    114 ALGEBRA .

    The divisors of 504 are now formed from the numbers in column II., and
    placed to the right of that column in the following manner :
    Place the divisor 1 to the right of column II., and follow this rule
    Multiply in order all the divisors which are written down by the next number
    in column II., which has not already been used as a multiplier : place the first
    new divisor so obtained and all the following products in order to the right of
    column II.

    380 S, the sum of the goth powers of the first n natural


    numbers is divisible by 2n + 1.
    Proof : x (x — 1°) (r? — 24) ... (2 * — 12 )
    constitutes 2n + 1 factors divisible by 2n + 1, by (366 ) . Multiply ont, re
    jecting x, which is to be less than 2n + 1. Thus, using (372),
    22" — S,220-3 + S222n --- ... S -102 + ( - 1 )" (\n ) = M (2n + 1).
    Pat 1, 2, 3 ... (n - 1) in succession for x , and the solution of the (1 - 1)
    equations is of the form S, = M (2n + 1) .
    THEORY OF EQUATIONS.

    FACTORS OF AN EQUATION .

    Generalform of a rationalintegral equation of thenth degree.


    400 Porn + 2247-1 + p2«n- + ... + Pn=1& + Pn = 0.
    The left side will be designated f (@ ) in the following
    summary.

    401 Iff (a ) be divided by x - a , the remainder will be f (a ).


    By assuming $ (Q ) = P (x — a ) + R .
    402 If a be a root of the equation f ( ) = 0 , then f (a ) = 0.

    403 . To compute f (a ) numerically ; divide f (w ) by a — a ,


    and the remainder will be f (a ). [401
    404 EXAMPLE. — To find the value of 4. 0 — 325 + 12x * — 20° + 10 when w = 2.
    | 4 – 3 + 12 + 0 - 1 + 0 + 10
    2 8 + 10 + 44 + 88 + 174 + 348
    4 + 5 + 22 + 44 + 87 + 174 + 358 Thus f (2) = 358.

    If a, b, c ... k be the roots of the equation f (x ) = 0 ;


    then, by (401) and (402 ),
    405 f(x ) = p. (x - a ) (x - 6 ) (x — c ) ... (x - k ).
    By multiplying out the last equation, and equating coefficients with
    equation (400), considering po = 1, the following results are obtained :
    116 THEORY OF EQUATIONS.
    -
    406 -- P1 = the sum of all the roots of f (x ).
    the sum of the products of the roots taken
    ? two at a time.
    Sthe sum of the products of the roots taken
    three at a time.

    en the sum of the products of the roots taken


    ( - 1)"Pr = 1 p at a time.
    ( - 1) " Pn = product of all the roots.

    407 The number of roots of f (x) is equal to the degree of


    the equation .

    408 Imaginary roots must occur in pairs of the form


    a + 84 – 1, a - 3 - 1.
    The quadratic factor corresponding to these roots will
    then have real coefficients ; for it will be
    24 — 2ax ta' + ß2. [405, 226
    409 If f (x ) be of an odd degree, it has at least one real root
    of the opposite sign to Pro
    Thus 27 — 1 = 0 has at least one positive root.
    410 If f (2 ) be of an even degree, and Pn negative , there is
    at least one positive and one negative root.
    Thus * — 1 has + 1 and — 1 for roots.

    411 If several terms at the beginning of the equation are of


    one sign, and all the rest of another, there is one, and only
    one, positive root.
    Thus 208 + 2x * + 3.x® +x * — 52 — 4 = 0 has only one positive root.

    412 If all the terms are positive there is no positive root.


    413 If all the terms of an even order are of one sign , and
    all the rest are of another sign , there is no negative root,
    414 Thus 2 * - 28 + x * -- ~ + 1 = 0 has no negative root,
    DISCRIMINATION OF ROOTS. 117

    415 If all the indices are even , and all the termsof the same
    sign , there is no real root ; and if all the indices are odd, and
    all the terms of the same sign , there is no real root but zero.
    Thus 2 * + this+ 1la=st 0eqhas co rereal
    uatino al root, nd 2200 ++
    root,aand 8 + x = 0 has no real root
    but zero . In this last equation there is no absolute term , because such a
    ould
    tetermrm wwould involve the zero power of a , which is even , and by hypothesis is
    wanting .

    DESCARTES RULE OF SIGNS.

    416 In the following theorems every two adjacent terms in


    f (x ),which have the same signs, count as one “ continuation
    of sign ” ; and every two adjacent terms, with different signs,
    count as one change of sign .

    417 f (x ), multiplied by (æ - a ), has an odd number of


    changes of sign thereby introduced, and one at least.
    418 f (x ) cannot have more positive roots than changes of
    sign , or more negative roots than continuations of sign .
    419 When all the roots of f (a ) are real, the number of
    positive roots is equal to the number of changes of sign in
    f (a ); and the number of negative roots is equal to the number
    of changes of sign in f ( – ).
    420 Thus, it being known that the roots of the equation
    2 * — 1023 + 3522 — 50x + 24 = 0
    are all real; the number of positive roots will be equal to the number of
    changes of sign , which is four. Also f ( - x ) = x* + 10x8 + 35x* + 502 + 24 = 0 ,
    and since there is no change of sign, there is consequently, by the rule , no
    negative root.

    4:21 If the degree of f (2 ) exceeds the number of changes of


    sign in f(@ ) and f ( - x ) together, by u , there are at least u
    imaginary roots.
    422 If, between two terms in f (a ) of the same sign , there
    be an odd number of consecutive terms wanting , then there
    must be at least one more than that number of imaginary
    roots ; and if the missing terms lie between terms of different
    118 THEORY OF EQUATIONS.

    sign , there is at least one less than the same number of


    imaginary roots.
    Thus, in the cubic equation 200 + 46 — 7 = 0 , there must be two imaginary
    roots .
    And in the equation wº– 1 = 0 there are, for certain , four imaginary roots.
    423 If an even number of consecutive terms be wanting in
    fa), there is at least the same number of imaginary roots .
    Thus the equation 0 + 1 = 0 has four terms absent ; and therefore four
    imaginary roots at least .

    THE DERIVED FUNCTIONS OF f (x).

    Rule for forming the derived functions.


    424 Multiply each term by the index of x, and reduce the
    index by one; that is, differentiate the function with respect
    to X .
    EXAMPLE. — Take
    f (x ) = 2 + 4 + 200 – 2* — * — 1
    f (x ) = 5x + 4.29 + 3x2 – 2x – 1
    20x3 + 1242 + 63 — 2
    f ( ) = 60x* + 24x + 6
    f* (a ) = 120x + 24
    f (x) = 120
    f (x ), f (x ), & c. are called the first, second, & c. derived functions of f (- ).

    425 To form the equation whose roots differ from those of


    f (x ) by a quantity a .
    Put x = yta in f (x ), and expand each term by the Binomial
    Theorem , arranging the results in vertical columns in the fol
    lowing manner :
    f (a + y) = (a + y) + (a + y) + (a + y) – (a + y)?– (a + y) — 1
    aº + a + aº – aº - 0 -1
    + ( 5a + 4a8 + 3a : - 2a - 1) y
    + + +

    + ( 10a: + 6a² + 3a - 1) ya
    + ( 10a + 4a + 1 ) yo
    + ( 5a + 1) ya
    + y
    TRANSFORMATION OF AN EQUATION. 119

    Comparing this result with that seen in (424),it is seen that


    426 f(a + y) = f(a) + fº(a)y

    80 that the coefficient generally of y" in the transformed


    equation is f*(a),

    427 To form the equation most expeditiously when a has a


    numerical value, divide f (x ) continuously by X - a , and the
    successive remainders will furnish the coefficients.
    EXAMPLE . — To expand f ( y + 2) when, as in (425),
    f (x) = x + 24 + — — 2 - 1.
    Dividerepeatedly by a – 2, as follows :
    11 + 1 + 1 - 1 - 1 - 1
    2 + 2 + 6 + 14 + 26 + 50
    11 + 3 + 7 + 13 + 25 + 49 = f (2 )
    | + 2 + 10 + 34 + 94
    1 + 5 + 17 + 47 + 119 = f' (2)
    1 + 2 + 14 + 62
    11 + 7 + 31 + 109 = 1
    2 + 2 + 18 That these remainders
    are the required coefficients
    1 + 9 + 49 is seen by inspecting the
    + 2 form of the equation (426 ) ;
    1 + 11 = Il for if that equation be di.
    vided by x - ary repeat
    edly, these remainders are
    1 = L (2) obviously produced when
    a = 2.
    Thus the equation , whose roots are each less by 2 than the roots of the
    proposed equation , is yo + 1ly* + 49y® + 109yº + 119y + 49 = 0.
    428 To make any assigned term vanish in the transformed
    equation, a must be so determined that the coefficient of that
    term shall vanish .
    EXAMPLE. - In order that there may be no term involving y * in equation
    (426 ),wemust have f* (a ) = 0 .
    • Find f* (a ) as in (424 ) ;
    thus 120a + 24 = 0 ; i a = - .
    The equation in (424 ) must now be divided repeatedly by x + } after the
    manner of (427 ) , and the resulting equation will be minus its second term .
    120 THEORY OF EQUATIONS.

    429 Note , that to remove the second term of the equation


    f (x ) = 0 , the requisite value of a is = - P1 ; that is, the
    npo '
    coefficient of the second term , with the sign changed , divided by
    the coefficient of the first term , and by the number expressing
    the degree of the equation .

    430 To transform f (x ) into an equation in y so that y = " (« ),


    a given function of X ,put x = p -'(y), the inverse function of y .
    EXAMPLE . — To obtain an equation whose roots are respectively three times
    the roots of the equation 28– 6x + 1 = 0. Here y = 3x ; therefore x = % ,
    and the equation becomes - 64 + 1 = 0, or yo — 544 + 27 = 0.

    431 To transform f ( x ) = 0) into an equation in which the


    coefficient of the first term shall be unity, and the other
    coefficients the least possible integers.
    EXAMPLE. — Take the equation
    288x8 + 240x * — 176x – 21 = 0.
    Divide by the coefficient of the first term , and reduce the fractions; the
    2315 11 7
    equation becomes = 0.
    96

    Substitute for æ,and multiply bykº;weget

    Next resolve the denominators into their prime factors,


    5k 11K 7k - 0.
    gå + 234 – 2. .17 – 20.3
    The smallest value must now be assigned to k , which will suffice to make
    each coefficient an integer. This is easily seen by inspection to be 2 .3 = 12,
    and the resulting equation is y + 10y — 887 – 126 = 0 ,
    the roots of which are connected with the roots of the original equation by
    the relation y = 12a .

    EQUAL ROOTS OF AN EQUATION .


    By expanding f (x + z) in powers of z by (405), and also
    (426 ), and equating the coefficients of z in the two ex
    EQUAL ROOTS. 121

    pansions,it is proved that


    f (x ) - f 1(x2 ) J (1)
    432 $"(a) = (2 - ) + 246 + 6) + & c .,

    from which result it appears that, if the roots a , b , c , & c . are


    all unequal, f ( x ) and f ' ( x ) can have no common measure in
    volving X . If, however, there are p roots each equal to a ,
    s roots equal to b , t roots equal to c , & c ., so that

    then f (x ) = po(x — a)" (x —b)*(x — c)'...


    433 8"(t) = if(d! + + L3 + &c.;
    and the greatest common measure of f (x ) and f'(x) will be
    444 (x — a )r -1(x — b ).-4 (x — c) -1...
    When x = a , f (x), f' (x), ... fr -1 ( ) all vanish. Similarly
    when x = b , & c.

    Practicalmethod of finding the equal roots.


    445 Let f (x ) = X , X , X, X , X ; ... Xm, where
    Xi = product of all tbe factors like (x — a ),
    x; = (x - a ) ,
    (2 - a) .
    Find the greatest common measure of f (x ) and f' (x ) = F (a ) say,
    F (x ) and Fi(x ) = F , (2 ) ,
    F (x ) and F ; (x ) = Fg (x),

    Lastly, the greatest common measure of Fm -1 (2 ) and Fm-1 (x ) = Fm (x ) = 1 .


    Next perform the divisions
    f (x ) • F ;(iv) = 0. (x ) say ,
    F ,(x ) = F ,(x ) = 0 (x),
    . ... .. . .. . . ..
    ... ... ... ...
    Fm -1 (x) = 1 = 4» (x ).
    And, finally , ( c) + 6 ( z) = X ,
    P : (x) = 4s ( ) = X.,
    . .. .. . . .. ...
    . .. .. . .. .
    P. -1 (x) = ºm (a ) = x -1,
    F -1 (x ) = p.. (x ) = X.. . [ T. 82.
    R
    122 THEORY OF EQUATIONS.

    The solution of the equations X , = 0 , X , = 0 , & c . will furnish all the


    roots of f (x ) ; those which occur twice being found from X , ; those which
    occur three times each, from Xg; and so on .
    446 If f (x ) has all its coefficients commensurable , X , X2, X3,
    & c . have likewise their coefficients commensurable .
    Hence, if only one root be repeated r times, that root must
    be commensurable .
    449 In all the following theorems, unless otherwise stated ,
    f (x ) is understood to have unity for the coefficient of its first
    term .

    LIMITS OF THE ROOTS.

    448 If the greatest negative coefficients in f (x ) and f ( - x )


    be p and q respectively, then p + 1 and – ( 2 + 1) are limits of
    the roots .
    449 If xn -r and -s are the highest negative terms in f (x)
    and f ( - x ) respectively , ( 1 + ♡p) and - (1 + 9) are limits
    of the roots .

    450 If k be a superior limit to the positive roots of f ( ) ,


    then 7 will be an inferior limit to the positive roots of f (x).
    an е то

    451 If each negative coefficient be divided by the sum ofall


    the preceding positive coefficients , the greatest of the fractions
    so formed + unity will be a superior limit to the positive
    roots .

    452 Newton 'smethod . — Putx = h + y in f (x ); then ,by (426 ),


    f(h + y) = f(h)+ uf"(h)+ 8°(H) + ... + **(h) = 0.
    Take h so that f (h ), f'(h ), f2 (h ) ... f" (h) are all positive ;
    then h is a superior limit to the positive roots.

    453 According as f (a ) and f (6 ) have the same or different


    signs, the number of roots intermediate between a and b is
    even or odd.
    INTEGRAL ROOTS. 123

    454 Rolle's Theorem . — One real root of the equation f' (2 )


    lies between every two adjacent real roots of f (x ).
    455 COR . 1. - f (w ) cannot have more than one root greater
    than the greatest root in f'(a ) ; or more than one less than
    the least root in f' (u).
    456 Cor . 2 .— If f (x ) has m real roots, f' (x ) has at least
    m - p real roots .
    457 COR . 3. - If fr (w ) has u imaginary roots, f (w ) has also
    u at least .
    458 COR . 4 . - If a, ß , y ... K be the roots of f"(2) ; then the
    number of changes of siyu in the series of terms
    f( ), f(a), f (B), f (y) ... f( - 00 )
    is equal to the number of roots of f (w ).

    NEWTON 'S METHOD OF DIVISORS.

    459 To discover the integral roots of an equation.


    EXAMPLE. — To ascertain if 5 be a root of 5 ) 105
    * — 6w* + 86x2 – 176x + 105 = 0 .
    - 176
    If 5 be a root it will divide 105. Add the quotient to the
    next coefficient. Result, – 155 . 731
    If 5 be a root it will divide – 155. Add the quotient to 86
    the next coefficient ; and so on. 5 ) 55
    If the number tried be a root, the divisions will be effectible 11
    to the end , and the last quotient will be - 1, or — Po, if po be - 6
    pot anity. ) -5
    - 1

    460 In employing this method , limits of the roots may first


    be found , and divisors chosen between those limits.
    461 Also, to lessen the number of trial divisors, take any
    integer m ; then any divisor a of the last term can be rejected
    if a - m does not divide f (m ) .
    In practice take m = + 1 and - 1.
    To find whether any of the roots determined as above are
    repeated , divide f (x ) by the factors corresponding to them ,
    and then apply the method of divisors to the resulting equation .
    124 THEORY OF EQUATIONS.

    EXAMPLE. — Take the equation


    2° + 2x – 17x* - 26x9 + 88x2 + 72.c — 144 = 0.
    Putting x = 1, we find f (1) = - 24. The divisors of 144 are
    1, 2, 3, 4, 6 , 8, 9, 12, 16 , 24 , & c.
    The values of a - m (since m = 1) are therefore
    0, 1, 2, 3, 5, 7, 8, 11, 15, 23, & c.
    imbers only 1, which it is of use the3.99
    Of these last numbers only 1, 2 , 3 , and 8 will divide 24 . Hence 2 , 3 , 4 , and
    9 are the only divisors of 144 which it is of use to try . The only integral
    roots of the equation will be found to be + 2 and + 3.

    462 If f (x ) and F ( X ) have common roots, they are con


    tained in the greatest common measure of f (x) and F (X ).
    mon

    463 If f (a ) has for its roots a , o (a) , b , ° (6 ) amongst others ;


    then the equations f (x) = 0 and f { Q ( x) = 0 have the common
    roots a and b .
    464 But, if all the roots occur in pairs in this way, these
    equations coincide.
    For example , suppose that each pair of roots, a and b, satisfies the equation
    a + b = 2r. We may then assume a - b = 2z. Therefore f (= + p ) = 0. This
    equation involves only even powers of z, and may be solved for z?.

    465 Otherwise : Let ab = z; then f ( c) is divisible by (x — a) (x - 6)


    = x2 – 2rx + z. Perform the division until a remainder is obtained of the
    form Pæ + Q , where P and Q only involve z.
    The equations P = 0 , Q = ) determine z, by (462) ; and a and I are found
    from a + b = 2r, ab = z.

    RECIPROCAL EQUATIONS.

    466 A reciprocal equation has its roots in pairs of the form


    a, — ; also the relation between the coefficients is
    Pr = Pn -rs or else Pr = - Pn -r.
    r9
    -

    467 A reciprocal equation of an eren degree, with its last


    - -- -

    term positire, may be made to depend upon the solution of an


    equation of half the same degree.
    BINOMIAL EQUATIONS. 125

    468 EXAMPLE : 4.xs — 24x5 + 5724 — 73x3 + 5702 — 24x + 4 = 0


    is a reciprocal equation of an even degree, with its last term
    positive.
    Any reciprocal equation which is not of this form may be
    reduced to it by dividing by x + 1 if the last term be positive ;
    and , if the last term be negative, by dividing by a - 1 or x ? – 1 ,
    so as to bring the equation to an even degree. Then proceed
    in the following manner :
    469 First bring together equidistant terms, and divide the
    equation by r ; thus

    4( + )– 24(x2 + 2)+ 57 (x + – 73 = 0.
    By putting a + 1 = y, and by making repeated use of the
    relation ** + * = (x + 1) – 2,the equation is reduced to
    a cubic in y, the degree being one-half that of the original
    equation .

    Put p for x + , and Pm for xmt


    470 The relation between the successive factors of the form
    Pm may be expressed by the equation
    Pm = PPm - 1 - Pm - 2.
    471 The equation for Pm, in terms of p, is
    Pu = p" –mp" = + m m
    ( 3) pm -s ...
    10 m (m - p - 1 ) ... (m — 2r + 1 ) ym
    P
    -2r ++ ...
    ...

    By (545), putting q = 1.

    BINOMIAL EQUATIONS.

    472 If a be a root of x " — 1 = 0 , then am is likewise a root


    where m is any positive or negative integer .
    473 If a be a root of " + 1 = 0 , then a2m + 1 is likewise a root.
    126 THEORY OF EQUATIONS.

    474 If m and n be prime to each other, m — 1 and " - 1


    have no common root but unity .
    Take pm - qn = 1 for an indirect proof.

    475 If n be a prime number, and if a be a root of 2 " - 1 = 0 ,


    the other roots are a , a ', a ... a ” .
    These are all roots, by (472 ). Prove, by (474 ), that no two can be equal.
    476 If n be not a prime number, other roots besides these
    may exist. The successive powers, however, of some root
    will furnish all the rest.
    4719 If 21 — 1 = 0 has the index n = mpq ; m , p , q being
    prime factors ; then the roots are the terms of the product
    (1ta ta' + ... tam - 1) ( 1 + B + 82 + ... + BP- 2)
    * (1 + y + y + ... + 49-?),
    where a is a root of XM – 1,
    2P 1:
    but neither a , b , nor y = 1. Proof as in (475).
    478 If n = m ", and
    a be a root of xm – 1 = 0,
    ß , 2011 - a = 0,
    y , 2m – B = 0 ;
    then the roots of 2 " - 1 = 0 will be the terms of the product
    ( 1 + a ta' + ... + am - 1) ( 1 + B + B4 + ... + B ^ - )
    * (1 + y + y + ... + ym -1).
    4792 + 1 = 0 may be treated as a reciprocal equation , and
    depressed in degree after the manner of (468).
    480 The complete solution of the equation
    20M – 1 = 0
    is obtained by De Moivre's Theorem . (757)
    The n different roots are given by the formula
    x = .cos 20*
    n + V = 1 sin 2017 n

    in which we must have the successive values 0, 1, 2, 3, & c.,


    concluding with h , if n be even ; and with " 5 ", if n be odd.
    CON
    CUBIC EQUATIONS. 127

    481 Similarly the n roots of the equation I

    21 + 1 = 0
    are given by the formula
    · < = cos(20+ 1) = v = 1 sin(2011)-
    n

    r taking the successive values 0, 1, 2, 3, & c., up to " > , if


    n be even ; and up to 7 , if n be odd .
    eve

    482 The number of different values of the product


    1 1

    is equal to the least common multiple of m and n ,when m and


    n are integers. .

    CUBIC EQUATIONS.

    483 To solve the general cubic equation


    2x2 + px + qx + r = 0 .
    Remove the term på by the method of (429). Let the trans ns

    formed equation be xv8 + qx + r = 0.


    484 Cardan' s method . — The complete theoretical solution
    of this equation by Cardan's method is as follows :
    Put @ = y + % (i.)
    38+ + (3yz + q) (y +z)+ r = 0.
    Put 3yz+ q = 0 ; :: y =
    Substitute this value of y , and solve the resulting quadratic
    in yº. The roots are equal to ys and respectively ; and we
    have, by (i.),

    435 -= {- +v + } + {-1 -VE + B }!


    128 THEORY OF EQUATIONS.

    The cubicmust bave one real root at least, by (409 ).


    2825,
    + 27 and n one
    Let m be one of the three value

    of the three values of { - - VA + }


    486 Let 1, a, a ’ be the three cube roots ofunity, so that
    a = - + } V -3, and «°= - 1 - 1 V -3. [472
    487 Then, since my = m / 1, the roots of the cubic will be
    m + n , am + a ’n , aºm tan.
    Now , if in the expansion of

    - - -
    -
    by the Binomial Theorem , we put
    M = the sum of the odd terms, and
    v = the sum of the even terms;
    then we shall have m = utv, and n = u - v ;
    or else m = uto V - 1, and n = u - YV - 1 ;
    according as vã + is real or imaginary.
    By substituting these expressions for m and n in (487), it appears that
    488 be positive, the roots of the cubic will be
    - r toV - 3 , - u - V - 3.
    be negative, the roots will be
    M , - u tvV3, - u - vV3.
    (ii.) IP * + ** = 0,the roots are
    2m , -m , -m ;
    since m is now equal to u .

    489 The Trigonometrical method . — The equation


    208 + gx + p = 0
    may be solved in the following manner, by Trigonometry,
    lan

    when + is negative.
    Assume x = n cos a. Divide the equation by nº; thus
    cos'a + o cosa + = 0.
    OS a

    But cosma - 2 cos a – Cos 3a = 0. By (657)


    BIQUADRATIC EQUATIONS. 129

    Equate coefficients in the two equations ; the result is


    C
    n = 4 4) , cos 3a = - 4r ( - )",
    a must now be found with the aid of the Trigonometrical
    tables.
    490 The roots of the cubic will be
    n cos a , n cos (žrta), n cos (žr — a ).
    491 Observe that, according positive or nega
    tive, Cardan 's method or the Trigonometrical will be practi
    cable. In the former case, there will be one real and two
    imaginary roots ; in the latter case, three real roots .

    BIQUADRATIC EQUATIONS.

    492 Descartes' Solution. — To solve the equation


    . * + 992 + rx + s = 0 .................. (i.)
    the term in re having been removed by the method of (429).
    Assume (x? + ex + f) (w ? — ex + g ) = 0 ............... (ii.)
    Multiply out, and equate coefficients with (i.) ; and the fol
    lowing equations for determining f, g , and e are obtained
    & +f = q + e", g -f = gf = 8 .........(iii.)
    493 @$ + 2qe* + (q — 48) e? —? ? = 0................ (iv.)
    494 The cubic in eº is reducible by Cardan's method , when the biquadratic
    has two real and two imaginary roots. For proof, take a tiß and matyas
    the roots of ( i.) , since their sum must be zero . Form the sum of each pair
    for the values of eſsee ( ii. ) ], and apply the rules in (488 ) to the cubic in e .
    If the biquadratic has all its roots real, or all imaginary, the cubic will have
    all its roots real. Take a + iß and - a Fiy for four imaginary roots of (i.),
    and form the values of e as before.

    495 If a’, f", y : be the roots of the cubic in e’, the roots of the biquadratic
    will be – } (a + 6 + y) , } (a + ß - y), Į (B + y - a ), } (x + a - B).
    S
    130 THEORY OF EQUATIONS.

    For proof, take w , x , y, z for the roots of the biquadratic ; then , by (ii.), the
    sum of each pair must give a value of e. Hence, we have only to solve the
    symmetrical equations
    y + z = a, w + x =
    z + x = ß, w + y =
    x + y = y, w + z = - Y

    +
    496 Ferrari's solution. To the left member of the equation
    X* + px? + qx + + rx + s = 0 ,
    add the quantity ax® + bx + , and assumethe result
    = ( *+ x + m)
    497 Expanding and equating coefficients, the following
    cubic equation for determining in is obtained
    8m — 4qmº + (2pr - 88) m + 4qs - pºs - p = 0 .
    Then æ is given by the two quadratics
    me+ x + m = + 240 to
    498 The cubic in m is reducible by Cardan's method when the biquadratic
    has two real and two imaginary roots. Assume a , 6 , y , & for the roots of the
    biquadratic ; then aß and yò are the respective products of roots of the two
    quadratics above. From this find m in terms of apya.

    499 Euler 's solution . — Remove the term in 23 ; then we


    have . * + qx? + rx + 8 = 0.
    500 Assume x = y + z + u , and it may be shewn that y , z ,
    and u are the roots of the equation

    # + ft + Toe = 0.
    501 The six values of y, z, and u, thence obtained , are
    restricted by the relation yzu =
    Thus x = y + z + u will take four different values.
    COMMENSURABLE ROOTS. 131

    COMMENSURABLE ROOTS.
    502 To find the commensurable roots of an equation.
    First transform it by putting x = into an equation of
    theforma ." + pian-1 + pqxn-2 + ... + p = 0 ,
    having Po = 1,and the remaining coefficients integers. (431)
    503 This equation cannot have a rational fractional root,
    and the integral roots may be found by Newton 's method of
    Divisors (459).
    These roots, divided each by k , will furnish the commen
    surable roots of the original equation .

    504 EXAMPLE. — To find the commensurable roots of the equation


    81x " — 207x * – 90° + 89x² + 2x – 8 = 0 .
    Dividing by 81, and proceeding as in (431), we find the requisite substitu
    tion to be
    The transformed equation is
    yö — 23y* — 9y® + 801y2 + 162y – 5832 = 0.
    The roots all lie between 24 and — 34, by (451).
    The method of divisors gives the integral roots
    6 , – 4 , and 3 .
    Therefore, dividing each by 9, we find the commensurable roots ofthe original
    equatiou to be į , - %, and j.
    505 To obtain the remaining roots ; diminish the transformed equation by
    the roots 6 , — 4 , and 3, in the following manner (see 427) :
    11 - 23 – 9 + 801 + 162 - 5832
    6 6 - 102 — 666 + 810 + 5832
    | 1 - 17 - 111 + 135 + 972
    - 4 + 84 + 108 - 972
    1 - 21 - 27 + 243
    3 3 - 54 - 243
    11 - 18 – 81
    The depressed equation is therefore
    y ? – 18y – 81 = 0 .
    Themm
    roots
    ens
    of which are 9 ( 1 + 2 ) and 9 ( 1 - 12) ; and, consequently, the
    ensurable roots of the proposed equation are 1 + 2 and 1 - 12.
    132 THEORY OF EQUATIONS.

    INCOMMENSURABLE ROOTS.
    506 Sturm ' s Theorem . - If f (a ) , freed from equal roots, be
    divided by f'(x), and the last divisor by the last remainder ,
    changing the sign of each remainder before dividing by it,
    until a remainder independent of x is obtained , or else a re
    mainder which cannot change its sign ; then f (x ), f' (x ) , and
    the successive remainders constitute Sturm 's functions, and
    are denoted by f (x ), fi(x ), fi(x ), & c. ...... fm (x ).
    The operation may be exhibited as follows:
    f (x) = 9. fi (x ) - f; (x ),
    fi (x ) = 93f (x ) – f (x ),
    f (x ) = 93f (x ) - f (w ),
    ... ... .. .
    fm-2(x) = \m-1fm-1(x ) – Fm (« ).
    507 NOTE . — Any constant factor of a remainder may .be
    rejected , and the quotient may be set down for the corres
    ponding function.
    508 An inspection of the foregoing equations shews
    ( 1) That fm (a ) cannot be zero ; for, if it were, f (a ) and
    fi (a ) would have a common factor, and therefore f (a ) would
    have equal roots, by (432 ).
    ( 2 ) Two consecutive functions, after the first, cannot
    vanish together ; for this would make fm (@ ) zero.
    (3 ) When any function, after the first , vanishes, the two
    adjacent ones have contrary signs.
    509 If, as a increases, f (x ) passes through the value zero,
    Sturm ' s functions lose one change of sign .
    For, before f (x ) takes the value zero , f ( x ) and fi (x ) have contrary signs,
    and afterwards they have the same sign ; as may be shewn by making h
    small, and changing its sign in the expansion of f (x + h ), by (426 ).
    510 If any other of Sturm 's functions vanishes, there is
    neither loss nor gain in the number of changes of sign .
    This will appear on inspecting the equations.
    511 RESULT. — The number of roots of f (a ) between a and b is
    equal to the difference in the number of changes of sign in
    Sturm 's functions, when x = a and when a = .
    INCOMMENSURABLE ROOTS. 133

    512 CoR. — The total number of roots of f (x ) will be found


    by taking a = too and b = - 00 ; the sign of each function
    will then be the same as that of its first term .
    - When the number of functions exceeds the degree of f (20 )
    by unity , the two following theorems hold :
    513 If the first terms in all the functions, after the first, are
    positive; all the roots of f (@ ) are real.
    514 If the first terms are not all positive ; then , for every
    change of sign, there will be a pair of imaginary roots.
    For the proof put x = t co and - , and examine the number of
    changes of sign in each case, applying Descartes' rule . (416 ) .
    515 If $ ( r) has no factor in common with f (w ), and if • (a )
    and fº(@ ) take the same sign when f (a ) = 0); then the rest of
    Sturm 's functions may be found from f (a ) and • (a ), instead
    of f'(w). functions
    For the reasoning in (509) and (510) will apply to
    the new .
    516 If Sturm 's functions be formed without first removing
    equal roots from fæ), the theorem will still give the number
    of distinct roots , without repetitions, between assigned limits.
    For if f (x ) and fi ( 2 ) be divided by their highest common factor (see 444 ),
    and if the quotients be used instead of f (a ) and fi (ic) to form Sturm 's func
    tions ; then , by (515 ) , the theorem will apply to the new set of functions,
    which will differ only from those formed from f (x ) and fi (a ) by the absence
    of the same factor in every term of the series.
    517 EXAMPLE. — To find the position of the roots of the equation
    2 * — 4x3 + 2 + 6x + 2 = 0.
    Sturm 's functions, formed according to f (x ) = 2 * — 4a8 + 2a + 6x + 2
    the rule given above, are here calculated . fi (w ) = 2X * - 6x* + 2 + 3
    The first terms of the functions are all f (x ) = 5.x * — 10x — 7
    positive; therefore there is no imaginary f (a ) = 2 - 1
    root.
    The changes of
    sign in the func a = - 2 oli 2
    tions, as 2 passes
    T

    through integral
    values, are exhi f (x ) =
    11+
    + + + + +

    bited in theadjoin fi(w ) =


    1+

    11+

    ing table . There · f (x) =


    +1

    are two changes of


    f: (x ) =
    +1

    sign lost while a


    + 1+

    passes from - 1 to fo (a ) =
    0 , and two more
    lost while x passes
    from 2 to 3 . There I Toffichanges
    of sign ... .. . S}| 4 4 2 2
    134 THEORY OF EQUATIONS.

    are therefore two roots lying between 0 and - 1 ; and two roots also between
    2 and 3.
    These roots are all incommensurable, by (503).

    518 Fourier 's Theorem . - Fourier’s functions are the fol


    lowing quantities f(x), f'(x ), F"(x) ...... "(« ).
    519 Properties of Fourier's functions. — As increases,
    Fourier 's functions lose one change of sign for each root of
    the equation f (x ) = 0 , through which x passes, and r changes
    of sign for r repeated roots.
    520 If any of the other functions vanish , an even number
    of changes of sign is lost .

    521 RESULTS . — The number of real roots of f (x ) between a


    and ß cannot be more than the difference between the number
    of changes of sign in Fourier' s functions when x = a , and the
    number of changes when x = ß .
    522 When that difference is odd , the number of intermediate
    roots is odd , and therefore one at least .
    523 When the same difference is even , the number of inter
    mediate roots is either even or zero.

    524 Descartes' rule of signs follows from the above for the
    signs of Fourier 's functions, when x = 0 are the signs of the
    terms in f (x ) ; and when x = 00 , Fourier's functions are all
    positive.

    525 Lagrange's method of approximating to the incom


    mensurable roots of an equation .
    Let a be the greatest integer less than an incommen
    surable root of f ( x ). Diminish the roots of f (x ) by a . Take
    the reciprocal of the resulting equation . Let b be the greatest
    integer less than a positive root of this equation. Diminish
    the roots of this equation by b , and proceed as before.
    526 Let a , b , c , & c . be the quantities thus determined ; then ,
    an approximation to the incommensurable root of f (x ) will be
    x = a + 1 1
    the continued fraction
    b + c +
    INCOMMENSURABLE ROOTS. 135

    5217 Newton 's method of approximation . - If c, be a quantity


    a little less than one of the roots of the equation f (x ) = 0 , so
    that f (a + h ) = 0 ; then cı is a first approximation to the
    value of the root. Also because
    f(a + h ) = f (c ) + hf'(q ) + * f" (C1) + & c....... (426 ),
    and h is but small, a second approximation to the root willbe

    In the same way a third approximation may be obtained from


    Ca, and so on .

    528 Fourier' s limitation of Newton 's method . — To ensure


    that C1, C2, C3, & c . shall successively increase up to the value
    Gth without passing beyond it , it is necessary for all values
    of x between c, and cth.
    (i.) That f (x ) and f ' ( ) should have contrary signs.
    (ii.) That f (a ) and f " (« ) should have the same sign .

    70 N

    Fig . 1. Fig . 2.
    A proof may be obtained from the figure. Draw the curve
    y = f (x ). Let OX be a root of the equation , and ON = C ;
    draw the successive ordinates and tangents NP, PQ , QR , & c .
    Then OQ = C , OS = Cz, and so on .
    Fig . (2 ) represents cz > OX , and the subsequent approxi.
    mations decreasing towards the root.

    530 Newton 's Rule for Limits of the Roots. - Let the co
    efficients of f (x ) be respectively divided by the Binomial
    coefficients, and let ao, ar, ag ... an be the quotients , so that

    f(x) = @o2a + nayan -1 + " (1


    1 .52 ) agæn- + ... + nan _jæ + Ano
    136 THEORY OF EQUATIONS.

    Let A1, A2, A3... Ar be formed by the law A , = 0 ; - 02-10. +1.


    Write the first series of quantities over the second , in the fol
    lowing manner :
    do, di, 22, az .... .. An - 1, ang
    A , A , A, Ag ...... An-1, An.
    Whenever two adjacent terms in the first series have the
    same sign , and the two corresponding terms below them in
    the second series also the same sign ; let this be called a
    double permanence. When two adjacent terms above have
    different signs, and the two below the same sign , let this be
    known as a variation -permanence.
    531 RULE . — The number of double permanences in the asso
    ciated series is a superior limit to the number of negative roots
    of f (a ) .
    The number of variation -permanences is a superior limit to
    the number of positive roots.
    The number of imaginary roots cannot be less than the
    number of variations of sign in the second series.

    532 Sylvester 's Theorem . — Let f (x + 1) be expanded by


    (426 ) in powers of x , and let the two series be formed as in
    Newton 's Rule (530 ).
    Let P (a ) denote the number of double permanences.
    Then P (a ) ~ P (u ) is either equal to the number of roots
    of f (a ), or surpasses that number by an even integer.
    NOTE. — The first series may be multiplied by n , and will
    then stand thus,
    fr (1), fn-1(4), 2 fn-2(1), (3 fn-8( ) ... In f(x).
    The second series may be reduced to
    SECO

    Gm(1), Gn-1 (1), Gn-2 (1) ... G (X ),


    U - 1

    fp 1 fr + 1
    where ге
    G , (1) = {fr(1)}" – " n= "- +po 1 fr-1 (1) fr+1 (1).
    533 Horner's Method . — To find the numerical values of the
    roots of an equation . Take, for example , the equation
    2 * — 4x3 + x2 + 6x + 2 = 0 ,
    and find limits of the roots by Sturm 's Method or otherwise.
    INCOMMENSURABLE ROOTS. 137

    It has been shewn in (517) that this equation has two


    incommensurable roots between 2 and 3 . The process of
    calculating the least of these roots is here exhibited .
    elebele Florovolo

    +6 + 2 ( 2:414213
    -6
    A, 20000
    -6 - 19584
    B, – 6000 4160000
    1104 - 2955839
    100 -- 4896 Ag 12041610000
    Labels

    176 1872 - 11437245184


    276 B , – 3024000 604364816
    192 68161 - 566003348
    468 - 2955839 38361468
    Lego

    208 68723 - 28285470


    48 67600 B , – 2887116000 10075998
    561 27804704 - 8485368
    68161 - 2859311296 1590630
    562 27895072
    D, 560 68723 B , - 2831416224
    563 139948 282843 ) 1590630 ( 562372
    567 6928600 - 283001674 1414215
    22576 139970 28284 ) 176415
    562 6951176 By – 282861704 169706
    22592 700 2828 ) 6709
    563 6973768 - 28285470 5657
    22608 700 282 ) 1052
    D, 5640 6996376 Bo — 28284770 848
    11 21 28 ) 204
    5644 69974 - 282845621 197
    11 2) 7
    NIC

    5618 69985 B , - 2828435


    11
    5652 Cp 69996
    Co 7
    D. 5,656 Root = 2:414213562372.
    METHOD. — Diminish the roots by 2 in themanner of (427).
    The resulting coefficients are indicated by A , B , C , D,.
    By Newton's rule (527), - | © ; that is , - 4. is an approximation to
    the remaining part of the root. This gives : 3 for the next figure ; •4 will be
    found to be the correct one. The highest figure must be taken which will
    not change the sign of A .
    Diminish the roots by :4. This is accomplished most easily by affixing
    ciphers
    of ' 4 .
    to A , B1, C , D , in the manner shewn, and then employing 4 instead
    Having obtained Ag, and observing that its sign is + , retrace the steps,
    138 TIIEORY OF EQUATIONS.

    trying 5 instead of 4. This gives A , with a minus sign , thereby proving the
    existence of a root between 2- 4 and 2-5 . The new coefficients are A2, B ., C2, Dz.
    - 1 gives 1 for the next figure of the root.
    B,
    Affix ciphers as before, and diminish the roots by 1, distinguishing the
    new coefficients as A3, B3, C3, Ds.
    Note that at every stage of the work A and B must preserve their signs
    unchanged. If a change of sign takes place it shews that too large a figure
    has been tried .
    To abridge the calculation proceed thus : - After a certain number of
    figures of the root have been obtained ( in this example four), instead of
    adding ciphers cut off one digit from B , two from C ., and three from Dr.
    This amounts to the same thing as adding the ciphers, and then dividing
    each number by 10000 .
    Continue the work with the numbers so reduced, and cut off digits in like
    manner at each stage until the D and C columns have disappeared .
    A, and By now alone remain , and six additional figures of the root are
    determined correctly by the division of A , by By.
    To find the other root which lies between 2 and 3, we proceed as follows:
    After diminishing the roots by 2, try 6 for the next figure. This gives A,
    negative ; 7 does the same, but 8 makes A , positive. That is to say, f (2.7)
    is negative, and f (2 : 8 ) positive. Therefore a root exists between 2: 7 and
    2:8, and its value may be approximated to , in the manner shewn.
    Throughout this last calculation A will preserve the negative sign .
    Observe also that the trial number for the next figure of the root given at
    each stage of the process by the formula – f (c)', will in this case be always
    too great, as in the former case it was always too small.

    SYMMETRICAL FUNCTIONS OF THE ROOTS OF


    AN EQUATION .

    NOTATION. — Let a , b , c ... be the roots of the equation


    f (x ) = 0 .
    Let sy denote am + 3m + ..., the sum of the math powers of
    the roots .
    Let Smp denotea" 1? + 2," a” + a " (P + ... through all the
    permutations of the roots , two at a time.
    Similarly let m.n., denote a " l c? + a" l" da + ..., taking all
    711 , P ,

    the permutations of the roots three at a time; and so on .


    SYMMETRICAL FUNCTIONS OF THE ROOTS. 139

    534 SUMS OF TIIE POIVERS OF THE ROOTS.


    Sm + Pism - i + p2Sm - 2 + ... + pm - 1 $i + mpm = 0 ,
    where m is less than n, the degree of f (a).
    Obtained by expanding by division each term in the value of f (x ) given
    at (432), arranging the whole in powers of x , and equating coefficients in the
    result and in the value of f' ( ), found by differentiation as in (421) .
    535 If m be greater than n , the formula will be
    Sm + pism -i + p2 Sm - 2 + ... + P , Sm - n = 0.
    Obtained by multiplying f (x ) = 0) by 2 * -", substituting for æ the roots
    a , b, c, & c . in succession , and adding the results .
    By these formulæ 81, 82, 83, & c. may be calculated successively .
    536 To find the sum of the negative powers of the roots, put
    m equal to n - 1 , n — 2 , n - 3 , & c. successively in (535 ) , in
    order to obtain s - 1, S - 2, S -3 , & c .

    537 To calculate s , independently .


    Rule : 8, = – x coefficient of x-" in the expansion of
    log ! ") in descending powers of x.
    Proved by taking f (x ) = (x - a ) (x – b ) (x — c ) ..., dividing by x “, and
    expanding the logarithm of the right side of the equation by (156) .

    538 SYMMETRICAL FUNCTIONS WHICII ARE


    NOT POIVERS OF THE ROOTS.
    These are expressed in terms of the sums of powers of
    the roots as under, and thence , by (534 ) , in termsof the roots
    explicitly ,
    Sn,p = $m.S; — 8min tp
    m , p m
    +p,
    539 8m,p,q = 8m8, 8, — Sm +pS, - Sm +qS, - $ +q.8m + 28m +p +q:
    The last equation may be proved by multiplying smo by
    8q ; and expansions of other symmetrical functions may be
    obtained in a similar way .

    540 If & (r) be a rational integral function of a, then the


    symmetrical function of the roots of f ( r ), denoted by
    140 THEORY OF EQUATIONS.

    • (a) + (1 ) + + (c ) + & c., is equal to the coefficient of 22-1 in


    the remainder obtained by dividing ♡ (x) $”(r ) by f ( x).
    Proved bymultiplying the equation (432) by 0 (0%),and by theorem (401).

    541 To find the equation whose roots are the squares of


    the differences of the roots of a given equation .
    Let F (a ) be the given equation , and S , the sum of the poth
    powers of its roots . Let f ( x) and s, have the same meaning
    with regard to the required equation .
    The coefficients of the required equation can be calculated
    from those of the given one as follows :
    The coefficients of each equation may be connected with the
    sums of the powers of its roots by (534 ) ; and the sums of the
    powers of the roots of the two equations are connected by the
    formula
    2r ( 2r - 1 ) ss - Ins. ..
    542 28, = nS»; – 2- S,Sr-1 + 2 1 .2 1) S. S2r-, - ... + nSer
    RULE . — 2s, is equal to the formal expansion of ( S – S ) 2r by
    the Binomial Theorem , with the first and last terms each mul
    tiplied by n , and the indices all changed to sufixes . As the
    equi-distant terms are equal we can divide by 2 , and take half
    the series.
    DEMONSTRATION. — Let a , b , c ... be the roots of F (x ).
    Let Q (x ) = (x — a )?! + (x - )2° + ... .... ........... (i.)
    Expand each term on the right by the Bin . Theor., and add, substituting
    S , S , & c. In the result change x into a , b , c ... successively, and add then
    equations to obtain the formula , observing that, by (i.),
    ♡ (a ) + ° ( ) + ... = 28 .
    If n be the degree of F (x), then in (1 - 1) is the degree
    of f ( ). By (96 ).
    543 The last term of the equation f (@ ) = 0 is equal to
    n ” F (a ) F' (B ) F ( y) ...
    where a, b, y, ... are the roots of I"(a). Proved by shewing
    that F ' (a ) F" (1) ... = n ” F (a ) F ( ) ...
    544 If F (x ) has negative or imaginary roots, f (x) must
    have imaginary roots.
    SYMMETRICAL FUNCTIONS OF THE ROOTS. 141

    545 The sum of the mth powers of the roots of the quad
    ratic equation XP - px + q = 0 .
    8. = pm -mpm -2g + m (m2 - 3)? np -4q- — ...
    1

    ...+ (- 1)"mm (m —- rp —- 1)... 2r + 1) pm - 2r0 + & c.


    1) ... (m — 2r+

    By (537) expanding the logarithm by (156 ).

    546 The sum of the mth powers of the roots of x " — 1 = 0


    is n if m be a multiple of n , and zero if it be not.
    By (537) ; expanding the logarithm by (156 ) .

    547 If (x) = ao + aza + azaº + & c. ............... (i.),


    then the sum of the selected terms
    Qmam + Qm+nQm+n + am +an wm +2n + & c.
    willbe s = {an-m * (ar) + B »-m * (Bx) + yn-m * (yx ) + & c.}
    where a, b , y, & c. are the nth roots of unity.
    For proof, multiply (i.) by a" -m , and change w into ax ; so with B , Y, & c .,
    and add the resulting equations.

    548 To approximate to the root of an equation by means of


    the sums of the powers of the roots.
    m + 1
    By taking m large enough, the fraction 8m+1
    Sm
    will approx X

    imate to the value of the numerically greatest root, unless


    there be a modulus of imaginary roots greater than any real
    root, in which case the fraction has no limiting value.

    549 Similarly the fraction SmSm +2 — Sm +1 approximates, as m


    Sm – 19m +1s.*
    increases, to the greatest product of any pair of roots, real or
    imaginary ; excepting in the case in which the product of the
    pair of imaginary roots, though less than the product of the
    two real roots , is greater than the square of the least of them ,
    for then the fraction has no limiting value.
    142 TIIEORY OF EQUATIONS.

    550 Similarly the fraction Sm8m +3 — 8m +18m +2 approximates ,


    SmoSm + 2 - 3m + 1
    as m increases , to the sum of the two numerically greatest
    roots , or to the sum of the two imaginary roots with the
    greatest modulus.

    EXPANSION OF AN IMPLICIT FUNCTION OF x.

    Let yº (Ax* + ) + y® ( B w + ) + ... + yº(Sx + ) = 0...... (1)


    be an equation arranged in descending powers of y , the co
    efficients being functions of X , the highest powers only of e
    in each coefficient being written .
    It is required to obtain y in a series of descending powers
    of x .
    First form the fractions
    a -b a - C a - d a - s
    an - a - 8 . . .. . Q - 0

    Let - = = t be the greatest of these algebraically , or


    if several are equal and greater than the rest, let it be the
    last of such . Then , with the letters corresponding to these
    equal and greatest fractions, form the equation
    Au° + ...... + Ku" = 0 ..................(3).
    Each value of u in this equation corresponds to a value of y ,
    commencing with wa'.
    Next select the greatest of the fractions
    -1 km k - $
    . .. ... ... ( 1 ).
    KT K -u " K
    1: -
    Let - " _ " = ť be the last of the greatest ones. Form
    K -v
    the corresponding equation Ku" + ... + Nu' = 0 .......... (5 ).
    Then each value of u in this equation gives a corresponding
    value of y , commencing with ua" .
    EXPANSION OF AN IMPLICIT FUNCTION. 143

    Proceed in this way until the last fraction of the series (2)
    is reached .
    To obtain the second term in the expansion of y , put
    y = x (u + yı) in ( 1) .....................(6 ),
    employing the different values of u , and again of ť and u , t"
    and u , & c. in succession ; and in each case this substitution
    will produce an equation in y and a similar to the original
    equation in y .
    Repeat the foregoing process with the new equation in y,
    observing the following additional rule :
    When all the values of t, t', t" , sc. have been obtained , the
    negative ones only must be employed in forming the equations
    in u . (7 ).
    552 To obtain y in a series of ascending powers of x .
    Arrange equation (1 ) so that a , ß , y, & c . may be in as
    cending order of magnitude, and a, b , c , & c . the lowest powers
    of e in the respective coefficients.
    Select t, the greatest of the fractions in ( 2 ), and proceed
    exactly as before, with the one exception of substituting the
    word positive for negative in ( 7 ).

    553 EXAMPLE .— Take the equation


    (28 + * *) + ( 3c? – 52 ) y + ( - 4x + 72°+ 2°) y? – y = 0.
    It is required to expand y in ascending powers of x .
    The fractions (2 )) are
    on – 03 -- 21 03 -- 21 ö3 _- 05 ; or 1, 1, and
    The first two being equal and greatest , we have t = 1.
    The fractions (4 ) reduce to - 2 – 5 = ý = ť .
    Equation ( 3 ) is 1 + 34 – 4u* = 0 ,
    which gives u = 1 and - ), with t = 1.
    Equation (5 ) is - 4u² — 20 = 0 ,
    and from this u = 0 and – 41, with ť = }.
    Wehave now to substitute for y, according to (6 ), either
    * ( 1 + y ), « ( - 1 + y ), afy, or ( - 4 + y ).
    Put y = x (1 + y ), the first of these values, in the original equation, and
    arrange in ascending powers of y , thus
    – 4 «* + ( - 528 + ) y + ( - 42 + ) y; – 10x*y; – 5x®y; – 2y = 0.
    The lowest power only of x in each coefficient is here written .
    144 THEORY OF EQUATIONS.

    The fractions ( 2) now become


    4 -3 4 -3 4 - 5 4 – 5 4 – 5
    0-1 0- 2 0 -3 0 -4 0 - 5°
    or 1, }, - , - - š.
    From these t = 1, and equation (3 ) becomes
    — 4 — 5u = 0 ; i. u = - 4 .
    Hence one of the values of y, is, as in (6 ), y = & ( - * + y ).
    Therefore y = x { 1 + x ( - 4 + y )} = * - *x* + ...
    Thus the first two terms of one of the expansions have been obtained.

    DETERMINANTS.

    554 Definitions. — The determinant | a, az is equivalent


    b , b₂
    to a ,b , - a , b , and is called a determinant of the second order.
    A determinant of the third order is
    | a, a, az 1 = a ,(bạcz — 63C2) + a )(634 —60) + az (b.cz – b2cı).
    | b , b, b
    | Ci C, C3
    Another notation is a ,b , cz, or simply (a ,b ,c ).
    The letters are named constituents , and the terms are
    called elements. The determinant is composed of all the
    elements obtained by permutations of the suffixes 1 , 2 , 3 .
    The coefficients of the constituents are determinants of
    the next lower order, and are termed minors of the original
    determinant. Thus, the first determinant above is the minor
    of cz in the second determinant. It is denoted by Cz. So the
    minor of a , is denoted by A1, and so on .

    555 A determinant of the nth order may be written in either


    of the forms below
    | a, 2 ... Q ... An | au 012 ... Qır ... Qin |
    | bi bz . .. bz . .. 021 022 ... dar ... Azn
    ... . .. . .. ... ... ... ... ...
    12 ... I ... In ! Ani An2 ... A nr ... Ann
    In the latter, or double suffix notation , the first suffix indicates
    the row , and the second the column. The former notation
    will be adopted in these pages.
    DETERMINANTS . 145

    A Composite determinant is one in which the || a, a , az ||


    number of columns exceeds the number of rows, || b , b2 bz ||
    and it is written as in the annexed example .
    Its value is the sum of all the determinants obtained by taking
    a number of rows in every possible way . exam Par
    i
    nanthas
    A Simple determinant suas orterms
    has single m le: tial,
    e thanfor its pconstituents.
    A Compound determinant has more than one term in some
    or all of its constituents. See (570) for an example .
    For the definitions of Symmetrical, Reciprocal, Partial,
    and Complementary determinants ; see (574) , (575) , and (576 ).

    General Theory.
    556 The number of constituents is na.
    The number of elements in the complete determinant is n .

    557 The first or leading element is aybacz ... In. Any


    element may be derived from the first by permutation of the
    suffixes.
    The sign of an element is + or – according as it has
    been obtained from the diagonal element by an even or odd
    number of permutations of the suffixes.
    Hence the following rule for determining the sign of an
    element.
    RULE. — Take the suffixes in order, and put them back to
    their places in the first element . Let m be the whole number
    of places passed over ; then ( - 1)" will give the sign required .
    Ex. - To find the sign of the element a ,b ,c,d ,ez of the determinant
    (a,bączd , ez).
    ay bg Cz d , eg
    Move the suffix 1, three places... ... 1 4 3 5 2
    , 2 , three places. .. . .. 1 2 4 3 5
    , , 3, one place ... ... 1 2 3 4 5
    In all, seven places ; therefore ( - 1) = - 1 gives the sign required .

    558 If two suffixes in any element be transposed , the sign


    of the element is changed .
    Half of the elements are plus, and half are minus.

    559 The elements are not altered by changing the rows into
    columns.
    If two rows or columns are transposed , the sign of the
    146 THEORY OF EQUATIONS.

    determinant is changed. Because each element changes its


    sign .
    If two rows or columns are identical, the determinant
    vanishes .
    560 If all the constituents but one in a row or column
    vanish , the determinant becomes the product of that con
    stituent and a determinant of the next lower order .

    561 A cyclical interchange is effected by n - 1 successive


    transpositions of adjacent rows or columns, until the top row
    has been brought to the bottom , or the left column to the
    right side. Hence
    A cyclical interchange changes the sign of a determinant
    of an even order only .
    The pth row may be brought to the top by r - 1 cyclical
    interchanges .

    562 If each constituent in a row or column bemultiplied


    by the same factor, the determinant becomes multiplied by it.
    If each constituent of a row or column is the sum of m
    terms, the compound determinant becomes the sum of m
    simple determinants of the same order .
    Also, if every constituent of the determinant consists of
    m terms, the compound determinant is resolvable into the sum
    of mº simple determinants.

    563 To express the minor of the oth row and leth column as
    a determinant of the n - 1th order.
    Put all the constituents in the 7th row and kth column equal
    to 0 , and then make r - 1 cyclical interchanges in the rows
    oooo

    and l - 1 in the columns, and multiply by ( - 1 ) (- + k)(n - 1).


    [:: = ( - 1) ^-1+k-1)(x-1).

    564 To express a determinant as a deter. 1000


    minant of a higher order. a 1 0 0
    . Continue the diagonal with constituents Be a
    of " ones," and fill up with zeros on one side, y Śh
    and with any quantities whatever (a , b , y , & c .)
    60

    on the other.
    DETERMINANTS. 147

    565 The sum of the products of each constituent of a


    column by the corresponding minor in another given column
    is zero. And the same is true if we read ' row ' instead of
    column. Thus, referring to the determinant in (555 ),
    Taking the pth and qth columns, Taking the a and c rows,
    Q ,A , + b, B , + ... +1L = 0 . aC + a ,C2+ ... + a , C , = 0 .
    For in each case wehave a determinant with two columns identical.
    566 In any row or column the sum of the products of each
    constituent by its minor is the determinant itself. That is ,
    Taking the pth column, Or taking the c row ,
    QA,+6,B,+ ... +11, = A. CC + cąc,+ ...+c,C, = A.
    567 The last equation may be expressed by Ec, C , = A .
    Also, if (apca) express the determinant | ap aq li then
    (agca) will represent the sum of all the determinants of the
    second order which can be formed by taking any two columns
    out of the a and c rows. The minor of (ap, C ) may be written
    (4p, Ca) , and signifies the determinant obtained by suppress
    ing the two rows and two columns of an and Co. Thus
    A = E (ap, Ca) ( A , , C . ). And a similar notation when three or
    more rows and columns are selected .

    568 Analysis of a determinant.


    RULE. — To resolve into its elements a determinant of the
    non order . Express it as the sum of n determinants of the
    (n - 1)th order by (560 ), and repeat the process with each of
    the new determinants .
    EXAMPLE :
    ay as ag an = a, b, b , b , 1 – 0 , 1 bg b, b , + az be bg bz – ag | b, b, bg
    b , b, b, b , Ca Cg Co Cg C G Ice Cz Cz G C₂ C₃
    G Ca Cg Id₂ d . d , l ld , d , d , 1 I da da dal
    Idy d , d , de
    Again, 1 b, b, b3 1 = b , ca Cs + bą | Cg C1 | + b3 C C |
    la cq c₂ | Id, dal ld d l l d, dal
    I d dy dz
    and so on . In the first series the determinants have alternately plus and
    minus signs, by the rule for cyclical interchanges (561) , the order being even .
    148 THEORY OF EQUATIONS.

    569 Synthesis of a determinant.


    The process is facilitated by making use of two evident
    rules. Those constituents which belong to the row and
    column of a given constituent a , will be designated “ a 's con
    stituents." Also, two pairs of constituents such as Ap, C , and
    aq, Co, forming the corners of a rectangle, will be said to be
    “ conjugate ” to each other ,
    RULE I. — No constituent will be found in the same term
    with one of its own constituents .
    RULE II. — The conjugates of any two constituents a and b
    will be common to a 's and l's constituents.
    Ex. — To write the following terms in the form of a determinant:
    abcd + bfyl + f ?k? + ledf + cghp + lahr + elpr
    - fhpr - ablr - ach? — lfhg - bdfa- efhl — cedp.
    The determinant will be of the fourth order ; and since every term must
    contain four constituents, the constituent 1 is supplied to make up the
    number in some of the terms. Select any term , as abcd , for the leading
    diagonal.
    Now apply Rule I.,
    a is not found with e, f,f, g, p , 0... (1). cis not found with f,f,1, r, 1,0 ...( 3).
    V is not found with e, h, h,p ,1,0...(2). d is not found with g,h,h,l,r,0...(4). .
    Each constituent has 2 (n - 1 ) , that is, 6 constituents belonging to it, since
    n 4 . Assuming, therefore, that the above letters are the constituents of
    a , b , c , and d , and that there are no more, we supply a sixth zero constituent
    in each case.
    Now apply Rule II. — The constituents common
    to u and I are e, p ; to a andcfifi to b and ~ 1, 0 ;
    to a and d — 4, 0 ; to b and deh,1 , 0 ; to c and d - , r, 0.
    The determinant may now be formed . The diagonala e f gl
    being abcd ; place e, p, the conjugates of a and b, either as
    in the diagram or transposed . f0
    Then f and f, the conjugates of a and c,may be written. 10 h d
    1 and 0 , the conjugates of b and c, must be placed as indicated, because
    1 is one of p 's constituents, since it is not found in any term with p , and
    must therefore be in the second row .
    Similarly the places of g aud 0, and of l and r, are assigned.
    In the case of 1 and d we have h , h , 0 from which to choose the two
    conjugates, but we see that U is not one of them because that would assign
    two zero constituents to b , whereas l has but one, which is already placed.
    By similar reasoning the ambiguity in selecting the conjugates l, r is
    removed .
    The foregoing method is rigid in the case of a completo determinant
    DETERMINANTS. 149

    having different constituents. It becomes uncertain when the zero con


    stituents increase in number, and when several constituents are identical.
    But even then , in the majority of cases, it will soon afford a clue to the
    required arrangement.

    570 PRODUCT OF TWO DETERMINANTS OF


    THE nth ORDER .
    (P ) (Q )
    | a , a , . .. anl a a , . .. an = | A , A , . .. Anh
    6 , 62 ... on B . B , ... BmL B , B , ... B ,
    . .. . .

    | 4 l ... In | | 1 day ... del L , L, ... L


    The values of A1, B , ... Li in A1 = dya, ta, a, t ... tanan
    the first column of S are an - B = aß. + 0 , ß , + ... tanßn
    nexed . For the second column .. .. .. .. ... ... ... .
    write B 's in the place of a ' s . l = aid , ta, d , t . .. tanden
    For the third column write c' s ,
    and so on .
    For proof substitute the values of A ,, B , & c. in the determinant S , and
    then resolve S into the sum of a number of determinants by (562) , and note
    the determinants which vanish through having identical columns.
    RULE. — To form the determinant S , which is the product of
    two determinants P and Q . First connect by plus signs the
    constituents in the rows of both the determinants P and Q .
    · Now place the first row of P upon each row of Q in turn ,
    and let each two constituents as they touch become products .
    This is the first column of S.
    Perform the same operation upon Q with the second row of
    P to obtain the second column of S ; and again with the third
    row of P to obtain the third column of S , and so on .
    571 If the number of columns, both in P and Q , be n , and
    the number of rows r, and if n be > r, then the determinant
    S , found in the same way from P and Q , is equal to the sum
    of the C (n , r ) products of pairs of determinants obtained by
    taking any r columns out of P , and the corresponding r
    columns out of Q .
    But if n be < r the determinant S vanishes .
    For in that case, in every one of the component determinants, there will
    be two columns identical.
    150 THEORY OF EQUATIONS.

    572 The product of the determinants P and Q


    formed in four ways by changing the rows into columns in
    either or both P and Q .
    573 Let the following system of n equations in 2 , X , ... More
    be transformed by substituting the accompanying values of
    the variables ,
    azxz + azx , + ...tanwn = 0 , x = a & ta, &z + ... tan Eng
    bıxı + bqXz + ... + b xn = 0, x = = B.$ + B &2 + ... + B En,
    .. . .. . . .. .. .

    4Xz+ lax:+ ... +I»«» = 0, Xn = 1,$ t1,$ +...+ 1, n.


    The eliminant of the resulting equations in & 2 ... Er is
    the determinant S in (570 ) , and is therefore equal to the
    product of the determinants P and Q . The determinant Q is
    then termed the modulus of transformation .

    574 A Symmetrical determinant is symmetrical about the


    leading diagonal. If the R ’s form the pth row , and the K 's
    the kth row ; then Rx = K , throughout a symmetrical deter
    minant.
    The square of a determinant is a symmetrical determinant.
    575 A Reciprocal determinant has for its constituents the
    first minors of the original determinant, and is equal to its
    n - 1th power ; that is ,
    Aj ... An1 = | a , ... an |n -1 Proof. — Multiply both sides
    of the equation by the original
    determinaut (555 ). Tbe con
    stituents on the left side all
    L ... Ln vanish excepting the diagonal
    of A ' s .

    576 Partial and Complementary determinants.


    If ” rows and the same number of columns be selected
    from a determinant, and if the rows be brought to the top ,
    and the columns to the left side, without changing their order,
    then the elements common to the selected rows and columns
    form a Partial determinant of the order r , and the elements
    not found in any of those rows and columns form the Com
    plementary determinant, its order being n - r .
    DETERMINANTS. 151

    Ex. - Let the selected rows from the determinant (a ,b ,c, d ,es) be the
    second, third, and fifth ; and the selected columns be the third, fourth , and
    fifth. The original and the transformed determinants will be
    | ay ay ag an ag and l bg bn bob bą |
    by ba be bi bo Cg Cn Cyc Cz
    - Ca Ca C Ca | ez es es a la
    d da da da do as an as a la
    .. la Cg eg es es dz d . d d de
    The partial determinant of the third order is (6364es), and its comple
    mentary of the second order is (a, d,).
    The complete altered determinant is plus or minus, according as the
    permutations of the rows and columns are of the same or of different class .
    In the example they are of the same class, for there have been four trans
    positions of rows, and six of columns. Thus ( - 1) ^0 = + 1 gives the sign
    of the altered determinant.

    577 THEOREM . — A partial reciprocal determinant of the poth


    order is equal to the product of the — 1th power of the
    original determinant, and the complementary of its corres
    ponding partial determinant.
    Take the last determinant for an example. Here n = 5, r = 3 ; and by the
    theorem ,
    B, B, B, = A ' la ag ' where B , C , E are the
    Og Og Og dm dal respective minors.
    E, E, E 1
    PROOF. — Raise the Partial Reciprocal to the original order five without
    altering its value, by (564 ) ; and multiply it by A , with the rows and
    columns changed to correspond as in Ex. (576 ) ; thus, by (570), we have
    | B, B , B, B , B , 1 bg bn bs ib bq = 1100 b, b , 1 = A'| ay ay
    C, C, C C , C , Co Co CGG Cg TOA OCCg 1d, d, 1
    E, E, E; E , E , | e, е, е, е, е, 00 A & ex
    0 0 0 I ag a , a, a, as | 10o o a, az
    10 000 i I dg d , do 1 d d | Jooo dy de

    578 The product of the differences between every pair of n


    quantities a , a , .. . Ang
    (az -az)(a , - as)(az - a ) ... (an - an) 1 1 1 ... 1
    X (az - az)(az - a ,)... (az - an) a az az ... An
    (az- a,) ... (az - an) > = a a az ... a ;

    X (an-1 -an) ) am-1 am- -?...am-11


    Proof. — The determinant vanishes when any two of the quantities are
    152 THEORY OF EQUATIONS.

    equal. Therefore it is divisible by each of the factors on the left ; therefore


    by their product. And the quotient is seen to be unity , for both sides of the
    equation are of the same degree ; viz ., în (1 - 1 ) . i

    579 The product of the squares of the Si ...


    differences of the same n quantities S | $ $, . .. Sn
    PROOF. - Square the determinant in (578 ), and
    write s, for the sum of the pth powers of the roots. | Sn - 1 Sn .. . $2n- 2
    580 With the same meaning for S1, S2 . .. , the same deter
    minant taken of an order r, less than n , is equal to the sum
    of the products of the squares of the differences of r of the n
    quantities taken in every possible way ; that is , in C (n , r )
    ways.
    Ex . : | $, $i = (a - a )' + (a, - as)' + & c . = (Q , - a.)?,
    | 81 8g |
    | 8, 8, 8, | = (ay - az) ' ( 4 - ) (ag – az) .
    8, 89 8g
    | 82 83 84
    The next determinant in order
    = ! (ay — a ,)' (az - az)' ( az - a )? (az - az): (az - a .): (az - a .)”.
    And so on until the equation (579) is reached .
    Proved by substituting the values of $1, 8, ... & c ., and resolving the deter
    minant into its partial determinants by (571) .

    581 The quotient of


    dox ” tamm -it ... taxm -* + . ..
    bow " +601 -1 + ... + b, xn -- + ...
    is given by the formula
    qowm-- + 91.wm- -1 + ... + 0,2 * -*-* + ...,
    where _ 1b, 00 ... a ,
    6 +1 b bo 0 ... aj
    bab, bo ... az
    | br br-1 br-2 ... bja ,
    Proved by Induction .
    ELIMINATION. 153

    ELIMINATION .

    582 Solution of n linear equations in n variables.


    The equations and the values of the variables are arranged
    below :

    4, 8 + 2,3 , + ... + 1, 2, = $ , A = Ak + B % + ... + L %.


    b, xz + b2x2 + ... + bm Xn = $a, X , A = A $1 + B , $2 + ... + L , $n
    . .. ... .. . . ..
    haz + 1,xz+ ... + lm Xn = tn 2, 4 = Ar& + B ,& + ... + L Én
    where A is the determinant annexed , and A1, B1, la ... an
    & c . are its first minors . . .. . . . . ..
    lli ... In
    To find the value of one of the unknowns X.p.
    RULE. — Multiply the equations respectively by the minors
    of the pth column, and add the results. X , will be equal to the
    fraction whose numerator is the determinant A , with its poth
    column replaced by E1, E2 ... En, and whose denominator is A
    itself.
    583 If &,, ... En and A all vanish, then X , X, ... Xn are in
    the ratios of the minors of any row of the determinant A .
    For example, in the ratios C : Cz : Cz : ... : Cn.
    The eliminant of the given equations is now A = 0 .

    584 Orthogonal Transformation .


    If the two sets of variables in the n equations (582) be
    connected by the relation
    X , + xy + ... + x = 8 + 8 + ... + 8 ...... (1 ),
    then the changing from one set of variables to the other, by
    substituting the values of the E' s in terms of the a 's in any
    function of the former, or vice versâ , is called orthogonal
    transformation .
    When equation ( 1 ) is satisfied , two results follow .
    I . The determinant A = £1.
    154 THEORY OF EQUATIONS.

    II. Each of the constituents of A is equal to the corre


    sponding minor, or else to minus that minor according as A is
    positive or negative.
    PROOF. - Substitute the values of & , £g ... En in terms of x , X , ... Xn in
    equation ( 1) , and equate coefficients of the squares and products of the new
    variables . Weget the n ? equations
    a +b + = 1 ) aga, + baby + = 0 ) aga , + bgb, + = 0 )
    ajag + bybę + = 01 a +b + = 1 Qgaz + bgba + = 0
    ajag + 6,63 + = 01 ayag + bgbg + = 0 Y a + b + = ly
    ajan + bybn + = 0 ) agan + b2bm + = 0 ) Azan + bgbm + = 0 )
    Also A = a,b, ... Form the square of the determinant A by the
    a , b, ... l, rule (570 ), and these equations show that the
    product is a determinant in which the only con
    agbg... lg stituents that do not vanish constitute a diagonal
    .. . . . . .. of ' ones. Therefore
    arbn ... In A = 1 and A = + 1.
    Again , solving the first set of equations fa 4 = , + ,0 + 4 ,0 + = A ,
    for a , (writing a ; as a , a ,, & c.), the second azd = 4 ,0 + 4 , + 420 + = A ,
    set for an, the third for ag, and so on , we
    bave, by (582 ), the results annexed ; which azd = 4 ,0 + 4 ,0 + Ag + = A ,
    proves the second proposition . & c. & c.

    585 Theorem . — The n - 2th power of a determinant of the


    nth order multiplied by any constituent is equal to the corre
    sponding minor of the reciprocal determinant.
    PROOF. — Let p be the reciprocal determinant of A , and B , the minor of B ,
    in p . Write the transformed equations (582) for the x' s in terms of the E' s,
    and solve them for £g. Then equate the coefficient of x , in the result with its
    coefficient in the original value of .
    Thus pšg = A (1,2 + ... + $ & r + ...), and $, = 6,27 + ... + byxor + ... ;
    i AB, = pbp = An-1b by (575) ; . By = An-2 bp.

    586 To eliminate æ from the two equations


    arm + bwm -1 + cwm- 2 + ... = 0 ............ (1),
    a'x" + b'xn-1 + c'wn–2 + ... = 0 ......... ... ( 2 ) .
    If it is desired that the equation should be homogeneous
    ịn x and y ; put Y
    instead of w , and clear of fractions. The
    following methods will still be applicable .
    ELIMINATION . 155

    I. Bezout's Method .— Suppose m > n .


    Rule . — Bring the equations to the same degree by multi
    plying (2) by wom -n. Then multiply (1) by a', and (2) by a, and
    subtract.
    Again , multiply (1) by aʼx + b', and (2) by (ax + b), and
    subtract.
    Again,multiply (1) by a'u + b'x + c',and (2 ) by (ax2 + bx + c),
    and subtract, and so on until n equations have been obtained .
    Each will be of the degree m - 1.
    Write under these the m - n equations obtained by multi
    plying (2 ) successively by X. The eliminant of the m equations
    is the result required .
    Ex. — Let the equations be (a + b 2 * + cæ8 + d2 * + ex + f = 0,
    I am + b'xº + c'x + ď = 0.

    The five equations obtained by the method, and their eliminant, by (583),
    are,writing capital letters for the functions of a , b , c , d , e, f,
    4 ,2 * + 3 ,28 + Cjx ? + Djæ + E , = 0 | A , B , C, D, E = 0.
    Aux* + B92.8 + Cgot + Dqx + E , = 0 A, B, C, D ,
    and
    Azx * + B928 + Cga* + D3 + E , = 0 A , B, C, D,
    a' *+ 6 '2 + c m + d'x = 0 Ia b í ď0
    a' x + 6' + 6 x + ď' = 0 ) O a' b ' ŕď
    Should the equations be of the same degree, the eliminant will be a sym
    metrical determinant.

    587 II. Sylvester's Dialytic Method .


    RULE . — Multiply equation ( 1) successively by x, n - 1 times ;
    and equation ( 2 ) m - 1 times ; and eliminate x from themtn
    resulting equations.
    Ex. — To eliminate a from an® + bx? + cx + d = 0
    px + qx + = |
    Themtn equations and their eliminant are
    px* + qx + r = 0 100pqr1 = 0 .
    pg" + qdº + rw = 0 Op q r o
    pæ* + qad + race = and pqr00
    ac + bc + c + d = 0 a b c d
    ax * + bx® + cx* + dx = 0 a b c do
    156 THEORY OF EQUATIONS.

    588 III. Method of elimination by Symmetrical Functions.


    Divide the two equations in (586) respectively by the
    coefficients of their first terms, thus reducing them to the
    forms f (x ) = 2 * + pi.am-1 + ... + Pm = 0,
    $ (x ) = x " + 9189-1 + ... + qn = 0 .
    RULE. — Let a , b , c ... represent the roots of f (x ). Form
    the equation • (a ) • (6 ) • (c ) ... = 0. This will contain sym
    metrical functions only of the roots a , b , c ... .
    Express these functions in terms of P1, P2 ... by (538), & c.,
    and the equation becomes the eliminant.
    Reason of the rule. — The eliminant is the condition for a common root of
    the two equations. That root must make one of the factors q ( a ) , o (6 ) ...
    vanish , and therefore it makes their product vanish .

    589 The eliminant expressed in terms of the roots a , b , c ...


    of f (x ), and the roots a , ß , y ... of o (a ), will be
    (a – a) (a – B) (a — y) ... (b — a) (6 - 3) (6 — y) ... &c.,
    being the product of all possible differences between a root of
    one equation and a root of another.
    590 The eliminant is a homogeneous function of the co
    efficients of either equation , being of the nth degree in the
    coefficients of f (x) , and of the mth degree in the coefficients
    of • (a ).
    591 The sum of the suffixes of p and q in each term of the
    eliminant = mn. Also, if p , q contain % ; if P2, 92 contain z ;
    if P3, 93 contain 7 , and so on , the eliminant will contain zmn
    Proved by the fact that p, is a homogeneous function of r dimensions of
    the roots a, b, c ..., by (406 ) .

    592 If the two equations involve x and y, the elimination


    may be conducted with respect to x ; and y will be contained
    in the coefficients P1, P2 ..., 91, 92 ... .

    593 Elimination by the Method of Highest Common Factor.


    Let two algebraical equations in x and y be represented by
    A = 0 and B = 0 .
    ELIMINATION . 157

    It is required to eliminate .
    Arrange A and B according to descending powers of x ,
    and, having rejected any factor which is a function of y only,
    proceed to find the Highest Common Factor of A and B .
    The process may be exhibited as follows :
    GA = 9. B + r7R ) C1, C2, C3, C, are the multipliers re
    coB = 0 , R , tr. R . quired at each stage in order to avoid
    fractional quotients ; and these must
    CzR, = 93R, + rsRg be constants or functions of y only.
    GR, = 9,Rx+ ro 91, 92, 93, 94 are the successive quo
    tients.
    riRi, r2R2,rzRg, r , are the successive remainders ; 11, 12, r3, rn
    being functions of y only .
    The process terminates as soon as a remainder is obtained
    which is a function of y only ; r , is here supposed to be such
    a remainder.
    Now , the simplest factors having been taken for C1, C2, C3, C4,
    we see that
    1 is the H . C . F . of 67 and rı ) The values of ~ and y,
    d. „ „ C and ra which satisfy simulta
    neously the equations
    da „ „ oil and rs A = 0 and B = 0 , are those
    obtained by the four pairs
    and ra of simultaneous equations
    dad , following :
    = 0 and B = 0 ... .......
    .. (1) The final equation in y ,
    which gives all admissible
    2 = 0 and R, = 0 ......... (2) values, is
    r ,rolg = 0 ,
    = 0 and R , = 0 ......... (3)| d ,dd, = 0 ,
    If it should happen that
    = 0 and Rs= 0 ......... (4) the remainder r, is zero,
    the simultaneous equa
    tions (1),(2),(3),and (4) reduce to
    1,=0 and 2 =0; j = 0 and i = 0; i =0 and # =0.
    158 THEORY OF EQUATIONS.

    594 To find infinite values of x or y which satisfy the given


    equations.
    Put x = _. Clear of fractions, and make z = 0.
    If the two resulting equations in y have any common
    roots, such roots, together with x = 0 , satisfy simultaneously
    the equations proposed.
    Similarly wemay put y =
    PLANE TRIGONOMETRY.

    ANGULAR MEASUREMENT.
    600 The unit of Circular measure is a Radian , and is the
    angle at the centre of a circle which subtends an arc equal to
    the radius. Hence
    arc
    601 Circular measure of an angle =
    radius
    602 Circular measure of two right angles = 3 :14159 ...37.
    603 The unit of Centesimal measure is a Grade, and is the
    one-hundredth part of a right angle .
    604 The unit of Sexagesimal measure is a Degree , and is
    the one-ninetieth part of a right angle .

    To change degrees into grades, or circular measure, or


    vice versâ , employ one of the three equations included in
    605 D G 2R
    90 100 a
    Where D , G , and C are respectively the numbers of degrees,
    grades , and radians in the angle considered .

    TRIGONOMETRICAL RATIOS .
    606 Let OA be fixed , and let the
    revolving line OP describe a circle
    round 0 . Draw PN always perpen
    dicular to AA . Then , in all posi
    tions of OP .
    ALN
    PN
    Op = the sine of the angle AOP,
    ON
    op = the cosine of the angle AOP ,
    PN
    on = the tangent of the angle AOP .
    160 PLANE TRIGONOMETRY.
    - - - -

    607 If P be above the line AA', sin AOP is positive.


    If P be below the line AA', sin AOP is negative.
    608 If P lies to the right of BB', cos AOP is positive.
    If P lies to the left of BB', cos AOP is negative.
    609 Note, that by the angle AOP is meant the angle through which
    OP has revolved from 04, its initial position ; ard this angle of revolution
    may have any magnitude. If the revolution takes place in the opposite
    direction , the angle described is reckoned negative.

    610 The secant of an angle is the reciprocal of its cosine,


    or cos A sec A = 1 .
    611 The cosecant of an angle is the reciprocal of its sine,
    or sin A cosec A = 1 .
    612 The cotangent ofan angle is the reciprocal of its tangent,
    tan A cot A = 1 .
    - - -

    Relations between the trigonometrical functions of the


    same angle.
    613 sin ’ A + cos’ A = 1. TI. 47
    614 sec? A = 1 + tan ’ A .
    615 cosec ' A = 1 + cot’ A .

    616 tan A = sin A 7606


    COS A

    If tan A = 0 va + 63

    617 sin A = / A .

    cos A = [606
    Vaʻ + 6
    tan A
    618 sin A = cos A = [617
    V1 + tan’ A V1 + tan ’ A
    TRIGONOMETRICAL RATIOS. 161

    619 The Complement of A is = 90° — A .


    620 The Supplement of A is = 180° — A .
    621 sin (90° – A ) = cos A,
    tan (90° — A ) = cot A ,
    sec (90° — A ) = cosec A .
    622 sin (180° - A ) = sin A ,
    cos ( 180° — A ) = - cos A ,
    tan (180° — A ) = - tan A .
    In the figure
    ZQOX = 180° – A . [607, 608
    623 sin ( - A) = - sin A .
    624 cos( - A ) = cos A .
    By Fig., and (607), (608).
    The secant, cosecant, and cotangent of 180° – A , and of
    - A , will follow the same rule as their reciprocals , the cosine ,
    sine, and tangent. [610-612

    625 To reduce any ratio of an angle greater than 90° to the


    ratio of an angle less than 90°.
    RULE . — Determine the sign of the ratio by the rules (607) ,
    and then substitute for the given angle the acute angle formed
    by its two bounding lines, produced if necessary.
    Ex. - To find all the ratios of 660°. B
    Measuring 300° ( = 660° — 360° round
    the circle from A to P , we find the
    acute angle AOP to be 60°, and P lies
    below AA , and to the right of BB'. A

    Therefore
    sin 660° = - sin 60° = - 1
    B :
    cos660° = cos 60° =
    and from the sine and cosine all the remaining ratios may be
    found by (610 -616).
    INVERSE NOTATION. — The angle whose sine is æ is denoted
    by sin - .
    162 PLANE TRIGONOMETRY .

    626 All the angles which have a given sine, cosine, or tan
    gent, are given by the formulæ
    sin ? x = nt + ( - 1)" 6 ... ............... (1),
    cos- ? x = 2n1 + 0 ........ ... (2),
    tan -? x = nu + 0 ... ... ... ...
    In these formulæ Q is any angle, which has a for its sine, cosine, or
    tangent respectively , and n is any integer .
    Cosec- 2 , sec- ' x , cot- ' x have similar general values, by (610 -612 ).
    These formulæ are verified by taking A , in Fig. 622, for 0, and making
    n an odd or even integer successively .

    FORMULÆ INVOLVING TWO ANGLES, AND


    MULTIPLE ANGLES .

    627 sin ( A + B ) = sin A cos B + cos A sin B ,


    628 sin ( A - B ) = sin A cos B - cos A sin B ,
    629 cos ( A + B ) = cos A cos B - sin A sin B ,
    630 cos ( A - B ) = cos A cos B + sin A sin B .
    Proof. — By (700) and (701) , we have
    sin C = sin A cos B + cos A sin B ,
    and sin C = sin ( A + B ), by (622) .
    To obtain sin ( A – B ) change the sign of B in (627) , and employ (623),
    (624), cos ( A + B ) = sin {(90° — A ) – B }, by (621).
    Expand by (628 ), and use (621), (623), (624). For cos ( A - B ) change the
    sign of B in (629).

    tan ( A + B ) – tan A + tan B


    631 1 - tan A tan B
    632 tan A – tan B
    tan ( A
    1 + tan A tan B
    cot ( A + B ) – cot A cot B - 1
    633
    cot A + cot B
    634. cot ( A - B ) = cot A cot B + 1
    cot B – cot A
    Obtained from (627 –630 ).
    MULTIPLE ANGLES. 163

    635 sin 2A = 2 sin A cos A . [627. Put B = A


    636 cos2A = cos A - sin ’ A ,
    637 = 2 cos’ A – 1 ,
    = 1 – 2 sin ’ A . [629, 613
    639 [637
    2 cos’ A = 1 + cos 2A .
    640 2 sin ’ A = 1 - cos 2A . [638

    641 [640
    sin A = V1= ços A
    642
    cos 4 = v1 +.ços A [639

    11 - cos A 1 - cos
    A sin A
    643 tan -2 1 + cos A sin A 1 + cos A
    [641, 642, 613

    1 - tan ’ A 2 tan
    646 cos A = sin A = [643, 613
    1 +4tan’ 1 + tan 4
    648
    64 cos A =
    com a conset
    cos 4 = 1 4+ tan
    con

    649 sin(45 + 4)= cos(45 – 4) = v1+ pin 4. con


    CO

    650 cos(45*+ 4)= sin(45° 4) = v1-sin 4. (642


    + Asin A =1 cos
    651 tan (45° + A ) = V1+ sin A = 1 cos
    1 - sin A
    A
    - sin A

    tan 2 2 tan A [631. Put B = A


    1 - tan’ A
    cot 2A = cot ? A - 1
    2 cot A
    164 PLANE TRIGONOMETRY.

    654 tan (45° + A ) = 1 + tan A


    1 - tan A

    655 tan (45° – A ) = 1 - tan A [631, 632


    1 + tan A
    656 sin 3A = 3 sin A - 4 sinº A .
    657 cos 3A = 4 cos' A – 3 cos A .
    tan 3A – 3 tan A - tan ' A
    658
    1 – 3 tan’ A
    By putting B = 2A in (627), (629), and (631) .
    sin UL ,
    659 sin (A + B ) sin (A – B) = sin’ A – sinº B
    = cos’ B – cos’ A .
    660 cos (A + B ) cos (A – B ) = cos’ A – sinºB
    COS

    = cos' B – sinº A .
    From (627), & c .

    661
    661 fire
    sin 4 + cos 4 = V1+ sin[Proved by squaring.
    662 sin 4 - cos4 = V1 - sin A.
    sin CO

    663 sin 4 = } {V1+ sin A - V1– sin A }.


    COS
    664 cos 4 = } {V1+ sin A + v1– sin A },
    when A lies between –45° and +45°.
    een

    665 In the accompanying diagram the


    signs exhibited in each quadrant are the
    signs to be prefixed to the two surds in + +
    thevalue of sin , according to the quad 1- + x +
    rant in which 4 lies.
    For cos - change the second sign.
    PROOF. - By examining the changes ofsign in (661) and (662) by (607).
    MULTIPLE ANGLES. 165

    666 2 sin A cos B = sin (A + B ) + sin ( A - B ).


    667 2 cos A sin B = sin ( A + B ) – sin ( A – B ).
    668 2 cos A cos B = cos( A + B ) + cos ( A – B ).
    669 2 sin A sin B = cos (A – B ) – cos (A + B ).
    COS

    [627 -630

    670 sin A + sin B = 2 sin 44 B cos47


    *
    8
    671 sin A – sin B = 2 cos 08
    2

    A + B A - B
    672 cos A + cos B = 2 cos 2
    0

    673 cos B - cos A = ?

    Obtained by changing A into 4 B, and B into 47 , in (666-669).


    It is advantageous to commit the foregoing formulæ to memory, in words,
    tbus 2 sin cos = sin sum + sin difference,
    2 cos sin = sin sum - sin difference,
    2 cos cos = cos sum + cos difference ,
    2 sin sin = cos difference – cos sum .
    sin first + sin second = 2 sin half sum cos half difference ,
    sin first - sin second = 2 cos half sum sin half difference ,
    cos first + cos second = 2 cos half sum cos half difference ,
    cos second - cos first = 2 sin half som sin half difference.

    674 sin (A + B + C )
    = sin A cos B cos C + sin B cos C cos A
    + sin C cos A cos B - sin A sin B sin C .
    675 cos ( A + B + C )
    = cos A cos B cos C - cos A sin B sin C
    - cos B sin C sin A - cos C sin A sin B .
    676 tan ( A + B + C )
    _ tan A + tan B + tan C - tan A tan B tan C
    1 - tan B tan C – tan C tan A - tan A tan B
    Proof.— Put B + C for B in (627), (629), and (631).
    166 PLANE TRIGONOMETRY.

    If A + B + 0 = 180°,

    677 sin A + sin B + sin C = 4 cos 4 cos cos


    sin A + sin B - sin C =
    sin A + sin B – sin C = 4 sin 4 sin cos como
    678 cos A + cosB + cos C =

    ‫ܬܟܬ‬
    cos A + cos B – cos C =
    cosA +cos B –cosC = 4cos cos. sinq –1.
    l
    ‫ܟܙܨ‬
    679 tan A + tan B + tan C = tan 2 tan B tan C.
    680 cot 4 + cot. + cot = cot cot. com
    681 sin 2A + sin 2B + sin 2C = 4 sin A sin B sin C .
    682 cos2A + cos2B + cos2C = – 4 cos A cos B cos C – 1.
    · General formulæ , including the foregoing , obtained by
    applying (666 –673).
    If A + B + C = a , and n be any integer,
    683 4 sin "A sinº sin
    = sin (1976 – nA ) + sin(1907 -nB ) + sin(" " – nc ) – sin ºm
    684 4 cos"A cosni cosmic
    = cos("07 –na) + cos(" * -nB) + cos(947.– nC ) + cos en
    If A + B + C = 0,
    685
    6854 sin"A sin mul sinq == -sinna–
    - sin nA - sin nB - sin nC .
    n -

    sinnВ– sin no .
    inA nB nc
    686 4cos " - cos " cos" = cosnA + cosnB + cosnC + 1.

    RULE. — If, in formulæ (683) to (686 ), two factors on the left


    be changed by writing sin for cos, or cos for sin , then , on the
    right side, change the signs of those terms which do not contain
    the angles of the altered factors.
    COMMON ANGLES. 167

    Thus, from (683),we obtain


    687 4 sin "A cos cos MC
    = -sin (1901 – na)+ sin (1901 –1B )+ sin( * – nC) + sin"
    A Formula for the construction of Tables of sines, co
    sines, & c.
    688 sin (n + 1) a - sin na = sin na - sin (n - 1 ) u - k sin na ,
    where a = 10" , and k = 2 (1 - сos a ) = '0000000023504.
    689 Formulæ for verifying the tables
    sin A + sin (72° + A ) - sin (72° — A ) = sin (36° + A ) - sin (36º – A ),
    cos A + cos (72° + A ) + cos (72° + A ) = cos (36° + A ) + cos ( 36° – A ),
    sin (60° + A ) - sin (60° – A ) = sin A .

    RATIOS OF CERTAIN ANGLES.


    (
    690 sin 45º= cos 45º= - tan 45º= 1.
    sin 60° = 3, cos 60°= , tan 60°= 13.
    sin 150 _ V3 - 1 15° - V3 + 1
    2v2 . 2 /2
    tan 15° = 2 - V37
    cot 15° = 2 + 3 ):
    sin 100 ✓5 - 1 18
    0 V5 + v5
    2 / 2 ;

    tan 18° = V 5 – 2, 5.
    694 sin 54° = V5+ 1, cos 54° = V5 22
    - V5
    2 v5
    tatann 54°
    54 == V5+2 5.
    695 By taking the complements of these angles , the same
    table gives the ratios of 30°, 75°, 72°, and 36º.
    168 PLANE TRIGONOMETRY.

    696 Proofs. — sin 15° isobtained from sin (45° — 30°),expanded by (628).
    697 sin 18° from the equation sin 2x = cos 3x , where x = 18°.
    698 sin 54° from sin 3x = 3 sin æ — 4 sinºx, where x = 18°.
    699 And the ratios of various angles may be obtained by taking the sam ,
    difference, or some multiple of the angles in the table, and making use of
    known formulæ . Thus

    12° = 30° — 18°, 71° = 15 , &c.,&c.

    PROPERTIES OF THE TRIANGLE .

    700 c = a cos Btb cos A.


    1701 A = sin Basic
    702 a ' = 6 + — 2bc cos A .
    Proof. — By Euc. II. 12 and 13, a ' = b' + c* — 2c. AD.

    703 COS A = b' + c - a?


    2bc

    ате
    If 8 = a +b + c, and A denote the area ABC,
    704 sin4 = V (8–6)bc,68 –c), cosa = v *(bc a).
    [6 + 1 , 642, 703, 9 , 10, 1.

    705 tan 4 = ( (956).


    706 sin A = Z Vs( –a)(s—b)(8–c). [633,704
    707 A = b4 sin A = Vs(s— a)(s—b)(8 —c), [707,706
    708 = }V26*c++ 2c*ư’ + 2a’b?— as– b*_c .
    PROPERTIES OF TRIANGLES. 169

    Let
    The Triangle and Circle.
    p = radius of inscribed circle ,
    re = radius of escribed circle
    touching the side a ,
    R = radius of circumscribing
    circle .
    709 r
    [From Fig.,a =" t
    710 7 =

    Bya=root +root om l
    [By s =" *+ ALTE ME
    a cos 5 cos
    [From a = r,tan . +r,tant

    713 Rezan a = abiem


    [ By (III. 20 ) and ( 706 )
    715
    = }vf(b+c)*secº4+(6 –c)'cosecº4)
    [ 702
    Distance between the centres of inscribed and circum
    scribed circles
    716 = VR — 2Rr. [936
    Radius of circle touching b , c and the inscribed circle
    717 p = r tan’ | ( B + C ). (By sin · + por
    170 , PLANE TRIGONOMETRY.

    SOLUTION OF TRIANGLES.

    Right-angled triangles are solved


    by the formulæ
    718 c = a + 62;
    la = c sin A ,
    719 b = c cos A ,
    ( a = b tan A ,
    & c.

    Scalene Triangles.
    720 Case I. — The equation
    VA

    [701
    sinA sin B
    will determine any one of the four
    quantities A , B , a , when the re
    maining three are known.

    721 The Ambiguous Case . - ...


    When , in Case I., two
    sides and an acute angle
    -- -

    opposite to one of them


    are given , we have, from
    the figure,
    sin C = ( sin A C' s

    Then C and 180°- C are the values of C and C', by (622).


    Also b = c cos A Eva - c? sin ’ A ,
    because = AD + DC.
    722 When an angle B is to be determined from the equation
    sin B = sin A ,

    and a
    is a small fraction ; the circular measure of B may be approximated
    to by putting sin (B + C ) for sin A ,and using theorem (796 ).
    SOLUTION OF TRIANGLES. 171

    723 CASE II. — When two sides b , c and the included angle
    A are known , the third side a is given by the formula
    a ’ = b + 62 - 2bc cos A , [ 702
    when logarithms are not used .

    Otherwise, employ the following formula with logarithms,


    724 1 2 6 +cu 2
    Obtained from 7 : = sin B + sin C (701), and then applying
    (670 ) and (671).
    B
    Po2 having been found from the above equation, and
    ve
    B+2 C being equal to 90°- 4,wehave
    725 B - B + C + B - C, C = B + C _ B -2 C .
    2 2

    B and C having been determined, a can be found by Case I.


    726 If the logarithms of b and c are known, the trouble of taking out
    log (6 — c) and log (6 + c) may be avoided by employing the subsidiary angle
    0 = tan -? ) , and the formula

    727 tan }(B -C) = tan (0 - 2 ) cot , [655

    Or else the subsidiary angle0 = cost , and the formula


    728 tan £(B —C) = tan cott [643

    If a be required without calculating the angles B and C , wemay use the


    formula
    [From the figure in 960, by
    729 a = (b+c)sin 1. drawing ang perpendicular
    drawing a perpendicular
    from B to EC produced .
    cos } ( B - C )
    730 If a be required in terms of b , c,and d alone, and in a form adapted
    to logarithmic computation, employ the subsidiary angle
    / - ot4bc
    0 = siu -+ Ti rji. .cos .? AA ),
    and the formula a = (b + 1) cos 0 . [702, 637
    172 PLANE TRIGONOMETRY.

    CASE III. — When the three sides are known, the angles
    may be found without employing logarithms, from the formula
    COS A - 6 + c - a ?
    731 260
    [ 703

    732 If logarithms are to be used , take the formulæ for


    S

    sin A , cos A , or tanA ; (704) and (705).


    QUADRILATERAL INSCRIBED IN A CIRCLE .
    cos Beat 0 - 0 - dº
    733 B

    2 ( ab + cd )
    From AC? = a + b = — 2ab cos B = c + d +
    2cd cos B , by (702), and B + D = 180°.
    734 sin B = ab
    - 2+ cd [613,733
    735 Q = V (s — a)(s — b )(s — c)(s — d )
    = area of ABCD ,
    and s = (a + b + c + d ) .
    Area = jab sin B + { cd sin B ; substitute sin B from last.
    2 _ (ac + bd ) (ad + bc)
    1736 [ 702, 733
    (ab + cd )
    Radius of circumscribed circle

    1737 V (ab + cd ) (ac + bd ) (ad + bc). [713, 734, 736

    If AD bisect the side of the triangle ABC in D ,


    4A
    1738 tan BDA = )

    739 cot BAD = 2 cot A + cot B .


    740 A D ’ = 1 (1? + + 2bc cos A) = į (6 + 0 - a ).
    If AD bisect the angle A of a triangle ABC,
    742 tan BDA = cot ByCatetan
    743
    SUBSIDIARY ANGLES: 173

    If AD be perpendicular to BC,
    744 AD _ bc sin A _ b’ sin C + cé sin B
    b + c
    745 BD 63 - 02 _ , tan B - tan C
    a tan B + tan

    REGULAR POLYGON AND CIRCLE.


    Radius of circumscribing circle = R .
    Radius of inscribed circle
    Side of polygon
    Number of sides
    COSE

    746 R = cosee .. r = cot . A


    Area of Polygon
    748 = {naº cot " = nRº sin 2 = nrº tan

    USE OF SUBSIDIARY ANGLES.


    749 To adapt a + b to logarithmic computation.
    Take 0 = tan-' ) ; then a +b = asecº0. - as

    750 For a – take = tan-- ( ) ; thus


    a - 6 = a / 2 cos(8
    cos
    + 45°).
    751 To adapt a cos C + b sin C to logarithmic computation .
    Take 0 = tan-tą ; then
    a cos C + b sin ( = / (a + 1 ) sin ( 0 + C ). [ By 617
    For similar instances of the use of a subsidiary angle, see (726 ) to (730).

    752 To solve a quadratic equation by employing a subsidiary


    angle .
    If x = — 2px + q = 0 be the equation ,
    [ By 45
    += »(1+ (1-3).
    174 PLANE TRIGONOMETRY.

    Case I.— If q be < pp, put = sin’d; then


    . * = 2p cos». , and 2p sin [639,640

    Case II.—Ifq be > p?,put ? = sec?0; then


    x = p (1 = ítan6), imaginary roots. [614
    Case III.— Ifq be negative, put fi = tan’0; then
    « = vq cot and –vatan , [644 , 645

    LIMITS OF RATIOS.
    753 sine = tani = 1,
    0
    2 T

    when 0 vanishes.
    V A
    For ultimately IN = A7 = 1.[601,606 -
    754 n sin = 0when n is infinite. Byputting for&in last. n

    755 (cosO) = 1 when n is infinite.


    Proof. — and expand the logarithm by ( 156) .

    DE MOIVRE 'S THEOREM.


    756 (cos atisin a ) cosBti sin B ) ... & c.
    = cos (a + B + y + ...) + i sin (a + B + y + ...),
    where i = V - 1. [Proved by Induction.
    1757 (cos 6 + isin 6)" = cosn0 + i sin no.
    Proof. — By Induction , or by putting a , b , & c. each = 0 in (756 ).

    · Expansion of cosno, & c., in powers sin @ and cos 0.


    1758 cos n0 = cos” 0 – C (n , 2) cos" -20 sin ? 0
    + C (n , 4 ) cos" - 0 sin * 8 - & c.
    759 sin n0 = n cos” -10 sin 6 - C (n , 3) cos” -38 sin 0 + & c .
    Proof. — Expand (757) by Bin . Th., and equate real and imaginary
    parts.
    TRIGONOMETRICAL SERIES. 175

    n tan 0 - C (n , 3) tan ' 6 + & c.


    760 tan n0 =
    - 1 - C (n , 2 ) tan’ 8 + C (n , 4 ) tan * 0 - & c.
    In series (758, 759), stop at, and exclude,all termswith indices greater
    than n . Note , n is here an integer.

    Let 8, = sum of the C (n , r) products of tan a, tanß, tan y,


    & c. to n terms.
    761 sin (a + b + y + & c.) = cosa cos ß ... (8 - 83 + 8; - & c.).
    762 cos(a + B + y + & c .) = cos a cos ß ... ( 1 - 82 + 84 - & c.).
    Proof. — By equating real and imaginary parts in (756 ).
    763 tan (a + + x + & c.) = $1 – 83 + 85 - 87 + & c .
    1 - 82 + 84 - 88 + & c .

    Expansions of the sine and cosine in powers of the angle.


    764 sin 6:= 0 – &c. cos 0 = 1 - top +& c.
    Proof.— Put for 6 in (757) and n = 00, employing (754) and (755).
    766 elo = cos 6 + i sin 6. e-iº = cos 6 - isin 8. By (150)
    1768 ' eil te-it = 2 cos 6. piº — e-10 = 2i sin 0 . .
    1770 itan 0 = e — e-19 1 + itan 6 = 228
    ei + e - 1 - i tan 8 -

    Expansion of cosa ( and sin" 0 in cosines or sines of


    multiples of 0.
    772 2n -1 cos” 0 = cos no + n cos (n — 2 ) 0
    + C'(n , 2) cos (n — 4 ) 8 + C (n , 3) cos(n - 6 ) + & c.
    773 When n is even ,
    21 - 1 ( - 1) in sin " 0 = cosnon cos (n - 2) 0
    IN
    + C (n , 2) cos (n — 4 ) 0 — C (n , 3) cos (n - 6 ) 0 + & c.,
    1774 And when n is odd,
    n - 1

    20-1 ( - 1 ) 2 sin " 0 = sin no - n sin (n — 2 ) 0


    + C (n , 2 ) sin (n — 4 ) 6 - C (n , 3) sin (n — 6 ) 6 + & c.
    176 PLANE TRIGONOMETRY.

    Observe that in these series the coefficients are those of the Binomial
    Theorem , with this exception : If n be even , the last term must be divided by 2.
    The series are obtained by expanding ( eis + )" by the Binomial Theorem ,
    collecting the equidistant terms in pairs, and employing (768) and (769).

    Expansion of cos no and sin no in powers of sin 0.


    1775 When n is even,
    cos n0 = 1 - sin’8 + **(n°4 ! -2%) sin* 8
    _n°(n°—2)(n*—
    6 ! 4*)sinº0 + &c.
    776 When n is odd ,
    cos no = cose{1-**2 ! sinº0 +(n°–1](n2–
    COS CO
    4 ! 3) sin
    (n - 1)(n° — 32)(n’ — 5º) sin 6 + & c.}.
    6 !
    bryny When n is even ,
    n cu
    .. . sin no = n cose{sin 8 –"*-
    3 ! *sinº8+
    (n ? — 22) (nº — 49) cins A _ (n° — 22) (nº — 4 ) (n? — 6²) cin76L be ?
    5 ! 7 !
    778 When n is odd,
    sin n0 = n sin A _ n ((nn * -— 1 ) sin3 in
    + (n° - 1 ) (nº - 3 )
    3! 5!
    n ( n - 1 ) (n° - 3° ) (n° - 5°) sin ? A + & c .

    Proof. — By ( 758), wemay assume, when u is an even integer ,


    cos n0 = 1 + A, sin : 0 + A ,sin * 6 + ... + A2, sinar 0 + ....
    Put 0 + x for 0 , and in cosno cosnx — sin nd sin no substitute for cos nx and
    sin nx their values in powers of nx from (764 ). Each term on the right is of
    the type A2, ( sin 0 cos x + cos O sin x )2r . Make similar substitutious for cosx
    and sinx in powers of x . Collect the two coefficients of a * in each term by
    the multinomial theorem ( 137 ) and equate them all to the coefficient of 2 *
    on the left. In this equation write cos' 0 for 1 - sina 8 everywhere , and then
    equate the coefficients of sinar O to obtain the relation between the successive
    quantities A2, and A2 +2 for the series (775 ).
    To obtain the series (777) equate the coefficients of a instead of those of x*.
    When n is an odd integer , begin by assuming, by (759),
    sin no = A , sin 0 + A , sinº0 + & c.
    TRIGONOMETRICAL SERIES. 177

    779 The expansions of cos no and sin no in powers of cos 0


    are obtained by changing e into ža - 0 in (775) to (778).
    780 Expansion of cosno in descending powers of cos 0.
    2cos no = (2 cos0)* - n (2cos6)*-*+ " (173)(2cos
    2 ! 6)----
    ...+ ( - 1)=n (n —r — 1)(n —rr —! 2)...(n – 2r + 1)(2 cos0)*-+*+
    upto the last positive power of 2 cos 0.
    Proof.— By expanding each term of the identity
    log (1 –2x)+log (1 - 3 ) = log (1 - -(2 + . - -)}
    by (156 ), equating coefficients of z ", and substituting from ( 768).

    783 sin a + c sin (a + B) + cº sin (a + 23 ) + & c. to n terms ns

    _ sin a - c sin (a - B ) - c " sin (a + nß ) + cm + 1 sin {a + 1 - 1ß }


    1 - 2c cos B + ca
    If c be < 1 and n infinite, this becomes
    784 = sin a - c sin (a – B )
    1 - 2c cos Bte?

    785 cos a + c cos (a + b ) + c* cos (a + 23 ) + & c. to n terms


    = a similar result, changing sin into cos in the numerator.
    786 Similarly when c is < 1 and n infinite.
    787 Method of summation . — Substitute for the sines or
    cosines their exponential values ( 768). Sum the two resulting
    geometrical series, and substitute the sines or cosines again for
    the exponential values by ( 766 ).

    788 €sin (a + B)+ sin(a + 2B) + sin(a + 38) + &c.to


    infinity = eccouf sin (ate sin B)= sin a.
    789 ccos(a + B)+ cos(a + 2B) + cos(a + 38) + &c.to
    infinity = ec cous cos (a +c sin B) – cosa.
    Obtained by the rule in (787).
    2 A
    178 PLANE TRIGONOMETRY.

    790 If, in the series (783) to (789 ), ß be changed into ß + , the signs of
    the alternate terms will thereby be changed .

    Expansion of 0 in powers of tan 0 (Gregory's series).


    791 0 = tan — tanº® + tanº0 – & c.
    The series converges if tan 6 be not > 1.
    Proof. — By expanding the logarithm of the value of e?* in (771) by ( 158).

    Formulæ for the calculation of the value of by Gregory's


    series.

    792 1 = tan-'4+tan--* = 4tan-' – tan- [192


    794 = 4 tan-'} –tan- + + tanum
    Proof. - By employing the formula for tan ( 4 + B ), (631).
    To prove that a is incommensurable.
    795 Conrert the value of tan 0 in terms of 8 from (764) and (765) into
    a continned imaction thns tan 8 - 91 b - 3 – 5 – 7 - & n ; or this resalt may

    may be obtained by putting it for y in (294), and by ( 770). Hence


    0 022
    tan o 3 – 5 - 7 - & c.
    Put for 0, and assume that t , and therefore , is commensurable. Let
    m ,m and » being integers. Then we shall have 1 =
    3n — 51 – in - & c .
    The continued fraction is incommensurable , by (177). But unity cannot be
    equal to an incommensurable quantity. Therelore r is not commensurable .

    1796 If sin x = n sin (x + a ), x = n sina + sin 2a + sin 3a + & c.


    797 If tan
    1-n
    x = n tany, x = y – m sin 2y + ", sin ty - sin 6y + & c.,
    where m =
    1tiu
    PROOF. — By subs'ituting the erpinential values of the sine or tangent (769)
    and (770 ), and then eliminating a .

    1798 Coefficient of x " in the expansion of eear cos bbæ


    " e cos æ – (a’ + 12 )- Cos no,
    = '4
    where a = r cos 6 and b = r sin 0 .
    · For proof, substitute for cos bæ from (768) ; expand by (150 ) ; pat
    a = rcos , b = r sin 0 in the coefficient of z " , and employ (757) .
    TRIGONOMETRICAL SERIES. 179

    799 When e is < 1, V1--es = ,l + 2b cos 0 + 25? cos 20 + 268 cos 30 + ...,
    1 - ecos
    where b = _ _ e
    1 + v1 - es
    26 2 0 1
    For proof, put e = 1 and 2 cos 0 = x + 5 expand the fraction in two
    series of powers of æ by the method of ( 257), and substitute from ( 768 ).

    800 sin a + sin (a + b ) + sin (a + 2B) + ... + sin {a + (n - 1) B }


    sin(a + " ! 8)sin" B

    801 cosa + cos (a + B) + cos(a + 2B ) + ... + cos{a + (n - 1)} }


    cos(a + ” B)sinº
    sin
    802 If the terms in these series have the signs + and -
    alternately , change ß into B + a in the results.
    Proof. — Multiply the series by and apply (669) and (666 ).

    803 If B = 2 in (800) and (801),each series vanishes.


    804 Generally, If ß = 27 , and if r be an integer not a
    multiple of n , the sum of the path powers of the sines or cosines
    in (800 ) or (801) is zero if r be odd ; and if » be even it is
    = C (r, a ); by (772) to (774).
    General Theorem . - Denoting the sum of the series
    805 c + cx + cx* + ... + cmx" by F (x ) ;
    then c cosa + g cos (a + B ) + ... + encos (a + nB) = 1 { e'* F (el ) + €-10F (e -if)} ,
    and

    806 csina + sin (a + B ) + ... + cmsin (a + nß)b ) = F ( !?) – e -t-Fie-is))}.


    Proved by substituting for the sines and cosines their exponential values
    (766), & c.
    180 PLANE TRIGONOMETRY.

    Expansion of the sine and cosine in factors.


    807 22n — 2.x” y" cos no + yin
    = {2 – 2xy cos6 + y } {x2 – 2xy cos(6 + 2 ) + y } ...
    to n factors, adding 2 to the angle successively.
    n

    PROOF. - By solving thequadraticon the left,weget x = y (cos no + isin no) * .


    The n values of x are found by (757) and (626 ), and thence the factors. For
    the factors of x " + y " see (480 ) .

    808 sin nø = 2"-1sin o sin (4 + )sin ($ + 3 )...


    as far as n factors of sines.
    Proof. – By putting x = y = 1 and 0 = 29 in the last.
    809 If n be even ,

    sin nø = 2*-1 sin$ cos6 (sin* . – sinº')(sin 2n –sinºm )&c.


    810 If n be odd, omit cos o and make up n factors, reckoning
    two factors for each pair of terms in brackets.
    Proof. — From (808), by collecting equidistant factors in pairs, and
    applying (659) .
    CO

    811 cos no = 21-1sin($ + .)sin($ + 2 ...to n factors.


    Proof.–Put ® + for in (808).
    812 Also, if n be odd,
    cos no = 2*--cos$ (sin 2.– sin®6)(sin –sinºo)...
    813 If n be even , omit cos g.
    PROOF. — As in (809 ).

    814 n = 2"-1sin..
    n sin 27
    n sin 3 ...sin (n-n 1)*.
    п

    Pk00F . — Divide (809) by sin o , and make o vanish ; then apply (754).

    815 sin6 = 0 {1- (9)} {1-(0 )} {1-(39)}.....


    816 cose = {1 - (29)"} {1- (30)} {1 - 69) ......
    Proof. — Put q = in (809) and (812) ; divide by (814 ) and make n
    infinite.
    ADDITIONAL FORMULÆ . 181

    817 € – 2 cos 0 + e +
    = 4 siu jo {1+ } {1+ 2 10;} {1+(110 }...
    Proved by substituting æ = 1 + y = 1 - 7 , and for 6in (807),
    making n infinite, and reducing one series of factors to utting
    z = 0.

    De Moivre's Property of the


    Circle . - Take P any point, and
    POB = 0 any angle ,

    BOC = COD = & c. = 27 ;


    OP = x ; OB = r .
    819 22n — 2.x " poll cos no + po2n
    = PB PC PD ... to n factors .
    By (807) and (702), since PBP = x2– 2xr cos6 + r", &c.
    820 If x = r, 27" sin 9,0 = PB .PC.PD ... &c.
    821 Cotes'sproperties. — If 8 = 24n ,
    21" gove = PB.PC .PD ... & c.
    822 X " + pon = Pa . Pb . Pc . .. & c .

    ADDITIONAL FORMULÆ .
    823 cot A + tan A = 2 cosec 2A = sec A cosec A .
    824 cosec2A + cot2A = cot A . sec A = 1 + tan A tan 4 .
    CO

    COS
    cos A = cos•4. – sinº 4 .
    tan A + sec A = tan(450+ 4 ).
    828 tan A + tan B = tan A tan B .
    cot A + cot B
    829 sec? A cosec - A = sec A + cosecº A .
    182 PLANE TRIGONOMETRY.

    830 If A + B + C = 7 ,
    tan B tan C + tan C tan A + tan A tan B = 1 .
    831 If A + B + C = 1 ,
    cot B cot C + cot C cot A + cot A cotB = 1 .

    832 sin " * + sin- * = . tan--- + tan =


    In a right-angled triangle ABC,C being the right angle,
    Da - 6 OD Zab
    833 cos 2B = tan 2B =
    a ’ + 62: a ’ - 62"

    834 tanļ4 = 1 ). R +r = }(a + b).


    In any triangle,
    835 sin }(A – B) = ^ = cos}C.
    CO
    cos } (A - B ) = a + sin }C.
    a sin A - B _ a' — 62 tan } A + tan B _ c.
    836 sin A + B = tan į A - tan B - a - 6
    837 }(a’ + 6² + cº) = bc cos A + ca cos B + ab cos C.
    838 Area of triangle ABC = }bcsin A
    2 sin B sin10C = } (a’ 12) sin A sin B
    sin A sin ( A - B )

    839 = 2004 , cos }A cos }B cos C.


    840 = i (a + b + c)? tan A tan _B tan C.
    With the notation of (709) ,
    841 r = } (a + b + c) tan ] A tanıB tan } C .
    842 = _ +abc
    2Rr Sa A = Vrrerore.
    b + c.
    843 a cos A + b cos B + c cos C = 4R sin A sin B sin C .
    ADDITIONAL FORMULÆ . 183

    844 Rtrri( a cot A + b cot Btc cot C ) = sum of per


    pendiculars on the sides from centre of circumscribing circle.
    This may also be shown by applying Enc. VI. D . to the circle described
    on R as diameter and the quadrilateral so formed .

    845 roror . = abc cos } A cos ? B cos C .


    846 r = / (ror.) + / (r.ra) + v (raro).
    847 r = ra1 + 1V + roce tan 4 = Von
    849 If O be the centre of inscribed circle,
    = a_+ 2b
    0A C + c ccooss } A .
    b c_
    850 a (b cos C - c cos B ) = 6° — cº.
    851 b cos B + c cos C = c cos ( B - C ).
    852 a cos A + b cos B + c cos C = 2a sin B sin C .
    853 cos A + cosB + cos C = 1 + 2a sin B sin C .
    a +b + c

    854 If s = } (a + b + c),
    1 - cos: a - cos’ b - cosé c + 2 cos a cos b cos c
    = 4 sin s sin ( s - a ) sin (s — b ) sin (s - c) .
    855 - 1 + cos'a + cos*6 + cosºc + 2 cos a cos 6 cos c
    = 4 cos s cos (s — a) cos (s — b ) cos (s — c).
    856
    4 cos cos. cos.
    = cos s + cos (s —a ) + cos (s — b) + cos (s — c).
    COS OS

    857
    = - sin s + sin (s — a ) + sin (8 — 6) + sin (s — c).

    858 rº= 6(1+ 2 + 3 +...)= 8(1+ + +...).


    PROOF. - Equate coefficients of 6 in the expansion of S . by (764) and
    (815 ) or of cos 6 by ( 765 ) and (816 ) .
    184 PLANE TRIGONOMETRY.

    859 Examples of the Solutions of Triangles.


    Ex. 1 : CASE II . (724) . - Two sides of a triangle b , c, being 900 and 700
    feet, and the included angle 47° 25', to find the remaining angles .
    w B - C b - c A
    23° 42' 30" ;
    btc 2

    therefore log tan }(B - C) = log cot 4 – log 8;


    therefore I tan ( B - C ) = L cot 23° 42' 30" – 3 log 2,
    10 being added to each side of the equation.
    .:. L cot 23° 42' 30" = 10.3573942* 1 . ( B - C ) = 15° 53' 19:55' *
    3 log 2 = •9030900 ) and 1 ( B + C ) = 66° 17 ' 30"
    .: L tan ( B - C ) = 9.4543042 ( . . B = 82° 10' 49.55"
    And, by subtraction , C = 50° 24 ' 10 :45" .
    Ex. 2 : Case III. (732). —Given the sides a, b, c = 7, 8, 9 respectively,
    to find the angles.
    tan A = VG-WATEVAMOV
    therefore - 10 + ; (log 2 - 1) = 9.650515 ;
    therefore A = 24° 5' 41.43" .*
    B is found in a similar manner, and C = 180° - A - B .

    Ex. 3. - In a right-angled triangle, given the hypotenuse c = 6953 and


    a side b = 3 , to find the remaining angles.
    Here cos84A == 6953: But
    _ 3F .. but, since A is nearly a right angle, it cannot be
    But, since
    determined accurately from log cos A . Therefore take

    V6953 ;
    therefore I sin 4 = 10 + ı (log 3475 – log 6953) = 9-8493913;
    therefore A = 44°59' 15:52",*
    therefore A = 89° 58 '31 04" and B = 0° 1' 28-96 ".
    * See Chambers's Mathematical Tables for a concise explanation of the
    method of obtaining these figures.
    SPHERICAL TRIGONOMETRY.

    INTRODUCTORY THEOREMS.
    870 Definitions. — Planes through the centre of a sphere
    intersect the surface in great circles ; other planes intersect
    it in small circles. Unless otherwise stated, all arcs are
    measured on great circles .
    The poles of a great circle are the extremities of the
    diameter perpendicular to its plane.
    The sides a , b , c of a spherical triangle are the arcs of
    great circles BC , CA , AB on a sphere of radius unity ; and
    the angles A , B , C are the angles between the tangents to the
    sides at the vertices , or the angles between the planes of the
    great circles. The centre of the sphere will be denoted by 0 .
    The polar triangle of a spherical triangle ABC has for its
    angular points A ', B ', C ', the poles of the sides BC , CA, AB
    of the primitive triangle in the directions of A , B , C respec
    tively (since each great circle has two poles ). The sides of
    A' B 'C' are denoted by a ', b', c'.
    871 The sides and angles of the
    polar triangle are respectively the
    supplements of the angles and
    sides of the primitive triangle ;
    that is,
    d ' + A = b' + B = 6 + C = 1 ,
    a + A = b + B = c + C' = .
    Let BC produced cut the sides A ' B ', B BS
    C'A ' in G , H . B is the pole of A 'C',
    therefore BH = Similarly CG
    therefore, by addition, a + GH = r and GH = A ’, because A ' is the pole of BC.
    The polar diagram of a spherical polygon is formed in the sameway, and
    the same relations subsist between the sides and angles of the two figures.
    2 B
    186 SPHERICAL TRIGONOMETRY.

    RULE . — IIence, any equation between the sides and angles


    of a spherical triangle produces a supplementary equation by
    changing a into a - A anil A into a - - a , & c .

    872 The centre of the inscribed circle , radius r., is also the
    centre of the circumscribed circle , radius R’, of the polar
    triangle , and 1 + R = .
    PROOF. - In the last figure, let O be the centre of the inscribed circle of
    ABC ; then OD, the perpendicular on BC , passes through A ', the pole of BC.
    Also, OD = r ; therefore O A ' = - . Similarly OB ’ = OC ' = - r ; there
    fore 0 is the centre of the circumscribed circle of A ' B ' C', and r + R ' = * .

    873 The sine of the arc joining a point on the circumference


    of a small circle with the pole of a parallel great circle, is equal
    to the ratio of the circumferences or corresponding arcs of the
    two circles.
    For it is equal to the radius of the small circle divided by the radius of
    the sphere ; that is , by the radius of the great circle .

    874 Two sides of a triangle are greater than the third .


    [By XI. 20.
    875 The sides of a triangle are together less than the cir
    cumference of a great circle . [By XI. 21.
    876 The angles of a triangle are together greater than two
    right angles.
    · For — A + + - B + - C is < 27, by (875) and the polar triangle.

    877 If two sides of a triangle are equal, the opposite angles


    are equal. [By the geometrical proof in (894).
    878 If two angles of a triangle are equal, the opposite sides
    are equal. [By the polar triangle and (877).

    879 The greater angle of a triangle has the greater side


    opposite to it.
    Proor. - If B be > A , draw tl:e arc BD meeting AC in D , and make
    LABD = A , therefore BD = AD ; but BD + DC > BC, therefore AC > BC.
    880 The greater side of a triangle has the greater angle
    opposite to it. [ By the polar triangle and (879).
    - - - -
    OBLIQUE -ANGLED TRIANGLES. 187

    RIGHT-ANGLED TRIANGLES .
    881 Napier's Rules. - In the triangle ABC let C be a right
    angle, then a , ( - B ), ( - c), ( - A ), and b , are called
    the five circular parts. Taking any part for middle part,
    Napier 's rules are —
    1. sine ofmiddle part = product of tangents ofadjacentparts .
    II. sine ofmiddle part = product of cosines of opposite parts.
    In applying the rules we can take A , B , c' instead of their
    complements, and change sine into cos, or vice versa , for those
    parts at once. Thus, taking b for the middle part,
    sin b = tan a cot A = sin B sin c.
    Ten equations in all are given by the
    rules.
    Proof.– From any point P in 0A, draw
    PR perpendicular to OC, and RQ to OB ;
    therefore PRQ is a right angle ; therefore OB
    is perpendicular to PŘ and QR, and therefore
    to FQ . Then prove any formula by proportion
    from the triangles of the tetrahedron OPQR ,
    which are all right-angled. Otherwise, prove
    by the formulæ for oblique -angled triangles.

    OBLIQUE -ANGLED TRIANGLES .


    882 cos a = cos b cos c + sin b sin c cos A .
    PROOF. - Draw tangents at A
    to the sides c, b to meet OB , OC
    in D and E . Express DE by
    (702) applied to each of tho
    triangles DAE aud DOE, and
    subtract. TRTA
    If AB and AC are both > 5 ,
    produce them to meet in A ', the
    pole of A , and employ the tri
    angle A'BC.
    If AB alone be > , pro
    dace BA to meet BC.
    ne supplementary formula , by (871) , is
    883 COS A = - cos B cos C + sin B sin C cosa .
    188 SPHERICAL TRIGONOMETRY.

    sin ( s — b ) sin (. —
    884 sin b sin c
    885 sin s sin ( s - a )
    sin b sin c
    886
    L s in (s — b ) sin (8 — c) where s = } (a + b + c).
    sin s sin ( s - a )
    Proof. ? ( 1 - сos A ) . Substitute for cos A from ( 872), and
    throw the nomerator ofthe whole expression into factors by (673). Similarly
    for cos
    The supplementary formulæ are obtained in a similar way,
    or by the rule in (871). They are
    cos ( S - B ) cos ( S - 0
    887 O

    sin B sin C
    1 - cos S cos ( S - A )
    888 sin = sin B sin C
    889 tan l - cos S cos ( S - A )
    cos ( S - B ) cos ( S - C )
    IL

    where S = } ( A + B + C ).
    890 Let orv sin s sin ($ — 12) sin ($ — 0 ) sin (s — c)
    = V1 + 2 cos a cos b cos c - cos’ a - cos? 6 - cos' c.
    Then the supplementary form ,by (871), is
    891 = v - cos S cos (S — A ) cos ( S — B ) cos ( S – C )
    = įv1 — 2 cos A cos B cos C - cos” A - cos” B - cos? C .

    892 sin A = - inci sin a =w inni


    [By sin A = 2 sin-- cos and (884,885),&c.
    893 The following rules will produce the ten formulæ
    (884 to 892) —
    1. Write sin before each factor in the s values of sin ,
    OBLIQUE-ANGLED TRIANGLES. 189

    cos , tan , sin A, and A , in Plane Trigonometry (704 –


    707), to obtain the corresponding formulæ in Spherical Trigo
    Con

    nometry .
    II. To obtain the supplementary forms of the five results,
    transpose large and small letters everywhere, and transpose
    sin and cos everywhere but in the denominators, and write
    minus before cos S .
    sin A sin B _ sin C
    894 sin a sin b sin c
    Proof. — By (882). Otherwise , in the figure of 882, draw PN perpendi
    cular to BOC, and NR, NS to OB, 00. Prove PRO and PSO right angles
    by I. 47 , and therefore PN = OP sin c sinB = OP sin b sin C .

    895 cosb cos C = cot a sin b - cot A sin C .


    To remember this formula , take any four consecutive angles
    and sides (as a , C , b , A ) , and , calling the first and fourth the
    extremes, and the second and third the middle parts , employ
    the following rule :
    RULE. — Product of cosines of middle parts = cot extreme
    side X sin middle side – cot extreme angle X sin middle angle .
    Proof. - In the formula for cos a (882) substitute a similar value for
    cosc, and for sin c put sin sin? A
    896 NAPIER’ S FORMULÆ .
    (1) tan } ( A - B ) = sin } (a − b) ont C
    sin ? ( a + b
    (2) tan } (AALR)
    + B ) –= cos
    cos }1 (aa 1− b% ) cot
    cot C .
    sin I ( A - B
    (3) tan } (a - b) = sin î ( A + B )
    ( 4 ) tan į (a + b ) = cos } ( A – B ) ....
    - cosį (A + B ) tan z
    Rule. - In the value of tan ( A - B ) change sin to cos to
    obtain tan ? ( A + B ). To obtain ( 3 ) and ( 4 ) from ( 1 ) and ( 2 ) ,
    transpose sides and angles , and change cot to tan .
    Proof. — In the values of cos A and cos B , by (883), put m sin a and
    m sin b for sin A and sin B , and add the two equations. Then put
    m - sin A sin B
    sin a + sin b ,' and transform by (670-672) .
    190 SPIERICAL TRIGONOMETRY.

    897 GAUSS' S FORMULÆ .


    sin } ( A + B ) _ cos ž (a − b )
    ē cos C COS 5c

    sin } ( A , B ) _ sin ž (a − b )
    cos IC sin c
    I DO

    cos } (A + B ) – cos } (a + b )
    sin IC cos C
    (1) cos } (A – B ) _ sin } (a + b )
    sin C sin c
    From any of these formulæ the others may be obtained
    by the following rule :
    RULE. - Change the sign of the letter B (large or small) on
    one side of the equation , and write sin for cos and cos for sin
    on the other side.
    Proof. — Take sin } ( A + B ) = sin ; A cos B + cos A sin ; B ,
    substitute the s values by (884, 885), and reduce.

    SPHERICAL TRIANGLE AND CIRCLE.

    898 Let be the radius of the in


    scribed circle of ABC ; r , the radius of
    the escribed circle touching the side a ,
    and R , R , the radii of the circumscribed
    circles ; then
    (1) tan r = tan ļA sin (s — a) = 0 .
    = 2sinsin A a sin }A sinBsin C
    2 cos ; A cos įB cos ; C
    25
    cos S + cos (S - A ) + cos (S - B ) + & c.
    Proof. — The first value is found from the
    right-angled triangle OAF, in which AF = s - a .
    The other values by (884 - 892 ).
    SPHERICAL TRIANGLE AND CIRCLE. 191

    899 (1) tan ra = tan A sin s = sincero


    (3) = 2.sin a sin ţA cos } B cos }C
    (+) = 2 cos ! A sin ? B sin ?
    28
    Tacos S - cos (S - A ) + cos (S - B ) + cos (S - C )*
    Proof. From the right-angled triangle OʻAF", in which AF' = 8.
    Note. — The first two values of tan re may be obtained from those of
    tan r by interchanging s and s - a .

    R = _ tan ža _ - - cos S
    900 (1) tan
    cos ( S - A )
    sin I
    = TT ja
    sin A cos žb cos žc
    _ 2 sin ja sin jb sin je

    _ - sin stsin (s - a ) + sin (s - 6 ) + & c.


    20

    PROOF. — The first value from the right-angled


    triangle OBD, in which Z OBD = S - A . The other
    values by the formulæ (887– 892).

    901 (1 ) tan R . = tan ğa = cos ( S - A )


    - cos S
    sin ja
    (B ) = in a sinih sin In

    w _ 2 sin ļa cos zb cosc

    (5 ) =_ $1
    sin s - sin (s – a ) + sin (s — b) + sin (s — c)
    20
    Proof. From the right-angled triangle O ' BD, in which Z O 'BD = 7 - S .
    192 SPHERICAL TRIGONOMETRY.

    SPHERICAL AREAS.

    902 area of ABC = (A + B + C - 1 ) ? = Er?,


    where E = A + B + C — , the spherical excess.
    Proof. — By adding the three lunes
    ABDC, BCEA, CAFB,
    and observing that ABF = CDE,
    weget (4 + + ) 2«r*= 2+1°+ 2ABO.
    903 AREA OF SPHERICAL POLYGON.
    n being the number of sides,
    Area = { Interior Angles - (n — 2) } po ?
    = {21 – Exterior Angles} po?
    = { 27 - sides of Polar Diagram } ?.
    The last value holds for a curvilinear area in the limit .
    Proof. — By joining the vertices with an interior point, and adding the
    areas of the spherical triangles so formed .

    904 Cagnoli's Theorem .


    sin 1 E = V sin s sin (s - a ) sin ( s b ) sin ( s - c ) ?
    2 cos a cos o cos c
    Proof. — Expand sin [ ] ( A + B) – (1 - C) ] by (628), and transform by
    Gauss's equations (897 i., iii.) and (669, 890 ) .

    905 Llhuillier 's Theorem .


    tan įE = v {tan is tan } (s - a ) tan ] (s — b ) tan ] (s — c ) } .
    Proof. — Multiply numerator and denominator of the left side by
    2 cos I ( A + B - ( + ) and reduce by (667, 668 ) , then eliminate (A + B)
    by Gauss's formulæ (897 i., iii.) Transform by (672, 673) , and substitute
    from (886 ).
    POLYHEDRONS. 193

    POLYHEDRONS.
    - - - -- -- - -- --- - - -
    Let the number of faces, solid angles, and edges , of any
    polyhedron be F , S , and E ; then
    906 F + S = E + 2.
    PROOF. - Project the polyhedron upon an internal sphere . Let m =
    number of sides, and s = sum of angles of one of the spherical polygons so
    formed. Then its area = { s - (m - 2 ) } r*, by (903 ). Sam this for all the
    polygons, and equate to 4art.
    - -- - - - - - - - --

    THE FIVE REGULAR SOLIDS.


    Let m be the number of sides in each face, n the number
    of plane angles in each solid angle ; therefore
    907 mF = nS = 2E .
    From these equations and (906 ), find F, S, and E in termsof m and
    n , thus ,
    1 mil 1 nil
    mn
    + = = + =
    2 2 Е т п 22

    1. >2;
    In order that F, S, and E may be positive, we must havem+n
    a relation which admits of five solutions in whole numbers , corresponding to
    the five regular solids . The values of m ,n , F , S, and E for the five regular
    solids are exbibited in the following table :
    n !FS E
    col
    co

    Tetrahedron ...
    ප|

    Hexahedron ... ...


    පාප ධ

    Octahedron ...
    Dodecahedron ...
    erCO
    ප ඵ

    Icosahedron . ...

    908 The sum of all the plane angles of any polyhedron


    SUT 2 n

    = 2m ( S - 2) ;
    Or, Four right angles for every vertex less eight right angles,
    2c
    194 SPHERICAL TRIGONOMETRY .

    909 If I be the angle between two adjacent faces of a


    regular polyhedron,
    sin } I = cos " • sin + . M

    Proof. -- Let PQ = a be the edge, and S


    the centre of a face, T the middle point of
    PQ , O the centre of the inscribed and circum
    scribed spheres, ABC the projection of PST
    upon a concentric sphere . In this spherical
    triangle,
    C = 5 , A = " , and B = * = PST. m
    Also STO = ; I.
    Now , by (881, ii.),
    cos A = sin B cos BC ;
    thatis, cos = sin sin ș1.
    n

    Q . e. d.

    If r, R be the radii of the inscribed and circumscribed


    spheres of a regular polyhedron ,

    910 r = tan }Icot R = tan }I tan


    Proof. – In the above figure, os = r, OP = R, PT = ; and
    OF .

    OS = PT cot " tan I. Also OP = PT cosec AC, and by (881,i.),


    m

    sin AC = tan BC cot A = cot ; I cot - ; therefore , & c.


    ELEMENTARY GEOMETRY.
    MISCELLANEOUS PROPOSITIONS.
    920 To find the point in a given line QY, the sum of whose
    distances from two fixed points S , S' is a minimum .
    Draw SYR at right angles to QY,
    making YR = YS. Join RS', cutting
    QY in P . Then P will be the required
    point.
    PROOF. - For, if D be any other point
    on the line, SD = DR and SP = PR.
    Bat RD + DS' is > RS'; therefore, & c.
    R is called the reflection of the point S,
    and SPS ' is the path of a ray of light K
    reflected at the line QY.
    If S, S ' and QY are not in the same plane , make SY, YR equal perpen .
    diculars as before, but the last in the plane of S ' and QY.
    Similarly, the point Q in the given line, the difference of whose distances
    from the fixed points Sand R ' is a maximum , is found by a like construction .
    The minimum sum of distances from S , S ' is given by
    (SP + S ' P )2 = SS2 + 4SY. S ' Y '.
    And the maximum difference from S and R ' is given by .
    (SQ - R 'Q ) = (SR')? – 4SY. R ' Y'.
    Proved by VI. D ., since SRR' S ' can be inscribed in a circle.

    921 Hence, to find the


    shortest distance from P
    to Q en route of the lines
    AB, BC , CD ; in other
    words, the path of the ray
    reflected at the successive
    surfaces AB, BC , CD.
    Find P7, the reflection of Pat
    the first surface ; then Pg, the
    reflection of P , at the second sur
    face ; next P3, the reflection of P ,
    at the third surface ; and so on if
    196 ELEMENTARY GEOMETRY.

    there be more surfaces. Lastly , join Q with Py, the last reflection , cutting
    CD in a . Join aP,, catting BOin b . Join bP , cutting AB in c . Join cP .
    PcbaQ is the path required .
    The same construction will give the path wben the surfaces are not, as
    in the case considered, all perpendicular to the same plane.

    922 If the straight line d from the vertex of a triangle


    divide the base into segments p , q , and if h be the distance
    from the point of section to the foot of the perpendicular from
    the vertex on the base , then
    6° + c = p + q* + 2d² + 2h (p — 9). (II. 12, 13.
    The following cases are important :
    (i.) When p = q, bº + c = 2q° +2°;
    i.e., the sum of the squares of two sides of a
    triangle is equal to twice the square of half
    the base, together with twice the square of the
    bisecting line drawn from the vertex .
    (ii.) When p = 2q , 26° + ¢* = 6q? + 3d? (II. 12 or 13)
    (iii.) When the triangle is isosceles,
    b = d = pq + d .

    923 If O be the centre of an equilateral triangle ABC and


    P any point in space. Then
    PA + PB² + PC* = 3 (PO? + 0A ).
    Proof. — PB + POʻ = 2PD + 2BD". (922, i.)
    Also PA² + 2PDP = 60D + 3POʻ, (922, ii.)
    and BO = 20D ;
    therefore, & c .

    CoR . - Hence, if P be any point on the


    surface of a sphere, centre 0 , the sum of the squares of
    its distances from A , B , C is constant. And if r , the radius
    of the sphere, be equal to 0A , the sum of the same squares is
    equal to 6r%.
    MISCELLANEOUS PROPOSITIONS. 197

    924 The sum of the squares of


    the sides of a quadrilateral is equal
    to the sum of the squares of the
    diagonals plus four times the square
    of the line joining themiddle points
    of the diagonals . (922, i.)
    925 COR. — The sum of the squares
    of the sides of a parallelogram is A
    equal to the sum of the squares of the diagonals.
    926 In a given line AC , to find a point X whose distance
    from a point P shall have a given
    ratio to its distance in a given
    direction from a line AB.
    Through P draw BPO parallel to the
    given direction . Produce AP , and make
    CE in tbe given ratio to CB. Draw PX
    parallel to EC, and XY to OB. There are
    two solutions when CE cuts AP in two
    points. [PROOF.— By (VI. 2 ) .

    927 To find a point X in AC,


    whose distance XY from AB parallel
    to BC shall have a given ratio to its
    distance XZ from BC parallel to AD.
    Draw AE parallel to BC, and having to
    AD the given ratio. Join BE cutting Að in
    X , the point required. [Proved by (VI. 2 ). BD z

    928 To find a point X on any


    line, straight or curved, whose
    distances XY, XZ, in given direc
    tions from two given lines AP , AB ,
    shall be in a given ratio .
    Take P any point in the first line.
    Draw PB parallel to the direction of XY,
    and BC parallel to that of XZ, making
    PB have to BC the given ratio . Join PÕ,
    cutting AB in D. Draw DE parallel to
    CB . Then AE produced cuts the line in
    X , the point required , and is the locus of
    such points. [ PROOF. — By (VI. 2) .
    198 ELEMENTARY GEOMETRY.

    929 To draw a line XY through a given point P so that


    the segments XP, PY, intercepted by a given circle, shall be
    in a given ratio .
    Divide the radius of the circle in that ratio ,
    and, with the parts for sides, construct a triangle
    PDC upon PC as base . Produce CD to cut the
    circle in X . Draw XPY and CY.
    Then PD + DC = radius ;
    therefore PD = DX ;
    But CY = CX ;
    therefore PD is parallel to CY ( 1.5 , 28 ); therefore
    & c., by (VI. 2 ).

    930 From a given point P in the


    side of a triangle, to draw a line PX
    which shall divide the area of the tri
    angle in a given ratio .
    Divide BC in D in the given ratio, and
    draw AX parallel to PD . PX will be the line
    required .
    ABD : ADC = the given ratio (VI. 1), and B
    APD = XPD (I. 37) ; therefore, & c.

    931 To divide the triangle ABC in a given ratio by a line


    XY drawn parallel to any given line AE.
    Make BD to DC in the given ratio . Then
    make BY a mean proportional to BE and BD,
    and draw YX parallel to EA.
    PROOF. — AD divides ABC in the given ratio
    (VI. 1). Now
    ABE : XBY :: BE : BD, (VI. 19 )
    or : : ABE : ABD ;
    therefore XBY = ABD.

    932 If the interior and exterior vertical angles at P of the


    triangle APB be bisected by straight lines which cut the base
    in C and D , then the circle circumscribing CPD gives the
    locus of the vertices of all triangles on the base AB whose
    sides AP, BP are in a constant ratio.
    MISCELLANEOUS PROPOSITIONS. 199

    PROOF. —
    The CPD = } (APB + BPE)
    = a right angle ;
    therefore P lies on the circumference of
    the circle, diameter CD (III, 31) . Also
    AP : PB :: A0 : OB :: AD : DB
    AC B
    (VI. 3, and A .), a fixed ratio.
    933 AD is divided harmonically in B and C ; i.e., AD : DB :: A0 : CB ;
    or, thewhole line is to one extreme part as the other extreme part is to the middle
    part. If we put a , b , c for the lengths AD , BD , OD, the proportion is
    expressed algebraically by a : b :: a - c : 6 - 6, which is equivalent to
    1 + 1 = 2.
    934 Also AP : BP = 08 : 00 = 00 : OB
    and AP? : BP = 0A : OB, (VI. 19)
    AP - ACP : 0P2 : BP - BOʻ. ( VI. 3, & B.)

    935 If Q be the centre of the inscribed circle of the triangle


    ABC , and if AQ produced meet the circumscribed circle ,
    radius R , in F ; and if FOG be a diameter, and AD perpendi
    cular to BC ; then
    (i.) FC = FQ = FB = 2R sin 4 .
    (ii.)Z FAD = FA0 = }(B - C),
    and _ CAG = \ ( B + C ).
    PROOF OF (i.)
    ZFQC = Q0A + QAC.
    But QAC = QAB = BCF; (III. 21)
    :: FQC = FOQ ; . . FO = FQ.
    Similarly FB = FQ.
    Also ZGOF is a right angle, and
    FGC = FAC = id ; (III. 21)
    :. FO = 2R sin
    936 If R , p be the radii of the circumscribed and inscribed
    circles of the triangle ABC (see last figure), and 0 , Q the
    centres ; then OQ = R - 2Rr.
    PROOF. - Draw QH perpendicular to A0 ; then QH = r . By the isosceles
    triangle AOF, OQ* = R - AQ .QF (922, iii.), and QF = FO (935, i.), and
    by similar triangles GFC, AQH , AQ : QH :: GF : FC ;
    therefore AQ .FC = GF .QH = 2Rr,
    200 ELEMENTARY GEOMETRY.

    The problems known as the Tangencies.


    937 Given in position any three of the following nine data
    viz., three points, three straight lines, and three circles, - it is
    required to describe a circle passing through the given points
    and touching the given lines or circles. The following five
    principal cases occur.
    938 I. Given two points A, B , and the straight line CD .
    ANALYSIS.— Let ABX be the required
    circle, touching CD in X . Therefore
    CX = CA.CB. (III. 36 )
    Hence the point X can be found, and the
    centre of the circle defined by the inter
    section of theperpendicularto CD through
    Å and the perpendicular bisector of AB.
    There are two solutions.
    Otherwise , by (926 ),making the ratio
    one of equality, and DO the given line. X
    Cor. — The point X thus determined is the point in CD at
    which the distance AB subtends the greatest angle . In the
    solution of (941) Q is a similar point in the circumference CD.
    (III. 21, & I. 16 )
    939 II. Given one point A and two straight lines DC, DE.
    In the last figure draw AOC perpendicular to DO, the bisector of the
    angle D , and make OB = 0A , and this case is solved by Case I .

    940 III. Given the point P , the straight line DE, and the
    circle ACF .
    Analysis. - Let PEF be the required
    circle touching the given line in E and the
    circle in F .
    Through H , the centre of the given
    circle, draw AHCD perpendicular to DE .
    Let K be the centre of the other circle.
    Join HK, passing through F , the point of
    contact. Join AF, EF, and AP, cutting
    the required circle in X . Then
    ZDHF = LKF ; (I. 27)
    therefore HFA = KFE (the halves of equal
    angles) ; therefore AF, FE are in the same D VEL
    straight line. Then, because AX .AP = AF . AE , (III. 36 )
    and AF. AE = AC .AD by similar triangles, therefore AX can be found .
    A circle must then be described through P and X to touch the given line,
    THE PROBLEMS OF THE TANGENCIES. 201

    by Case I. There are two solutions with exterior contact, as appears from
    Case I. These are indicated in the diagram . There are two more in which
    the circle AC lies within the described circle. The construction is quite
    analogous, C taking the place of A .
    941 IV . Given two points A , B and
    the circle CD .
    Draw any circle through A , B, cutting the
    required circle in C , D . Draw AB and DC ,
    and let them meet in P . Draw PQ to touch
    the given circle. Then , because
    PC .PD = PA . PB = PQ”, (III. 36 )
    and the required circle is to pass through
    A, B ; therefore a circle drawn through A , B , Q
    must toacb PQ , and therefore the circle CD ,
    in Q ( III. 37 ), and it can be described by Case
    1. There are two solutions corresponding to
    the two tangents from P to the circle CD .

    942 V . Given one point P , and two circles , centres A and B .

    M M o I K
    ANALYSIS. — Let PFG be the required circle touching the given ones in F
    and G . Join the centres QA, QB. Join FG , and produce it to cut the
    circles in E and H , and the line of centres in O . Then , by the isosceles
    triangles, the four apgles at E , F , G , H are all equal ; therefore AE, BG are
    parallel, and so are AF, BH ; therefore AO : BO : : AF : BH, and O is a
    centre of similitude for the two circles. Again , ZHBK = 2FLK , and
    FAM = 2FNM (III. 20 ) ; therefore FNM = HLK = HGK ( III.21) ; there
    fore the triangles OFN, OKG are similar ; therefore OF . OG = OK . ON ;
    therefore, if OP cat the required circle in X , OX . OP = OK . ON . Thus
    the point X can be found , and the problem is reduced to Case IV .
    Two circles can be drawn through P and X to touch the given circles.
    One is the circle PFX . The centre of the other is at the point where EA
    and HB meet if produced, and this circle touches the given ones in E and H .
    943 An analogous construction, employing the internal centre of simili.
    tude Oʻ, determines the circle which passes through P, and touches one given
    circle externally and the other internally . See ( 1047 - 9 ) .
    The centres of similitude are the two points which divide the distance
    between the centres in the ratio of the radi . See (1037 ).
    2D
    202 ELEMENTARY GEOMETRY.

    944 CoR. — The tangents from 0 to all circles which touch


    the given circles, either both externally or both internally,
    are equal.
    For the square of the tangent is always equal to OK . ON or OL . OM .
    945 The solutions for the cases of three given straight lines
    or three given points are to be found in Euc. IV., Props. 4 , 5 .
    946 In the remaining cases of the tangencies, straight lines
    and circles alone are given . By drawing a circle concentric
    with the required one through the centre of the least given
    circle , the problem can always be made to depend upon one
    of the preceding cases ; the centre of the least circle becoming
    one of the given points .
    947 DEFINITION . - A centre of similarity of two plane curves
    is a point such that, any straight line being drawn through it
    to cut the curves, the segments of the line intercepted between
    the point and the curves are in a constant ratio .
    948 If AB, AC touch a circle at B and
    C , then any straight line AEDF, cutting
    the circle , is divided harmonically by
    the circumference and the chord of con
    tact BC .
    Proof from AE .AF = AB”. (III. 36 )
    ABP = BD . DC + AD , (923)
    and BD . DC = ED .DF. (III. 35)
    949 If a , ß , y, in the same figure, be the
    perpendiculars to the sides of ABC from
    any point E on the circumference of the
    circle, then By = a '.
    Proof Draw the diameter BH = d ; then EBP = Bd, because BEH is a
    rightangle. Similarly EOʻ = yd . But EB . EC = ad (VI. D .), therefore & c .

    950 If FE be drawn parallel to the base


    BC of a triangle , and if EB , FC intersect
    in 0 , then
    AE : AC :: EO : OB : : FO : 00.
    By VI. 2 . Since each ratio = FE : BC.
    Cor. - If AC = n . AE, then
    BE = ( n + 1) OE .
    MISCELLANEOUS PROPOSITIONS. 203

    951 The three lines drawn from the angles of a triangle to


    the middle points of the opposite sides, intersect in the same
    point, and divide each other in the ratio of two to one.
    For, by the last theorem , any one of these lines is divided by each of the
    others in the ratio of two to one,measuring from the same extremity, and
    must therefore be intersected by them in the same point.
    This point will be referred to as the centroid of the triangle.

    952 The perpendiculars from the angles upon the opposite


    sides of a triangle intersect in the same point.
    Draw BE, CF perpendicular to the sides,and let
    them intersect in 0 . Let A0 meet BC in D . Circles
    will circumscribe AEOF and BFEC, by ( III. 31) ;
    therefore ZFA0 = FEO = FCB ; ( III. 21).
    therefore ZBDA = BFC = a rightangle;
    i.e., AO is perpendicular to BC, and therefore the
    perpendicular from A on BC passes through 0 .
    O is called the orthocentre of the triangle ABC. B
    COR. — The perpendiculars on the sides bisect the angles
    of the triangle DEF, and the point 0 is therefore the centre
    of the inscribed circle of that triangle
    Proof. From ( III.21), and the circles circumscribing OEAF and OECD.
    953 If the inscribed circle of a triangle ABC touches the
    sides a , b , c in the points D , E , F ; and if the escribed circle
    to the side a touches a and b , c produced in D ', E , F ' ; and if
    s = } (a + b + c) ;
    then
    BF" = BD' = CD = $ - 0,
    and AE = AF = $ ;
    and similarly with respect to
    the other segments .
    PROOF. — The two tangents from
    any vertex to either circle being equal,
    it follows that CD + c = half the
    perimeter of ABC, which is made up
    of three pairs of equal segments ;
    therefore CD = s - C.
    Also
    AE + AF = AC + CD + AB + BD
    = 2s ;
    therefore AE ' = AF = s. .
    204 ELEMENTARY GEOMETRY.

    The Nine- Point Circle .


    954 The Nine-point circle is the circle described through
    D , E , F , the feet of the perpendiculars on the sides of the
    triangle ABC. It also passes through the middle points of
    the sides of ABC and the middle points of 04, OB, OC ; in
    all, through nine points .
    PROOF. — Let thecircle
    cut the sides of ABC
    again in G , H , K ; and
    OA, OB, OC in L , M , N .
    ZEMF = EDF (III. 21)
    = 20DF (952, Cor.) ;
    therefore, since OB is the
    diameter of the circle cir
    cumscribing OFBD (III.
    31) , M is the centre of
    that circle ( III. 20) , and

    S
    therefore bisects OB.
    Similarly OC and VA 2
    are bisected at N and L . B
    Again , Z MGB = MED ( III. 22) = OCD, (IJI. 21), by the circle circum
    scribing OECD. Therefore MG is parallel to OC, and therefore bisects BC.
    Similarly H and K bisect CA and AB.
    955 The centre of the
    nine-point circle is the
    middle point of OQ , the
    line joining the ortho
    centre and the centre
    of the circumscribing
    circle of the triangle
    ABC .
    For the centre of the N . P .
    circle is the intersection of
    the perpendicular bisectors
    of the chords DG , EH, FK ,
    and these perpendiculars
    bisect OQ in the same point
    N , by (VI. 2 ).
    956 The centroid of
    the triangle ABC also lies on the line OQ and divides it in
    R so that OR = 2RQ .
    Proof. The triangles QHG , OAB are similar, and AB = 2HG ; there
    fore A0 = 2GQ ; therefore OR = 2RQ ; and AR = 2RG ; therefore R is the
    centroid , and it divides OQ as stated (951).
    CONSTRUCTION OF TRIANGLES. 205

    957 Hence the line joining the centres of the circumscribed


    and nine-point circles is divided harmonically in the ratio of
    2 : 1 by the centroid and the orthocentre of the triangle .
    These two points are therefore centres of similitude of the
    circumscribed and nine-point circles ; and any line drawn
    through either of the points is divided by the circumferences
    in the ratio of 2 : 1. See (1037 .)
    958 The lines DE , EF, FD intersect the sides of ABC in
    the radical axis of the two circles.
    For, if EF meets BC in P , then by the circle circumscribing BCEF,
    PE . PF = PC . PB ; therefore (III. 36 ) the tangents froni P to the circles
    are equal ( 985) .
    959 The nine-point circle touches the inscribed and escribed
    circles of the triangle .
    PROOF. — Let O be the orthocentre , and I, Q
    the centres of the inscribed and circumscribed
    circles . Produce AI to bisect the arc BC in T.
    Bisect AO in L , and join GL, cutting AT in S .
    The N . P . circle passes through G , D , and
    L (954) , and D is a right angle. Therefore
    GL is a diameter, and is therefore = R = QA
    (957). Therefore GL and QA are parallel.
    But QA = QT, therefore
    GS = GT = CT sin 4 = 2R sin®4 (935,i.)
    Also ST = 2GS cos 0
    (8 being the angle GST = GTS).
    N being the centre of the N . P . circle, its B bo
    radius = NG = { R ; and r being the radius of
    the inscribed circle, it is required to shew that
    NI = NG - r .
    Now NI = SN + SI° — 2SN . SI cos 0. (702)
    Substitute SN = R - GS ;
    SI = TI - ST = 2R sin 4 - 2GS cos e ;
    and GS = 2R sin ’ A , to prove the proposition .
    If J be the centre of the escribed circle touching BC, and r , its radius, it
    is shewn in a similar way that NJ = NG + ra.

    To construct a triangle from certain data.


    960 When amongst the data we have the sum or difference
    of the two sides AB , AC ; or the sum of the segments of the
    base made by AG , the bisector of the exterior vertical angle ;
    or the difference of the segments made by AF, the bisector of
    206 ELEMENTARY GEOMETRY.

    the interior vertical angle ; the following construction will


    lead to the solution .
    Make AE = AD = AC. Draw
    DH parallel to AF, and suppose
    EK drawn parallel to AG to meet
    the base produced in K ; and
    complete the figure. Then BE
    is the sum , and BD is the differ
    ence of the sides.
    BK is the sum of the exterior
    segments of the base, and BH is
    the difference of the interior seg .
    ments. < BDH = BEC = A ,
    LADO = EAG = } (B + C ) ,B É
    ZDCB = {DFB = } (C - B ).
    961 When the base and the vertical angle are given ; the
    locus of the vertex is the circle ABC in figure (935) ; and the
    locus of the centre of the inscribed circle is the circle , centre
    F and radius FB . When the ratio of the sides is given ,
    see (932 ).
    · 962 To construct a triangle when its form and the distances
    of its vertices from a point A ' are given .
    Analysis.— Let ABC be the required triangle. On
    A ' B make the triangle ABC” similar to ABC, so that
    AB : A ' B :: CB : C ' B . The angles ABA', CBC" will
    also be equal; therefore AB : BC :: AA : CC , which
    gives CC", since the ratio AB : BC is known. Hence
    the point C is found by constructing the triangle
    ACC'. Thus BC is determined, and thence the tri
    angle ABC from the known angles.

    - 963 To find the locus


    of a point P , the tan
    gent from which to a
    given circle, centre A ,
    has a constant ratio to
    its distance from a given
    point B .
    Let AK be the radius of
    the circle, and p : q the
    given ratio . On AB take
    AC, a third proportional to
    AB and AK , and make
    AD : DB = pº : .
    With centre D , and a radias
    TANGENTS TO TWO CIRCLES. 207

    equal to a mean proportional between DB and DC, describe a circle. Itwill


    be the required locus.
    Proov. - Suppose P to be a point on the required locus. Join P with
    A, B, C, and D .
    Describe a circle about PBC cutting AP in F , and another about ABF
    cutting PB in G , and join AG and EF . Then
    PK = AP – AKP = AP? — BA . AC (by constr.) = AP? — PA . AF (III. 36 )
    = AP .PF (II. 2) = GP . PB (III. 36).
    Therefore, by hypothesis18 ,,
    pa : q = GP . PB : PB$ = GP : PB = AD : DB (by constr.) ;
    therefore ZDPG = PGA (VI. 2) = PFB (III. 22) = PCB (III. 21).
    Therefore the triangles DPB, DCP are similar ; therefore DP is a mean pro
    portional to DB and DC. Hence the construction.
    964 COR . — If p = 9 the locus becomes the perpendicular
    bisector of BC , as is otherwise shown in (1003).

    965 To find the locus of a point P , the tangents from which


    to two given circles shall have a given ratio . (See also 1036 .)
    Let A , B be the centres, a , b the radii
    (a > b ), and p :q the given ratio . Take c, so
    that c : b = p : q, and describe a circle with
    centre A and radius AN = Va’ - C . Find the
    locus of P by the last proposition , so that
    the tangent from P to this circle may have
    the given ratio to PB. It will be the re
    quired locus.
    Proof. — By hypothesis and construction ,
    PK ( PK + AP? - a' + c AP? _ AN
    - PT 2 - PT + 72 BP BP2
    CoR . — Hence the point can be found on any curve from
    which the tangents to two circles shall have a given ratio .

    966 To find the locus of the point from which the tangents
    to two given circles are equal.
    Since , in ( 965) , we have now p = 9, and therefore c = b , the construction
    simplifies to the following :
    Take AN = (a - 6?), and in AB take AB : AN : AC. The perpen
    dicular bisector of BC is the required locus. But, if the circles intersect,
    then their common chord is at once the line required. See Radical Axis
    (985) .
    208 ELEMENTARY GEOMETRY.

    Collinear and Concurrent systems of points and lines .


    967 DEFINITIONS. — Points lying in the same straight line are
    collinear. Straight lines passing through the same point are
    concurrent, and the point is called the focus of the pencil of
    lines.
    Theorem . - If the sides of the triangle ABC , or the sides
    produced , be cut by any straight line in the points a , b , o
    respectively , the line is called a transversal, and the segments
    of the sides are connected by the equation
    968 (Ab : 6C ) (Ca : aB ) (Bc : CA ) = 1.
    Conversely , if this relation holds, the points a , b, c will be
    collinear .
    PROOF. — Through any vertex A draw AD
    parallel to the opposite side BC, to meet the
    transversal in D , then
    Ab : 60 = AD : Ca and Bc : A = aB : AD
    (VI. 4 ), which proves the theorem .
    Note. In the formula the segments of the
    sides are estimated positive, independently of B
    direction , the sequence of the letters being pre
    served the better to assist the memory . A point may be supposed to travel
    from A over the segments Ab, 6C , & c. continuously , until it reaches A again .
    969 By the aid of (701) the above relation may be put in
    the form
    (sin A Bb : sin bBC ) (sin CAa : sin a AB ) (sin BCc: sin cCA ) = 1
    970 If O be any focus in the plane of the triangle ABC , and
    if A0, BO , CO meet the sides in a , b , c ; then , as before,
    (Ab : 6C) (Ca : aB) ( Bc : CA ) = 1 .
    Conversely , if this relation holds, the lines Aa, Bb, Cc will
    be concurrent.
    PROOF. — By the trans
    versal Bb to the triangle
    AaC , we have ( 968)
    (Ab : 8C) (CB : Ba )
    x (a0 : 04) = 1.
    And , by the transversal
    Cc to the triangle AaB,
    ( Bc : CA) (AO : 0a)
    x (uC : CB ) = D1e. pa
    Multiply these equations ?
    together.
    COLLINEAR AND CONCURRENT SYSTEMS. 209

    971 If bc, ca, ab, in the last figure, be produced to meet the
    sides of ABC in P , Q , R , then each of the nine lines in the
    figure will be divided harmonically , and the points P , Q , R
    will be collinear.
    Proof. — (i.) Take 6P a transversal to ABC; therefore, by (968),
    (CP : PB) (BC : A ) (Ab : 6C ) = 1 ;
    therefore, by (970), CP : PB = Ca : aB.
    (ii.) Take CP a transversal to Abc, therefore
    (AB : Bc) (cP : Pb) (BC : CA ) = 1.
    But,by ( 970), taking O for focus to Abc,
    (AB : Bc) (cp : pb) (6C : CA) = l ;
    therefore cP : Pb = cp : pb.
    (iii.) Take PC a transversal to A0c, and b a focus to AOc; therefore, by
    (968 & 970), (Aa : a0) (OC : Cc) (cB : BA) = 1,
    and (Ap : p0 ) (OC : Cc) (cB : BA) = 1 ;
    therefore Aa : a0 = Ap : p0 .
    Thus all the lines are divided harmonically .
    (iv.) In the equation of (970 ) put Ab :60 = AQ : QC the harmonic
    ratio, and similarly for each ratio , and the result proves that P , Q , R are
    collinear, by ( 968) .
    CoR . - If in the same figure qr , rp, pq be joined, the three
    lines will pass through P , Q , R respectively.
    PROOF. — Take ( as a focus to the triangle abc, and employ (970 ) and the
    harmonic division of bc to show that the transversal rq cuts lc in P .

    972 If a transversal intersects the sides AB , BC, CD , & c .


    of any polygon in the points a , b , c, & c . in order , then
    (Aa : aB) (Bb : 6C) (Cc : cD ) ( Dd : dE ) ... & c . = 1 .
    Proof.-- Divide the polygon into triangles by lines drawn from one of the
    angles, and, applying (968 ) to each triangle, combine the results.

    973 Let any transversal cut the sides of a triangle and


    their three intersectors AO , BO , CO (see figure of 970 ) in the
    points A', B ', C', a ', b', c', respectively ; then, as before,
    (A 'B' : B'C') (C'a' : a'B ) ( B’é : c'A ') = 1.
    Proof. — Each side forms a triangle with its intersector and the trans
    versal. Take the four remaining lines in succession for transversals to each
    triangle, applying ( 968) symmetrically , and combine the twelve equations.
    2 E
    210 ELEMENTARY GEOMETRY.

    974 If the lines joining corresponding vertices of two tri


    angles ABC, abc are concurrent, the points of intersection
    of the pairs of corresponding
    sides are collinear, and con
    versely .
    Proof. — Let the concurrent lines
    Aa , Bb, Cc meet in 0 . Take bc,
    ca, ab transversals respectively to
    the triangles OBC , OCA , OAB , ap
    plying (968 ), and the product of the
    three equations shows that P , R , Q
    lie on a transversal to ABC .
    975 Hence it follows that, if the lines joining each pair of
    corresponding vertices of any two rectilineal figures are con
    current, the pairs of corresponding sides intersect in points
    which are collinear.
    The figures in this case are said to be in perspective, or in
    homology , with each other. The point of concurrence and
    the line of collinearity are called respectively the centre and
    axis of perspective or homology. See (1083).
    976 Theorem . — When three perpendiculars to the sides of a
    triangle ABC , intersecting them in the points a , b, c respec
    tively, are concurrent, the following relation is satisfied ; and
    conversely , if the relation be satisfied , the perpendiculars are
    concurrent.
    Abº - bC + Caº - aBº + Btº - c4° = 0.
    Proof. — If the perpendicnlars meet in 0 , then Abs — bCº = A0 ? – OC",
    & c. ( 1.47) .
    EXAMPLES. — By the application of this theorem , the concurrence of the
    three perpendiculars is readily established in the following cases : --
    ( 1) When the perpendiculars bisect the sides of the triangle
    ( 2 ) When they pass through the vertices. (By employing I. 47.)
    ( 3 ) The three radii of the escribed circles of a triangle at the points of
    contact betweeu the vertices are concurrent. So also. are the radius of the
    inscribed circle at the point of contact with one side, and the radii of the two
    escribed circles of the remaining sides at the points of contact beyond the
    included angle.
    In these cases employ the values of the segments given in (953 ) .
    (4 ) The perpendiculars equidistant from the vertices with three con
    current perpendiculars are also concurrent.
    (5 ) When the three perpendiculars from the vertices of one triangle upon
    the sides of the other are concurreut, then the perpendiculars from the
    vertices of the second triangle upon the sides of the first are also concurrent.
    Proof. — If A , B , C and A', B , C are corresponding vertices of the tri
    angles, join AB', AC , BC , BA’, CA , CB , and apply the theorem in conjunc
    tion with ( I. 47) .
    TRIANGLES CIRCUMSCRIBING A TRIANGLE. 211

    Triangles of constant species circumscribed to a triangle .


    977 Let ABC be any triangle, and ( any point ; and let
    circles circumscribe AOB, BOC, COA. The circumferences
    will be the loci of the vertices of a triangle of constant form
    whose sides pass through the
    points A , B , C .
    PROOF. - Draw any line b Ac from circle
    to circle, and produce bc , cB to meet in a .
    The angles AOB, COA are supplements
    of the angles c and b ( III. 22 ) ; therefore
    BOC is the supplement of a ( I. 32 ) ; there
    fore a lies on the circle OBC . Also, the
    angles at O being constant, the angles a , b, c
    are constant.

    978 The triangle abc is a maximum when its sides are per
    pendicular to 0A , OB , OC .
    Proof. — The triangle is greatest when its sides are greatest. But the
    sides vary as Oa , Ob , Oc, which are greatest when they are diameters of the
    circles ; therefore & c., by (III. 31).
    979 To construct a triangle of given species and of given
    limited magnitude which shall have its sides passing through
    three given points A , B , C .
    Determine O by describing circles on the sides of ABC to contain angles
    equal to the supplements of the angles of the specified triangle . Construct
    the figure abco independently from the known sides of abc, and the now
    known angles OLC = 0AC, OaC = OBC , & c. Thus the lengths Ou, Ob, Oc
    are found, and therefore the points a, b, c, on the circles, can be determined .
    The demonstrations of the following propositions will now be obvious.

    Triangles of constant species inscribed to a triangle .


    980 Let abc, in the last figure, be a fixed triangle, and
    any point. Take any point A on be, and let the circles cir
    cumscribing 0 Ac, OA cut the other sides in B , C . Then
    ABC will be a triangle of constant form , and its angles will
    have the values A = Oba + Oca , & c . (III. 21. )
    981 The triangle ABC will evidently be a minimum when
    0A , OB, OC are drawn perpendicular to the sides of abc.
    982 To construct a triangle of given form and of given
    limited magnitude having its vertices upon three fixed lines
    be, ca, ab .
    212 ELEMENTARY GEOMETRY.

    Construct the figure ABCO independently from the known sides of ABC
    and the angles at 0, which are equal to the supplements of the given angles
    a , b , c . Thus the angles O AC, & c . are found , and therefore the angles Obc ,
    & c ., equal to them ( III. 21 ) , are known. From these last angles the point 0
    can be determined , and the lengths 0A, OB, OC being known from the inde
    pendent figure , the points A , B , C can be found.
    Observe that, wherever the point O may be taken , the angles AOB, BOC
    COA are in all cases either the supplements of, or equal to, the angles c, a , b
    respectively ; while the angles aOb, 10c, coa are in all cases equal to CEC,
    A Í a, B = b.
    983 NOTE. - In general problems, like the foregoing, which
    admit of different cases, it is advisable to choose for reference
    a standard figure which has all its elements of the same affec
    tion or sign . In adapting the figure to other cases, all that
    is necessary is to follow the same construction , letter for
    letter, observing the convention respecting positive and
    negative, which applies both to the lengths of lines and to
    the magnitudes of angles, as explained in (607 — 609).

    Radical Axis.
    984 DEFINITION . — The radical axis of two circles is that
    perpendicular to the line of centres which divides the dis
    tance between the centres into segments , the difference of
    whose squares is equal to the difference of the squares of the
    radii .
    Thus, A, B being the centres , a , b the radii, and IP the
    the radicalaxis, Al - BI = a’ — .
    p (1)

    985 It follows that, if the circles intersect, the radical axis


    is their common chord ; and that, if they do not intersect, the
    radical axis cuts the line of centres in a point the tangents
    from which to the circles are equal ( I. 47).
    To draw the axis in this case, see ( 966 ).
    RADICAL AXIS. 213

    Otherwise : let the two circles cut the line of centres in C , D and C", D '
    respectively. Describe any circle through C and D , and another through
    Cand D , intersecting the former in E and F . Their common chord EF
    will cut the central axis in the required point I .
    Proof. — IC . ID = IE . IF = IC . ID ( III. 36 ) ; therefore the tangents
    from I to the circles are equal.

    986 Theorem . — The difference of the squares of tangents


    from any point P to two circles is equal to twice the rectangle
    under the distance between their centres and the distance of
    the point from their radical axis, or
    PK ? — PT = 2AB . PN .
    NLP

    PROOF.
    PK -- PT: = (AP? – BP2) - (a’ — 6 ) = (AQ? – BQ”) – (AI – B12),
    by (I. 47) & (984 ). Bisect AB in C, and substitute for each difference of
    squares, by ( II. 12 ) .

    987 Cor. 1 . - If P be on the circle whose centre is B , then


    PK = 2AB . PN
    988 Cor. 2. - If two chords be drawn through P to cut the
    circles in X , X', Y, Y' ; then, by (III. 36 ),
    PX .PX' – PY.PY' = 2AB . PN .
    989 If a variable circle intersect two given circles at con
    stant angles a and B, it will intersect their radical axis at a
    constant angle ; and its radius will bear a constant ratio to
    the distance of its centre from the radical axis. Or
    PN : PX = a cos a – b cos ß : AB ,
    214 ELEMENTARY GEOMETRY.

    Proof. — In the same figure, if P be the centre of the variable circle, and
    if PX = PY be its radius ; then, by (988 ),
    PX (XX’ – YY') = 2AB . PN .
    But XX' = 2a cos a and YY' = 26 cos ß ;
    therefore PN : PX = a cos a - b cos ß : AB,
    which is a constant ratio ifthe angles a, ß are constant.
    990 Also PX : PN = the cosine of the angle at which the
    circle of radius PX cuts the radical axis . This angle is
    therefore constant.
    991 Cor . - A circle which touches two fixed circles has its
    radius in a constant ratio to the distance of its centre from
    their radical axis.
    This follows from the proposition by making a = B = 0 or 27 .

    If P be on the radicalaxis ; then (see Figs. 1 and 2 of 984 )


    992 (i.) The tangents from P to the two circles are equal,
    or PK = PT. ( 986 )
    993 (ii.) The rectangles under the segments of chords
    through P are equal, or PX . PX' = PY . PY'. (988)
    994 (ii.) Therefore the four points X , X ', Y , Y' are con
    cyclic (III. 36 ) ; and, conversely , if they are concyclic , the
    chords XX ' , YY' intersect in the radical axis .
    995 DEFINITION . — Points which lie on the circumference of
    a circle are termed concyclic .
    996 (iv .) If P be the centre, and if PX = PY be the radius
    of a circle intersecting the two circles in the figure at angles
    a and ß ; then , by ( 993) , XX' = YY', or a cos a = b cos ß ;
    that is , The cosines of the angles of intersection are inversely
    as the radii of the fixed circles.
    997 The radical axes of three circles (Fig . 1046), taken two
    and two together, intersect at a point called their radical centre,
    Proof. — Let A , B , C b : tie centres , a , b , c the radii, and X , Y , Z the points
    in which the radical axes cut BC , CA, AB. Write the equation of the defini
    tion (984) for each pair of circles. Add the results, and apply (976 ) .

    998 A circle whose centre is the radical centre of three


    other circles intersects them in angles whose cosines are
    inversely as their radii (996 ).
    INVERSION. 215

    Hence, if this fourth circle cuts one of the others or


    thogonally , it cuts them all orthogonally .

    999 The circle which intersects at angles a , ß , y three fixed


    circles, whose centres are A , B , C and radii a , b , c , has its
    centre at distances from the radicalaxes of the fixed circles
    proportional to
    b cos ß - c cos y c cosy - a cosa a cos a — b cos ß
    ВС CA AB
    And therefore the locus of its centre will be a straight line
    passing through the radical centre and inclined to the three
    radical axes at angles whose sines are proportional to these
    fractions.
    Proof. — The result is obtained immediately by writing out equation (989)
    for each pair of fixed circles .

    The Method of Inversion .


    1000 DEFINITIONS. — Any two points P , P', situated on a
    diameter of a fixed circle
    whose centre is O and radius
    k , so that OP . OP' = k ”, are
    called inverse points with re
    spect to the circle , and either
    point is said to be the inverse
    of the other . The circle and óto
    its centre are called the circle
    and centre of inversion , and
    k the constant of inversion .
    1001 If every point of a plane figure be inverted with
    respect to a circle , or every point of a figure in space with
    respect to a sphere, the resulting figure is called the inverse
    or image of the original one.
    Since OP : k : OP', therefore
    1002 OP : 0P ' = 0P2 : k = k : 0P2.
    1003 Let D , D ', in the same figure, be a pair of inverse
    points on the diameter OD'. In the perpendicular bisector of
    DD', take any point Q as the centre of a circle passing through
    D , D ', cutting the circle of inversion in R , and any straight
    line throughº0 in the points P , P . Then , by (III. 36 ),
    OP .OP = OD . OD = ORP (1000 ). Hence
    216 ELEMENTARY GEOMETRY.

    1004 (i.) P , P are inverse points ; and , conversely , any


    two pairs of inverse points lie on a circle.
    1005 (ii.) The circle cuts orthogonally the circle of inver
    sion (III. 37) ; and, conversely , every circle cutting another
    orthogonally intersects each of its diameters in a pair of
    inverse points .
    1006 (ii.) The line IQ is the locus of a point the tangent
    from which to a given circle is equal to its distance from a
    given point D .
    1007 DEF. — The line IQ is called the axis of reflexion for
    the two inverse points D , D ', because there is another circle
    of inversion , the reflexion of the former, to the right of IQ ,
    having also D , D ' for inverse points.

    1008 The straight lines drawn from any point P , within


    or without a circle (Figs. 1 and 2 ), to the extremities of any
    chord AB passing through the inverse point Q , make equal
    angles with the diameter through PQ . Also, the four points
    0 , A , B , P are concyclic , and QA . QB = Q0 . QP .

    TP 0

    Proof. — In either figure OP : 0A : OQ and OP : 0B : 02 (1000 ),


    therefore, by similar triangles, ZOPA = OAB and OPB = OBA in figure
    ( 1) and the supplement of it in figure ( 2 ). But OAB = OBA ( I. 5 ), there
    fore, & c.
    Also, because ZOPA = OBA, the four points 0 . A , B , P lie on a circle in
    each case (III. 21), and therefore QA . QB = Q0 .QP (III. 35, 36 ).

    1009 The inverse of a circle is a circle , and the centre of


    inversion is the centre of similitude of the two figures. See
    also (1037) .
    PROOF. - In the figure of (1043), let O be the point where the common
    tangent RT of the two circles, centres A and B, cuts the central axis, and let
    any other line through O cut the circles in P , Q P ', Q . Then, in the demon
    stration of (942) , it is shown that OP . 00 = 0Q . OP = ka, a constant
    quantity . Therefore either circle is the inverse of the other, k being the
    radius of the circle of inversion .
    INVERSION . 217

    1010 To make the inversions of two given circles equal


    circles .
    RULE. — Take the centre of inversion so that the squares of
    the tangents from it to the given circles may be proportional to
    their radii (965 ).
    Proof. — (Fig . 1043) AT : BR = OT : OR = OT : k ”, since OT : k : OR .
    Therefore OT? : AT = ki® : BR , therefore BR remains constant if OT & AT.

    1011 Hence three circles may be inverted into equal circles,


    for the required centre of inversion is the intersection of two
    circles that can be drawn by (965) .

    1012 The inverse of a straight line is a circle passing


    through the centre of inversion .
    Proof. — Draw OQ perpendicular to the
    line, and take P any other point on it. Let
    Q , P ' be the inverse points. Then OP . OP =
    0Q . OQ '; therefore, by similar triangles,
    20PQ = OQP, a right angle ; and OQ' is
    constant, therefore the locus of P is the
    circle whose diameter is OQ '.
    1013 EXAMPLE. — The inversion of a poly
    gon produces a figure bounded by circular
    arcs which intersect in angles equal to the
    corresponding angles of the polygon, the
    complete circles intersecting in the centre
    of inversion .

    1014 If the extremities of a straight line P'Q ' in the last


    figure are the inversions of the extremities of PQ , then
    PQ : PQ = (OP . OQ) : V (OP . OQ ).
    PROOF. - By similar triangles, PQ : PQ = OP : OQ and PQ : PQ =
    OQ : 0P . Compound these ratios.
    1015 From the above it follows that any homogeneous
    equation between the lengths of lines joining pairs of points
    in space, such as PQ . RS. TU = PR . QT. SU , the same
    points appearing on both sides of the equation , will be
    true for the figure obtained by joining the corresponding
    pairs of inverse points.
    For the ratio of each side of the equation to the corresponding side of the
    equation for the inverted points will be the same, namely ,
    ✓ (OP. OQ . OR ...) : ✓ (OP . OQ . OR ...).
    2F
    218 ELEMENTARY GEOMETRY.

    Pole and Polar .


    1016 DEFINITION . — The polar of any point P with respect
    to a circle is the perpendicular to the diameter OP (Fig. 1012)
    drawn through the inverse point P .
    101 It follows that the polar of a point exterior to the
    circle is the chord of contact of the tangents fron the point ;
    that is , the line joining their points of contact.
    1018 Also, P Q is the polar of P with respect to the circle,
    centre 0 , and PQ is the polar of Q . In other words, any
    point P lying on the polar of a point Q ', has its own polar
    always passing through Q '.

    1019 The line joining any two points P , p is the polar of


    Q', the point of intersection of their polars.
    Proof. -- The point Q lies on both the lines P ' Q', p' Q ', and therefore has
    its polar passing through the pole of each line, by the last theorem .
    1020 The polars of any two points P , p , and the line joining
    the points form a self-reciprocal triangle with respect to
    the circle, the three vertices being the poles of the opposite
    sides. The centre () of the circle is evidently the orthocentre
    of the triangle (952). The circle and its centre are called the
    polar circle and polar centre of the triangle.
    If the radii of the polar and circumscribed circles of a
    triangle ABC be 7 and R , then
    72 = 4R² cos A cos B cos C .
    PROOF. - In Fig. (952) , O is the centre of the polar circle, and the circles
    described round ABC, BOC, COA, AOB are all equal ; because the angle
    BOO is the supplement of A ; & c . Therefore 2R . OD = OB . OC (VI. C )
    and = 0A . OD = 0A . OB . OC = 2R . Also , OA = 2R cos A by a diameter
    through B , and (III . 21).

    Coaxal Circles.
    1021 DEFINITION . — A system of circles having a common
    line of centres called the central axis , and a common radical
    axis , is termed a coaxal system .
    1022 If O be the variable centre of one of the circles, and
    COAXAL CIRCLES. 219

    OK its radius, the whole system is included in the equation


    012 – OK ? = + 8 ?,
    where 8 is a constant length .

    1023 In the first species (Fig . 1),


    01 — OK = 82,
    and & is the length of the tangent from I to any circle of the
    system ( 985 ) . Let a circle, centre I and radius d, cut the
    central axis in D , D '. When 0 is at D or I ', the circle whose
    radius is OK vanishes. When 0 is at an infinite distance,
    the circle developes into the radical axis itself and into a line
    at infinity .
    The points D , D' are called the limiting points.
    1024 In the second species (Fig. 2),
    OR – 01 = 82,
    and 8 is half the chord RR common to all the circles of
    the system . These circles vary between the circle with
    centre I and radius d , and the circle with its centre at infinity
    as described above. The points R , R ' are the common points
    of all circles of this system . The two systems are therefore
    distinguished as the limiting points species and the common
    points species of coaxal circles.
    220 ELEMENTARY GEOMETRY.

    1025 There is a conjugate system of circles having R , R for


    limiting points, and D , D ' for common points, and the circles
    of one species intersect all the circles of the conjugate system
    of the other species orthogonally (1005 ).
    Thus, in figures (1 ) and ( 2), Q is the centre of a circle
    of the opposite species intersecting the other circles or
    thogonally .
    1026 In the first species of coaxal circles, the limiting
    points D , D ' are inverse points for every circle of the system ,
    the radical axis being the axis of reflexion for the system .
    Proof .— (Fig . 1) 01 – 02 = OK ,
    therefore OD . OD' = OK , (II. 13 )
    therefore D, D are inverse points (1000 ).
    1027 Also, the points in which any circle of the system
    cuts the central axis are inverse points for the circle whose
    centre is I and radius d. [Proof. — Similar to the last.
    1028 PROBLEM. - Given two circles of a coaxal system , to
    describe a circle of the same system - (i.) to pass through a
    given point; or ( ii.) to touch a given circle ; or ( ii .) to cut a
    given circle orthogonally.
    1029 I. If the system be of the common points species, then , since the
    required circle always passes through two known points , the first and second
    cases fall under the Tangencies. See (941).
    1030 To solve the third case, describe a circle through the given common
    points, and through the inverse of either of them with respect to the given
    circle, which will then be cut orthogonally , by (1005 ) .
    1031 II. If the system be of the limiting points species, the problem is
    solved in each case by the aid of a circle of the conjugate system . Such a
    circle always passes through the known limiting points, and may be called a
    conjugate circle of the limiting points system . Thus,
    1032 To solve case (i.) - Draw a conjugate circle through the given point,
    and the tangent to it at that point will be the radius of the required circle.
    1033 To solve case (ii.) — Draw a conjugate circle through the inverse
    of either limiting point with respect to the given circle, which will thus be
    cut orthogonally, and the tangent to the cutting circle at either point of
    intersection will be the radius of the required circle .
    1034 To solve case (iii.) - Draw a conjugate circle to touch thegiven one,
    and the common tangent of the two will be the radius of the required circle.
    1035 Thus, according as we wish to make a circle of the system touch , or
    cut orthogonally, the given circle , we must draw a conjugate circle to cut
    orthogonally, or touch it.
    CENTRES AND AXES OF SIMILITUDE. 221

    1036 If three circles be coaxal, the squares of the tangents


    drawn to any two of them from a point on the third are in
    the ratio of the distances of the centre of the third circle from
    the centres of the other two .
    PROOF. — Let A , B , C be the centres of the circles ; PK , PT the tangents
    from a point P on the circle , centre C , to the other two ; PN the perpen
    dicular on the radical axis. By ( 986 ),
    PK ? = 2A0 . PN and PT? = 2BC . PN ,
    therefore PK’ : PT ? = AC : BC.

    Centres and axes of similitude.


    1037 DEFINITIONS. — Let 00ʻ be the centres of similitude
    (Def. 947) of the two circles in the figure below , and let any
    line through O cut the circles in P , Q , P ', Q . Then the
    constant ratio OP : 0P = OQ : OQ is called the ratio of
    similitude of the two figures ; and the constant product
    OP . OQ ' = 0Q . OP is called the product of anti -similitude.
    See (942), ( 1009), and ( 1043).
    The corresponding points P , P' or Q , Q on the same
    straight line through O are termed homologous, and P , Q ' or
    Q , P are termed anti -homologous.
    1038 Let any other line Opqp' be drawn through 0 .
    Then , if any two points P , p on the one figure be joined , and
    if P ', p', homologous to P , p on the other figure, be also joined ,
    the lines so formed are termed homologous. But if the points
    which are joined on the second figure are anti-homologous to
    those on the first, the two lines are termed anti-homologous.
    Thus, Pq, Q 'p are anti-homologous lines.
    1039 The circle whose centre is 0 , and whose radius is
    equal to the square root of the product of anti-similitude, is
    called the circle of anti -similitude.

    1040 The four pairs of homologous chords Pp and Pp',


    Qq and Q ' , Pq and P' , Qp and Qp' of the two circles in the
    figure are parallel. And in all similar and similarly situated
    figures homologous lines are parallel.
    ProoF. - By (VI. 2) and the definition (947).
    222 ELEMENTARY GEOMETRY.

    1041 The four pairs of anti-homologous chords, Pp and


    Q 'q , Qq and Pp , Pq and Qp , Qp and Pğ , of the two circles
    meet on their radical axis .
    PROOF. — OP . OQ = Op . Oq = k ,
    where k is the constant of inversion ; therefore P , P , Q , ý are concyclic ;
    therefore Pp and Q 'll meet on the radical axis. Similarly for any other pair
    of anti-homologous chords.
    1042 COR. — From this and the preceding proposition it
    follows that the tangents at homologous points are parallel ;
    and that the tangents at anti-homologous points meet on the
    radical axis . For these tangents are the limiting positions of
    homologous or anti-homologous chords. (1160 )
    1043 Let C , D be the inverse points of 0 with respect to
    two circles , centres A and B ; then the constant product of
    anti- similitude
    OP .OQ or OQ .OP' = 0A . OD or OB . OC.

    CA TOT
    CENTRES AND AXES OF SIMILITUDE. 223

    PROOF.— By similar right-angled triangles,


    04 : 01 : 0C and OB : OR : OD ;
    therefore 0A . OD = OB . OC ... ...... (1),
    and also 04 .0C = OT = OP . OQ, (III. 36)
    and OB . OD = 0R² = OP . OQ”;
    therefore OA . OB . OC . OD = OP. OQ . OP . OQ,
    therefore & c., by ( 1).
    1044 The foregoing definitions and properties ( 1037 to
    1043 ), which have respect to the external centre of similitude
    0 , hold good for the internal centre of similitude 0', with the
    usual convention of positive and negative for distances
    measured from ' upon lines passing through it.
    1045 Two circles will subtend equal angles at any point on
    the circumference of the circle whose diameter is 00 ', where
    0 , 0 are the centres of similitude (Fig. 1043). This circle is
    also coaxal with the given circles, and has been called the .
    circle of similitude.
    Proof. — Let A, B be the centres, a, b the radii, and K any point on the
    circle, diameter 00'. Then, by (932),
    KA : KB = AO : B0 = AO : BOʻ = a : b,
    by the definition ( 943) ;
    therefore a : KA = b : KB ;
    that is, the sines of the halves of the angles in question are equal, which
    proves the first part . Also , because the tangents from K are in the constant
    ratio of the radii a , b , this circle is coaxal with the given ones,by (1036, 934 ).

    1046 The six centres of similitude P , P, Q , q , R , r of three


    circles lie three and three on four straight lines PQR , Pqr,
    Qpr, Rpg , called axes
    of similitude.
    PROOF. – Taking any
    three of the sets of points
    named, say P , q , r, they are
    shewn at once to be col.
    linear by the transversal
    theorem ( 968 ) applied to
    the triangle ABC.
    For the segments of its
    sides made by the points
    P, 9, r are in the ratios of
    the radii of the circles.
    224 ELEMENTARY GEOMETRY.

    1047 From the investigation in (942), it appears that one


    circle touches two others in a pair of anti-homologous points,
    and that the following rule obtains :
    RULE . — The right line joining the points of contact passes
    through the external or internal centre of similitude of the two
    circles according as the contacts are of the same or of different
    kinds.
    1048 DEFINITION. - Contact of curves is either internal or external ac
    cording as the curvatures at the point of contact are in the same or opposite
    directions.

    1049 Gergonne' s method of describing the circles which touch


    three given circles.
    Take Pgr, one of the four axes of similitude, and find its poles a , b, y
    with respect to the given circles, centres A, B, C (1016 ). From O , the
    radical centre, draw lines through a, b, y, cutting the circles in a, a , b, b ,
    c, 6 . Then a , b , c and a , b , c will be the points of contact of two of the
    required circles.
    PROOF. — Analysis.- Let the circles
    E , F touch the circles A , B, C in
    a , b , c, a', b', '. Let bc, b 'c' meet in
    P ; ca , c' á' in q ; and ab , a'b ' in r.
    Regarding E and F as touched by
    A , B , C in turn, Rule (1047) shews
    that aa', bb', cc' meet in 0 , the centre
    of similitude of E and F ; and ( 1041)
    shews that P, 7, and lie on the
    radical axis of E and F .
    Regarding B and C , or C and A ,
    or A and B , as touched by E and F
    in turn , Rule ( 1047) shews that P , q, r .
    are the centres of similitude of B and P /
    C , C and A , A and B respectively ;
    and ( 1041) shews that () is on the radical axis of each pair, and is therefore
    the radical centre of A , B , and C .
    Again , because the tangents to E and F , at the anti-homologous points
    a , a ', meet on Pqr, the radical axis of E and F (1042) ; therefore the point
    of meeting is the pole of aa' with respect to the circle A (1017). Therefore
    au passes through the pole of the line Par (1018). Similarly , bü' and cc
    pass through the poles of the same line Pqr with respect to B and C . Hence
    the construction .
    1050 In the given configuration of the circles A , B , C , the
    demonstration shews that each of the three internal axes of
    similitude Por, Qrp, Rpq ( Fig . 1046 ) is a radical axis and
    common chord of two of the eight osculating circles which
    can be drawn. The external axis of similitude PQR is the
    ANHARMONIC RATIO . 225

    radical axis of the two remaining circles which touch A , B ,


    and C either all externally or all internally .
    1051 The radical centre 0 of the three given circles is also
    the common internal centre of similitude of the four pairs of
    osculating circles. Therefore the central axis of each pair
    passes through 0 , and is perpendicular to the radical axis.
    Thus, in the figure, EF passes through 0 , and is perpen
    dicular to Pqr .
    Anharmonic Ratio.
    1052 DEFINITION. — Let a pencil of
    four lines through a point o be cut
    by a transversal in the points A , B ,
    C, D . The anharmonic ratio of the
    pencil is any one of the three frac
    tions
    AB . CD AB . CDAD . BC
    AD . BC OF AC . BD OF AC . BD
    1053 The relation between these three different ratios is
    obtained from the equation
    AB . CD + AD . BC = AC . BD.
    Denoting the terms on the left side by p and q, the three anharmonic
    ratios may be expressed by
    P : 9 , P : p + q, 9 : p + q .
    The ratios are therefore mutually dependent. Hence, if the identity
    merely of the anharmonic ratio in any two systems is to be established, it is
    immaterial which of the three ratios is selected .
    1054 In future, wben the ratio of an anharmonic pencil { O , ABCD is
    mentioned , the form AB . CD : AD . BC will be the one intended, whatever
    the actual order of the points A , B , C , D may be. For, it should be observed
    that, by making the line OD revolve about 0 , the ratio takes in turn each of
    the forms given above. This ratio is shortly expressed by the notation
    {0, ABCD }, or simply {ABCD }.
    1055 If the transversal be drawn parallel to one of the lines, for instance
    OD , the two factors containing D become infinite, and their ratio becomes
    unity. They may therefore be omitted . The anharmonic ratio then reduces
    to AB : BC. Thus, when D is at infinity , wemay write
    { O, ABC 00 } = AB : BC.
    1056 The anharmonic ratio
    AB . CD sin AOB sin COD
    AD . BC sin AOD sin BOC '
    2 a
    226 ELEMENTARY GEOMETRY.

    and its value is therefore the same for all transversals of the
    pencil.
    PROOF. — Draw OR parallel to the transversal, and let p be the perpendi
    cular from A upon OR . Multiply each factor in the fraction by p . Then
    substitute p . AB = 0A . OB sin AOB, & c . (707) .
    1057 The anharmonic ratio (1056 ) becomes harmonic when its value is
    unity . See ( 933 ) . The harmonic relation there defined may also be stated
    thus : four points divide a line harmonically when the product of the extreme
    segments is equal to the product of the whole line and the middle segment.
    Homographic Systems of Points.
    1058 DEFINITION. — If x , a , b , c be the distances of one
    variable point and three fixed points on a straight line from a
    point 0 on the same ; and if a , a , b , c be the distances of
    similar points on another line through 0 ; then the variable
    points on the two lines will form two homographic systems
    when they are connected by the anharmonic relation
    1059 (« — a ) (b — c) _ (it' - a ) (6 - c )
    (x - c) (a — b ) (ax ' — c ) (a' — 6 )
    Expanding, and writing A , B , C , D for the constant coeffi.
    cients , the equation becomes
    1060 Axx' + Bx + Cx + D = 0.
    From which
    Cr + D and X - _ BirtD
    1061 X =
    Att B Ax + C
    1062 THEOREM . – Any four arbitrary points X1, X. , X3, X , on
    one of the lines will have four corresponding points é , a ,a's, x'
    on the other determined by the last equation , and the two sets
    of points will have equal anharmonic ratios.
    Proof. — This may be shown by actual substitution of the value of each z
    in terms of x', by (1061), in the harmonic ratio {x cylgl;}.
    1063 If the distances of four points on a right line from a
    point 0 upon it , in order, are a , a , ß , ß ', where a, ß ; a', ß are
    the respective roots of the two quadratic equations
    ax? + 2hx + 6 = 0 , a'r ? + 2h'x + b' = 0 ;
    the condition that the two pairs of points may be harmonically
    conjugate is
    1064 ab'ta'b = 2hh'.
    INVOLUTION 227

    Proof.— The harmonic relation,by (1057), is


    ( a - a ') (6 - 0 ') = (a - B ) (a' - 1 ).
    Multiply out, and substitute for the sums and products of the roots of the
    quadratics above in terms of their coefficients by (51, 52) .

    1065 If un, u , be the quadratic expressions in ( 1063 ) for two pairs of


    points, and if u represent a third pair harmonically conjugate with u , and ua,
    then the pair of points u will also be harmonically conjugate with every pair
    given by the equation un + du , = 0 , where 1 is any constant. For the con
    dition (1064 ) applied to the last equation will be identically satisfied .

    1066 DEFINITIONS. — Pairs of inverse points PP', QQ , & c., on


    the same right line , form a system in involution , and the rela
    tion between them , by (1000), is
    OP . OP = OQ . OQ = & c . = k”.
    P Q B
    The radius of the circle of inversion is k , and the centre
    O is called the centre of the system . Inverse points are also
    termed conjugate points .
    When two inverse points coincide, the point is called a
    focus.
    1067 The equation OP2 = 72 shows that there are two foci
    A , B at the distance k from the centre, and on opposite sides
    of it , real or imaginary according as any two inverse points
    lie on the same side or on opposite sides of the centre.
    1068 If the two homographic systems of points in (1058)
    be on the same line, they will constitute a system in involu
    tion when B = C .
    Proof. — Equation (1060 ) may now be written
    Axé + H (2 + x') + B = 0 ,
    (2+ 1 )(x + 1 ) = - 1 = ,
    a constant. Therefore - is the distance of the origin O from the centre
    of inversion . Measuring from this centre, the equation becomes g = k”,
    representing a system in involution .

    1069 Any four points whatever of a system in involution on


    a right line have their anharmonic ratio equal to that of their
    four conjugates .
    228 ELEMENTARY GEOMETRY.

    Proof. — Let p , p'; , 4 ; r, r'; 8, s' be the distances of the pairs of inverse
    points from the centre.
    In the anharmonic ratio of any four of the points, for instance {poʻrs),
    substitute p = k + p , a = k = 9 , & c ., and the result is the anharmonic
    ratio { p 'qr's'} .

    1070 Any two inverse points P , P are in harmonic relation


    with the foci A , B .
    __ P _ B_
    Proof. — Let P, º be the distances of P, P from the centre 0 ; then
    pp' = k ”, therefore ? tha p' tk _ ktp .
    p' - k k - D
    that is, (933)
    BP = BP
    1071 If a system of points in involution be given , as in
    ( 1068), by the equation
    Axx' + H (x + x ) + B = 0 ... ............ (1) ;
    and a pair of conjugate points by the equation
    ax? + 2hx + b = 0 .. . ............ (2 ) ;
    the necessary relation between a , h , and b is
    1072 Ab + Ba = 2Hh.
    PROOF. — The roots of equation ( 2 ) must be simultaneous values of x , at in
    ( 1) ; therefore substitute in ( 1)
    * + = - 21 and vá'= (51)
    1073 CoR.- A system in involution may be determined from two given
    pairs of corresponding points.
    Let the equations for these points be
    ax ? + 2hx + 6 = 0 and a'w + 2hx + b' = 0 .
    Then there are two conditions ( 1072),
    Ab + Ba = 24h and AB + B ^ = 271 ,
    from which A, H , B can be found.
    A geometrical solution is given in (985 ). C , D ; C , D are, in that con
    struction, pairs of inverse points, and I is the centre of a system in involution
    defined by a series of coaxal circles (1022) . Each circle intersects the
    central axis in a pair of inverse points with respect to the circle whose centre
    is 0 and radius d.

    1074 The relations which have been established for a system of collinear
    points may be transferred to a system of concurrent lines by the method of
    ( 1056 ) , in which the distance between two points corresponds to the sine of
    the angle between two lines passing through those points.
    METHODS OF PROJECTION . 229

    The Method of Projection .


    1075 DEFINITIONS.— The projection of any point P in space
    (Fig . of 1079 ) is the point p in which a right line OP , drawn
    from a fixed point o called the vertex, intersects a fixed plane
    called the plane of projection .
    If all the points of any figure, plane or solid , be thus pro
    jected , the figure obtained is called the projection of the
    original figure.
    1076 Projective Properties. The projection of a right line is
    a right line. The projections of parallel lines are parallel. The
    projections of a curve, and of the tangent at any point of it , are
    another curve and the tangent at the corresponding point.
    10717 The anharmonic ratio of the segments of a right line
    but two transversals of the same anharmonic pencil. (1056 )
    1078 Also , any relation between the segments of a line
    similar to that in (1015) , in which each letter occurs in every
    term , is a projective property. [Proof as in (1056 ).
    1079 Theorem . - Any quadrilateral PQRS may be projected
    into a parallelogram .
    CONSTRUCTION .
    Produce PQ , SR to
    meet in A , and PS,
    QR to meet in B .
    Then , with any
    point o for vertex,
    project the quadri.
    lateral upon any
    plane pab parallel to
    OAB. The projected
    figure pqrs will be a
    parallelogram ,
    PROOF. - The
    planes OPQ , ORS in .
    tersect in OA , and
    they intersect the
    plane of projection
    which is parallel to
    04 in the lines pq,
    rs. Therefore pq and
    T8 are parallel to 0 A ,
    and therefore to each
    other. Similarly , ps,
    qr are parallel to OB.
    230 ELEMENTARY GEOMETRY.

    1080 Cor . 1. — The opposite sides of the parallelogram pars meet in two
    points at infinity, which are the projections of the points A , B ; and AB
    itself, which is the third diagonal of the complete quadrilateral PQRS, is
    projected into a line at infinity.
    1081 Hence, to project any figure so that a certain line in it may pass to
    infinity - Take the plane of projection parallel to the plane which contains the
    given line and the vertex .
    1082 Cor. 2.— To make the projection of the quadrilateral a rectangle ,
    it is only necessary to make AOB a right angle .
    On Perspective Drawing .
    1083 Taking the parallelogram pqrs, in (1079), for the original figure,
    the quadrilateral PQRS is its projection on the plane ABab . Suppose this
    plane to be the plane of the paper . Let the planes OAB, pab, while
    remaining parallel to each other, be turned respectively about the fixed
    parallel lines AB, ab. In every position of the planes, the lines Op, Oq, Or,
    Os will intersect the dotted lines in the same points P , Q , R , S . When the
    planes coincide with that of the paper, pqrs becomes a ground plan of the
    parallelogram , and PQRS is the representation of it in perspective.
    AB is then called the horizontal line, ab the picture line, and the plane of
    both the picture plane.
    1084 To find the projection of any point p in the ground
    plan.
    Rule. - Druw pb to any point b in the picture line, and draw OB parallel
    to pb, to meet the horizontal line in B Join Op, Bb, and they will intersect in
    P , the point required .
    In practice, pb is drawn perpendicular to ab, and OB therefore perpendi
    cular to AB. The point B is then called the point of sight, or centre of vision ,
    and O the station point.

    1085 To find the projection of a point in the ground plan ,


    not in the original plane, but at a perpendicular distance c
    above it .
    RULE. - Take a new picture line parallel to the former, and at a distance
    above it = c coseca, where u is the angle between the original plane and the
    plane of projection . For a plane through the given point, parallel to the
    original plane, will intersect the plane of projection in the new picture line
    so constructed .
    Thus, every point of a figure in the ground plan is transferred to the
    drawing.
    1086 The whole theory of perspective drawing is virtually included in
    the foregoing propositions. The original plane is commonly horizontal, and
    the plane of projection vertical. In this case, cosec a = l, and the height of
    the picture line for any point is equal to the height of the point itself above
    the original plane.
    The distance BO, when B is the point of sight, may be measured along
    AB, and bp along ab, in the opposite direction ; for the line Bb will continue
    to intersect Op in the point P .
    METHODS OF PROJECTION . 231

    Orthogonal Projection .
    1087 DEFINITION . — In orthogonal projection the lines of
    projection are parallel to each other, and perpendicular to the
    plane of projection . The vertex in this case may be consi
    dered to be at infinity.
    1088 The projections of parallel lines are parallel, and the
    projected segments are in a constant ratio to the original
    seginents.
    1089 Areas are in a constant ratio to their projections.
    For, lines parallel to the intersection of the original plane and the plane
    ofprojection are unaltered in length , and lines at right angles to the former
    are altered in a constant ratio . This ratio is the ratio of the areas, and is
    the cosine of the angle between the two planes.

    Projections of the Sphere.


    1090 In Stereographic projection , the vertex is on the sur
    face of the sphere, and the diameter through the vertex is
    perpendicular to the plane of projection which passes through
    the other extremity of the diameter. The projection is there
    fore the inversion of the surface of the sphere (1012) , and the
    diameter is the constant k .
    1091 In Globular projection , the vertex is taken at a dis
    tance from the sphere equal to the radius V2, and the
    diameter through the vertex is perpendicular to the plane of
    projection .
    1092 In Gnomonic projection ,which is used in the construc
    tion of sun -dials , the vertex is at the centre of the sphere.
    1093 Mercator's projection ,which is employed in navigation ,
    and sometimes in maps of the world , is not a projection at
    all as defined in (1075 ) . Meridian circles of the sphere are
    represented on a plane by parallel right lines at intervals
    equal to the intervals on the equator. The parallels of lati.
    tude are represented by right lines perpendicular to the
    meridians, and at increasing intervals , so as to preserve the
    actual ratio between the increments of longitude and latitude
    at every point.
    With r for the radius of the sphere , the distance, on the chart, from the
    equator of a point whose latitude is 1 , is = r log tan (45° + } ^ ) .
    232 ELEMENTARY GEOMETRY.

    Additional Theorems.
    1094 The sum of the squares of the distances of any point
    P from n equidistant points on a circle whose centre is O and
    radius r . = n (r ? + 0P2) .
    Proor. – Sum the values of PB”, PC?, & c., given in (819 ), and apply (803).
    This theorem is the generalization of (923 ) .
    1095 In the same figure, if P be on the circle , the sum of
    the squares of the perpendiculars from P on the radii OB , OC ,
    & c., is equal to įnr .
    PROOF. — Describe a circle upon the radius through P as diameter, and
    apply the foregoing theorem to this circle .
    1096 Cor. 1 . — The sum of the squares of the intercepts on the radii be
    tween the perpendiculars and the centre is also equal to înr . (1. 47 )
    1097 Cor . 2. — The sum of the squares of the perpendiculars from the
    cquidistant points on the circle to any right line passing through the centre
    is also equal to lnr .
    Because the perpendiculars from two points on a circle to the diameters
    drawn through the points are equal.
    1098 Cor. 3. — The sum of the squares of the intercepts on the same
    right line between the centre of the circle and the perpendiculars is also
    equal to įnra. (I. 47 )
    If the radii of the inscribed and circumscribed circles of a
    regular polygon of n sides be r, R , and the centre 0 ; then ,
    1099 I. The sum of the perpendiculars from any point P upon the sides
    is equal to nr.
    1100 II. If p be the perpendicnlar from O upon any right line , the sum
    of the perpendiculars from the vertices apon the same line is equal to np.
    1101 III. The sum of the squares of the perpendiculars from P on the
    sides is = n (72 + OP ).
    1102 IV . The sum of the squares of the perpendiculars from the vertices
    upon the right line is = n ( p’ + R *).
    Proof. — In theorem I., the values of the perpendiculars are given by
    g - OP cos (0 + 20 + ), with successive integers for m . Add together the n
    values, and apply ( 803).
    Similarly, to prove II. ; take for the perpendiculars the values
    p - R cos(0+ 2mm).
    To prove III. and IV ., take the same expressions for the perpendiculars ;
    square each value ; add the results, and apply (803, 804).
    For additional propositions in the subjects of this section ,
    see the section entitled Plane Coordinate Geometry .
    GEOMETRICAL CONICS.

    THE SECTIONS OF THE CONE.

    1150 DEFINITIONS.- A Conic Section or Conic is the curve


    AP in which any plane intersects the surface of a right cone.
    A right cone is the solid generated by the revolution of
    one straight line about another which it intersects in a fixed
    point at a constant angle.
    Let the axis of the cone, in Fig . ( 1 ) or Fig . (2 ), be in the plane of the
    paper, and let the cutting plane PMXN be perpendicnlar to the paper . ( Read
    either the accented or unaccented letters throughout.) Let a sphere be inscribed
    in the cone, touching it in the circle EQF and touching the cutting plane in
    the point S , and let the cutting plane and the plane of the circle EQF inter
    sect in XM . The following theorem may be regarded as the defining property
    of the curve of section.
    1151 Theorem . — The distance of any point P on the conic
    from the point S , called the focus, is in a constant ratio to
    PM , its distance from the line XM , called the directrix , or
    PS : PM = PS : PM ' = AS : AX = e , the eccentricity .
    [ See next page for the Proof.]
    1152 COR. — The conic may be generated in a plane from
    either focus S , S', and either directrix XM , X ' M ', by the law
    just proved.
    1153 The conic is an Ellipse, a Parabola , or an Hyperbola ,
    according as e is less than , equal to , or greater than unity .
    That is , according as the cutting plane emerges on both sides
    of the lower cone, or is parallel to a side of the cone, or in
    tersects both the upper and lower cones.
    1154 All sections made by parallel planes are similar ; for
    the inclination of the cutting plane determines the ratio
    AE : AX .
    1155 The limiting forms of the curve are respectively — a
    circle when e vanishes, and two coincident right lines when e
    becomes infinite.
    2 A
    234 GEOMETRICAL CONICS.

    Proof OF THEOREM 1151. — Join P, S and P, 0 , cutting the circular section


    in Q , and draw PM parallel to NX . Because all tangents from the same
    point, O or A , to either sphere are equal, therefore RE = PQ = PS and
    AE = AS. Now , by (VI. 2 ), RE : NX = AE : AX and NX = PM ;
    therefore PS : PM = AS : AX, a constant ratio denoted by e and called
    the eccentricity of the conic .

    Referring the letters either to the ellipse or the hyperbola in the


    subjoined figure , let C be the middle point of AA' and N any other point on
    it . Let DD , RR ' be the two circular sections of the cone whose planes pass
    through C and N ; BCB and PN the intersections with the plane of the
    conic. In the ellipse, BC is the common ordinate of the ellipse and circle ;
    but, in the hyperbola , LC is to be taken equal to the tangent from C to the
    circle DD'.
    1156 The fundamental equation of the ellipse or hyperbola
    is PN ? : AN .NA' = BC : ACP.
    PROOF. — PN ? = NR . NR and BC = CD .CD (III. 35, 36 ). Also, by
    similar triangles (VI. 2 , 6 ), NR : CD = AN : AC and NR : CD' = A 'N : A ' C .
    Multiply the last equations together.
    THE ELLIPSE AND HYPERBOLA. 235

    1157 Cor. 1 . - PN has


    equal values at two points
    equi- distant from AA .
    Hence the curve is sym
    metrical with respect to
    AA and BB .
    These two lines are
    called the major and mi
    nor axes, otherwise the
    transverse and conjugate
    axes of the conic .
    When the axes are
    equal, or BC = AC, the
    ellipse becomes a circle ,
    and the hyperbola be
    comes rectangular or A
    equilateral.

    1158 Any ellipse or hyperbola is the orthogonal projection


    of a circle or rectangular hyperbola respectively .
    Proof. — Along the ordinate NP, measure NP = AN . NA' ; therefore by
    the theorem PN : P ' N = BC : A0. Therefore a circle or rectangular hyper
    bola, having AA' for one axis , and having its plane inclined to that of the
    conic at an angle whose cosine = BC : AC, projects orthogonally into the
    ellipse or hyperbola in question , hy (1089). See Note to (1201).

    1159 Hence any projective property (1076 – 78), which is


    known to belong to the circle or rectangular hyperbola , will
    also be universally true for the ellipse and hyperbola respec
    tively.

    THE ELLIPSE AND HYPERBOLA.

    Joint properties of the Ellipse and Hyperbola .


    1160 DEFINITIONS. — The tangent to a curve at a point P
    (Fig . 1166 ) is the right line PQ, drawn through an adjacent
    point Q , in its ultimate position when Q is made to coincide
    with P .
    The normal is the perpendicular to the tangent through
    the point of contact.
    236 GEOMETRICAL CONICS.

    In (Fig . 1171) , referred to rectangular axes through the


    centre C ( see Coordinate Geometry ) ; the length CN is called
    the abscissa ; PN the ordinate ; PT the tangent ; PG the nor
    mal ; NT the subtangent ; and NG the subnormal. S , S are
    the foci ; XM , X ' M ' the directrices ; PS, PS the focal dis
    tances, and a double ordinate through S the Latus Rectum .
    The auxiliary circle ( Fig . 1173) is described upon AA' as
    diameter.
    A diameter parallel to the tangent at the extremity of
    another diameter is termed a conjugate diameter with respect
    to the other .
    The conjugate hyperbola has BC for its major, and AC for
    its minor axis (1157).
    1161 The following theorems (1162) to (1181) are deduced from the
    property PS : PM = e obtained in ( 1151) .
    The propositions and demonstrations are nearly identical for the ellipse
    and the hyperbola, any difference in the application being specified .

    1162 CS : CA : CX, and the common ratio is e .

    I ÄR LA

    AS =– AX
    Proof.— By(1151), e = AX A’S =– }(A’SEAS)
    (A XH č oror cu
    AX) == CS C
    1163 In the ellipse the sum , and in the hyperbola the dif
    ference, of the focal distances of P is equal to the major axis ,
    or PS + PS = AA .
    Proof.- With the same figures wehave, in the ellipse, by (1151),
    PS + PS PS + PS AS + A'S = AA', therefore & c.
    = PM + PM ' = XX' , and also e = AX + A ' X = XX" there.
    For the hyperbola take difference instead of sum .
    THE ELLIPSE AND HYPERBOLA . 237

    1164 CS = AC - BC² in the ellipse.


    [For BS = AC, by (1163) .
    CS = ACP + BC% in the hyperbola .
    [ By assuming BC . See (1176 ).
    1165 BCP = SL . AC.
    Proof. — (Figs. of 1162) SL : SX = CS : CA, (1151, 1162)
    . : SL . AC = CS . SX = CS (CX - CS) = CAP- CS (1162) = BC (1164).
    1166 If a right line through P, Q , two points on the conic ,
    meets the directrix in Z , then SZ bisects the angle QSR .

    PROOF. - By similar triangles, ZP : ZQ = MP : NQ = SP : SQ 151),


    therefore by (VI. A .)
    1167 If PZ be a tangent at P , then PSZ and PSZ' are
    right angles.
    PROOF. — Make Q coincide with P in the last theorem .
    1168 The tangentmakes equal angles with the focal dis
    tances .
    Proof. - In (1166 ), PS : PS = PM : PM ' (1151) = PZ : PZ ; therefore,
    when PQ becomes the tangent at P, ZSPZ = SPZ', by (1167) and (VI. 7 ).
    1169 The tangents at the extremities of a focal chord inter
    sect in the directrix .
    Proof. — (Figs.of 1166). Join ZR ; then, if ZP is a tangent, ZR is also ,
    for (1167) proves RSZ to be a right angle .
    1170 CN . CT = AC? .

    Proor.—(Figs.1171.) TS = PS (VI.3,4.)= PM"(1151) = NX


    therefore TS” + TS _ NX' + NX 2CT _ 2CX
    TS - TS = NX -NÝ Or 208 20N
    therefore CN . CT = CS . CX = AC?. (1162)
    - - -
    238 GEOMETRICAL CONICS.

    1191 If PG be the normal,


    GS : PS = GS : PS = e .

    G NS X T TXA SN

    ProOF. - By (1168) and (VI. 3, A .),


    GS – GS - GS' + GS _ 2CS ..
    PS = PS* = PS PS = 2CA == e.e. (1162)
    But, for the hyperbola, change plus to minus.
    1172 The subnormal and the abscissa are as the squares of
    the axes, or NG : NC = BC : AC .
    Proof. — (Figs.1171.) Exactly as in (1170), taking the normal instead
    ofthe tangent,we obtain CS = ex : ca = CS = CSE (1162),
    .. CNOCG = CAZOS, OF NG = RC (1164).
    1173 The tangents at P and Q , the corresponding points on
    the ellipse and auxiliary circle , meet the axis in the samepoint
    T . But in the hyperbola , the ordinate TQ of the circle being
    drawn, the tangent at Q cuts the axis in N .

    Proof. – For the ellipse : Join TQ . Then CN . CT = CQ? (1170 ) ; tbere


    fore CQT is a right angle (VI. 8 ) ; therefore QT is a tangent.
    For the hyperbola : Interchange N and T .
    THE ELLIPSE AND HYPERBOLA. 239

    1174 PN : QN = BC : AC.
    Proof.- - ( Figs. 1173 ). NG . NT = PN ?, and CN . NT = QN . (VI. 8 )
    Therefore NG : NC = PN ? : QN ?; therefore, by (1172).
    This proposition is equivalent to (1158 ), and shows that an ellipse is the
    orthogonal projection of a circle equal to the auxiliary circle.
    1175 Cor. — The area of the ellipse is to that of the auxiliary
    circle as BC : AC (1089) .
    1176 PN : AN . NA = BC ? : AC .
    PROOF. — By (1174 ), since QN = AN . NA (III. 35, 36 ) . An independ
    ent proof of this theorem is given in (1156 ) . The construction for BC in
    the hyperbola in (1164 ) is thus verified.
    1177 Cn . Ct = BC .
    ProvF. — (Figs. 1173.)
    PN Cn . Ct PN ? PN PN
    CT = NT : :: CN . CT = CN .NT = ON : (VI. 8) = ĂN . NA (III. 35, 36).
    Therefore,by (1170) and (1176 ), Cn .Ct: AC? = BC? : AC%.
    1178 If SY, SY' are the perpendiculars on the tangent,
    then Y , Y ' are points on the auxiliary circle , and
    -

    SY . S ' Y' = BC ?.
    - - - -

    Proof. — Let PS meet SY in W . Then PS = PW (1168). Therefore


    SW = AA' (1163) . Also , SY = YW , and SC = CS . Therefore CY = iSW
    = AC. Similarly CY' = AC. Therefore Y , Y ' are on the circle.
    Hence ZY' is a diameter (III. 31) , and therefore SZ = S ' Y ', by similar
    triangles ; therefore SY . SZ = SA SA (III. 35, 36 ) = CS” ~ CÀ (II. 5 )
    = BC? (1164).
    1179 Cor. If CE be drawn parallel to the tangent at P ,
    then PE = CY = AC.
    1180 PROBLEM. — To draw tangents from any point 0 to an
    ellipse or hyperbola .
    CONSTRUCTION. — (Figs. 1181.) Describe two circles ,one with centre ) and
    radius OS, and another with centre S and radius = AA', intersecting in M ,
    240 GEOMETRICAL CONICS.

    M . Join MS , M ' S . These lines will intersect the curve in P, P , the


    points of contact. For another method see (1204) .
    PROOF. - By ( 1163) , PS + PS = AA' = SM by construction . There
    fore PS = PM , therefore ZOPS = OPM (1. 8 ) , therefore OP is a tangent
    by (1168) .
    Similarly PS = PM', and OP is a tangent.
    1181 The tangents OP, OP subtend equal angles at either
    focus.

    Proof. — The angles OSP, OSP are respectively equal to OMP, OM ' P ,
    by ( I. 8 ), as above ; and these last angles are equal, by the triangles OSM ,
    OSM ', and ( I. 8 ). Similarly at the other focus.

    Asymptotic Properties of the Hyperbola .


    1182 DEF. — The asymptotes of the hyperbola are the
    diagonals of the rectangle formed by tangents at the vertices
    A , A ', B , B '.
    1183 If the ordinates RN , RM from any point R on an
    asymptote cut the hyperbola and its conjugate in P , P , D , D',
    THE HYPERBOLA. 241

    then either of the following pairs of equations will defineboth


    the branches of each curve
    RN? - PN ? = BC2 = P ' N ? – RN ... . ..
    RM² - DM = AC? = D'MⓇ- RM ............(2).
    PROOF. — Firstly, to prove (1 ) : By proportion from the similar triangles
    RN ? BC2 PN ?
    RNC, OAC, we have CN ? – AC2 - ON ? - ACP
    by (1176 ), since AN . NA = CN : - AC?. By (II. 6)
    Therefore RN ?AC
    — PN ? =_ AC?
    BC? by the theorem (69) ;
    therefore RNP - PNP = BC%.
    Also,by (1176 ), applied to the conjugate hyperbola, the axes being now
    CN - AC – CNS by similar triangles ;
    reversed , PNP - BC2 = BCP = RN ” wy
    therefore P 'N ? – BOʻ = RN? or P'N ? — RN ? = BC?.
    Secondly , to prove (2 ) : By proportion from the triangles RMC, OBC, we
    RM AC DU ?
    have
    CMI = BC - CM - BC ?
    by (1176), applied to the conjugate hyperbola, for in this case we should
    have BM .MB = CM - BC?.
    for RM ?DC12
    – DM = AC3
    262 ; therefore RM - DII = AO ".
    Therefore
    Also, by (1176), since CM , D '][ are equal to the coordinates of D',
    CM BC2 CM
    D' M ” – AC AC? - RMMan by similar triangles;
    therefore D'M ' - ACP = RM or D' M - RM = AC?.
    1184 Cor. 1. - If the same ordinates RN , RM meet the
    other asymptote in r and ro', then
    PR . Pr = BC and DR . Dr' = AC?. ( II. 5)
    1185 Cor . 2 . - As Rrecedes from C , PR and DR con
    tinually diminish . Hence the curves continually approach
    the asymptote .
    1186 If XE be the directrix , CE = AC.
    Proof. - CE : C0 = CX : CA = CA : CS and CS = CO. (1164)
    1187 PD is parallel to the asymptote .
    RN BC: RNP - PN ? PN3
    FROOF.- RM * = ACU = RM - DM Du (69 ).
    Therefore RN : PN = RM : DM ; therefore,by (VI. 2).
    2 1
    242 GEOMETRICAL CONICS.

    1188 The segments of any right line between the curve and
    the asymptote are equal, or QR = qr.

    Proof. QR : QU = qR : qU ' ) Compound the ratios ,


    and Qr : Qu = qr : qui s ' and employ (1184 ).
    1189 Cor. 1. - PL = Pl and QV = qV.
    1190 Cor. 2. - CH = HL . Because PD is parallel to IC .
    (1187)
    1191 QR . Qr = PL = RV2 - QV? = QV2 - RV?.
    Proof. — QR : QU = PL : PE Compound theratios. Therefore,by( 1184),
    and Qr : Qu = Pl : Pe J QR . Qr = PL . Pl = PL (1189 ) .

    1192 4PH . PK = CS?.


    Proof. - PH : PE = CO : 002 :. PH. PK : PE . Pe = CO’ : 002
    and PK : Pe = Co : 00 5 = CS? : 4BC?; therefore, by (1184).

    Joint Properties of the Ellipse and Hyperbola resumed .


    If PCP' be a diameter, and QV an ordinate parallel to the
    conjugate diameter CD (Figs. 1195 and 1188 ) .
    1193 QV2 : PV . VP = CD2 : CP2.
    This is the fundamental equation of the conic, equation (1176) being the
    most important form of it .
    THE ELLIPSE AND HYPERBOLA . 243
    Otherwise :
    In the ellipse, QV2: CP - CV ? = CDP: CP2
    In the hyperbola , QV2: CV - CP = CD’ : CP ;
    and QʻV?: CV2+ CP = CDP: CP2.
    Proof. — (Ellipse. Fig . 1195 .) — By orthogonal projection from a circle.
    If C, P , P , D , Q , V are the projections of c, p , p , d , q, v on the circle ;
    qu = pv . vp ' and cd = cps. The proportion is therefore true in the case of
    The circle. Therefore & c.,by (1088).
    (Hyperbola . Fig . 1188 ) —
    CDRV PURV + PL _ QV Q V
    CP: = cy2 = CP2 = CVI + CPI = CV _ CP or cy? + CP2 (1191)

    1194 The parallelogram formed by tangents at the extremi


    ties of conjugate diameters is of constant area , and therefore,
    PF being perpendicular to CD (Figs. 1195 ),
    PF . CD = AC . BC .
    Proof. — ( Ellipse.) — By orthogonal projection from the circle ( 1089) .
    (Hyperbola . Fig . 1188.) — CL . Cl = 4PH . PK = Co . Co (1192) ; there
    fore, by (VI. 15), ALCI = OC0 = AC . BC.

    If PF intersects the axes in G and G ”,


    1195 PF . PG = BC and PF . PG ' = AC ?
    17

    Proof.- PF. PG = PK . PN = Cn .Ct = BC (1177). Similarly for


    PF . PG .
    1197 COR. - PG . PG ' = CD ' = PT . Pt. By (1194 )
    244 GEOMETRICAL CONICS.

    1198 The diameter bisectsall chords parallel to the tangent


    at its extremity.
    Proof. — (Ellipse. Fig . 1195.) — By projection from the circle ( 1088 )
    QV = VQ . (Hyperbola .) By (1189.)
    1199 Cor . 1 . – The tangents at the extremities o f any chord
    meet on the diameter which bisects it.
    PROOF. — The secants drawn through the extremities of two parallel chords
    meet on the diameter which bisects them (VI. 4 ), and the tangents are the
    limiting positions of the secants when the parallel chords coincide.
    1200 Cor. 2. - If the tangents from a point are equal, the
    diameter through the point must be a principal axis. (I. 8 )
    1201 Cor . 3 .— The chords joining any point Q on the curve
    with the extremities of a diameter PP', are parallel to con
    jugate diameters , and are called supplemental chords.
    For the diameter bisecting PQ is parallel to P ' Q (VI. 2 ). Similarly the
    diameter bisecting P ' Q is parallel to PQ .

    1202 Diameters are mutually conjugate ; If CD be parallel


    to the tangent at P , CP will be parallel to the tangent at D .
    • Proof. — (Ellipse. Fig. 1205.) — By projection from the circle ( 1088).
    NOTE. — Observe that, if the ellipse in the figure with its ordinates and
    tangents be turned about the axis it through the angle cos-' (BC = AC ), it
    becomes the projection of the auxiliary circle with its corresponding ordinates
    and tangents.
    ( Hyperbola . Fig . 1188.) — By (1137, 1189) the tangents at P , D meet the
    asymptotes in the same point L . Therefore they are parallel to CD , CP (VI. 2 .)

    If QT be the tangent at Q , and QV the ordinate parallel


    to the tangent at any other point P ,
    1203 CV .CT = CP

    PROOF. — CR bisects PQ (1199). Therefore PW is parallel to QR .


    Therefore, by. (VI. 2), OV : CP = CW : CR = CP : CT.
    THE ELLIPSE AND HYPERBOLA. 245

    1204 COR. — Hence, to draw two tangents from a point T, we may find
    CV from the above equation, and draw QVQ parallel to the tangent at P to
    determine the points of contact Q , Q .

    Let PN , DN be the ordinates at the extremities of con


    jugate diameters, and PT the tangent at P . Let the ordinates
    at N and R in the ellipse , but at T and C in the hyperbola ,
    meet the auxiliary circle in p and d ; then
    1205 CN = dR , CR = pN .
    1. de

    TAN
    PROOF. — ( Ellipse.) Cp, Cd are parallel to the tangents at d and p (Note to
    1202). Therefore pod is a right angle . Therefore pNC, CRd are equal
    right-angled triangles with CN = dR and CR = pN .
    (Hyperbola.) CN . CT = AC? ( 1170 ),
    and DR .CT = 2ACDT = 2CDP = AC .BC (1194);
    :. ON = 10= PN (1174) - N = PI = CF (similar triangles).
    But
    bu PN ) ; : : CR = pN . Also Cp = Cd ; therefore the tri
    angles CpN , DCR are equal and similar; therefore CN = dR and dR is
    parallel to pN .
    1206 CoR.- DR : dR = BC : AC.
    Proof. — (Ellipse.) By (1174 ). (Hyperbola .) By the similar right-angled
    triangles, we have dR : pN = CR : TN = DR : PN ;
    therefore dR : DR = pN : PN = AC : BC ( 1174 ) .

    In the same figures,


    1207 (Ellipse.) CNP + CR² = AC ?; DR + PNP = BC?.
    1209 (Hyperbola .) CN — CRⓇ = AC? ; DR — PNP = BC?.
    Proof. – Firstly, from the right-angled triangle CNp in which pN = CR
    (1205) .
    Secondly , In the ellipse, by (1174), DR + PN ? : dRº + pN ’ = BC ? : AC?,
    and dR? + pN = AC , by (1205) . For the byperbola, take difference of squares.
    246 GEOMETRICAL CONICS.

    1211 (Ellipse.) CP2 + CD2 = AC + BC %.


    1212 (Hyperbola .) CP2 - CD = AC ’ – BC ?.
    Proof. — (Figs. 1205.) By (1205 — 1210 ) and (I. 47), applied to the
    triangles CNP, ČRD.

    The product of the focal distances is equal to the square


    of the semi-conjugate diameter , or
    1213 PS . PS = CD '.
    PROOF. — ( Ellipse. Fig . 1171.) 2PS . PS' = ( PS + PS') ? – PS? – PS"
    = 4AC2 — 2CŠ — CP2 (922, i.) = 2 (AC? + BC2- C ) (1164)' = 2CDº(1211).
    ( Hyperbola .) — Similarly with 2PS . PS = PS? + PS” — ( PS" — PS) = & c.
    1214 The products of the segments of intersecting chords
    Q0q, Q 'Oq' are in the ratio of the squares of the diameters
    parallel to them , or
    02 .09 : 0Q'.Oq' = CD' : CD'?.
    Proof. — (Ellipse.) By projection from the circle (1088) ; for the propor
    tion is true for the circle, by (III. 35, 36 ).
    (Hyperbola . Fig . 1188.) Let O be any point on Qq. Draw I0i parallel
    to Ee, meeting the asymptotes in I and i ; then
    OR . Or - 0Q . Oq = QR . Qr (II. 5) = PL ( 1191) ......... (1).
    Now OR _ PL na Or _ Pl. . OR . Or _ PL - CD (1104
    " OI PE '14 Oi - Pei : 01. 0i = PE . Pe = BC2 (118 ).
    OR . 0i
    Therefore 01. PĽ2 _=
    Or -- BC CDa ; or, by (1), _ OQ . Oq - CD
    Ber
    01. 0i – BC2 - BC**
    Similarly for any other chord Q 'Od drawn through 0 .
    Therefore Q . Oq : OQ . O4 = CD : CD ?.
    1215 COR . — The tangents from any point to the curve are
    in the ratio of the diameters parallel to them .
    For, when ( is without the curve and the chords become tangents , each
    product of segments becomes the square of a tangent.

    1216 If from any point Q on a tangent PT drawn to any


    conic (Fig . 1220 ), two perpendiculars QR , QL be drawn to the
    focal distance PS and the directrix XM respectively ; then
    SR : QL = e.
    Proov. - Since QR is parallel to ZS (1167) , therefore, by (VI. 2) ,
    SR : PS = QZ : P2 = QL : PM ;
    therefore SR : QL = PS : PM = e .
    Cor . — By applying the theorem to each of the tangents from Q , a proof
    of (1181) is obtained .
    THE ELLIPSE AND HYPERBOLA . 247

    1217 The Director Circle. — The locus of the point of inter


    section , T , of two tangents always at right angles is a circle
    called the Director Circle .
    --
    --
    -

    VS
    I-

    PROOF. — Perpendiculars from S , S' to the tangentsmeet them in points


    Y, Z , Y', Z ', which lie on the auxiliary circle . Therefore, by (II. 5 , 6 ) and
    (III. 35, 36 ), TC* „ AC = TZ . TZ' = SY. SY' = BC?. (1178)
    Therefore TC? = AC? + BC , a constant value.

    NOTE. — Theorems (1170 ), (1177 ) , and (1203 ) may also be deduced


    at once for the ellipse by orthogonal projection from the circle ; and, in all
    such cases, the analogons property of the hyperbola may be obtained by a
    similar projection from the rectangular hyperbola if the property has already
    been demonstrated for the latter curve.

    1218 If the points A , S ( Fig . 1162) be fixed , while o is


    moved to an infinite distance, the conic becomes a parabola .
    Hence, any relation which has been established for parts of
    the curve which remain finite, when AC thus becomes infinite ,
    willbe a property of the parabola .
    1219 Theorems relating to the ellipse may generally be
    adapted to the parabola by eliminating the quantities which
    become infinite, employing the principle that finite differences
    may be neglected in considering the ratios of infinite quantities .
    EXAMPLE. — ID (1193), when P is at infinity, VP becomes = 2CP ; and
    in (1213) PS' becomes = 2CP. Thus the equations become
    * = 2CP* and PS = COP
    Therefore QV2 = 4PS . PV in the parabola .
    248 GEOMETRICAL CONICS.

    THE PARABOLA .

    If S be the focus, XM
    the directrix , and P any point
    on the curve, the defining pro
    perty is

    1220 PS = PM
    and e = 1. (1153) Í xu s
    1221 Hence
    MA
    AX = AS.

    1222 The Latus Rectum = 4AS.


    PROOF . SL = SX (1220 ) = 2AS.

    1223 If PZ be a tangent at P , meeting the directrix in Z ,


    then PSZ is a right angle .
    Proof. - As in (1167) ; theorem (1166 ) applying equally to the parabola .

    1224 The tangent at P bisects the angles SPM , SZM .


    Proof.- PZ is common to the triangles PSZ, PMZ; PS = PM and
    ZPSZ = PMZ (1223).
    1225 COR , ST = SP = SG . (I. 29, 6 )

    1226 The tangents at the extremities of a focal chord PQ


    intersect at right angles in the directrix .
    Proof. — (i.) They intersect in the directrix, as in (1169 ).
    (ii.) They bisect the angles SZM , SZM ' (1224 ), and therefore include a
    right angle.

    1227 The curve bisects the sub - tangent. AN = AT.


    Proof . — ST = SP (1225) = PM = XN , and AX = AS.

    1228 The sub -normal is half the latus rectum . NG = 2AS.


    PROOF. - ST = SP = SG and TX = SN (1227). Subtract.
    THE PARABOLA . 249

    1229 PNP = 4AS . AN .


    Proof.— PN = TN . NG (VI. 8) = AN . 2NG (1227) = 4AS. AN ( 1228 ).
    Otherwise, by (1176 ) and (1165) ; making AC infinite. See (1219).

    1230 The tangents at A and P each bisect SM , the latter


    bisecting it at right angles .
    Proof. — (i.) The tangent at A , by (VI. 2 ), since AX = AS.
    (ii.) PT bisects SM at right angles , by ( I. 4), since PS = PM and
    ZSPY = MPY.
    1231 COR. SA : SY : SP . [ By similar triangles.

    1232 To draw tangents from a point 0 to the parabola .


    ConstrucTION . - Describe a circle, centre - - - İM
    O and radius Os, cutting the directrix in
    M , M '. Draw MQ, M ' Q ' parallel to theaxis,
    meeting the parabola in Q , Q . Then OQ ,
    oll will be tangents .
    Proof. — OS, SQ = OM , MQ (1220); 0
    therefore ,by ( I. 8 ), ZOQS = OQM ; there
    fore OQ is a tangent (1224 ) . Similarly
    OQ' is a tangent.
    Otherwise, by (1181). When & moves
    to infinity , the circle MM' becomes the -- -
    directrix .

    1233 CoR. 1. — The triangles SQO , SOQ' are similar, and


    SQ : 50 : SQ '.
    Proof. - < SQO = MQO = SMM ' = SOQ'. (III. 20)
    Similarly SQʻO = SOQ.
    1234 Cor . 2 .- The tangents at two points subtend equal
    angles at the focus ; and they contain an angle equal to half
    the exterior angle between the focal distances of the points .
    Proof. LOSQ = 0SQ , by (Cor. 1) .
    Also ZQOQ = SOQ + SQ0 = 1 - OSQ = {QSQ .

    1235 DEF. — Any line parallel to the axis of a parabola is


    called a diameter .
    1236 The chord of contact QQ' of tangents from any point
    O is bisected by the diameter through 0 .
    PROOF. This proposition and the corollaries are included in (1198 – 1200 ),
    by the principle in (1218) . An independent proof is as follows.
    2 K
    250 GEOMETRICAL CONICS.

    The construction being as in (1232) ,


    we have ZM = ZM '; therefore QV = VQ
    (VI. 2).
    1237 Cor . 1. - The tangent
    RR' at P is parallel to QQ ' ; and
    OP = PV.
    PROOF. — Draw the diameter RW .
    QW = WP; therefore QR = RO (VI. 2) .
    Similarly QʻR ' = R O.
    1238 Cor . 2 . - Hence , the dia
    meter through P bisects all chords parallel to the tangent at P .

    If QV be a semi- chord parallel to the tangent at P,


    1239 QV2 = 4PS . PV.
    This is the fundamental equation of
    the parabola , equation (1229) being the
    most important form of it.
    Proof. — Let QO meet the axis in T .
    By similar triangles (1231),
    SRP = SQR = STQ = POR ;
    and ZSPR = OPR (1224 ) . Therefore TL
    PRʻ = PS. P0 = PS.PV and QV = 2PR.
    Otherwise : See (1219) , where the
    equation is deduced from ( 1193) of the
    ellipse.
    1240 Cor . 1. - If v be any other point, either within or
    without the curve,on the chord QQ', and pv the corresponding
    diameter, vQ .vQ ' = 4ps . pv. (II. 5)
    1241 Cor. 2 . — The focal chord parallel to the diameter
    through P , and called the parameter of that diameter, is equal
    to. 4SP. For PV in this case is equal to PS.

    1242 The products of the segments of intersecting chords,


    Q0q, Q 'Oq', are in the ratio of the parameters of the diameters
    which bisect the chords ; or
    02 . 09 : 0Q'.Oq' = PS : P ' S .
    Proof. — By (1240), the ratio is equal to 4P8 .20 : 4PS. po.
    CD =_ 78
    Otherwise : In the ellipse ( 1214), theratio is = con PS ((1213)
    PS . PS
    PS
    So when S' is at infinity and the curve becomes a parabola (1219).
    ON CONSTRUCTING THE CONIO . 251

    1243 CoR . — The squares of the tangents to a parabola from


    any point are as the focal distances of the points of contact.
    Proof.— As in (1215). Otherwise, by ( 1233) and (VI. 19).

    1244 The area of the parabola cut off by any chord QQ is


    two-thirds of the circumscribed parallelogram , or of the tri
    angle formed by the chord and the tangents at Q , Q '.
    PROOF.— Through Q. 9, 4 , & c., adjacent points
    on the curve, draw right lines parallel to the
    diameter and tangent at P . Let the secant Qa
    cut the diameter in 0 . Then , when q coincides
    with Q , so that Qq becomes a tangent, we have
    OP = PV (1237). Therefore the parallelogram
    Vq = 2Uq, by (I.43), applied to the parallelogram
    of wbich oQ is the diagonal. Similarly vq = 2uq',
    &c. Therefore the sum of all the evanescent par
    allelograms on one side of PQ is equal to twice
    the corresponding sum on the other side ; and
    these sums are respectively equal to the areas
    PQV, PQU. - (NEWTON, Sect. I., Lem . II.)

    Practical methods of constructing the Conic .


    1245 To draw the Ellipse ,
    Fix two pins at S, 8 (Fig. 1162). Place over them a loop of thread
    baving a perimeter SPS = S8 + AA . A pencil point moved so as to keep
    the thread stretched will describe the ellipse, by (1163).
    1246 Otherwise. - ( Fig . 1173.) Draw PHK parallel to QC, cutting the
    axes in H , K . PK = Að and PH = BO (1174 ). Hence, if a ruler PHK
    moves so that the points H , K slide along the axes, P will describe the ellipse.

    1247 To draw the Hyperbola .


    Make the pin S (Fig . 1162) serve as a pivot for one end of a bar of any
    convenient length. To the free end of the bar attach one end of a thread
    whose length is less than that of the bar by AA'; and fasten the other end of
    the thread to the pin S. A pencil point moved so as to keep the thread
    stretched , and touching the bar, will describe the hyperbola, by (1163).
    1248 Otherwise : - Lay off any scale of equal parts along both asymptotes
    (Fig. 1188 ), starting and numbering the divisions from C, in both positive
    and negative directions.
    Join every pair of points L , I, the product of whose distances from C is
    the same, and a series of tangents will be formed (1192 ) which will define
    the hyperbola . See also (1289).
    252 GEOMETRICAL CONICS.

    1249 To draw the Parabola .


    Proceed as in (1247), with this difference : let the end of the bar, before
    attached to S , terminate in a “ T -square,” and be made to slide along the
    directrix (Fig . 1220), taking the string and bar of the same length .
    1250 Otherwise : - Make the same construction as in (1248 ), and join
    every pair of points, the algebraic sum of whose distances from the zero
    point of division is the same.
    PROOF.-- If the two equal tangents from any point T on the axis (Fig .
    1239) be cut by a third tangent in the points R , r ; then RQ may be proved
    equal to rT, by (1233), proving the triangles SRQ, SrT equal in all respects.
    1251 Cor. — The triangle SRr is always similar to the isosceles triangle
    SQT.

    1252 To find the axes and centre of a given central conic.


    (i.) Draw a right line through the centres of two parallel chords. This
    line is a diameter, by (1198) ; and two diameters so found will intersect in
    the centre of the conic.
    (i .) Describe a circle having for its diameter any diameter PP of the
    conic, and let the circle cat the curre in Q . Then PQ , P Q are parallel to
    the axes, by (1201) and (III. 31).

    1253 Given two conjugate diameters, CP, CD , in position


    and magnitude : to construct the conic.
    On CP take PZ = CDP - CP ; measuring
    from C in the ellipse, and towards C in the
    hyperbola (Fig . 1188 ). A circle described
    through the points C , Z , and having its
    centre 0 on the tangent at P , will cut the
    tangent in the points where it is intersected
    by the axes.
    PROOF. - Analysis : Let AC, BC cut the
    tangent at P in T , t. The circle whose Z
    diameter is Tt will pass through C ( III. 31),
    and will make
    CP .PZ = PT. Pt(III.35, 36) = CDº(1197).
    Hence the construction .

    Circle and Radius of Curvature.


    1254 DEFINITIONS. — The circle which has the same tangent
    with a curve at P (Fig . 1259), and which passes through
    another point Q on the curve, becomes the circle of curvature
    when Q ultimately coincides with P ; and its radius becomes
    the radius of curvature.
    CIRCLE OF CURVATURE. 253

    1255 Otherwise. - The circle of curvature is the circle which


    passes through three coincident points on the curve at P .
    1256 Any chord PH of the circle of curvature is called a
    chord of curvature at P .
    1257 Through Q draw RQ' parallel to PH , meeting the
    tangent at P in R , and the circle in Q ', and draw QV parallel
    to PR. RQ is called a subtense of the arc PQ .

    1258 THEOREM .— Any chord of curvature PH is equal to


    the ultimate value of the square of the arc PQ divided by the
    subtense RQ parallel to the chord : and this is also equal to
    QV - PV.
    PROOF. — RQ = RPP : RQ (III. 36 ) . And when Q moves up to P , RR
    becomes PH ; and RP, PQ, and QV become equal because coincident lines.

    1259 In the ellipse or hyperbola , the semi- chords of cur


    vature at P , measured along the diameter PC, the normal
    PF , and the focal distance PS, are respectively equal to
    CD CD ? CD2
    CP PF , AC :
    the second being the radius of curvature at P .

    Proof. — (i.) By (1258), PI = 273 = VPCD


    CP (1193) = 2CD in the CP

    limit when VP becomes PP = 2CP.


    (ii.) By the similar triangles PHU, PFC (III. 31), we have
    PU . PF = CP . PH = 2CD", by (i.)
    (ii.) By the similar triangles PIU, PFE (1168) , we have
    PI. PE = PU . PF = 2CD', by (ii.) ; and PE = AC ( 1179).
    254 GEOMETRICAL CONICS.

    1260 In the parabola , the chord of curvature at P (Fig .


    1259) drawn parallel to the axis , and the one drawn through
    the focus, are each equal to 4SP, the parameter of the dia
    meter at P (1241).
    Proof.- By (1258 ). The chord parallel to the axis = QT : PV = 4PS
    (1239) ; and the two chords are equal because they make equal angles with
    the diameter of the circle of curvature.
    1261 Cor. — The radius of curvature of the parabola at P.
    (Fig. 1220) is equal to 28P = SY.
    Proof. — (Fig. 1259.) JPU = {PIsec IPU = 2SP sec PSY (Fig . 1221).
    1262 The products of the segments of intersecting chords
    are as the squares of the tangents parallel to them (1214 - 15 ) ,
    ( 1242 -43 ) .
    1263 The common chords of a circle and conic (Fig. 1264 )
    are equally inclined to the axis ; and conversely, if two chords
    of a conic are equally inclined to the axis, their extremities
    are concyclic.
    Proof. — The products of the segments of the chords being equal ( III.
    35, 36 ), the tangents parallel to them are equal (1262). Therefore, by (1200).

    1264 The common chord of any conic and of the circle of


    curvature at a point P , has the
    same inclination to the axis as
    the tangent at P . TO
    PROOF. - Draw any chord Qq parallel
    to the tangent at P . The circle circum
    scribing PQq always passes through the
    same point p (1263) , and does so, there
    fore, when Qq moves up to P , and the
    circle becomes the circle of curvature.

    1265 PROBLEM .-- To find the centre of curvature at any given


    point of a conic .
    • First Method . — (Fig . 1264.) Draw a chord from the point making the
    same angle with the axis as the tangent. The perpendicular bisector of the
    chord will meet the normal in the centre of curvature, by (1264 ) and ( III. 3 ).
    1266 Second Method . - Draw the normal PG and a perpendicular to it
    from G , meeting either of the focal distances in Q . Then a perpendicular to
    the focal distance drawn from Q will meet the normal in 0 , the ceutre of
    curvature.
    MISCELLANEOUS THEOREMS. 255

    Proof. — ( Ellipse or Hyperbola.) By (1259),


    the radius of curvature at P

    = F = 60 PG (1195) = PG (1194)
    = PG sec'SPG = PO.
    For AC = PE (by 1179).
    (Parabola .) By (1261) . The radius of curvature
    = 2SP - SY = 2SP sec SPG = PO.
    For 2SP = PQ, because SP = SG (1125) .

    Miscellaneous Theorems.
    1267 In the Parabola (Fig . 1239) let QD be drawn perpendicular to PV,
    then QD' = 4AS. PV. (1231, 1239)
    1268 Let RPR be any third tangent meeting the tangents OQ, OQ ' in
    R , R '; the triangles SQO, SRR , SOQ are similar and similarly divided by
    SR, SP, SR (1233- 4 ).
    1269 CoR . OR .OR = RQ .RQ .
    1270 Also , the triangle PQQ = 20RR . (1244)
    With the same construction and for any conic ,
    1271 OQ : OQ = RQ .RP : RQ . PR. (1215, 1243)
    1272 Also the angle RSR = iQSQ'. (1181)
    1273 Hence, in the Parabola ,the points 0 , R , S , R are concyclic, by (1234 ).

    1274 In any conic (Figs. 1171), SP : ST = AN : AT.


    Proof.- ST
    SP = ST
    SE = CA (69, 1163) = CN (1170) = AN (69).
    1275 COR . - If the tangent PT meets the tangent at A in R , then SR
    bisects the angle PST (VI. 3 ).
    1276 In Figs. (1178) , SY', S' Y both bisect the normal PG .
    1277 The perpendicular from S to PG meets it in CY.
    1278 If CD be the radius conjugate to CP, the perpendicular from D
    upon CY is equal to BC.
    1279 SY and CP intersect in the directrix .
    1280 If every ordinate PN of the conic (Figs. 1205) be turned round N ,
    in the plane of the figure, through the same angle PNP , the locus of P ' is
    also a conic, by (1193). The auxiliary circle then becomes an ellipse , of
    which AC and BC produced are the equi- conjugate diameters .
    If the entire figures be thus deformed , the points on the axis AA remain
    fixed while PN , DR describe the same angle. Hence CD remains paraliel to
    PT. CP , CD are therefore still conjugate to each other.
    256 GEOMETRICAL CONICS.

    Hence, the relations in (1205- 6 ) still subsist when CA, CB are any con
    jugate radii. Thus universally ,
    1281 PN : CR = DR : CN or PN .CN = DR . CR.
    1282 If the tangent at P meets any pair of conjugate diameters in T, T',
    then PT. PT is constant and equal to CD'.
    PROOF. — Let CA, CB (Figs. 1205 ) be the conjugate radii, the figures
    being deformed through any angle. By similar triangles,
    PT : PN = CD : DR
    PT : CN = CD : CR } ,' . therefore PT. PT': PN . CN = CD : DR . CR.
    Therefore PT. PT' = CD , by (1281).
    1283 If the tangent at P meets any pair of parallel tangents in T , T ,
    then PT . PT = CD , where CD is conjugate to CP.
    PROOF. — Let the parallel tangents touch in the points Q , Q '. Join PQ ,
    PQ', CT, CT". Then CT,CT are conjugate diameters (1199, 1201) . There
    fore PT. PT = CD (1282).
    1284 Cor.- QT.QT = CD”, where CD is the radias parallel to QT.
    1285 To draw two conjugate diameters of a conic to include a given
    angle . Proceed as in (1252 ii.), making PP in this case the chord of the
    segment of a circle containing the given angle ( III. 33).
    1286 The focal distance of a point P on any conic is equal to the length
    QN intercepted on the ordinate through P between the axis and the tangent
    at the extremity of the latus rectum .
    Proof. — (Fig. 1220 ). QN : NX = LS : SX = e and SP : NX = e.
    1287 In the hyperbola (Fig. 1183). CO : CA = e. - (1162, 1164).
    If a right line PKK ' be drawn parallel to the asymptote CR, cutting the
    one directrix XE in K and the other in K '; then
    1288 SP = PK = e . CN - AC ; SP = PK = e . CN + AC.
    PROOF. - From CR = e .CN (1287) and CE = AC (1186) .
    1289 Cor. — Hence the hyperbola may be drawn mechanically by the
    method of (1249) by merely fixing the cross-piece of the T -square at an
    angle with the bar equal to BCO .

    1290 DEFINITION.— Confocal conics are conics which have the same foci.
    1291 The tangents drawn to any conic from a point T on a confocal conic
    make equal angles with the tangent at T .
    Proof. — (Fig . 1217.) Let T be the point on the confocal conic.
    SY : SZ = SZ' : SY' (1178).
    Therefore ST and S ’ T make equal angles with the tangents TP , TQ ; and
    they also make equal angles with the tangent to the confocal at T (1168 ) ,
    therefore & c .
    1292 In the construction of (1253), PZ is equal to half the chord of
    curvature at P drawn through the centre C (1259 ).
    H

    DIFFERENTIAL CALCULUS.

    INTRODUCTION .

    1400 Functions. - A quantity which depends for its value


    upon another quantity a is called a function of a . Thus, sin x ,
    log x , a “, a + ax + ac* are all functions of a . The notation
    y = f ( ) expresses generally that y is a function of X . y = sinx
    is a particular function .
    1401 f (a ) is called a continuous function between assigned
    limits , when an indefinitely small change in the value of x
    always produces an indefinitely small change in the value of
    f (a ).
    A transcendental function is one which is not purely
    algebraical, such as the exponential, logarithmic , and circular
    functions ar, log x , sin x, cos x , & c.
    If f (2 ) = f ( - x ), the function is called an even function .
    If f (x ) = - f ( - x ) , it is called an odd function .
    Thus, a " and cos x are even functions, while ai and sin x are odd functions
    of x ; the latter, but not the former, being altered in value by changing the
    sign of x.

    1402 Differential Coefficient or Derivative. - Let y be any


    function of x denoted by f (x ), such that any change in the
    value of x causes a definite change in the value of y ; then x
    is called the independent variable , and y the dependent variable.
    Let an indefinitely small change in a , denoted by dx, produce
    a corresponding small change dy in y; then the ratio dir , in
    the limit when both dy and dx are vanishing , is called the
    differential coefficient, or derivative, of y with respect
    to X.
    • 2 L
    258 DIFFERENTIAL CALCULUS.

    1403 THEOREM . — The ratio dy : dx is definite for each value


    of , and generally different for different values.
    Proof. — Let an abscissa ON
    (Defs . 1160 ) be measured from 0
    equal to x,and a perpendicular or
    dinate NP equal to y . Then,what
    ever may be the form of the function
    y = f ( x ) , as x varies, the locus of
    P will be some line PQL. Let
    OM = a , MQ = ý' be values of x
    and y near to the former values. A то л м
    Let the straight line QP meet the
    axis in T ; and when coincides with P , let the final directiou of QP cut the
    . axis in T” .
    Then QS or - y _ PN
    v And ultimately, when QS and SP vanish,
    And,
    SP " &' - 2 NT
    they vanish in the ratio of PN : NT'. Therefore 1 = P = tan PT’ N , a
    definite ratio at each point of the curve, butdifferent at different points.

    1404 Let NM , the increment of æ , be denoted by h ; then,


    when h vanishes, y = f(a +h) –f(x) = f'(a ),
    a new function of a , called also the first derived function ,
    The process of finding its value is called differentiation .

    1405 Successive differentiation . If or f”(x) be differ


    entiated with respect to x , the result is the second differential
    cocfficient of f ( r ), or the second, derived function ; and so on
    to any number of differentiations. These successive functions
    may be represented in any of the three following systems of
    notation :

    f'(x ), f" (x ), f'' (Q ), fiv(x ), . .... F"(x );


    yre 922 43 yur ..... Ynz *
    The operations of differentiating a function of x once,
    twice, or n times, are also indicated by prefixing the symbols

    os )... (3 );
    or, more concisely , d. , dere ... dux.
    * See note to (1487) .
    INTRODUCTION . 259

    1406 If, after differentiating a function for r , æ be made


    zero in the result , the value may be indicated in any of the
    following ways : Ž U ), Yx03 , dxo .

    If any other constant a be substituted for æ in Yx , the


    result may be indicated by . Y.za .
    1407 Infinitesimals and Differentials . — The evanescent
    quantities dx , dy are called infinitesimals ; and , with respect
    to x and y , they are called differentials . dar, d ’y are the second
    differentials of x and y ; dx',dy the third,and so on.
    1408 The successive differentials of y are expressed in terms
    of die by the equations
    dy = f" (xx ) dx ; dy = f " ( x ) dxP; & c., and d " y = f" (x ) dx” .
    Since f ' (x ) is the coefficient of dx in the value of dy, it has
    therefore been named the differential coefficient of y or f (x ).*
    For similar reasons f" ( x ) is called the second , and f" (a ) the
    nth differential coefficient of f (x ), & c .
    1409 Two infinitesimals are of the same order when their
    ratio is neither zero nor infinity.
    If dx , dy are infinitesimals of the same order, dx , dy , and
    dady will be infinitesimals of the second order with respect to
    da, dy ; dx , dx dy, & c . will be of the third order, and so on .
    da, dx , & c . are sometimes denoted by x, ï , & c.
    1410 LEMMA. - In estimating the ratio of two quantities, any
    increment of either which is infinitely small in comparison
    with the quantities may be neglected .
    Hence the ratio of two infinitesimals of the same order is
    not affected by adding to or subtracting from either of them
    an infinitesimal of a higher order .
    Example.—I dy da dx* = dxdy –dæ = d ,for dx is zero in comparisonwith
    the ratio ay . Thus,in Fig. (1403), putting PS = dæ, QS = dy; we have
    ultimately
    PV RS
    , by dy(1258),
    - kdx
    QRdy =inkdz?, where k is a constant. Therefore
    the limit hy the principle iust enun
    NT - PS dx dx in the limit, by the principle just enunciated ;
    that is , QR vanishe s in compari son with PS or QS even whin those lines them
    selves are infinitely small.
    * The name is slightly misleading , as it seems to imply that f ' ( c) is in
    some sense a coefficient of f ( x ) .
    260 DIFFERENTIAL CALCULUS.

    DIFFERENTIATION .
    DIFFERENTIATION OF A SUM , PRODUCT, AND QUOTIENT.
    Let u , v be functions of x , then
    1411 d(u +v) – du + dv.
    dx

    1412

    1413
    (5)= (s die meist ".
    Proof. — (i.) d ( u + v) = ( u + du + v + dv) - (u + v ) = du + dv.
    (ii.) d (uv) = (u + du) (v + dv ) - uv = vdu + udv - dudv,
    and, by ( 1410), dudv disappears in the ultimate ratio to dx.
    (iii.).)
    (üü a ( u ) = u + du - u – vdu - udv
    ( ) = + dv - ö = Tu + dv) v '
    therefore & c., by ( 1410) ; v dv vanishing in comparison with vi.
    Hence, if u be a constant = c,
    1414 cde anda (O ) = -

    DIFFERENTIATION OF A FUNCTION OF A FUNCTION .


    If y be a function of % , and z a function of x ,
    1415 dy – dy dz
    dx da dx
    PROOF. — Since, in all cases , the change dx causes the change dz, and the
    change dz causes the change dy ; therefore the change dæ causes the change
    dy in the limit.
    Differentiating the above as a product, by (1412), the successive differ
    ential coefficients of y can be formed. The first four are here subjoined for
    the sake of reference. Observe that ( y.) . = 42: z.
    1416 Yx = Y % •
    1417 Y2 = yzw + y:mre.
    1418 430 = Yzz + 3y2zcz + : 23
    1419 yw = you are on + 6y3. mare + yz (3 * + 47.7sv) + Yamato
    DIFFERENTIATION . 261

    DIFFERENTIATION OF A COMPOSITE FUNCTION.


    If u and v be explicit functions of x , so that u = ° (x ) and
    v = * (Q ) ,
    1420 dF (u , v ) _ dF du , dF do .
    dx = du dx av dx
    Here dF in the first term on the right is the change in
    F (u , v) produced by du , the change in u ; and dF in the
    second term is the change produced by dv, so that the total
    change dF (u , v) may be written as in ( 1408)
    dF,+dF, = dum du + ddvF dv.
    DIFFERENTIATION OF THE SIMPLE FUNCTIONS.
    Since dy = f(a + h) - f @ ) when h vanishes,we have the
    following rule for finding its value :
    1421 RULE. - Expand f ( x + h ) by some known theorem in
    ascending powers of h ; subtract f (x ) ; divide by h ; and in
    the result put h equal to zero .
    The differential coefficients which follow are obtained by га

    the rule and the theorems indicated .

    1422 = n2* 1.
    y = x*. .
    PROOF. - Here If ((x + h ) - f ( x ) _ ( x + h )" — * * _ næ" -'h + C (n , 2 ) xn -22 ? +
    h
    ( 125) = næ -' + 0 (n, 2) an-2h + ... = ng *- , when h vanishes.
    1423 Cor.
    da = n (n - 1)...(n =r + 1)a*-5. delen = [n .
    1424 y = log. ; = alog,a
    1
    PROOF.— By (145) , -log, (x + hh ) - logaa _ 1 1 h
    a log , a
    Expand the logarithm by (155) .
    1495 Cor døy = ( - 1)"- n - 1 Put n = - 1 in (1423)
    and r = n - 1 .
    - 20" log a
    262 DIFFERENTIAL CALCULUS.

    1426 y = a*; die = a*log ,a.


    DOF
    PROOP: * = ( -1). Expand cºby(149).
    1427 Cor. boy = a*(log,a)". .
    Function. Derivative.
    Method of Proof by Rule ( 1421)
    and Limits (753 ).
    1428 sin x . cos x . Expand by (627, 629), and
    1429 cos x . - sin x . put l - cosh = 2 sin ,
    1430 tan x . secº x . Espand by (631),observing
    (1410).
    1431 cot x . - cosecº x . By cotx = tana
    ,and (1415 ).
    1432 sec x . tan x secx. By secæ = COS and (1415).
    1433 cosecx . - cotxcosecx . Similarly .
    :

    1434 sin - 2 x 1 1 If sin -?x = y, x = sin y ,


    therefore
    cos -1 ‫که‬ — x ?): dx = cos y = V1 - ** ;
    1436 tan -2x )
    cot-1 x 5 therefore and even
    1438 sec - 1 x
    cosec-1 x ) XV ( x - 1 ) Similarly for the rest.

    EXAMPLES.
    1440 (Vx); = (2 %), = 10 = (1422)

    1441 (E ).= (x=”), = =n2=*- = - MES (1422 )


    1442 {(a + x?)' (6 + xº)?} = 3 (a + 22): 22(6 + 2*) + 2 (6 + x") 3x*(a + x2)8
    = 6x ( a + x+)’ (6 .+ x ) ( + ax + 2x®). ( 1412, ’15, '22)

    ) _ (e + e -r) (ette- ') – (e* – e- ) (e - e -t) - _ 4


    vi
    14 43 f (e 7e-1 ). = ' te netete tje (e* + e - -)
    ( 1413, 1426 )

    1444 d.(log tan«)'= 2log tan a tan i sec x =: 4: lossin tan


    22 *. (1415,"24,"30).
    SUCCESSIVE DIFFERENTIATION . 263

    Some differentiations are rendered easier by taking the logarithm of the


    function. For example ,
    11 - ming ; therefore log y = { log (1 - *) - log (1 + x²) ;
    1445 y = V (1 + 2)3 ?
    therefore
    1 – 2x 3 2x
    y dx 2 (1 - 2 ) 2 (1 + x*)
    therefore = .v =— 92x ( 2 - 7 ) – 2: ( 2 — 2 ?)
    1 – 2* (1 + x ?) (1 - x2)
    1446 y = ( sin x )"; therefore log y = x log sin x ;
    therefore : : 9. = log sinæ + in cos &; sin a
    (1415,*24,"28)
    therefore The = (sin x) * (log sin x + ä сota ).
    Otherwise, by (1420), y : = < (sin x )+- cos x + (sin x )" log sin (1426 )
    = (sin æ )* (x cotx + log sin x ) .

    SUCCESSIVE DIFFERENTIATION .

    1460 Leibnitz's Theorem . - If n be any integer ,


    (ym)nz = Ynzz + ny(n -1)x Zx + C (n , 2 ) Y (n = 2)* * 2. + ...
    ... + C (n , r ) Y(n ,r)z rx + ... + yznx
    Proof. — By Induction (233). Differentiate the two consecutive terms
    C (1 , r) Yin-r)z Pre + C (n , r + 1) Y(n=- -1)2 71p+1)z
    and four terms are obtained, the second and third of which are
    C (n , r ) Yin -r)= 2( +12 + C (1 , r + 1) Yin -rizZ(p+1)=
    = { C (n , r ) + C ( n, r + 1 ) } yin -r) + % (p+ 1)= = C (n + 1, r + 1) Y(177-777)2. %( +1)71
    by (102) .
    This is the general term of the series with n increased by unity. Similarly,
    by differentiating all the terms the whole series is reproduced with n in
    creased by unity.

    DIFFERENTIAL COEFFICIENTS OF THE nth ORDER.


    1461 (sin ax)næ = a " sin (ax + int ). By Induction
    1462 ( cos ax )nx = a" cos (ax + int ). and (1428) .

    1463 (ear)nx = a" eau. (1426 )


    1464 (eary)nx = eas (a + d.)" y,
    where, in the expansion by the Binomial Theorem , dry is to
    be replaced by Yrx (1460, '63)
    264 DIFFERENTIAL CALCULUS. .

    1465 (@az cos bx)nx = pheat cos (bx + nd ),


    where a = r cos $ and b = r sin .
    Proof. — By Induction . Differentiating once more,weobtain
    grow eat {a cos (bx + no) — 6 sin (bæ + no)}
    = gomo+lenz {cos o cos (bx + ng ) - sin o sin (bx + no)}
    = yon +lear cos (bx + n + 10) .
    Thus n is increased by one.

    1466 (x2 -1 log x)n = n - 1 = x. (1460),(283)


    ( 1423)
    1467 n + 1

    1468 (tan -+ x)n. = ( - 1)^-1|n — 1 sin" 0 sin no,


    where 0 = cot- x .
    PROOF.— By Induction . Differentiating again, we obtain (omitting the
    coefficient)
    (n sin" -18 cos 0 sin no + n cos no sin " 0 ) 0 ,
    = n sin " -10 (sin në cos 0 + cosno sin 0 ) ( - sin : 0 ) .
    Since, by (1437), 0, = - (1 + ac?) - = - sin?0.
    Therefore (tan - x )(n+ 1)2 = ( - 1)" [n sin " +l sin (n + 1) 0 ,
    n being increased by one.

    ) na = (- 1)"\n sin*+18sin(n+ 1)0. (1436,1468)


    + x +)
    1469 (( 1 1.
    Sin 72 + 1

    1470 ( 12 ) = (- 1)"|n sin*+18 cos(n + 1)0.


    nude
    1

    Proor.–By(1460), (172). = (it'w) +"(Ita) - nt (n - 1) =

    Then by ( 1469).

    1471 Jacobi's Formula .


    din -1)= (1 - x2)n -1 = ( - 1)" -11 . 3 ... (2n - 1) sin (n cos=- x ) = n .
    Proof. — Let y = 1 - 2° ; therefore
    (yº+ b)nz = - (2n + 1) (xy" - 1) (n- 1)z• Also (y"+ )nz = (yy" - 1)ngo
    Expand each of these values by (1460) and eliminate (yn-b )(n- 2) , the deriva .
    tive of lowest order. Call the result equation ( 1 ) . Now assume (1471) true
    for the value n . Differentiate and substitute the result, and also (1471) on
    the right side of equation ( 1) to obtain a proof by Induction.
    THEORY OF OPERATIONS. 265

    1472 Theorem . - If y, z are functions of x , and n a positive


    integer,
    zynz = (ym)n. - n (y c)(n-1)= + C (n , 2)(yz2x)\n–2)..... + ( - 1)" Yzne
    Proof. — By Induction . Differentiate for æ , substituting for zxYne on the
    right its value by the formula itself.

    PARTIAL DIFFERENTIATION .
    1480 If u = f (x , y ) be a function of two independent vari
    ables, any differentiation of u with respect to x requires that y
    should be considered constant in that operation , and vice versâ .
    Thus, or Wer signifies that u is to be differentiated succes
    sively twice with respect to x, y being considered constant.
    05 .
    1481 The notation ♡ 42x3y* signifies that u is to be
    dæ’dys or
    differentiated successively twice for x , y being considered
    constant , and the result three times successively for y, æ being
    considered constant.
    1482 The order of the differentiations does not affect the
    final result , or Uxy = Uy.c.
    f (x + h . 1 ) - f (:r, y ) in limit. (1484)
    PROOF. - Let u = f (xy) ; then w, = !

    Uzy
    dur _ f( x + h , y + k ) - f( x, y + k ) - f (x + h, y) + f (ir, y) in 1;
    - dy I hk
    Now , if uy bad been first formed, and then Wyx , the same result would have
    been obtained . The proof is easily extended . Let u , = v ;
    then Urry = V xy = Vyz = Uyer ; and so on .

    THEORY OF OPERATIONS.
    1483 Let the symbols 0 , 1, prefixed to a quantity , denote
    operations upon it of the same class, such as multiplication or
    differentiation . Then the law of the operation is said to be
    distributive , when
    đ ( + ) = + (a ) + ( y) ;
    * See note to (1487) .
    2 M
    266 DIFFERENTIAL CALCULUS.

    that is, the operation may be performed upon an undivided


    quantity , or it may be distributed by being performed upon
    parts of the quantity separately with the same result.
    1484 The law is said to be commutative when
    QYX = YPX ;
    that is, the order of operation may be changed , 0 operating
    upon Yx producing the same result as y operating upon ºx .
    1485 0m x denotes the repetition of the operation 0 m times,
    and is equivalent to $ $ ... a to m operations. This definition
    involves the index law ,
    0°°* x = 0m +nx = 0n+mx ,
    which merely asserts ,that to perform the operation n times in
    succession upon r , and afterwards m times in succession upon
    the result, is equivalent to performing it mtn times in suc
    cession upon x .
    1486 The three laws of Distribution , Commutation , and the
    law of Indices apply to the operation of multiplication , and
    also to that of differentiation ( 1411 , ’12). Therefore any
    algebraic transformation which proceeds at every step by one
    or more of these laws only, has a valid result when for the
    operation of multiplication that of differentiation is sub
    stituted .
    1487 In making use of this principle, the symbol of dif
    ferentiation employed is to , or simply do prefixed to the
    quantity upon which the operation of differentiating with
    respect to x is to be performed . The repetition of the opera
    tion is indicated by
    py does mos e dandy3 , & c., prefixed to the
    function . An abbreviated notation is dz, der, där, der3v, & c .
    Since d , Xdx = 0 , in the symbolic operation of multiplication ,
    it will be requisite , in transferring the operation to differentia
    tion , to change all such indices to suffixes when the abbreviated
    notation is being used .

    NOTE . — The notation Y , Y2x, Uzz3y, d2f3y, & c. is an innovation. It has, how .
    ever,the recommendations of definiteness, simplicity, and economy of time in
    writing, and of space in printing. The expression as requires at least
    fourteen distinct types,while its equivalent zday requires but seven. For
    THEORY OF OPERATIONS. 267

    such reasons I have introduced the shorter notation experimentally in these


    pages.
    All such abbreviated formsofdifferential coeficients as y' 7 " 7 ' ... or yyy...,
    though convenient in practice, are incomplete expressions, because the inde
    pendent variable is not specified .
    The operation d.2 3y and the derived function U233yn would be more
    accurately represented by (d ); and (uz) , the iudex as usual indicating the
    repetition of the operation. But the former notation is simpler , and it has
    the advantage of separating more clearly the index of differentiation from the
    index of involution.
    In the symbols yo and Y2x, the figure 2 is an index in each case : in the
    first, it shows the degree of involution ; in the second , the order of differentia
    tion. The index is omitted when the degree or the order is unity, since we
    write y and yz.
    The suffix takes precedence of the superfix. y means the square of ye.
    d . (y ) would be written (yº) , in this notation .
    As a concise nomenclature for all fundamental operations is of great
    assistance in practice, the following is recommended : or y ,may be read
    " y for x," as an abbreviation of the phrase, “ the differential coefficient of y
    d 'y dlu du
    for, or with respect to, .” Similarly, Y då?' dxdy' daʻdys? q, or the shortened
    forms Y22, Uzy, U273y, may be read “ y for two x ,” “ u for wy," " u for two x , three
    y," and so forth .
    The distinction in meaning between the two forms Yne and Yzz is obvious.
    The first (in which n is numerical and always an integer) indicates n succes .
    sive differentiations for x ; the second indicates two successive differentiations
    for the variables x and z .
    The symbols and dạy or Yz09 may be read, for shortness ,
    daxi
    "y for a zero," " y for two x zero "; dize ay (xy) can be read “ d for two a
    three y of g (xy ).”
    Although the notation v , is already employed in a totally different sense
    in the Calcalus of Finite Differences, my own experience is that the double
    signification of the symbol does not lead to any confusion : and this for the
    very reason that the two meanings are so entirely distinct. Whenever the
    operation of differentiation is introduced along with the subject of Finite
    Differences, the notation ay must of course alone be employed .

    Thus, in differentiation , we have


    1488 THE DISTRIBUTIVE LAW d.(u + v) = d,u + d20. (1411)
    1489 THE COMMUTATIVE LAW d, (d,u ) = d , (d ,u )
    THE COMMUTATIVE

    or dzyu = dyxu . (1482)


    1490 THE INDEX LAW damuu = q***u,
    that is , dmadnxu = dim +n)z U . (1485)
    268 DIFFERENTIAL CALCULUS.

    1491 EXAMPLE .—
    (d , - dy)' = (do - d.) (d . - d ,) = d .d . - d ,d , — d, d + d, d, = d2e - 2dry + dzy
    Here ded , - dyd , or dry = dyz by the commutative law . (1489 )
    dod , = dne by the index law . (1490 )
    Also (d): - 22.7y + d2y ) 16 = d . u — 2dry u + day U , by the distributive law .
    Therefore, finally, (dx - d ,)? u = d2 u — 2dry u + d2yU .
    Similarly for more complex transformations.
    1492 Thus d, may be treated as quantitative, and operated upon as such
    by the laws of Algebra ; dy being written do , and factors such as dedy, in
    which the independent variables are different, being written dry, & c.

    EXPANSION OF EXPLICIT FUNCTIONS.

    TAYLOR'S THEOREM. —EXPANSION OF f (x + h).


    1500
    f(x +h) = f(x)+ hf'(x)+ * 3"(x)+ ... + F"(x + 6h),
    where 0 is some quantity between zero and unity, and n is
    any integer.
    PROOF. — (i.) Assume f ( x + h ) = A + Bh + Cha + & c.
    Differentiate both sides of this equation , - first for x , and again for h - and
    equate coefficients in the two results.
    1501 (ii.) Cox's Proof. — LEMMA. — If f ( x ) vanishes when x = a, and also
    when x = b, and if f (x ) and f' (x ) are continuous functions between the
    same limits ; then f' (a ) vanishes for some value of x between a and b .
    For f' (:-:) must change sign somewhere between the assigned limits (see
    proof of 1403) , and , being continuous, it must vanish in passing from plus to
    minus .
    1502 Now , the expression
    f(a+ )=f(a)—-f"(a)-.. ..- f"(a)
    20* + 1 n + 1 ser
    m +T } f ( a + h ) - f (a ) - hf ( a ) — ... -
    In + 1 2"
    vanishes wben x = 0 and when æ = h . Therefore the differential coefficient
    with respect to a vanishes for somevalue of x between 0 and h by the lemma.
    Let oh be this value. Differentiate , and apply the lemma to the resulting
    expression , which vanishes wben x = () and when x = 00h . Perform the
    same process n + 1 times successively , writing Oh for ooh, & c., since 0 merely
    stands for some quantity less than unity . The result shews that
    i Intl
    fool(a+ x) = {f(a + h) -f(a) — hf'(a)– ... - f"(a) }
    vanishes when x = 0h . Substituting Oh and equating to zero, the theorem
    is proved .
    EXPANSION OF EXPLICIT FUNCTIONS. 269

    1503 The last term in (1500 ) is called the remainder after


    n terms. It may be obtained in either of the subjoined forms,
    the first being due to Lagrange ,

    mg (x + 0h) ora (1 – 0)n-1f"(x + 6h).

    1504 Since the coefficient diminishes at last without


    limit as n increases ( 239, ii.), it follows that Taylor's series
    is convergent if f" (x ) remains finite for all values of n .

    1505 If in any expansion of f (x + h ) in powers of h some


    index of h be negative, then f (x ), f' (x ), f" (a ), & c. all become
    infinite.
    1506 If the least fractional index of h lies between n and
    n + l ; then fu + 1 (a ) and all the following differential coeffi
    cients become infinite .
    Proof. — To obtain the value of f" (Q ), differentiate the expansion n times
    successively for h, and put h = 0 in the result.

    MACLAURIN ’ S THEOREM .
    Put æ = 0) in (1500), and write x for h ; then , with the
    notation of (1406 ),

    1507 f(x) = f(0)+ af"(0)+ 10.1"(0)+ ...+ fu(Ox),


    where 0, as before, lies between 0 and 1.DU

    Putting y = f (x ), this may also be written


    1508 y = y + x dy + * * dPY + dºy + & c.
    1509 NOTE.- If any function f (x ) becomes infinite with a finite value of
    it, then f ( x ), s” (a ), & c. all become infinite. Thus, if f (x ) = sec -' ( 1 + x ),
    f'(x ) is infinite when x = (1438 ). Therefore $ " (0 ) , f ' (O ) , & c . are all
    infinite, and f ( ) cannot be expanded by this theorem .

    Bernoulli's Series. Put h = - x in (1500) ; thus,


    1510 FO) = f(x) — æf"(x) + * 23"(x) – 1 95F"(x) + &c.
    270 DIFFERENTIAL CALCULUS.

    1511 If • ( + k ) = 0 and $ (y) = x ; then


    k = - xy. + 1.2 Yu - 1.2.3 Yxe+ & c.
    Proof. — Let y = p -' (x ) = f (w ), and let y + k = f (x + h ) ;
    therefore * + h = q (y + k ) = 0 .
    Therefore y +k = f(0) = f(x) —æf'(x)+ 1 , 8"(x) –&c., by (1510);
    which proves the theorem .
    EXPANSION OF f (x + h , y + li).
    Let f(xy) = U. Then , with the notation of ( 1405),
    1512
    f(x +h, y + k) = u + (hu ,+ ku,)+ +2(1°u2 + 2hku, + k+uz )
    + 1.2.3(hova + 3h*ku + 3hk*u_,+kus)+ &c.
    1513 The general term is given by (hdo+ kd,)" u,
    where, in the expansion by the Binomial Theorem , each index
    of dr and d , is changed into a suffix ; and the coefficients
    dr, d2x , & c . are joined to u as symbols of operation (1487 ) ;
    thus u , is to be changed into Uzr .
    Proof. — First expand f (x + h, y + k) as a function of (x + h ) by (1500) ;
    thus, f (x + h , y + k ) = f (x , y + k ) + hf: (x, y + k ) + - , hf24 (x, y + k ) + & c .
    Next, expand each term of this series as a function of (y + k ). Thus,
    writing u for f (xy) ,
    f ( , y + k) = u t ku, 13 t kun t .. .
    +

    +
    +

    hf.(x, y + k ) = hun + hkuzy + hku, tehku,3, + . .. .. .


    13
    +

    $20 (2x, y + k ) = t h ?kuzzy h *k * 2 2y + ... ... ...


    to

    1 2 2
    +

    t
    fc (@, y + k )
    = Wu + e3Wkusy
    hkuzey t . . . .. .
    +... ... ... ... ... ...
    fw (2 , y + k ) = a h *unt ... ... ... ... ... ... ... ... ...
    The law by which the terms of the same dimension in h and k are formed,
    is seen on inspection . They lie in successive diagonals ; and when cleared
    of fractions the numerical coefficients are those of the Binomial Theorem .
    EXPANSION OF EXPLICIT FUNCTIONS. 271

    The theorem may be extended inductively to a function of


    three or more variables . Thus, if y = f ( x , y , z ) , we have
    1514 f (x + h , y + k , x + 1) = u + (hu, tku, + luz)
    + i (h’uz: + k +uzy + P+uzz + 2kluyz + 21huza + 2hkury) + ... ,
    the general term being obtained as before from the expression
    (hd,+ kd,+ ld .)"u.
    1515 CoR . - If u = f (xyz) be a function of several inde.
    pendent variables, the term (hur + ku, + luz ) proves, in con
    junction with (1410 ), that the total change in the value of u ,
    caused by simultaneous small changes in X , Y, Z , is equal to
    the sum of the increments of u due to the increments of x, y , %
    taken separately and superposed in any order.
    This is known as the principle of the superposition of small
    quantities.
    1516 To expand f ( x , y ) or f (x , y , z , ... ) in powers of x , y ,
    & c., put x , y , z each equal to zero after differentiating in
    (1512) or ( 1514 ) , and write x , y , ... instead of h , k , & c .
    1517 Observe that any term in these series may be made
    the last by writing x + oh for X , y + 0k for y , & c ., as in (1500) .

    SYMBOLIC FORM OF TAYLOR'S THEOREM .


    The expansion in ( 1500) is equivalent to the following
    1520 f (x + h ) = ende f ( ).
    PROOF. — By the Exponential Theorem ( 150), writing the indices of de as
    suffixes ( 1487),
    eidaf(x) = (1 + hd . + 3h?dx. + ...)f(x) = f(x ) + hf=(x ) + {kºf20(x) + ...,by (1488).
    CoR. - Af(x ) = f (x + h ) - f (x ) = (etc. — 1) f (x ),
    therefore Af(x ) = (ekdo — 1) f ( ),
    and generally A ” f (x ) = (end. — 1)" f (x ),
    the index signifying that the operation is performed n times
    upon f ( x ).

    1521
    PROOF.
    Similarly $ (x + h; y + k ) = ekde+kdvf (x, y).
    end,+ kd, f (x , y) = { 1 + hd, + kd , + 1 (hd. + kdv) ' + 3 (hd. + kd,)' + ... } $ (x , y )
    = f (x + h , y + k ), by (150) and (1512).
    272 DIFFERENTIAL CALCULUS.

    1522 And,generally, with any number of variables,


    f ( x + h, y + ku + 1...) = ehd,+kd,+10,*** f ( x,y , ...).
    Cor. As in (1520 ),
    Af( x, y , ...) = (eħd +kd,**** — 1) f (x , y ,z ...).

    1523 If u = f ( , y ) = ( r, o ), where x = r cos 0 , y = rsin ;


    and if x ' = rcos (@ + w ), y ' = rsin (0 + w ) ; then f ( x' y') is
    expanded in powers of w by the formula
    f (x ', y ') = e®(zdy -yda) f (x , y ).
    Proof. — By (1520 ), r being constant,
    ° (r, 0 + w ) = ewdo o (1,0 ) = ewdof (x , y ).
    Now x and y are functions of the single variable 0 ; therefore
    Un = UzXn+ u ,yo = u , ( ấr sin 6) + u , (r cos 0 ) = xu, - yuz.
    The operation d , will be transformed by the same law (1492) ; therefore
    do = æd, - yd,; therefore
    f (a ',y ) = @ (zd,- yd 2) f(x, y ) = 1 + w (xu, - yu,) + w *(z*Uzy - 2xyuxy + yºllze) + & c.

    1524 EXAMPLES. — The Binomial, Exponential, and Loga


    rithmic series for (1 + x )”, a®, and log ( 1 + x ), (125 , 149, 155),
    are obtained immediately by Maclaurin 's Theorem (1507) ; as
    also the series for sin x and cos a (764), and tan - 1 x (791).
    The mode of proceeding, which is the same in all cases, is
    shewn in the following example ; the test of convergency
    (1504) being applied when practicable .
    2
    62x + & c .
    1525 tan x = x + x + 1 2 + 1372
    7
    +
    Obtained by Maclaurin 's theorem , as follows:- Let
    f (x ) = tan x = y Therefore y , = % and 2 , = 2yz ;
    f ( x ) = sec x = z ) y and z being used for shortness.
    f" (.x ) = 2 secºx tan x = 2yz ,
    ' (x) = 2 (zyr + yzr) = 2 (2° + 2y+z ),
    fiv (x ) = 2 (4yz + 4yzº + 4yºz ) = 8 (2yz + yºz ),
    f ' ( 20 ) = 8 ( 22° + 8yºz? + 3y?z? + 2y +z ) = 8 (228 + 11yºz + 2y +z ),
    fri (x ) = 8 (12yz + 22yz® + ...) = 272yz + ...,
    fri (x ) = 272 * + ... & c.,
    the terms omitted involving positive powers of y, which vanish when x is
    zero, and which therefore need not be computed if no term of the expansion
    higher than that containing x ? is required .
    EXPANSION OF EXPLICIT FUNCTIONS. 273

    Hence, by making a = 0 , and therefore y = 0 and z = 1, we obtain


    f (0 ) = 0 ; f ( 0 ) = l ; " (0 ) = 0 ; f'" (0 ) = 2 ; Fiv (0) = 0 ; f ' (0) = 16 ;
    fri( 0 ) = 0 ; frii(0 ) = 272.
    Thas the terms up to w ?may be written by substituting these values in
    (1507).
    In a similar manner , may be obtained
    se
    1526 secx = 1 + + ** + 92 + &c..........
    Methods of expansion by Indeterminate Coefficients.
    92

    1527 RULE 1 . - Assume f ( x ) = A + Bx + Cx² + & c . Differ


    entiate both sides of the equation . Then expand f' (x ) by some
    known theorem , and equate coefficients in the two results to
    determine A , B , C , 8c.
    1528 Ex. sin “? x = x + 51 203 , 1 . 3 1.3 , 1 . 3 . 5 x ?
    + 57 +T t + & c.
    Obtained by Rule I. Assume
    sin - x = A + Bx + Cr? + Dw8 + E .c* + F2 + ..........
    Therefore, by ( 1434), ( 1 - x *) * = B + 2Cx + 3D.x2 + 4 Ecº + 57x* + .........
    But,by Bin. Th.(128), (1–2*)** = 1+ x + 1, 2 + 2820 +..............
    1
    1.3
    Equate coefficients ; therefore B = 1; C = 0 ; D = . - . ; E = 0 ; F = 52 . 4 . 5 ;
    & c. By putting x = 0, we see that A = f (0 ) always. In this case
    A = sin - 0 = 0 .

    In a similar manner, by Rule I.,


    1529 esins = 1 + x + m2 - 3.x * - 81
    8.25 3.28
    - TE ++ && cc.............

    1530 RULE II. - Assume the series, as before, with unknown


    coefficients . Differentiate successively until the function re
    appears. Then equate coefficients in the two equivalent series.
    1531 Ex. — To expand sin æ in powers of x .
    Assume sin x = A + Bx + Cx? + D .x® + Ec* + Fx + ......
    Differentiate twice, cosc = B + 2Cx + 3D .c? + 4E23 + 5 Fu* + ........
    - sin x = 2C + 3 . 2Dx + 4 . 3E .c* + 5 .4728 + ...... ...
    Put x = 0 in the first two equations ; therefore A = 0 , B = 1,
    Equate coefficients in the first and third series.
    2 N
    274 DIFFERENTIAL CALCULUS.
    -
    Thus - 20 = A, :: 0 = 0 ; -3.2D = B, · D = - ;
    –4.3E = C, : E = 0 ; -5.4F = D, :: F = 1213.4. & c.
    Therefore sinæ = + 1 .45- &c., as in (764).
    1532 RULE III. — Differentiate the equation y = f ( x) twice
    with respect to x , and combine the results so as to form an
    equation in y , yx, and y2x: Next assume y = A + Bx + Cx' + & c .
    Differentiate twice , and substitute the three values of y , yx, yux
    so obtained in the former equation . Lastly , equate coefficients
    in the result to determine in succession A , B , C , & c .
    1533 Ex. - To expand sin mo and cosmoin ascending powers
    of sin 0 or cos 0 .
    These series are given in (775–779). They may be obtained by Rule III.
    as follows :
    Put x = sin 0 and y = sinmo = sin (m sin -' x ).
    m
    Therefore Yo = cos (m sin -? ) – (1434 ) ...... ...
    ✓1- 23
    mo + cos (m sin -2x ) __ m x
    Y 2 = - sin (m sin - .0
    (1 - x )}
    Therefore, eliminating cos (m sin- « ), (1 – ~ ?) 42: - xyz + m 'y = 0 .........(ii.)
    Let y = A + Ayx + Aqxº + ...... + A ,& " + ......... ............ (iii.)
    Differentiate twice, and put the values of y, Y., and yze in equation (ii.) ;
    thus 0 = mi' ( A + A , + A, 2* + A3x® + ... + Ana" + ...............)
    – ( A , + 2A,« + 3A3x ? + ... + nArx* -+ + ..............)
    + ( 1 – 22) {24, + 2 .343x + ..........
    + (1. – 1) nA. 2" -2 + n (n + 1) An+12 " -1 + (n + 1) (n + 2 ) An+2# " + ... }.
    Equating the collected coefficients of æ" to zero, we get the relation
    _ n - ma
    A2 +2 = 01 A , ......... ... ......... ... ... (iv .)
    (n + 1 ) (n + 2)
    Now , when x = 0 , y = 0); therefore A = 0, by (ii.). And when x = 0 ,
    Yr = m , by (i.) ; and therefore A , = m , by differentiating (ii .). The relation
    ( iv .) furnishes the remaining coefficients by making n equal to 0, 1, 2, 3, & c.
    in succession ,
    Cosmo is obtained in a similar way.

    1534 RULE IV . - F'orm the equation in y , yx, and yes , as in


    Rule III. Take the nth derivative of this equation by applying
    Leibnitz's formula (1460 ) to the terms, and an equation in
    EXPANSION OF EXPLICIT FUNCTIONS. 275

    Yin+2)x, y (n+ 1)x, and ynx is obtained . Put x = 0 in this ; and


    employ the resulting formula to calculate in succession y3x0 ,
    Y410, 8c. in Maclaurin 's expansion ( 1507).
    1535 Ex. eaasinsin^- * _= 1 + ax + 45
    a (a ’ + 1 )
    x +, "a 1"
    1 .2 .3
    m3

    a ( a ? + 22) 4 , a (a ’ + 1 ) ( a “ + 3 ) + & c.
    + 1. 2 .3 .4 * * + 1. 2 . 3 . 4 . 5
    Obtained by Rule IV . Writing y for the function , the relation found is
    (1 – xº) 422 - xyz - ały = 0.
    Differentiating n times, by (1460), we get
    (1 – 2°) Yin +2)2 – (2n + 1) XY(n+1; – (a ’ + nº) ynx = 0 .
    Therefore Yin +2370 = (u ? + nº) yngo , a formula which produces the coeffi .
    cients in Maclauriu 's expansion in succession when you and Y200 have been
    calculated .

    1536 ARBOGAST'S METHOD OF EXPANDING $ (*),


    where < = a + ajx + 13 x + 3 49 + &c. .........(i.)
    Let y = 0 (2 ). When x = 0 , y = º ( a ) ; therefore , by
    Maclaurin 's theorem (1508 ),

    y = f (a) + ay: + 942. + Yar + &c. ......(ii.)


    Hence, in the values of Yea Yzr, & c.,at (1416 ), « has to be
    put = 0 .
    Now , when x = 0 , x = a ; therefore yz, Y22, & c . become d '(a ),
    " (a ), & c . ; and 7.ro, 72.70, 7320, & c . become an, An, Az, & c . Hence
    yzo = $ (a ) az,
    Yzco = $ " (a) a + $ (a )az,
    yu = $ " (a ) a + 38" (a ) a ,a + $ (a )az, & c.
    1537 EXAMPLE. — To expand log ( a + bx + cu? + 0.0% + & c.).

    Here a = b, ax= 24, = 6d,$'(a)= x 9"(a)= - 49"(a)


    Therefore yo = Yo= - + =
    Therefore, substituting in (ii.),we obtain as far as four terms,
    log(a+bx+cz -...)=loga+ * + - )*'+ ( + )x +
    276 DIFFERENTIAL CALCULUS.

    1538 Ex. 2. — To expand (a + aja + agie?+ ... + amore")" in powers of x .


    Arbogast's method may be employed ; otherwise, we may proceed as
    follows. Assume (a , + aza + age? + ... Q „ X” )' = 4 , + 42x + A©¢* + ............
    Differentiate for æ ; divide the equation by the result ; clear of fractions,
    and equate coefficients of like powers of x .

    BERNOULLI'S NUMBERS.

    1539 = 1- +B,1 -B. + &.,


    where B ,, By, & c. are known as Bernoulli's numbers. Their
    values , as far as B18, are

    They are found in succession from the formula


    1540 nB,-1 + C (n, 2) Bu- + C (n ,3) Bm-st ...
    ... + C (n ,2) B - in + 1 = 0,
    the odd numbers B3, B3, & c. being all zero. его .

    Proof. —Let y =
    ROOF

    Then,by (1508),
    y = x +ym2 +9.20 + Yzo *+ yer alan+ &e.
    Here Yarro = ( - 1) "+1B. Now y = 1 and yo = - , by (1587). Also
    ye* = y + x . Therefore, by (1460), differentiating n times,
    et { yne + nyin -1)= + C (n, 2) Yin -2)2 + ... + nyz + y } = Yng.
    Therefore wYin -1} 0 + C (1 , 2) Yin -2,00 + ... + nyro + y = 0 .
    Substitute B , -1, B1-2, & c ., and we get the formula required .
    By, B5, B ,, & c. will all be found to vanish . It may be proved, à priori,
    that this will be the case : for
    1541
    Therefore the series (1539 ) wanting its second term is the expansion of the
    expression on the right. But that expression is an even function of x (1401 ) ;
    changing the sign of a does not alter its value. Therefore the series in ques
    tion contains no odu powers of x after the first.
    1542 The connexion between Bernoulli's numbers and the sums of the
    powers of the natural numbers in ( 276 ) is seen by expanding ( 1 - 0 ) - in
    powers of e', and each term afterwards by the Exponential Theorem ( 150) .
    EXPANSION OF EXPLICIT FUNCTIONS. 277

    1543
    # = 4–B,2–1,4 +B.21–1, –B.(2-1) +&c.
    1544

    21= 2{B.(2–1) g–B.(2-1) d+B.2 –1; -&c.}.


    .

    Phoor. c1= 231 and + =1-4 47 and by(1530).


    19

    1545 1+ x +of the .......= 2352 By


    i et + 1 ,
    Proof. - In the expansion of S (1540) substitute 2i0 for x , and it
    becomes the expansion of ocot (770). Obtain a second expansion by differ
    entiating the logarithm of equation (815, sino in factors). Expand each
    term of the result by the Binomial Tbeorem , and equate coeflicients of like
    powers of 0 in the two expansions.

    STIRLING ' S THEOREM .


    1546 $ (x + h ) - ° (x ) = h¢}(x ) + Azh {¢ (x + h ) — *' (x)}
    + Agh? {$ " (x + h) - 8 " (x ) } + & c .,
    where A2n = ( - 1)" B2n + |2n and A2n+ 1 = 0 .
    Proof. — A1, A2, A3, & c. are determined by expanding each function of
    x + h by ( 1500 ) , and then equating coefficients of like powers of x . Thus

    -4,= 0; 13 - 2 -4 = 0; 1 - 4 - 4 -4 = 0; &c.
    To obtain the general relation between the coefficients : put 9 ( 2 ) = ,
    since Ay, Ag, & c. are independent of the form of o . Equation (1546) then
    produces - = l - Azh — Agh ?— Ag79 — & c.;
    and ,by (1539), we see that, for values ofn greater than zero,
    42n +1 = 0 and 4.2n = ( - 1)" B2n + | 2n .

    BOOLE'S THEOREM .
    1547 + (a + b ) - (2 ) = A { ( + 1 ) + + (a ) }
    + Agh? {$ ” (x + h ) + $" (x )} + & c.
    PROOF. - A1, Ag, Ag, & c. are found by the same method as that employed
    in Stirling's Theorem .
    278 DIFFERENTIAL CALCULUS.

    For the general relation between the coefficients, as before, make (w ) = e",
    and equation (1547) then produces
    e - 1
    t = Azh + A22° + Agh* + & c.;
    and, by comparing this with (1544), we see that
    A2 = 0 and A2n-1 = (–1)n- B, 29 ,1 | 212

    EXPANSION OF IMPLICIT FUNCTIONS.


    1550 DEFINITION. — An equation f (x, y) = 0 constitutes y
    an implicit function of x . If y be obtained in terms of x by
    solving the equation , y becomes an explicit function of x .

    1551 LEMMA. -— If y be a function of two independent


    variables x and z ,
    d { F (y) y ;} = d. { F ( y) yx }.
    PROOF. — By performing the differentiations, we obtain
    F'(y) y-y: + F (y) Yzz and F '(y) y:yz+ F (y) Yra,
    which are evidently equal, by (1482).

    LAGRANGE' S THEOREM .
    1552 Given y = x + xº (y), the expansion of u = f (y ) in
    powers of x is
    n - 1

    f(y) = f(-) + x® (-) fº(-) + ... + [ {$ (x)}"f"(*)] +


    Proof. — Expand u as a function of x , by ( 1507) ; thus, with the notation
    of (1406 ), u = + x Ur0 + *o -U200 + ... + Unxo + & c .
    Here u , is evidently f (x).
    Differentiating the equation y = z + æ° (y) for x and z in turn , we have
    Yo = (y ) + æd' (y ) Y. and y : = 1 + æ $' (y) Y :
    Therefore Y . = 4 (1) yz; and, since Un = f'(y) Y: and u, = f' (y) y.,
    therefore also Uz = 0 (y ) Uz...... ...... (i.)
    The following equation may now be proved by induction, equation (i.)
    being its form when n = 1.
    Assume that Unie = din -1): [{ 7 ( ) }" u ,] ........... .......... ( ii.)
    Therefore Win +1)= = d (n-1).d., [ {♡ (7 ) }" u ,] (1482)
    = dim-1)od . [ {$ (y ) }" ur] ( 1551) = du: [ {ø (y ) }" +\w ; ], by (i.)
    EXPANSION OF IMPLICIT FUNCTIONS. 279

    Thus, n becomes n + 1. But equation (i .) is true when n = 1 ; for then


    it is equation (i.); therefore it is universally true.
    Now , since in equations (i.) and (ii.) the differentiations on the right are
    all effected with respect to z, x may be made zero before differentiating instead
    of after. But, when x = 0, U , = f'(x) and o (y ) = 9 ( ), therefore equa
    tions (i.) and (ii.) give
    Ug = (2) f'(z) ; Uneo = d(1 -1): [{ø ( ) }" f' ( ) ].

    1553 Ex.1.–Given ye— ay +b = 0: to expand log y in powers of 1


    Here
    herefore, in Lagrange's formula,
    a = b; = ; (y) = logy; $(y) = yº; and y = x+ 2y". and

    Therefore 1o = log2; 1 = x 1 = z ;
    Uingo = dın-1)=(23* 1) = (3n —1)(3n —2)...(2n + 1)z?".
    Uz

    Therefore , substitating the values of x and z, (1552) becomes


    b 32 1 (3n - 1) (3n — 2 ) ... (2n + 1 ) 7, 21 1
    log y = sg a= + a 2n+1) la ...
    18

    1 . 2 ... n

    1554 Ex. 2.— Given the same equation : to expand y" in powers of
    f(y) is now y", and, proceeding as in the last example,we find
    S 5 21 , n n + 5 ) 64 1 , n (n + 7) (n + 8) 10 1
    Y = " + " i ūt 1.2 atat 1.2.3 O aš
    , n (n + 9 ) (n + 10 ) (n + 11) 18 1
    1.2 .3.4 as at Faco )
    +
    is

    + 3
    + 12 31,60 ,1
    • If n = 1 , y + & c. .. .. ..
    a' aa 0as
    3 + 55

    CAYLEY'S SERIES FOR


    1555 - 1

    de = A -431-6(=) +...+ 41533) [={&i=)}*-*).+...,


    where A = (0
    Proof. - Differentiate Lagrange's expansion (1552) for 2, noting that
    a = 1–20 ( ) Replace« by Y C9 Put f(y)= o);and therefore
    f'( ) = =% , since f is an arbitrary function. Then make y = 0.
    280 DIFFERENTIAL CALCULUS.

    LAPLACE'S THEOREM.
    1556 To expand f(y) in powers of a when
    y = F (x + x¢ (y) } .
    Rule . — Proceed as in Lagrange's Theorem , merely sub
    stituting F (z ) for z in the formula .

    1557 Ex. 3. — To expand e' in powers ofa when y = log (z + å sin y).
    Here f (y ) = e" ; F (x ) = logz ; $ (1) = sin y;
    In the value of Unxo (1552), 9 (2 ) becomes ¢ { F (x)} = sin log z;
    § (2) becomes f { F (x ) } = el082 = z ; therefore f' (x ) = 1.
    Thus the expansion becomes
    ev = z + å sin log + ... + d(n-1):(sin log z)".

    1558 Ex. 4 . - Given sin y = x sin (y + a ) : to expand y in powers of x .


    Here y = sin -' (x sin y + a ), with 2 = 0 .
    f (y) = y ; F (z) = sin -' ; ( y) = sin (y + a ).
    (z) in (1552) becomes ø {F (x) } = sin (sin -'z + a).
    f (z) becomes f { F (2)} = F (z ) = sin - z ; therefore F” (z) = ( 1 - 2 ) .
    Thus y = x sin ( sin -' z + a ) (1 – 2 ) –
    + 1x?d ,{sinº (sin- z + a )(1 - 22) - } + *2®d sinº(sin -' z + a )(1 - z") } } +
    with put = 0 after differentiating. The result is, as in (796 ),
    y = x sin a + 1x* sin 2a + sin 3a + & c.

    BURMANN 'S THEOREM .


    1559 To expand one function f (y) in powers of another
    function + (y ).
    Rule . — Put x = * (y) in Lagrange's expansion , and there
    fore ø (y) = (y – z) = 4 (y); therefore
    1560 fm) = f(3 +4 (w {4753"(1)}, +. .
    ...+** **dentro (147 (w } .+ &c.
    Here y = % signifies that after differentiating , is to be sub
    stituted for y .
    EXPANSION OF IMPLICIT FUNCTIONS. 281

    1561 Cor . 1.— Since x = 4 ( ), y = +-+(@); therefore ( 1560)


    becomes, by writing æ for * (y),
    ${4}(»)}= f(-)+...+ de {(473) *(w)} +... > ,y = 2

    But since the variable y is changed into z after differentiating ,


    it is immaterial what letter is written for y in the second
    factor of the general term .

    1562 Cor.2.— If f(y) be simply y, the equation becomes omes

    **(x)==+=( 7).. -..try {( 5 ) }.. + yaz

    1563 Cor. 3.- If z = 0, so that y = æp(y), we obtain the


    expansion of an inverse function ,

    * *( )= =(4% . .+. .tapi dapat {(4% )"}...+


    Y = 0 = 0

    1564 Ex. 5. — The series (1528) for sin - a may be obtained by this
    formula ; thus,
    Let sin - x = y, therefore x = sin y = y (y), in (1563); therefore
    21 y2l
    sin -' x = a Tsin
    ( + & c.
    yl , y =0 1. 2 \ sin ' y / 80 ' 1 . 2 . 3 \sinº y / 2y0

    1565 Ex.6.—If y = 1+ v 5 = 1-V6 -3"),then,by Lagrange's


    theorem (1552),since y = + wefind
    nin + 2r - 1 /
    nu

    . .. +
    gov= ) +n(*)** +...+ "\27277?(
    Ir Intr )* *+. . ...
    Put x = 2 / t, thus
    1566 ( 1 - V (1 - 4t) in nin + 2r - 1
    : t" + nt" +' + ... + {" +" +
    . 2 ir ntr
    Change the sign of n, thus
    156 " ( 1 + V (1 — 41) , nin - p - 1 .
    1 - nt + ... ( - 1)" A ll £ . . . .. . .
    I 2 Ir In - 2r

    This last series, continued to " + 1 or 1+ 1 terms, according asn is even Ven

    or odd, is equal to the sum of the two series, as appears by the Binomial
    theorem .
    20
    282 DIFFERENTIAL CALCULUS.

    Also, by Lagrange's Theorem ,


    . . . . . . ..
    1568 logy = log + (*) +...+ 7 ( )*+.
    or, by putting x = 2 vt,
    1 - / ( 1 - 4t)
    1569 log - L = t + ... + ... .. .. .
    " 5 2t

    1570 Ex. 7 . - Given xy = log y ; to expand y in powers of x .


    The equation can be adapted as follows:
    y = er , therefore xy = xey.
    Pat xy = ý , therefore y' = xe" , from which, by putting z = 0 in (1552),
    y' may be expanded , and therefore y.

    Ex. 8. — To expand eay in powers of yeby.


    Here x = yesu, ° (y) = %yeby = e-bu, if wetake z = 0. Therefore
    1571 ev= 1+ ayeb + a(a — 20)y a(a— 36)•vynos + ......
    ABEL'S THEOREM .
    1572 If (a) be a function developable in powers of ee; then
    $ (x + a) = $ (x) + a® (x +6) + " 47 ") "(x + 26)+ ...
    a (a - rb ) r- 1
    .. . . . . . . + 16)r
    1. 2 ... $"(x + rb) +
    ProOF. — Let $ (y ) = A, + A , em + A ,e?' + A ,484 + ........................ (i.)
    Put y = 0, 1, 2 , 3 , & c . in (1571) , and multiply the results respectively by
    A0, A , et, Age , & c. Then the theorem is proved by equation (i.).

    1573 Cor. — If ( ) = ", Abel's formula gives


    (x + a)" = " + na (x + 6 )n -1 + C (n , 2) a (a — 26) (x + 26 )n -2 + ...
    .................. + C ( n ,r ) a (a - rb) -1( x + rb)" -" + & c.

    INDETERMINATE FORMS.
    rms LE .
    1580 Forms Rule.— If $(3)
    * ( x ) be a fraction which
    takes either of these forms when x = a ; then $+ (a(a )) = ”,(a
    t (a )) or
    or
    te
    , the first determina iating
    fraction obtained by different
    Ye " (a )
    INDETERMINATE FORMS. 283

    the numerator and denominator simultaneously and substituting


    a for x in the result.
    1581 But at any stage of the process the fraction may be
    reduced to its simplest form before the next differentiation .
    See example (1589) .
    PROOF. — (i.) By Taylor's theorem (1500 ), since q (a ) = 0 = 4 ( a ),
    Q (a + h ) _ \ (a ) + hº' (a + 0h ) _ ¢ (a + on) _ ' (a )
    (a + h ) + (11) + ht' (u + on ) (a + on ) (a )'
    when h vanishes.
    (ii.) If y (a ) = 4 (a ) = 0 ,
    which is of the first form , and therefore
    () (1414) = f (a ) C). Therefore ® (a) =f (0)
    1582 Vanishing fractions in Algebra are of the indeter
    minate form just considered , and may be evaluated by the
    rule , or by rejecting the vanishing factor common to the
    numerator and denominator.
    _ O _ ( x - a ) (x2 + axta”)
    Ex. — When x = a ; - a' mū - (x — a ) (x + u ) = 2a = z 4.

    1583 Form 0 x co. Rule.— If $ (x)X4(x) takes this form


    when x = a, put q (a) * * (a) = (a) – ma , which is of the
    form o
    1584 Forms 0°, 00 ', 1" . RULE. — If {$ ( x )}+ (x) takes any of
    these forms when x = a , find the limit of the logarithm of the
    expression . For the logarithm = 4 (a ) log o (a ), which , in each
    case , is of the form 0x00 .

    1585 Form 6 - 0 . Rule. — If $ (x) – 4 (x) takes this form


    when x = a,we have e*(a) (a) = m = 0 ; and if the value of
    e

    this expression be found to be c,by (1580 ), the required value


    will be log c.
    1586 Otherwise: $(a) —4 (a) = º(a) {1 - @ }, which is of
    the form 0 X 0 (1583).
    284 DIFFERENTIAL CALCULUS.

    1587 Ex.1.–With a= 0, y = I = ; = < (1580) = 1.


    Also , with x = 0 ,

    1588 Ex. 2. — With x = 1 ;


    cot(mv)log« = tan (72) = 7 = cos
    "9*' (****x ) (1580) =
    TX

    1589 Ex. 3. — With x = 0 ;


    m (lon) - ( log x ) " _O n (log 2 )" - 1 in
    me - a - ma-m 7 - m )" x-mu
    by (1581) , differentiating n times and reducing the fraction to its simplest
    form after each differentiation .

    1590 Ex. 4 . — With x = 0 ; y = (1 + ax ) : = 1" .


    By(1584), log y = log (1+ ax) = = 1 _+ ax. (1580) = a ; :: y = e".
    1591 Ex. 5. — With « = * ; y = (x − x )sin« = 0°.
    By (1584 ), log ) = sin a log ( T - a) = log (T -2 ) = = = sin2
    cosecx co (T )cos e
    (1590) - 0 - 2 sin x cos x.
    . -a
    - cos x — ( — ) sin = 0 ; : : y = 1. .
    1592 If f( ) and a becomeinfinite together, then
    f(x) = f'(x) = f(x + 1) - f(x).
    Proop.= f(0) = = )(1580) = f( + 1)–f(w) (1404),
    since, when x = 00 , h may be taken = 1.

    Indeterminate forms involving two variables.


    1592 Rule. — First : If the values x = a, y = b make the frac
    tion P(2,9 =
    en + (x , y)
    ; the true value is = * , if p, and 4, both
    vanish .
    1593 Secondly : If pe : 4z = , : 4 , = k , the true value of
    the fraction is k .
    Proof . — (i.) By (1703) 01 (x(2,, y4)) =– *9 ,x ++ 4yvé
    ,19+,
    Y and
    an y being an arbitrary
    function ofx , — that is, independent of a , - the value of the fraction is inde
    terminate unless , and y , both vanish . .
    ( ii.) If we substitute di = kur and Q = kyy, the fraction becomes = k .
    JACOBIANS. 285

    JACOBIANS.

    1600 Let u , v, w be n functions of n variables x, y, z (n = 3).


    The following determinant notation is adopted :
    | u , u , uz | 1x u X, Xw
    d (uvw ) d (xyz)
    V , V, V , = d (xyz ) yu y » ую d (uvw )
    |W , W , W Zu Zo Zo
    The first determinant is called the Jacobian of u , v , w with
    respect to X , Y , Z , and is also denoted by J (uvw ), or simply
    by J .

    1601 THEOREM. — d (uvw ) x d (wym ) = 1.


    d (xyz ) d (uvw )
    Proof. - If the product of the two determinants be formed by the rule in
    (570 ), first changing the columns into rows in the second determinant (559) ,
    the first column of the resulting determinant will be
    U ,2x + 4yYx + 2 . = 2x ) and the whole Un Vu Wu ' 11001
    1 , 2, + 2 , + 4, 8 = 1. } , determinant | up 0 , 1 , 1 = 1 0 1 0 = 1.
    U;& n + 4 , 4x + , = Uw ) will be | Un V . Wow ! 1001

    1602 If u , v , w are n functions of n variables a, ß , y (n = 3),


    and a , ß , y , functions of x , y , % ;
    d (uvw ) v d (aby) _ d (uvw )
    daßy) d (xyz) d (xyz)
    ProOF. — Form the product of the two determinants, changing columns
    into rows in the second as in ( 1601). The first column of the resulting
    determinant will be
    U9, + usb , + 4 ,7x = U , ) and the whole de- | Uz V . We a com
    u ,a , tuli , tuyy = U , E , terminant will | W , V, W , =
    | U , V, W , d (xyz)
    since rowsand columns may be transposed (559 ).
    1603 CoR. - If a , ß , y are only given as implicit functions
    of x , y , z , by the equations p = 0 , x = 0 , x = 0, involving the
    six variables ; then
    d (Px4) x d (aßy) = ( - 118 d (PX4)
    d (aßy) * d (xyz) = ( - ) d (xyz )
    Proof.-- By (1737), 0 .9z + paß . + 0 , 7 = = - 9 ., where , is the partial
    derivative of $ . Thus uz Vz, W ., in the determinant, are now replaced by
    503; - X =, - 42; so with y and z ; and by changing the sign of each element,
    the factor ( - 1 )' is introduced (562) .
    286 DIFFERENTIAL CALCULUS.

    1604 If u , v, w , n functions of n variables x , y, z (n = 3 ), be


    transformed into functions of ?, n , % by the linear substitutions
    x = aš+ a,n + a: s ) then d (uru ) = m d (urw )
    y = b $ + b2n + b35 ; d ( $m5) d (xyz )
    % = c $ + c2n + c35 ) or J ' = MJ,
    where M is the determinant (a , b , c ,) called the modulus of
    transformation (573).
    Proof. — J = | U, U, U l M = a b c J' = Uy Un uz
    V, V, V , ag ba ca
    |W , W, W , lazb₂c₃1 Iwewnwel
    Form the product MJ by the rule in (570 ) . The first element of the
    resulting determinant is u tu, bitu = Uzde + + 4 .25 = Uy: Similarly
    for each element. Then transpose rows and columns, and the determinant
    J' is obtained.
    1605 When the modulus is unity, the transformation is said
    to be unimodular.

    1606 If, in ( 1600), « (uvu ) = 0 , where p is some function ;


    then J (uvw ) = 0 ; and conversely,
    Proof. — Differentiate o for x , y, and z separately, thus
    0,92 + 0 ,02 +PW, = 0 ;
    similarly y and z ; and the eliminant of the three equations is J (uvw ) = 0 .

    1607 If u = 0, v = 0 , w = 0 be a number of homogeneous


    equations of dimensions m , n , p in the same number of vari.
    ables x , y , % ; then J (uvu ) vanishes, and if the dimensions
    are equal Jr, J,, J, also vanish .
    Proof. — By(1624 ), xu + yu , + zu , = mu)
    XV. + yv, + zv, = nu ; . . Jx = A ,mu + Benv + Cipw .
    æWe + yw , + zw . = pw )
    By (582), A , B , C , being the minors of the first column of J. Therefore ,
    if u , v, w vanish, J also vanisbes.
    Again , differentiating the last equation, J + xJ, = A ,mu, + B nv, + Cipu ,.
    Therefore, if m = n = p, J + xJ, = m (Aju , + B ,02+ C w .) = m ).
    Therefore J, vanishes when J does.

    1608 If u = 0 , v = 0 , w = 0 are three homogeneous equations


    of the second degree in x , y , z , their eliminant will be the
    determinant of the sixth order formed by taking the eliminant
    of the six equations u , v, w , JZ, J,, Jr.
    QUANTICS. 287

    PROOF. — J is of the third degree, and therefore Jr, J., J, are of the second
    degree, and they vanish because u , v , w vanish, by ( 1607) . Hence u , v , w ,
    Ja, Jy, J, form six equations of the form (x , y , z ) = 0 .

    1609 If n variables x , y , % (n = 3 ) are connected with n


    other variables ? , n , % , by as many equations u = 0 , v = 0 , w = 0 ;
    then dx dy dz _ d (uvw ) . d (uvw )
    dě dn disn5) - d (xyz)
    Proof. — By (562) we have
    dx dy dz. | Uz Uyul |UzXE U ,Yn U .25 l lus un uz
    V , V, Vs = V , XE V y Yn V .25 = 1 vs Vin vs
    Wn t WyYn W ,75 Iwewnwe

    QUANTICS.

    1620 DEFINITION. — A Quantic is a homogeneous function of


    any number of variables : if of two, three variables, & c ., it is
    called a binary, ternary quantic, & c . The following will
    illustrate the notation in use. The binary quantic
    ax® + 3bx+y + 3cxyº + dys
    is denoted by (a ,b , c,dXx, y ) when the numerical coefficients
    are those in the expansion of (x + y ) . When the numerical
    coefficients are all unity , the same quantic is written
    (a , b , c,d8X ,y ) . When the coefficients are not mentioned ,
    the notation (x , y ) is employed.
    EULER’ S THEOREM OF QUANTICS.
    If u = f ( , y) be a binary quantic of the nth degree , then
    1621 Xux + yu , = nu .
    1622 . a *U2 + 2xyu ry + yºu2y = n (n - 1) u .
    1623 (æd + yd,)" u = n (n - 1) ... (n - r + 1) u . (1492)
    • Proof. — In ( 1512) put h = ax, k = ay ; then, because the function is
    homogeneous, the equation becomes
    ( 1 + a )" u = uta (xu, + yu,) + fa * ( z0 %u2x + 2xyuxy + yʻuzy) + .........
    Expand ( 1 + a )", and equate coefficients of powers of a .
    288 DIFFERENTIAL CALCULUS.

    The theorem may be extended to any quantic , the quan


    tities on the right remaining unaltered . Thus, in a ternary
    quantic u of the nth degree,
    1624 xur + yu, tzu , = nu ; and generally
    1625 (xde + yd , + zd .)" u = n (n - 1) ... (n - r + 1) u .
    DEFINITIONS.
    1626 The Eliminant of n quantics in n variables is the
    function of the coefficients obtained by putting all the quantics
    equal to zero and eliminating the variables (583, 586 ) .
    1627 The Discriminant of a quantic is the eliminant of its
    first derivatives with respect to each of the variables (Ex .
    1631) .
    1628 An Invariant is a function of the coefficients of an
    equation whose value is not altered by linear transformation
    of the equation , excepting that the function is multiplied by
    the modulus of transformation (Ex. 1632).
    1629 A Covariant is a quantic derived from another quantic ,
    and such that, when both are subjected to the same linear
    transformation, the resulting quantics are connected by the
    same process of derivation (Ex. 1634 ) .
    1630 A Hessian is the Jacobian of the first derivatives of a
    function .
    Thus, the Hessian of a ternary quantic u , whose first
    derivatives are Ur, U , , Uz, is
    | U2x Ury Urz
    d (u , u,uz) = I U ya Uzy Wyz
    d ( xyz )
    | Uzx Uzy U22 |
    1631 Ex. — Take the binary cubic u = ax + 3bx*y + 3cxy? + dy". Its first
    derivatives are
    Uz = 3ax: + 6bxy + 3cy?, 13a 6b 3c 01
    Uy = 3bx ' + 6cxy + 3dv . 10 3a 6b 3c
    36 6c 3d 0
    Therefore (1627) the discriminant of u 10 36 6c 3d
    is the annexed determinant, by (587) .
    1632 The determinant is also an invariant of u , by ( 1638 ) ; that is, if u
    be transformed into v by putting x = 0 & + on and y = a'& + ß'n ; and, if a
    corresponding determinant be formed with the coefficients of v , the new
    determinant will be equal to the original one multiplied by (aß' - a'ß )'.
    QUANTICS. 289

    1633 Again, Uge = bax + 6by, Uzy = 6cx + 6dy, Uzy = 6bx + 6cy.
    Therefore, by (1630), the Hessian of u is
    UzzU2y — uły = (ax + by) (cx + dy) - (bx + cy )
    = (ac — _*) x + + (ad — bc) xy + (bd — c*) y .
    1634 And this is also a covariant; for, if u be transformed into v, as
    before, then the result of transforming the Hessian by the same equations
    will be found to be equal to U21 V2y — uzy. See (1652).

    1635 If a quantic, u = f (x , y,z ...), involving n variables


    can be expressed as a function of the second degree in
    X , X , ... X.,- 1, where the latter are linear functions of the
    variables , the discriminant vanishes.
    PROOF. — Let u = px + { X , X , + xX,X , + & c.,
    where X , = aja + by + 62 + & c.
    The derivatives Up W ,, & c. must contain one of the factors X , X , ... Xn- , in
    every term , and therefore must have, for common roots , the roots of the
    simultaneous equations X , = 0, X , = 0 , ... X : -1 = 0 ; n - 1 equations being
    required to determine the ratios of the n variables. Therefore the dis
    criminant of u, which (1627) is the eliminant of the equations u , = 0, U, = 0,
    &c., vanishes, by (588).
    1636 Cor. 1. - If a binary quantic contains a square factor,
    the discriminant vanishes ; and conversely .
    Thus, in Example (1631), if u has a factor of the form (Ax + By) , the
    determinant there written vanishes.
    1637 COR. 2 . — If any quadric is resolvable into two factors ,
    the discriminant vanishes.
    An independent proof is as follows:
    Let u = XY be the quadric, where
    X = (ax + by + cz + ...), Y = (a'x + b'y + c'z + ...).
    The derivatives Ur, Uy, U , are each of the form pX + qY, and therefore have
    for common roots the roots of the simultaneous equations X = 0 , Y = 0 .
    Therefore the eliminant of U = 0, 4, = 0, & c. vanishes (1627).

    1638 The discriminant of a binary quantic is an invariant.


    PROOF. — A square factor remains a square factor after linear transforma
    tion. Hence, by (1630 ), if the discriminant vanishes, the discriminant of
    the transformed equation vanishes, and must therefore contain the former
    discriminant as a factor (see 1628) . Thus the determinant in ( 1631) is an
    invariant of the quantic u .

    The discriminant of the ternary quadric


    1639 u = ax® + by? + cx ?+ 2fyz + 2gzx + 2hxy
    2 P
    290 DIFFERENTIAL CALCULUS.

    is the eliminant of the equations


    1640 ju, = ax + hy + g = 0 ) that is, the
    la h 81
    žu, = hx + by + fx = 0 y% ; determinant \h ) f
    lu , = gx + fy + cx = 0 ) lg fcl
    1641 = abc + 2fgh - af" — by2 - ch ” = A .
    1642 The following notation will frequently be employed .
    The determinant will be denoted by A , and its minors by
    A , B , C , F , G , H . Their values are readily found by differ
    entiating A with respect to a , b , c , f, g , h ; thus,
    A = bc - fº = A , B = ca - g ? = As, C = ab - hº = A .,
    F = gh - af = { As, G = hf - bg = 14,, H = fg - ch = itron
    1643 The reciprocal determinant is equal to A ’ or
    A HG | la h g |2
    HB F = h bf By (575).
    G F C lg f c

    1644 The discriminant of the quaternary quadric


    u = ax? + by® + cm + dwº + 2fyz + 2gzr + 2hry
    + 2pxw + 2qyw + 2rzw
    is the eliminant of the equations
    1645 fun = ax + hy + gx + pw = 0 , that is, a " & P
    lu, = hx + by + fz + qw = 0 the ho fa
    lu , = gx + fy + cm + rw = 0 deter lg for
    zuv = px + qy + rx + dw = 0) minant Ip q r d

    The determinantwill be denoted by A', and , by decom


    posing it by (568), we have
    1646 A ' = d . 1 - Ap — Bqº - Cr — 2Fqr – 2Grp — 2Hpq.
    1647 A ' = } (pA ; + 94 + ra,) + d . A .

    1648 THEOREM .- If $ (ry ) be a quantic of an even degree ,


    $ (d ,, - d . ) $ ( x , y )
    is an invariant of the quantic .
    QUANTICS. 291

    PROOF.— Let the linear substitutions ( 1628) be


    æ = ax + bn, y = a'& + B 'n .. ...... .. . .. . .. . . .. ( i.)
    Solve for & and n. Find &z, Ey, " x, Ny, and substitute in the two equations
    1 . d, dęty + dnny; de = de &x + dynx.
    The result is
    d , = {ada + b ( - dę) } = M ; - d, = {a'dn + ( - dę) } = M ......(ii.),
    where M = ab' - a 'b, the modulus of transformation. Equations (i.) and (ii.)
    are parallel, and show that the operations d , and — d , can be transformed in
    the same way as the quantities x and y ; that is, if ♡ (x , y ) becomes 4 (5, n ) ,
    then « (dy, - d .) becomes 4 (dn, - d ) M ", where n is the degree of the
    quantic p. But (dy, - dr) ° (x, y) is a function of the coefficients only of
    the quantic 9 , since the order of differentiation of each term is the same as the
    degree of the term ; therefore the function is an invariant, by definition ( 1628).

    1649 EXAMPLE. — Let (3, y) = ax* + bx8y + exy8 + fy ". The quantic must
    first be completed ; thus, ° (xy) = ax* + bwºy + cx *y* + exy + fy ", (c = 0 ) ; then
    p (dys - dz) y (x , y ) = (adoy - bdzyx + cdayze - edyse + fdar ) • ( , y )
    = a . 24f — 6 .6e + c.4c - e.6b + f .24a = 4 (12af , 3be + cº) .
    Therefore 12af - 3be + ca is an invariant of Q, and = (12AF - 3BE + 04) = M ,
    where A , B , C, E , F are the coefficients of any equation obtained from o by a
    linear transformation .
    But if the degree of the quantic be odd, these results vanish identically .

    1650 Similarly, if $ (@, y), 4 (x , y ) are two quantics of the


    same degree, the functions
    $ (dy, — dx) + (x , y) and 4 (dy, — d .) $ (x , y )
    are both invariants.
    1651 Ex.- If $ = ax?+ 2bxy + cy? and = áʻx2 + 26ʻxy + cy ; then
    (adzy - 2bdoy + cdze) (a'x?+ 26'xy + c'y?) = ac' + ca' – 216, an invariant.
    1652 A Hessian is a covariant of the original quantic .
    PROOF. — Let a ternary quantic u be transformed by the linear substitu
    tions in ( 1604 ) ; so that u = Q (x , y , z ) = 4 (E, n, 5) . The Hessians of the
    two functions are d (Uz1,02) ond d (uç uz) (1630). Now
    d ( xyz) d (Ens)
    d (ugu ,uz) = yd (ugu,ug) - vd(1,0,0.) = M :d (1, ,u.).
    d (Eng) d (xyz) : d ($15) d (xyz)
    The second transformation is seen at once from the form of the determi.
    nant by merely transposing rows and columns ; the first and third are by
    theorem ( 1604 ). Therefore, by definition ( 1629), the Hessian of u is a
    covariant.

    1653 Cogredients. Variables are cogredient when they are


    subjected to the same linear transformation ; thus, x , y are
    292 DIFFERENTIAL CALCULUS.

    cogredient with x', y when


    x = af + bn ? and a ' r af + bn ' .
    y = $ + dn y' = ct'+ dr' so
    1654 Emanents. If in any quantic u = q (x , y), we change
    æ into x + px', and y into y + py', where x ' , y' are cogredient
    with x , y ; then , by ( 1512),
    $ (x + px', y + py')
    = u + p (x 'dx + y'd ,) u + ipº (x 'do + y'd ,) u + & c.,
    and the coefficients of p , p , p', ... are called the first , second ,
    third , . .. emanents of u .

    1655 The emanents, of the typical form (a'd + y'd,)” u , are


    all covariants of the quantic u .
    Proof. — If, in 0 (x", y ), we first make the substitutions which lead to the
    emanent, and afterwards make the cogredient substitutions, we change
    æ into x + px', and this into ax + bn + p (ax' + bn').
    And if the order of these operations be reversed, we change
    æ into aš + bn, and this into a ( + pE ) + 6 (n + pn') .
    The two results are identical, and it follows that, if Q ( , y) be transformed
    by the same operations in reversed order , the coefficients of the powers of p
    in the two expansions will be equal, since p is indeterminate . Therefore , by
    the definition ( 1629 ), each emanent is a covariant.

    1656 For definitions of contragredients and contravariants, see ( 1813 – 4 ).


    For other theorems on invariants, see (1794 ), and the Article on Invari
    ants in Section XII.

    IMPLICIT FUNCTIONS.

    IMPLICIT FUNCTIONS OF ONE INDEPENDENT VARIABLE .


    If y and z be functions of x , the successive application of
    formula (1420 ) gives , for the first, second , and third deriva
    tives of the function $ (y , z ) with the notation of (1405 ) ,
    1700 0x( yz ) = $wy: + m.
    1701 02: (ym) = $2y + 20vxy.c% + + $22 + Ⓡuyz +.pozze
    IMPLIỚIT FUNCTIONS. 293

    1702 Post (yz) = 43x4 + 302yzy_mx + 3y2zYcm + 63


    + 3 ($ 299x + puzzx) y2c + 3 (42zmx + 2yYx)220
    + $ ,43 + $ 2 % 3x
    By making z = x in the last three formulæ , and conse
    quently z = 1 , 22 = 0 , or elseby differentiating independently ,
    we obtain
    1703 x (xy) = $ yc + 02
    1704 Pix (xy) = $ 2,92 + 20 yyx + $ 2x + 4Y2x.
    1705 ir (xy) = $ 3yy? + 302392 + 3 $ y2r9z + $ 82
    + 3 (62,9x + Pry) Y2x + $ , Yarı
    1706 In these formulæ the notation Qc is used where the differentiation
    is partial, while , ( x , y ) is used to denote the complete derivative of g (x , y ).
    with respect to x . Each successive partial derivative of the function (y , z )
    (1700 ) is itself treated as a function of y and z , and differentiated as such by
    formula (1420 ) .
    Thus, the differentiation of the product dyYzin (1700 ) produces
    (Qy) <yz + 0 , 427 = (P24 Yz + Py: 7 .) Y : + PyY2r.
    The function o , involves y and z by implication . If it should not in fact
    contain z , for instance, then the partial derivative 0 . vanishes. On the other
    hand, Yuz Yuz, & c. are independent of z ; and 2 ., 227, & c. are independent of y.

    DERIVED EQUATIONS.
    1707 If • ( y ) = 0 , its successive derivatives are also zero ,
    and the expansions (1703 - 5 ) are then called the first , second ,
    and third derived equations of the primitive equation q (xy ) = 0 .
    In this case, those equations give, by eliminating Yx,
    1708 dy = dy = 24.,4.6.– $: 03 – 0. 02.
    dr "

    1710 Similarly , by eliminating Yo and Yzx, equation (1705)


    would give y3x in terms of the partial derivatives of $ (xy) .
    See the note following ( 1732).

    1711 If $ (xy) = 0 and one = 0; var dir

    Ty

    1712 and dy = 3$226 Parody


    PROOF. - By ( 1708 ), 0 , = 0. Therefore (1704) and (1705) give these
    values of yze and Yas.
    294 DIFFERENTIAL CALCULUS.

    1713 If & and , both vanish , y , in (1708 ) is indeterminate .


    In this case it has two values given by the second derived
    equation (1704 ) , which becomes a quadratic in yz.
    1714 If 02. , prv, and 0.2, also vanish , proceed to the third
    derived equation (1705 ), which now becomes a cubic in yr,
    giving three values, and so on .
    1715 Generally, when all the partial derivatives of $ ( x , y ) of
    orders less than n vanish for certain values x = a , y = b, we
    have , by (1512), q (a , b ) being zero,
    © (a + h ,b + 2 ) = (hd. + kd,)" (xy),a;8 + terms of
    higher orders which may be neglected in the limit. ( x , y are
    here put = a , b after differentiation .) Now , with the notation
    of 1406), hoc,a + kpuno = 0 ;
    therefore e

    h
    = _ºy.70, = dy ;
    the values ofwhich are therefore given by the equation
    1716 (hd + kd,)" $ (x,y),a,0 = 0.
    1717 If y, becomes indeterminate through x and y vanish
    ing, observe that = % in this case, and that the value of
    the latter fraction may often be more readily determined by
    algebraic methods.
    If x and y in the function + ( x, y ) are connected by the
    equation 4 (x , y ) = 0 , y is thereby made an implicit function
    of x , and we have
    1718 $ ( x , y ) = 0.14 , - 4xby

    1719 $21(x ,y) = {(4 ,4, 4,9% )* + ($ , 42%,) yang


    - 2 (Øyxy, - ¥yxby) 4x4,} = V .
    PROOF. — (i.) Differentiate both o and y for a , by (1703), and eliminate yg
    ( ii.) Differentiate also , by (1704 ), and eliminate y, and yes.

    If u , y , z are functions of a , then , as in (1700) ,


    1720 $ e (uyz ) = $uur + by + ,7x.
    · IMPLICIT FUNCTIONS. 295

    1721 $20 (uyz) = $ zuu? + $ 249 % + $ 2cm


    + 24 + 20zum ,uz + 20 uyU.xyz
    + 44 U2x + 4yY2x + 0 ,72x •
    1722 To obtain px (xyz) and $ 20 (xyz), make u = æ in the
    above equations.

    Let U = $ ( x , y , z , $ ) be a function of four variables con


    nected by three equations u = 0 , v = 0 , w = 0, so that one of
    the variables, & , may be considered independent.
    1723 Wehave, by differentiating for },
    $ex: + $, $ $ + $ 22x+ $ _ U _ = 0 dU _ d (buvw ) 1 .
    U ,Xx+ u, y£ + u ,2x + u = 01. d d (xyzť)
    Ve X : + v, yx + v , % + ve where J - d (uvw )
    W . X + w ,ya + w . m + °d(ass)
    1724 dx = _ d (uvw ) 1. di d (uvw ) 1 .
    (&yz ) T d (xĖz ) Ti

    Observe that U stands for the complete and pt for the


    partial derivative of the function U .
    Proof. — (i.) U is found by taking the eliminant of the four equations,
    separating the determinant into two termsby means of the element 0 : - Ug,
    and employing the notation in ( 1600).
    (ii.) mg, yự, and zų are found by solving the last three of the same equa
    tions,by (582).

    IMPLICIT FUNCTIONS OF TWO INDEPENDENT VARIABLES.


    1725 If the equation • ( x , y , z ) = 0 alone be given , y may
    be considered an implicit function of a and 2 . Since x and %
    are independent, we may make z constant and differentiate
    for x ; thus, for a variation in a only , the equations (1703 -5 )
    are produced again with • (@ , y, z) in the place of q (x , y ).
    1726 If x be made constant, z must replace x in those equa
    tions as the independent variable.
    Again , by differentiating the equation $ (xyz) = 0 first for
    4, making z Cconstant,
    C
    and the result for % , making a constant,
    296 DIFFERENTIAL CALCULUS.

    we obtain
    1727 $ xz + yxy: + øyzyz + $ 299xy. + $ ,Yz: = 0 .
    From this and the values of Yx and yz, by (1708),
    1728 = 0 , 0, 02 + 0, 0, 0 . - 07: 0 ;, - 0 -02

    1729 If x , y , z in the function p (x , y , z) be connected by the


    relation + (x , y, ) = 0 , y may be taken as a function of two
    independent variables x and %. We may therefore make z
    constant, and the values of pc ( x , y , z ) and 02x ( x, y ,z ) are
    identical with those in (1718, ’19) if x , y, z be substituted for
    X , y in each function .
    1730 By changing « into the same formulæ give the
    values of 0 : (x , y , z ) and 022 (.C, y , z ) .
    1731 On the samehypothesis , if the value of $ ( x, y , z) , in
    forming which z has been made constant, be now differen
    tiated for z while x is made constant, each partial derivative
    Pro ty, & c . in (1718 ) must be differentiated as containing x , y ,
    and % , of which three variables x is now constant and y is a
    function of z .
    The result is
    1732 pxz (x ,y ,z ) = {(frzy, - 4.x264 - ($ yx4, - 44: $y)4243
    - ($ y24.0 - \ y $ y)4240 + ($2,4 , - 42, 6 , 424:4 , } = 4 %.
    In a particular instance it is generally easier to apply such rules for
    differentiating directly to the example proposed, than to deduce the result in
    a functional form for the purpose of substituting in it the values of the partial
    derivatives.
    1734 EXAMPLE. — Let • (x , y ,z) = lx + my + nz and * (x , y, z ) = 2 ? + y + z?
    = 1, x and z being the independentvariables ; P . (x , y, z ) aud 0 + ( x , y , z ) are
    required . Differentiating o , considering z constant,
    9.(*,y,z) = l+move = 1–m ; since late -
    d .C -
    Pzz (x , y , z ) = -my - Xyz = my? + .za .
    y
    a result which is otherwise obtained from formula ( 1719 ) by substituting the
    values 0 . = l, 0, = m , 9 : = n ; 022 = P2y = 023 = 0 ;
    te = 2x, y , = 2y , it . = 2z ; ze = 42 = 42 = 2.
    Again , to find 92: (x , y , z ), differentiate for z , considering a constant in
    the function
    de = 1–m ; thus het ons = +mount = -met since it - -
    IMPLICIT FUNCTIONS, 297

    1735 Let U = ° (@ , y , z , & , n ) be a function of five variables


    connected by three equations u = 0 , v = 0 , w = 0) ; so that two
    of the variables , n may be considered independent. Making
    n constant, the equations in (1723 ), and the values obtained
    for Uy, X , Y , Zg, hold good in the present case for the varia
    tions due to a variation in Ę , observing that $ , u , v , w now
    stand for functions of n as well as of Ę .
    1736 The corresponding values of Un, X ,, Yng zin are obtained
    by changing & into n .

    IMPLICIT FUNCTIONS OF n INDEPENDENT VARIABLES .


    1737 The samemethod is applicable to the general case of
    a function of n variables connected by r equations u = 0 , v = 0 ,
    w = 0) . .. & c .
    The equations constitute any n — 70 of the variables we
    please, independent : let these be g , n , 8 .. .. The remaining ro
    variables will be dependent : let these be x , y , % . .. ; and let
    the function be U = ° (x , y, % ... ?, n , % ...) .
    For a variation in & only, there will be the derivative of the
    function U , and r derived equations as under.
    1738 $ + buyg+ 0 7 + ... + pe = Uf,
    Uz Xx + u , yz + u mg + ... +
    Vnxx + v, yx + v . m + ... + ve = 0 ,
    . .. ... ... ... & c .,
    involving the r implicit functions Xy, yx,zę, & c . The solution
    of the r equations, as in (1724 ), gives
    1920 dx _ _ d (uvw ...) d (uvw ...) 1 . & c .
    1739 df = - ] ($yz ...) ?
    1740 where J = d (uvw ...) Also dU _ d (puvw ) 1 .
    Te v = d (xyz ...) Also aš = 2 (xyz6)
    The last value being found exactly as in (1723).
    1741 With & replaced by n we have in like manner the
    values of Xn, Y . , Zn, U , ; and similarly with each of the inde
    pendent variables in turn .

    1742 If there be n variables and but one equation


    9 (X , , % ...) = 0, therewillbe n - 1 independent and onedepen
    2 Q
    298 DIFFERENTIAL CALCULUS.

    dent variable . Let y be dependent. Then for a variation in


    & only of the remaining variables, the equations (1703 -5 )
    apply to the present case , º standing for o (x , y , % .. .) . If x
    be replaced by each of the remaining independents in turn ,
    there will be, in all, n - 1 sets of derived equations.

    CHANGE OF THE INDEPENDENT VARIABLE.

    If y be any function of x , and if the independent variable


    & be changed to t, and if t be afterwards put equal to y , the
    following formulæ of substitution are obtained , in which pays:

    1760
    =
    Differentiating these fractions, we get
    1762 Brip = - = PP,
    1765 mn = *,{94® –259.}=R14( – 239.}.
    1766 = 3213,5 *v*sy = pp; + 1°92,
    so; Ex. — If x = r cos 0 and y = r sin 0 ; then
    ly r , sin A + + cos 0 dy p ? + 2ri - rra
    1768
    1700 dior , cos 0 - r sin 8 : dx = (rocos 6 7 sin 03.
    PO
    Proof. — Writing for t in (1760) and (1762), we have to find 2 , yo,
    X 209 Yze ; thus,
    2 . = r , cos 0 r sin 0 ; 224 = r. ,cos 0 – 2r, sin 0 - r cos 0 ;
    Yo = r, sin 0 + r cos 0 ; 920 = rse sin 0 + 2r, cos 0 - r siu 0 .
    Substituting these values , the above results are obtained.

    of To change the variable from æ to t in (a + 6x)" Ynz, where


    ( a + 1x) = e , employ the formula
    1770 (a + bx)" Ynz = b (d ,- n - 1) (d, — n — 2) ... (d, - 1) ye.
    in which , multiplication by deby the index law signifies the
    repetition of the operation de (1492 ) .
    CHANGE OF THE INDEPENDENT VARIABLE. 299

    Proof. – di {( a + bx )" y. } = {n ( a + b.x )*- byn + (a + bx)" Yin +1jz } & c.


    Now bx, = e = a + bt. Substitute this, and denote (u + bx )" yne by U. ;
    therefore de (Um) = nU. + Une or U, . = b (1. - n ) Un.
    Therefore U . = U (d — ~ — 1) Un- and Un-1 = (d . — 1 — 2 ) Un-2, & c.,
    and finally U , = b (do- 1 ) U ,
    But U , = (a + bx ) ye = bze ' = bye.
    Therefore Un = " (d . - 1 — 1) (d . — 1 — 2 ) ... (d . - 1) Ye.
    1771 Cor. - (a + x )" Ynx and w " ynx are transformed by the
    same formula by putting b = 1.

    1772 Let V = F ( 1, y), where x , y are connected with g, n


    by the equations u = 0 , v = 0 . It is required to change the
    independent variables x , y to g, n in the functions V , and V, .
    1773 RULE . — To find the value of V. - Differentiate V , u , v ,
    cach with respect to x , considering Ě, n functions of the inde
    pendent variables x , y ; and form the eliminant of the resulting
    equations ; thus,
    1774 VE + V.,7. - V, = 0 ) VE V, - V. ||
    up& c + u ,netu, = 0 % ; :: Up Un Uz
    V štu,netvi = ) | vg Vn Vr |
    Similarly , to find V ,.
    1775 Ex. - Let x = r cos 0 and y = r sin 6 ; then
    1, = V, cos 0 – V, sin 6, V, = V, sin 6 + v. cos 6.
    PROOF. — uzrcos 0 - x, v = rsin 0 — y, I V. V. - V 1
    and the determinant in (1774 ) takes the form cos 0 - r sin 0 - 1 = 0.
    annexed by writing r and 0 instead of & and n . sin o p cos 0 0
    A similar determinaut gives V ,
    To find V2 , substitute V , cos 0 — V .Sio in the pl. ce of V in the value of
    V, ; and in differentiating for r and 0, consider V , and V , as functions of
    both r and 1 . Similarly , to find V.,, and Voy. Thus,
    1777 V» = V.,cos*v + (1 V.- V.) 2sin 960369+ v,siya0 + v, sin*e.
    1778 V,,= V,,sin’0 -CV.- v.) 2sin Q cos@ +v,60$*® + v..cos*®.
    By addition these equations give .
    1779 Vet V2, = V2 20 .
    01035)
    300 DIFFERENTIAL CALCULUS.

    1780 Given
    2. U. 2 :
    V = f (x , y , z ) and ?, n , Ś known functions of
    V . V . V , are expressed in termo
    , V , are expressed in terms of V , V , V , by
    the formulæ
    dV _ d (Vm ) . DV _ d (ĚVÝ) . J. DV – (SMV ) . J .
    d d ( xyz) , dnd( xyz) º dd(xyz)
    PROOF. — Differentiate V as a function of ( VES + Vnit Vibe = Vz,
    &, n , Ś with respect to independent variables )
    x , y, z. The annexed equations are the result . 3 V45,
    & + V , , + VS = Vys
    Solve these by (582)with the notation of (1600). ( 15. + V n + V ;5. = V ..

    1781 Given V = f (x , y , z ) , where x , y , z are involved with


    E , n , & in three equations u = 0 , v = 0 , w = 0 , it is required
    to change the variables to g, n , % in Vr, V . , and V ,.
    Applying Rule ( 1773) to the case of three variables,we have
    VŠx + V , & + V ;50 - V = 0 , VE V , V , - V21
    Up &c + u ,net uz tur = 0 / . Ug Un uz uxl _
    = 0.
    V¢ &x + v, N.+ + vs Ex + vr = 0 % vş vn Vs Ve
    WEŠx + w ,nc + wSc + wr = 0 wę w , ws wil
    The determinant gives V , in termsof V , V , V ,and the deri
    vatives of u , v , w . V , and V . are found in an analogousmanner .
    1782 Similarly with n equations between 2n variables.
    1783 Ex. — Given
    x = r sin 0 cos ; y = r sin 8 sin ; = r cos 0.
    The equations u , v, w become •
    r sin 0 cos0 - x = 0 ; r sin 0 sin p - y = 0 ; r cos 0 – = 0.
    Writing r, 0, $ instead of !, n , 5 , the determinant becomes
    V, V. V - V.
    sin 0 cos o rcos 0 coso - r sin 0 sin o -1
    sin o sin q r cos 0 sin o p sin 0 cos 0 0
    cos 0 - psin 6 0
    From which V , is obtained . Similarly , V , and V , ; and, by an exactly similar
    process, the converse forms for V , V ., and Va. The results are
    * In writing out a determinant like the above, it will be found expeditions
    in practice to have the columns written on separate slips of paper in order to
    be able to transpose them readily . Thus, to find the coefficient of V ., bring
    the second column to the left side, and , since this changes the sign of the
    determinant, transpose any two other columns, so that the coefficient of V .
    may be read off in the standard form as the minor of the first element of the
    determinant.
    CHANGE OF THE INDEPENDENT VARIABLE. 5301

    1784 V , = V, sin 6 cos + v .cos 8 cos - V. sing .


    1785 V , = V, sin 6 sin + .cos 6sin$ + v .Com
    1786 V, = V,cos 6 – V sino
    1787 V, = Vesin 0 cos0 + V , sin 0 sin + V . cos0 .
    1788 V. = Ver cos 0 cos 0 + V , r cos 0 sin ø – Vor sin 0.
    1789 V. = - V r sin 0 sin 0 + V, r sin 0 cos p .
    1790 To find V , directly ; solve the equations u , v , w , in ( 1783), for r, 0,
    and q ; the solution in this case being practicable ; thus,
    p = / (xº + y + z ), 0 = tan - V (271 ) , p = tan -1 Y .
    Find T2, 0 %, 0 , from these , and substitute in Ve = V ,% + V , 0 , + VoPr. Simi.
    larly, V , and V ,. Also V , = VzX + V ,Yr + V .7 . Similarly, V , and Ve.
    1791 To obtain View , substitute the value of V , in the place of V, in the
    value of Vy, in (1785), and, in differentiating Vư, V ., V ., consider each of
    these quantities a function of r, 0, and 0 .

    To change the variables to r , 0, and 4 , in V28 + V2, + V25,


    the equations (1783) still subsisting . Result
    1792 V2 + 12, + V2:
    = Vo + V + (V,cot 6 + Vx + Voscosec°6).
    Proof. — Put r sin 0 = P, so that a = f coso and y = p sin o ,
    therefore,by (1779), Vx + V,, = V2 + 1 V.+ Vzp.................(i.).
    P

    Also, since r = r cos 8 and p = r sin 0, we have, by the same formula,


    Vu + V2 = Vx + 10 + V.......................(ii.).
    Add together (i.) and (ii.), and eliminate Vo, by (1776), which gives
    V. = V,sin0 + V.COSO
    If V be a function of n variables x , y, ... connected by
    the single relation , a? + y* + x + ... = r...................... (i.).
    1793 V + V , + V2 + & c. = V2 + " < V,
    302 DIFFERENTIAL CALCULUS.

    Proof.— V. = V.1. = V. , since r = , by differentiating (i.),


    therefore V, = . + ,* * = V. + V (1 - ).
    Similarly V = v * +0.61 - ),do.
    Thus, by addition ,
    Vu +V, +&o.= v.,* +/ + - + v.( –2*+ * +--) = Vrt"= 1V,
    LINEAR TRANSFORMATION .
    1794 If y = f (x , y, z), and if the equations u , v, w in
    (1781) take the forms
    x = a ;$ + bin + c5) ( 4 $ = AqX + A , y + Ază ,
    y = a , $ + b3n + c2$ , then 3 An = Bx + By+ B . ,
    x = az $ + 639 + c35 ) Cas = CX + C , y + Cző ,
    by (582), A being the determinant (a,b2c3), and 4, the minor
    of an, & c .
    1795 The operations de, d , , d , will now be transformed by
    the first set of equations below ; and dy, din, d , by the second
    set.
    d , = (A dz+ B ,d , + Cdz) = A ) dž = ad. + azd , + azd ,
    d , = ( A dş + B ,d , + C ,d ;) ; A { , d , = b ,de + b , d , + bzd . { .
    de = (Azdę + B ;d , + Cd ) A do = cd, cd , + cd. '
    Proof. — By d , = d . + d , + de5o and dě = d .x8 + dyYg + d,2g; and the
    values ofEn, , & c., from the preceding equations.

    1797 From ( 1795), Vi = a,Va+ a,V, + azVe Operating


    again upon Vš, we have
    Ves = (and: + a» , + azd .) Vi = a (Vg): +az(Vg), +az(V :) ,
    and by substituting the value of Vų, and similarly, with Van,
    V25, we obtain the formulæ ,
    1798
    V2 = a; Veta V2, + a V2 + 2a,a ,Vya + 2aza , V .x + 2aja, Vzy )
    V20 = 1; 12x + ; V2, + 6 V2: + 26 ,63 Vyz + 2636 , Vzx + 26,6 , Vxy { .
    Vx = c Vetc; V. + Vez + 2c,c Vyc + 2c3c, Vex + 2cc, Vxy !
    CHANGE OF THE INDEPENDENT VARIABLE. 303

    ORTHOGONAL TRANSFORMATION.
    1799 If the transformation is orthogonal (584), we have
    X2 + 42 + 2 = $ + P + 6 ;
    and since, by (582, 584 ), A = 1 , A , = 41, & c . ; equations
    (1794 ) now become
    1800 x = , + bin + c5 š = ax + ay + azz )
    y = a, & + b .n + c25 % , n = bx + b ;y + bză .
    % = az$ + 637 + 638) = x + c2y + cz3 )
    And equations (1795) become
    1802 do = a d + b ,d , + cd, d = a d , tad, + azd,
    d , = azdę+ bąd + cado , d , = b,d , + b ,d , + bzd .
    d , = azdę+ b3d , + czdz) de = d . + c, d , +cd. )
    The double relations between x , y , z and g, n , % , in the six
    equations of (1800 - 1 ) , and the similar relations in (1802 - 3 )
    between ded,,d , and dd, dy, are indicated by a single diagram
    in each case ; thus,
    1804 Ś m Š
    X a , b , ci de la, b , a
    y a , b , c, d , a, 62 Cz
    z l az b₂ C₃ d, az bz cz
    1806 Hence, when the transformation is orthogonal, the
    quantities x , y , z are cogredient with dx, d , , dz, by the defini.

    1807 Extending the definition in ( 1629) , it follows that any


    function u = (x , y, ), when orthogonally transformed , has,
    for a covariant, the function « (dr, d ,, d ) u . That is , if by
    the transformation ,
    u = $ (x , y , z ) = ¥ ( $, n , S),
    then also (dz, dy, dz) u = ¥ (dě, dn, dz) u .
    1808 But if u be a quantic, then , as shown in ( 1648 ) ,
    ♡ ( d.,, d , , d ,) u is always a function of the coefficients only of u ,
    and the covariant is , in this case, an invariant.
    1809 Ex. — Let u or 9 (x, y, z) = axº + by* + cz* + 2fyz + 2gzx + 2hæy, .
    = 2 {a ? + 2° + c? + 262 + 2gº + 2K* }, and this is an invariant of u .
    304 DIFFERENTIAL CALCULUS.

    1810 When V = f(x,y ,z).is orthogonally transformed ,


    V2 + V2, + V2 = V2 + V2n + V25.
    Proof. — By adding together equations (1798 ), and by the relations
    ai + b3 + ci = 1, & c., and a , Ag + b, bę + GC3 = 0, & c.,
    established in (584 ).

    1811 If two functions u , v be subjected to the same ortho


    gonal transformation , so that
    u = $ (x ,y , z ) = ($, , ) and v = v (x ,y ,z ) = * ($, n , S) ;
    then $ (dx, d ,, d .) v = 0 (dị, d ,, dz) v .
    1812 Ex. - Let u = ax + bya + cze + 2fyz + 2gzx + 2hxy = 0
    = a' ? + b'na + c'59 + 2f"n5 + 29'55 + 2H'En = 0 ,
    and let v = x + y + z = + 1° + 5 = 4 and 4 .
    Then (d ,dy d .) v = avuz + bv2y + cV,, + 2fvya + 290:- + 2hVry ,
    and « (de,d.,dz) v = a'vae + b'van tevas + 2f'uns + 2g'vsx + 2h'ven
    But 02z = 2, and 1,2 = 0, & c. Hence the theorem gives a + b + c
    = a + b ' + ' ; in other words, a + b + c is an invariant.

    1813 Contragredient. — When the transformation is not


    orthogonal, ( 1795 ) shows that do is not transformed by the
    same, but by a reciprocal substitution , in which ai, bu, ci are
    replaced by the corresponding minors A1, B , C . In this case
    de, d ,, d , are said to be contragredient to x , y ,z .

    1814 Contravariant. - If , in (1629) , the quantics are sub


    jected to a reciprocal transformation instead of the same, we
    obtain the definition of a contravariant.

    1815 When z is a function of two independent variables a


    and y , the following notation is often used :

    dže= p, dig= % te meer,


    elementi del mon = s, depending = t.
    Let Q (x , y , z ) = 0 . It is required to change the inde
    pendent variables from x , y to z , y . The formulæ of trans
    MAXIMA AND MINIMA . 305

    formation are

    1816 ene ro -- dex


    -
    1819 spq - tp - qr - ps
    1819 deg = 25p9 Tipsarq'; dydž po
    PROOF. — Formulæ (1761, 1763) give , and X22, because, since y remains
    constant, o may be considered a function of only two variables, x and % .
    Formulæ (1708 - 9 ) give x, and X , in terms of partial derivatives of ,
    since z is now constant,and o may be taken as a function of the two variables
    1 and y .
    But ¢ (x, y, z) = 0 is equivalent to y (x , y ) — % = 0 ; and the partial
    derivatives of o with respect to ac and y are the same as those of y ; and
    therefore the same as those of z when x and y are the independent variables.
    Hence z may be written for g in the formulæ .
    Lastly, dydz ļ - plel2 ) = _ I ) d – 9Px - Pqx 1 = qr - ps
    pla dz psp po
    The independent variable is here changed from z to x , without reference
    to the equation ý = 0 ; and this is allowable, because y is constant for the
    time being in either case .

    MAXIMA AND MINIMA.


    Maxima and Minima values of a function of one
    independent variable.
    1830 DEFINITION . — A function ® (v ) has a maximum value
    when some value x = a makes ¢ (x ) greater than it is made
    by any value of a indefinitely near to a. Similarly for a
    minimum value, reading less for greater .
    1831 ILLUSTRATION. — If the
    ordinate y in the figure be al
    af-

    ways drawn = f (a ), it has


    maximum values at A , C, E ,
    and minimum at B and D
    (1403 ).
    NOTE . — For the algebraic
    determination of maxima and
    minima values by a quadratic
    equation, see (58 ).

    1832 RULE I. - A function $ ( x ) is a maximum or minimum


    when & ' ( x ) vanishes, and changes its sign as x increases from
    plus to minus or from minus to plus respectively.
    2 R
    306 DIFFERENTIAL CALCULUS.

    1833 RULE II .- -- Otherwise 0 ( x ) is a marimum or minimum


    when an odd number of consecutive derivatives of $ (x ) vanish ,
    and the next is minus or plus respectively .
    Proof. — (i.) The tangent to the curve in the last figure becomes parallel
    to the x axis at the points A , B , C , D , E as x increases ; therefore, by (1103 ),
    tan 0, which is equal to f ( v), vanishes at those points, while its sign changes
    in the manner described .
    (ii.) Let f " (e) be the first derivative of f (x ) which does not vanish when
    x = a , n being even ; therefore, by (1500), f (a + h) = f (a ) + f" (a + th ).
    The last term retains the sign of f" (a ), when h is small enough, whether h
    be positive or negative, since n is even . Therefore f (x ) diminishes for any
    small variation of a from the value a if f " (a ) be negative, but increases if
    f" (a ) be positive. Hence the rule .

    1834 NOTE. — Before applying the rule for discovering a


    maximum or minimum , we may evidently
    (i.) reject any constant factor of the function ;
    (ii.) raise it to any constant power, paying attention to sign ;
    (iii.) take its reciprocal ; maximum becoming minimum , and
    vice versa ;
    (iv.) take the logarithm of a positive function .
    1835 Ex. 1. — Let ( ) = x? — 7x * — 35x + 1,
    therefore ¢'(x ) = 7x9 — 28.03 — 35 = 7 (23 — 5) (23 + 1).
    Also º" (:-) = 7 (625 — 12x²). Therefore x = 75 makes '(x ) vanish, and
    0 " ( x ) positive ; and therefore makes o (a ) a minimum .

    1836 Ex. 2 . — Let $ ( x) = (x − 3)** (x - 2)" . Here


    $ (2 ) = 14 (-x – 3)13 (2 — 2)" + 11(2 - 3) *(x - 2)10 = (x — 3)13(2 — 2)10 (254 - 61),
    and we know , by (444 ) or by (1460 ), that, when x = 3 , the first thirteen
    derivatives of ¢ ( v ) vanish ; and 13 is an odd number. Therefore o (x ) is
    either a maximum or minimum when x = 3 .
    To determine which, examine the change of sign in o ' (x ) . Now ( 2 - 3 ) 18
    changes from negative to positive as x increases from a value a little less
    than 3 to a value a little greater, while the other factors of ¢' (a ) remain
    positive. Therefore , by the rule, 0 (2) is a minimum when æ = 3.
    Again , as a passes through the value 2, 9 ' ( 2 ) does not change sign , 10
    being even . Therefore x = 2 gives no maximum or minimum value of o (a ).
    Lastly, as x passes through the value of the signs of the three factors
    in ''(x ) change from ( - ) ( + ) ( - ) to ( - ) ( + ) ( + ) ; that is, °'(x) changes
    from + to — ; and, consequently , q ( a ) is a maximum .

    1837 Ex. 3. - Let q (x, y) = ** + 2x’y — yø = 0 . To find limiting values


    of y.
    MAXIMA AND MINIMA . 307

    Here y is given only as an implicit function of x . Differentiating, in


    order to employ formulæ (1708, 1711) ,
    Que = 4 .8 + 4xy, $ 22 = 12.c? + 4y, 0 , = 2x2 – 3y? ;
    Yo = 0 makes 0 , = 0. Solving this equation with Q ( x, y ) = 0, we get
    x = £1, y = - 1 when Y , vanishes .
    And then yw = - 021 = _ 12 – 34 =– 8, positive ; therefore ,when w = + 1,
    0, 2 -
    y has – 1 for a minimum value.
    Similarly , by making y the independent variable, it may be shewn that,
    when y = - , x has both the maximum and minimum values + 6.

    1838 A limiting value of $ (x , y),


    subject to the condition y (x , y ) = 0 ......... (i.),
    is obtained from the equation 0 :4, = ,4: ...... (ii.)
    Simultaneous values of æ and y, found by solving equa
    tions ( i.) and ( ii. ) , correspond to a maximum or minimum
    value of p .
    Proof. — By (1718), o being virtually a function of x only ; and,by ( 1832 ),
    8 .(27) = 0 .
    1839 Ex. — Let Q (x, y) = xy and ų (x, y) = 2x — xy + y = 0......(i.)
    Eqnation (ii.) becomes y ( 3y2 — x ) = x (62? — y ).
    Solving this with (i.), we find y3 = 2.28 and 2 ? ( + — / 2 ).
    Therefore x = 13/ 2 , y = 1 / 4 are values corresponding to a maximum
    value of q . That it is a maximum , and not a minimum , is seen by inspecting
    equation (i.)
    1840 Most geometrical problems can be treated in this way, and the
    alternative of maximum or minimum decided by the nature of the case.
    Otherwise the sign of 0 .x (ay ) may be examined by formula (1719) for the
    criterion , according to the rule .

    Maxima and Minima values of a function of two


    independent variables.
    1841 RULE I. - A function p (x , y ) is a marimum or minimum
    when & and , both vanish and change their siyns from plus to
    minus or from minus to plus respectively, as x and y increase .
    1842 RULE II. - Otherwise, d , and oymust vanish ; 02602 - šv
    must be positive, and 02x or pay must be negative for a maximum
    and positive for a minimum value of p .
    Proof. — By ( 1512), writing A , B , C for $ 20, $ rys 24, we have, for small.
    changes h , k in the values of x and y,
    ♡ (@ th , y + k ) - ° (x , y ) = hqx + køy + } ( Al2 + 2Bhli + Cla) + terms
    which may be neglected, by (1410) .
    308 DIFFERENTIAL CALCULUS.

    Hence, as in the proof of (1833), in order that changing the sign of h or


    k shall not have the effect of changing the sign of the right side of the equa
    tion , the first powers must disappear, therefore 0 , and o , must vanish . The
    next term may be written , by completing the square, in the form
    2Ā Ulat( A tº + B ,) tal
    + AC - Bº }; and, to ensure this quantity retaining its
    sign for all values of the ratio h : k , AC - B * must be positive. © will then
    be a maximum or minimum according as A in the denominator is negative
    or positive.
    It is clear that A and B might have been transposed in the proof. Hence
    B must have the same sign as A .

    1843 A limiting value of $ (x , y, z ),


    subject to the condition y ( x , y , m ) = 0 ......... (i.),
    is obtained from the two equations
    1844 $ 4, = 0,42.. ... (ii.), 6:4 , = ,4 . ...... (ii.) ;
    ten .,
    1846 or, as they may be written : ......
    Home ta ...... (iv.)
    Simultaneous values of x , y , z , found by solving equations
    (i., ii., iii.) , correspond to a maximum or minimum value of p .
    PROOF .— By ( 1841), o being considered a function of two independent
    variables x and z, and , by (1729, 1730),
    Qrlæ , y , z ) = 0 gives (ii.) , and 0 :(x , y , z ) = 0 gives (i .)
    The criterion of maximum or minimum in (1842) may also be applied
    without eliminating y by employing the values of 02 and 02. in ( 1719, '30 ) .

    1847 Ex. — Let $ (x , , z ) = x + + y + 22


    and { (x, y,z ) = ax* + by® + cz° + 2fyz + 2gzæ + 2hxy - 1 = 0..(i.)
    Equations (ii.) and (iii.) here become
    Ep , say, (iv.)
    ax + hy + gz hx + by + fz gæ + fy + cz
    Therefore, by proportion (70) and by (i.),x + y +z = R-1 = p.
    From equations (iv .) we have
    1848 ax + hy + g2 = Rx ) la -Rh gi
    hæ + by + fz = Ry { ; : 1 h 1- R f = 0,
    gæ + fy + cz = Rz ) |g c-R
    1849 or ( R - a) (R - 1) (R — c) + 2fgh
    - (R — a )82 – (R — b) g – (R — c)k = 0, or (see 1641)
    R3 – R (a + b + c) + R (bc + ca + ab - fa - ge — hº) — A = 0 ,
    MAXIMA AND MINIMA. . 309

    This cubic in R is the eliminant of the three equations in


    x, y, z . It is called the discriminating cubic of the quadric (i.),
    and its roots are the reciprocals of the maxima and minima
    values of ? + yº + z .
    1850 To show that the roots of the discriminating cubic
    are all real.
    Let Ry, R , be the roots of the quadratic equation
    1851 If
    | R - B R f1=
    -c / =
    (R - 6)( R - c) — 4* = 0 .................(v.)
    Ry > b and c, and b and c > Rz.
    Make R = R , in the cubic, and the result is negative, being minus a
    square quantity, by ( v .) . Make R = Ry, and the result is positive. There
    fore the cubic has real roots between each pair of the consecutive values + 00 ,
    Ry, Rz, – 20 ; that is, three real roots . But since the roots are in order of
    magnitude, the first must be a maximum value of R , the third a minimum ,
    and the intermediate root neither a maximum nor a minimum .

    Maxima and Minima values of a function of three


    MU

    or more variables.
    1852 Let • (xyz ) be a function of three variables. Let
    P2r2 0.272 022, Quz, Par , Pry be denoted by a , b , c , f, g , h ; and let
    A , B , C , F , G , H be the corresponding minors of the deter
    minant A , as in (1642) .
    1853 RULE I. - * ( x , y , z ) is a maximum or minimum when
    Pue Qv, $ , all vanish and change their signs from plus to minus
    or from minus to plus respectively, as x , y , and z increase .
    Otherwise
    1854 RULE II. — The first derivatives of $ must vanish ; A
    and its coefficient in the reciprocal determinant of A must be
    positive ; and o will be a maximum or minimum according as
    a is negative or positive. Or, in the place of A and a , read B
    and l or C and c.
    Proof. — Pursuing the method of ( 1842), let , . Ś be small changes in
    the values of x, y , z. By (1514),
    g (x + , y + , z + 5) - 0 ( x, y, z )
    = čp . + nq , + 69: + } (ał? + bnp + cą? + 2f1;5 + 236 + 2hEn ) + & c.
    For constancy of sign on the right, 47, Oy, 0, must vanish . The quadric ,
    may then be re -arranged as under by first completing the square of the terms
    in &, and then collecting the terms in 5, 91, and completing the square. It
    thus becomes S - F %)? BC - F ? ,
    (as thn + 95) 1+ (Cn S0 + C5f:
    310 DIFFERENTIAL CALCULUS.

    Hence, for constancy of sign for all values of £, n , $, it is necessary that o


    and BC - F ? should be positive. This makes B also positive. By symmetry ,
    it is evident that A , B , C , BC - F , CA - Gʻ, AB - HP will all be positive.
    The sign of a in the first factor then determines, as in ( 1842), whether o is a
    maximum or a minimum .

    1855 The condition may be put otherwise. Since


    BC - F ? = aA by (577) , the condition that BC2 – F" must be
    positive is equivalent to the condition that a and A must have
    the same sign . Hence we have also the following rule :
    1856 RULE III. - Pv, ,must vanish ; the second of the four
    determinants below must be positive, and the first and third
    must have the same sign : that sign being negative for a max
    imum and positive for a minimum value of $ ( x, y , z ).
    1857 $2.co 022 .wylo 14.2. Pry dezle | $20 fry fra Pew l
    I dye Py Dye Czy Dyz Pyx + 2y buz Dyw
    1 Pax Pay 02: Pax Day 2: $ w
    | Pwa Pwy Peoz $ 200
    1858 The theorem can be extended in a similar manner to
    $ (X , Y ,Z , W , ... ) , a function of any number of variables. Form
    the successive Hessians of $ (1630) for one , two, three , & c .
    variables in order as shown above ; then
    1859 RULE. — In order that ♡ ( x , y , z , w , . ..) may be a max
    imum or minimum , xv, 9., 0w , & c .must vanish ; the Hessians of
    an even order must be positive ; and those of an odd order must
    have the same sign , that sign being negative for a maximum and
    positive for a minimum value of the function 0 .
    For a demonstration in full, see Williamson's Diff. Calc., 4th Edit., p .433.

    1860 Ex. - Required a limiting value of the function


    u = ax' + by+ + cz? + 2fyz + 29zx + 2hay + 2pæ + 244 + 2rz + d .
    The condition in the rule produces equations (1 ), (2 ), (3 ). Equation ( 1)
    results from Euler's theorem (1621), thus ; introducing a fourth variable w ,
    as in ( 1645) , we have æu : + yu, + zu , + wu, = 2u ,
    which reduces to (4 ) by means of ( 1), (2 ), (3 ), and the value of u , putting
    w = l.
    1861 , = ax + hy + gz + p = 0 ...... (1) ) la h g P1
    qu, = hx + by + fx + 9 = 0 ......(2 ) ( . . 1 hbf a
    to

    quz = gæ + fy + cz tp = 0......(3) ( . g fc r
    10

    px + 1y trz + d = U ... ... (4 ) ) par d - u !


    MAXIMA AND MINIMA. 311

    The determinant is the eliminant of the four eqnations, by (583), and is


    equivalent, by the method of (1724 , Proof. i.), to A ' — Au = 0 , or u = A ' : A
    (Notation of 1646 ) .
    To determine whether this value of u is a maximum or minimum , either
    of the conditions in (1854, '6 ) may be applied ; and since, in this example,
    U2 = 2a, U2, = 26, & c., the letters a , b, c, f, g, h may be considered identical
    with those in the rule .

    1862 To determine a limiting value of $ (x , y , z , .. .) , a func


    tion of m variables connected by n equations un = 0 , 12 = 0 , .. .
    Un = 0 .
    RULE. — Assume n undetermined multipliers 11, 12,... ,
    with the following m equations :
    Petd, (u ), + 12(u , st ... + in (un), = 0 ,
    ... + , (ul), + 1, (uz), + ... + in (uny = 0,
    . .. . .. . .. ... .. . .. .
    making in allmtn equations in mtn quantities, x , y , z, ...
    and 11, 12, ... dn. The values of x , y , 2, ..., found from these
    equations, correspond to a maximum or minimum value of o.
    Prook. — Differentiate ™ and U7, U2, ... Un on the hypothesis that x , y ,%, ...
    are arbitrary functions of an independent variable t. Multiply the resulting
    equations, excepting the first, by 11, 19, .. , in order,and add them to the value
    of or. The coefficients of Xc, Yo, 24, ... may now be equated to zero , since the
    functions of t are arbitrary, producing the equations in the rule.

    1863 Ex. 1. — To find the limiting values of me = æ? + y + x2 with the


    conditions . Ax? + By + Cz2 = 1 and lx +my + nz = 0.
    Here m = 3, n = 2; and, choosing 1 and je for the multipliers, the equa
    tions in the rule becoine
    · 2x + 2 Anx + ul = 0 ......(1) ) Multiply ( 1), (2 ), (3) respectively by
    2y + 2 Bay tum = 0 ......( 2) x , y, z, and add ; thus vi disappears ,
    2z + 2C4z + pn = ...... (3) and we obtain
    x + y + 2° + (Ax® + By® + Cza) 1 = 0, therefore i = - .
    Substitute this in (1), (2), (3 ) ; solve for a', y,z , and substitute their values
    in
    lx + my + nz = 0.
    . 22 m no
    1864 a _1 + 2
    The resultJU 18is 4,2 + cij = 0, a quadratic in ra.
    The roots are the maximum and minimum values of the square of the
    radius vector of a central section of the quadric Av? + Byº + Cz? = 1 made by
    the plane læ + my + nz = 0 .

    1865 Ex. 2 . - To find the maximum value of u = ( + 1 ) (y + 1) ( + 1) ,


    subject to the condition N = a * b c".
    312 DIFFERENTIAL CALCULUS .

    This is equivalent to finding a maximum valua of


    log (x + 1) + log (y + 1) + log (2 + 1),
    subject to the condition log N = x log a + y log b + z log c.
    The equations in the rule become
    1
    i + 1 log a = 0 ; - 1 ta log b = 0 ; z- + 1 + 1 log c = 0.
    By eliminating 1, these are seen to be equivalent to equations (1946).
    Multiplying up and adding the equations,wefind X , and thence x + 1, y + 1 ,
    z + l ; the values of which, substituted in u , give, for its maximum value,
    u = {log (Nabe) }3 = 3 log« log b log ca.
    Compare (374), where a , b, c and x , y ,z are integers.
    Continuous Maxima and Minima .
    1866 If px and Qy, in (1842), have a common factor, so that
    px = P4 ( x , y), 0 , = Q + (x , y ).........(i.) ,
    where P and Q may also be functions of x and y ; then the
    equation 4 (x , y ) = 0 determines a continuous series of values
    of x and y . For all these values p is constant, but, at the same
    time, it may be a maximum or a minimum with respect to any
    other contiguous values of ø , obtained by taking æ and y so
    that + (xy) shall not vanish .
    1867 In this case, 027024 — 2, vanishes with t , so that the
    criterion in Rule II . is not applicable.
    Proof. — Differentiating equation (i.), we have
    02 = P 4 + P4c ? Prv = P, + + P4, ?
    02y = Q , + + Qy , I Pyx = Qx + Q4,
    If from these values we form 022027 - ºry * Pyxe will appear as a factor of
    the expression.
    1868 Ex. -- Take z as q (xy ) in the equation
    z? = a’ – b* + 26 / (x2+ 4*) —~* — yº.................... (i.),
    : * . = a(vety) -1) and us;= u (v cene ty5 –2).
    The common factor equated to zero gives a? + y' = 1 , and therefore z = + a ...(i .)
    Here a is a continuous maximum value of z , and - a a continuous minimum .
    Eqnation (i.) represents, in Coordinate Geometry, the surface of an anchor
    ring, the generating circle of radius a having its centre at a distance b from
    the axis of revolution Z . Equations (ii.) give the loci of the highest and
    lowest points of the surface.

    For the application of the Differential Calculus to the


    Theory of Curves, see the Sections on Coordinate Geometry.
    INTEGRAL CALCULUS.

    INTRODUCTION .

    1900 The operations of differentiation and integration are


    the converse of each other. Let f (x ) be the derivative of
    Q (x ); then (a ) is called the integral of f (x ) with respect to x .
    These converse relations are expressed in the notations of the
    Differential and Integral Calculus, by

    dp(x ) = f(x) and by \ f(x )dx = $ (x).


    1901 THEOREM . — Let the curve y = f (r) be drawn as in
    ( 1403 ), and any ordinates Ll, Mm , and let OL = a , OM = b ;
    then the area LMml = º (b ) - ° (a ).
    PROOF. — Let ON be any value of x , and PN m
    the corresponding value of y , and let the area
    ONPQ = A ; then A is some function of æ . Also,
    if NN' = dx, the elemental area NN 'PP = dA
    = ydæ in the limit; therefore A = y. Thus A
    is that function of x whose derivative for each
    value of « is y or f (x ) ; therefore d = 0 (2 ) + C , O NN M
    where C is any constant. Consequently the area
    LMml = Q (6 ) - ° ( a ), whatever O may be.
    The demonstration assumes that there is only one function (2 ) corres
    ponding to a given derivative f (x ). This may be formally proved .
    If possible , let y (a ) have the same derivative as o (a ) ; then, with the
    same coordinate axes, two curves may be drawn so that the areas defined as
    åbove, like LMml, shall be y (x ) and y (x ) respectively , each area vanishing
    with x . If these curves do not coincide, then , for a given value of x , they
    have different ordinates, that is, ' ' ( ) and ' ( ) are different, contrary to the
    bypothesis . The curves must therefore coincide, that is, o (s ) and (ə ) are
    identical.
    2 s
    314 INTEGRAL CALCULUS.

    1902 Since ( 1) - ( a ) is the sum of all the elemental areas


    like NN ' P ' P included between Ll and Mm , that is , the sum of
    the elements ydx or f ( x ) dx taken for all values of x between
    a and b , this result is written
    f ( x )dx = $ (6 ) - ° (a ).
    1903 The expression on the left is termed a definite integral
    because the limits a , b of the integration are assigned .*
    When the limits are not assigned , the integral is called inde
    finite.

    1904 By taking the constant C = 0 in ( 1901), we have the


    area ONPQ = $ (x) = \ f(x)dx.
    NOTE. — In practice , the constant should always be added to
    the result of an integration when no limits are assigned .

    MULTIPLE INTEGRALS.
    1905 Let f(x,y,z) be a function of three variables ; then
    x : Cy2 C22
    the notation " L" f(x,y,z)dx dy dz
    dxi uyi o 'zi

    is used to denote the following operations.


    Integrate the function for z between the limits z = 21, 2 = 22,
    considering the remaining variables x and y constant. Then ,
    whether the limits 2 , 7 , are constants or functions of x and y ,
    the result will be a function of x and y only . Next, consider
    ing x constant, integrate this function for y between the limits
    y , and Y2, which may either be constants or functions of x .
    The result will now be a function of x only. Lastly , integrate
    this function for a between the limits x , and xz.
    Similarly for a function of any number of variables.

    1906 The clearest view of the nature of a multiple integral


    is afforded by the geometrical interpretation of a triple in
    tegral.

    * The integralmay be read “ Sum a to b, S (x ) dx ” ; ſ signifying " sum .”


    INTRODUCTION . 315

    Taking rectangular coordinate axes , let the surface


    z = ° (x , y ) (XYom in the figure) be drawn, intersected by the
    cylindrical surface y = . 4 ( ) (RMNnm ), and by the plane
    X = Xı (LSml) . The volume of the solid OLMNolmn bounded
    by these surfaces and the coordinate planes will be
    (x ) prip* (2) * (x,y)
    Jo Jo
    $ (x ,y) dx dy = Jo1" Jo| Jo
    dx dy dz.
    PROOF.— Since the volume cut off by any plane parallel to OYZ, and at a
    distance x from it, varies continuously with a, it must be some function of x .
    Let V be this volume, and let d V be the small change in its value due to a
    change dæ in 2. Then, in the limit, dy = PQqpx da , an element of the solid
    shown by dotted lines in the figure. Therefore
    dV _ pery
    a = PQgp = (2) Q (x , y ) dy, by (1902 ),
    dæ
    2 being constant throughout the integration for y . The result will be a
    function of x only . Making a then vary from 0 to x , we have, for the whole
    pancy(x ) pasi (x ) = (x , y )
    volume, | 0
    volume, S {$ * «, )dy}ds= ["{ (S*"dz]ay}da,
    Jo ( do LJO
    since ( x , y ) = % . With the notation explained in (1905) , the brackets are
    not required , and the integrals are written as above.

    1907 If the solid is bounded by two surfaces 2, = 0, (x , y ), 2, = 02 (x , y ),


    two cylindrical surfaces Y = 4 . ( w), yo = 4 : (x ), and two planes x = x1, x = x ,,
    the volume will then be arrived at by taking the difference of two similar
    integrals at each integration , and will be expressed by the integral in (1905).
    If any limit is a constant, the corresponding boundary of the solid
    becomes a plane.
    316 INTEGRAL CALCULUS.

    METHODS OF INTEGRATION .
    INTEGRATION BY SUBSTITUTION.
    1908 The formula i ( x ) dx =

    where z is equal to f (x ), some function chosen so as to facili


    tate the integration .
    Rule 1. — Put x in terms of z in the given function , and
    multiply the function also by x ; then integrate for z .
    If the limits of the proposed integral are given by x = a ,
    x = b , these must be converted into limits of z by the equation
    x = f - (2 ) .

    The following rule presents another view of the method of


    substitution , and is useful in practice .
    1909 Rule II.- If Q(x ) can be expressed in the form F(z)2 ;
    then 56 (x)dx = F (z)xdx = ( F (z)dx.
    dx
    Ex. 1. — To integrate - Substitute % = x + 7 (zº + a ') ;
    ✓ (x ? + u ? ) :

    therefore ante = 1 + Mitteis = 2+


    dic y) = metais
    Saten da Se = logu = log (x + v(&*+a%)}.
    Ex. 2 . 5 + 2.08 . 5. -6 + 2.7 - 3
    x + an J 2 - 3 + c ^2
    dx = - log (2 -6 + a - ) .

    Here • =***+ 7-, F(x) = - , * = - (52-6+ 22-2).


    ( 1 - cą ? - dx [ x -2 _ c da
    Ex. 3 .
    J1 + ca ? ( 1 + ax ? + c*r *) ] -' + cx ✓ (c* ? + x -2 + a )
    d . ( - 1 + cr ) de
    + cx ) V { (2 - ' + cx )' + a - 2c}
    Joox
    lo g æv (a – 2c) +1 ✓ cixa
    ( 1 + ax ? + c>z )
    = V _(a -1 _2c) log
    - (2c1 - a ) Cos - V (20
    or v (2c - a ) 1 + ca
    By (1927) or (1926 ). Here « = 2 -' tcx .
    METHODS OF INTEGRATION . 317

    In Examples (2 ) and ( 3 ) the process is analytical, and leads to the dis


    covery of the particular function z , with respect to which the integration is
    effected . If z be known, Rule I. supplies the direct, though not always the
    simplest,method of integrating the function .

    INTEGRATION BY PARTS.
    1910 By differentiating uv with respect to , we obtain the
    generalformula ( u,vdx = uv - ſuv.dx .
    The value of the first integral is thus determined if that
    of the second is known .
    RULE. — Separate the quantity to be integrated into two
    factors. Integrate one factor, and differentiate the other with
    respect to x . If the integral of the resulting quantity is
    known, or more readily ascertained than that of the original
    one, the method by “ Parts” is applicable .
    1911 NOTE. — In subsequent examples, where integration by Parts is
    directed , the factor which is to be integrated will be indicated . Thus, in
    example (1951), “ By Parts Seardw ” signifies that ear is to be integrated and
    sin be differentiated afterwards in applying the foregoing rule. The factor 1 is
    more frequently integrated than any other,and thisstep will be denoted by Sdx .
    INTEGRATION BY DIVISION .
    1912 A formula is
    f(a + ba )"dx = a f(a + bæ*)p+1+ 0 2*(a + b2M)p-1dx :
    The expression to be integrated is thus divided into two terms,
    the index p in each being diminished by unity, a step which
    often facilitates integration .
    Similarly, S (n + bx" + cx " P d.o
    = a (a + bx" + cam )? -? + bæ " ( a + bx" + cx™)p -1 + cx " (a + bæ " + cm" )? -?.

    1913 Ex. — To integrate (x * + a%) dx.

    By Parts (dx, fv (x +a")dx = = v (x + a )- JSve(a + a )


    By Division, V (zº+a")dx = 1.70*a?) + a?)*
    Therefore, by addition,

    [ vce*+a%)dx =tëV( + )+15avrete


    = {xv (z’ + a?) + fa’log {x + V (z? + a?)},by (1909, Ex. 1).
    318 INTEGRAL CALCULUS.

    INTEGRATION BY RATIONALIZATION.
    1914 In the following example , 6 is the least common de
    nominator of the fractional indices. Hence, by substituting
    % = w , and therefore x , = 6z", we have
    rat - 1 , 73 - 1 dx , = 6
    Txt 2 - 1 dm
    Am J 22
    di - 1 "

    = 65 (x +2 – 2 +x2–2 + 1 - 2 1+ 1 )) d .
    Each term of the result is directly integrable by (1922) and
    (1923). For other examples see (2110 ).

    INTEGRATION BY PARTIAL FRACTIONS.


    1915 Rational fractions can always be integrated by first
    resolving them into partial fractions. The theory of such
    resolutions will now be given .
    1916 If $ (w ) and F ( x ) are rational algebraic functions of x ,
    • (a ) being of lowest dimensions, and if F' (x ) contains the
    factor (x -- a ) once, so that
    F (x ) = (x — a )y (x)..................... (1) ;

    1917 then ole = + and = . ...(2).


    Proof. — Multiply equation (2) by ( 1), thus
    q (ə ) = A4 (2 ) + (x — a ) x (x ).
    Therefore, putting x = a , • (a ) = A4 (a ). Also , by differentiating (1 ), and
    putting x = a afterwards, F ' (a ) = (a ). Therefore A = º (a ) = F (a ).

    1918 Again , if F (x) contains the factor (x — a ), n times, so


    that F (x ) = (x — a)" \ (x ).
    (x) – A , A.
    Assume
    F (x) + (x — a)" ++ 1(x _40
    ,-- +...+
    — a)"–1 T * - 7x40
    - a +* * (x )*
    To determine A1, A, ... An. Multiply by (x – a)"; put
    x = a and differentiate, alternately .

    1919 If F (x ) = 0 has a single pair of imaginary roots inary


    STANDARD INTEGRALS. 319

    at iß ; then ,applying (1917), let


    $ (a + i = A - iB ; : (a ) = A + iB ;
    and the partial fractions corresponding to these roots will be
    A - iB AtiB _ 2A (x — a ) + 2BB
    x - a - iß * x - a + iB (x - a )” + B +
    For practical methods of resolving a fraction into partial fractions in the
    different cases which occur, see (235 - 238 ) .

    INTEGRATION BY INFINITE SERIES.


    When other methods are not applicable , an integral may
    sometimes be evaluated by expanding the function in a con
    verging series and integrating the separate terms.
    Ex. de = loga +ax + 4* + momento +&c. (150)

    STANDARD INTEGRALS.

    1921 Some elementary integrals are obtained at once from


    the known derivatives of simple functions. Thus the deri
    vatives ( 1422 – 38) furnish corresponding integrals. The fol
    lowing are in constant use :
    + 1
    M i
    O
    C

    1922 ſamdr = * S du = log x.


    1924 ſade = log ſe= e .
    - OT

    1926 Savetas= c0-1c,or 4 sin- . [Saba.


    1927 Savetass = Lloga+v(* * *). [Substituto ?
    &
    320 INTEGRAL CALCULUS.

    1928 Sport as = log{w + V(x*+< )}. (1909,Br.1)


    iz
    -

    1929 Svekatec ) = sin- or – cos' (1434)


    © x de
    1930 Joos = v (a’ Ex?).
    1931 By Parts,Division, and adding results (1913),weobtain
    TV(x * + a?)dx = 1x V (x *£a?) + ja ’ log {x + v (x2 + a%)} .
    1932 By Parts, Division , and difference of results,
    . x² dx
    J = (x + a?) = }xv (x*£a”) F }a' log {x + v (x2 + a )}.
    1933 ( V (a?— xº)dx = £a*sin-1 * + 1x v (a? —x4).
    1934 Subtes = fa*sin- * - *.« V(a?— xº); As in(1932)

    1935 Si te 2 tan- or - cotone (1436)


    1936 S on za log (By Partial fractions
    1937 Some alog at I Do.
    1938 (sin « dx = - cosx. ſcosadx = sin x.
    COS a COS

    1940 ſtanædx = - log cosx. . ſcot« dx = logsinæ .


    1942 (secxdx = logtan (5 + ). Scosecxdx = log tano Cose

    METHOD. - (1940 , '2 ), substitute cosx. (1941, '3 ), substitute sin 2 .

    1944 ſsin --«dx = x sin-+x + v (1–a“).


    1945 cos - 2 x dx = x cos - x - V ( 1
    CIRCULAR FUNCTIONS. 321

    1946 ( tan-' « dx = x tan-- r - } log (1 + x ).


    1947 | cot -1 x dx = x tan -+x + } log (1 + x?).
    1948 sec-*«dx = x sec-'x – log (x + V (122 – 1)}.
    1949 ( cosec -1x dx = x cosec -1x + log {x + v (x2 - 1)}.
    Method. — (1944) to (1949), integrate by Parts, f de.
    1950 log x dx = x log x — X. [By Parts, do

    1951 Satocoa = v =65 tau-1{ tan ğ 147%)},


    1952 or
    1 ✓ (bta) + 1 (b - a ) tanix
    V ( b - u ”) v (6 + a ) - V ( b - a ) tan x '
    according as a is > or < b .
    [Subs. tanza, and integrate by (1935 or '37)

    VARIOUS INDEFINITE INTEGRALS.


    GENERALIZED CIRCULAR FUNCTIONS.

    1954 sin ” dx . cos” dx . cosec" dx.


    METHOD. — When n is integral, integrate the expansions in (772 - 4 ).
    Otherwise by successive rednction, see (2000) . For ſcosec" dx, see ( 2058).
    tan ” - 18 tan " -3x tan " - 5 x
    1957 tan ” xdx = 1 - 1 - 1 - 3 + 9 - 5 - & c.
    PROOF. — By Division ; tan” x = tan”- 2 x sec' x — tan ” -% X , the first term of
    which is integrable ; and so on .

    1958 (9 _ dr
    CO2)* = Tab
    J (a + b cos x ) S (a − b cosz) 1dz.
    METHOD. - By substituting Similarly with
    sin æ in the place of cos x, substitute 11 — .
    1959 | cos”x dr.· Jsin
    ſcos”x dx. J cos
    nëdr. sin ”x
    na dx.
    dx. I( sin
    sin
    sinn”x d.r.
    COS NX sin nr J cos nx nx dr. SI I

    2 T
    322 INTEGRAL CALCULUS.

    METHOD. - By (809 & 812), when p and n are integers, the first two
    functions can be resolved into partial fractions as under , p being < n in the
    first and < 1 - 1 in the second. The third and fourth integrals reduce to one
    or other of the former by substituting ja - X .
    1062 cos x _ 1 Fraï ( - Ir sin (2r
    1 )r : 1 sin 1 )00 cos? (21 - 1) 0
    (2r -- 11
    1963 cos nie n r cos æ - cos ( 2r - 1) 0 , with 0 =
    1964 cos”x = 1
    sin ne n sinx
    n- (- 1) +1sin º s” 18, with 0 = 1 .
    cos xrô- cocos rů
    The fractions in (1963 ) are integrated by ( 1952) ; those in (1964 ) by (1990 ). :

    Formulæ of Reduction .
    1965 ( cosnxdx = 2 ( cos (n - 1)« dx _ ſcos(n – 2). dr
    cos” x cos”- X Jcosx
    ” a
    rcos nx dr =
    sin (n - 1 ) x 'cos (n — 2 ) x
    1966 sinº X J sin ? - 18 J sinº
    ' sin nx dx = ( cos ( n - 1) x ( sin (n - 2 ) X
    1967
    1967 sine dx = 21 sin ” - X J sin ” x
    Csin nxen ' sin (n - 1) x ' sin (n — 2) x
    1968 J cos”æ dx = 2JI cos” - X cos ' x
    Proof. — In (1965) . 2 cos x cos (n - 1) x = cos nx + cos (1 – 2 ) x , & c.
    Similarly in ( 1966 – 8 ).

    1969 sin” x sin nx dx


    sin ? x cos nx
    - -
    p +n 1 + 1, sinº-1x cos(n - 1) x dx.
    1970 cos” x sin nx dx
    cos x cos nr p
    cos” — x sin ( n - 1 ) x dx.
    p + n Jo
    p + n

    1971 sin ” x cos nx dx


    sin ” x sin nx p
    p +n p + n
    - sinº-1 x sin (n - 1) x dx.
    1972 COSP x cosnx dx
    nt
    sin nx
    cos? Y t , m P
    p I cosa -l x cos (n - 1) x dx.
    p + nJ
    CIRCULAR FUNCTIONS. 323

    Proof. — ( 1969). By Parts, sin nx dx. In the new integral change


    cos nz cos into cos (12 — 1) x - sin næ sin x . By successive reduction in
    this way the integralmay be found . Similarly in ( 1970 – 2) .
    Otherwise, expand sinºx or cos x in multiple angles by (772 – 4 ), and
    integrate the terms by the following formulæ .

    1973 – 1975
    sin på sin nx dx = 21 (sin
    \
    (p —n)x _ sin (pp + +n)x
    p - n n
    ).
    ):
    and so with similar forms, by (666 - 9).

    1976 cosCOSpedo
    Nx and ( sincospanxdw are found from
    ſicos px — sin pa dn = 212P+n -1 dz
    COS nx J 1 + z2n
    when p and n are integers, by equating real and imaginary
    parts after integrating the right side by (2023) .
    PROOF. — Put cos a ti sin a = z ; therefore izdx = dz. Multiplying nume
    rator and denominator of the fraction below by cos nx + i sin nx, we get
    cos px + i sin px _ , cos (p + n ) x + i sin ( p + n ) . _ . * n i
    cos na 1 + cos 2nx + i sin 2næ 1 + zan
    . ZP +n -1 dz
    therefore
    cos no I 1 + 220

    1978 J[ cossinpånxdy and J( sinsinpanadw are found in the same


    ſcos pr + i sin px. 2p + n - 1 du
    way from
    J sin na 1 - 2, 20
    Proof. As in (1976 ), by multiplying numerator and denominator of
    cos pe ti sin poche orci sin na
    ople by cos nx - i sin næ ,
    sin na

    1980 cos w dx and / sin x dr


    J V ( cos no J ( cos nx )
    cos x + i sin e
    Putting y = 9 (cos nx ) ;, we
    we find tan nx = i (lấy”) , and therefore

    ydo = 21 Hence,multiplying by i,wehave


    ſic♡os(cosx—sin«
    nx )
    dx = J12 dy- y"
    The real part and the coefficient of i in the expansion of the integral on
    the right by (2021, ²2 ), are the values required .
    324 INTEGRAL CALCULUS.

    dir
    1982 [Subs. tan a
    J a cos* x + b sinºa
    1982 (awet in a= Thaytan"(tana ). [Subs,tana
    1983 Ja + b tanmtane {b log (a cos x + b sin x ) + ax}. [Subs. tan x
    a ' + 72 10
    ſ dx 1 tan -1 V (a + b)
    J a + b sinºa ✓ (a + ub) cot x va' [Substitute cota

    1985 Spit corte do = tan"'(acosa)– coec. [Substitute acose


    1986 . cossx de tan - t a sin
    [ Subs. sina
    J 1 - u : cos'c a ( 1 — ^ ^) at v (1 - a )

    1987 [cosa V(1– aº sinºx)dx = }sinæ v (1–aºsin d)+ sin '(asinx).


    [ Substitute a sin 3
    1988 | Bin /(1 -asinºx)
    1 - a
    4 = - 3 cosz / 1 -asinº )
    2a
    " log a cosa + V ( l - a' sin ' x ) } . [Subs: a cos z
    1989 sin « (1 – a*sinºx)?dx = - cosx (1 – a`siu®x);
    ti ( l - a ) sin x / ( 1 - a ' sinºa ) dx. [ Subs. a cos a

    1990 siu a (a +
    dr cos x )
    = log {(a cosecxa +’ -162cotx) tanº ;x}.
    0 - b cosa 1, sin x
    Ву
    sin x (a + b cos it ) (a - 6 ) sin a (u - 6 ?) (a + b cus x )
    tan xdir - 1 cos rv ( h - m )
    1991 V (a + b tau? a ) = (b - u ) V6
    [ Subs. cos x ✓ (b - a )

    1992 ( v (a+b sinº») lx = vb cos-1 Vbcosz


    sin x ✓ (a + b)
    - va log { va cotx + v (a cosec’ x + b ) }.
    METHOD. - By Division (1912), making the numerator rational, and in
    tegrating the two fractions by substituting cot x aud cos x respectively .

    1993 dr osi dr - dt — },
    Jatal cos & + ccos 22 ml] 2ccosa + b - m J 2c cose + 6 + m ) '
    where m = { 12 — 2c (a — c ) }. Then integrate by (1953).
    - dx do
    1994 (1953 )
    Jacos x + b sin x +c Jv ( a ’ + b ') cos 0 + c
    Method. - Substitute 0 = 2 - a , where tan a
    EXPONENTIAL AND LOGARITHMIC FUNCTIONS. 325

    1995 PF (sin x , cos r )dr


    Ja cos x + b sin « to'
    F being an integral algebraic function of sin x and cos x.
    Methon. — Substitute () = x - a as in (1994 ), and the resulting integral
    takes the form f ( sin 0, cosa ) 170 _ ſ o (cos A ) 110
    1215 sin 0 ( cos 6 )
    J A cos 0 + B J A cos 0 + B ' J A cos 0 + B ?
    since f contains only integral powers of the sine and cosine, and may there
    fore be resolved into the two terms as indicated .
    To find the first integral on the right, divide by the denominator and
    integrate each term separately. To find the second integral, substitute the
    denominator .

    F ( cos x )dx
    1996 ( a + b , cos x ) (an + 6, cos x ) ... (arton COS X
    where F' is an integral function of cos x .
    METHOD. — Resolve into partial fractions. Each integral will be of the
    form A + 6.Cosa (1951).
    PA cos x + B sin r + 0
    1997
    a cos x + 6 sin x + c
    METHOD. — Let • (a ) = a cos x + b sin xtc ; i . O'( x) = - a sin x + b cos x .
    Assame A cos x + B sin x + C = 10 (7 ) + uo' (x ) + v . Substitute the values of
    9 (cm) and ' ( x ), and equate the coefficients to zero to determine 1, M , v. The
    integral becomes

    S{1+ugo(2)+ p cos }de = 18+ulog (a)+ Seco dai,


    and the last integral is found by (1991).

    EXPONENTIAL AND LOGARITHMIC FUNCTIONS.


    1998 |eF( x ) dx can be found at once when F (x) can be
    expressed as the sum of two functions, one of which is the
    derirative of the other, for

    Ter{$ (x) + $'(x )}dx = e*+ (x).


    1999 ſea cos"bxdx and ſersin "bxdx are respectively = are

    a cosbx + nb sin brCenx


    . cos ” -1 WX + n ( n - 1 ) 62
    a ’ + n°6 a ’ + nl - Joex cos” — 2 bx dx.
    326 INTEGRAL CALCULUS.

    and
    \" - 1) bº ( par sinn-2 bxdx.
    a sin br — nb cos6X ear sinº -1 bx ,+ n" (n
    a ” + n -6 e si 6X + +n juat sinab
    Proof. — In either case, integrate twice by Parts, S
    Otherwise, these integrals may be found in terms of multiple angles by
    expanding sin " x and cos" by (772– 4 ), and integrating each term by (1951- 2 ).

    2000 le* sin ” x cos” x dx is found by expressing sin” x and


    cos” x in terms ofmultiple angles .
    Ex. : Sed sin®æ cos xdx. Put eio = z in (768),
    (2isin x )" ( 2 cos x ) = (3 - 2- 1)' (2 + z -1) :
    = (z — -1)* (z? – 7-2): = (37 —27) — 3 (25 — 2-5) + (23 — 2 -5) + 5 (2 — 2- 1) ;
    .:. 2º em sinº x cos” x = ex (sin 73 — 3 sin 5x + sin 3x + 5 sin x ).
    Then integrate by (1999).
    2001 THEOREM. — Let P , Q be functions of « ; and let
    SPdx = P1, SP,Q_dx = P,, SP Qqdx = Ps, & c. Then
    ( PQ"dx = P, Q" — NP,Q »-1 + n (n - 1) P ;Q *-? – ...In Pn +1: n + 1•

    Proof.— Integrate successively by Parts,SPdæ, & c.


    ucces

    2002 THEOREM . - Let P , Q , as before , be functions of x ; and


    en

    let P, = ( ); P = ): Ps= (m ), &c. Then


    P P,
    Jordx = - (n - 1) Q.Q - - (n - 1)(n = 2) Q.Q---
    22

    Pz 1 CP . .

    (n - 1)(n - 2)(n - 3) Q : Q»–3***... * n - 1 . 1-1dx.


    Proof.— Integrate successively by Parts, 1 – (n -1)Qxdx.
    EXAMPLES.

    2003 m
    fem-1 (log x?)"dx
    av

    = F ( - 1***"(:= 1)7-4-..+(-1)-..).
    METHOD . 2

    Method.–By(2001). P,= ,P = P.= & .


    ALGEBRAIC FUNCTIONS. 327

    2004
    7m + n .r
    zv nedx = { " {1 + nx log æ + }(næ log ~)*+ &c.}dx,
    [By (149)
    and each term of this result can be integrated by (2003).
    2005 * xm - 1dx
    + 1
    J (log x )"
    ml- n + 2 mº - n +3

    co
    = > " (n - 1 + (n - 1) (n - 2) + (0 - 1)(n — 2)(n - 3) To
    m ” - i Cam - 1dx
    in - 1 J log x •
    METHOD . — By ( 2002). Px = x " , P , = mxm -?, P , = m *zm- ', & c.

    The last method is not applicable when n = 1 . In this


    case , writing l for log x ,
    m212 m313
    006 rom - i dx ' = log l + ml +
    m * 14
    J log x 1 . 22T 1 . 2 . 32 " 1 . 2 . 3 . 42
    METHOD : am- 1 em log x
    Toga - Expand the numerator by (150) , and inte
    grate .
    See also (2161 -6 ) for similar developments of the expo
    nential forms of the same functions .

    PARTICULAR ALGEBRAIC FUNCTIONS.


    2007 nr
    S 1
    1) = n=1 { _ , - # }
    n (12 + 1 ) ... 2 (n - 1 ) ,
    ' n - 2 ( (1 - x )* - 1 . 2 ... (n − 1)
    n being even. (1918)

    [Subs. V (1—r")
    200 1. dx V2 + V (1 + x2)
    2009 50-25%(1+
    ( 1 – 2*) ✓ ( 1 + a ?)) = Tz log v2ti204
    ✓ ( l — w *).**
    dx x / 2 + 1 (TP- 1 )
    J (1 + x®) V (x2 - 1) ✓ (1 + x )
    328 : INTEGRAL CALCULUS.

    , 11x? (ae - br ) ae > be


    2011 (c + exc ") v (a + bx*) (ace - bc?) V I aec' + ac \ ' .
    ev ( a + bx?) + x1 (1C? — ace) ae 5 bị
    = V(l ?–ace)log vic tex";
    2012 Save the a long Via4luni)–va. [Subs.
    = log HXV2 V2 + ( 1 + 2+ )
    2013 J(2(1 -(1+) rº)dx
    (1 + x*) 1 - 22 [Subs.1- 2
    2014 $ 4 +250 ) = 7 since it [Subs.1.*
    2015 (1 + 2 ") de
    (1 + x ) + xV2 In - XV22
    1 - 2* 1- 1+a S . .
    1 x²ax vil + 4) + 212 . -12 / 22
    2016
    J (l — æ*) ✓ (1 + x *) 1 - x? 1+ ?
    [Substitute zv (1 + x ) = x 2 in ( 2015 - 6 )

    I do log
    (27 _ 1)* - -* - i tan - 1 _ 2
    L =
    2017 sa
    J ( 1 - x ”) ( 2.c: — 1)? (2.22 – 1)* + x (2x² - 1)
    [ Substitute x = z (2.x2 - 1 )

    - da
    2018 i = tan
    J ( 1 + x+) { ✓ (1 + x*) — x }} { v ( 1 + 2*) — x *}*
    [ Substitute vivilte')-x }
    2019
    Top =log va+ 1=2– Jistan-*2 1* - . [Subs.W1=2)
    stitut
    dx ng i 1+ 2
    7

    a reduces to ( 2019) by substituting


    2020 (1 + x))Vltme+3.5
    J (1+ ) roc
    ( 1 + 34 + 3.< )
    INTEGRATION OF *
    x" + 1
    If I and n are positive integers , and 1 - 1 < n ;* then , n
    being even ,

    * If l = n, the value of the integral is simply - log (x*– 1).


    INTEGRATION OF am 329
    x" + 1

    2021 2242 = log(0– 1)+ (-1'log(x+ 1)


    OS

    + EcosrlBlog(x2–2xcosrB+ 1)– £sinrißtan-1*siostry


    where ß = * , and I denotes thatthe sum of all the terms ms

    obtained by making r = 2, 4 ,6 ...n – 2 successively , is to


    succes

    be taken .

    If n be odd ,
    xl- 1dx
    2022 S i = log (xX -- 1) 1)

    + cosrlBlog (x2 – 2x cosrß + 1) – ? sinrißtan -1* si-ncosrß


    rß
    with r = 2, 4,6 ...n - 1 successively.
    If n be even ,

    2023 Sid = - cos rlſ log(22–2xcos rß + 1)


    - CC

    + 2 sin rlßtan--*-—cos rß,


    with r = 1, 3, 5 ... n - 1 successively .
    If n be odd ,
    2024 Stdr = ( 1) = log (x + 1)
    Ecosrißlog (x2 – 2x cosrß + 1) + 2 sin rlßtan -12 sincosrß
    rß
    with r = 1, 3, 5 ...n — 2 successively.
    PROOF. — (2021 - 4 ). Resolve into partial fractions by the method
    men + 1 into partial i
    of (1917). Wehave º (a ) = ?-1 = "am
    " , since a ” = F1. The different
    " F (a ) na" - } . n '.
    values of a are the roots of 2 " = 0 , and these are given by x = cos rß +
    i sin rß , with odd or even integral values of r. (See 480 , 481 ; 2r and 2r + 1
    of those articles being in each case here represented by r.) The first two
    terms on the right in ( 2021) arise from the factors a + 1 ; the remaining
    terms from quadratic factors of the type
    (x - cos rß - i sin rß) (x - cos rß + i sin rß) = (x - cos rp )' + sin rß.
    These last terms are integrated by ( 1923) and ( 1935) . Similarly for the
    cases (2022 - 4 ) .
    2 u
    830 INTEGRAL CALCULUS.

    2025 If, in formulæ (2021-4 ), 7 2 8(3# - rß ) sin rlß be


    added to the last term for the constant of integration , the
    integral vanishes with x , and the last term becomes
    In - 1 X sin rß
    sin rib tan
    " . 1 - x cos rß '
    reading – in (2021- 2 ), and + in (2023–4 ).
    real

    1 + 1

    2026 Scarpent (5) Sif da,


    where az" = bu". Then integrate by (2023 -4 ).

    2027 dx = + Esin mrß tan-1« sincosrßrß.


    J X" + 1 n

    where ß = 7 : n , and r = 1, 3 , 5 , ... successively up to n - 1


    or n - 2 , according as n is even or odd .
    2028
    aº - 1 - - m-1
    - dx = E cosmrß .log ( x - 2x cos rß + 1) ,
    . * +1 n

    with the same values of r ; but when n is odd , supply the


    additional term ( - 1)" 2 log (x + 1) = n .
    PROOF. – Follow the method of ( 2024, Proof) .
    Similar forms are obtainable when the denominator is 2 " - 1 .

    2029
    xl- 1dx
    And An = cos (n - 1) $ .tan - 4
    sin Ø

    - }£ sin (n - 1) $.log (x2– 2x cos$ + 1),


    where $ = 0 + n and r = 0 , 2, 4 ,... 2 (n - 1) successively.
    But if the integral is to vanish with x , write tan -- ,1 -* Xsincos o
    as in (2025).
    Proof. — By the method of (2024 ). — The factors of the denominator are
    given in (807), putting y equal to unity .
    INTEGRATION OF $ " (a + bx")&
    m
    831

    ( 21+4(log
    20300' Ja " +1
    æ)" dx = F ? ( B)" { cos (r18 + 3mm)
    x log (x ? — 2x cosrß + 1) - 2 sin (rlß + ima )tan
    van
    -1* sincos P },
    rß } '
    with the values of ß and r in (2021- 4 ).
    PROOF. — Differentiate the equations (2021- 4 ) in times with respect to l,
    by (1427) and (1461 - 2 ). If m be negative, integrate m times with respect
    to l,and the same formula is obtained by (2155 -6 ).

    In a similar manner, from ( 2027 – 8 ) and (2029), the general


    terms may be found for the integrals
    2032
    [ { - ( - 1)" x " -m - 1} (log x ) on and ?- ? (log x ) " dx
    X" + 1 J 22n — 2.x " cosn0 + 1 %

    INTEGRATION O

    2035 Rule I. — When m + is a positive integer, integrate


    n

    by substituting z = p(a + bx^)ă. Thus m + 1


    9 - 1

    fix "(a + ban)ădx = 2 (2 +2-1( 7972) *-'dz.


    Expand thebinomial,and integrate the separatetermsby (1922).
    2036 But if the positive integer be 1 , the integral is known
    at sight, since m then becomes = n - 1 .

    2037 RULE II.


    RULE When mtl
    IL -_ When - + P is a negative integer,
    n q
    substitute z = (ax -" + b ) Thus m + 1

    2+(a+ banyído= - 3 *- (429)


    a * **d .
    Expand and integrate as before.
    2038 But, if the negative integer be – 1, the integral is
    found immediately by writing it in the form
    ſz*** (a2-*+ b) dx = (2---1(ax+*+ uj das
    na( + )(ax=*+ b) +
    332 INTEGRAL CALCULUS.

    EXAMPLES.

    2039 To find Sæt (1 + 2+)# dx. Here m = }, n = }, p= 2,q = 3, m + 1 = 3,


    a positive integer. Therefore, substituting y = (1 + xt) , x = (yö — 1)',
    X , = 6yº (y — 1), and the integral becomes
    (yº– 1)y de dy = 6 y*( ?– 1) dy,
    the value of which can be found immediately by expanding and integrating
    the separate terms.

    2040 <°(a +bat)t dla = i& (a+bx+)*.


    For " n
    = 1 ( 2036 ) ; that is, m + 1 = n , and the factor 28 is the derivative
    of fat.
    2041 ſ (1 + 1x)?
    by dx, or | x - * ( 1 + x ) dx. Here m = - 1, n = 1, p = 2 ,
    J
    q = 3, n + 2 = – 2, a negative integer. Therefore, substitute
    m ?!9
    y = ( x= + + 1 ) , x = (1 - 1) -2, ~,, = - 6y2 (78 – 1)-& Writing the integral in
    the form below , and then substituting the values, we have
    Sæ- (x - * + 1)* x,dy = – 6 jy* (78 – 1) dy,
    which can be integrated at once .

    2042 | x -? (12 + bxn) - dx. Here " + 1 + = - 1:


    J æ ( a + bx" ) n 9
    therefore, by ( 2038 ), the integral
    = 12 - -1(ax-" + b)- dx = - na
    log (a.x=" + b).
    REDUCTION OF S. m (a + bx”)” dx .
    When neither of the conditions in (2035, 2037) are ful
    filled , the integral may be reduced by any of the six following
    rules, so as to alter the indices m and p , those indices having
    any algebraic values.
    2043 I. To change m and p into m + n and p - 1.
    Integrate by Parts , ſxm dx.
    2044 II. To change m and p into m - n and p + 1 .
    Integrate by Parts, ſxn-1 (a + bx")" dx.
    2045 III. To change m into m + n.
    Add 1 to p . Then integrate by Parts , ſxm dx ; and also
    by Division , and equate the results.
    REDUCTION (a + bx" ) dx. 333

    2046 IV . To change m into m - n .


    Add 1 to p , and subtract n from m . Then integrate by
    Parts , jx dx ; and also by Division , and equate the results.
    2047 V . To change p into p + 1 .
    Add 1 to p. Then integrate by Division , and the new
    integral by Parts, ſx" -1 (a + bx^)”.
    2048 VI. To change p into p - 1.
    Integrate by Division , and the new integral by Parts,
    5x -1 (a' + bxº)p!
    2049 MNEMONIC TABLE FOR THE SAME Rules.
    I. m + n , p - 1 By Parts (m ).
    II. m - n , p + 1 | By Parts ( p).
    III. (p + 1), Parts (m ) and Division .
    +

    IV . m - n (p + 1 , m - n ), Parts (m ) and Division.


    1

    V. P + 1 (p + 1), Division , and the new integralby Parts (p ).


    VI. ' p- 1 Division, and the new integral by Parts (p).

    By applying the rules, Formulæ of reduction are obtained .


    Thus, any of the six values below may be substituted for the
    integral (a + bx ”)” dx .

    2050 - 2055
    I 2 + 1 ( a + bx") bnp
    * * * " ( a + bx") p- 1 dx.
    m + 1 m + 1

    II. Adom -n +1(a + b.x*)p +1 _ m - n + 1 [xm -n (a + bx")p+1dx.


    bn ( p + 1 ) bn (p + 1) J
    ram + i( a + bx ) p +1 6m
    ( tntnpt mtn
    (a + bx”) dx .
    a (m + 1 ) a (m + 1 )
    v 2 -n + i (a + bxa) p + 1 a (m - n + 1)
    m -n (a + bx ") dx .
    b (m + np + 1) b (m + np + 1 ) J
    grm +1( a + brin ) p + 1 , m + n + np + 1 mm ( a + bx*)p +1dx .
    an ( p + 1 ) an (p + 1) J
    Mm + 1 (a + burn)P anpp
    Tama + bw " )p - 7 dx.
    mtnp + 1 ' m + np + 1 "
    334 INTEGRAL CALCULUS.

    EXAMPLES.
    2056 To find va’ — 12
    * *) dx.
    , Apply Rule I. or Formula I. ; thus
    J 203
    32- (a?—xº)}dx = - 12-?(a* —24)!+ ! Sæ- (a* —zº)- dx (1927).
    2057 To find J((a ? mida.
    — 2 ?) ?
    Apply Rule II.or Formula II.; thus
    $26( – )= da = 2°(a* — )- - 3fzº(a* —2,)-4dx (1934).
    2058 Scosec" o do. Substituting sin 6 = x, the integral becomes
    sin -mo do dx = x -m (1 — xº) - ! dx.
    dæ
    Apply Rule III.; thus, increasing p by 1 and integrating , first by Parts
    Sæ-mdx, and again by Division ;
    x -m ( 1 – 22) dx = " " ( 1 – X+) '
    1- m '
    22- (1 - x2) dx,

    fa * (1 –xº)!dx = <-" (1 –29)= de – 22-n(1 –x2) +da.


    Equating the results, we obtain
    2059 2 -1 (1 - 2 ) - ! dx = (1 - 2 ) + 22-m (1 – w*) - dx.
    1 - m
    By repeating the process, the integral is made to depend finally upon
    2 - * (1 — xº) - dx or 1 — ?) - dx ,
    according as m is an odd or even integer (1927, '29).
    2060 Ssinmodo is found in a similar manner by Rule IV . The integral
    to be evaluated is ſąm (1 -xº) - !dx ; and the integral operated upon is
    $2m-2 ( 1 – 2 ?)!dx . Otherwise apply Formula IV . See also (1954).

    2061 To find cor. Apply Rule V . p = - r, and increasing p by


    1, we have, first, by Division,
    $(&*+a+ -*dx = {x?(x2+ *)**de +a?|(x2+a )-*dx.
    Integrating the new form by Parts, 5x (xº+ aº)-"da,wenext obtain
    $2°(x++ a?)- dx = *** ta?)l- *dx.
    2 (1 - r ) 2 (1 - r ) )
    Substituting this value in the previous equation, we have, finally,
    2062
    ſ dx 2r - 3 dx
    J (2° + a”)= = 2 (r – 1) a (na” + a?)r-it a "(2r — 2) J (7* + a’yr-1°
    REDUCTION nº @ cos O do. 335

    This equation is given at once by Formula V . Thus r is


    changed into r - 1 , and by repeating the process of reduction ,
    the original integral is ultimately made to depend upon (1935 )
    for its value if r be an integer .

    Another formula for this integral is


    - dx ( - 1 )" - 1 dr- 1 / 1o tan - 1
    X
    2063
    J (x ? + B )* 1 . 2 ... ( r VA

    Proof. — Write B for a in ( 1935 ), and differentiate the equation r - 1


    times for by the principle in ( 2255 ).

    2064 To find S (a’ + x?) " dx. Apply Rule VI. By Division, we have
    fle + )”da = a*|(4*+22)#--!dx + [**(a*+x2)dn–1da.
    The last integral, by Parts, becomes
    1**(a*+2*) -"dx = + <(a*+x?tu - [ *+x2) ".nJ

    Substituting this value in the previous equation,weobtain

    2065 $(++a+'" dx = ( * 47*9**+, qui S(a++x)---da,


    a result given at once by Formula VI.
    If n be an odd integer, we arrive, finally, by successive
    e Successive

    reduction in this manner , at S (a ’ + x+)* dx (1931) .

    2066 The integral Ssin” O cos odo is reducible by the fore


    going Rules I. to VI., if, in applying them , n be always put
    equal to 2 ; if p be changed into p + 2 instead of p + l ; and
    if Division be always effected by separating the factor
    cos? 0 = 1 - sin 0 .
    Proor. S sin” 6 cos” O do = $ 2" (1 — x?)}(p-1 dx, where x = sin 0. Thus
    n = 2 always, and the index } ( p - 1 ) is increased by 1 by adding 2 to p .

    Thus, Rule I. gives the formula of reduction


    2067
    sin "Acos? 6d0 – sin " +10 cosp -18 p -1CSsinº
    .inº ++ 2200Ccos”
    O
    -2000 .
    m + 1 m + 1
    336 INTEGRAL CALCULUS.

    But the integral can be found by substitution in the fol


    lowing cases :
    If r be a positive integer ,

    2068 cos? +1x sin’xdx = [(1 – 2* z dz,where z = sin x. re

    + 1
    ere 2 = COST .
    2069 | sin? + cos”. de = - 1 - sy = ds,where= =cosz.
    If m + p = - 2r,

    2070 sin” x cos”xdx = (1 + z2)=-1 " dz, where z = tanz.


    FUNCTIONS OF a + bx + cx”.
    The seven following integrals are found either by writing
    2071 a + bx + ca = {(2cx + b)2 + 4ac — bº} - 40,
    and substituting 2cx + b ; or by writing
    2072 a + bæ — ca = {4ac + b? — ( 2cx — b)2} = 4c,
    and substituting 2cx — b .
    - dx 2cx + b - V ( - 4ac)
    2073. Ja+ bx + cm = 7 (b - 4ac) log 2cx + 6 + V – 4ac)
    2cx + b
    OF V (4ac — 65) tan - (4ac0*)'
    according as bº > or < 4ac (2071, 1935–6 ).
    2014 C dx ✓ (1 ? + 4ac) + (2cx - 6 )
    Ja + bx - cx ? — 72 + 400 ✓ (6 + 4ac) - (2cx - 6 )
    (2072, 1937 )

    2075

    Svattester)= to log(268+6+2/0v(a+bx +cx')}.


    1 dx sin - 1 2cx - b
    2076
    2076 Svatele-ca5
    ✓ ( a + bx - cxº) = te sine V (4ac + 6 )
    (2071- 2, 1928 - 9 )
    FUNCTIONS OF a + bx + ca ". 337

    2077 v (a +bx + c«^)dx = {e-+ | V (12° + 4uc — b )dy.

    ola
    2078 v (a +bx— ca“)dx = $c-+ | V (xac+U* — )dy,
    where y = 2cx + b . The integrals are given at (1931– 3 ).
    I dx dy
    2079 = 22p - 1 cp - 1
    (a + b + cx ?) ” | ( y : + 40c - 62)
    [By (2071), the integral being reduced by (2062– 3).
    p (lx + m ) dx
    2080 J ( a + bx + cx?)
    1 | (2cx + 6)dx umblr_ dx
    = 2cJ (a +bx + ca%)p + ( m 2c ) J (a +bx + ca?)p*
    The value of the second integral is ( a + bx + cx2) 1 -P ; (1 - 2 ),
    unless p = l , when the value is log ( a + bic + caº) . For the
    third , see (2079) .
    METHOD . — Decompose into two fractions, making the numerator of the
    first 200 + b ; that is, the derivative of a + bx + ca?.

    2081 Sotho dx may be integrated as follows:


    I. If bº > 4ac, put a and B for – b ( —4ac), and. and ,
    2c
    by Partial Fractions, the integralis resolved into
    clan ) {T na ta dx - Sapta dx }. (1986)
    II. If bº < 4ac, put a = n²and 2V (ac) – b = mº, and the
    integralmay be decomposed into
    0892 2mn
    208 1 c {( J[ ( x- pn) æ + qm dx - 1 (2 - pn ) x - qm dx {,
    + mx + n ge — mx + n
    the value ofwhich is found by (2080).
    III. If = 4ac,
    2083 J|a_ _+ bxdx + t
    cmd = 2 2a +h
    6x27 . (2ab tan -'(rv 2 ).(2062)
    338 INTEGRAL CALCULUS.

    C x dx 2a
    2084
    Ja + bx + cx * b ( 2a + bx )
    X )
    2085
    2085 Ja
    Satriateur
    + bx + ca4 = ,27;tan-"(vv ) 2aza +furs
    bx")

    REDUCTION OF Sæm ( a + bx" + c22n)” dir.


    Note. In the following Formulæ of Reluction , for the sake of clearness,
    am (a + bx " + cx?n )" is denoted by (m , p ) , and the integral merely by (m , p ) .

    2086 (m + 1)|(m,p) = (m + 1,p)


    - bnp | (m + n ,p - 1) - 2enp | (m + 2n,p – 1) ....... (1).
    2087 bn (p + 1) ſ (m ,p) = (n =n + 1,2 + 1)
    -(m , n + 1) ſ(m – n,p + 1) – 2en (p + 1)ſ(m + n,p)...(2).
    2088 2en(p + 1) ſ (m ,p) = (m — 2n + 1,p + 1)
    :: -(m — 2n + 1)|(m — 2n,p + 1) –bn(p + 1)/(m –n,p)...(3).
    2089 (m + up + 1) ſ (m ,p) = (m + 1,p)
    tanp | (m ,p - 1) -cnpſ(m + 2n,p - 1 . ...........
    2090 (m + 2np + 1) ſ (m ,p) = (m + 1, p)
    + 2anp ſ(m ,p –1) +bnp ſ (m +»,p – 1) ....... ...(5).
    2091 6(m +up + 1) (m,p) = (m –n + 1,p + 1)
    -a(m n + 1) |(m –n,p)– e(m + 2np + n + 1) J(m. .+. .n,p)
    .. (

    2092 bn (p + 1) | (m , p) = - (m - n + 1,p + 1)
    +(m +2np+ n + 1){(m –n,p + 1) –2an(p + 1)|(m .–n,p)
    . .. ( 7 ) .
    REDUCTION OF Sæm(a + " + cæºr)” dæ. 339

    2093 en (p + 1) ſ (m,p) = (m –2n + 1,p + 1)


    +an (p + 1)/(m —2n,p)– (m + np - n + 1) ſ(m –2n,p + 1)
    ... .. . (8 ).

    2094 an (p + 1) S (m , p) = - (m + 1, p + 1)
    +(m + np + n + 1) ſ(m .p + 1) +cn(p + 1)ſ(m + 2n,p)...(9).
    2095 2an (p + 1) ſ (m , p) = - (m + 1, p + 1)
    +(m +2np+ 2n+ 1) (»,p +1)–bn(p+ 1)$(m. ..+n,p)
    ... (10) .

    2096 a(m + 1) (m ,p) = (m + 1,p + 1)


    –b(m + up+ n + 1)f(m +n,p)–c(m +2np + 2n + 1)(m + 2n,p)
    ....(11).
    2097 c(m + 2np + 1) | (m , p) = (m — 2n + 1, p + 1)
    –6 (m +up -n + 1)|(m –n,p) —a(m —2n +1)/(m...—.. .2n,p)
    ( 12).
    Proor. — By differentiation , we have
    2098
    S (m , p) = m S (m — 1, p) + bnp S (m + n - 1, p - 1) + 2cnpS (m + 2n - 1, p - 1).
    Formulæ (1 ), ( 2), and (3 ) are obtained from this equation by altering the
    indices m and p , so that each integral on the right, in turn, becomes ſ (m , p ).
    Again , by division ,
    2099 S (m ,p ) = a ſ (m ,p - 1) + of(m + n,p - 1) +cf(m + 2n,p - 1)...(A ).
    And, by changing m into mắn, and p into p + 1,
    2100 S (m -n , p + 1) = a ſ (m - n ,p) + 0 % (m ,p) + c , (m + n ,p) .....(B ).
    Formulæ (4 ) to ( 12 ) may now be found as follows :--
    (4 ), by eliminating ſ (m + n , p - 1) between (1) and (A ) ;
    (5 ), by eliminating S (m + 2n, p - 1) between (1) and (A ) ;
    (6 ) , by eliminating ſ (mớn, p + 1) between ( 2) and ( B ) ;
    (7) , by eliminating S (m + n , p ) between (2) and ( B ) ;
    310 INTEGRAL CALCULUS.

    (8), from (4 ), by changing m into m - 2n , and p into p + 1;


    (9 ), from (4 ), by changing p into p + 1 ;
    ( 10 ), from ( 5), by changing p into p + 1 ;
    (11), from (6 ), by changing m into m + n ;
    (12) , from (6 ), by changing m into mấn .

    If a and B are real roots of the quadratic equation


    atbæ" + ca2n = 0 , then , by Partial Fractions,
    0101 dx 1 sp x dx parmdx
    "Ja + bx" + cx ?n = c (a - B ) J x" — a TJ x" — B } '
    and the integrals are obtained by ( 2021-2 ).
    But, if the roots are imaginary ,
    m + 1
    C xm dx 2n 2 dx
    2102
    Ja + bil " + cxın J1 - 2 % " cos no + x ?n ?
    Z
    ca
    where cos no = - 21 u0) and == (*)**.
    2103 Val p is reduced to (2079–80) by (2097).
    dx dy
    2104 5)
    J (x + h ) v (a + bx + cxº) J✓ ( A + By + Cy
    where y = (x + h ) - , A = c, B = b ^ 2ch, C = a - bh + ch .
    dx cos -11 + hx _ 1 cosh -z1 + hx
    2105 ath '
    J (x + h ) Væ* - 1 ✓1h - x + h vh - 1
    2106 | dx sin -21 + hx
    J (a + h ) VI - a ✓h - 1 x+ h
    [By (2181) .
    METHOD. - Substitute (x + h ) - , as in (2004). Observe the cases in which
    h = l.

    210 - dx yr - dy
    J ( x + h ) ' v ( a + bx + cx“) JV ( A + By + Cy')
    with the same values for A , B , C , and y as in (2104). The
    integral is reduced by (2097) .

    (latm ) dx
    2108 J ( 22 + B ) v (a + bx + cx )
    INTEGRATION BY RATIONALIZATION . 341

    METHOD. — Substitute O by putting a = ß tan (0 + y ), and determine the


    constant y by equating to zero the coefficient of sin 20 in the denominator.
    | L cos 6 + M sin A je
    The resulting integral is ofthe form J ( P + Q cos 20)
    Separate this into two terms, and integrate by substituting sin 0 in the first
    and cos 0 in the second .

    $ (x ) dx
    2109
    F (x ) / (a + bx + cx?)'
    where • (a ) and F (@ ) are rational algebraic functions of x ,
    the former being of the lowest dimensions.
    METHOD. — Resolve (2) into partial fractions. The resulting integrals
    are either of the form (2107) , or else they arise from a pair of imaginary roots
    of F (x ) = 0 , and are of the type - ( Ax + B ) dx _ Substitute
    | { (x - a ) + p ?} v (a + bx + ca*)
    2 - a in this, and the integral ( 2108) is obtained .

    INTEGRATION BY RATIONALIZATION .

    In the following articles, F denotes a rational algebraic


    function . In each case, an integral involving an irrational
    function of is , by substitution , made to take the form
    JF (z ) dz. This latter integral can always be found by the
    method of Partial Fractions (1915 ) .

    2110 + b = ( a + bc ) , & c ., dc.


    + gx ) + gx )

    Substitute a + bxle = , where l is the least common de


    f + g~
    nominator of the fractionalindices; thus,
    nom

    _ fz' a dx _ 1z2- 1 (of - ag) |a + bxa _ , & C.


    b - gz?' dz (6 - gz?)? f + gæ /
    the powers of z being now all integral.

    2 nmn

    2011 fevr {u, certainment ,&o.}dr.


    Reduce to theform of (2110) by substituting æ“.
    342 INTEGRAL CALCULU' S .

    2112 [ F {v (a + V mæ+ n)}dr. Subs. 7 (a + V mæ+ n).


    2113 [ F {x, V (bx Ecx )}dx. Substitute
    2
    x=; z? I
    62
    And therefore ✓ ( bx + cx ') = (z? Fc)?

    2114 ( F {x, / (a + bx + cxº)} dx .


    Writing Q for a + bx + ca®, F may always be reduced to
    V ēQe,' in which A, B, C, D are constants or
    the form A + Bv
    C + Dv'
    rational functions of x. Rationalizing this fraction , it
    takes the form L + M Q. Thus the integral becomes
    ( Lde + ( MvQda, the first of which two integrals is
    rational, while the second is equivalent to Medt, which
    is of the form in (2075).

    2115 OTHERWISE. - (i.) When c is positive, the integral may be made


    rational by substituting
    - a - cza d e _ 2c (bz - cz ? — a) , a - 0
    a V (a + bx + caº) =
    2c2 - 6 ' dz (2cz — ) cz - b

    (ii.) When cis negative, let a , B be the roots of the equation


    a + bx - cu? = 0 , which are necessarily real ( a , b , and c being now all
    positive), so that a + bæ - cxº = c (x - a ) ( 6 - 2 ). The integral is now made
    rational by substituting
    caz? + B dx _ 2 ( a – B ) cz _ ( B - a ) cz
    . cz? + 1 ' diz - (cz* + 1) V (a + bx - cz*) = W .
    In each case the result is of the form | F (z) dz.

    2116 " F {va, Va + b.x" + cxən} dx,


    when " + " is an integer, is reduced to the form (2114) by
    n
    San

    substituting an.
    2117
    SF{«, / (a + bx),V (f + x:x)}dx. Substitute = = +
    INTEGRALS REDUCIBLE TO ELLIPTIC INTEGRALS. 343

    - f22 _ 2V (ag – bf)


    erore ,,2 x =
    and, therefore va + bx ) =
    9:2 — 6 ✓ (92? – 6 )
    vag - bf) dx 2 (bf - ag) %
    vif + gx
    ✓ (972 — b ) di (922 — 1)2
    The form SF {z , (922 — 6 )} dz is obtained, which is
    comprehended in (2114 ) .

    2118 »" F {x", ✓ (a + b+x?), bx" (a + b+q2")} dx,


    when m + is an integer, is reduced to the form SF ( ) dz
    n

    by substituting z = bæ" Iv (a + ?z ), and therefore e

    z= (a+W* *)
    m + 1
    REQ
    a legea nr.( )* * (* *).
    2119 ( x"(a + b.2%)* F (x»)dx isrationalized bysubstituting
    ao

    either (at m + 1 m- + 1 , p
    AN " ccording as n
    or +
    - N
    isintegral,whether positive or negative.
    2120 2m-1 F {a ”, x“, (a + bx ) }dv,when n is either a
    positive or negative integer, is rationalized by substituting
    (a + baasi.

    INTEGRALS REDUCIBLE TO ELLIPTIC INTEGRALS.


    2121 ( F { x, v (a + bx + cxº + dv + ex*)} dx.
    Writing X for the quartic, the rational function F may
    always be brought to the form
    P +QX
    “ P ' + QVX
    and this again , by rationalizing the denominator, to the form
    344 INTEGRAL CALCULUS.

    M + N / X , where P , Q , P ', Q ', M , N are all rational functions


    of x . ſ Mdx has already been considered (1915 ).
    CNY ( Rdz
    NXdx = 1 dx or ere R is rational.
    X Where
    By substituting x = 27991 , and determining p and q' so
    1 + y'
    that the odd powers of y in the denominator may vanish, the
    last integral is brought to the form
    p Rdy
    2122
    J7 (a + byj” + cy ")"
    R being a rational function of y , may be expressed as the
    sum of an odd and an even function ; thus the integral is
    equivalent to the two
    2123 | _ yF;(yº)dy _ F ,(yº)dy
    Jū (a + by+ + cy ) + J / (a + by' + cy“)*
    The first integral can be found by substituting vy.
    The second ,by substituting 1 , for
    for y’, can bemade to
    depend upon three integrals of the forms
    ee ns

    - dx PV1 - kl? dx
    2124
    V1 - x ? . 1 - k ? x ? JV1 - X
    dx
    J ( 1 + nx?) ✓1 - x². 1 - ka?

    By substituting p = sin - x , the above become


    do
    2125 V1 - k sin o dd ,
    JV1 - k sin ' '
    p_ dø
    J ( 1 + n sin' ) v1 - k - sin ' s
    These are the transcendental functions known as Elliptic
    Integrals. They are denoted respectively by
    2126 F (k , d ), E (k , d ), [] (n , k , b).
    INTEGRALS REDUCIBLE TO ELLIPTIC INTEGRALS. 345

    APPROXIMATIONS TO F (k , ' ) AND E (k , ' ) IN SERIES.


    When k is less than unity , the values of F (k , p ) and
    E (k, ), from the origin p = 0, in converging series, are
    3 .k + 1 . 3 .5 .k6
    F (k , =
    4 . 23 * 2. 4 .6 . 25 48 + ...
    1 . 3 . 5 ... n - 1 kn Ro .
    + G
    2 .4 .6 ... n 21
    2128 E (k , ) = 2 . 4 . 23
    « At
    T 2 . 4 .6 . 25

    + (- 1
    +11. 3 .5 ... n - 3 kn
    2 . 4 . 6 . ..n 2n - 1 Ano
    where
    sin 20 - 6 , [ n being an even integer .

    4 = sing _4sim24 + 36
    4,= sinof
    6 _6 sim + 15 sim 26 – 106,
    sin ng n sin ( n - 2 C (n , 2 ) sin ( n — 4 ) 6
    4.=sinones_ sin(
    mn=2
    - 2 )01C(
    T 1,2)sin(
    0 - 4 n=4)
    C ( n , 3 ) sin (n - 6 ) 6 .
    n - 6 9 + ... + ( - 1)}" } C (n , žn) $ .
    PROOF. — In each case expand by the Binomial Theorem ; substitute from
    ( 773) for the powers of sing, and integrate the separate terms.
    The values of F (k , p ) and E (k , o ), between the limits
    $ = 0, p = , are therefore
    2129

    (x, ) = {1+ ( )* +(1: *x +(1.3.69** +&c.}.


    2130
    E(* ) = 7 {1-(1)- -(44)5 - 1.3.5) 3 =&c.}.
    But series which converge more rapidly are
    2131
    F(* 7)= =\+^ {1+(+)* +(4:3)- +(6.3:5)* +&c.}
    346 INTEGRAL CALCULUS.

    2132
    E(1,3)= 214n){1+(1)x +(4+)2 +(1.3 )? +&c.}
    there ere n - 1 - V (1 — kº)

    2133 J / (a + bx F+ (xca) dx
    + * + uxt), when l'(a) can be ex
    pressed in the form (z - 7) +(c+ + ), is integrated by
    substituting æ + ?
    im
    Ifb is negative, and F( )oftheform (x+ 1)+(--- );
    substitute a
    2134
    F ( x ) dir
    v (a + bx + cx' + d + ext + bx + ar")
    Substitute - = %.

    Hence
    Henco sopa
    V 103 = (2013-2017
    12V7 — 2 2V7 + 2 )diz,
    !
    and the integral takes the form
    POVP- 4 Və + 2 - V = — 2 dz.
    J / 14 ( 3 – 3x) + b ( x - 2 ) + c + d} 2 - 4
    where P , Q are rational functions of % . Writing Z for the
    cubic in % , we see that the integral depends upon
    ' P d: and PQ ( x + 2 ) d :
    JvZ (3 + 2 ) JvZ (3 + 2 )
    the radicals in which contain no higher power of z than the
    fourth . The integrals therefore fall under (2121).

    F ( r ) der
    2135
    J7 (a + bx + cx' + b " + ax")*
    Expressing F ( r) as the sum of an odd and an even func
    tion , as in (2123) , the integral is divided into two ; and , by
    substituting x ” , the first of these is reduced to the form in
    ( 2121) , and the second to the form in (2134 ) with a = 0 .
    INTEGRALS REDUCIBLE TO ELLIPTIC INTEGRALS. 347

    F ( r ) dx
    2136 V (a + bx + ca ? + dw )
    Put x = yta, a being a root ofthe equation a + bx + ca ? + Ww = 0) ;
    and , in the resulting integral, substitute zy for the denominator.
    The form finally obtained will be
    ( ( P + Q Va + B:*)dz,
    which falls under (2134), P and Q being rational functions
    of z .

    2137 F (x ) dx
    ) ^ ( a + bx + + cx4)
    Expressing F ( ) as the sum of an odd and an even func
    tion , as in (2123 ), two integrals are obtained . By putting
    the denominator equal to z in the first , and equal to xz in the
    second , each is reducible to an integral of the form
    ( (P + Q Va + Bz*) dz ,
    which falls under (2121).

    2138 Svedberg - . [Expandby (2127).


    2139 Svit = P ( 136).
    Proov. - Substitute cos '. in (2138), and 2 tan -' x in (2139).
    2140
    dx
    J7(20x — a”)(6 ~) = 7
    516-— 03 ko6F(178
    1 ) or vžF(V2 *),
    according as b is > or < 2a.
    Proov. - Substitute accordingly, x = 2a sin ? q or a = b sin ' .

    - dx ja - b o ).
    2141
    ✓ ( x )(x - 6 )(x + c) ( a + c ) lato
    PROOF. - Substitute x = Q - (a , b ) sinºq, æ being < a and > b .
    343 INTEGRAL CALCULI' S.

    SUCCESSIVE INTEGRATION.
    2148 In conformity with the notation of (1487) , let the
    operation of integrating a function v , once, twice, . .. n times
    for x, be denoted either by
    So, [2.00.... 2 , . or byd-,, d_2 ...d_us)
    Je
    the notation d -, indicating an operation which is the inverse
    of dr. Similarly, since Yr, Yara 43.r, & c . denote successive
    derivatives of y , so Y - 2, Y –2x2 Y - 37, & c . may be taken to repre
    sent the successive integrals of y with respect to X .
    2149 Since a constant is added to the result of each in
    tegration , every integral of the 9th order of a function of a
    single variable x must be supplemented by the quantity
    arxn - 1 a . rin - 2
    mītn9 + ... + an -id + an = Vnd
    0,
    where an, d ,, dz ... an are arbitrary constants ,

    Examples.
    The six following integrals are obtained from (1922) and
    (1923).
    When p is any positive quantity,
    2150 L 120= ( + 1)(2 +2)...(p+n)+ 1...
    n2

    When p is any positive quantity not an integer, or any


    positive integer greater than n ,
    2151 ( - 1)"
    v nr (p - 1)(y - 2)...( - 11).2.2-- + J.. .
    When p is a positive integer not greater than n , the fol
    lowing cases occur —

    2152 September (
    22
    "logx +10
    pro

    2153 Sub = 1, *(v log x=«)+[...


    (p + (p + 1) =
    SUCCESSIVE INTEGRATION . 349

    2154 South >27"{So-._(vlog«)– } +S4 0. (p

    For the integral within the brackets, see (2166).


    The following formula is analogous to (1461-2)
    2155 Lsing }ax = e.sg}(ax– žum)+ S_0, V nc

    SUCCESSIVE INTEGRATION OF A PRODUCT.


    Leibnitz 's Theorem ( 1460) and its analogue in the Integral
    Calculus are briefly expressed by th
    2157 Dnx (uv) = (d: + 8.)"uv D -na (uv ) = (d.: + 8 )-" uv ;
    where D operates upon the product uv, d only upon u , and 8
    only upon v. Expanding the binomials, we get

    2159 D.. (wv) = ung0 + nu(n-1)x6. + ( . )W1n e)xV2v + &c.


    2160 D -nx(uv) = u_nxv — nu_(n+1)xVx + " 17 _u_(n+2)ąVzx - & c.
    PROOF. — The first equation is obtained in (1460). The second follows
    from the first by the operative law (1483) ; or it may be proved by Induction ,
    independently ,as follows
    Writing it in the equivalent form
    (uv ) = UVN - U in-
    vrt ( n + 1) | UV2z - & c. ... ... ( i.),
    Jnx J NX 1 + 1) . J (n + 2 )

    make n = l ; then
    Uv) = uv - uvrt uVor - & c . .. . . . . .... ...... ( ii.) ,
    Jæ J23 32
    a result which may be obtained directly by integrating the left member suc
    cessively by Parts. Now integrate equation ( i.) once more for x , integrating
    each term on the right as a product by formula (ii.), and equation (i.) will
    be reproduced with (n + 1) in the place of n.

    2161 | earze" = ea"(a+ d.)-*w* + 0. Or, by expansion ,


    Un. und

    2162
    | aºg" = { em –nm -- + %( +1)1 . 2 a( - 1) am-
    a / " – &c.?
    S + [ne0." .
    · If m be an integer, the series terminates with ( - 1)" (n) = am .
    350 INTEGRAL CALCULUS.

    Similarly , by changing the sign of m ,


    2163 t
    axmma , N1 ((n n. + 1) m ( m + 1) . s +
    a " 1 . 2 a *.x +2 ) Jnu

    Proof. — Putting u = eux, v = 2m in (2158), the formula becomes


    J na
    parzm = (d : + 8x) - qatx = (dream tear ox) -*q?" = eat (a + or) -". .
    Here ear is written before or within the brackets, because à does not
    operate upon ear. Observe, also, that the index - n affects only the opera
    tive symbols d , and dr, but it therefore affects the results of those operations.
    Thus, since deux produces ae" x, the operation d , is equivalent to a X , and is
    retained within the brackets, while the subject eat, being only now connected
    as a factor with each term in the expansion of ( a tox) - ", may be placed on
    the left.

    2164 ſecar"de= * {2*=** * + n(n =1)x=-==&c.}


    2165 Sum dx = cp + +1)+&c.} n + 1 2m + 2

    Proof.— Make n = 1 in (2162) and (2163).


    ( p + ) m

    2166 S 2"(loga)" =Secreta *q".


    Jnx und [Subs.loga.
    Hence the integral of the logarithmic function may be
    obtained from that of the equivalent exponential function
    (2161).
    Fór another method, see (2003–5).

    HYPERBOLIC FUNCTIONS.
    2180 DEFINITIONS. — The hyperbolic cosine, sine, and tan
    gent are written and defined as follows :
    2181 cosh x = } (e" + e- ) = cos (ix ). (768)
    2183 sinh = (e -e-*) = −i sin (i ).
    2185 tale = em - -
    er te- = - itan (ix ). (770)
    HYPERBOLIC FUNCTIONS. 351

    By these equations the following relations are readily


    obtained.
    2187 cosh ( = 1 ; sinh ( = 0 ; cosh co = sinh co = 0 .
    2191 coshºc - sinh? 2 = 1.

    2192 sinh ( + y ) = sinh ở cosh + cosh & sinh .


    2193 cosh (x + y) = cosh x cosh y + sinh x sinh y.
    2194 tanh (x + y) _= 1tanh c + tanh y
    + tanh tanh y

    2195 sinh 2 = 2 sinh a cosh .


    2196 cosh2x = coshºc - sinh? a.
    2197 = 2 cosh ” x - 1 = 1 + 2 sinh ? x .

    2199 sinh 3 = 3 sinh + 4 sinh .


    2200 cosh 3x = 4 coshºx - 3 cosh x .

    2 tanh .
    2201 tanh2 = 1 + tanhČ x
    3tanh c + tanhºa
    2202 tanh3x =
    1 + 3 tanhřx

    CO
    2203 sinh = V cosh –1; cosh = cosh +1.
    2205 cosh x - 1 cosh x - 1 sinh x
    cosh x + 1 sinh x cosh x + 1°

    2208 1 + tanhºc. sinh 2 = 2 tanh đa


    cosh x = 1 - tanh x 1 tanhº go

    INVERSE RELATIONS.
    2210 Let u = cosh x , . . x = cosh - ? u = log (u + vu — 1).
    2211 v = sinh x , . x = sinh -1 v = log ( v + + 1) .
    2212 w = tanhx, ::x = tanh-?w = įlog ( + 4 ).
    352 INTEGRAL CALCULUS.

    GEOMETRICAL INTERPRETATION OF tanh S.


    2213 The tangent of the angle which a radius from the centre
    of a rectangular hyperbola makes with the principal axis, is
    equal to the hyperbolic tangent of the included area .
    - P
    Proof. — Let o be the angle, r the radius, and S the area, in the hyperbola
    ai — y : = 1 or på = sec 20 ; then
    sec 20d0 = { log tan ( + + 0 ) . (1942)

    Therefore 628 = 1 +tan ; therefore tan0 = = = tanh S. (2185)

    VALUE OF THE LOGARITHM OF AN IMAGINARY QUANTITY.


    2214 log (a + ib) = } log (a’ + 6?) titan - .
    PROOF. —
    atib
    log - = itan -19 By (771).
    Phoor – log' vpenting=
    ( a* + 63) logV =itaua . By(771).

    DEFINITE INTEGRALS.

    SUMMATION OF SERIES BY DEFINITE INTEGRALS.


    2230 f (x )dx = [f (a ) + f (a + dx) + ... + f (a + ndx)] dx,
    where n increases and dx diminishes indefinitely , so that
    ndx = b - a in the limit .

    2231 Ex. 1. — To find the sum , when n is infinite, of the series


    1 1 , 1 , -
    1 P lus,
    ... ... +
    nT " n + T 1 * n + 2 . ...** ntni

    detant en tanto ...+ Sam de = log 2.


    2232 Ex. 2. — To find the sum ,when n is infinite,ofthe series MO

    wit1 + winget .. . Pat n = d., then


    dx
    1+(12)*+ 1+ 4202)2+...+ 1+(male)== .17 = (1935)
    THEOREMS RESPECTING LIMITS OF INTEGRATION . 353

    THEOREMS RESPECTING THE LIMITS OF


    INTEGRATION .

    2233 [ Substitute a - 3 .
    2233 % *(x)dx = S®(a–a)dx. [Substitutoa ---

    2234 $ $(x) dx = 2[**( ) dx ,


    or zero, according as ® (@ ) = + (a — m ) for all values of x
    between 0 and a .

    If • (a ) = $ ( - a ), that is, if y (« ) be an even function


    (1401) for all values of x between 0 and a .

    2236 $.$(x) dx = S*®(6)dx = 18 $(x) dx .


    co со

    If q (x ) = - ( - x ), that is , if o (a ) be an odd function


    for all values of x between 0 and a .

    2238 L *(x)dx= -S®(ə)dx and SJ - a $(x)dx = 0.


    B - 1 min edo=- ainade and I ainsade =0.
    Given a < c < b, and that æ = c makes $(x) infinite, the
    value of $ (@) dæ may be investigated by putting u = 0,
    after integrating, in the formula
    ( c - u

    2240 $ (x)dx = f****( )dx + 8° $ (x)dx.


    If the function $ (a ) changes sign on becoming infinite , this
    expression , when u is an indefinitely small quantity, is called
    the principal value of the integral.
    2 z
    354 INTEGRAL CALCULUS.

    Ex.- L =Liam + = - * * * - +2 = 0,
    which is the principal value. If, however, u be made to vanish, the expres.
    sion takes the indeterminate form 00 – 00 .
    it ( 2 ) de
    2241 Given a < c < b , the integral . 30u will always be
    Ja (x - C )"
    finite in value while n is less than unity .
    Proof. — Let u in (2240) be taken so near to c in value that y ( r) shall
    remain finite and of the same sign for all values of x comprised between
    cĒM. Then the part of the integral in which the fraction becomes infinite,
    potu multip
    and which is omitted in (2240), will be equal to (x dx _ mu
    - c)» lt ipli
    lied by
    by
    a constant whose value lies between the greatest and least values of (x )
    wbich occur between (c - u ) and t (c + m ). By integration it appears that
    the last integral is finite in value when n is < 1 .

    2242 | f(x )dx = (b — a ) f {a + 6 (6 — a)},


    where 0 lies between 0 and 1 in value.
    The equation expresses the fact that the area in (Fig . 1901), bounded by
    the curve y = f (x ), the ordinates f ( a ), f (b ), and the base b - a is equal to
    the rectangle under b - a and some ordinate lying in value between the
    greatest and least which occur in passing from f (a ) to f (b ) .

    If 4 (x ) does not change sign while æ varies from x = a to


    x = b,
    2243 f(x)y (x ) dx = f(a + (6 — a)} ( x ( )dx .
    2244 If ¢ {x, 5.) isa symmetricalfunction ofx and ,
    ( 1 ) dr
    ox,
    0

    Proof. — Separate the integral into two parts by the formula


    and substitute – in the last integral.

    METHODS OF EVALUATING DEFINITE INTEGRALS .

    2245 RULE I. - Substitute a new variable, and adjust the


    limits accordingly .
    For examples, see numbers 2291, 2308, 2342, 2345, 2416 , 2425, 2457,
    2500, 2605, & c.
    METHODS OF EVALUATING DEFINITE INTEGRALS. 355

    2246 RULE II. — Integrate by Parts ( 1910 ), so as to intro


    duce a known definite integral.
    For examples, see numbers 2233, 2290, 2430 , 2453, 2465, 2484 -5 ,
    2608 -13, 2623, 2625, & c .

    2247 RULE III. — Differentiate or integrate with respect to


    some quantity other than the variable concerned ; if a known
    integral is thus obtained , evaluate it, and then reverse the
    operation of differentiation or integration before performed
    with respect to the secondary variable .
    For examples, see numbers 2346 -7, 2364, 2391, 2417, 2421 -4 , 2426 , 2428,
    2497–8 , 2502 -4 , 2571, 2575 –6 , 2591, 2604, 2614, 2017 - 8, 2632, & c.

    2248 RULE IV . – Substitute imaginary values for constants ,


    and thus transform the expression into one capable of inte
    gration .
    For ecamples, see numbers 2490, 2494 , 2577, 2594 , 2598, 2603, 2606 ,
    2615, 2641 - 2.

    2249 RULE V . — Expand the function , if possible , in a finite


    or converging series, and integrate the separate terms.
    For escamples, see numbers 2395 – 7 , 2402– 3 , 2418– 9, 2479, 2506, 2571,
    2593, 2598 , 2614, 2620, 2625, 2629, 2630 - 2, 2639.

    2250 RTLE VI. — Decompose the integral into a number of


    partial integrals , and change all these by some substitution
    into integrals having the same limits. By summing the
    resulting series, a new integral is obtained which may be a
    known one.
    For examples, see numbers 2341, 2356-61, 2572, 2638.

    2251 RULE VII. — Separate the function to be integra 'ed


    into two factors, and replace one of them by its value in the
    form of a definite integrał taken between constant limits with
    respect to some new variable. The double integral so obtained
    may frequently be evaluated by changing the order of integra
    tion as explained in (2261) .
    For examples, see numbers 2507, 2510 , 2573, 2619.
    2252 RULE VIII. — Multiply a known definite integral which
    is discontinuous between certain values of a constant which it
    356 INTEGRAL CALCULUS .

    contains, by some function of that constant, such that the


    integral of the product with respect to the constant is known .
    A new definite integral may thus be obtained .
    For examples, see numbers 2518 , 2522.

    Particular artifices not included in the foregoing roles are employed in


    2293, 2305, 2310 , 2:314 - 5 , 2317, 2367 - 9 , 2404 - 15 , 2422, 2429, 2456 , 2495 ,
    2514 , 2518, 2585, 2600, 2626 , 2635, 2637.
    Additional formulæ for integration will be found at 2700 , et seq.

    DIFFERENTIATION UNDER THE SIGN OF


    INTEGRATION .

    Let u = ( f( )dx, where a,b, and f(x) are independent


    of each other; then
    2253 = f(@) and dem = -f(a).
    PROOF. — Let u = $ (b ) - ° (a ).
    Therefore Up = $ (6) = f (b ) and Ug = – ø (a ) = - f (a ) .

    Let u = Sºf(x,c)de. Then, when a and / are inde ne n a are

    pendent of c,
    2255 v.= $ {f(x,c)} dx and « = 1"{ (x,c)}md.c.
    Proop - = {fºr(2,o+h)ds–[°s( ,c)ds} + h
    po f (x, c + h ) - f ( c. c) ale
    dx (since h is constant relatively to x ) =- pdf
    4 (.x , c ) ,
    La de

    But if a and b also are functions of c,

    2257 Si{df(4.c);de+f(0,0) –5(a,e,second


    PROOF. — The complete derivative of u with respect to c will now be
    Uc tu,be + ugac But Wp = f (b, c) and wg = - f (a , c ), by (2253– 4 ).
    APPROXIMATE INTEGRATION . 357

    INTEGRATION BY DIFFERENTIATING UNDER THE SIGN OF


    INTEGRATION .

    2258 Ex. 1.– [2"6 +dx = [(eca)nadx = d.va(cardx (2256)


    = dna (eas - ) = eat (x +da)" a"!,
    by (1464), a and æ being transposed.

    2259 Ex. 2.– [2" sinbxdx = duaſca sinbxdx.


    The last integral is given in ( 1999), putting n = 1.

    2260 Ex. 3.
    [au+=+de= f(as) dx = (ſa=dx),= (lon),a fost lo-Joga .
    INTEGRATION UNDER THE SIGN OF INTEGRATION .
    When the limits are constant,
    rya Pre

    2261 Jxin " f( ,y) dxdy = Dyi" Ixi


    ** f(x ,y) dydr.
    That is, the order of integration may be changed .
    But an exception to this rule occurs when , at any stage
    of the integration , an infinite value is produced . The double
    integrals above will not then have the same value.

    APPROXIMATE INTEGRATION .
    BERNOULLI'S SERIES.
    2262 S f(x)dir = af(a)- f(a)+ 12 38"(a)–...
    nn sn

    + 1:3) ,* pn-(a)+ 1=1); 1.2*5"(a)d.o.


    PROOF. - Integrate successively by Par & c. Or change
    f'(x ) into f (x ) in (1510).

    2263

    (•)de= ( –a)f(a)+ 4 7*:"(a)+ 0 "(a)+&c.


    358 INTEGRAL CALCULUS.

    Proof. — Put f (a ) for $ ' (u ) in the expansion of the right side of equation
    (1902) , by Taylor's theorem (1500 ) ; viz.,
    f (x ) dx = q (b ) - (a ) = (b - a ) 6 '( a ) + (b(0;1 -. 2a ) ? 0" (a) + & c.

    The following is a nearer approximation :


    Let (6 — a ) = nh, where n is an integer ; then
    2264 ( f(x)dx = h{1.50)+1f(a)+f(a+ h)+...+916 –h)}
    B (50m)=f(a)}+** {$"m)=f"()}
    * {f"(6) — fº (a)} + & c.
    Proof. — Expand (enido — 1) : (emaz – 1) by ordinary division, and also
    by (1539) , and operate upon f ( x) with each result ; thus, after multiplying
    by h , we obtain , by (1520 ) ,
    h{f(x)+f(x+h)+f(x+ 2h)+...+f(x+n – 1h)} h. 21 2B.
    = { f ( x + nh) - f (20 212 1 . 2 . 3 .4 43.
    which expression , by changing x into a and æ + nh into b, is equivalent to
    nh

    the above,since d --{f(x + nh) - f(x)} = ( f(x)do.


    2265
    'a + h
    (" T"fordr _ 2hf(a , h) , 3h+f'(a - 2h ) , 4hf"(a - 3h)
    " 12 . T 1 . 2 .3 + 1 . 2 . 3. 4 F & c.
    Proof. - Assume x = cehr. Then is equal to the coefficient of
    in the expansion of - Thus

    mart
    x = c + 2h1 35 ,31 432
    + 1 2 3 + 1. ) 3.4
    Substitute d , for x , and therefore de har for c in this equation, and operate
    with it upon Jo (x + h) dx, employing (1520) . Finally , write f (x ) for (x ),
    and a for x .

    A more general result, obtained in the same way, is


    Pa

    2266 f ( x ) dx = nhf ( a - h ) + n (

    + (n + 3) f" (a – 3h) + & c.


    THE INTEGRALS B (1, m ) AND T (n ). 359

    THE INTEGRALS B (1, m ) AND r (n ).

    EULER'S FIRST INTEGRAL B (1, m ).


    The three principal forms are
    2280 1. B (1,m ) = 22- (1 — 2 )» –1dx = B (m ,]). [By(2233)
    2281 II. B(1,m) = 101. )+mdx. (Bysubstituting 1-2inI.
    - 1

    2282 III. B(1,m )= 12*12 mdx.[By substituting 772 inI.


    + m

    When I and m are positive, and I is an integer ,


    1(1- 1)
    2283 B (1, m ) = min)
    If m be the integer, interchange l and m . If both l and
    m are integers, the forms are convertible .
    Proof. — Integrate (2280 ) by parts, thus,
    ( 2 - (1–2)»--dx = ? ( 2 -2(1–2)».
    m Jo

    Repeat this step successively.

    EULER'S SECOND INTEGRAL T (n ).


    n being a real and positive quantity , n - 1

    2284 f(n) = 1* -*-*-idx = ['(log )* *dx.


    The second form being obtained by substituting e -* in the first.

    2286 f (1 ) = 1, r (2) = 1.
    2288 r (n + 1) = nl (n ) = n (n - 1) ... (n — r ) T (n - r ).
    2290 r (n + 1) = in , when n is an integer.
    PROOF. — By Parts, e - *xn -'dx = net
    " Jon+ Jo
    e - *x "d:c .
    The fraction becomes zero at each limit, as appears by (1580 ) , differentiating
    the numerator and denominator, each r times, and taking r > n .
    360 INTEGRAL CALCULUS.
    n - 1
    1

    2291 *e* * de = T(m)= * *(log 1)**de.


    PROOF. - Substitute kæ in the first integral, and so reduce it to the form
    (2284) . In the second integral, substitute – log x , reducing it to the former,
    When n is an integer, (2291) may be obtained by differentiating n - 1
    times for k the equation Jo
    e-kidx = 7

    When u is an indefinitely great integer,


    1 . 2 . 3 .. . u
    2293 f (n" ?) = n (n + 1) ... (n + x )'
    Proof.— log1 = lim . (1 - x+) (1583). Give itthis value in (2285),
    and then substitute y = xx ; thus, in the limit,
    r (n) = v*- |'(1 – )»-'da = v* (*z -!(1 – 3)---dy. Then, by (2283),
    changing u finally into p + 1 in the fraction.

    log F (1 +n) IN A CONVERGING SERIES.


    2294 Let n be < 1, u an indefinitely great integer, and
    S, = 1+ 2 + 3 +... then
    log (1 + n ).
    2295 = (logu - S )n + {Sın?— } Synº + 1S.n — }S:nº+ & c.
    2296 = }log in mm +(log – S,)n –}$;n – }$x*— &c.
    2297 = {log innfly) +n (1 +log - - s )
    . + (1- S.)+ (1– 8.)+ &c.
    2298 = flog ini . -flog +-4227843n
    - 0673530ne — 0073855n5 – 0011927n ? — 0002231nº – & c.
    Hem 1 . 2. 3 ... u
    Proof . — By (2293), 1 (1 + n ) = (n + 1)(n + 2) ... (n + w )'
    since n tuti
    thati = 1,when k = c . Whence
    O

    log P(1+ n) = ulu –1(1+u)–1(1+ % )–1(1+ % )-...-1(1+ ).


    THE INTEGRALS B (1, m ) AND I (n ). 361
    Developing the logarithms by (155), the series (2295) is obtained . The next
    series is deduced from this by substituting
    Sonº + 18,n * + žSono + 1 Son8 + & c. = log na — log sin nn ,
    a result obtained from (815) by putting 0 = nt and expanding the logarithms
    by (156 ) .
    The series (2297) is deduced from the preceding by adding the expression
    0 = - log
    61 +- 1 + n 3
    & c., from ( 157).

    2305
    ww _ r ( ) r (m )
    B (l , m
    r (l + m )
    PROOF. — Perform the integrations in the double integral
    Jo Jo
    e-$(y+1)gel+m -1y2-1dxdy,
    first for x, by formula (2291), and then for y, by (2281), and the result is
    B ( , m ) r (l + m ) . Again perform the integration, first for y , by (2291), and
    the result is r (i) T (m ), by (2284).
    NOTE . — The double integral may be written by the following rule :
    Write xy for x in r (1), and multiply by the factors of r ( m + 1 ). We
    thus obtain e- ( y) -lxerºx” da day,
    Jo Jo
    which is equivalent to the integral in question .
    2306 B (1, m ) B (l + m ,n ) = B (m , n ) B (m + n , l)
    = B (n , l) B (n + 1 , m )
    2307 r (1) T (m ) (n ). . [By (2305).
    T (l + m + n )
    2308 *21–1(a — x)m-1dx = a +m-1B(1,m ). [Substitute *
    If p and q are positive integers , p < 9 , and if m = 4P + 1.
    por
    - 29
    xulp
    2309 Schengende = mysinma
    Jo 1 + x²4 sin mai
    poo 2 2p п
    2310
    2310 Sli
    Jo 1 m- 2129,dx = 2,tanma
    2qtan MT
    PROOF. - (i.) In (2023) putl = 20 + 1, n = 29, and take the value of the
    integral between the limits # oo . The first term becomes log1 = 0 ; the
    second gives the series
    7 . lt
    sin
    . 317 ( 29 – 1 ) lan 7
    Il 29 29 29 q sin ma'
    by (800 ). The integral required is one-half of this result, by (2237).
    (ii.) (2310) is deduced in a similar manner from (2021).
    3 A
    362 INTEGRAL CALCULUS.

    com - 1

    2311 l. do = sind nur SA de tal ma MT

    where m has any value between 0 and 1 .


    Proof. – By substituting 2,29 in ( 2309 -10). Also, since m = " T ', by
    taking the integers p and q large enough, the fraction may, in the limit, be
    made equal to any quantity whatever lying between () and 1 in value.

    2313 T (m ) [ (1 – m ) = sin m being < 1.


    Proof. — Patl+ m = 1 in the two values of B (1, m ) (2282) and (2305) ;
    thus, r (m ) (1 - m ) = - dx = sin
    - "ma, by (2311 ) .
    Jo 1x +

    2314 Cor. 1 (9) = V1.


    The following is an independent proof:
    r(3) = {** ****do = 2 e dy = 2[ e dz.
    Now form the product of the last two integrals, and change the variables to
    r, 0 by the equations
    y = r sin 01 from which , by (1609), dydz =_ ";d (12) drdo = rdrdo. Hence
    % = r coses ! !

    {r (3)}" = 4 | e-us="dydr = 4 ( 1" -+*rdrdo = a;


    Jo Jo Jo Jo

    the limits for r and 0 being obtained from


    que = y +z", tan 0 =

    2315 r(4)r( )re*)...(4-1) = 27


    Proof. — Multiply the left side by the same factors in reversed order, and
    apply (2313) thus

    -Cr(1- 1 ( ) r (1 - )...1 (",??) r (1-" ; ?)


    2 ** (n - 1)= = by (814).

    2316 „I'mor(») ( + ) .. (x+";}) = 127)**


    THE INTEGRALS B (1, m ) AND I (n ). 363

    Proof . — Call the expression on the left (x ) . Change x to x + r, where


    p is any integer, and change each Gamma function by the formula
    T ( x + r ) = 21" T ( ) (2288 ). The result after reduction is o ( r ). Hence
    g (x ) = Q (x + r ), however great r may be. Therefore y (x ) is independent of
    2 . But, when x = - , (a ) takes the value in question by (2315 ). There
    fore 0 (sv) always has that value.
    The formula may also be obtained by means of (2294 ).

    NUMERICAL CALCULATION OF F (x ).
    2317 All values of f ( x ) may be found in terms of values
    lying between r (0 ) and r (1 )."
    When x is > 1, formula ( 2289) reduces f (x ) to the value
    in which x is < 1 ; and when a lies between 1 and 1, formula
    (2313) reduces the function to the value in which a lies
    between 0 and i .
    Values of r (a ), when x lies between 0 and 1 , can also be
    made to depend upon values in which a lies between į and } ,
    by the formulæ ,

    2318 1 (x ) = 21-2x / I (2.r )


    21="comm(171)
    Proof. — To obtain (2318 ), make n = 2 in (2316 ). To obtain (2319),
    put m = į (1 + x) in ( 2313 ), and change æ into ji in (2318 ), and then
    eliminate 1 ((1 +2 ).
    ).

    Methods of employing the formule --


    2320 (i.) When æ lies between ſ and 1 , reduce f (x ) to
    r (1 - x ), by (2313 ).
    2321 ( ii.) When a lies between and , reduce by ( 2
    the limits on the right of which will then be į and š .
    2322 (iii.) When a lies between 0 and 1 , reduce by (2318 ) ;
    r (Ita ) will then involve the limits and , and will be
    reducible by case (ii.)
    If 2x is < d , reduce 1 (2x) by (2318 ), writing 2a for X . If
    this gives 4x < 6 , reduce again by the same formula , writing
    4. for x , and so on .
    364 INTEGRAL CALCULUS .

    2323 The figure exhibits the


    curve whose equation in rect
    angular coordinates is y = f (x ) .
    Let the unit abscissa be OA =
    AB = 1. Then the ordinates
    AD, BC are also each = 1 by
    (2286 – 7 ) .
    The minimum value of f (x )
    is approximately0.8556032,cor
    responding to å = 1:4616321.
    The values of log I (a ) in
    the table at page 30 correspond
    to ordinates taken between
    AD = T ( 1 ) and BC = T ( 2 ) .

    INTEGRATION OF ALGEBRAIC FORMS.


    Pixl -itqm - 1
    2341 Jo (1 + x )?+ m ** dx = B (1, m ).
    Proor. - Add together (2281) and (2282 ). Separate the resulting in
    tegral into + 1., and substitute in the last part.
    * 1 x ? – 1 ( 1 - x )m - 1 B (1 , m )
    2342 11tm
    a " (1 + a ) [Substitute x +ax
    ( x + a )' + m x ta'

    n2 - 1

    2343 92– 1–2» ---dx = 1 BC ..). [Substituto 2".


    n in

    2344 ( ' _ 22-1( 1 — ~ )m -1


    +mdx = B(1,m ).[Subs.(a–652+
    Jo {ax + 6 (1 - x)}l+mºd
    The integral is also equivalent to
    sin2 –18 cos2m - 10
    40, and similarly in other cases.
    i+mdo,
    J. (a sin ’ 6 + b cos 6)/+m
    po pre - 1 dx 1 be
    2345 Substitute
    Jo ( a + bx )" + 1 = nalni at bæ
    INTEGRATION OF ALGEBRAIC FORMS. 365

    poo xn -1 dx 1(m -1) 1


    2346 Jo (a + bx )" +n + n (m ) aminº
    Proof. — Differentiate (2345) m - 1 times for a . (2255)
    m + n - 1

    2347 m < 1.
    So 27 **dx = C(m,n)sin mer
    n + 1

    Proof.— Substitute x = ay in (2311),and then differentiate n times for a .


    poxm - 1de a MT * xm -1de T 4
    cot
    т .
    ec
    Jo 1 + x " n n Jo 1 - X" n n
    where m and n are any positive quantities, and m is < n .
    Proof.— Change m into - n (2311- 2), and then substitute an .

    When n is positive and greater than unity,


    Osea

    2350 Seit n

    cosech. She is a cot...


    Proof.— Substitute am in (2311–2) and change m into
    CO

    2352 Sophie = cose Soviete = cot,I.


    OOF .
    Proof.–Substitute witzin(2350), and it in(2351).
    When m lies between 0 and 1 ,
    MTT TO

    2354 S. - sec mon = - tan mening


    Proof. — Make n = 2 and write m + 1 for m in (2348–9 ).
    Olom - 1 - X - dx =
    2356 °120m - 1 + x -m Pixm -1 — x -m
    Jo 1 + sin mt Jo 1 - X tan mơ?
    Per
    where m lies between 0 and 1.
    Proof. - Separate (2311-2) each into two integrals by the formula
    1 = 1 + 1 , and substitute x-" in the last integral.
    Jo 1

    Otherwise, in (2601) substitute e-" , and change a into ra - **.


    366 INTEGRAL CALCULU'S .

    ዝ
    0

    2358 (97de= see my de = tan


    Proof. – From (2354 -5 ) by the method of (2356).
    p .mm -1 + pn -m -1

    :199
    MT
    2360

    1
    2360 S7
    Jo 1 + 20 ="dx = cosec non
    firm - 1 - pn - m -1
    2361 Jo 1 - X" n
    dx = cot T.
    n

    PROOF. — In the same way, from (2348 – 9) .


    Pixn -m - 1 - pn +m -1
    2362
    Jo 1 + xan
    Pixn - m -1 - pn + m - 1
    2363 Jo 1 - 2PM

    Proof. — In (2601) substitute e 2n and put a = 21 * In (2595) substi


    tute e n and put a = m * .

    ( n - 11))
    T
    2364 Stations in-T) 2 - , n being an integer.
    (1 - 1)

    Proof. — By successive reduction by ( 2062), or by differentiating


    *+ a 1 , 1 - 1 times with respect to a’. (2255 )

    236
    fi wonalde T cosecMTT
    d'o ( 1 + x ) ( 1 - x )" (1 + 0 ) on. (Subs. 1tbi (2311)
    Cil 7 - 1 npna - 1)
    2367
    Jol1 - X: - 1 *..
    - 2 " )dx
    ) = log n.
    PROOF. — The value when a = 1 is log n . The difference, when the value
    is a, is (122-1-- 12r - n **
    Jo 1 - 2 JO 1 -
    which , by substituting x" in the second integral, is seen to be zero .

    F(x) being any integralpolynomial,


    2368 JP*? F (x )dx = AT, where A is equal to the constant
    -1 ✓ (1 - x²)
    term in the product of F(*) and the expansion of ( 1 - 11-) i
    n in
    LOGARITHMIC AND EXPONENTIAL FORMS. 367

    PROOF. — By successive reduction by (2053), we know that


    [ F (x ) dic -
    14002001 = $ (x) (1— xº)+ 1 vez ..........(1),
    where o (2 ) is some integral polynomial and A is a constant. Therefore the
    integral in question = AT . To determine A write the last equation thus,
    si (1- ) *do =to$(a)(1- 2 ) + 11 (1- ) da.
    Expand each binomial; perform the integrations and equate the coefficients
    of the two logarithmic terms in the result.

    F (x) being an integralpolynomial of a degree less than n ,


    2369 S ) du = m -1 2 + {F(c)log – }.
    Peoop – 15.400) dea in-II. ( 4 )}d . (255)
    Bat = f (x,c) + *_ ),where f is of a dimension lower than n – 1
    (421), and therefore din -1) f ( x, c ) = 0. Hence the integral on the right
    in - l ph F ( c ) , dn b- c)
    don-1 a wc log 6 + 0

    INTEGRATION OF LOGARITHMIC AND


    EXPONENTIAL FORMS.

    ( 12P dx
    2391 %*x®log x dx = ( -p 11 = log ( p + 1 ) .
    + 1) Jo log x
    Proof. — These are cases of (2292). Otherwise ; to obtain the first
    integral differentiate , and to obtain the second integrate , the equation
    | zo da = 1 , with respect to p (2255 and 2261).

    2393 (log x)"dx = (- 1)" (0r (n+ 1)*+1*


    + 1)

    ProoF. - See (2292). Otherwise, when 1 is either a positive or negative


    integer , the value may be obtained, as in (2391), by performing the differ.
    entiation or integration there described, n times successively, and employing
    formulæ (2166 ) , and (2163 ) in the case of integration.

    ( + XP -- XP log .lt
    2394 9+ 1
    [ By (2392) .
    Jo log x
    368 INTEGRAL CALCULUS.

    (log x ) 2n -1 22n – 2 po 22n - 1


    239 - dx =
    Jo 1- X do e

    f (log x )2n -- 1 dx = 22n - 1 - -1 B . , 721 = x 2n - 1


    2397
    Jo 1 +x 2n e" + 1
    Proof. — Expand by dividing by 17x, and integrate by (2393) ; thus

    +
    p (log x )2n -1 , = -
    - dx
    Jo 1-2
    ſ'(log
    J . 1 +7)*-*dx
    2 = - |2n –1 (1– 2 + - + &e.).
    The first series is summed in (1545) . The difference of the two series mul.
    tiplied by 221 -' is equal to the first ; this gives the value of the second series.

    2399
    Stone de = (10g(1)-- = -1-1 -1-3 - 60.= - .
    pilog x dr = - 1 + 02
    2400
    Jo 1 + x
    +

    Proof. — As in (2395– 7 ), making n = 1.


    The series (2399) may also be summed by equating the coefficients of 0°
    in (764) and (815) .

    2401Slovej dv = - 1- 3 - 1 - -&.=
    Proof. — The integral is half the sum of those in (2399, 2400).
    pil , ltr
    2402 Jo[' log(1+
    ir 1)dx = Jo X
    log
    1 - r
    dr =

    Proof. — Expand the logarithms by ( 155) and (157 ) and integrate the
    terms. The series in (2400 - 1 ) are reproduced .

    2404 Let ( log (1 - x ) dx = $ (a ), a being < 1.


    Substitute 1 - x = y ; therefore, writing 1 for log,
    • (a ) = p ly dy - p ly dy _ pl-aly dy
    Ji-al - y J. 1 - y Jo 1 - y
    The second integral by (2399),and the third by Parts, make the right side
    = - +la1(1–a)– (**?(1-2)dy. Therefore
    JO Y
    LOGARITHMIC AND EXPONENTIAL FORMS. 369

    2405 $ (a) + $ (1 – a) = log a log (1 – a ) – mº.


    If a = }, ."log (1
    X
    -x )dx = }(log2) –
    Again, y(x) =f*?(1–a)do, :(by 2253)$*(x)= 2(1–2)...(i.);
    2.0(12 = = dx *(31)= -(1+z)•Ityda
    by(1),
    < də 1 -
    : *(1 ) = .(Io
    1 - x x (x - 1))= D'ITA + Dota 1 - 2 1 - 2 To 1 - 22

    = } {l(1 – )}?– ( ).
    Put i, for æ; then
    alat

    2406 0 (–a)+$ (142) = }{log(1+x)}!.


    pa

    Also, °( ) = - ( (1+ * + + + &c.)


    = - (*+ + 9+9 + &c.).
    Hence $(®)+®(-3)= - ] (* + * + +&c.) = {$(2).
    Eliminate $ ( -x ) by (2406); thus
    Na

    2407 + (HT +41)+ 7 +( )—+(x) = }{log (1+x)}":


    Let x',and thereforea = - ] + V5 = $ say,
    :.by(2407), 9(12)– 4 (B) = } {1(1 +B)}”.
    or $(1 –B) (B) = } (u )”; [: Bº= 1-B and 1+B = ;
    and by (2405) °(1 –B)+ (B) = 2(13) – . *',
    : $(3) = (13) – } *°and 9 (1–B) = (13) – 15-",that is,
    5 - 1

    2408 . log( = )dx=(log v5=1) -


    3 - 15

    2409 090g(=*)dx =(log Vg=!)-


    2410 Let a be > 1, then • (a ) contains imaginary elements, but its
    value is determinate . We have
    o (a ) = 1 (1 - x ) carpa 1 (1 x ) , pa pa ni + 1 (x - 1)

    3 B
    370 INTEGRAL CALCULUS .

    the integration by 2399, and 1 ( - 1) = ni by (2214). The last integral


    = xila+ * {1x+2(1- )} " = wila + ](Va®)+ ( 2(1- )
    Substitute 1 = y in the last integral,and it becomes
    $ 40,5w ay =[]?1.5)ay-50245w ay = -- -+( ).
    1 Y

    Hence , when a is > 1 ,


    2411 $(a)+ ® (a ) = - * +wi loga + }(loga)".
    If a = 2, this result becomes,by employing (2405 ),
    2412 Selog(1-x)dx
    Jo X = -- +wilog2.

    2413 Let (w)= So 2, logit . dv.


    Therefore + (x)= ( + ), d:4 (173)= TK
    therefore + (x)+ + (17%) = ( (+1+ 2 + 1 :)dx, therefore
    2414 4(x) +4 (17% ) = {log log 7 + x.
    The constant vanishes, by ( 2403) and (2401), putting x = 1.

    Let x = ! s , and therefore a = 2 - 1; then, by (2414 ),


    2 - 1

    2415 1 + x dx
    2415 L og
    31 - x x = =}{log(72–13°.
    2416 Slog 1 *) dx = log2.
    Proof . – Substitute o = tan -Px ; then y (2233) ,

    JO F1(1+ tang)dø = 1*2{1+tan(5 –c)} $ = 1*21+kang dig.


    2417 By differentiating or integrating the equations (2341)
    to (2363) with respect to the index m , the integrals of func
    LOGARITHMIC AND EXPONENTIAL FORMS. 371

    tions involving log x are produced ; thus, from (2356 ), by


    integrating for m between the limits i and m , we have
    pirm -l — X “ dx = log tan MTT
    2 :
    2418 (1943)
    Jo ( 1 + x ) log x
    Otherwise , this result may be arrived at by forming the expansion of the
    fraction in powers of x , and integrating the terms by (2392 ) ; the reduction
    is then effected by 815 -6 .
    In a similar manner, we obtain themore general formula
    10
    2419 p (209 – 1+ xP)20"log
    - ) dx com n + p _ m + 2p n + 3p
    x m + p n + 2pm + 3p "

    ? xm - 1 - mu - m - 1 MT
    2420
    Jo ( 1 + x " ) log x
    dx = log tann:
    PROOF. — Integrate ( 2360) for m from 2n to m .

    2421
    | XP — X " + wf(r - p ) log x a .
    (log x )
    = (p + 1)log(p + 1) – (2 + 1) log (r + 1) + (r - p) {1 + log (q + 1)}
    Proof. — Integrate (2394) for p between the limits r and p .

    2422 (q — r)xP + (r - p) x + + (p — 9) x "dr


    (log x )
    = log {( p + 1)(p+ 1)(q=r)(g + 1)(4+1)(r-p)(r + 1)(-+1)(p-9)} .
    Proof. — Write (2421) symmetrically for r, p ; p , q and q, r. Multiply
    the three equations, respectively, by q, r, P, and add, reducing the result by
    (2394 ).

    2423 plog (1 + aⓇx®) dx = log (1 + ab).


    Jo 62 + x
    PROOF. - Differentiate for a , and resolve into two fractions. Effect the
    integration for x , and integrate finally with respect to a .

    2424
    1 log (1+ 2)10g(1+ ) =25C + 4)log (1+ 4)=27
    Proof. —In (2423) put a = 1, and substitute 1 = y; multiply up by b,
    E

    and integrate for b between limits 0 and , a nd in the result substitute by.
    372 INTEGRAL CALCULUS.

    e - k .c ?
    2425 [ Substitute
    2425 | e-**dx = į VE
    JO
    [Substitute kæhe.
    ?.

    2n + 1
    2426 e -kr® 22n dx _= 1. 3 ... (2n
    k 2 VT.
    2n + 1
    Proof. — Substitute ke?. Otherwise , differentiate the preceding equation
    n times for k .

    2427 ( * c" de = loga-logo.


    2428 1645 _60–6)e^)dx=e-b+logowanie
    Proof-—Making n = 1in(2291),(**-*rde= 1 Integrate this for a
    JO

    between limits b and a to obtain (2427) ; and integrate that equation for b
    between limits b and c to obtain (2428).

    2429 Lemon _(4-6)e-* dx=a+b+blogg


    Proof.— Make c = a in (2428 ).
    Otherwise. - Integrating the first term by Parts, the whole reduces to
    re-at - e - bx7 pa e -ar - e -bx
    L & to
    The indeterminate fraction is evaluated by ( 1580) and the integral by ( 2427).

    2430 1° e-b mode


    - e-** -4_ (a — b)e e-ee -_ (a − 20
    b)*e-a* dx
    = i {a’ + 362 - 4ab+ 26*(loga - logb)}.
    Proof.--By twosuccessive integrations by Parts, ſx-edz, &c.,
    * * * ** dla = - * + combe*" + ][*"** *d* *dx.
    Also

    Substitute these values , and make e = 0 . The vanishing fractions are found
    by (1580 ) , and the one resulting integral is that in (2427) .
    In a similar manner the value of the subjoined integral may be found .
    INTEGRATION OF CIRCULAR FORMS. 373

    2431
    pose - bx ear b) e- ax
    (a - be - as (a − b) e- 4.* ( a − b) e-at? dx .
    1 . 2 .x2 1 . 2 .3 x3

    INTEGRATION OF CIRCULAR FORMS.


    NOTATION. — Let arm signify the continued product of n
    factors in arithmetical progression , the first of which is a , and
    the common difference of which is b, so that
    2451a ," = a (a + b)(a + 26) ... {a + (n - 1)6 }.
    Similarly , let
    2452 . a{") = a (a − b)(a − 26) ... {a - (n - 1) 6 }.
    These may be read , respectively, “ a to n factors, differ
    ence b ” ; “ a to n factors, difference minus b .”
    sin ” xdr - n - 1 rin
    2453 'o n Jo
    " sin " - 2 x dx .

    PROOF. — By (2048 ) ; applying Rule VI., we have, by division ,


    sin"xdx = (sin*-*zdo -ſsin*-** cosºuda,
    sinn -1x cos a I sin " ad.x .
    sin ”- x cos x dx = $ n - 1 n - 1J
    1 ſt
    Therefore sin ” -? x cos x dx =n - 1 J.
    sin ”xdx.
    The substitution of this value in the first equation produces the formula .

    If n be an integer, with the notation of (2451),


    1 (n )

    Jo

    Proof.— By repeated application of formula ( 2453) .

    Wallis's Formula . — If m be any positive integer, we have

    2456 2m (423 }'> ; >(2m -1){ } }".


    (m ) (m
    374 INTEGRAL CALCULUS.

    And since the ratio of these limits to each other constantly


    approaches unity as m increases, the value of either of them
    when m is infinite is in .
    Ex. - With m = 4, f. lies in magnitude between
    22 .42.69. 8 22. 42.62.7
    1 .32.52. 72 and 1 . 32.52.72
    Proof.— Put 2m = n , then
    sina-lædx, sin " x dx, and sin " - xdx
    n Jo
    are in descending order of magnitude ; the first and second because sin x is
    < 1 ; the second and third by (2453 ) ; then substitute the factorial values by
    (2454 -5 ).

    2m - 1 T

    2457 "tan* - podø = 2 sin ma [Subs.a= tanº in (2311).


    IT

    2458 ["sin"xdx= 2 "sin®æde. [By(234).


    TT

    2459 "sin" x cos”xdx = 2 *"sin" x cos”xdx or zero,


    co

    according as p is an even or odd integer.


    acc n even
    [By(2234).
    2461 S "sin"x cos'wdr = {" sin”rcos"vdx. [By(283).
    2462 ("sin "x cos”zde = }B (" +1,2
    2 +2 1).[Subs.sinte(2280).
    Let either of the integers n and p , in (2461) , be odd , and
    the other either odd or even ; thus , let n be odd and
    = 2m + 1 , then
    2463 sin2m + 1 x cos x dx =_ 25m )
    (2451)
    JO (p + 1 )cm +1)
    Proof .-- Transposing the indices by (2461), we bave, by Parts (2067),
    "sinºx cosểm•«dx = p27,
    + llo"sin”** cos**- ædæ.
    By repeating the reduction , the integral finally arrived at is
    2464 I sin 2m x cos x læ =
    1 + 2m + 1
    INTEGRATION OF CIRCULAR FORMS. 375

    If both the indices are even , then


    2 12
    2465 sinºm X cos P x dx = - (2451)
    L"sin ” ecosøvdx = 1, 29SEK (m + p )
    (2451)
    Proof. - Reduce by Parts as before. The final integral is sinºm + 2P & dx ,
    the value of which is given at (2455).
    2466 Should either of the indices be a negative integer, the
    value of the integral is infinite, as the foregoing reduction
    shows, for the factor zero will then occur somewhere in the
    denominator.

    2467 sin nx sin px dx = cos nx cospædx = 0,


    when n and p are unequal integers.
    2n
    2469 sin nx cospx dx = - or zero ,
    2469 'o sinne cos prde = 2* orzero,
    according as the difference of the integers n and p is odd or
    even . [ By (1973–5 ).

    2470 *sinºnxdx = " cos?nxdx = 3h,


    when n is an integer .
    Proof. — Express in terms of cos 2nx, and then integrate.
    n - 1

    2472 ("sin"xdx = {"cos"xdx = L*(1–29*'dr.


    The following four integrals (2473– 9 ) all vanish for in
    tegral values of n and p excepting in the cases here specified .

    2474 S*sin”x sin nædx = ( -1) ? c(0,97")


    Jo

    when p and n are both odd, and n is not greater than p.


    2475 But if p be even , and n odd, the value is
    wuns 1 C ( p , 1) , C (p , 2)
    +) 2p-2 (n" —p - (p - 2)2 + n — (p — 4)2 ***
    :::(-1) C( P)?
    376 INTEGRAL CALCULUS.

    2476 ["sin x cos nædx = (–1)*C(p,"7 ")} ,


    when p and n are both even , and n is not greater than p .

    2477 But if p be odd and n even, the value is


    (- 1) 7 111 - C(p,1)(p — 2) + C (p,2)(p —4) _
    22 -2 ( p ’ - n ? ( p - 2) - n ' T (p — 4) — n?

    ...(-1)='C {p,B( =1}}}


    COS

    2478 cos”xcosnxdx = C(j,%>");


    when p and n are either both odd or both even , and n is not
    greater than p .

    2479 (*cos”a sin nedx,when pan is odd,takes the value


    ( 1 C (p , 1) , C (p, 2) we ?
    21 -2 {n --p2 + n - (p 2)2 + n2 – (p — 4) + & c.}'
    the last term within the brackets being
    C ( p, R :) when p is odd and C (p, 2 ) when p is even
    n even , or n odd .
    - and
    n - 1
    Proof. — (For 2474 to 2479.) — Expand by (772 -4 ),and apply (2467 -2470 )
    to the separate terms.

    COROLLARIES.— n being any integer,


    COS n V COS
    2480 ( cos"ocos nædv = 7
    "cos”e cos nxdx = On + 1

    SI

    2482 Csinºsx cos2nædx = ( –1)


    'o

    Jo
    "sin® +1 sin(2n + 1).xdx = ( - 1)".* *i

    2484 cos” x cos nx dx = P 1 cos” - ? x cos nx dx .


    pé -no
    INTEGRATION OF CIRCULAR FORMS. : 377

    2485
    JO
    cos” x sin nxdx = pé - n cosp -2x sin nxdx —p ’ - n2
    Proof.— (For either formula) By ] s « dx ; and the new integral
    of highest dimensions in cosx,by Parts, ſcos" -læ sin ædx.

    2486 " cos*- x cos nxdx = 0. " Cos* ?.x sin nxdx =in - ]
    Proof.— Make p = r in (2484–5).
    When k is a positive integer,
    2488 Jo
    cogn–2 x cosnx dx = 0,
    2489 I cos*+2k x cosnxdx = (n + 2k )-1 _ _ ~ + 2k + 10
    F 1 (k ) 21+ 2k + 1.
    PROOF. — The first, by putting p = n - 2 , n — 4 , ... n - 2k successively in
    (2484 ) and employing (2486 ). The second, by putting p = 1 + 2, n + 4 ,
    ... 1 + 2k successively and employing ( 2481) .

    When k is not an integer,

    2490 ( cos*-2 x cosnxdx = 22 -1+1 sin ka B(n - 2k + 1,k).


    - n + 1

    PROOF. — In (2706 ) take f ( a ) = a * -2%, and transform by (766 ) . The co


    efficient of i vanishes by (2239) , and the limits are changed by (2237) .
    22 23
    2491 cos” x sin n .rdr =
    + + ... + ? ).
    T
    X

    2" ' 2
    <

    Proof. By successive reduction by (1970 ), making m = n , avd the


    integral definite.

    2492 When pand n are integers, one odd and the other even ,
    I cos”.x cosnodr = ( - 1)!(p+n + 1(P) S + , with p odd ,
    (n - p ) p +1) 1 - , with p even .
    Proof. — Reduce successively by (2484 ). The final integral, according as
    p is odd or even , will be
    | cos x cosnx dir – cosants ( - 1 ) in or ita
    I cosna do — sinnt ( - 1 ) (n - 1)
    1n - i 12

    - 30
    378 INTEGRAL CALCULUS.

    COS

    2493 ( cos'u cosnxdx =1 ( sin"rcos-“xd«,


    where n and p are any integers whatever such that p - n
    is > 0 .
    Proof . — When p - n is odd ,each integral vanishes, by (2478 ) and (2459).
    When p - n is even, let it = 2k; then, by (2488 ),
    cos”+25x cos nx dx = ('"n +l (K27:)
    ) n + 24 + 1
    ( n + 22 ).*)
    (n + k )

    = \ " F2" ) - |* sinan x cosak x dx, (by 2465 ) .


    1," ? Jo
    But n + 2k = p, and by (2234 ) the limitmay be donbled . Hence theresult.
    CT
    2494 x cos” –2 x sin nx dx =
    * 2"(n - 1):
    Proof. — In (2707) , put k = l and f (x ) = xn -2. Give ein its value from
    (766 ). The imaginary term in the result vanishes, and the limits are changed,
    by (2237). Finally , write x instead of 0 .

    2495
    Tr”(cosa )sin”.edx = 1.3...(2n – 1) (cos x)cos na da.
    Proof. — Let z = cosa. By (1471) , we have
    d.n-1)=(1 –2*j*-+= (– 1)n- 1.3. .(2n — 1)sinne............(i.)
    Also, by integrating n times by Parts,
    Ⓡ $"(z)(1 –22)*-}dz = ( - 1)"( f(-)dnz(1 –z*)*++dz
    Then substitute 2 = cos x .
    = - 1.3...(2n -1){ (a)a (sinnm)dz, by (i.)
    Otherwise . --Let f (z ) = 4 , + Ajz + Az? + & c. = 8A ,zº,
    :: 8"(z) = Ep (p - 1) ...(p =1 + 1) 4 22-",
    :: f(cos x) cosnedx = £4, "cos”x cosnada
    cos e) sin ?” d:t, by ( 2493).
    1 .3 ... (2n

    dx
    2496 ( 1982)
    Jo a cos” x + lº sinºx = 2at
    INTEGRATION OF CIRCULAR FORMS. 379

    2497
    cos' xdr
    to Ta * cost sin rj2 = 407: ( Differentiate (2496) for a .

    sinº x dr
    į: [ Differentiate (2496 ) for 6 .
    (a ’ cos” x + bé sin x )2

    2499 S cercos artfain = )


    [ Add together ( 497 - 8 )

    2500 lis (cara ' com x )3 = bala +


    sin es
    cos“ x + bos' sin + ).
    dx T 15 3
    3 ntit 51).
    2501 Jo (a' cos x + b+ sin ’ x )* _= 3206 ( tu i
    ( 2500 ) and (2501) are obtained by repeating upon (2499) the operations
    by which that integral was itself obtained from (2496 ) .

    ol tan - 1 ax
    2502 atvlta ) .
    2502 Satanak,de
    Jo XV ( 1 - 2 ) = log (a+ v1+4“).
    Proof. — Denote the integral by u .

    sa Patatesvian = veltas [by(2008 )


    :u= ( 40) = .log(u+ V1+a"). (1928)
    po tan - ax
    2503 (1 + a ).
    2503 SJo X221744
    ( 1 + x ) do = log (1+a).
    Proof. – Differentiate for a . Integrate for a by partial fractions, and
    then integrate for a .

    2504 Stan a tan- log { 1+ %)*(1+ ) }.


    Proof. — From (2503) weobtain
    The the thing dea y love( +o).
    Integrate for a between limits q and co , and in the result substitute bæ .
    380 INTEGRAL CALCULUS.

    f tan - ax - tan - bx
    2505
    2505 Stan "avztan"b"de=
    Jo
    logemen
    Proof. -- Applying (2700), 0 (0) here vanishes. Also, by Parts,we have

    since bu is infinite and therefore tan -' (? in every element of the


    integral. Hence the required value is

    pa x sinx T
    2506
    Jo 1 + cos’ x
    Proof. — (i.) Substitute 7 - a , and the integral is reproduced, and is
    thus shown to be

    =2 1 cos+,dy = - (tan -'cosr - tan-'cos0) =


    (ii.) Otherwise , expand by dividing by the denominator, and integrate
    each term of the result by Parts. Employing (2478 ) we obtain the series

    * (1- 3 + 1 - } + &c....) = (By(2945).

    2507 La costa de = V1 = LTsite dv.


    Proof. — By the method of ( 2251) , putting

    the integral becomes


    e-*0*cosedxdy = 3 | | --="cos dydx (2261)
    = že (2610) = V . (2348)
    The second integral is obtained in a similar manner.

    2509 ſcos dy = 17 = "sinyjdy.


    Proor. - Substitute y’, and (2507 – 8) are produced .
    INTEGRATION OF CIRCULAR FORMS. 381

    When n and p are integers,


    por sin " x
    2510 | 2o -le -2x sin ” x dzdx .
    Jo XD 1 . 2 .. . ( p
    The integration for æ in the double integral is given in
    (2608 – 9 ) , and the original integral is thus reduced to the
    integral of a rational fraction .
    Proof. — By the method of (2251), putting
    mer
    zP - dz . [By (2291).
    rm5.cavado. [By(2201).
    2511 Ssign de = 21 [By(2510).
    sin ' x [ By (2510)
    2512
    2512 LS97.*dx= 65.(2 + 19 +9 = 1
    22 & (2081).

    2513 cas qo-scos prde= log


    Proor. - By ( 2700 ). Transforming the numerator by (673), and putting
    } ( P + q ) = a , i ( p - 9 ) = b, this becomes

    2514 sinar sin bi do = 10 at


    2515 L cos queacos pådx = ;( –9).
    Proof. — Integrate (2572) for r between the limits p and q.

    If a and b are positive quantities ,


    2516 ( sinax cosbx dx = 1 or 0,
    do X
    according as a is > or < b.
    Proof. — Change by (666), and employ (2572) .

    2518 *sin av sin bx dx = wis or


    according as a or b is the least of the two numbers.
    382 INTEGRAL CALCULUS.

    Proof. – From (2515) , exactly as in (2513).


    Otherwise, as an illustration of the method in (2252), as follows. De
    noting the integral in (2516 ) by u, we have, (i.) when bis > a ,

    fudb = l + ab+(°026= Tahun


    thatis, a El Sº sin accos b#Abda = ( sin av sin bé dé.
    JOJO
    ( 2261)

    (ii.)When )is <a, pºudb= (*7 db=


    If a is a positive quantity,
    po sinº x cos ax ,
    2520 0 ,
    2520 So sin’w cos axdx = (2 -a) or 0,
    aca
    according as a is or is not less than 2 .
    Proof. — sinº x cos ax = i sin æ {sin (1 + a ) æ + sin (1 – a) x } ;
    and the result then follows from (2318), the value of the integral being in
    thetwo cases - I = 0 and 1 - (2- 1) = (2 -a).
    2522
    psin ’ x sin ar
    2522 Lsin*
    Jo .csin
    23 av dx = or p (1- 4),
    according as a is > or < 2 .
    Proof. — Denote the integral in ( 2520 ) by u ; then, when a is > 2 , the
    present integral is equal to

    Inda= [ (2–a)da+ fºoda =


    And,whenais <2, ( uda= [ (2–a)da = 1 - 2

    INTEGRATION OF CIRCULAR LOGARITHMIC


    AND EXPONENTIAL FORMS.

    2571
    pe- ar sin rx dx = tan

    Proof. — Differentiate for r, and integrate by (2584).


    Otherwise. — Expand sin rx by (764), and integrate the terms by (2291) .
    Gregory's series (791) is the result.
    CIRCULAR LOGARITHMIC AND EXPONENTIAL FORMS. 383

    pos sin rx
    2572
    L"sin
    Jo Xra dx =
    PROOF. — ( i.) By making a = 0) in (2571).
    (ii.) Otherwise . By the method of (2250 ) . First , observing that the in
    tegral is independent of r, which may be proved by substituting rx , let r = l.
    Then (* sin x dn – f* sin x art (2* sin x 20 + [3" sin ajat
    Jo 20 Jo 22 Ја Ј 20 х
    Now , n being an integer, the general term is either
    p2017 sin - " – sin ydy _-, by substituting x = (2n - 1) + + y,
    J (PR - 1) Jo ( 2n - 1 ) + y'
    plan - 1;# sin ja = sin ydy
    J (2x - 2)
    , by substituting x = (2n - 1) + - Y;
    Jo ( 2n - 1 ) # -1t
    * sin x or
    y # ty
    y
    37 -Y 3 + T51 - Y - & c . 2 dy
    u

    = 1["siny tan dy(2013) =[ uin? dy= 5


    2573 Se nedo = e = = Sony do.
    Proof:-(i.) By(2251),putting ita = 2[20-4-20ydy (2291),
    the integral takes the form
    JO

    21/ cosrae-tod iydedy = 2 |***ye-u' cosrædyde


    COS

    = v*( *-7 dy(2614) = " "(2604).


    ens ha dic ,
    po sin ax cos ba ;
    (ii.) Otherwise. By the method of (2252), putting u =
    it follows from (2516 ) that

    Therefore e = $.Sosinca,cas de nadado= 1.It aida, by(2583).


    2575
    To nijedo =
    po sin rx
    2576 (1 - e=").
    do x (1 +x )dx =
    Proof. - For (2575) differentiate, and for ( 2576) integrate equation
    (2573) with respect to r .
    384 INTEGRAL CALCULUS.

    r (n ) sin ( n tan - 1 )
    2577 Se- zon- song (bx)dx = (a’ + 62) in cos( n tan
    10

    Proor.–By (2291), (**-**2*-*da = T(m).


    Put k = a + ib , and a = r cos 0, b = r sin 0 ; thus
    - (a ibt
    De-ash* -'de= (cosno–i sinne)F[ ,
    Jo

    by (757). Substitute on the left side for e -ibs from (767 ), and equate real
    and imaginary parts. Otherwise, as in (2259) .

    2579 2- sin (ha)dr= T(m) sin(9 ) COS

    Proof. — Make a = 0 in ( 2577 ).

    2581 - dx =
    sin /ma
    r ( m ) 2 cos ( 2
    Proof. — Put n = 1 - m in (2579),and employ
    En r (1–0) = sinna = sinmai
    Sunn

    2583 1| e - a sin bx dx = e - at cos bx dx =


    0 a : + 62 a ' + 62:
    Proof. — Make n = 1 in (2577- 8 ).
    Otherwise. - Directly from (1999), putting n = 1, and -- a for a .

    2585 (* cogº-n-10 sin"-10 sin (pe,de = (p* r— (n)f(n)sin (nt),


    ) cos( 2 )
    where n is a positive integer > 1.
    Proof.- In (2577),pat tan-1) = 8,thus,writing p for n,
    OS O sin
    le-ax 20p -1sin bxdx = " aP ?) CCOS O sin po.
    Multiply this equation by 11-'alb = a" tan " - 10 sec 6de.
    aud integrate fronı b = 0 to 00 , by (2579). Then the corresponding limits
    in the integration for A will be 0 and in .
    CIRCULAR LOGARITHMIC AND EXPONENTIAL FORMS. 385

    2587 S*in*-*o*ion(PⓇ)do = Foto( ), -1


    Proof. — Put n = p - 1 in (2585 ).

    2589 5*e* * *sions}(otan6) dx = '(n)cosmo sin } no.


    Proof. — In (2709), let 9 (x ) = cos(x tan () ;
    :.by (765), 4. = 1, 4, = - tan 4, = tango , & c.; 4, 4 , & o.
    vanishing. Therefore
    2 -322-1 cos (x tan 6 ) dx
    1 -a(1 ..1)tanº0+
    .2 q tan 0– tanº0+...= 's x xa -1 dx

    The series on the left = } ( 1 + i tan 6 ) -a + } (1 - i tan 6 ) -a , which by the


    values (770) and ( 768 ) reduces to cos al cosa 0. Then change a into n .
    Similarly , with sine in the place of cosine.

    2591 S * ****sinbe dx =tan -tan


    PROOF. — Integrate (2583 ) for a between a = a and a = ß .

    2592 Sekscos ax sin"vdx,


    Jo

    where n is any positive integer .


    See (2717 –20 ) for the values of this integral.

    em – e - m
    2593S teresin mædx = 3 e +2 cosa te-> -

    a being < 1.
    PROOF. — The function expanded by division becomes
    (eºx te-ar ) sinmx (e - * + e -372 + e -5# * + & c.)
    Multiply in and integrate by (2583). The result is
    sm is in

    ^ { (2n - 1 ) - a } + m2 T - { (2n - 1) * + aj' +m2·


    But this series is also produced by differentiating the logarithm of equation
    (2953) . Hence the result.
    3 D
    386 INTEGRAL CALCULUS.

    perx - e - ax sin a
    2594 - CoSmx dx =
    Jo e* x - e -mx e ” + 2 cos a te - m
    Proof. — Change m into id in (2593 ), thus
    r (248 + e -4 ) (26.6 — e -01) dx = _ sin 0
    ho em I – e - cos a + cos e
    Now change a into im and write a instead of 0 .

    erx - e -tared psinmxdx lem - e - m


    2595 "el
    Jo e* - e -* * ens - e-** = 4ełm + e-im
    Proof. - Make m = 0 in (2594), and a = 0 in (2593 ).

    petite - -TX sin me , _ 1 em te -km


    2597
    Jo e*x - e -T 2 em - e im
    Proof.— Make a = in (2593).
    ( x2n -1dx 220 _ 1
    2598 Jo ex - e - Tz 4n
    PROOF. — Expand sin mx on the left side of (2596 ) by (764 ). The right
    side is = – ditan (žim ) by (770 ). Expand this by (2017), and equate the
    coefficients of the same powers of m .

    poso pax - e - at 2 (en - e -m ) sin a


    2599 sin mxdx =
    muu Jo exte - . e2m + 2 cos 2a te- am

    2600 p eir
    Jo e e--amfr.
    ti t
    te COS mx dx =
    2 (em te- m ) cosa
    em + 2 cos 2a te- amº
    Proor. — To obtain (2599), pnt atin and a - successively for a in
    equation (2593), and take the difference of the results. (2600) is obtained
    in the same way from (2594 ) .

    par te- ax
    2601 - dx = sec a .
    Jo earte - fax
    Proof. - Make m = 0) in ( 2600 ).

    2602 Lºsin(cv)*dx = (*cos(ca)*dx = 2. 2


    CIRCULAR LOGARITHMIC AND EXPONENTIAL FORMS. 387

    Proof.—-By (2425) [ e-a?e dx =


    Put a = 172c.
    V2 Substitute on the left from (766), and equate real and
    imaginary parts.

    2604
    PROOF. — Denote the integral by u . Differentiate the equation for a , and
    substitatea in theresulting integral to prove that du = - 2u, and there
    fore u = Ce-20. When a = 0, we get
    Pe-rºda = 0, :: C = }/ * (2425).
    - 2ak
    2605
    2805 • ae = =.
    Proof. – Substitute x ✓k, and integrate by ( 2604 ).

    cosec

    2606 12:(* *) o[(x2+ )sin6]dæ


    - 2a coso COS /
    = YT Techno sin
    o (2asin6+ 9).
    PROOF. - In (2605) put k = cos 0 + isin 0 ; substitute from (766 ), and
    eqnate real and imaginary parts.

    2608 le-as sin2n+1xdx = ,(a’ + 1 )(1 a’. 2 .+ 33²). .....(2n(a’+ +1)2n


    0
    +

    2609
    1 . 2 . 3 ... 2n
    e -ax sinền x dx =
    ud ma(a’ + 2°)(aº + 4?)... (a’ + 2n”)
    2610
    le-ar cos2n + 1 x dx = a a (2n + 1) 2n .
    Fa’+ (2n + 1)2+ (a’+ 2n + 1°)(a’ + 2n – T")
    a (2n + 1) 2n (2n - 1) (2n - 2) al 2n + 1
    t ... +
    (a' + 2n + 14) (u ? + 2n – I^) (a + 2n - 3 ) (a ’ + 2n + 14) ... (u ' + 1 )
    388 INTEGRAL CALCULUS.

    a 2n (2n - 1)
    2611 vle
    o
    cos* x dx = q* + (2n): + (942 ) (97-421
    )
    1 a 2n (2n - 1) ( 2n — 2 ) ( 2n - 3) a ' 2n
    (a ’ + 21 ") (a’ + 2n - 2") (a’ + 2n – 4 ) + ... + (a + 2nº) ... (a ’ + 24)
    Proof Of (2608 - 11) . — Reduce successively by ( 1999 ). The integral
    part after each reduction disappears between the limits in the cases (2608 - 9 ),
    but not in the cases ( 2610 - 1 ) . See also (2721) .

    2612
    AT

    e -as cos2n +1x dr = 1 . 2 . 3 ... (2n + 1)


    ( a + 1) (a + 3 )... (a ’ + 2n
    2613
    : e - at cos2n x dx = . 1 . 2 . 3 ... 2n
    a (a ’ + 22)(a ’ + 4 ?) ... (a ’ + 2n ") "
    Proof. — By successive reduction by (1999) .

    2614 e-** cos2hvdx=


    Proof. — Denote the integral by u , then

    die = - - **2x sin 2bx dx = -12.00- %s'cos2xda = –202 ,


    the second integration being effected by parts, Se-a r 2x dx. Therefore
    log u = log 0 ;aud 0 = 0 gives C = (2425).
    Otherwise. — Expand the cosine by (765), and integrate the terms of the
    product by ( 2426) . Thus the general term is
    ( - 1)" (26 )" | e-a?s? ?rde = ( - 1) (26) * 1.3 .5 ... (2n - 1) ve.
    2n J. +) 12n 2 n +dą2n+ 1
    16121 1
    = (- 1 - , which givesthe required result by (150).
    1 . 2 ... n ' "

    2615 (2181)
    e-* **cosh 25x = vela) .
    PROOF . - Change l into ib in ( 2014 ) .
    CIRCULAR LOGARITHMIC AND EXPONENTIAL FORMS. 389

    2 2
    0
    P

    2017 ſera sin 2bxdx = W "e ".


    24

    2618 Se-*****sin(26x +}na)dx = I do (be=se).


    Proof. — To obtain (2617 ), put a = 1 in (2614 ), and differentiate for b .
    To obtain (2618 ), differentiate, in all, 1 + 1 times for b .

    2619 [ cosame***dx= loga.


    Proof. — By (2251) putting y (2291), and changing the
    order of integration, the integral becomes
    (cosa –e-4?)e-*vdydx = 1 | (e-su cosa — e=1&vp/e)dydx
    584, 2291) = log a .
    = [ (iigi-atu)dy(2584, 2291)=loga.
    2620 JO
    log (1 – 2a cos x + a”) dx = 0),
    when a is equal to , or less than, unity ; but is equal to
    2n log a , when a is greater than unity.
    PROOF. — (i.) a = 1. By (2635), since
    log 2 (1 - сos x ) = log 4 + 2 log sin fx .
    (ii.) a < 1. By integrating (2922) from 0 to * .
    (iii.) a > 1. As in (2926 ), integrating from 0 to .

    2622 log (1 - n cos x )dx.


    When n is less than unity, the values of this integral depend
    on those of (2620). See (2933 ).
    pro æ sin x dx
    2623 1. 1 - 2a cos x + q*
    a =- ā-log (1+ a), or log (1+ 2),
    according as a is less or greater than unity .
    Proof. — Integrate ("log (1 – 2a cos2 +a )dx by Parts, ſix, and apply
    (2620 ).

    2625 " cosrx log (1 – 2a cos x + °)dx = - ", or — "a ",


    according as a is less or greater than unity.
    390 INTEGRAL CALCULUS.

    PROOF. - Substitute the value of the logarithm obtained in (2922) . T


    integral of every term of the resulting expansion , excepting the one in which
    u = r, vanishes by (2467).

    pa sin x sin rx dx mar- 1 ma - ( + 1)


    2627 J. 1 — 2a cos x + a ? -
    2627 2
    according as a is less or greater than unity .
    Proof.— Integrate (2625) by Parts, S cos rą dr.
    po cos rx dx ma "
    2629 J. 1 — 2a cos æta? = -- ! a being < 1 .
    PROOF. — The fraction = cos ræ ( 1 + 2a cos + 2a’ cos 20 + 2a cos 3x + ... )
    = ( 1 - a'), by ( 2919), and the result follows as in ( 2625).

    o dr a 1 + aec
    2630 1 + 41- 2a cos cx ta' : ) 1 - ae- c'

    PROOF. — Expand the second factor by (2919), and integrate the terms
    by (2573).

    2631 log (1 –2a coscæ + a”)dx = log (1- ae-o).


    Proof.- Expand the numerator by (2922), and integrate the terms by
    (2573).

    - X sin cx dx
    10 (1 + x )(1 - 2a cos cx + a“) = 2 (ec - a):
    Proof. — By differentiating (2631) for c.
    Otherwise. — Expand by (2921), and integrate the terms by ( 2574) .

    6333
    263 [** log(1 + ccosx ) dx = - (cos -2c) {.
    COS X
    Proof. — Put a = 1 in ( 1951), and take the integral between the limits
    O and jr , then integrate for between limits 0 andno
    c; the result is
    ta
    par log ( 1 + ccosx ) 2 _ ope vi
    1
    Jo
    VI + 6
    db,
    Jo cosa
    and the integral on the right is found by substituting cos-'b.
    CIRCULAR LOGARITHMIC AND EXPONENTIAL FORMS. 391

    palog ( 1 + cco dx = a sin -1c.


    2634 COS X
    Proof. — As in ( 2633), by taking 0 and a for the limits ofa .

    2635 "Tog sin æde = {1 83 =Soviet sects des


    Proof. - sin x dx = cos . dx (2233). Add these integrals and sub .
    stitute 2x, applying (2234 ) to the result.

    2637 vlog sin ædx= log


    0

    Proor.- *z log sin« dx = ( (1 –2)• logsinada,by(2233). Equatethe


    JO

    difference of these integrals to zero.


    71

    2638 * x log sinºxdx = -nºn*log 2, n being an integer.


    Proof. — Method of (2250), CT
    In
    I xl sinº x dx = cl sinº x dx + æl sinºx dx + ... + 1 .cl sin * x dx
    JO J (n - 1) *

    = ælsin ’ « dx + | (+ + y ) I sinºy dy + ... + { (n - 1) + + y } I sin ' y dy.


    Each integral reduces by (2635 ) and (2637) ; for example,
    (a + y ) l sin ’ y dy = 2 (x + y ) sin ydy = 24 Isin ydy + 2 ylsin y dy
    do
    = - 27 ’ log 2 — * ° log ? = - 37" log 2.
    The result is – {1 + 3 + 5 + ... + (2n - 1)} ? log 2 = – nạrºlog 2.

    2639 Sein mido= 116 – + 3 = ).


    Proof. — Develope sin nex by (764) ; integrate the terms by (2396), and
    sum the series by (1539).

    2640 Lo siento deagua


    392 INTEGRAL CALCULUS.

    Proof. — Develope sin ma by (764); integrate the terms by (2398 ). The


    resulting series is = m + 22; cosec ima, by (2918 ),which is equivalent to the
    above by (769).

    264
    picos (m log x) — cos (n log x ) dx = { log 1 La
    + ma
    - log x

    2642 ( sin (m log x ) — sin (n log x ) dx = tan -1m - tan -? n .


    log x
    Proof.— Put prim and q = in in (2394), and equate corresponding
    parts . See (2214 ).

    2643
    pisin (n log x ) r = tan - n . Pivers (n logr) dx =

    1
    + 2'
    Jo log x log x
    Proof. — Put m = 0 in (2641) and (2642) .

    MISCELLANEOUS THEOREMS .

    FRULLANI'S FORMULA.

    2700 J*$(ax)= ( )dx = $(0) log + **(01)da,


    h being = 0, and the last term generally = 0.
    PROOF. — In the integral | 21" Qu dz substitute 2 = ac and > = br,
    and equate the results thus,

    pocelde-120942 = *269)drabio
    ON

    ir papa core) con de-1 0n)ds =[ 20 ds=0(0)log in


    Then make h infinite . For applications see (2513) and (2305).
    MISCELLANEOUS THEOREMS. 393

    2701 1* (an);}(b)dx =(a−b)(**(*)do


    +60)(log –a+b)–(a−b)$©:+ 84 4Chas) da,
    with h = .
    -

    Posto a {di agen as}=L sun is very


    (2257)= 1*©CP)de=9°(0)la— (),
    by making b = 1 in the proof of (2700 ). Integrate for a between limits

    and ,tras decode- de


    =(a –1) * (a)dic +f*(0){ala–616–a+ }= (a–1),8( ),
    and the wet is =f*$(ar) 60m)do- 10 colde.
    POISSON 'S FORMULÆ .
    moo (" f (a + eit) + f(a + e-it) dx = 24 ,5(a +.c),
    2702 do 1 - 2c cos x + c 1 - 02
    c being < 1 .
    ProoF. - By Taylor's theorem (1500), and by (2919), the fraction is equal
    to the product of the two expansions

    2 {f(a) + fº(a)cosæ + 1. f” (a)cos2x + 2f" (a)cos3x +... }


    and {1 + 2c cos x + 2c cos 2x + 2c cos 3x + ...}
    divided by (1 — c ). By ( 2468) the integral of every term of the product
    vanishes, except when it is of the form 2 cos’nx, and this is = 7 , by
    (2471 ). Hence the result.

    2703

    + 972 + f(ate it)(1 – c cos x) dx = + {f(a + c) + F(a)}.


    2704
    ("f(a+ ext) -f(ate-it) sin x dx = in {f(a + c) - f(a)}.
    Jo 1 - 2c cos x + c

    ( 2921,OOF: - As
    2467, in ).(2702), adding
    2470 unity to each side of (2919),and employing
    3 E
    394 INTEGRAL CALCULUS.

    ABEL'S FORMULA.
    Given that F (x + a ) can be expanded in powers of e-a,
    then
    F ( x + iat) + F ( x - iat) dt = F ( x + a ).
    2705
    1 +ť
    Proof. — Assume F (x + a ) = A + A ,e-a + Age -2a + Age-3a + & c.,
    . . F (x + iat) + F (x – iat) = 2A + 2A , cos at + 2 A , cos 2at + & c .
    Substitute and integrate by (1935) and (2573).
    Ex. — Let F (x) = , then at te = (hta)
    KUMMER’ S FORMULA.
    2706 ( f(2x cos6ei") ezikodo = sin ko ("(1 - 2 ) -1} (x2) dz.
    PROOF. — If n = xe2i8 , then xth = 2x cos 6 eit by (766 ). Substitute these
    values in the expansion of f (x + h ) by (1500 ) ; multiply by elike and in
    tegrate ; thus, after reducing by (769 ),
    I f (2x cos Ocio) e2ikedo = sin kr sf
    . ( r ) æf"(. ) . _ 7* ?"
    k k + 1 1 . 2 . (k + 2 ) j
    Again , putting h = – in (1500), multiplying by gk -1 do, and in
    tegrating, we have p* -? f ( x – xo) do = the foregoing series within the
    brackets. Equating the two values and changing o into 1 - 2, the formula
    is obtained .
    For an application see (2490 ) .

    2707 When k is an integer ,

    | $(2x cos6ei")e?iko8de = " coskm ('(1 –2)*-15(.«z)dž.


    PROOF. -- Divide eqnation ( 2706 ) by sin ka , and evaluate the indeterminate
    fraction by (1580 ) , differentiating with respect to k .
    For applications see (2490 ), ( 2491) .

    2708 If I be a function of a so chosen that

    Sxf(t, le)de = C.(°V (",0)de.........(1.),


    and if the series
    Af(«, 0) + 4,f(x,1) + A ,F (1, 2) + & c. ... = ° ( )...(ii.),
    MISCELLANEOUS THEOREMS. 395

    where o is a known function , then


    " X $ ( x)dx
    A ,Co + A ,C + A , C + & c. ... = 4 ...(iii.)
    Xf (x , 0) dx
    Proof. — Multiply (ii.) by X , and integrate from a to b, employing (i.)

    2709 If the sum of the series


    4, + 42x + 422° + A3x } + & c. ... = $ (2 )
    be known, then
    Ac + Aja + Aga (a + 1 ) + Aza (a + 1)(a + 2 ) + & c....
    e-*ga-1 $(x )dx
    e - xa - 1 dx

    Proof. -- In (2708) let X = e- –1 and f (a, k ) = x*. Then since,by


    Parts,we have ſe->mark=de= a (a + 1)... (a + k - 1) ( -*ma-'da,
    it follows that Cx = a ( a + 1) ... (a + k - 1). Hence, conditions (i.) and (ii.)
    Fel

    being fulfilled, result (ii.) is established.


    For an application see (2589 ).

    THEOKEM.- Let f(x + iy) = P + iQ ... .............


    2710 Then dų dx dy = [ Ᏸ Ꮅ dᏢ
    lx . . . . . .
    Jo dy

    Pb ( dP ,
    2711 11
    Ja Ja dy
    dx dy = -
    Ja Ja dx
    dy dx . .. . .. ... .. .
    PROOF. — Differentiating (i.) independently for x and y ,
    f (x + iy ) = Px + iQr; if (x + iy ) = P , + iQy,
    . . Px + iQx = Qv - iP, . Po = Qy and Q . = - Py.
    Hence by (2261) the equalities (ii.) and (iii.) are obtained .
    Ex. — Let f (x + iy ) = e- ( +iųjº = e- r*eve (cos 2xy - i sin 2.cy) .
    Here Pre- ** cos 2xy, Q = - e -x*et sin 2xy, therefore, hy (iii.),
    per(efucos 262—c* cos 2ax) dx = 18ct*(e- sin 2by - e-a sin 2ay)dy.
    Ja

    Put a = a = 0 , b = 00 ; therefore
    for(es"cos262–1)do=0, :.605 *cos2ada = 1- >do.
    do
    396 INTEGRAL CALCULUS.

    CAUCHY’S FORMULA.
    2712 Let ( 2**F(x4)dx = Azu, n being an integer,then
    S* *P{(x -2)}do
    = 4,+"(4,7")4,+("7)*4.+(17304 +&o.
    Proor.–In the integral| **F (-+)da= 24., substitute 2= -
    and it becomes | ( - +)*(* + ]) F {(0- 1) } = 20.pn ...(i.)
    Let the integral sought be denoted by Can, then
    Ser{( - ) } = 1** ** {(0- 1) }der = Cum
    This is proved by substituting in the first integral. Therefore by addition
    [( + 1) F {(0- 1 ) } = 2C . . .....(ii.)
    Now, in the expansion of cos(2n + 1)0 (776),put 2 cos 0 = æ + and
    2i sin 20
    , where x = eio by (768 -9 ), and multiply the equation by

    F { (1 - 2) } and integratefrom x = 0 to v = . Then, by (i.) and


    (ii.),the required result is obtained .
    2713 Ex. - Let F (x ) = e-ar, then
    Azn = | x ?ne- az' lx = 11.. 3 ... (2n -- 11)) va and 4, = 5
    2n +law +
    Therefore
    Le- a V'T n (n + 1 ) , (n - 1 ) ) , (n - 2 )(6) , (n - 3 )(8)
    2 va 1 4a 1(2) 4 43 + 113)43 % +

    FINITE VARIATION OF A PARAMETER.

    2714 Theorem (2255 ) may be extended to the case of a


    finite change in the value of a quantity under the sign of
    integration .
    MISCELLANEOUS THEOREMS. 397

    Let a be independent of a and b , and let A be the differ


    ence caused by an increase of unity in the value of a , then

    14p(x,a)dx = 46°$(x,a)de.
    2715 Ex.1. [ -u du = : *8e- dæ = .1, that is
    1 .
    Joemarler - 1) dx = a (a + 1)*
    Also , by repeating the operation,
    4"e-asdx = s».I , that is
    2716 l (e - 2 - 1)" dx = ( - 1)" in
    J. e-a (e- * — 1)" = a (a + 1 ) ... (a + n )

    2717 Ex. 2 . - In (2583- 4 ) put k for a and (2a – m ) for b, then


    e-kr A sin ( 2a , m ) x dx = A . 2a - m .. ... .. .. . ( i.) ,
    ** + (2a – m ).
    k ? + (2a – m )? ............ (ii.),
    e-k* A cos (2a — m ) « dx = 47
    In (ii.) let m = 2p , an even integer,then
    A ?! cos ( 2a - 2p ) < = cos (2a + 2p) < - 2p cos ( 2a + 2p - 2)x + ...
    ... # cos (2a — 2p ) .
    = cos 2ax [ cos 2px — 2p cos ( 2p - 2 ) x + ( 2p , 2 ) cos (2p - 4 ) ~ — ...
    . . + cos 2px ]
    - sin 2ax (sin 2px 2p sin (2p - 2) x + ...
    ... – sin 2px ],
    The coefficient of cos 2ax, in which equidistant terms are equal, is
    = ( - 1) ° 22° sina) x (773) ; wbile the coefficient of sin 2ax vanishes because
    the equidistant terms destroy each other. Therefore
    App cos ( 2a - 2p) x = ( - 1)" 22 cos 2ax sin ’' x .
    Hence (ii.) becomes
    2718 | e-k* cos 2ax sinºxdx = 6 22p
    k + (2a - 2p)
    2719 Again, in (i.) let m = 2p + 1, an odd integer, then
    42p + 1 sin (2a – 2p - 1) x = sin (2a + 2p + 1) x – (2p + 1 ) sin (2a + 2p - 1 ) x
    + C (2p + 1 , 2 ) sin (2a + 2p - 3 ) x — ... — sin (24 - 2p - 1 ) *
    = sin 2ax [cos (2p + 1) x - (2p + 1 ) cos (2p - 1) x + ... - cos (2 + 1 ) x ]
    + cos 2ax (sin (2p + 1) x - (2p + 1) sin (2p - 1) 2 + ... + sin (2p + 1) X ].
    The coefficient of sin 2ax vanishes as before, while that of cos 2ax is
    = ( - 1) " 22p + sin2p + 1 x ( 774).
    398 INTEGRAL CALCULUS.

    Therefore equation (i.) becomes


    2720
    P
    I e -kx cos 2ax sinºp +1x dx = ! 22p + T A2
    4
    +1 _ 2a - 2p - 1
    k + (22 — 2p - 1)2*
    To compute the right member of equation (2718 ), we have
    4 +(24 2p)* = * Lm +(2a + 2p)*
    2p , C (2p, 2 ) 1 1.
    K* + (2u + 2p - 2)* * ** + (2a + 2p — 4)» . * * + (2a- 2p)" ;
    Let a = 0, then the equidistant terms are equal,and we obtain in this case
    2721 A2P __ ( - 1 )P 1 .2 ...2p . 22p
    k + (2a - 2p) Tk (k + 4 )(k + 16 ) ... {kº + (2p );}
    Thus formula ( 2609) is proved.
    Similarly , by making a = 0 in (2720 ) after expansion , formula (2608) is
    obtained.

    Let p be any integer, and let q and a be arbitrary , but


    q < 2p in (2722), and < 2p + 1 in (2723).
    2722
    pocos 2ax sinar X a .
    Jo *X91+1
    +1 *dx = ( - 1) | 42p _
    + (2a?+1—
    7

    20 )?dz.
    il 24T (9 + 1 ) Jo
    2723
    * * cos 2ax sinºp2p ++11 xx dx
    29 + 1 dr

    ( - 1) ( A2p +1 (22 — 2p — 1) dz.


    = 22p +11 (g + 1 ) Jo m + (20 — 22) " *
    where A has the signification in (2714).
    Proof. — Employing the method of (2510), replace
    +1 by r (q + 1) I e -xzz? dz ,

    q being integral or fractional; therefore


    * cos 2ax sin ?” 2 2 - 1 Ip i cos 2ax sin ? ze- ** z? dz da ,
    29+1 ( + 1) J. Jo
    by changing the order of integration . Substitute the value in (2718 ) for
    the integral containing x , writing the factor z ? under the operator A , since it
    is independent of a.
    Similarly , with 2p + 1 in the place of p, we substitute from (2720) .
    MISCELLANEOUS THEORDMS. 399

    It may be shown that, whenever a > p , formula (2722)


    reduces to
    2724
    p* cos2arsin !Xan = (- 1)p+lay A27 (2a - 2p )?.
    X4 + 1
    22p+15(4 + 1)sin LT
    For a complete investigation, see Cauchy’s “ Mémoire de l'Ecole Poly
    technique,” tome xvii.
    2725 Ex. — Let a = 2 , p = 1, q = j,
    po cos 4x sin ’ x 2 +
    Joals A ’ (2a - 2 ) },
    (4) sin
    and A’(22 —2)} = (2a + 2) – 2(2a)*+ (24 –2)+ = 65 –2.4#+ 2 .

    FOURIER’S FORMULA.
    ch sin ax
    2726 J. sinx* $ (x )dx = $(0),
    when a = and h is not greater than i .
    Proof. — (i.) Let ° (x) be a continuous, finite, positive quantity , de
    creasing in value as x increases from zero to h.
    psin ax pf 211 37 . pro
    ° ( 2 ) dx = 1° + 1° + 1° + .. . + 1 .. (i.),
    - UT
    Jo sin x

    a
    being the greatest multiple of " contained in h. The terms are alter
    nately positive and negative, as appears from the sign of sin ax. The fol
    lowing investigation shows that the terms decrease in value. Take two
    consecutive terms
    (n + 1) (n + 2) 5
    sin ax
    sina P (2x ) dx, J (n º+ 1) . sisinn ax $ (x ) dx.

    Substituting x - * in the second integral, it becomes


    (n+1)7 sin ar _ o (a +
    da ,

    and since o decreases as w increases, an element of this integral is less than


    the corresponding element of the first integral.
    400 INTEGRAL CALCULUS.

    Now , by substituting ax = y, we have


    (n + 1)
    a sin ax o (2 ) dx = 1 (n + 1) + sinu "" + 1)
    sin æ $ ( 2) dw = si
    asin
    ny o ( )
    lalu
    dy = (0 ) [ "***"sina dy ...(ii.),
    P$(0)
    ya
    Jo a sin
    when a is infinite, because then « ( ) = $(0) and asin = y.
    Hence the sum of n terms of (i.) may be replaced by ° (0 ) Jo"* siny Y dy,
    which ,when n is infinite, takes the value o ( 0 ) 11 by (2572) ; while the sum
    of the remaining terms vanishes, because (the signs alternating ) that sum is
    less than the n + 1th term , which itself vanishes when n is infinite .
    ( ii.) If Q (x ) , while always decreasing, becomes negative, let ( be a con
    stant such that 0 + 0 (2 ) remains always positive while x varies from 0 to h .
    The theorem is true for C + 0 ( c ) , and also for a function constant and equal
    to O , and it is therefore true for the decreasing function o whatever its sign .
    If ö (x ) is a function always increasing in value, - (a ) is a decreasing
    function . The theorem applies to the last function, and therefore also to $ (x ).

    2727 CoR . — Hence the same integral taken between any


    two limits lying between zero and T , vanishes when a is
    infinite.

    2728 pa sin ax + (x ) dx
    Jo sin x
    = 7 { $(0)+ *(+) +$(26 )+...+ $(n - 1)+ ++ (nt)},
    when a is an indefinitely great odd integer , and no is the
    greatest multiple of a less than h . But when a is an indefi
    nitely great even integer , the second and alternate terms of
    the series have the minus sign .
    Proof. (* sin az y (x)dx = f**sin aty (x)dx + ſ“ sin 9, (x)dx......(i.),
    decompose the second integral into 2n others with the limits 0 to 1 , a to ,
    # to # , ... (2n - 1 ) # to nt ; and in these integrals put successively x = y,
    7 - y, 7 + y , 27 - y, 27 + y, ... na - y. The new limits will be 0 to è , in to
    O alternately, with the even terms negative, so that, by changing the signs of
    the even terms, the limits for each will be 0 to zł . Also, if a is an odd in
    er, sin ax is changed into ray by each substitution , so that (i.) becomes
    teger sin a sin y
    på sin ay { o (y) +
    J . sin11!y {1997TY
    (y) + " (7(* - y) + 0 (1 + y) + ... + ° (na - y) } dy
    + " since ®(a)do........(ii.)
    si a minus
    But, when a is even , the substitution of ra Fy for c makes 5sinn yy
    MISCELLANEOUS THEOREMS. 401

    whenever r is odd. The limit of the first part of (iii.) is


    { (0) + 20 (1) + 20 (24 ) + ... + 20 (n - 1) + ° (na ) }, by (2726 ).
    In the last part of (iii.) pat æ = na + y, and the integral becomes
    sin y 4p (nx + y)dy = $ (na), ifh — na is $ , by (2725).
    If h - nn lies between # and # , decompose the integral into two others ;
    the one with limits 0 to ji will converge towards įtº (NT ), while the other
    with limits ja to h - na becomes, by putting y = 7 — % ,
    S y [ (n + 1) — ]dz = 0,
    J (n + 1) 7 - sin zº
    the limit by (2727). Hence the last term of (iii.) is jᢠ(nm ). Substi.
    tuting these values, (2728 ) is obtained .

    2729 Ex.--By (2614), ( e-a*s*cos 2bxdx = the


    Put b = 0, 1, 2 ... n successively, and add , after multiplying the first equa
    tion by }, thus
    e - a x (f + cos 2x + cos 4x + ... + cos 2n x ) dx

    = {tteåte ...temp.
    The left side = { [ -a-ze sin (2n + 1) « dx, by (801),
    sins
    and, if n = 00 , becomes
    - 472 92222
    lite-re + e-4r*a* +e-Pr* *+...},by (2728) ;
    : : a {ite- n* * + e- 4na + e - omPa + ... } = te as te a? te a? + ...} .

    Put ra = a and 1 = B;therefore


    2730 Va{ite- * + e-40° + 4–90° +...}
    = VB { + e-ste-489+ e-989+ ...},
    with the condition aß = .
    1

    2731 J.0 *singX a®$ (x)dx = 4(0),


    when a is an infinite integer.
    PROOF. — The integral may be put in the form
    "sin at + (x)dx,where (x) = sin 2 ( ),
    Jo sin
    3 F
    402 INTEGRAL CALCULUS.

    therefore, by ( 2726 ), when h is $ , and by (2728), if h is > fx, the value


    is 170 ( 0 ) , since in (2728 ) 0 (7 ), • (27 ), & c. all vanish. But (0 ) = 0 (0 ).
    Hence the theorem is proved .

    When a and ß are both positive ,


    2732 * sin ae$(x)dx = 0 = 1J - *a sinX ar$(«)dv.
    2733 sin ar$ (x)dx= "$ (0).
    J - a

    Proop-—(1.) (*= 1 - 19 = 7400 – 54(0),by(2729).


    (1)ſ*.= f.+6 = (0)+ 54cm),
    by substituting - « in the second integral.
    rach
    2734 JO JOT * (x)cosuxdudx = " $ (0), when a = 0 .
    COS
    Proor. sinar *cos uxdu. Substitutethis in (2731).
    20

    When a and B are positive, the limit when a is infinite of


    pa CB
    2735 SS*+( )costucosuxdudå,
    L 'Ova
    Pa CB
    or of (*0
    **(x) sin tu sin urdudx,
    a
    un Uz

    is žtº (t), if t lies between a and B , 17° (t) if t = a , and zero


    for any other value of t.
    Prof. — When a = 0 webave, by (668), and integrating with respect to u ,
    CP sin a (z - t) w B sin a ( x + t )
    Il giacos uc cos tu dx du = ; Ja x - 7 P (x )dx + 3 Ja
    ? PO | x +t
    Io(@)de
    Jado
    B - t sin az (z + t ) dz + 1 PB + isin
    Jane Jatt 2
    ? (z — t)dz .........(i.),
    by substituting 2 = x - t and z = x + t in the two integrals respectively .
    When a is infinite, the limit of each integral is known.
    When a and B are positive and t lies between them in value, the
    limit of (i.) is T o (t), by (27:32- 3)...................... ... .. .. .. ... . .. ... ... .. . .)
    When a and ß are positive and t does not lie between them , the
    value is zero, by (2732 )......... ..... (iii )
    MISCELLANEOUS THEOREMS. 403

    Ifartin (i.), the first integral becomes = tº (t) by (2731), and


    the second vanishes as before; so that the value, in this case , is 0 (t) ...( iv .)
    The same demonstration applies in the case of (2736 ), transforming by
    (669) instead of (668).

    Hence, by (ii.), if t be always positive,


    COS
    2737 S**(x)cos tu cosuxdudx = 5, $(8)
    vo

    ) sin tu sin ux du dx.


    JO 0

    2739 Ex. - Let (de) = e-ax,


    1 e-ascos tu cosux du da = e-at.
    JOJO

    Therefore, by (2584), la du = e-at,


    which is equivalent to (2574 ), with t = 1.

    The expressions in (2737 –8 ) being even functions of u ,we have, supposing


    t to be always positive,
    0 ( ) cos tu cosuæ du dx = $ (t) = ( ) sin tu sin ux du dæ ... (i.)
    - Jo

    Replacing ♡ (c ) by 0 ( - 2 ), and afterwards substituting — 2 , these


    equations become
    • (a ) cos tu cos ux du dx = 10 ( - t)
    = - ľ ſ º(a) sin tu sin ux du de.........(ii.)
    From (i) and (ii.), by addition and subtraction , we get
    2740 ſ ° (x)cos tu cosuxdu dx = - [9 (t) +® ( –t)],
    2741 • ( a ) sin tu sin ux du dx = a ( v (t) - ° ( - ) ).

    Whence, by addition ,

    2742 U - 00 - 00
    $ (x ) cosu (t - x )dudx = 27° (t),
    the original formula of Fourier's.
    404 INTEGRAL CALCULUS.

    THE FUNCTION } (x ).
    The function d , log f (x ) is denominated * (a ).
    1 1 1
    2743 ( x ) = logu xte
    X X + 1 x + 2
    when u is an indefinitely great integer .
    Proof. — By differentiating the logarithm of ( 2293).

    2744 Cor. 4 (1) = logu - 1 - 5 .. . -


    uti'
    when = 0 ,
    = – 0:577215,66-1901,532860,60 ... (Euler).
    All other values of 4 ( x ), when x is a commensurable
    quantity , may be made to depend upon the value of 4 ( 1 ) .

    When æ is less than 1 ,


    21745 * ( 1 - x ) - 4 ( x ) = a cotTX .
    PROOF. — Differentiate the logarithm of the equation
    T' (2 ) T (1 — 2 ) = + = sin ax (2313).

    2746 y (x) +4 (x + + ) + 4(x + 3 ) + ...+4 (x + " )


    = ny (nr) - n log n .
    Proof. — Differentiate the logarithm of equation (2316 ).

    fraction . To computethe value of ¥( ) when I is a proper


    2747
    Find 4 (2 ) from the two equations
    2748 *(1- 4)- (%) = - cotla. (2743)
    2749
    *(1-4)+«(4)= 2{y(1)–log4 +cos zemlog(2versa)
    O

    +cosSur 10g(2 versiun)+ cos Spa log(2 verg )+&c.}.


    THE FUNCTION y (x ). 405

    The last term within the brackets,when q is odd, is


    cos(9 – q1)pe log (2vers(9– 91)");
    /
    2 vers

    and when q is even , the last term is + log 2 according as p


    is even or odd .
    Proof.— Equation (2743) may be written
    tiu + 2
    ¥ (x) = x +1+2 2 + pe utl ?
    Me being an indefinitely great integer. 3
    Replace a successively by = , 1 ; where q is any in
    teger ; thus
    +

    - 9 + 12 - - Lt
    9 + 1 3q + 1 '
    Sco

    7
    972 +2+
    ... ... ... ... ... ... ... . .. , .....(i.)
    ¥181) - +12– 2 – 1+13– 34–1 +14- 12-1 +1}
    4 (1) = - ' 1 + 12 – } + 1? - } +14 - + 1- )
    Now , if o be any one of the angles 27q 47 67 2 (9 - 1) + we
    1 = cos qp = cos 290 = cos 3q9 = & c......................(ii.),
    coso = cos (9 + 1) 9 = cos (29 + 1) 9 = cos(39 + 1) • = & c....... (iii.),
    cosº + cos 2® + cos 39 + ... + cos (9 – 1) 9 + 1 = 0 by (803)......(iv.)
    By means of the relations (ii.) and (iii.), equations (i.) may be written
    cos 04 (1 ) = - q cos q + cos$ 12 - I , cos (q + 1)¢ + cospl} - ,
    cos 204 (4 ) = - cos 2€ + cos2012 - , fa cos(q +2)4 +cos2013 –,
    cos 3d 3° + cos 3912 – cos ((997+ :3) ®
    4 +I 3, cos + cos 3913 - ,

    cos(9–1904 (952) = - cos(g- 1)9 +cos(9 –19012


    29 - 51 cos
    " ( 29 - 1) + cos (9 – 1 )pl} - ,
    (1 ) = - 1 + 12 - 1 + -
    Upon adding theequations, the coefficient of each logarithm vanishes,by (iv.)
    The remaining terms on the right form a continuous series, and we have
    cosq4 (1) +cos 204(3) +...+ cos(9 – 1944 (431)+ (1)
    = - 9 { cos + 1 cos 20 + cos 39 + in inf.}
    = iq log (2 – 2 cos ) by (2928 ).... . .. ... .. . .. . .. . ( v .)
    406 INTEGRAL CALCULUS.

    Let 2 * = w . Then, by giving to o in equation (v.) its different values w ,


    2w , 36 ... (9 – 1) w ,we obtain 2 - 1 linear equations in the unknown quanti
    ties y (1.), * (* )... (9 -). To solve these equations for * (? ).
    p being an integer less than 9, multiply them respectively by
    cos pw , cos 2pw ... cos (9 – 1 ) pw,
    and join to their sum equation (2746 ), after putting æ = = and n = 9 .
    0

    The coefficient of ) in the result,k being any integer less than q, is


    cos pw cos kw + cos 2pw cos 2kw + ... + cos (9 – 1) pw cos (q - 1) kw + 1.
    By expanding each term by (668), we see by (iv .) that this coefficient
    vanishes excepting for the values k = 9 - p and k = p , in each of which
    cases it becomes = 19. Hence, dividing by 19, we obtain
    + (2 - 2 ) + + ( 2 ) = 24 (1) — 24q + cos pwl(2 – 2 cos w )
    + cos 2pwl (2 – 2 cos 2w ) + ...... + cos ( 1 - 1) pwl {2 – 2 cos(q - 1) w }.
    The last term = cos pwl(2 — 2 cos w ) = the third term ; the last but one
    = cos 2pwl( 2 — 2 cos 20 ) = the second term , and so on , forming pairs of
    equal terms. But, if q be even , there is the odd term
    cos }qpw log (2 – 2 cos įqw ) = + 2 log 2 ,
    according as p is even or odd.

    EXAMPLES. - By (2748 – 9) we obtain


    2750 4 (X) = 4 (1) – 3 log 2 + ✓ ( ) = + (1) — 3 log 2 – ,
    2752 + (3) = ¥ (1) – } log 3 + , + (3) = < (1) – log 3
    2754 9 (3 ) = 9 ( 1 ) – 2 log ( 2 ).

    DEVELOPMENTS OF 4 (a + r).
    When x is any integer,
    1
    2755 y (a + x) = y (a) + a tati + a + 2 + ... + ate- 1
    Proof. — By (2289), putting n = a + 2 - 1 and r = x - 1,
    r (a + 2 ) = (a + 3 - 1 ) (a + x — 2) ... ( a + 2) ( a + 1) a f (a ).
    Differentiate the logarithm of this equation with respect to a .

    wall X X ( x — 1) , x ( x - 1) ( x - 2 )
    2756 y (a + x) = *(a)+ a - 2aa + 1 + 3a (a + 1)(a + 2)
    2 _+ & c.
    x (x - 1)(x - 2)(x - 3)
    4a (a + 1)(a + 2)(a + 3) *
    THE FUNCTION \ (x ). 407

    If x be a positive integer, the number of terms in this


    series is finite , and the value of y (a + x ) can be found from
    that of ( a ).
    Hence, by this or the preceding formula , in conjunction
    with ( 2747), the value of 4 ( N ) , when N is any commensurable
    quantity , may be found in terms of t ( 1) .
    PROOF. — Let y (a + x ) = A + Bx + Cx (x - 1) + Dx (x - 1) (x - 2 ) + & c .
    Changes into æ + 1 ; then ,
    Ay ( a + 2 ) = v (a + x + 1) - 4 (a + x ) = d , {log r (a + x + 1) - log T (a + x ) }
    = d, log (a + 2 ) (2288) = atas
    Az = 1, A . ( x - 1) = 2x , Ax (ie – 1) (2 - 2 ) = 3x (x - 1 ), & c. Therefore
    = B + 2Cx + 3Dx(x - 1) + 4E+ (x − 1)(x - 2) + ,
    A = = 20 + 2. 3Dx + 3 .4Ex (x - 1) + ,
    = 2 . 3D + 2 . 3 . 4 . Ext,

    4 .In=
    a txo 2.3.4E+.
    Put x = 0 in each equation to determine the coefficients A, B , C , D , & c.;
    1 1 1
    thus A = (a ), B :
    aa+1āma(a + 1 )'
    2.3D = : 1 = a
    -1 = 72_
    a (a + 1) ( a + 2 )'
    a + 1)
    2 2 .3
    2 .3 . 4E on .
    Ta ( a + 1 ) ( a + 2 ) a ( a + 1 ) (a + 2 ) (a + 3 )' “

    SUMMATION OF SERIES BY THE FUNCTION 4 (x ).


    2757 Formula I. 6 + 6*c+ ofzet...to fine
    = - * ( +1)+ +n + 1).
    PROOF.- Let S, denote the n terms of the series to be summed. We have
    Sw.1- 8. + ( + +1) = = [4 (* +n +2) =< (* +n+ 1)](2288)
    Sol ( +n+2)= 8.- ( +n +1).
    Hence the difference is independent of n,and therefore
    Sony Co +n + 1) = 8 - 4 ( + 1) = 5 - 4 (2 +1).
    408 INTEGRAL CALOALUS.

    2758 Ex. 1+ + 1+...+ 2 +1 =1–14(1)++ (n+3).


    ንተስ
    2759 Formula II. ( -66+ 6f2.-...- 07(2x + 1)
    ( %. + {* (* )- + ( 52 )++ (+22 + ) -+( x +1)3.
    Proof. — The series is equivalent to
    a
    + - + a Sa + a
    . . . ti
    B 6 + 2c ' 0- + 4c + . .. 70
    " + + 2nc 70 + 0 ' + 0 + 2c- "" b + c + 2nc s
    and the result follows by Formula I.

    2760 Formula III.


    1 1
    - ... in inf.
    7 ~ 6 + 6 + 6 + 20

    = bcôtes + 2.{*(*780)– (47,20)}.


    PROOF. — Make n = co in Formula II. The last two terms become equal.

    2761 Ex. 1. - In (2760 ) let b = c = 1, then


    1-1+}- + & c. = iti { ( 2) — ( ) } = log 2.
    For Ý (2) = 1 + 4 ( 1), by (2755) ; \ ( ) = 2 + 4 (1) — 2 log 2, by (2754 -5 ).
    2762 Ex. 2. - In ( 2760) let b = 1, c = 2, then
    1 - } + } - { + & c. = { + 14 (7 ) - 24

    2763 (1 + a) = * dx + 4 (1).
    PROOF. + (1 + a ) = 4 (1 w a (a - 1) , ala - 1)( a — 2 ) _ &
    ! - & c. [by (2756)
    2. 1 . 2 3 . 1. 2 . 3
    But 1-(1-a)" = a– 4(0.51)2+ a(a=!).49–2)39– &c.,
    therefore (1+ a) = ['1- (1 –a)"dx + (1).
    Substitute 1 - x in the integral.

    2764 4 (1 + a) - 4 (1 + 6) = (* * = 7° dx. [By(2763)


    THE FUNCTION y (x ). 409

    Ex. — Put b = - a ; then


    +(1+ a)–4 (1–a) = ă +4 (a)–4 (1–a) (2756) = 1 --cotxa (2745). a

    2765 Therefore $ –1 do = -- cot wa.


    *(x) AS A DEFINITE INTEGRAL INDEPENDENT OF ¥ (1).
    2766 ( ) = -S"(logu + 4 )du.
    Proof. ♡ ( )= log- - - 71 -...- + with p = 0 (2743).
    p1zx -1 _ 24 + - 1
    But Teti Tætu - 1 1 -%
    - dz,
    by actual division and integration .
    Also logr =S (1
    o )da (2367).
    :: 4 (a ) =
    1 - 2 -1 PS200 * 4 - 1 - Meizk --1 ){ dz, u = 00 ......(
    J. 1 - % 17 1 — % 1 - zle )
    Put z = yu in the first integral, therefore
    pyu - YM - 1
    p . 1 - yMallu- 4 - dy = uly - Y dy .
    Jo 1 - yu Jo 1 - yu
    Replace y by , and suppress the term common with the second integral
    i rzet +H -- 1 uzmx - 1 )
    of (i.), and we get y -% 12h )
    Pat zk = U, and this becomes
    13

    Cu ( 1 - 4)

    But when u = 0 the product (1 u ) has – log u for its limit (1584);
    and vă = 1. Hence the result.

    2767
    2767 +(x)=[ { 4- Italde
    2

    Proof. I(e)=[ ***- de; d_T(w)= [-==- logràr.


    poe- a -ere a- zaz
    But, by (2427), log % T= a
    da ,
    3a
    410 INTEGRAL CALCULUS.

    e -a e - a

    ::d:P(*)={ 1•7 **** *dzda


    020

    = [(-)|[ --- (17w ] (2201),


    wbich establishes the formula since
    dor (x ) = f (x ) = d , log r (x ) = (x ).

    2768 log f(x)= [ [6 –1e4-** ]


    Mri - -1
    2769 - C
    Jo L 1 - %
    = CCT - +1]
    2770 (x) = CI - ES
    Proof. — Integrate ( 2767) for x between the limits 1 and x , observing
    that log r ( 1) = 0) ; thus
    log I (X) = 1*{(x- 1)e-a_ (1+0)-?- (1+0)-*}sada
    log ( 1 + a )
    .
    Subtract from this the equation obtained from it by making x = 2 , and
    multiplying the result by ® – 1. We thus obtain
    -2 (1 + a ) - ? - ( 1 + r ) -27 da
    log f (x ) = - 1) (( 1 +1 a. ) -?
    (x L1 - !
    a J log (1 + a )'
    Substitute & = log (1 + r ), and (2768 ) is the result. To obtain (2769 ), sub
    stitute % = ( 1 + a ) -? Lastly , (2770 ) is the result of differentiating (2768 )
    for æ .

    NUMERICAL CALCULATION OF log T (x ).

    2771 The secondmember of ( 2768) can be divided into two parts,one of


    which appears under a finite form , and the other vanishes with x . If we put

    P = ( - 1- 1- 2-1) , and Q = E(1–0-5)'


    then log (x) = {"(P + Qe=se)de ...................(i.)
    If Q be developed in ascending powers of Ę, the terms which contain
    negatire indices are + = R say.
    NUMERICAL CALCULATION OF log f (x ). 411

    Put F(x) = (P + Re-tr)


    = [ [((0–1)- 1-2- )e-f+ (1 + )e-tes] .. ...(ii.),
    and o(a)= f ( -E)e-tedf= [ ( + - - 1 - 7)e-to...(ii)
    Then , by (i.), log l' (2 ) = F (x ) + (x ) ........................ ( iv.)
    F ( ) can now be calculated in a finite form , and a (x ) will have zero for its
    limit as x increases.

    First, to show that F (1) and ~ (1 ) can be exactly calculated .


    *(t)= ( ( - - - - ) -1
    and,by substituting je,
    0(t) = [ ( - :- )e-pastu . ..(v.)
    Again , putting x = 1 in (iii.), we have

    and, by substituting 45,


    *(1)= 1 (2-4 - 3- 3) * . ....(vi);
    po 1 1
    w (1 ) = ( 1 - e -262E ..... ...( vii.)
    The difference of (vi.) and (vii.) gives

    0 = 1 (1 -2 * 1. 4) .....(vi),
    e- e -2 e §
    since
    Tre- 1 - 0 -2 1 - e - 28
    Subtract ( viii.) from (v .), thus
    - 2

    0(1)= 116* * * _43%)a = –log2(2420).


    Also , by (iv .), F ( 1) + (i ) = 1r ( ) = ila , . . F (1) = { log ( 27 ) — 5 ...(ix.)

    F (2 ) may now be found by calculating F (2 ) - F (3 ) as follows:


    B (ii), F(x)– F(t) = ('[( -+)o-f+ (1 + )(e-te-e-*)]
    = 1* **---* **(2–3)0=*ds+ 1*2**pe= *dig
    Jo Jo

    = - x + (x - ) loga (2427 -8),


    :: F (x ) = } log (27 ) + (x — ) log x — « , by (ix .) ;
    : . by (iv.) Jog l' (x ) = { log (24 ) + (x – } ) log — to (@ )...............(x .) ;
    2772 :: 1 (2 ) = e - 8 202 – V (27 ) ew (x) ........................(xi.)
    412 INTEGRAL CALCULUS.

    When « is very large, ew (r) differs but little from unity. For o (s ) diminishes
    without limit as x increases, by the value (iii.)
    Replacing a (x) in (x .) by its value (iii.), and observing that
    log I (x + 1) = log æ + log r (2 ),
    we get log (x + 1 ) = } log (24 ) + (x + 1) log x _ x
    po 1 1 1 1 e - &x de
    " 11 - e - ........ (xii.)
    2
    Now , by (1539),
    | 1 1 1 1 B, BE B . . 22n - 2 B 323
    \ 1 - eE- E 2) 1.2 1. 2 . 3.47 1 ... 2n + 1 ... 2n + 2 '
    where 0 is < l. Also
    1o-te > = 24. 8e-** dt= 0,2, 21.
    So that equation (xii.) produces

    2773 log[(a+1)= log(26) +(x+ 4)logx=x


    B
    Bo B . OB2n
    , 2n ++2
    2
    1 . 2.x 3 . 4. 3 (2n + 1) (2n + 2) 2:2n + 1°
    s, hoaseries
    iThis hergent, tthe
    ry termis,wdivergent, he teterms
    r increasing indefinitely . The comple
    mentary term , which increases with n and is very great when n is very great,
    is , however, very small for considerable values of 1 . For instance, when
    a = 10 , the values obtained for log I (11), by taking 3, 4, 5 , or 6 terms of
    the series, are respectively,
    16•090820096 , 16 - 104415343, 16 -104412565, 16 :104112563.

    CHANGE OF THE VARIABLES IN A DEFINITE


    MULTIPLE INTEGRAL .

    2774 Let x , y , z be connected with Ě , n , % by three equations


    u = 0 , v = 0 , w = 0.
    Then , when the limits of the integral containing the new
    variables can be assigned independently , we have

    CON*F(x,y,x)dxdydz
    u' r ' Viu21 d (urw )
    ' ' r'52 d ( Ems) at dadi.
    = - 1
    . ' $iuni I $i d (uvu )
    d (xyz)
    TRANSFORMATION OF A MULTIPLE INTEGRAL. 413

    where 0 is what F becomes when the values of x , y ,% , in terms


    ofĚ, n , %, obtained by solving the equations u , v , w , are sub
    stituted ,

    Proof. F(x,y,z)dxdydz = ({{ (8, ,5) dęi dydn dzdzdn


    d. dg.
    To find Xr, consider n and 5 constant, and differentiate the three equations
    U , v , w for $, as in (1723) . To find yn, consider S and 2 constant, and differ
    entiate for 71. To find 2x, consider x and y constant, and differentiate for % .
    We thus obtain
    d (uvw )
    dx dy dz _ _ d (Eyz ) d (125) d (GEN ) d (Eng)
    dę dn de d (uvw ) d (uvw ) d (uvw ) d (uvw )
    d (xyz) d (yzá) d (z&n ) d (xyz)
    observing that two interchanges of columns in a determinant do not alter
    its value or sign (559).
    Similarly in the case of any number of independent variables .

    When, however, the limits in the transformed integral


    have to be discovered from the given equations, the process
    is not so simple .
    In the first place, we shall show how to change the order
    of integration merely .
    2775 Taking a double integral in its most general form , we shall have
    po p$ (w ) B $ (v )
    I F (x, y) dxdy = 21 F (x, y) dy dx ...............(i.)
    Jadv (x ) Jazy (y )
    The right member will generally consist of more than one integral, and
    denotes their sum . The limits of the integration for æ may be, one or both ,
    constants, or, one or both , functions of y. V is the inverse of the function ,
    and is obtained by solving the equation y = 4 (2 ), so that < = ¥ (y) . Simi
    larly with regard to q and .
    An examination of the solid figure described in ( 1907) , whose volume
    this integral represents, will make the matter clearer. The integration, the
    order of which has to be changed , extends over an area which is the projec
    tion of the solid upon the plane of xy , and which is bounded by the two
    straight lines w = a, x = b, and the two curves y = 4 (x ), y = Q (x ).
    The summation of the elements PQqp extends from a to b , and includes
    in the one integral on the left of equation (i.) the whole of the solid in
    question .
    But, on the right, the different integrals represent the summation of
    elements like PQqp , but all parallel to OX, between planes y = a , y = ß , & c .
    drawn through points where the limits of a change their character on account
    of the boundaries y = 4 (æ ), y = 0 (2 ) not being straight lines parallel
    to OX .
    414 INTEGRAL CALCULUS.

    2776 EXAMPLE. — Let the figure represent the pro $ (* )


    jected area on the xy plane, bounded by the curves -- ---
    y = y (x ), y = 0 ( x ) , and the straight lines x = a , x = b .
    Let y = 0 ( 2 ) have a maximum value when < = c.
    The values of y at this point will be o ( c ), and at the
    points where the straight lines meet the curves the
    values will be q (a ), (b ), 4 (a), 4 (6 ).
    According to the drawing, the right member of
    equation (i.) will now stand as follows, U being written
    for F (x, y),
    (4(a)so • ( a ) ro 7° (6) po po (c) ,( )
    Jy (6) JY (y) Jy (a ) Ja
    U dy dr + Uly dx + 1
    Jo (6) ] $, (y)
    Jo (a) . $ (y)
    Udydx.
    The four integrals represent the four areas into which the wbole is divided
    by the dotted lines drawn parallel to the X axis . In the last integral, 0 , (y )
    and 0. (y) are the two values of a corresponding to one of y in that part of
    the curve y = 0 ( x ) which is cut twice by any x coordinate .

    2777 To change the order of integration in a triple integral,


    from %, y ,a to y , x, 7, we shall have an equation of the form
    fra (42(x ) (02(7,y) 12 74 , (z) $ 2(2, 3 )
    Jiri d 47 (2 ) J$1 (x,y)
    F (x,y,z) dxdydz = JzJ4,
    1 I F (x, y, z) dzdx dy
    ......... ( ii.)
    (z) J & ,(2,3)
    Here the most general form for the integrals whose sum is indicated by ?
    is that in which the limits of y are functions of z and a , the limits of x fanc
    tions of z, and the limits of z constant. Referring to the figure in (1906 ),
    tbe total value of the integral is equivalent to the following . Every element
    dady dz of the solid described in (1907) is multiplied by } (xyz), a function
    of the coordinates of the element, and the sum of the products is taken.
    This process is indicated by one triple integral on the left of equation (ii );
    the limits of the integration for z being two unrestricted curved surfaces
    % = 0 , (x, y ) , z = 02 (x , y ) ; the limits for y , two cylindrical surfaces y = ų , (2 ),
    y = 4 . (x ) ; and the limits for x , two planes x = xy, x = xz.
    But, with the changed order of integration, several integrals may be
    required . The most general form which any of them can take is that shown
    on the right of equation (ii. ) Solving the equation z = 0 , (x , y ), let
    y = 0 , (x, x ), y , = 0 , (2 , 2 ) be two resulting values of y ; then the integra
    tion for y may be effected between these limits over all parts of the solid
    where the surface % = % . (it, y ) is cut twice by the same y coordinate.
    The next integration is with respect to x , and is limited by the cylindrical
    surface, whose generating lines, parallel to OY, touch the surface z = 0 .(x , y ).
    At the points of contact, a will have a maximum or minimum value for each
    value of z ; therefore d ,,P . (æ , y ) = 0 . Eliminating y between this equation
    and that of the surface, we get x = 4 , ( 2 ), x = Y , ( z ) for the limits of x.
    Lastly, the result of the previous summations is integrated for z between
    two parallel planes % = % , z = , drawn so as to include all that portion of
    the solid over which the limits for x and y, already determined, remain the
    same.
    TRANSFORMATION OF A MULTIPLE INTEGRAL. 415

    The remaining integrations will take place between % = %, and similar


    successive parallel planes ; and , according to the portion of the solid which
    any two of these planes intercept, the limits of x for that integral will be one
    or other of the bounding surfaces, cnrved or plane, the limits of y , one or
    other of the curved surfaces .

    The general problem to change the variables in a multiple


    integral, and determine the limits from the given equations,
    may now be solved .
    2778 First, in the case of a double integral,
    xa pua(x )

    Jai Jai(c )
    .................. (iii).
    to change from x , y to &, n , having given the equations u = 0 ,
    v = 0 , involving the four variables.
    To change y for n, eliminate & between these equations ; thus y = f (2,9)
    and dy = f» («,n) dn. Substituting these values, we shall have
    F (x, y) dy = F {2 ,f (x,n) } fn (x, n) dn = F, (x, n) dn.
    Also, if n, corresponds to Yı, the equations y = M, (x ) and y = f (x, y ) will
    give n = y , ( x ). Similarly ng = 42 (2 ).
    Hence the integral ( iii.) may now be written
    222 (+2 (x ) fna ,(n )
    T F1 (x , n ) dxdn E = F ( , n ) dn dx ...... ... (iv .) ,
    31241(20) JnJY , (n )
    the form on the right being obtained by changing the order of integration , as
    explained in (2775 ).
    Next, to change æ for £, eliminate y between the equations u = 0 , v = 0) ;
    thus, x = g (£, n ) and dx = 9£ (£, n ) dž. Substituting as before, we shall have
    Fi(@ , n) dx = F , (E, n) ds.
    Also, &, corresponding to ® , the equations xy = (n ) and X , = g (X1, n ) pro
    duce & = m (n ), and in the same way q = m , (n ). Hence , finally ,
    fx8 Fuz (or ) ana m2(n )
    I F (x,y) dx dy =
    Jai Jui (x )
    F (€, n) dn d &............ (v.)
    1. Ini dm ,(n )
    In the last transformation from a to ], the most general form of the
    integrals which may be included under has been chosen . When any of the
    limits of u are constants, the process is simplified.

    2779 Again , to change the variables from x , y, z to ?, n , 8 ,


    in the triple integral,

    na pasospol* " F (x, y,z) da dydz...............(vi.),


    Jænd41(x)) xi (r,y)
    having given the equations u = 0 , v = 0, w = 0 between the
    six variables x, y ,% , &, n, %.
    416 INTEGRAL CALCULU 8.

    First, to change from % to 5, eliminate & and n between the three


    equations, and let the resulting equation be % = f (x , y , 5). From this
    dz = fs(x, y, S)ds; therefore
    F (x, y,z) dz = F {2, y, f ( ,y, 5)} fz( ,y, :) az = F (x, y, 5) d .
    Also, if 5 corresponds to the limit zu the equations 24 = xı(e, y) and
    % = f (x, y, ši) give % = 0, (x, y). Similarly 5, = (x, y).
    The integral (vi.) may therefore be written
    fro.[42(x) [°2( ,y) ( 52 14 (5) ,(5,- )
    Fi(x , y, 5) dx dyds = Fi (x , y, 5) d'$dx dy
    Jai J 41(-2) J+1(x,y) J5J4 (3) 20,(5,8) .......... (vii.),
    the last form being the result of changing the order of integration , as ex
    plained in (2777). We have now to change from y to "); we therefore
    eliminate z and & from the equations u, v, w , obtaining an equation of the
    form y = f (5,x, ), and proceed exactly as before . The result, as respects
    the general form of integral in (vii.), will be
    152 74215) (12(5x)
    1 1 F ,(c, 9, 5) d5dx dn ..................(viii.)
    J51 )¥.(5) Jaz (59 )
    The order of me and has now to be changed by (2775 ) . Since is a
    constant with respect to integrations for x and n , y , (5 ), 4 , ( ) will also be
    constants, while 1 , (5 , 2 ), 1, (5, . ) will be functions of the single variable x .
    Suppose n = 1,(5, x) gives æ = ,(5,n). Similarly, x = 4 ,( ,n) may be
    the other limit .
    At the point where x = 4 (6 ) and n = 1, (6, ), we shall obtain by
    eliminating x, say, n = M ( %) . Similarly , from x = \ , (5) and n = 1 , ( ( c)
    suppose, we get n = ma(% ) for the next limit ; then a general form for the
    transformed integralwill be
    pa fuz(5) (5 n )
    T F ,(x, , 5) dźdndx ......................(ix .)
    JS1 Jui(5) JA, (5, 9 )
    It now remains to change from the variable « to Ę. Eliminating y and z
    between the equations u , v, w , we have a result of the form æ = f ( E, , ) .
    Substituting for æ and dx as before,we arrive finally at the form
    PS2 puo(5) pua (s,n) _
    T F , (5, 9, 8) djdndě ..................... (x.)
    Iši dur(5) In (Sn)
    It should be noticed that the limits x = \ , (3,n ), x = 1 ,(5, n), in (ix .), are
    not necessarily different curves. They may, in some of the partial integrals ,
    be different portions of the same curve. This was exemplified in the last
    integral of (2776 ).

    MULTIPLE INTEGRALS .
    The following theorems, (2825 ) to (2830 ), which are
    given for three variables only , hold good for any number.
    Let x, y , z be quantities which can take any positive values
    MULTIPLE INTEGRALS. 417

    subject to the condition that their sum is not greater than


    unity ; then
    M P1 - X 1- -
    r (l) r (m ) r (n )
    2825 JOJO JO *30-ty»- *-1dxdydz = T (l + m + n + 1)
    Here x + y + z = 1 is the limiting equation.
    Proof. - Integrate for 2 ; then for y by (2308) ; finally for æ by (2280),
    and change to the gamma function by (2305 ).

    2826 - --17--dędndt = a'B”y" " )" G )" %)


    1m - 1 in

    2826 87°°°° sdh45 Par r * +1)


    when

    when (%) + (3)'+ (*) = 1 is the limiting equation.


    a

    DOF
    Proof.–Substitute * = ( ) = (3 ) ,« = ( )',and apply(2825).
    2827 When the limiting equation is simply & + n + = h,
    the value of the last integral becomes
    2l+ m + n r (1) T (m ) r (n ) '
    r (l + m + n + 1)

    2828 The value of the same integral, taken between the


    limits h and h + dh of the sum of the variables, is
    hl+m +n - 1 I (1) T (m ) T (n ) .
    r (l + m + n )
    Proof. — Let u be the value in (2827); then, by Taylor's theorem , the
    value required is
    du dh – (11ml 31+ m + n - 1 ( ) r ( m ) r ( n ) 22
    un dh = (1 + m + 11) h ?+m + r (l + m + n + 1)
    which redaces to the above, by (2288) .

    1n

    2829 | 21-4ym 1x *–15(x + y + z)dxdydz


    = T(1) f(m )[(n) 70
    T (l + m + n )
    h ’ + m + - 1 dh,
    Jó
    3 h
    418 INTEGRAL CALCULUS.

    if x + y + z = h and h varies from () to c . In other words, the


    variables must take all positive values allowed by the condi
    tion that their sum is not greater than c.
    Proof.– For each value of'h the integration with respect to x, y, z gives,
    f (1 ) 2l++ mm ++ n -1
    - I' ( ) ( m ) 1 (n ) ,,
    by (2828), 1 (l + m + n )
    the variations of x , y , z not affecting h. This expression has then to be in
    tegrated as a function of h from 0 to c .

    -11 - 1Yn - 14

    2830 077- -8{(1) + ( ) + (3) }dědndt


    1 m n .
    _ a'Bºy» ( ) ( ) C )ceannhétq + -1

    par r( + * + % ) Jo
    with the limiting equation (
    wi th the limiting equation ( ) +(3) + (x) = c.
    la

    Proor.-- From (2829) by substituting = )",&c.


    2831 If x , y , z be n variables, taking all positive values
    subject to the restriction # * + y* + + ... $ 1 ; then
    drdydz. & c . opt(n + 1)
    1 V (1 — * - * - - - & c.). 2" T {} (n + 1)} "
    But, if negative values of the variables are permitted , omit
    the factor 2 " in the denominator .
    Proof. —1 In (2830) pnt l = m = &c. = 1; a = B = &c. = 1 ; p = q = &c. = 2 ;
    f (n ) = ; c = 1 ; and the expression on the right becomes
    { r ( ) }n hon - 1
    dh.
    2" T (3n) J. ( 1 - 1 )
    The integral isis. = B ( f"
    The in , î1)) (228
    (2200 ) = m ( n ) ( 2) (2305 ).
    Hence the result. (n + 1) ?
    But if negative values of the variables are allowed under the same re
    striction , ' + + + ... $ 1, each element of the integral will occur 2 " times
    for once nnder the first hypothesis. Therefore the former result must be
    multiplied by 2".
    MULTIPLE INTEGRALS. 419

    2832 If n positive variables, x , y , * , & c ., are limited by the


    condition as + y + 7 + & c . $ 1 , then

    1 ° (ax + by + cz + & c.)dxdydz ...


    (n -1)
    _ (kø)(1 – 82)$(n-1)dŚ,
    2n-11 { } (n + 1)} Jo
    where l = a + 22 + cº + & c.
    PROOF. - Change the variables to £, n , Ś by the
    orthogonal transformation (1799) , so that 2 y % ...
    a + + 0% + & c . = k ”, and ax + by + cz + & c. = kĘ.

    ole
    ole
    a b

    ole
    ole
    The integral then takes the form . k ki T ...
    ||| 0 (kë) dědnds... with + 1° + 52 + & c . $ 1. a .
    Now , integrate for n, Š, & c., considering & con nl To že ...
    stant, by adapting formula (2826) . The limiting sla" 6" ["
    equation is

    (vci-45) + (vc - + &c.ton–1terms;


    = 1. *-*|
    Therefore put l = m = & c. = 1 ; p = q = & c . = 2 ; a = B = & c. = V (1 -- ).

    The result is J. ° (kt) 2n=1 T {} (n - 1) + 1} “ S,


    which is equivalent to the value above.

    2833 With the same limiting equation for n variables and


    the same value of k ,
    or (ax + by + cx + & c.) dx dydz ...
    JII (1 - x - — * * - & c.) “ *
    = 21-iT(In ).J. " (kf)(1 - 5%){n-1d .
    Proof. - Making the same orthogonal transformation as in (2832) , the
    • (kx) dxdnds ...
    integral changes to
    *** V (1 — * - » - & c.)'
    Considering & constant, the integration for the remaining variables is effected
    by (2830). Adapting the integral to that formula, we have
    q ( )d& r dnds ...
    V(1 –2 ))] {1 - 1 - 12 - &c.}
    with
    with Wi- ) + ( á ) + &c.,ton– 1 terms,$ !
    for the limiting equation .
    420 INTEGRAL CALCULUS.

    Here l = m = & c. = 1 ; p = q = & c. = 2 ; a = b = & c. = (1 - 6 ) ;


    f (h ) = ic = 1 ; and the reductions are similar to those in (2832).

    2834 If in (2832 – 3 ) negative values of the variables are


    admitted (since the limiting equation is satisfied by such ) ,
    each element of the integral with respect to n , % , & c . will then
    occur 2n -1 times, and therefore the result in each case must
    be multiplied by 2n - 1, and the limits of the integration for &
    will be – 1 and 1 instead of 0 and 1 .

    EXPANSIONS OF FUNCTIONS IN CONVERGING


    SERIES.

    The expansion of a function by Maclaurin 's theorem (1507 )


    can be at once effected if the nth derivative of the function is
    known , or if merely the value of the same, when the inde
    pendent variable vanishes, is known . Some nth derivatives of
    different functions, in addition to those given at (1461– 71 ) ,
    are therefore here collected . When the general value would
    be too complicated , the value for the origin zero alone
    is given .

    DERIVATIVES OF THE nth ORDER .


    The following is a general formula for calculating the nth
    derivative of a function of a function.
    If y be a function of z , and z a function of x ,

    2852 = ( - )..
    where r = 1 , 2, 3, ... n successively, and a is put = z in each
    term of the expanded binomial, after differentiation.
    Proof.— Assume yna = 4,9z+ 22 92 + *4x +..+ in yner
    To determine any coefficient Ar, form r equations from this by making
    y = % , 7", 2 , ... 2 " in succession : multiply these q equations respectively by
    rz - , - ( 1, 2 ) z -?, C (r , 3 ) - , ... ( - 1 ) +1 , -", and add the results. All the
    coefficients excepting A, disappear. This is shown by differentiating the
    equation (1 – 2 )" = 1 – 9x + 0 (1, 2) 2 – 0 (r, 3) '+ ... * *
    EXPANSIONS OF FUNCTIONS. 421

    successively for æ , and making x zero after each differeutiation. Thus, finally,
    4 = (- 1)-* { ins – C(p,2)(29)ne + C(1,3)(29)m ....(- 1)rus(2) c}
    nI

    = dno ( - 1) ,
    with a put = %, after expanding and differentiating the binomial.

    2853 EXAMPLES. — The formula may be applied to verify


    equations ( 1416 – 19) .
    Jacobi's formula (1471) may also be obtained by it.
    2854 d(n+1)ą sin-1x
    1 .3 ... (2n - 1 ) S п
    = 2"(1 — x )"(1 — *) 1 – 2n –12 + 2n – 1.2 = 32
    1 . 3 . 5 C (n , 3 )
    2n - 1 . 2n - 3 . 2n - 5 22+... + 2"}, where 2 = 11
    PROOF. ( sin -> x )(n +1)x = { ( 1 , 2 ) = ( 1 + x ) } }nt (1434).
    Expand the right member by (1460).

    2855 dnx tan - 1 X . This derivative is obtained in (1468 ) .


    The following is another method , which also includes the
    result in ( 1469) .

    detan*'a = 1te = {eti o };


    In – 1 ( - 1 ) ^ S 1 1
    : . by (1425) dn tan -' x =
    2i ( (x + 1) ^ (2 - 3)
    Put x = coto, therefore x Ei= (1 + x ?) ( cos 0 + i sin 6 ), which values
    substituted in ( 1) convert the equation , by (757), into
    (tan - 2 )ng = ( – 1 )n -1 |n - 1 sin” O sin no.
    1° COS

    2856 d .& {eucosa cos(x sina)} = ex cosa cos (x sin a + na).


    PROOF. - By Induction .

    LAGRANGE'S METHOD .
    2857 Lemma. — The nth derivative of a function u = f (a )
    will, by Taylor's theorem (1500 ) , be equal to 1. 2 ... n times
    the coefficient of h " in the expansion of f (a + h ) in powers of
    h by any known method .
    422 INTEGRAL CALCULUS.

    Let u = a + bx + cx?, and therefore Ug = b + 2cx ; then


    drx (a + bx + ca?)” is equal to either of the following series,
    with the notation of (2451– 2).
    (r - 1) cu no- P)CPUP
    2858 1 (m) un -ru". + 1 ( ) 1 (r - 2p) U -P
    2859 or, putting 2n = m and 4ac — lº = q ,
    TU

    cie
    - 1 2P
    (2
    mon=r{us+n3qu ' ..+C(m.paruo *+...}.
    PROOF. - Changing æ into ath in u”, it becomes (u tu h + chº)". Then ,
    by (2857 ), drxu " will be = 1 r times the coefficient of h " in the expansion of
    this trinomial. (2858 ) is the result, and it may be obtained by expanding
    { (u + uzh) + ch' }" as a binomial, and collecting the coefficients of k from the
    subsequent expansions. The value (2859) is found by taking

    (u +ugh +cle)* = "" {(1+ + (2%) }",


    expanding, collecting coefficients of h", and multiplying by 1. 2 ... , as
    before.

    2860 Ex.- To find dnz (a’ + x )". Applying formula ( 2859), we have
    u = a' + a , uz = 22, q = 2a, r = n . Therefore
    duz(a*+ zº)" = (2n) ? C an
    {c"+ "?(2(n = 1) a?aʻxn-
    I m * (m - 1)^ ( n - 2) ( – 3 ) , , -4
    1 . 2 . 2n ... ( 2n - 3)

    - 2r
    02

    2861 dekaº= q*v*{a"(2x)"+...+ * ***(20)*+r+&c.},


    with r = 1, 2, 3 , & c. in succession.
    PROOF. - By the method of (2857). Putting ed (2 +h )2 = ear e2axh qan , expand
    the factors containing h by (150 ), and from the product of the two series
    collect the coefficients of h ".

    2862 du cosx* = (2x)"cog(x+ +" ) +...


    n

    ...+ (2.1)-- cio (28+(n=>]=)+&c.


    EXPANSIONS OF FUNCTIONS. 423

    PROOF. — dnx (cos xº + i sin æ ?) = ( nxeir? Expand the right by (2861),


    putting in-* = etin-r)}", since, by (766), edi* = i sin = i. Also put
    eitt+ i(n -r )}* = co (12 — r ) + (1 — 1 ) #
    2 2 si
    and then equate real and imaginary parts.

    2864 dmth = ( - 1)"[emo + {n + 1 – 2" }e(n=1)o


    + { C (n + 1, 2 ) — 2" (n + 1 ) + 3"} eín -2)e
    + { C( + 1, 3 ) - 2°C( + 1, 2) + 3 (n + 1 ) - 4* } en-3)= + &c. |
    PROOF. — Let u be the function . By differentiating u it is seen that
    (ef + 1)* •lUnx = A , ent + An-jen -1}* + Ar-2 e(n -3)* + ... + 4 ,et,
    the A 's being constants. To determine their values, expand u = (e* + 1) - ",
    and also (e* + 1)" +!, by the Binomial theorem ; thus
    Une = ( - 1) " {e - * — 2"e -23 + 3" e- 31 — 4" e -63 +. & c.},
    (e* + 1)* +1 = e(n +1)x + (n + 1) end + C (n + 1, 2) e(n - 1)x + C (n + 1, 3) eln -2)* + & c.
    From the product of the two expansions the coefficients Am, An-1, & c. may
    be selected.

    2865 duzo tan -'x = (- 1) 7 |n - 1 or zero,


    according as n is odd or even .
    Proof. - By Role IV . (1534 ). The first and last differential equations
    (see Example 1535) are, in this case,
    ( 1 + x ) 420 + 2 .cy : = 0 ...... (i.) ; Yin + 2) zo + n (n + 1 ) Ynzo = 0......(ii.) ;
    with yzo = 1 and Y2x0 = 0.
    Otherwise. By ( 1468), putting a = 0 .

    2867 dnxo sin -2x = 1. 32.52.... (n — 2 ) or zero ,


    according as n is odd or even .
    ac

    PROOF. — By differentiating (1528).


    Otherwise. — As in ( 2865) where equations (i.) and (ii.) willbecome in this
    case (1 – x ') Yzx = xyz ......(i.) Yin +2)zo = n 'ynxo ... ...(ii.)

    2869 duro (sin - x ) = 2 .22.4².62... (n — 2)? or zero,


    according as n is even or odd .
    Proof. — As in (2865) ; equations (i.) and (ii.) being identical with those
    in ( 2867) .
    424 INTEGRAL CALCULUS.

    2871 dnxo cos(m sin ? x )


    ( - 1) mở( mº - 2)( ° – 4 ) ...[ mº - (x - 2) ],
    or zero; according as n is even or odd and > 0 if even .
    2873 dnxo sin (m sin -2x )
    = ( - 1) = m (mº — 1)(mº— 3 ) ... [mº— (n — 2) ],
    or zero ; according asn is odd or even and > 1 if odd.
    2875 dnco cos (m cos - x )
    = (- 1) 7 m (mº— 1)mº— 3°)...[mº–(n — 2)*]sin ma,
    2876 or
    = ( - 1)7 m2(mº — 22)(mº—42)... [mº— (n — 2)*] cost,
    according as n is odd and > 1, or even and > 0 .
    2879 dugo sin (m cos-1x )
    = (- 1)2m2(mº — 22)(m²— 42)... [mº— (n — 2) ] sin Žma,
    2879 or
    n + 1

    = ( - 1) 7 m (m²– 1)(mº — 3 )... [m ? – (n — 2) ] cos žmn ,


    according as n is even and > 0 , or odd and > 1.
    Observe that, in (2871– 3 ), sin - ' 0) = 0, and in (2875– 9 ),
    cos -10 = 5 , are the only values admitted .
    Proof. — For (2871–9). As in (2865) ; equations (i.) and (ii.) now be
    coming in each case
    (1 — æ ) Yzz — æyo + mºy = 0 ...... (i.) Y (n +2)30 = (nº - mº) Yndo ... ... (ü .)
    Otherwise. By the method of (1533) .

    2880 Let y = æ cotx, then


    I

    n cos " : = y sin " +... + yrxoC (n,r)sin " * + + ...


    ...+9n-1)zon sin
    with integral values of r, from 0 to n - 1 inclusive.
    EXPANSIONS OF FUNCTIONS. 425

    2881 Thus , denoting Ynxo shortly by Yn, we find , by making


    n = 1, 2 , 3 , & c . successively in the formula ,
    2 8 32
    Y: = - , y = - Y =- Ys = - 115
    PROOF. — Take the nth derivative of the equation « cos x = y sin x by
    (1460), reducing the coefficients by (1461- 2), and putting æ finally = 0.
    2882 The derivatives of an odd order all vanish . This may be shown
    independently, as follows :
    Let y = $ (x ), then 0 (x ) is an even function of x (1401) ; therefore
    024 * (x ) = - $ 2n +) ( - 2 ) ;
    :: 92n +2 (0 ) = – 2n+2 (0) ; i. $ 20+2 (0) = 0.
    N - 1

    2883 dnso{(1 +x2)*sin(m tan-+x)} = (- 1)"? m ?" or zero,


    according as n is odd or even .
    N - 2

    2885 dndo{(1 + x2) { cos (m tan -+ x)} = ( - 1) 7 mm or zero,


    according as n is even or odd .
    Proof. - As in (2865). Equations (i.) and (ii.), both for (2883) and
    (2885), are now ( 1 + ?) 422 - 2 ( - 1 ) xya + m (m - 1) y = 0................(i.),
    and Yin :2)zo = - (m , n ) (m - n - 1) Ynco ........................ (ii.)
    Formula ( ii.) gives the factors in succession , starting with yo = 0 , yz = m
    in (2883) ; and with yo = 1, yz = 0 in (2885).

    2887 duxo{(1 + x2) 2 cos(m tan -1x)} = ( - 1)2min) or zero,


    according as n is even or odd .
    Proof. — Change the sign of m in (2885) .
    Note. - In formulæ (2883- 7) zero is the only admitted value of tan - ? 0 .

    2889 d.00(24 ) = (- 1)* B, orzero,


    according as n is even or odd; by (1539).
    2891 When p is a positive integer,
    duco(a® est cosbæ) = n(?(a*+7°) ?"cos{(n –p)tan-1 }
    or zero,according as n is > or < p .
    Proof. — Put y = eat cos bx and x = x ” in (1460 ), employing (1465).
    3 I
    426 INTEGRAL CALCULUS.

    MISCELLANEOUS EXPANSIONS.
    The following series are placed here for the sake of
    reference, many of them being of use in evaluating definite
    integrals by Rule V . (2249). Other series and methods of
    expansion will be found in Articles (125 – 129) , (149 – 159) ,
    ( 248 – 295 ) , (756 – 817 ) , (1460) , (1471- 1472), (1500 - 1573) .
    For tests of convergency , see (239 - 247) .
    Numerous expansions may be obtained by differentiating
    or integrating known series or their logarithms. These and
    other methods are exemplified below .

    1 1 1 1
    2911 cotx =
    X T -x otx 27 - x
    1.
    - - & c.
    20 + x 31 - XT 35 + x
    Proof. — By differentiating the logarithm of equation (815).

    2912 -cotme= +estatitas


    at W + 37 & c.
    ilt
    to
    + a+ 2 + x+ 3 + 2 +3 +&c.
    Proof. — By changing x into na in (2911).

    2013 tane =te tatinio


    T
    attiat & c.

    Proof. — By changing w into įt - a in ( 2911).

    2914 cosec x =
    2914 coseex = N +' T -- X -no-2
    + x 27 x
    nt & c.
    T 20 tx T 37 - x 3 + x 4 X
    I
    Proof. — By adding together equations (2911, 2913), and changing *
    into ix.
    EXPANSIONS OF FUNCTIONS IN SERIES. 427

    2915
    2915 Sin
    sinme
    та от 1 1 - m 1 + m

    -2-m +2+m + 3+* - 37m -&o.


    PROOF. - By putting x = ma in (2914).

    6
    20

    2916 cotx= - 2 PB3: _ By – ko.


    For proof see (1545 ). The reference in that article (first edition ) should
    be to ( 1541) not (1540) .

    29117

    2918
    2918 cosec
    cosec2x == 1= +2(2+1)By
    t
    _ 2(28
    4 – 1) B.23 + 2 (2 - 1) Bxxx+ &c.
    16
    PROOF. — By (2916 ) and the relations
    tan x = cotx — 2 cot 2d , cosec x = cotiv - cot x .

    2919 _0
    y = 1 + 2a cos x + 2aº cos 2x + 2a cos 3x + & c .
    1 - 2a cos x ta? = 1 +20
    COS x
    2920
    1 - 2a cos x + a ?
    a Ita2? os
    - 1 - 0 ((cosxta
    Oti c , cos2x + a' cos 3x ta' cos 4x + & c ...)
    2921
    sin x
    - = sin c + a sin 2x + aº sin 3caº sin 4c4 & c .
    1 - 2a cosx + aº
    PROOF. — By (784 -6 ) making a = 6 = x and c = a .

    When a is less than unity and either positive or negative,


    428 INTEGRAL CALCULUS .

    2922
    COST COST
    -log(1 + 2a cosæ + aº) = acosa cos2x + cos3x — &o.
    2923 tan--lta asinə
    cos x
    = asin æ – *sin 2x + " sin 3x — &c.
    a si

    PROOF. — Putting . = a ( cos a tisin x ), we have


    log ( 1 + x ) = log (1 + a cosa + ia sin x )
    = { log (1 + 2a cosx + a +) + i tan -- 1 a+ sin 2 (2214),
    a cosa
    and also log (1 + x ) =
    Substitute the value of z and equate real and imaginary parts.
    2924 Otherwise. — To obtain (2922),
    log ( 1 + 2a cos x + a®) = log (1 + aelt) + log (1 + ae-it).
    Expanding by (154 ), the series is at once obtained by (768).
    2925 Otherwise. — Integrate the equation in (786 ) with respect to a , after
    changing a and ß into x , and c into - a .

    2926 When a is greater than unity, put


    log (1 + 2a cos x + aº) = log a ' + log (1 + 2a- cos x + a -?),
    and the last term can be expanded in a converging series by
    (2922).
    2927
    log 2 cos 3x = cos x - cos 2x + { cos 3x – 1 cos4x + & c.
    2928
    · log 2 sin įx = - cos x — cos 2x – į cos 3x – 4 cos4x - & c.
    2929 x = sin x — 1 sin 2x + } sin 3x — 1 sin 4x + & c.
    2930 i (1 - x ) = sin x + i sin 2x + } sin 3x + i sin 4x + & c .
    Proof. — ( 2927–30) Make a = + 1 in (2922– 3).

    2931 = sin x + ž sin 3x + } sin 5x + & c .


    2932 q =2 2 (1 + { - } - { + in - & c.).
    Proof. — Add together (2929-30) , and put x = in .

    When n is less than unity , and a = 1 + 1 ( 1 + na),


    2933 log(1 + cosx ) = log (1 + 2acosx + a ) -log (1 + aº),
    EXPANSIONS OF FUNCTIONS IN SERIES. 429

    and is therefore equal to twice the series in (2922), minus


    log ( 1 + a ?) . But if a be greater than unity , expand , as in
    (2926), by
    2934 log ( 1 + n cos x )
    = log (1 + 2a-l cos x + a -9) + log a’ – log (1 + a ”).
    2935
    (1 + 2a cos x )" = A + A , cos x + A , cos 2x + A , cos 3x + & c.,
    where
    A = 1 + C (n , 2 ) 2a² + ... + 0 (n , 2p ) C (2p, p ) a?! + ...,
    A, = 2a {n + C (n , 3) 3a ' + ... + C (n , 2p + 1) C (2p + 1, p ) a P + ... },
    and
    A - , (n - p + 1 ) a - rA
    Arti =
    ( n + r + l) a
    If n be a positive integer, the series terminates with the
    n + 1th term , and the values of A and A , are also finite.
    Proof. — Differentiate the logarithm of the first equation ; multiply np
    and equate coefficients of sin rx after transforming by (666 ) ; thus Aprl is
    obtained .
    To find A and Aj, expand (1 + 2a cos x )" by the Binomial Theorem , and
    the powers of cos x afterwards by ( 772) . .

    LEGENDRE'S FUNCTION Xm.


    2936 ( 1 - 2ax + a ) - - = 1 + Xa + X , a ' + ... + X , a " +
    with An - 2nn
    1 d " (x2 - 1)" .

    PROOF. — Expand by the Binomial Theorem , and in the numerical part


    of each coefficient of a” express 1 . 3 . 5 ... 2n - 1 as 2n = 2" | n .

    Consecutive functions are connected by the relation


    2937 d ,Xn+1 = (2n + 1) X , + d , Xn-1.
    PROOF. — Differentiate the factor once under the sign of differentiation in
    the values of X , +1 and X n - 1 given by the formula for X , in (2936).

    A differential equation for X , is


    2938 (1 — 29) d2, X , - 2x d , X , + n (n + 1) X , = 0 .
    430 INTEGRAL CALOTLUS.

    2939 When p is any positive integer ,


    1° + 2° + 3® + ... + (n - 1)"
    s mp+ 1 nip B ,NP- 1 Bino -s B .np- ó
    R1p + 1 2 p + 12 p - 1 4 1p - 3 + 16 p - 5
    concluding, according asn is even or odd,with
    ( - 1)}p+1 B,1
    IP
    or ( - 1)+(p+1)IpB-12)
    .-19.. ?.
    Proor. = = . Expand the left side by division, and
    each term subsequently by (150). Again , expand the first factor of the right
    side by (150 ), and the second by ( 1539), and equate the coefficients of ze in
    the two results.
    See (276 ) for the values of the series when p is 1, 2 , 3 , or 4 . But the
    general formula there is incorrectly printed .

    Let the series (2940–4 ) be denoted by San, Sam, 8an, s2n+ 1,as
    under, n being any positive integer; then
    2n - 1

    2940 = 1+ + + x+&o...= 2 *B
    2941 s =1- * + +&c... P *B,
    2942 s = 1+ # +3 + + + 80...= ="B..
    OF
    Proof.— (i.) Son is obtained in (1545).
    (ii.) S. - S., = 2 (21 +a + &e.) = 2 $ . This give S.
    (iii.) 820 = } ( San + San ).

    2943 s = 1 + 32n + + & c. = ad 2n- 1)= COTTx


    4 | 2n - 1
    x = 1
    2944 sén+1 = 1 2 + 1 en + 1 2n + 1
    _ ad2nd cot Tax
    = 4+*+ 1 | 2n ?
    x = 1.
    PROOF. — By differentiating equation (2912) successively, and putting
    a = { in the result. To compute de cot # d , see (1525 ) .
    EXPANSIONS OF FUNCTIONS IN SERIES. 431

    2945 The following values have been calculated by formulæ


    (2940 – 4 ).
    4eod
    col
    !A.cla

    S. SE SES .= 9.50
    317 127/773
    a
    S = 12 T°
    30240 1209600
    70 17.03
    960 161280
    55 6177
    24

    1536 184320
    .

    cos c - cos a
    2946
    2946_ 1 - cos a * =(1 )[:- *- :][1- 27**+ a)?]]
    sa
    [1- 442-a):][1-(4 tu..]...&o,
    COS
    sa
    2947 [1- 6 ][1-(4+ ][ - ]
    x [1- 63 **a;][ - Con w.:)...&c.
    Proof. Cos1 x- cos
    — cosa a = sind(a + z) sin } (a — ~ ). Expand the sines by
    sina la
    (815 ). The two n + 1th factors of the numerator divided by the corresponding
    ones of the denominator reduce to

    c
    Ta

    =(1-2 – )(1+ *+a)(1+22-a)(1-2 .ta)


    = (1- (217
    2m2 -a)
    - a )?=)(1-12moitaja).
    (2n7 + a ) )
    Similarly with (2947) employing (816).
    COS
    2948 cose+tan sina = (1+ 2 )(1- 2 )(1+ xma)
    * (1-327a)(1+ 52-a)(1- 27 ). & ,
    2949 cos2–cot, sino = (1-2 ) (1+ 2+.)(1-2 )
    * (1+a ta)(1- )(1+og na...&c.
    Proor. cose +tan sin = = cos(34
    cos ja ). Expand the cosines by (816),
    and reduce. Similarly with (2949), employing 815.
    432 INTEGRAL CALCULUS.

    2950 = 0[1+ ( )'][1+ ( *)'][1+ (37) ....


    2951 40 *= [1+(* )][1+(3*)'][1+ ]. ..
    Proof.— Change 0 into ix in (815) and (816 ).
    p* – 2 cos a te- -
    2952 2 (1 – cos a )
    = [1+ (a )'][1+ (372- )"][1+ (**a)'][1+ (12 -a)"]...&e.
    * * + 2 cos a tes
    2953 2 ( 1 + cos a )

    = [1+ ( ) ][1+ (212)'][1+ (3 - )"] [1+ (371a)"]...&c.


    Proor. — Change into ix in (2946–7).

    FORMULÆ FOR THE EXPANSION OF FUNCTIONS


    IN TRIGONOMETRICAL SERIES.
    2955 When x has any value between 1 and - 1,
    $(x) = 3S,* (u)dv + 7 = = $(w)cos no (v =x)dv...(i.),
    where n must have all positive integral values in succession
    from 1 upwards
    But, if x = l or - 1, the left side becomes lp (l) + 10 ( - 1).
    Proof. — By (2919) we have, when h is < 1 ,
    1 - h
    1 - 2h cos 0 + ha = 11 + 2h cos 0 + 2h * cos 20 + 21* cos 30 + & c....
    Put 0 = + (0 -2) ; multiply each side by 0 (u), and integrate for v from
    - 1 to l; then make h = 1. The left side becomes, by substituting ; = v - 2 ,
    p (1 — h?)$ (v) do - -- (1 — h®) (2 + z) dz
    J -11 - 2h cos " (V ) h+ J - 1-2 (1 - h ): + 4h sin
    When h = 1 each element of the integral vanishes, excepting for values of v
    which lie near to 2 . Therefore the only appreciable value of the integral
    arises from such elements, and in these z will have values near to zero , both
    positive and negative, since x has a fixed value between I and - 1. Let these
    values of z range from - ß to a. Then between these small limits we shall
    have sin 72
    21 = 412 , and p (x + 6 ) = (20),
    EXPANSIONS OF FUNCTIONS IN SERIES. 433

    and the integral takes the form


    (x ) pa do
    (1 - h )
    J - B ( 1 - h ): + ??
    (1 + h ) n - AT Vh wanciB hi
    - VI 2lop ( x ),
    1 (1 - h ) l (ich )
    when h is made equal to unity , which establishes the formula .
    In the case, however, in which æ = l, sin " vanishes atboth limits,that
    is, when z = 0 and when 3 = - 21. We have therefore to integrate for %
    from – B to 0 , and also from – 21 to - 21 + a , a and B being any small
    quantities. The first integration gives lø ( l) as above, putting a = 0 . The
    second integration , by substituting y = x + 21, produces a similar form with
    limits ( to a , and with o ( x - 21) in the place of o (.r ) giving ly ( - 1) when
    x = l. Thus the total value of the integral is 10 (1) + 10 ( - 1) . The result
    is the same when a = - 1.
    That the right side of equation (i.) forms a converging series appears by
    integrating the general terms by Parts ; thus
    (1 ) cos nA (1:-— » ) , i sin 17 (v — 2 ) )
    ‫ ر‬.
    - J - 1
    $"(v) sin 17 (v=2)dv,
    which vanishes when n is infinite , provided ¢ (v ) is not infinite.
    Hence the multiplication of such terms by l " when n is infinite produces
    no finite result when h is made = 1, although 1° is a factor of indeterminate
    value,

    2955a A function of the form p (x ) cosnr, with n infinitely


    great, has been called “ a fluctuating function ,” for the reason
    that between any two finite limits of the variable a , the func
    tion changes sign infinitely often , oscillating between the
    values o (a ) and ~ (~ ). The preceding demonstration shows
    that the sum of all these values, as x varies continuously
    between the assigned limits, is zero .
    By similar reasoning, the two following equations are
    obtained .
    2956 If æ has any value between
    1 m2
    0 and 1,
    $(x) = 3 (**(v)dv + 1 3+ 4(u)cos1+ (v= x)do...(2).
    VO

    But if x = 0 , write 4 (0) on the left; and if x = l, write


    14 ( ).
    If æ has any value between 0 and 1,
    2957 o = * *+(v)do + 12.75** (u)cos1+ ( + x)dv...(3).
    3 K
    434 INTEGRAL CALCULUS.

    But if x = 0 , write 20(0 ) on the left ; and if x = l, write


    ( ).

    2958 °(x)= 78 (v)do+ { cos" mi"| cos * ..(v)de


    . ... ( 4 ) .
    This formula is true for any value of x between 0 and í,
    both inclusive .
    But if a be > 1, write 0 ( x ~ 2ml) instead of p ( x ) on the
    left, where 2ml is that even multiple of l which is nearest to x
    in value.
    If the sign of x be changed on the right, the left side of
    the equation remains unaltered in every case .
    2959 $ (x ) = ? sin "T" ("sin97% $(1) do......... (5).
    lalo

    This formula holds for any value of a between 0 and 1


    exclusive of those values.
    If æ be > l , write + ( x ~ 2ml) instead of $ ( 2 ) on the left,
    + or – according as x is > or < 2ml, the even multiple of l
    which is nearest to x in value.
    But if x be 0 or 1 , or any multiple of 1, the left side of this
    equation vanishes.
    If the sign of x be changed on the right, the left side is
    numerically the same in every case, but of opposite sign .
    Proof.----For (2958 -9) . To obtain ( 1 ) take the sum , and to obtain (5 )
    take the difference, of equations ( 2 ) and (3 ) . To determine the values of
    the series when æ is > l, put x = 2ml + é', so that a ' is < l .
    EXAMPLES .
    For all values of x , from 0 to a inclusive,
    cos 3x cos 5 .x
    2960 x =
    PROOF. - In formula (4 ) put « ( ix ) = x and 1 = , then

    p'ocos uv dv= [vsinw +cos,ne] = - or0,


    according as n is odd or even .
    Similarly, by formula (5 ), equation (2929) is reproduced.

    For all values of æ , from – 7 to ha inclusive.


    2961 r = # {sinæ –sing " + sing5" –&c.}.
    Proof. — Change x into - x in (2961) .
    435

    2962 теаr — e - ar sin x 2 sin 2x 3 sin 3 .0


    2 er - e -dr Caº + 1 + 2 + 4 + 3
    Proof. - In formula (5 ) put o ( c ) = eat – e - ax and l = r ; then
    Unr1n (elli -- e - an )
    - e-ar) sin nvdv = ( - 1)" +1? a +

    2963 If $ ( r ) be not a continuous function between a = 0


    and x = l, let the function be p (ix ) from x = 0 to x = a, and
    * ( ) from x = a to a = 1 ; then , in formulæ ( 1) and (5 ) , we
    shall have p (@ ) or ( ) respectively on the left side, according
    to the situation of x between 0 and a , or between a and l.
    But, if x = a , we must write tipa) + 7 ( a ) } for the left
    member.
    Proof. — In ascertaining the value of the integral in the demonstration
    of (2955 ) , we are only concerned with the form of the function close to the
    value of x in question . Hence the result is not affected by the discontinuity
    unless x = a . In this case the integratiou for z is from - ß to 0) with o ( x )
    for the function , and from 0 to a with y (x ) for the function , producing
    10 (a ) + 14 (a ) .

    2964 Hence an expression involving æ in an infinite


    series ofsines of consecutive multiples of " may be found,
    such that, when a lies between any of the assigned limits
    (0 and a , a and b , b and c , . .. k and 1 ), the series shill be equal
    respectively to the corresponding assigned functions
    f (x), f (x), 13( ) ... fu(x),
    provided that the integrals
    ( sin("T").fi(*)de,
    0
    sin "T"/>(+)dx,...Ssin ("7")! ( )dx
    can all be determined .
    2965 The same is true, reading cosine for sin ? throughout,
    with the additional proviso [as appears from fornuli ( 1) ] that
    the integrals
    Diri(s)dan 5°/a(w)de,...SS.(«)dlv
    can also be determined .
    2966 Ex . 1 . – To find in the form of a series of cosines of multiples of
    z a function of x which shall be equal to the constants a , , or , according
    as a lies between 0 and a , u and b, or b and it.
    436 INTEGRAL CALCULUS.

    Formula (4) produces, putting l = t,


    = {{ d +[°8 +[vdo}
    + 387 . cos nx {{"acos nudo+[°3cosnede+ ycosnoda }
    = 1 {a (a– B)+ 0(3 — )+ wy}
    2 na 1 cosnx (a - B ) sin na + (B - v ) sin nb + y sin nn }.

    2967 Ex. 2. — To find a function of a having the value c, when æ lies


    between 0 and a, and the value zero when æ lies between u and I.
    By formula (4), we shall have
    NTTV NU cl Na
    cos ( V ) dx = cl cos CONEdv =
    NTT

    since ® (v) = c from 0 to a, andso zero from a to l.


    2c in
    + sin
    Therefore (x)= + {sina'cos + sin 27acos
    TU

    : + ş sin stacos 87* +&c.}.


    When x = a, the value is } [o (a) + 0 ] = ic, by the rule in (2963). This
    may be verified by putting a = - 1 in (2923).
    2968 Ex. 3 . - To find a function of x which becomes equal to ka when
    æ lies between 0 and jl, and equal to k (1 — ) when a lies between il and I,
    By formula (4 ) ,
    ni NTV
    IGNTV = 1 kv cos T: ( 1 - 0 ) cos dv .
    (V ) cos

    This reduces to (2cos" –cosn –1) = - 4 or 0,


    according as n is, or is not, of the form 4m + 2. Also

    sp(w)do=ht"udo+1(1,42-v)do="
    .: 9(a)= *_ ? cos amet + cosGope+ bancos Town+ &c.}.
    APPROXIMATE INTEGRATION .
    curve
    2991 Let | f (x)dx be the integral, and let the curve
    y = f (a ) be drawn. By summing the areas of the trapezoids,
    whose parallel sides are the n + 1 equidistant ordinates
    APPROXIMATE INTEGRATION. 437

    Yu, Yı, ... Yn, we find, for a first approximation,


    b
    f(x)dx = 5 " (y. + 2y + 2y2+ ... + 2yn-1 + yn)......(i.)
    1

    SÍMPSON 'S METHOD .


    2992 If y, be the ordinate intermediate between yo = f (a )
    and y = f (b ) , then , approximately ,

    f(x)dx = 6 " ( + 44 +y:)..........(ii.)


    PROOF. — Take n = 3 in formula (i.) ; write ya for Ys, and suppose two
    intermediate ordinates each equal to 1 . The area thas obtained is equal to
    what it would be if the bounding curve were a parabola having for ordinates
    Yo, Yı, ya parallel to its axis. Otherwise by Cotes 's formula (2995).
    2993 A closer approximation , in terms of 2n + 1 equi
    distant ordinates, is given by Simpson's formula ,
    Yun + 4 (y . tyz + ... + y2n - 1)
    + 2 (ya + y :+ ... + yzn-»)]......(iii.)
    Proof. — Wehave

    f (x ) dx = f (x ) dx + f (x ) d . + ... + f (2 ) dæ .
    72

    Apply formula (ii.) to each integral and add the results, denoting by Yr the
    value of y corresponding to x = .

    2994 When the limits are a and b , the integral can be


    changed into another having the limits 0 and 1 , by sub
    stituting x = a + (b - a ) y .

    COTES’S METHOD .
    Let n equidistant ordinates, and the corresponding
    abscissæ , be
    Yo, Yu, Ya ... Yn–1» Yn and 0,nnn
    , 2 ... — 1, 1.
    2995 A formula for approximation will then be
    f(x)dx = A.4 + Ayyı + ...+ Ayy, + ... + Anyn (iv.),
    ( 12 )

    where A , = ( - 1)"+" 1 (NX )-1 dx. (2452)


    ro in - r Jo NX - 70
    438 INTEGRAL CALCULUS.

    Proof. — The method consists in substituting for f (x ) the integral


    function
    (nm)7 - '1 yo...
    , (n.x ) "
    + (x ) = ( - 1)" Lnx in
    + ( - 1 )" —
    (nz - 1 )Ľuer
    ...+ (- 1)" (12 -30;}n yu}
    q taking all integral values from 0 to n inclusive. When x = r, we have
    ų (r = Yr ; so that 1 (a ) has n + 1 values in common with f (x ). The
    approximate value of the integral is therefore (x ) dr, and may be written
    as in (iv .)
    By substituting 1 – x, it appears that
    r1 ( m2) )
    S edx = ( - 1 )" i dr ;
    Jo nx - r Jon & — (n - r )
    and therefore A , = Ar- r. Consequently it is only necessary to calculate
    half the number of coefficients in (iv .)
    2996 The coefficients corresponding to the values of n from 1 to 10 are
    as follows. Every number has been carefully verified , and two misprints in
    Bertrand corrected ; namely , 2089 for 2989 in line 8 , and 89500 for 89600 in
    line 11.

    n = 1: 4, = A,

    n = 2: 4 = 4, = 4, =
    n = 3: 4,= 4,= 1 4, = 4,=
    n =4: 4, = 4,= 1 = 4, = 2 A,=
    u = 5: 4, = d ; = .. 4, = 4, = 35 A, = A $ =
    » = 6: 4,= 4 = 4,= 4, = 4.= 4,=20 43=1*
    3577

    n = 7: 4,= 4, = 17230 4,= 4,= 17230


    2989
    A, = A3 = Az = A = 17280
    980 2944 464
    n = 8 : 4 , = A3 = 28350 ' A = A , = 141751
    41 AA , = A : = - i
    A; = A = 5248 2835
    14175 '4
    APPROXIMATE INTEGRATION . 439

    2857 157 + 1
    A - A.
    n = 9 : 40 = A , = 89600 : 4, = 1 89600' 42 - do 2240'
    1209 2889
    13 - 16 5000' 14 15 ~ 44800 *

    n= 10: A = 470= 26067 , 4, = 49, =- 1426575


    9688° 16175
    4 , = A9 = - 199584
    5675 4825 17807
    Az = Az = 12474 44 = 4 E = - llvos A5 = 24948

    GAUSS'S METHOD.
    2997 When f ( x ) is an integral algebraic function of degree
    2n, or lower, Gauss's formula of approximation is

    $ 5(x) = 4,f(x))+...+ 4,5(2-)+...+ 4,f(x ) (v.),


    where xo ... X.p ... X'n are the n + 1 roots of the equation
    4 (x ) = dinti)e {1x0+1(x — 1)"+1} = 0 ......... (vi.),
    and A1, =- 12? (x* — X .) ... (x — X -1)(x — Xr+1) ... (x — X ) de
    J. (X . - xo) ... (x , — X r-1)( x- — Xr+1)... (x - - xn)... (vii.)
    The formula is evidently applicable to a function of any
    form which can be expanded in a converging algebraic series
    not having a fractional index in the first 2n terms. The result
    will be the approximate value of those terms.
    Proof.— Let (c) = (x — x.)( 2 — ~ ) ... (2 — «n),
    and let f ( 0) = Q4 (22 ) + R ... ....... ... .... (viii.),
    where f ( x ) is of the 2nth degree, Q of the n - 1th , and R of the nth, since y (x )
    is of the n + 1th degree.
    Then the method consists in choosing a function ų (a ) of the n + 1th degree,
    so that Qu ( x ) dx shall vanish ; and a function R of the nth degree, which
    shall coincide with f (x ) when x is any one of the n + 1 roots of \ (x ) = 0 .
    (i.) To ensure that (x ) dx = 0 . We have, by Parts, successively ,
    writing N for + (x), and with the notation of (2148),
    SAPN = /1 - .(2 -1,0)
    =+* 1 -pop-" + (p -115.6--*|.-)
    = & c. & c.
    = ="| N –pa -| n +p(p -1)29- n -...+lp . ...
    J 20 ( p + 1) ix )
    Now Q4 (x ) is made up of termslike vº * (x ) with integral values of p from
    O to n - 1 inclusive. Hence, if the value (vi.) be assumed for 4 ( c ) , we
    440 INTEGRAL CALCULUS.

    see, hy (ix .), that | Qu (x)dx will vanish at both limits, because the factors
    x and x - 1 will appear in every term .
    (ii.) Let R be the function on the right of equation ( v .) Then , when
    = xr, we see, by (vii. ), that A , = 1, and that the other coefficients all vanish .
    Hence R becomes f (.1) whenever x is a root of y (x ) = 0 .

    The values of the constants corresponding to the first six values of n ,


    according to Bertrand , are as follows. The abscissæ values, only, have been
    recalculated by the author.

    n = 0: 2 = 5, A , = 1.
    n = l: x, = 21132487, A, = A , = 5, log = 9:6989700 ;
    2 , = .78807513 .
    n = 2 : 2 , = · 11270167, 4, = A , = is, log = 9.4436975 ;
    = 5 ;
    = 88729833, 4 , = s , log = 9.6478175.
    n = 3 : : 2 . = .06943184 , A, = A3 = 1739274, log = 9.2403681;
    = •33000948, A, = A, = -3260726, log = 9.5133143 ;
    = .66999052 ;
    = .93056316 .
    n = 4 : = .04691008 , A , = A , = .1184634, log = 9:073583 + ;
    = -23076531, A , = Az = 2393113, log = 9:3789687 ;
    = ') , A , = 2814411, log = 9 :4539975 ;
    = 76923 166 ;
    = .95308992.
    n = 5 : = .03376524, A , = Az = .0856622, log = 8.9327895 ;
    = .16939531, A, = A , = :1803808, log = 9-2361903;
    = '38069041, A , = Az = 2339570, log = 9: 3691360 ;
    = :61930959 ;
    = .83060469 ;
    = .96623476 .
    As a criterion of the relative degrees of approximation obtained by the
    foregoing methods, Bertrand gives the following values of
    ſ log (1 + r) dx = log2 = - 2721982613.
    J. 1 +
    Method of Trapezoids, 271.2837.
    Simpson 's method , n = 10 , 27-22012.
    Cotes's n = 5, 2722091.
    Gauss's n = 4, 2721980.
    For other formulæ ofapproximation, see also p. 357.
    CALCULUS OF VARIATIONS.

    FUNCTIONS OF ONE INDEPENDENT VARIABLE .

    3028 Let y = f (x ), and let V be a known function of x , y ,


    and a certain number of the derivatives
    chief object of the Calculus of Variations is to find the form
    of the function f (a ) which will make

    U= Jxo Vdx.... ..........(i.)


    a maximum or minimum . See (3084).
    enote Yæ, Yarı Yur, & c. by p , q, r , & c.
    For a maximum or minimum value of U , SU must vanish .
    To find 80 , let sy be the change in y caused by a change in
    the form of the function y = f ( x ), and let op , dq, & c . be the
    consequent changes in p , q, & c .
    Now , P = y.ro
    Therefore the new value of p , when a change takes place in
    the form of the function y , is
    p + op = (y + 8y): = y2 + (8y).,,
    therefore åp = (89)ą; that is, 8(dy ) = d (8y).
    Similarly , . dq = (8P )
    or = (89 )x, & c . ......... .........(ii.)
    Now dU = (* 8Vda (1483). Expand by Taylor's theorein ,
    rejecting the squares of dy, 8p, 89, & c., and we find
    re

    8U = 1* (V,8y + V,&p + V,89 + ...) dx,


    Jaxo

    or, denoting V, V , V, ... by N , P, Q , ...,


    8U = 1" (Nay + P &p + Q8q + ...) dx ............(iii.)
    3
    442 CALCULUS OF VARIATIONS.

    Integrate each term after the firstby Parts, observing that by


    (ii.) ( 8p dx = dy, & c., and repeat the process until the final
    cess

    integrals involve dydx. Thus


    | N8y dx is unaltered ,
    |Pàpda = Pêy -SP28yde,
    5 Q&p dx = Qëp — Qu8y + fQz-dyde,
    ( Rårdx = R&q — RĄ8p +Rzęy - | Rusdyda,
    .. . . .. .. . ...
    3029 Hence, collecting the coefficients of dy,dp, dq, & c.,
    .... .... 8U = (* (N - Pc + Q2: — R $ + ...) Sydx
    + 8y. (P - Q : + R2 - ...), — dy ( P - Qz + R2 - ...).
    + 8p. ( Q - R , + S24 - ...) - 8p . ( Q - R , + 52 - ...),
    + 891 ( R - S + T20 - ...), – 89. ( R - S , + 7 , - ...). + & c . (iv.)
    The terms affected by the suffixes 1 and 0 must have a
    made equal to X , and , respectively after differentiation .
    Observe that Pr, Qız, & c . are here complete derivatives ;
    y , p , q, r , & c., which they involve, being functions of x.
    Equation (iv .) is written in the abbreviated form ,
    Cri

    3030 SU = (* K8y8x + H . - H . ................. (v.)


    Uto

    The condition for the vanishing of 8U, that is, for mini
    mum value of U , is
    3031 K = N - P2 + Q26 – Rx + &c. = 0 .........(vi.),
    3032 and H - H , = 0 ........................ (vii.)
    Proof. — For, if not, wemust have
    | Kdy dx = H . - H , ;
    that is, the integral of an arbitrary function (since y is arbitrary in form ) can
    be expressed in terms of the limits of y and its derivatives ; which is impos
    sible. Therefore H , - H , = 0 . Also K = 0 ; for, if the integral could vanish
    without K vanishing, the form of the function dy would be restricted , which is
    inadmissible .
    FUNCTIONS OF ONE INDEPENDENT VARIABLE. 443

    The order of K is twice that of the highest derivative contained in V . Let


    n be the order of K , then there will be 2n constants in the solution of equa
    tion (vi.) and the same number of equations for determining them . For
    there are 2n terms in equation ( vii.) involving dy , dyo, dp , & c . If any of
    these qnantities are arbitrary, their coefficients must vanish in order that
    equation (vii.) may hold ; and if any are not arbitrary, they will be fixed in
    their values by given equations which , together with the equations furnished
    by the coefficients which have to be equated to zero , will make up, in all, 2n
    equations.

    PARTICULAR CASES.
    3033 1. — When V does not involve cc explicitly , a first
    integral of the equation K = 0 can always be found . Thus,
    if, for example ,
    V = ( , p, q, r, 8),
    a first integral will be
    V = Pp
    + Qp: – Qxp
    + Rp2- - R . P : + R2& P
    + Spx - S P2x + S2 Pe -Sp + C .
    The order of this equation is less by one than that of (vi.)
    Proof. — We have V . = Np + Pq + Qr + Rs.
    Substitute the value of N from (vi.), and it will be found that each pair of
    terms involving P , Q , R , & c . is an exact differential.

    3034 II. — When V does not involve y , a first integral can


    be found at once , for then N = 0, and therefore K = 0 , and
    we have Pe - Q2x + R34 - & c. = 0 ;
    and therefore P - Qu + R2 - & c . = A .
    3035 III. — When Vinvolves only y and p ,
    V = Pp + A , by Case I.
    3036 IV . - When V involves only p and q,
    V = Qq + Ap + B . See also (3046).
    PROOF. K = - Pq + Q2 = 0, giving, by integration, P = Qu + A .
    Also V . = Pq + Qr = Aq + Qxq + Qr.
    Integrating again , we find V = Qq + Ap + B ,
    a reduction from the fourth to the second order of differential equations.
    444 CALCULUS OF VARIATIONS.

    3037 Ex. - To find the brachistochrone, or curve of quickest descent,


    from a point () taken as origin to a point x,yu, measuring the axis of y down
    wards. *
    Velocity at a depth y = V2gy.
    Therefore time ofdescent = *dæ.
    Here v + A , by Case III .
    ✓ { y (1 + p ') } "
    By reduction, y(1+ p) = 1 = 2a, an arbitrary constant.
    That is, since p = tan 0, y = 2a cos? O, the defining property of a cycloid
    having its vertex downwards and a cusp at the origin
    H . - H . reduces to – - {( phy) - ( phụ) } = 0.
    If the extreme points are fixed , dy , and dy, both vanish .
    The values x1, Yu, at the lower point, determine a .
    Suppose x ,, but not yı, is fixed. Then dy, is arbitrary ; therefore
    its coefficient in (3 ) (P - Qz + & c.), must vanish ; that is, (V2) , = 0, or
    P } = 0, therefore 2, = 0, which means that the tangent at the
    i Vy (1 + p?)
    lower point is horizontal, and the curve is therefore a complete half cycloid .
    3038
    historia oy,both vanishirections, èpo,ops
    In the example of the brachistochrone, it is useful to notice that
    ( i.) If the extreme points are fixed , oy , oy , both vanish .
    ( ii.) If the tangents at the extreme points have fixed directions, dpo, op ,
    both vanish.
    (ii.) If the curvature at each extremity is fixed in value, dpo, oqo, &pu, ēqı
    all vanish.
    (iv .) If the abscisse 2 , X , only have fixed values, dyo, èy , are then
    arbitrary, and their coefficients in H , – H , must vanish.

    3039 When the limits Xog X , are variable, add to the value of
    dU in (3029) Vida, – Vºdxo.
    Proof. — The partial increment of U, due to changes in 22 and 2o, is
    du da + au dx, = V,dæ; – V,dło. By (2253).
    3040 When X and Yu, wo and y , are connected by given
    equations, y = * (Q ), Yo = x (xo).
    RULE . - Put
    8y = {v}(x3) — pı} dx, and dy, = {x'(x ) - po} dx ,
    * The Calculus of Variations originated with this problem , proposed by John Bern ulli
    in 1696 .
    FUNCTIONS OF ONE INDEPENDENT VARIABLE . 445

    and afterwards equate to zero the coefficients of dx, and dxo,


    because the values of the latter are arbitrary .
    PROOF. - Y. + ày, being a function of xz,
    ( y + dy .) + dz, ( . + &yı) dx , = 4 (x + dx,) = 4 (27) + 4 '(a ) diy;
    therefore èy , + p,dx, = 4 (2 ) dxı, neglecting opdx;.
    Ex. - In the brachistochrone problem (3037), the result thus arrived at
    signifies that the cycloid is at right angles to each of the given curves at its
    extremities.

    3041 If V involves the limits Xo, X , Yo, 41, Po, Pi, & c ., the
    terms to be added to dU in (3029) , on account of the varia
    tion of any of these quantities , are

    da, " {Ve +VP1 + Vp121 + ...} dæ


    + dæoſ" {V .+ VyPo+ V2.60+ ...} dæ
    + ( { Vw8yo + V,,dy + Vodpo + Vp dpi + &c.}dx.
    In the last integral, 8yo, 841, 8po, & c. may be placed outside
    the symbol of integration since, they are not functions of x . ,
    Hence, when V involves the limits Xo, X1, yo, 41, Po, Pu, & c.,
    and those limits are variable, the complete expression for
    8 U is

    3042 8U = S" {N - P. + Q2.– R3 + &c.}8ydx


    + {v +S"(V.+ V„Ps+ V, 2 +...)dx}dư;
    - {V.-8"(v..+ V»Po+ V» Go +...)dx}dx,
    + { P - Q.+ R3 - ...),+ Sºv,,dx }dya
    -{{P - Qs+ Ru -...) -S „dv}8y.
    + {(Q – R.+ S .-...) + S*v, dx }&p
    - {(Q - R : + S20 - ...). — * v ,dx } po + &c.
    446 CALCULUS OF VARIATIONS.

    3043 Also, if y = 4 (x ) and yo = x (x ) be equations re


    stricting the limits, put
    dyn = {4'(;) — Pa} dạy and 8yo = {x'(mo)– Po} deo. (3040)
    The relation K = 0) is unaltered , and , by means of it, the
    additional integrals which appear in the value of H . - H .
    become definite functions of x .

    3044 Ex.- To find the curve of quickest descentof a particle from some
    point on the curve y = x (x ) to the curve Yı = 4 (x ) .
    As in (3037), t = 1 fxi 11 + p ? 11 + pS
    -, and contains y ,
    ✓ (29) J...Vy - yo Vy - yo
    P, and yo. Equation (3042) now reduces to

    ou = [ (N –P.)dyda+ ,da –{ .- ProPoda}da,


    + Proy –{P.-|*Vydx}oyo......(1).
    Now K = 0 gives N - P , = 0 ; therefore V = Pp + A (3035) ;
    1 + p3 - - pa - + A.
    therefore
    Vy - yo
    *
    ✓ ( y - yo) ( 1 + p ?)
    Clearing of fractions,and putting A = --✓ ( 2a )
    this becomes

    (y - yo)( 1 + p%) = 2a ......... .. . . .. . . . ( 2 )

    Also
    Also PP == Vo
    V, == 7y-y)(1+2")} = 142a) ..
    Hence
    ce
    V= ; Vy = - V, = - N = - P ,(by K = 0),
    therefore e = Po - P , = Po - P . ........ ......... (4).
    ✓ ( 2a )

    Substituting the values ( 2), (3 ), (4 ), in (1), the condition 7 , - H , produces


    (1 + pi)dx – (1 + PoPx) dx, + p, dy, - p,dyo = 0.
    Next, put for dy, and dy , the values in (3040) ; thus the equation becomes
    { 1 + P1 “' (xx) } dx - { 1 + P . x '(x ) } dx = 0 ................ (5 ) ;
    dxı, dx, being arbitrary, their coefficients must vanish ; therefore
    Pit' (x1) = - 1 and Pix' (x ) = - 1 .
    That is, the tangents of the given curves y and x at the points x ,yo and my,
    are both perpendicular to the tangent of the brachistochrone at the point xyi.
    Equation (2 ) shews that the brachistochrone is a cycloid with a cusp at
    the starting -point, since there y = Yo, and therefore p = 0 .
    FUNCTIONS OF ONE INDEPENDENT VARIABLE. 447

    OTHER EXCEPTIONAL CASES.


    (Continued from 3036.)
    3045 V . --Denoting y ,Yvo Y2x ... Yno by Y, P1, P2 ... Pri
    and V, V , Vp ... V by N ,P1, P , ... Pui
    let the first m of the quantities y, P1, P2, & c. be wanting in the
    function V ; so that
    V = f(®, Pm,Pm+1 ... Pon).
    Then K = dmo P.mo — dựm +1)?Pm+1 + ... ( — 1)»-mdmg Pn = 0,
    TA

    which equation , being integrated m times, becomes


    Pm - d.Pm +1+ d20 Pm +2– ... (- 1)^-mdin-m)o Pne
    | = Co +cº+ ... + m–13”- ......(i),
    a differential equation of the order 2n - m .
    3046 VI. — Let æ also be wanting in V, so that
    V = f (Pm Pm +1... Pn);
    then K = 0 is the same as before, and produces the same
    differential equation (i.) From that equation take the value
    of Pm , and substitute it in
    V . = PmPm+ 1 + Pm +1Pm+2 + ... + Pn Pn +1.
    Each pair of terms, such as Pm + 2 Pm +s — d26Pm + 2 Pm + 1, is an
    exact differential; and we thus find
    V = c + Pm+ 1Pm+1 + (Pm +2Pm + 2 — d ,Pm +2Pm+1) + ...
    + (PnPr — d ,PnPn -1 + d2, P Pn -2) — ... ( – 1)n-m -1dın-m -1)= P „Pm+1
    + (co +Cæ + ... + Cm -jam -1) Pm +,dx.
    The resulting equation will be of the order 2n - m - 1 , or
    m + 1 degrees lower than the original equation .

    3047 VII. — If V . be a linear function of pn, that being


    highest derivative it contains, P , will not then contain
    Therefore dne P , will be, at most, of the order 2n - 1. In
    deed , in this case , the equation K = 0 cannot be of an order
    higher than 2n - 2 . (Jellett, p. 44 .)
    448 CALCULUS OF VARIATIONS.

    3048 VIII. — Let pn be the lowest derivative which V in .


    volves ; then , if Pm = f ( x ), and if only the limiting values of
    x and of derivatives higher than the mth be given , the problem
    cannot generally be solved . ( Jellett, p . 49.)

    3049 IX . - Let N = 0 , and let the limiting values of x alone


    be given ; then the equation K = 0 becomes
    Pr – Q2x + R20 - & c . = 0,
    or, by integration , P - Qx + R2, - & c. = c,
    and the two conditions furnished by equating to zero the co
    efficients of dyn, dyo, viz .,
    ( P - Qc + & c.), = 0 , ( P - Q + + & c.), = 0 ,
    are equivalent to the single equation c = 0 , and therefore
    H - H , = 0 supplies but 2n - 1 equations instead of 2n , and
    the problem is indeterminate.

    3050 Let U = " V dæ + V', where


    V = F (x, y , p , q ...) and V ' = f (xo, X , Yo,Yu, Po,P1, & c.)
    The condition for a maximum or minimum value of U arising
    from a variation in y , is, as before , K = 0 ; and the terms to
    be added to H . - H , are
    Vdxo + V % 84 + V6.8p . + ... + Vadå + Vydy. + & c .
    If the order of V be n , and the number of increments dxo, 840,
    & c. be greater than n + 1 , the number of independent incre
    ments will exceed the number of arbitrary constants in K , and
    no maximum or minimum can be found.
    Generally , U does not in this case admit of a maximum
    or minimum if either V or V ' contains either of the limiting
    values of a derivative of an order = or > than that of the
    highest derivative found in V . ( Jellett, p . 72.)

    FUNCTIONS OF TWO DEPENDENT VARIABLES.


    3051 Let V be a function of two dependent variables y , z ,
    and their derivatives with respect to x ; that is , let
    V = f ( x , y , p , q ... 2 , p . q .... ............... ( 1) ,
    FUNCTIONS OF TWO DEPENDENT VARIABLES. 449

    where p , q, ... , as before , are the successive derivatives of y ,


    and p ', s', ... those of % .
    Then , if the forms of the functions y , z vary, the condition
    for a maximum or minimum value of U or “ Vdx is
    8U = * (K8y + K '8x)dx + H. - H .+ H - H '= 0 ...(2).
    Here K ', H ' involve , p’, q', ..., precisely as K , H involve
    y , p, q , ... ; the values of the latter being given in (3029).

    3052 First, if y and «z are independent, equation (2) ne


    cessitates the following conditions :
    K = 1, K ' = 0, H . - H .+ H , - H ' = 0 ...... (3).
    The equations K = 0 , K = 0 give y and z in terms of a ,
    and the constants which appear in the solution must be deter
    mined by equating to zero the coefficients of the arbitrary
    quantities 8yo, dy, dpo, op. ... 8zo, 8zz, opó, dpa's ...,
    which are found in the equation
    H . - H .+ H ,' - H / = 0 .................. (4 ).

    3053 The number of equations so obtained is equal to the


    number of constants to be determined .
    Proof. – Let V = f (x , y , Y ., Yze ... Ynch Z , % 224 ... *mx),
    K is of order 2n in y, and . . of form ® (@ , y, Yo ... Y2nx, %,7g ... %(m +m ): ) ... (i.),
    K ' is of order 2m in z, and ... of form © (x , y , Yx ... Y (n +n )xe , ... % 2mx) ... (ii.)
    Differentiating (i.) 2m times, and (ii.) mun times, 3m + n + 2 equations are
    obtained , between which , if we eliminate % , kz .. . 213m + n ) , we get a resulting
    equation in y , of order 2 (m + n ) , whose solution will therefore contain
    2 (m + n ) arbitrary constants. The equations for finding these are also
    2 (m + n ) in number, viz., 2n in H . - H , and 2m in Hi - H ..

    3054 Note. — The number of equations fordetermining the constants is not


    generally affected by any auxiliary equations introduced by restricting the
    limits. For every such equation either removes a term from ( 4 ) by an.
    nulling some variation (dy, dp , & c . ) , or it makes two terms into one ; in each
    case diminishing by one the number of equations, and adding one equation,
    namely itself.

    3055 Secondly, let y and be connected by some equation


    3 m
    450 CALCULUS OF VARIATIONS.

    (xyz ) = 0. Y and > are then found by solving simultane


    ously the equations
    $ ( x, y, z) = 0 and K : $ , = K ' : 0 ..
    PROOF. — (2 , y, z ) = 0 , and therefore 0 (s , y + ày, x + 82 ) = 0, when the
    forms of y and z vary . Therefore gydy + pz8z = 0 (1514 ). Also Koy + K '&z = 0,
    by (2 ). Hence the proportion.

    3056 Thirdly, let the equation connecting y and a be of the


    more general form
    $ (x, y,p ,q ... % ,pige...) = 0...... ........(5).
    By differentiation , we obtain
    Pydy + Pp8p + qq&q + ... 028z + $p &p' + 9q80 + ... = 0... (6 ).
    If (which rarely happens) this equation can be integrated so
    as to furnish a value of £z in terms of dy , then op ' , dq', & c .
    may be obtained, by simple differentiation , in terms of Sy, op.
    Generally, we proceed as follows :
    .8V = N8y + P8p + Q &q + ... + N '8z + Pf8p' + Q+8q' + ... ...(7).
    Multiply (6) by , and add it to (7), thus
    8V = ( N + 19,)dy + ( P + 10p) 8p + ...
    ... + (N ' + 102) 8z + ( P' + 10p) 8p' + ... ......(8 ).
    The expression for 8U will therefore be the same as in (2 ), if
    we replace N by N + 10y, P by P + 10p, & c., thus
    Ci

    3057 SU = ("[ {(N + 10,) - (P + 10p) + ...} dy


    + { (N ' + 10 .) - ( P ' + 10p)2 + .. } 82 ] d.o
    + { P + 10 , - (Q + 10g) + ...}idy.
    - {P + 0 , - (Q + 0a). + ...}. dyo
    + { Q + 10 , - (R + 101) +. .}i Sp.
    - { Q + 16, - ( R + 10 +)2 + ...}. dp
    & c. & c.
    + similar terms in P , Q ... p , a ... & c. ...(I).
    3058 To render SU independent of the variation Sz, we must
    FUNCTIONS OF TWO DEPENDENT VARIABLES. 451

    then equate to zero the coefficient of 8x under the sign of


    integration ; thus
    N ' +10: - ( P' + 10p)2 + ( Q' +104)2. - & c. = 0......(10),
    the equation for determining 1 .
    3059 Ex. (i.)— Given V = F( ,y,P,9...2),where
    % = fvdx and v = F (x , y, p , q ...).
    The equation o is now 2 - Svdx = 0 or v — , = 0,
    Py = Vy PP = Vp , Qq = vq, & c.,
    92 = 0 , pe = - 1, 0g = 0, the rest vanishing .
    Substituting these values in (9), we obtain
    & U = 1" [{N +10, - (P + 1v ) + (Q + 1v ) - ...} 8y + { N' +1 ;}8:]dæ
    + {P + 10, - ( Q + 102)2 + ...};dy, - { P + 10, - ( Q + Ava)x + ... }ody.
    + {Q + 10, - (R + Avr).& + ...}, 8p. - {Q + 10 , - ( R + 1v.): + . ..}. èp, + & c.
    For the complete variation DU add V dx, – Vdxo. To reduce the above so
    as to remove dz, we must put N ' + 1, = 0, and therefore = - | Ndx. Let
    ^ = u be the solution, u being a function of x , y, p , q ... %. Substituting this
    expression for , the value of ĉ U becomes independent of dz.

    Ex. (ii.) — Similarly, if z in the last example be = fmv (2148), o becomes


    P- 2Dz = 0 ; and, to make N ' + ' po vanish, we must put 1 = -SN'.
    3061 Ex. (ii .) – Let ✓1 + y + zdx ........... ..........(1)
    Here N = 0 ; N ' = 0 ; P =_ _ Q = 0;
    ✓ 1 + p + p?? ✓1 + p + p ?'
    = 0 ; and the equations K = 0, K’ = 0 become
    PR = 0 , Pe = 0 , or ✓1 + Ppå + p = 0,
    V1 + p' + p h = 6 .
    olving these equations, we get
    Ye = m ; Zz = n ; ory = mx + A ; % = nx + B .

    02 First, if X1, 91, W , X , Yo, zo be given , there are four equations to


    etermine m , n , A , and B .
    This solves the problem , to find a line of minimum length on a given
    ved surface between two fixed points on the surface.

    63 Secondly, if the limits x,, x, only are given, then the equations
    are
    ( P ), = 0, ( P ). = 0 , ( P ), = 0 , ( P ), = 0,
    e only
    unde equivalent
    termin ed .
    to the two equations m = 0 , 1 = 0 , and A and B remain
    452 CALCULUS OF VARIATIONS.

    3064 Thirdly, let the limits be connected by the equations


    0 (21, 91,2,) = 0 , y (xo,Yo, 7o) = 0.
    We shall have (4 :, + Pupa + 02,pí) dx, + Oy,dy , + 4z, 8z, = 0.
    Substitute Ox, = m097, 02 = n47, P. = m , pi = n ; thns
    (1 + mm , + nn,) dx, + m ,dy, + n , ezz = 0 .
    Eliminate dx, by this eqnation from
    . V,dx, + ( P ),dy, + ( P '), èz, = 0,
    and equate to zero the coefficients of dy , and dzı ; then
    mV = ( P ), (1 + mm , + nn ) ; n V = ( P ), ( 1 + mm , + nn ).
    Replacing V , P , by their values, and solving these equations for m and n ,
    we find m = m , n =
    Similarly from the equation (xo, Yo, 2.) = 0 we derive m = mo, n = ng
    Eliminating 1, 91, 22, , Yo,z , between these equations, and
    y = mx; + A ; 2q = nzi + B ; yo = mx, + A ; 2. = nx, + B ;
    Q (x1, 91, 2 ) = 0 ; ( lo, Yo, z ) = 0 ;
    four equations remain for determining m , n , A , and B .

    3065 On determining the constants in the solution of (3056).


    Denoting P , 9, r ... by P1, P2, P3 ... , we have
    V = F (x , y ,P1,P2... Pn, ,pí, pa... Pm) ;
    and for the limiting equation ,
    $ ( x, , P1, P2, ... Pr, z,pí, pa, ... pm') = 0 .
    V is of the order n in y and m in %.
    o is of the order n ' in y and m ' in z.
    3066 RULE I. - If m be > m ', and n either > or < n ', the
    order of the final differential equation will be the greater of the
    two quantities 2 ( m + n '), 2 (m ' + n ) ; and there will be a
    sufficient number of subordinate equations to determine the
    arbitrary constants.
    3067 Rule II. - I/ m be < m ', and n < n', the order of the
    final equation will generally be 2 (m ' + n ') ; and its solution
    may contain any number of constants not greater than the least
    of the two quantities 2 ( m ' — m ), 2 (n' - n ).
    For the investigation, see Jellett, pp. 118 — 127.

    3068 If V does not involve ix explicitly, a single integral of


    order 2 (m + n ) - 1 may be found . The value of V is that
    given in (3033), with corresponding terms derived from z .
    FUNCTIONS OF TWO DEPENDENT VARIABLES. 453

    Proof. - dV = Ndy + P , dp, + ... + Pudpn + N 'dz + Pidpí + ... + Pmdpm.


    Substitute for N and N ' from the equations K = 0, K ' = 0 , as in (3033),
    and integrate for V .

    RELATIVE MAXIMA AND MINIMA.


    3069 In this class of problems, a maximum or minimum
    101
    ma im m

    value ofan integral, U . = V dx , is required , subject to the


    condition that another integral, U, = (* V,da, involving the
    same variables, has a constant value.
    RULE. — Find the maximum or minimum value of the func
    tion U , + aU , ; that is, take V = VitaV2, and afterwards
    determine the constant a by equating U , to its given value.
    For examples, see (3074), (3082).
    GEOMETRICAL APPLICATIONS.
    3070 PROPOSITION I. — To find a curve s which will make
    F (x,y) ds a maximum or minimum , F being a given function
    of the coordinates x , y .
    The equation (5), in (3056 ), here becomes
    p + pº = 1 ;
    where p = 2 , p = Y., x and y being the dependent variables, and s the in .
    dependent variable .
    In (3057), we have now ,writing u for F (x, y),
    N = Uqs N' = Uys Pe = 2p, Op = 2p';
    the rest zero . The equations of condition are therefore
    Ur - d , (^« .) = 0 and U , — d, ( ^yn) = 0 ............
    Multiplying by x, y, respectively, adding and integrating, the result is
    = u,
    the constant being zero.*
    Substituting this value in equations (1), differentiating uz, and uyn, and
    putting Ug = U2% . + WyYo we get
    y. (Uzy - ) = UIzg........ . ... .. . . . . ( 2 ) ,

    2 .(U,2 ,— _zy.) = uyzs.............. .. .......... ..


    • See Todhunter's " History," p. 406 .
    NS
    454 CALCULUS OF VARIATIO .

    Multiplying (2 ) by y , and (3) by xom and subtracting, we obtain finally


    w ( ,225 — ,428) = uzys — U,#g, or
    3071 u – du dy du dx .. ... .. . ... .. .. .. ( 4 ) ,
    ē dx ds dy ds
    p being the radius of curvature .
    To integrate this equation , the form of u must be known.
    and , by assigning different forms, various geometrical theorems
    are obtained .

    3072 PROPOSITION II. - To find the curve which will make

    F (x,y)ds+ ( f(x,y)dx ..................(1)


    a maximum or minimum , the functions F and f being of given
    form .
    Let F (x , y ) = u and f (x , y ) = v .
    Equation (1 ) is equivalent to S ( u + vxg) ds.
    In (3057) we now have V = u + vp ; and for ¢ , p + p = 1, as in (3070) .
    Therefore N = Uz + pvz ; P = V, = v ; Op = 2p ;
    N ' = Uy + pv, ; p = 2p' ; the rest zero.
    Therefore, equating to zero the coefficients of dx and dy, the result is the two
    equations Uz + pvc - (v + 1p ), = 0 ,
    U , + pv, - ( p ) , = 0 ;
    or d , (Axg) + v, = uz + 2 ,0z)
    d. (^ys) = U , + Vya
    Multiplying by X , Y , respectively, adding, and integrating, we obtain , as in
    (3070 ), l = u , and ultimately ,
    3073 1 ( du dy du dx , dv
    mã dx ds dy ds + dy ) :
    3074 Ex. — To find a curve s of given length,such that the volume of the
    solid of revolution which it generates about a given line may be a maximum .
    Here S (yºx , - a²)ds must be a maximum , by (3069), a ' being the arbitrary
    constant. The problem is a case of (3072),
    u = a*, U , = 0, Ug = 0, v = y , v, = 2y.
    Hence equation (3073) becomes 1 = 2
    Giving p its value, – (1+2)*(where
    PP , p- = dx%),
    ) and integrating,the result

    ; from which = [ 'a'(yº-+ (y?b?)+dy33)


    + 63 ; from which
    = y®7+%
    FUNCTIONS OF TWO INDEPENDENT VARIABLES. 455

    FUNCTIONS OF TWO INDEPENDENT VARIABLES.

    3075 Let V = f (x , y ,z , p , q, r, s, t),


    in which x , y are the independent variables, and p, 9, 7, s, t
    stand for Zz, Zy, 222 , 7qy , Z2y respectively ( 1815 ) , z being an in
    determinate function of x and y .
    Let U = Jxo ( " Vdxdy,
    1" dyo
    and let the equation connecting ~ and y at the limits be
    $ ( x , y ) = 0. The complete variation of U , arising solely from
    an infinitesimal change in the form of the function % , is as
    follows :
    Let V ., Vp, & c . be denoted by Z , P , Q , R , S , T .
    Let = ( P - R . - 19,) 8z + 188q + Rap,
    * = ( Q - T , - 18.) 8z + 1S8p + 789,
    x = (Z - P . - Qy + R2 + Sxv + Txv) 82.
    The variation in question is then
    Cri /

    3076 8u = S*( v=n4y=ro+ Pu=vody -Py=» % )dæ 1X


    c =

    + [S*$dy] + S*S*xdx dy.


    LUV xr
    Jr == ro J
    J .to Jyo

    Proor.- o *P*vdx dy = $*{"ov de dy


    Jxo J vo JXoJyo

    Ixolyo
    (Z8z + Pdp + Qaq + Rör + S &s + Têt)dx dy

    = Js0 Jyo ( do + dyauto+x}de dy


    as appears by differentiating the values of ¢ and y . But

    by(2257),and a dy = vyava–Yvave
    Hence the result.

    3077 The conditions for a maximum or minimum value of


    O are, by similar reasoning to that employed in (3032),
    p = 0, = 0, x = 0.
    456 CALCULUS OF VARIATIONS.

    GEOMETRICAL APPLICATIONS.
    3078 Proposition I. — To find the surface, S, which will
    POS

    make || F (x, y,z)dS a maximum orminimum , F being a given


    function of the coordinates x , y , z . [ Jellett, p . 276.
    Here, putting F (x, y , z ) = u, V = uvi + p + q° ;

    2 = V1+pP+ + PE vitemetigh Q = vi that


    and V , V , V , are all zero.
    de laseu dese)+x(4(1,+Orange,
    + p* + q*)
    ( 1 + p ?) t - pas
    V1 + p + q* ( 1 + pº + q?) }
    The equation x = 0 or Z - P . - Qy = 0 gives
    ( 1 + q?) r – 2pgs + ( 1 + p ?) t 1
    (1 + p * + 9°)? Uv1 + p* + q !
    If R , R ' be the principal radii of curvature, and l, m , n
    the direction cosines of the normal, this equation may be
    written

    3079 + + + + + (mme + mo din + une = 0,


    and according to the nature of the function u different
    geometrical theorems may be deduced .

    3080 PROPOSITION II. - To find the surface S which willmake


    sf F (x, y,z) dS + {\f(x,y,z)dædy
    ma
    a maximum or minimum ; F and f being given functions of
    a umum III

    the coordinates x , y , z .
    Let F ( x, y, z ) = u and f (x, y, z ) = v . Proceeding throughout as in
    (3078 ), we have V = uv1+ p + q* + v,
    Z = V1 + p + q* uz + v .,
    and the remaining equations the same as in that article if we add to the
    resulting differential equation the term - ! on the left.
    APPENDIX . 457

    This equation may then be put in the form


    3081 + =- o meget
    where l, m , n are the direction cosines of the normal to the
    surface.

    3082 Ex. - To find a surface of given area such that the volume contained
    by it shall be a maximum .
    By (3069),the integral |(z -av1+ p + g )dady
    must take a maximum or minimum value. The problem is a case of (3080).
    Wehave u = - a , v = z, Ug = 0 , Ug = 0 , Ug = 0, 0, = 1 ;
    and the differential equation of the surface (3081) reduces to
    (1+ q*) r –2078 + (1 +pº)t + = (1 +p +q0)& = 0 ;
    3083 or +1

    APPENDIX .

    ON THE GENERAL OBJECT OF THE CALCULUS OF


    VARIATIONS.
    3084 DEFINITIONS. — A function whose form is invariable is
    called determinate, and one whose form is variable, indeter
    minate .
    Let du be the increment of a function u due to a change
    in the magnitude of the independent variable , du that due to a
    change in the form Lof
    hen
    the function , Du the total increment
    from both causes ; then
    Du = du + ou .
    Thus, in (3042), the terms involving dx, and dixo constitute du ,
    and the remaining terms du ; the whole variation being Du .
    du is called the variation of the function u .
    3085 A primitive indeterminate function , w , of any number
    of variables is a function whose variation is of arbitrary but
    constant form ; in other words, 8u = 0 .
    3 N
    458 CALCULUS OF VARIATIONS.

    3086 Let v = Fu be a derived function , — that is, a func


    tion derived by some process from the function u ; F denoting
    a relation between the forms, but not between the magnitudes,
    of u and v.
    The general object of the Calculus of Variations is to
    determine the change in a derived function v , caused by a
    change in the form of its primitive u .
    The particular derived functions considered are those
    whose symbols are d and , denoting operations of differ
    entiation and integration respectively .
    SUCCESSIVE VARIATION.
    3087 Let the variation of the variation , or second variation
    of V due to a change in the form of the involved function ,
    y = f (x ), be denoted by 8 (8V ) or 82V ; the third variation by
    8 V , and so on .
    By definition ( 3085), y being a primitive indeterminate
    function , and dy its variation, dy = 0 ... ... ( 1 ) .
    3088 The second variation of any derivative of y is also
    zero, i.e., dép , 829, & c . all vanish.
    · Proof.- q (yns) = 8 (@ynx) = {8 (8y) }ue = (SⓇY)n2 = 0 by ( 1).
    3089 If V = f (x , y , p, q , r , & c. ...), where y is a primitive
    indeterminate function of x , then
    8» V = (Syd, + Spdp+ 8qd,+ ...)" V,
    where, in the formal expansion by the multinomial theorem ,
    dy , dp , & c . follow the law of involution , but the indices of d , ,
    dip, & c . indicate repetition of the operation dy, dp, & c. upon V .
    · Proof. – First, 8V = ( èy d , + dpd, taqd, t ...) V .
    In finding 02V , each product, such as dy d , V , is differentiated again as a
    function of y , p , q , & c. ; but, since the variations of dy, dp, & c . vanish by (2 ),
    it is the same in effect as though òy , dp , & c. were not operated upon at all.
    They accordingly rank as algebraic quantities merely, and therefore
    ?V = (dyd, + op dp + oqda + ...) ' V .
    Similarly for a third differentiation ; and so on.

    A IMMEDIATE INTEGRABILITY OF THE FUNCTION V .


    3090 DEF. — When the function V (3028) is integrable
    without assigning the value of y in terms of x , and therefore
    APPENDIX . 459

    integrable whatever the form of the function y may be, it is


    said to be immediately integrable, or integrable per se .
    3091 The requisite condition for V to be immediately
    integrable is that K = 0 shall be identically true.
    PROOF. — v dx must be expressiblo in the form
    $ (x, 91P191 ...) - ° (Toyo Poqo ...),
    where o is independent of the form of y . Hence, a change in the form of y,
    which leaves the values at the limits unaltered , will leave
    8 Vdx = 0 ; that is, Koy = 0 .
    But the last equation necessitates K = 0 , since oy is arbitrary. And K = 0
    must be identically true, otherwise it would determine y as a function of x .
    DIFFERENTIAL EQUATIONS.

    GENERATION OF DIFFERENTIAL EQUATIONS.


    3050 By differentiating ordinary algebraic equations, and
    eliminating constants or functions, differential equations are
    produced . Some methods are illustrated in the following
    examples.

    3051 From an equation between two variables and n


    arbitrary constants , to eliminate the constants .
    RULE. — Differentiate r times (r < n ), and from the r + 1
    equations any r constants may be eliminated , and thus C ( n , r )
    different equations of the rth order (3060) obtained , involving
    dry dr- ly
    3 , & c.
    dxri dxr-1" Only r + 1, however,of these equationswill be
    independent. By differentiating n times and eliminating the
    constants , a single final differential equation of the nth order
    free from constants may be obtained .
    3052 Ex.- To eliminate the constants a and b from the equation
    y = ax' + bx ...... ......... (i.)
    Differentiating, we find dy = 2ax + b .... ............................ (ii.)
    Eliminating a and b in turn, we get
    con +bæ = 2y, a due = axº+ y.
    dx .... (iii, iv.)
    Now , differentiating (iii.) and eliminating b produces the final equation of
    the second order,
    vzdy 22 % + 2y = 0........ ..... ....(v.)
    The same equation is obtained by differentiating (iv.) and eliminating a .
    3053 To eliminate the function o from the equation 2 = ° (u),
    where v is a function of x and y . We have
    Kp = $ '(v) V= = '(v) ,.
    Therefore 2t = %, Ur.
    GENERATION OF DIFFERENTIAL EQUATIONS. 461

    3054 To eliminate o from u = ° (v ), where u and v are func


    tions of x , y , z.
    Consider x and y the independent variables, and differ
    entiate for each separately , thus
    Ur + uz . = ' (v) (vx + v ,ze),
    U , + % , = $ ' (v ) (v, + vozy),
    and, by division, $ ' (v ) is eliminated.

    3055 To eliminate Pi, $29 ... Pn from the equation


    F (x, y,z, qı(az), $2(az), ... Pu (an)} = 0,
    where an,uz, ... an are known functions of x , y, z.
    Rule . — Differentiate for x and y as independent variables,
    forming the derivatives of F of each order, up to the (2n - 1) th
    in every possible way ; that is , F ; Fr, F , ; F2 , Fry , F2 ; 8c.
    There will be 2n ? unknown functions, consisting of p1, 02, ... Or
    and their derivatives, and 2n + n equations for eliminating
    them .

    3056 To eliminate ], $ 1(E), Ⓡ2 ( ), ... Pn( ) between the


    equations
    F {x, y,z, £, 41(k), 41(E) ...Pu( )} = 0,
    f {x , y , z , & , 01( ), P2( ) ... On (F) } = 0 .
    RULE. — Consider zand functions of the independent
    variables x , y , and form the derivatives of F and f up to the
    2n - 1th order in the manner described in (3055 ). There will
    be 4nº + n functions, and 4n² + 2n equations for eliminating
    them .

    3057 To eliminate p from the equation


    F {x , y ,z , w , ° (a, b ) } = 0 ,
    where a , ß are known functions of x , y, z, w .
    RULE . — Consider x , y , z the independent variables . Dif
    ferentiate for each , and eliminate $ , Pas De between the four
    equations.
    462 DIFFERENTIAL EQUATIONS.

    DEFINITIONS AND RULES .


    3058 Ordinary differential equations involve the derivatives
    of a single independent variable.
    3059 Partial differential equations involve partial deriva
    tives, and therefore two or more independent variables are
    concerned .
    3060 The order of a differential equation is the order of the
    highest derivative which it contains.
    3061 The degree of a differential equation is the power to
    which the highest derivative is raised .
    3062 A Linear differential equation is one in which the
    derivatives are all involved in the first degree.
    3063 The complete primitive of a differential equation is
    that equation between the primitive variables from which the
    differential equation may be obtained by the process of differ
    entiation .

    3064 The general solution is the name given to the complete


    primitive when it has been obtained by solving the given
    differential equation .
    Thus, reverting to the example in ( 3051) , equation (i.) is the complete
    primitive of (v .)which is obtained from (i.) by differentiation and elimination .
    The differential equation (v .) being given, the process is reversed .
    Equations (iii.) and (iv .) are called the first integrals of (v .), and equation
    (i.) the final integral or general solution .

    3065 A particular solution , or particular integral, of a


    differential equation is obtained by giving particular values to
    the arbitrary constants in the general solution .
    For the definition of a singular solution., see (3068 ) .
    3066 To find when two differential equations of the first
    order have a common primitive.
    RULE. — Differentiate each equation , and eliminate its
    arbitrary constant. The two results will agree if there is a
    common primitive, which, in that case , will be found by elimi
    nating yg between the given equations.
    Ex. - Apply the rule to equations (iii.) and (iv .) in (3052) .
    SINGULAR SOLUTIONS. 463

    3067 To find when two solutions of a differential equation ,


    each involving an arbitrary constant, are equivalent.
    RULE . — Eliminate one of the variables. The other will also
    disappear, and a relation between the arbitrary constants will
    remain .
    Otherwise, if V = 0 , v = be the two solutions: V and v
    being functions of the variables, and C and c constants ; then
    dv dv dv dv
    dx dy = dy dx
    is the required condition .
    Proof. – V must be a function of v. Let V = º (u) ; therefore Ve = pove
    and V , = powy; then eliminate doo .

    Ex. - tan -' (x + y ) + tan - ' ( — y ) = a and 2 + 2x = y2 + 1 are both solu


    tions of 2xy yz = 2 + y + 1. Eliminating y, x disappears, and the resulting
    equation is b tan a = 1.

    SINGULAR SOLUTIONS.
    3068 DEFINITION. — “ A singular solution of a differential
    equation is a relation between x and y which satisfies the
    equation by means of the values which it gives to the differ
    ential coefficients Yu, Yur, & c., but is not included in the com
    plete primitive." See examples (3132 – 3 ).

    3069 To find a singular solution from the complete primitive


    9 (x , y , c) = 0 .
    RULE I. – From the complete primitive determine c as a
    function of x , by solving the equation y = 0 , or else by solving
    X = 0 , and substitute this value of c in the primitive. The
    result is a singular solution , unless it can also be obtained by
    giving to c a constant value in the primitive .
    3070 If the singular solution involves y only , it results from
    the equation y . = 0 only , and if it involves x only , it results
    from x = 0 only. If it involves both x and y , the two equa
    tions x = 0 , y . = 0 give the same result .

    TNA )
    464 DIFFERENTIAL EQUATIONS.

    3071 When the primitive equation • (xyc) = 0 is a rational


    integral function , $ c = 0 may be used instead of x = 0 or
    y = 0.
    Proof. — Let 9 (x, y , c) = 0 be expressed in the form
    y = f (x , c) .... ..... . .. . .. . .. . . .. ( 1 ) .
    Then, if c be constant, Yo = fr .......... .......... (2 ) ;
    and, if c varies, Yr = fr + f.Co ....... ..... ( 3 ).
    Wben c is constant, the differential equation of whicb ( 1) is the primitive is
    satisfied by the value of y , in ( 2 ) . But it will also be satisfied by the same
    value of ye when cis variable , provided that either fe = 0 or fx = 0 , and in
    either case a solution is obtained which is not the result of giving to c a
    constant value in the complete primitive ; that is, it is a singular solution .
    But f . = 0 is equivalent to yo = 0, and fx = 00 makes ya = 00 , and therefore
    2 = constant.

    GEOMETRICAL MEANING OF A SINGULAR SOLUTION .


    3072 Since the process in Rule I. is identical with that
    employed in finding the envelope of the series of curves
    obtained by varying the parameter c ' in the equation
    $ (x , y, c) = 0 ; the singular solution so obtained is the equa
    tion of the envelope itself .
    An exception occurs when the envelope coincides with one
    of the curves of the system .
    3073 Ex. — Let the complete primitive be
    y = cx + V1- C , therefore y = x - - C 5 ; y = 0 gives cr -,
    ✓1 - ' v1 +x2
    Substituting this in the primitive gives y = V1+ x*, a singular solution . It
    is the equation of the envelope of all the lines that are obtained by varying
    the parameter c in the primitive ; for it is the equation of a circle, and the
    primitive, by varying c, represents all lines which touch the circle . See also
    ( 3132– 3 ).

    3074 “ The determination of c as a function of x by the solution of the


    equation yo = 0 , is equivalent to determining what particnlar primitive has
    contact with the envelop at that point of the latter which corresponds to a
    given value of 2 .
    “ The elimination of c between a primitive y = f (x, c) and the derived
    eqnation y. = 0, does not necessarily lead to a singular solution in the sense
    above explained .
    “ For it is possible that the derived equation yo = 0 may neither, on the
    one hand, enable us to determine c as a function of x, so leading to a singular
    solution ; nor, on the other hand, as an absolute constant, so leading to a
    particular primitive.
    SINGULAR SOLUTIONS. 465

    “ Thus the particular primitive y = ect being given , the condition yo = 0


    gives et = 0 , wbence c is too if æ be negative, and - 00 if x be positive.
    It is a dependent constant. The resulting solution y = 0 does not then
    represent an envelope of the curves of particular primitives, nor strictly one
    of those curves. It represents a curve formed of branches from two of them .
    It is most fitly characterised as a particular primitive marked by a singularity
    in the mode of its derivation from the complete primitive.”
    [ Boole's “ Differential Equations,” Supplement, p . 13.

    DETERMINATION OF A SINGULAR SOLUTION FROM THE


    DIFFERENTIAL EQUATION.
    3075 RULE II. - Any relation is a singular solution which ,
    while it satisfies the differential equation , either involves y and
    makes p, infinite, or involves x and makes ( ) infinite .

    3076 “ One negative feature marks all the cases in which a solution
    involving y satisfies the condition P , = 0 . It is, that the solution , while
    expressed by a single equation , is not connected with the complete primitive
    by a single and absolutely constant value of c .
    “ The relation which makes p, infinite satisfies the differential equation
    only because it satisfies the condition yg = 0, and this implies a connexion
    between c and x , which is the ground of a real, though it may be unimportant,
    singularity in the solution itself.
    “ In the first, or, as it might be termed, the envelope species of singular
    solutions, c receives an infinite number of different values connected with the
    value of x by a law . In the second, it receives a finite number of values also
    connected with the values of x by a law . In the third species , it receives a
    finite number of values, determinate, but not connected with the values of x .”
    Hence the general inclusive definition
    3077 “ A singular solution of a differential equation of the
    first order is a solution the connexion of which with the com
    plete primitive does not consist in giving to c a single constant
    value absolutely independent of the value of x .”
    . [Boole's “ Differential Equations,” p . 163, and Supplement, p. 19.

    RULES FOR DISCRIMINATING A SINGULAR SOLUTION OF


    THE ENVELOPE SPECIES.

    3078 RULE III. — When P , or ( ) is made infinite by


    equating to zero a factor having a negative index, the solution
    “ may be considered to belong to the envelope species.”
    3079 “ In other cases, the solution is deducible from the
    - 3o
    466 DIFFERENTIAL EQUATIONS.

    complete primitive by regarding c as a constant of multiple


    value, - its particular values being either, 1st, dependent in
    SO
    some way on the value of x, or, 2ndly , independent of x, but
    still such as to render the property a singular one.”
    [Boole's “ Differential Equations,” p . 164.
    3080 RULE IV .- A solution which , while it makes p , infinite
    and satisfies the differential equation of the first order, does not
    satisfy all the higher differential equations obtained from it , is
    a singular solution of the envelope species.
    m -1
    Ex. : Y : = my * has the singular solution y = 0 when m is > 1.
    Now Yra = m (m - 1) ... (m - r + 1) y " ,
    and, when r is > m , the value y = 0 makes Yrq infinite. The solution is,
    therefore, by the rule of the envelope species .
    3081 RULE V . - " The proposed solution being represented by
    u = 0 , let the differential equation , transformed by making u
    and x the variables, be uz + f (x , u ) = 0 .
    rud
    Determine the in
    tegral Jo 94
    D
    as a function of x and u, in which U is either
    equal to f (x , u ) or to f (x , u ) deprived of any factor which
    neither vanishes nor becomes infinite when u = 0. If that
    integral tends to zero with u , the solution is singular " and of
    the envelope species. [ Boole, Supplement, p . 30.
    3082 Ex. — To determine whether y = 0 is a singular solution or par
    ticular integral of Yx = y (log y ) .
    po dy
    Here u = y, and Joy (log y) = log y
    As this tends to zero with y, the solution is singular.
    Verification . The complete primitive is y = ec-s, and no constant value
    assigned to c will produce the result y = 0 .

    3083 Professor De Morgan has shown that any relation


    involving both x and y , which satisfies the conditions 1 , = 0 ,
    Pa = 00 , will satisfy the differential equation when it does not
    make yır, as derived from it, infinite ; that it may satisfy it
    even if it makes yze infinite ; and that, if it does not satisfy the
    differential equation , the curve it represents is a locus of
    points of infinite curvature, usually cusps, in the curves of
    complete primitives. [Boole, Supplement,p . 35.
    FIRST ORDER LINEAR EQUATIONS. 467

    FIRST ORDER LINEAR EQUATIONS.

    3084 M +Ny = 0, or Mdx + Ndy = 0,


    M and N being either functions of x and y or constants .
    SOLUTION BY SEPARATION OF THE VARIABLES.
    3085 This method of solution , when practicable , is the
    simplest, and is frequently involved in other methods.
    Ex. xy ( 1 + x *) dy = ( 1 + y ?) dx,
    therefore ydy = _ de
    i + ja = * (1 + 22)
    and each member can be at once integrated .

    HOMOGENEOUS EQUATIONS.
    3086 Here M and N , in (3084 ) , are homogeneous functions
    of x and y , and the solution is affected as follows:
    RULE. — Put y = vx, and therefore dy = vdx + xdv, and
    then separate the variables. For an example , see (3108).
    EXACT DIFFERENTIAL EQUATIONS.
    3087 Mdx + Ndy = 0 is an exact differentialwhen
    M , = N
    and the solution is then obtained by the formula
    S Mdx + S { N - d , (SMdx )} dy = C .
    · PROOF. — If V = 0 be the primitive, we must have V , = M , V, = N ;
    therefore Vxy = M , = Nc. Also V = SMdx + $ (y ), (y ) being a constant
    with respect to x .
    Therefore N = V , = d ,S M dx + $' (y),
    therefore $ (y ) = S {N - d , SMdx} dy + C .

    3088 Ex. (x2 – 3x*y)dx + (99 –2 )dy = 0.


    Here M , = - 3x = Nc. Therefore the solution is
    C= y + {39– –0,(* –-x+y)} dy
    = -=”y+fvody = ** 4*–zy.
    468 DIFFERENTIAL EQUATIONS.

    3089 Observe that, if Mdx + Ndy can be separated into two


    parts, so that one of them is an exact differential, the other
    part must also be an exact differential in order that the whole
    may be such .
    3090 Also , if a function of x and y can be expressed as the
    product of two factors, one of which is a function of the
    integral of the other, the original function is an exact differ
    ential.

    3091 Ex.- Y
    cos Y dx = Y cos Y dy = cost.ydeady
    Y Y = 0.
    Here is the integral of the second factor. Hence the solution is
    sin = C.
    INTEGRATING FACTOR FOR Mdx + Ndy = 0 .
    When this equation is not an exact differential, a factor
    which will make it such can be found in the following cases.

    3092 I. — When one only of the functions Mx + Ny or


    Mx - Ny vanishes identically, the reciprocal of the other is an
    integrating factor.

    3093 II. - If , when Mx + Ny = 0 identically, the equation


    is at the same time homogeneous, then x - (n+ 1) is also an in
    tegrating factor.

    3094 III. - If neither Mx + Ny nor Mx – Ny vanishes identi


    cally, then ,when the equation ishomogeneous,w in, is an
    integrating factor ; and when the equation can be put in the
    form © (xy)xdy + x (xy)ydx = 0,MEN is an integrating
    factor.

    Proof. — I. and III. — From the identity


    M dx + Ndy = } {(Mx+ Ny)d log ay + (Mx– Ny)d logo},
    FIRST ORDER LINEAR EQUATIONS. 469

    assuming the integrating factor in each case, and deducing the required
    forms for M and N , employing (3090) .
    II.- Pat v = , M = x "® (v), N = x ** (u), and dy = ædv + vdx in
    Mdx + Ndy and Mx + Ny.
    3095 The general form for an integrating factor of
    Mdx + Ndy = 0 is
    p M -N do,
    M = è Nur - Mv,

    where v is some chosen function of x and y ; and the condition


    for the existence of an integrating factor under that hypothesis
    is that
    3096 My - Nu must be a function of v .
    No. – Mo,
    PROOF. — The condition for an exact differential of Mudx + Nudy = 0 is
    (Mu ), = ( Nu) . ( 3087) . Assume = ¢ (v ) , and differentiate out ; we thus
    obtain D'v = M , - N ,
    Nur - Mv,

    The following are cases of importance.


    3097 I. — If an integrating factor is required which is a
    function of x only , we put u = ° ( ), that is, v = x ; and the
    necessary condition becomes
    M

    N
    712 must be a function of x only.

    3098 II. - If the integrating factor is to be a function of xy,


    the condition becomes, by putting xy = v ,
    M , - N must be a function of xy only .
    Ny - Mx miles

    3099 III. — If the integrating factor is to be a function of


    , the condition is
    a®(N. - M.) mustbea function of me
    Mx + Ny
    If Mæ + Ny vanishes , (3092 ) must be resorted to.
    L
    470 DIFFERENTIA EQUATIONS.

    In this and similar cases, the expression found will be a


    function of vry if it takes the form F (v) when y is re re .

    placed by va .
    3100 IV . - Theorem . — The condition that the equation
    Mdx + Ndy = 0 may have a homogeneous function of a and y
    of the nth degree for an integrating factor , is
    * * ( N . - M ,) + nNx –
    Mx + Ny * = F (u), where u = 4
    3101 The integrating factor will then be obtained from
    x = x*eſplus du
    PROOF. — Pat u = v = x " 4 (u ) in (3097) , thus

    Perform the differentiations, and, by reduction, we get


    { (u ) = x* (N . - M ,) + n Ne
    V (u ) = Mæ + Ny
    The right member must be a function of u in order that (u ) may be found
    by integration .

    3102 Ex. - To ascertain whether an integrating factor, which is a homo


    geneous function of a and y , exists for the equation
    ( y' + axy”) dy - ay®dx + (x + y ) (x dy — y dx ) = 0.
    Here M = - (ay® + xy + y²), N = (y + axy* + xy + 2 ) .
    Substituting in the formula of (3100), we find that, by choosing n = - 3,
    the fraction reduces to ax*y * — Bxyº .
    oy , and, by putting y = ux, it becomes u
    a function of u .

    se ne du = - –3logus
    y

    H = a = ax2) = - 4- 7,
    the integrating factor required. It is homogeneous, and of the degree - 3
    in x and y , as is seen by expanding the second factor by (150 ).

    3103 If by means of the integrating factor u the equation


    uMdx + uNdy = 0 is found to have V = C for its complete
    primitive, then the form for all other integrating factors will
    be uf ( V ), where f is any arbitrary function .
    FIRST ORDER LINEAR EQUATIONS. 471

    PROOF. — The equation becomes


    uMf ( V ) dx + uNf ( V ) dy = 0 .
    Applying the test of integrability (3087), we have
    {u Mf ( V ) }, = {uNf ( V ) }z .
    Differentiate out, remembering that
    (u M ), = (uN ) , V = un , Ve = uM ,
    and the equality is established .
    3104 GENERAL RULE . — Ascertain by the determination of an
    integrating factor that an equation is solvable , and then seek to
    effect the solution in somemore direct way.

    SOME PARTICULAR EQUATIONS.


    3105 (ax + by + c) dx + (a'x + b'y + c') dy = 0 .
    This equation may be solved in three ways.
    1. – Substitute x = [ - a, y = n - B ,
    and determine a and ß so that the constant terms in the new
    equation in & and n may vanish .

    II. — Or substitute ax + by + c = ť , a'x + b 'y + c' = n .

    3106 But if a : a ' = b : b ', the methods I. and II. fail. The
    equation may then be written as a function of ax + by .
    Put z = ax + by , and substitute bdy = dz - a dx , and after
    wards separate the variables x and z.

    3107 III. - A third method consists in assuming


    (An + C ) d & + ( A'E + C')dn = 0 ,
    and equating coefficients with the original equation after sub
    stituting & = x + miy, r = x + mzy.
    mi, My are the roots of the quadratic
    amº + (6 + a ') m + b' = 0 .
    The solution then takes the form
    {(am ,– a')(x + m ,y) + cm , - c'}am -a = C.
    {(am , -a')(x + m ,y) + cm ,- c'}ama-a?
    AL
    RENTI EQUAT
    IONS .
    472 DIFFE

    3108 Ex. (3y - 7 + 7) dx + (74 - 3x + 3 ) dy = 0.


    Put x = & - a , y = n - B , thus
    (311 — 78) d & + (77 – 35) dn = 0 ........................ (i.),
    with equations for a and B , 7a - 3B + 7 = 0 ; 3a — 78 + 3 = 0 ;
    therefore a = - 1 , B = 0..... ......................... ... (ii.)
    (i.) being homogeneous, put n = vě, and therefore dn = vdx + &dv (3086) ;
    :: (709– 7)[d8+ (70 – 3)Edv = 0, or + 2 = 3 dv = 0.
    The second member is integrated, as in (2080 ) , with b = 0, and, after
    reduction, we find 5 log (n + 3) + 2 log (n ) = 0 .
    Putting & = 0 - 1 and n = y , by (ii.) the complete solution is
    (y + 2 – 1) (y — < + 1) ' = 0 .

    3109 When P and Q are functions of a only , the solution


    of the equation
    Pden

    dy +Py = 0
    dx is y = Ce Spade .........(i.)
    bymerely separating the variables. L
    G

    3110 Secondly , the solution of


    Pd .c Pd. c

    de+ Py = Q is y = e-frét{c + fQe8 Pax dx}.


    This result is obtained by the method of variation of
    parameters.
    RULE. — Assume equation (i.) to be the form of the solution ,
    considering the parameter C a function of x . Differentiate (i.)
    on this hypothesis, and put the value of y , so obtained in the
    proposed equation to determine C .
    Thus, differentiating (i.),we get yz = Cee-/ P4s — Py,
    therefore Q = Cce Space, therefore C = (QSP48dx + C".
    Then substitute this expression for C in equation (i.).
    Otherwise, writing the equation in the form ( Py - Q )dx + dy = 0 , the
    integrating factor el Ple may be found by (3097).
    3111 Ya + Py = Qy"
    is reduced to the last case by dividing by y" and substituting
    % = yl-".
    FIRST ORDER LINEAR EQUATIONS. 473

    * 3212 P dx + P ,dy + Q (x dy - y dx) = 0 .


    P1, P , being homogeneous functions of a and y of the pth
    degree , and Q homogeneous and of the oth degree, is solved
    by assuming
    Py = x + (? ), P, = x ( ), Q = xºx (% ).
    Put y = va , and change the variables to x and v. The result
    may be reduced to
    da X4 ( v ) _ x (v ) 29-p + 2
    do 7 ° (v) + 04 (v) (u) + (v )'
    which is identical in form with (3211), and may be solved
    accordingly.
    3213 (A , + BX + C1y ) (ady - y dx ) , ( A, + Byx + Cay ) dy
    - (Ag + Bg:c + Oxy ) dx = 0 .
    To solve this equation, put x = & ta, y = 9 + ß ,
    and determine a and ß so that the coefficients may become homogeneous,
    and the form of (3212 ) will be obtained .

    RICCATI'S EQUATION .
    3214 Ux + buº = cx " . .. .. ...... . .. ( A ) .
    Substitute y = ux, and this equation is reduced to the
    form of the following one, with n = m + 2 and a = l. It is
    solvable whenever m (2t + 1) = - 4t, t being 0 or a positive
    integer.
    3215 XYx - ay + byº = ca " ... ... ... (B ).
    1. — This equation is solvable , when n = 2a , by substituting
    y = vxº, dividing by 22a, and separating the variables. We
    dv
    thus obtain
    C - bv2
    = xa-1da.
    Integrating by (1937) or (1935), according as b and c in
    equation (B ) have the same or different signs, and eliminating
    n (B)have the same or differ
    v by y = vaca, we obtain the solution
    2.za J (be)
    Ce a + 1
    3216 2.1 " / bc) ...... ........ ( 1 ) ,
    Ce a - 1

    * The preceding articles of this section are wrongly numbered. Each number and
    reference to it, up to this point, should be increased by 100. The sheets were printed off
    before the error was discovered .
    3 P
    474 DIFFERENTIAL EQUATIONS.

    3217 or y = 1 (- 6)antan{c _**//= be)}. ...(2).


    3218 II. — Equation (B ) may also be solved whenever ever

    n - 2a
    2n
    = t a positive integer.
    RULE. - Write z for y in equation ( B ), and nt + a for a
    in the second term , and transpose b and c if t be odd .
    Thus, we shall have
    X2, - (nt + a ) = + 622 = ca" (when t is even ) ...... (3 ) ,
    Xez : - (nt + a ) z + cz = bæ" (when t is odd) ...... (4 ).
    Either of these equations can be solved as in case (I.),when
    n = 2(nt + a), that is, when " 520 = t. 2 having been de
    termined by such a solution, the complete primitive of (B )
    will be the continued fraction
    2
    21

    v = 6 + ta+242 +...+(0–1)nta + ..(S),


    where k stands for b or c according as t is odd or even.

    3219 III. — Equation (B ) can also be solved whenever


    Can N

    n + 2a
    2n = t a positive integer. The method and result will be
    the same as in Case II., if the sign of a be changed throughout
    and the first fraction omitted from the value of y . Thus
    Xin

    Proof. — Case II. - In equation ( B ), substitute y = A + 2 *, and equate


    the absolute term to zero. This gives A = or 0.
    Taking the first value, the transformed equation becomes
    dæne - (n + a)yn +cy; = bæ".
    Next, put y = " + 2 + 2 ", and so on. In this way the 4th transformed
    equation (3) or (4) is obtained with a written for the tth substituted variable ye
    FIRST ORDER NON -LINEAR EQUATIONS. 475

    Case III. — Taking the second value, A = 0 , the first transformed equa
    tion differs from the above only in the sign of a ; and consequently the same
    series of subsequent transformations arises, with - a in the place of a . The
    saccessive substitutions produce (5 ) and (6 ) in the respective cases for the
    values of y .

    3220 Ex. Ux+ u? = cx - *. (3214 )


    Putting u = du = XY ,
    and the equation is reduced to xyz - y + y* = cæš of the form (B ). Here
    a = 1, b = 1, n = ģ, and " 2n
    + 20 .= 2 , Case III. By the rule in (3218),
    changing the sign of a for Case III., equation (3 ) becomes
    62, - 3x + z = cæš.
    Solving as in Case I., we put z = vi}, & c. ; or, employing formula (1)
    directly,
    Cebvcx} + 1
    cukio

    z = 4cm - Td ; and then , by (6 ), y


    Ceo veas - 1 '
    +

    is the final solution .

    FIRST ORDER NON -LINEAR EQUATIONS.

    3221 Type
    n - 1

    ) + P (%)**+...+P.- +P.= . ..(1).


    where the coefficients P1, P2, ... P , may be functions of x
    23 mar ns

    and y .

    SOLUTION BY FACTORS.
    3222 If (1) can be resolved into n equations,
    Yx - P1 = 0, y: - P2 = 0, ... Yx - P » = 0 ...... (2),
    and if the complete primitives of these are
    Vi = C1, V, = C2, ... Vº = (n...............(3),
    then the complete primitive of the original equation will be
    ( V, – c)(V , - c) ... ( V . — c) = 0...............(4).
    476 DIFFERENTIAL EQUATIONS.

    Proof. - Taking n = 3 , assume the last equation. Differentiate and


    eliminate c . The result is
    (V , - V3) ( V; - V ) ’(V , - V,) dV,dv,dy, = 0 ............... (5 ).
    By (2 ), dv, = M (Ye - ) dx , & c., where My is an integrating factor. Sub
    stitute these values in (5 ), rejecting the factors which do not contain differ
    ential coefficients, and the result is
    ( x - 2 ) (2x - 2 ) ( x - 2 ) = 0,
    which is the differential equation ( 1) .

    3223 Ex. — Given y + 3y = + 2 = 0 .


    The component equations are yx + 1 = 0 and yz + 2 = 0 , giving for the
    complete primitive
    (y + x - c) (y + 2x – c) = 0 .

    SOLUTION WITHOUT RESOLVING INTO FACTORS.


    3224 Class I. — Type $ ( x , p ) = 0 .
    When x only is involved with p , and it is easier to solve
    the equation for x than for p , proceed as follows.
    RULE. — Obtain x = f (p ). Differentiate and eliminate dx
    by means of dy = pdx. Integrate and eliminate p by means
    of the original equation .
    Similarly, when y = f(p ), eliminate dy, & c.
    3225 Ex. - Given x = ayx + był, i.e., a = ap + bp ..................... ( 1 ),
    dx = adp + 26 pdp, therefore dy = pdx = apdp + 2bp*dp,
    therefore y = " + 2% *+ C.
    Eliminating p between this equation and ( 1), the result isthe complete primitive
    ( ax + 6by – bc) = (bay - 4x® – ac) (a ’ + 4bx ).

    3226 CLASS II. — Type


    x®(p )+ yų (p) = x (p).
    RULE. — Differentiate and eliminate y if necessary. Inte
    grate and eliminate p by means of the original equation .
    If the equation be first divided by (p ), the form is
    simplified into
    3227 y = x ^ (p ) + x (p ).
    Differentiate, and an equation is obtained of the form
    X , + Pir = Q , where P and Q are functions of p .
    This may be solved by (3210) , and p afterwards eliminated .
    FIRST ORDER NON -LINEAR EQUATIONS. 477

    3228 Otherwise , a differential equation may be formed


    between y and p , instead of between x and p .
    3229 Or, more generally , a differential equation may be
    formed between a or y and t, any proposed function of p , after
    which t must be eliminated to obtain the complete primitive.

    3230 Clairaut's equation, which belongs to this class , is of


    the form y = px + f (p ).
    Rule. — Differentiate, and two equations are obtained
    (1)...... Pa = 0, and :: p = c; (2)......2 +f'(p ) = 0.
    Eliminate p from the original equation by means of (1), and
    again by means of (2 ). The first elimination gives y = cx + f (c),
    the complete primitive. The second gives a singular solution .
    PROOF. — For, if Rule I. (3169) be applied to the primitive y = cx + f (c),
    we have æ + f' (c ) = 0 ; and to eliminate c between these equations is the
    elimination directed above, c being merely written for p in the two equations.

    3231 Ex. 1. y = px + xv1 + p ?.


    This is of the form y = xº ( p), and therefore falls under (3227). Differ
    ai ep
    entiating, we obtain a dp + dx V1 + p ' t dp_ = 0.
    ✓ (1 + p )
    since dy = pdx ; thus
    ( cd+es + ifjs)dp+ et = 0,
    in which the variables are separated.
    Integrating by (1928), and eliminating p, we find for the complete
    primitive x + y * = C .c.

    3232 Ex. 2. y = px + V b? – aʼpº.


    This is Clairant's form ( 3230 ). Differentiating, we have
    die {a -vera heim )} = 0.
    The complete primitive is y = cx + ✓ (62 – aʼc*) ;
    and the elimination of p by the other equation gives for the singular solution
    a ’ye — 6% 2 = a *b?, an hyperbola and the envelope of the lines obtained by
    varying c in the complete primitive, which is the equation of a tangent.

    3233 Ex. 3. To find a curve having the tangent intercepted between


    the coordinate axes of constant length .
    478 DIFFERENTIAL EQUATIONS.

    The differential equation which expresses this property is


    yV1 + p * -
    1 + p' = a ,
    P
    ap
    | $ = pe+* V1+ 2 ** ... ... ....... ... ... .. . ..

    a
    Differentiating gives = . ... .. . . ..
    dæ " (1 + p )} } =
    Eliminating p between ( 1) and ( 2) gives,
    1st, the primitive y = cx + ✓1
    + ca
    2nd, the singular solution 25 + y3 = aš ...... . ..... ....... ( 4 ) .
    (3) is the equation of a straight line; (4 ) is the envelope of the lines
    obtained by varying the parameter c in equation ( 3).

    3234 Class III. — Homogeneous in x and y .


    Type *** ( _.. ) = 0.
    RULE. — Put y = vx , and divide by x ” . Solve for p , and
    eliminate p by differentiating y = vx ; or solve for v, and
    eliminate v by putting v = ; and in either case separate the
    variables.

    3235 Ex. y = px + xV1 + p .


    Substitute y = væ , and therefore p = v + xvz. This gives v = p + v1 + p .
    Eliminate p between the last two equations, and then separate the variables.
    dx , 2vdv = 0 ,
    The result is
    a Tito
    from which æ (v? + 1) = C or 2 + y = Ca .
    The same equation is solved in (3131) in another way.

    SOLUTION BY DIFFERENTIATION .
    3236 To solve an equation of the form
    F {$ (x, y,yx), y ( x, y, yu)} = 0 .
    RULE. — Equate the functions p and respectively to arbi
    trary constants a and b . Differentiate each equation , and
    eliminate the constants. If the results agree, there is a common
    HIGHER ORDER LINEAR EQUATIONS. 479

    primitive (3166 ),which may be found by eliminating y , between


    the equations d = a , p = b , and subsequently eliminating one
    of the constants by means of the relation F (a , b ) = 0 .

    Ex. 2 - yy: + f (y*— y*y;) = 0.


    Here the two equations a - yy: = a, f (yº - y’ yx) = b,
    on applying the test, are found to have a common primitive. Therefore,
    eliminating Yx, we obtain
    f { y* — (ve — a ) } = b .
    Also, by the given equation , a + b = 0.
    Hence the solution is f { y: - (x + b)?} = b .

    HIGHER ORDER LINEAR EQUATIONS .

    3237 Type 1 + P 4 +...+ P(n-1) A4 + Pm3 = Q ,


    where P,... P , and Q are either functions of x or constants.
    LEMMA . - If y 42, ... Y , be n different values of y in terms
    of x , which satisfy (3237), when Q = 0 , the solution in that
    case will be y = Cıyı + C2y2 + . .. + CnYn.
    PROOF. — Substitute 41, 42, ... Yn in turn in the given equation. Multiply
    the resulting equations by arbitrary constants , C1, C2, ... On respectively ;
    add, and equate coefficients of P1, P2, ... Pn with those in the original
    equation .

    LINEAR EQUATIONS WITH CONSTANT COEFFICIENTS.


    3238 ynx + a,y(n-1)x + ...+ ain-1)9x +any = Q ......... (1).
    3239 Case 1. — When Q = 0 .
    The roots of the auxiliary equation
    m " + am " - + ... + an -ım + an = 0 ... ............ (2 )
    being mon, mg, ... Mn, the complete primitive of the differential
    equation will be
    3240 y = C em * + Cem® + ... + Cyema............ (3).
    If the auxiliary equation (2) has a pair of imaginary roots
    480 DIFFERENTIAL EQUATIONS.

    (a + ib), there will be in the value of y the corresponding


    terms
    3241 Aear cosbx + Bear sin bx .................. (4 ).
    If any real root m ' of equation ( 2 ) is repeated r times, the
    corresponding part of the value of y will be
    3242 (A , + Axx + A2x” + ... + Ar_1 r=1) em'x.
    And if a pair of imaginary roots occurs 9 times, substitute
    for A and B in ( 3241) similar polynomials of the r – 1th degree
    in a .
    PROOF. - ( i.) Substituting y = Cemx in ( 1) as a particular solution , and
    dividing by Cema, the auxiliary equation is produced, the roots of which
    furnish n particular solutions, y = , emix, y = C , emt, & c., and therefore, by
    the preceding lemma, the general solution will be equation ( 2 ) .
    ( ii.) The imaginary roots a Eib give rise to the terms Сeax +ibx + C 'ear- ibs,
    which , by the Exp. values (766 ), reduce to
    ( C + C'') eat cos bx + i ( C - 0') eux sin bx.
    (iii.) If there are two equal roots m , = mi, put at first m ? = mith . The
    two terms Ce* * + Coelm ,+h )x become em # ( C , + Coehr). Expand ear by (150 ) ,
    and put 0 , + C , = A , Cgh = B in the limit when h = 0 , C , = 00 , C = - 0 .
    By repeating this process, in the case of r equal roots, we arrive at the form
    ( A , + Azx + A.9x2 + ... + A . -12*-!) emit ;
    and similarly in the case of repeated pairs of imaginary roots.

    3243 Case II. - When Q in (3238 ) is a function of x .


    First method. - By variation of parameters.
    Putting Q = 0 , as in Case I., let the complete primitive be
    y = Aa + BB + Cy + & c . to n terms...... ...... (6 ) ,
    a , B , y being functions of a of the form emx. The values of
    the parameters A , B , C , . .. , when Q has its proper value
    assigned , are determined by the n equations
    3244 Ara + B+B + to n terms = 0,
    Agar + BxBx + = 0,
    Ajazz + B B2x + y = 0,
    ... .. .

    A.,^(n-1)x + B_ B(n-1)x + „ = Q,
    A ,, Bx, & c. being found from these equations, their integrals
    must be substituted in (6 ) to form the complete primitive.
    HIGHER ORDER LINEAR EQUATIONS. 481

    Proof. — Differentiate (6) on the hypothesis that A , B , 0 , & c. are func


    tions of a ; thus
    9x = ( Aar + BB . + ...) + (42a + BxB + ...).
    Now , in addition to equation (1), n - 1 relations may be assumed between
    the n arbitrary parameters. Equate then the last term in brackets to zero,
    and differentiate y , in all, n - 1 times, equating to zero the second part of
    each differentiation ; thus we obtain
    Yo = Aug + BB. + & c. and Aga + B2B + & c. = 0 ,
    928 = Aaz: + BB28 + & c. and Arax + B B. + & c. = 0 ,
    ... .. . .. . . .. . .. . .. .. . .. .
    Yin-1)= = Aa(n-1)x + BB1n-1) + & c. and (n=2)x + B. Pin 2)x + &c. = 0.
    The n quantities A ., Bx, & c . are now determined by the n - 1 equations on
    the right and equation (1) . For, differentiating the value of y (n-1)xy we have
    Yux = {Aanz + BBnæ + & c.} + { A ,A (n-1)x + BrBin - 1)2 + & c.},
    and if these values of Yxı Yara ... Ynx be substituted in (1 ), it reduces to
    AxA(n-1)x + B . Bın–1)x + & c. = Q ,
    for the other part vanishes by the hypothetical equation
    Ynx + a,y(n-1)x + ... tany = 0,
    since the values of yx, ... Yin -1) ą, and the first part of ynx, are the true values
    in this equation.

    3245 Case II. - Second Method.-- Differentiate and eliminate


    Q . The resulting equation can be solved as in Case I. Being
    of a higher order, there will be additional constants which
    may be eliminated by substituting the result in the given
    equation .
    3246 Ex. - Given 924 - 742 + 12y = x .... .. ... . ( 1 ) .
    1st Method . - Putting x = 0 , the auxiliary equation is mi- 7m + 12 = 0 ;
    therefore m = 3 and 4. Hence the complete primitive ofyże - 7yx + 12y = 0
    y = Ae*t + Bett. . ........ ... .. ... ( 2 ) ,

    The corrected values of A and B for the primitive of equation ( 1) are found
    from

    Açe ++ 4B,6*
    31,6X Boek == xfi
    07, ::
    " 4 . = - xe- e and A = 30,71e- +a.
    4 + 1
    B = xeb and B = - 3 Te-b
    16
    + b.

    Substitating these values of A and B in (2), we find for its complete primi.
    tive y = aeľ + bet + 12x144+ ?
    3247 2nd Method. 72 – 79. + 12y = ☆ ................. .... (1).
    . . . . . . . . . .

    3o
    482 DIFFERENTIAL EQUATIONS.

    Differentiating to eliminate the term on the right, we get


    943 - 743x + 12y2 = 0 .
    The aux. equation is m * — 7m3 + 12m = 0 ; therefore m = 4 , 3, 0 , 0 .
    Therefore y = Aete + Be3 + Cx + D ............ .. .. . . .... . . ( 2 ) ;
    Yz = 4Ae** + 3 Be3 + C ; Yze = 16 Ae** + 9Bewe.
    Substitute these values in (1); thus C = ; D = ;
    therefore , substituting in ( 2 ), y = Aellt ū as before.

    3248 When a particular integral of the linear equation


    (3238 ) is known in the form y = f ( x ) , the complete primitive
    may be obtained by adding to y that value which it would
    take if Q were zero.
    Thus, in Ex. (3247),
    022 9 12 * 144 iss aa particular
    particula integral of ( 1) ; and
    the complementary part Ae x + Best is the value of y when the dexter is zero.

    3249 The order of the linear equation (3238) may always


    be depressed by unity if a particular integral of the same
    equation , when Q = 0 , be known.
    Thus, if Ya + P 1922 + PgY : + Pxy = Q ...........................( 1),
    and if y = % be a particular solution when Q = 0 ; let y = vz be the solution
    of ( 1). Therefore, substituting in (1 ),
    (* + P472x + Pg7x + Pqm ) v + & c. = Q ,
    the unwritten terms containing VF, Vzx, and var
    The coefficient of v vanishes, by hypothesis ; therefore, if we put vz = l ,
    we have an equation of the second order for determining u . u being found,
    v = ſudx + C .

    3250 The linear equation


    (a + bx)" Yux + A (a + bx)" -"Y(n-1)& + B (a + ba")" -"y(n -9z + ...
    ... + Ly = Q ,
    where A , B , ... L are constants, and Q is a function of x , is
    solved by substituting a tbx = e , changing the variable by
    formula (1770) , and in the complete primitive putting
    t = log (a + bx).
    Otherwise, reduce to the form in (3446 ) by putting
    a + b = 1 , and solve as in that article.
    HIGHER ORDER NON -LINEAR EQUATIONS. 483

    HIGHER ORDER NON -LINEAR EQUATIONS.

    3251 Type F(x,y, y con ...com ) = 0.


    SPECIAL FORMS.
    3252 F (x, Yræ Y(r+1) ...Yna) = 0.
    When the dependent variable y is absent, and Yrp is the
    derivative of lowest order present, the equation may be de
    pressed to the order n - qo by putting Yrr = % . If the equation
    in z can be solved , the complete primitive will then be
    in z ca

    y = J + S 0 (2149)
    3253 F (y , Yræ Y(r +1)x ... ynx ) = 0 .
    If be absent instead of y, change the independent vari
    able from a to y , and proceed as before .
    Otherwise, change the independent variable to y , and
    make p ( = yx) the dependent variable .

    For example, let the equation be of the form


    3254 F (y, Yu, Yaz, Yxv) = 0 ........ .. (1).
    (i.) This may be changed into the form
    F (y, Xy, Way, X3y) = 0 by (1761, '63, and '66 ) ;
    and the order may then be depressed to the 2nd by (3252). The solution
    will thus give x in terms of y .
    3255 (ii.) Otherwise, equation (1) may be changed at once into one of
    the form F ( y, P , Pya Pay) = 0 , by (1764 and ’67),
    the order being here depressed from the 3rd to the 2nd. If the solution of
    this equation be p = ° (y, G , ca), then, since dy = pdx, we get, for the com
    plete primitive of (1 ), r = 1 dy .tes.
    10 (y, C1, c.
    C)

    3256 Ynx = F (x ).
    Integrate n times successively , thus
    y = [ F(x) +01 –1 + 0229-2 + ...+ cm.
    n .
    484 DIFFERENTIAL EQUATIONS.

    3257 Yzv = F (y ).
    Multiply by 2y, and integrate, thus
    dy
    (dy ) = 2 [F( )dy +cı, x = Sv Et .
    { 2 ( F ( y)dy + }
    3258 Ynx = F {Y (n -1) } ,
    an equation between two successive derivatives .
    Put Yin -1)= = % , then zz = F ( ), from which

    * = SFO)to..................(1).
    If, after integrating, this equation can be solved for z so
    that z = ° ( , c), we have y(n - 1)2 = ° (x , c), which falls under
    (3256).
    3259 But if cannot be expressed in terms ofæ , proceed as
    follows :

    960-) = z; Y» » = [zda = [f ";


    %(n=)== Shay Fe * -- --
    Finally, y =fFoFE - SH "
    the number of integrations and arbitrary constants introduced
    being n - 1.
    3260 Ynx = F {Y(n-1)x}.
    Put Yin - 2) 2 = % ; then 22. = F (x ), which is (3257) , the
    solution giving æ in terms of z and two constants . If z can be
    found from this in terms of x and the two constants, we get
    % or Yin - 2) 2 = ° ( , C1,C2),
    which may be solved by (3256 ).
    3261 But if z cannot be expressed in terms of x , proceed as
    in (3259).
    DEPRESSION OF ORDER BY UNITY.
    3262 When F (x , y , yx, Yzx , ...) = 0
    is rendered homogeneous by considering
    X , Y, Yæ , Yıcı Ysx , & c.
    HIGHER ORDER NON -LINEAR EQUATIONS. 485

    to be of the respective dimensions 1, 1, 0 , - 1, - 2, & c. ; put


    x = e', y = ze', & c.
    The transformed equation will contain the same power of e in
    every term , and will reduce to the form
    F (z , Ze,Z209 ...) = 0 ,
    the order of which is depressed by unity by putting % o = u .
    3263 When the original equation is of the 2nd order , the
    transformed equation in u and z may be obtained at once by
    changing x , y , yx, Yux into 1, %, u + % , u + uuz, respectively .
    The solution is then completed , as in example ( 3264).
    Proof. — We have a = e°; y = ze” ;
    Yo = % + % ; 92: = e -º (22 + zo) ; Y3x = e -> (239 — % ) ; and so on .
    The dimensions of x , y, yx, & c . with respect to e , are 1, 1, 0, - 1, - 2, & c.
    Therefore the same power of e' will occur in every term of the homogeneous
    equation .

    3264 Ex. : * * Y2 = (y - xyz ) .


    Making the above substitutions for x, y, Yx and Y2x, the equation becomes
    220 + % = - .
    Pat 70 = u ; thus
    28 + u = - Up = - uug, therefore 2° + 1 = - 12, u + = - dz ,

    therefore tan -?u = a - % (1935), therefore 2 = u = tan (a - x),


    therefore do = cot (a - x ) dz, therefore 0 = - log b sin (a - x ) (1941).
    But
    O = log and =
    therefore bx = cosec (a – ), or bx = sec (c+ %),
    by altering the arbitrary constant.

    3265 When F (x, y, Ya, Yzoo.. ) = 0


    is made homogeneous by considering x , y , ym, Yor, & c . to be of
    the respective dimensions 1, n , n - 1 , n - 2 , & c . ; put
    x = e , y = zeno,
    and depress the order by putting zo = U , as in (3262).

    3266 When the original equation is of the 2nd order, the


    486 DIFFERENTIAL EQUATIONS .

    final equation between u and z may be obtained at once by


    changing
    X , Y, Yx,y2 into 1, m , u + nz,uu , + (2n - 1) u + n (n - 1) % ,
    respectively .
    3267 Ex.: Y2xyz = xy ...........
    ....... ............. (1)
    With the view of applying (3265 ), the assumed dimensions of each
    member of this equation , being equated, give
    n – 2 + 1 - 1 = 1 +n, therefore n = 4 .
    Thus x = eº ; y = ze“ ; Yx = e) (20 + 4x) ; Y2x = e2 (228 + 720 + 12 ) .
    Substituting in (1), e disappears ; and by putting zo = U , Z2, = uuz, the equa
    tion is reduced to
    (u ? + 4uz) du + (7u? + 40uz + 4872 — 2 ) dz = 0 ,
    which is linear and of the 1st order. This equation is also obtained at once
    by the rule in (3266 ).

    3268 When F (x, y, Yx, Yır, ...) = 0


    is homogeneous with respect to y , yx, Y2x, & c., put y = e " ,
    and remove e as before by division. The equation between u
    and x will have its order less by unity than the order of F .

    3269 Ex.: Y2 + Pya + Qy = 0 .......... ... (1),


    P and Q being functions of x .
    Here y = ed" ; Yo = uy ; Y2x = (u : + uP) y.
    Substituting , the equation becomes uz tu ' + Pu + Q = 0 , an equation of the
    1st order. If the solution gives u = Q (x , c), then 5° (« ,c) dx = logy is
    the complete primitive of ( 1) .

    EXACT DIFFERENTIAL EQUATIONS.

    3270 Let dU = $(x, y, Ya, Yzve... Yna)dx = 0


    be an exact differential equation of the nth order . The highest
    derivative involved will be of the 1st degree.
    3271 RULE FOR THE SOLUTION (Sarrus). — Integrate the term
    involving ynx with respect to y (n -1x only , and call the result U .
    Find dui, considering both x and y as variables. dU - DU ,
    will be an exact differential of the n - 1th order.
    MISCELLANEOUS METHODS. 487

    Integrate this with respect to y(n -1), only, calling the result
    U2, and so on.
    The first integral of the proposed equation will be
    U = U , + U2 + ... + Un = C .
    3272 Ex.: Let dU = {y ? + (2xy - 1) yö + xy2x + x *yox } dx = 0 .
    Here U = x ’yzz, dU , = (2xy2x + a *yor ) dx ;
    du - dU, = {y® + (2xy - 1) Y: - xyzx } dx = 0 ;
    . Un = - xyz dU , = - (4x + xyzx) dx, du - d0, - dU, = (y² + 2xyyr) da ;
    therefore Ug = xy , and U = x*yze — Xyz + xy = 0
    is the first integral.
    3273 Denoting equation (3270) by dU = Vdx, the series of
    steps in Rule (3171) involve and amount to the single condi
    tion that the equation
    N - Pc + Q2x – R3x + & c. = 0 ,
    with the notation in (3028) , shall be identically true. This
    then is the condition that V shall be an exact 1st differential.
    3274 Similarly, the condition that V shall be an exact 2nd
    differential is
    P - 2Q . + 3R2: - 45%: + & c. = 0.
    3275 The condition that V shall be an exact 3rd differential
    is Q - 3R , + 3.482, - 34. + &c. = 0,
    and so on . [Euler, Comm . Petrop., Vol. viii.

    MISCELLANEOUS METHODS.

    3276 Yzx + Pyx + Qy = 0 . ................... (1)


    where P and Q are functions of x only.
    The solution is y = [e-5Pdv (25 Qe=?[Pdeda)-"dx.
    Proof. — Put eJPdx = 2, and multiply (1) by z ; then, since 2, = P2,
    zy. + Qzy} = 0. Put zyz = u - , . uur = Qz- , . u = V (2SQz- dx), & c.
    3277 Y2x + QyX + R = 0 ........................(1),
    where Q and R are functions of y only.
    488 DIFFERENTIAL EQUATIONS.

    The solution is x = felad (2 6 Reledydy)-+dy.


    Proof. — Put Jady = 2, and multiply (1) by z.
    . . (zyz). = Rz, ii zy : (zy.). = Rzłyą : : (zyn)' = 2 ( Rz’dy, & c.

    3278 Y2x + Pyx + Qy% = 0 ,


    where P , Q involve x only .
    Put yo = % , and the form (3211) is arrived at.
    3279 Y2x + Py: + Qy' = 0 .
    This reduces to the last case by changing the variable from æ
    to y by (1763).
    3280 For a few cases in which the equation
    yx + Py: + Qy + R = 0
    can be integrated , see De Morgan's “ Differential and Integral
    Calculus,” p. 690 .
    3281 Y2x = ax + by .
    Put ax + by = t (1762–3 ). Result t2 = bt. Solve by (3239)
    or (3257).
    3282 ( 1 - x2) Y2. — xyx + q*y = 0 .
    Put sin -1x = t, and obtain Yz + q*y = 0 .
    Solution , y = A cos (q sin - 2x ) + B sin (q sin - » ).
    3283 (1 + ax?) Y2x + axy. + q y = 0.
    Pu t ſ da
    Put zoma = t, and obtain Yu + q*y = 0 as above.
    JV ( 1 + ax )

    3284 Liouville's equation , Yzx + f (x )yz + F (y) y = 0 .


    Suppress the last term . Obtain a first integral by (3209), and
    vary the parameter. The complete primitive is
    ) dic

    les plusdedy = A fe-Sola)dedx + B.


    3285 Jacobi's theorem . — If one of the first integrals of the
    equation Y2x = f (x , y) is Y: = $ (x , y , c).........(i., ii.),
    APPROXIMÁTE SOLUTION . 489

    the complete primitive will be


    18.(dy - pdx) = c.
    PROOF. — Differentiating ( ii.), we obtain 0 + 00, = f ( , y), and differen
    tiating this for c, 4x6 + 0 .9, + $ yo = 0 . But, by (3187) , this is the condition
    for ensuring that gedy - Pe9dx = 0 shall be an exact differential ; therefore po
    is an integrating factor for equation (ü .), Y : - (x , y, c) = 0 .

    Equations involving the arc s , having given


    3286 ds = dæ? + dya or Sq = v1+ ye.
    3287 s = ax + by.
    Here vi + ye = a + byx. Find y, from the quadratic equation .
    3288 X 95 = a .
    Changefrom 8 to æ by (1763) ; :: – 8 *82 = a , ..8°= 2ax + c,
    or 1+ y:= 2abte y = [vaate - 1)dw +e'.
    APPROXIMATE SOLUTION OF DIFFERENTIAL EQUATIONS
    BY TAYLOR ' S THEOREM .
    3289 The following example will illustrate the method :
    Given Y2x = xyz + y, : : 43v = (2 + 2) Yx + xy.
    Generally, let ynx = Anyx + Bny ; y(n + 1). = An + 19x + Bu +1y .
    But,by differentiation,
    Y(n +1)2 = ( ,æ + 4n + Bn) yx + (4 , + Bh) y ,
    i . Anti = Anæ + A + Brand Bn+1 = An + Bn.
    But A , = x , B , = 1, . Ag = x* + 2, By= x ;
    A4 = 208 + 5x, B = x2 + 3 , & c .
    Now , when x = a , let y = b and y. = p ; then , by Taylor's
    theorem ( 1500 ) ,

    y = a +P(23– a) +(49p + B%b)( )* +(4.p + B b)(mma)


    + & c.,
    which converges when æ — a is small. [De Morgan,p.692.
    3 R
    490 DIFFERENTIAL EQUATIONS.

    SINGULAR SOLUTIONS OF HIGHER ORDER


    EQUATIONS.

    DERIVATION FROM THE COMPLETE PRIMITIVE.


    3301 Let Ynx = 4 (x , y, Yx, Y2x ... Yin -1) +) ............... (1)
    be the differential equation, and let its complete primitive be
    y = f (x ,a ,b, c ... 8 ).....................(2),
    containing n arbitrary constants .
    3302 Rule. — To find the general singular solution of (1),
    eliminate abc ... s between the equations
    y = f, yı = fq, yax = fax ..., Y(n-1)= = f(n-1)s......(3)
    and fa fax fu2x ... fa(n-1)
    fi fox fox ... fo(n-1)3 = 0 ............ (4).
    f, for fax ...fo(n=1)3
    The result is a differential equation of the n - 1th order , and
    the integral of it, containing n - 1 arbitrary constants , is the
    singular solution ,
    Proov. - Differentiate ( 2 ), considering the parameters a , b ... s variable ,
    thus yo = fz + fqaz + ... + 8,8 . Therefore, as in (3171) ,
    Yo = fx if fa Qz + fo bz + ... fo 8, = 0,
    Y2x = f2 if faxQz + foxbx + ... frase = 0 , as well;
    and so on up to ynx = frx Eliminating az, bos ... 8, between the n equations
    on the right, the determinant equation (4 ) is produced with the rows and
    columns interchanged .

    3303 Ex. : yxyz+ 1x*y2 - y: - ( : — XY2r)* = 0 ............... (1 ) .


    The complete primitive is y = 9** +bx +a'+ bº..........................(2).
    From which ........................
    Yo = ax + b ......... .
    and the determinant equation is
    | 3x2 + 2a, x = 0 or + 2bx = 2a .......
    2C + 2b , il
    SINGULAR SOLUTIONS OF HIGHER ORDER EQUATIONS. 491

    Eliminating a and b from (2 ), ( 3 ), and (4 ), we get the differential equation


    4dy + ( 2x + *) dx = ✓ (1 + x ) dx ................ (5 ),
    ✓ (16y + 4x* + 24)
    the integral of which , and the singular solation of (1), is
    ✓ ( 16y + 4.c + x*) = x / ( 1 + xa) + log {æ + V (1 + x<)} + C .
    [ Boole, Sup., p . 49.

    3304 Either of the two ' first integrals ' (3064 ) of a second
    order differential equation leads to the same singular solution
    of that equation .
    3305 The complete primitive of a singular first integral of
    a differential equation of the second order is itself a singular
    solution of that equation ; but a singular solution of a singular
    first integral is not generally a solution of the original equa
    tion .
    Thus the singular first integral (5 ) of equation ( 1 ) in the
    last example has the singular solution 16y + 4a2 + 2 * = 0 , which
    is not a solution of equation ( 1) .

    DERIVATION OF THE SINGULAR SOLUTION FROM THE


    DIFFERENTIAL EQUATION .
    3306 RULE. — Assuming the same form (3173), a singular
    solution of the first order of a differential equation of the nth
    order will ma
    make d (ynx) , infinite ; a singular solution of the
    ke d (y(n-1)=)
    second order will mak 7 d (ynx) d (ynx) Ynx), both infinite ;
    d (y(n-1)=)? d (y(n-2) ) "
    and so on . [ Boole, Sup., p. 51.

    3307 Ex. — Taking the differential equation (3303) again ,


    y — æyx + ja ’yz: — yže - (Y : — XY22)' = 0.................. ( 1 ).

    and differentiating for yz and Y2x only,


    { 1x * + 2x (Y . — XY22) – 2yze} d (722) — {æ + 2 (9: - Xyze) } d ( ) = 0 .
    The condition (Y2x ) = 0 requires
    d (4 )
    ba ? + 2x (Y : - @ y2r) — 2422 = 0 .
    Substituting the value of Y2x obtained from this in equation ( 1 ) , and rejecting
    the factor (x + 1 ), the same singular integral as before is produced (3303,
    equation 5 ).
    492 DIFFERENTIAL EQUATIONS.

    EQUATIONS WITH MORE THAN TWO VARIABLES.

    3320 Pdx + Qdy + Rdz = 0 ................ ( 1).


    P , Q , R being here functions of x , y, %, the condition that this
    equation may be an exact differential of a single complete
    primitive is
    3321 P (Q: - R ,) + Q (R . - P .) + R (P, - Qx) = 0.
    PROOF. — Let u be an integrating factor of Pdx + Qdy + Rdz = 0 . Then
    - uPdx = uQdy + pRdz, and , by (3187) , for an exact differential, wemust
    have (uQ) , = ( u R ) , Write this symmetrically for P , Q , and R , differentiate
    out, and add the three equations after multiplying them respectively by P ,
    Q , and R .
    To find the single complete primitive of equation (1).
    3322 RULE. Consider one of the variables z constant, and
    therefore dz = 0. Integrate, and add $ (z ) for the constant of
    integration . Then differentiate for x , y, and z , and compare
    with the given equation ( 1 ). If a primitive exists, $ (2 ) will be
    determined in terms of z only by means of preceding equations.
    The complete primitive so obtained is the equation of a
    system of surfaces, all of the same species, varying in position
    according to the value assigned to the arbitrary constant.
    3323 Ex. : (x - 3y - x ) dx + (2y — 3x) dy + (2 a ) dz = 0 ...............(
    Condition (3321) is satisfied ; therefore, putting dz = 0 , we have
    (x - 3y - x ) dx + (2y — 3.x ) dy = 0 .
    Applying (3187), M , = - 3 = Nx, and integration gives
    fx? — 3xy - zx + y + ° (z) = 0 .
    Differentiating now for x , y, and z,
    (x - 3y - x) dx + (2y – 3x)dy + {$ (z ) — « }dz = 0.
    Equating coefficients with (1), $' ( ) = , therefore \ (x ) = {z? + C .
    Hence the single complete primitive is
    god + 2y ' + ? – 6xy – 2z2 = C ,
    the equation of a system of surfaces obtained by varying the constant C .

    3324 When the equation Pdx + Qdy + R dz = 0 is homo


    geneous, put x = uz, y = vz . The result , when the coefficient
    of dz vanishes, is of the form
    3325 Mdu + Ndv = 0 ,
    EQUATIONS WITH MORE THAN TWO VARIABLES. 493

    solvable by (3184). Otherwise it is of the form


    3326 dz = M du + Ndv,
    and the right will be an exact differential if a complete primi.
    tive exists.

    3327 Ex.: yz dx + zæ dy + xy dz = 0 ... ... ... .(1) .


    Condition (3321) is satisfied. Putting
    x = uz, y = vz , dx = udz + zdu, dy = vd2 + zdv,
    ( 1) becomes dz _ du dv _
    T + 3u + 30 = 0 ,
    and the solution is log (zułot) = C or xyz = 0 .
    When the equation
    Pdæ + Qdy + Rdz = 0 .....................(1)
    has no single primitive :
    3328 Rule. — Assume • (a , y, z ) = 0 .....................
    and differentiate ; thus
    Pedx + ® ,dy + p,dz = 0 .................. (3).
    The form of ¢ being given , eliminate z and dz from ( 1) by
    (2 ) and (3 ). The result, being of the form
    Mdx + N dy = 0 ,
    can be integrated , and the solution taken with ( 2 ) constitutes a
    solution of equation (1 ) , and represents a system of lines (by
    varying the constant of integration ) drawn on the surface
    $ (x, y , z ) = 0.

    3329 Ex. : ( 1 + 2m ) xdx + (1 - x ) y dy + zdz = 0 .


    The condition (3321) not being satisfied, assume + y + z = god as the
    function © , therefore ædx + ydy + zdz = 0 ; and by eliminating % and dz,
    2mdx - y dy = 0 , the integration of which gives y* — 4mx = C , a cylindrical
    surface intersecting the spherical surface in a system of curves (by varying
    C ), whose projections on the plane of æy are parabolas.

    The condition that


    3330 Xdx + Y dy + Zdz + Tdt = 0 ,
    where X , Y , Z , T are functions of x , y, % , t, may be an exact
    494 DIFFERENTIAL EQUATIONS.

    differential,may be shewn,in amanner similar tothatof(3321) ,


    to be expressed by any three of the equations
    3331 Y (Z - T,) + Z (T , - Y ) + T ( Y , - 2 , ) = 0,
    Z ( T. - X ) + T ( X , – 22) + X (26 – T; ) = 0 ,
    T ( X , - Y,) + X ( Y - T,) + Y ( T - X ,) = 0,
    X ( Y , - Z ,) + Y (2 , - X ,) + Z ( X , Y ,) = 0 ,
    the fourth being always deducible from the other equations.
    on
    If this condition is fulfilled ,the solution of equation (3330 )
    is analogous to (3322 ).
    Integrate as if % and t were constant, and therefore dz and
    dt zero , adding for the constant of integration © (z , t).
    Differentiate next for all the variables , and determine by
    comparison with the original equation .
    3332 If a single primitive does not exist, the solution must
    be expressed by simultaneous equations in a manner similar
    to that of (3328).

    SIMULTANEOUS EQUATIONS WITH ONE


    INDEPENDENT VARIABLE .

    GENERAL THEORY.
    3340 Let the first of n equations between n + 1 variables be
    Pdx + P dy + P dz + ... + Pydw = 0 ......... (1) ,
    where P , P , ... P, may be functions of all the variables.
    Let æ be the independent variable. The solution depends
    upon a single differential equation of the nth order between
    two variables .
    Solving the n equations for the ratios da :dy :dz : & c ., let

    & Qi Q2
    dy _ Q. dx = % , ...... dwdx =_ QnQ '
    Differentiate the first of these equations n - 1 times, substi.
    tuting from the others the values of 2p ... W ,, and the result
    SIMULTANEOUS EQUATIONS. 495

    S
    is n equations in Y« Y2x ... Ynx, and the primitive variables
    X, Y , % ... W .
    Eliminate all the variables but x and y , and let the differ
    ential equation obtained be
    F (x , y , Yx ... Ynx ) = 0 .
    Find the n first integrals of this, each of the form
    F ( x , y , Yr ... Y (n - 1) 2 ) = C , and substitute in them the values of
    Yra Y2x , .. . Ynx, in terms of x , y, z ... W , found by solving the n
    equations last mentioned . Thus a system of n primitives
    is obtained , each of the form F (X , , % ... W ) = C .

    3341 The same in the case of three variables.


    Here n = 2 . Let the given equations be
    P dx + Qidy + R dx = 0 ,
    P dx + Qzdy + R , dz = 0 .
    - dx dy dz
    3342 Therefore
    herefore QR – Q,R R ,P , – P,R P1Qz – PyQi
    From these let y = + (x,y,z), 2. = 4 (x,y,z).
    Therefore Y2x = 4x + PyYx + 03% no
    Substitute the value of zr, and eliminate z by means of
    Yx = (2 , y , z ). An equation of the 2nd order in 2, y , Yue Y2x
    is the result. Let the complete primitive of this be
    y = x (a ,a , b ). Then we also have $ (x , y ,z ) = dex (x , a , 1).
    These two equations form the complete solution .

    FIRST ORDER LINEAR SIMULTANEOUS EQUATIONS WITH


    CONSTANT COEFFICIENTS.
    3343 In equations of this class, the coefficients of the
    dependent variables are constants, but any function of the
    independent variable may exist in a separate term .
    Such equations may be solved by the method of (3340 ),
    but more practically by indeterminate multipliers.

    3344 Ex. (1): dx +7 –y = 0, . + 2x + 5y = 0.


    Multiply the second equation by m and add . The result may be written

    d(v zmy)+(2m +7){x +3m +2y} = ...... ....(1).


    496 DIFFERENTIAL EQUATIONS.

    To make the whole expression an exact differential, put = m . This


    + 7

    gives m = =1+ 1, m'= = 1 ........ ........(2);


    (1) now becomes d (x dt+ my) + (2m + 7)(x + my) = 0,
    and the solution is æ + my = ce=(2m +7) and a + m 'y = c'e- 2m + 7)t.
    Solving these equations, and substituting the values (2 ),
    iy = ce-(6-i)t — c'e-(6-i)t = e-6t{ (c - c') cost- i (c + c ) sin t},
    ia = e-« {( 2 +itc) cost+ ( 7 - ) sint} ,
    or y = e-6 (C cost - C'sint), x = ke-6 { (C + C") cost+ (0 - C ) sin t}.
    3345 Ex. 2 : X : + 5x + y = e , Yt + 3y - x = ezt.
    Multiply the second equation by m , and add to the first
    d (a + my) + (5 - m ) 3s x + 1 + 3m
    y = et + met
    dt 5 - m )
    Put 1 + 3m
    " = m , thus determining two values of m ,and put æ + my = z ; thus
    5- m
    ze + (5 — m ) = e + met. This is of the form (3210) .
    Note . — The equations of this example, written in the symmetrical form
    dy
    of (3342), would be 7 dx _ -
    e' – 5x – y
    – = dt.
    el + x – 3y

    3346 General solution by indeterminate multipliers.


    Let
    be given with
    P1 = & + by + 42 + dı,
    P , = 0 ,2 + 12y + c9z + da,
    Ps = Aga + b3y + c2z + dz.
    Assume a third variable t and indeterminate multipliers l, m , n such that
    dt _ Idx + m dy + n dz _ 1dx + m dy + ndz
    T TIP, +mP, + nP, (lx + my + nz + r) * .......... . ( 1).
    The last fraction is an exact differential, and, to determine 1, l, m , n , r,
    we have
    a,l + azm + azn = 11, 1a - 1 ag ag i
    6, + b, m + bzn = am , I b, b, - . bg = 0.
    cl+ cą m + czn = in, La Cg Cs - 1
    dil + d2m + dyn = ir,
    SIMULTANEOUS EQUATIONS. 497

    The determinant is the eliminant of the first three equations in l, m , n .


    The roots of this cubic in furnish three sets of values of l, m , n , r , which,
    being substituted in the integral of (1 ), give rise to three equations involving
    three arbitrary constants ; thus,
    et = (læ + may + nz + r ) , cpt = (13x +May + ng + ry)"?,
    Cgt = (lge + mgy + ngz + rs) ^s.
    Eliminating t, we find for the solution two equations involving two
    arbitrary constants .
    A similar solution may be obtained when there are more than three
    variables .

    3347 To solveo P . -dx xP - P . dy - dz .


    - up = P . - > p = & c . ... ( 1),
    where Prax + by +c, Pi = djæ + by + cı, &c.
    Assame prat + on + ch, p. = 2,8 + bın + as, &c.,
    and take dě = dn = ds ................................. (2 ),
    1 Pi P2 P
    the solution of which is known by (3346). Substitate & = æs, n = ys, and
    these equations become
    xd5+ ždx – yds+ <dy – dç
    Pipp
    and therefore Gdx = Śdy – ds
    P1 - æp Pa - YPP
    Dividing numerators and denominators by 5 , the first equation in ( 1) is pro
    duced, and therefore its solution is obtained by changing &, n in the solution
    of (2) into us and ys.
    Certain simultaneous equations in which the coefficients
    are not constants may be solved by the method of multipliers .
    Thus,
    3348 Ex. (1): xx+ P (ax + by) = Q , ye+ P (a'x + l'y) = R ,
    P, Q , R being functions of t. Multiply the second equation by m , add, and
    determine m as in (3344 ) . The solution is obtained from
    ta

    2 +my = e ia+ma"|Pdt{0 + 1 éla+ma",Jedt(Q + mR) dt}, (3210)


    with two values of m .

    3349 Ex.(2): 2 + ( -y) = 1, yet 1 (2+ 5y) =t


    are equations solvable in a similar manner, and the results are
    x +y= $ (6 +6 + $), « +2y= (65+ 6 + .
    [ Boole, p . 307.
    3 s
    498 DIFFERENTIAL EQUATIONS.

    REDUCTION OF ORDER IN SIMULTANEOUS EQUATIONS.


    3350 THEOREM . - n simultaneous equations of any orders
    between n dependent variables and 1 independent variable are
    reducible to a system of equations of the first order by sub
    stituting a new variable for every derivative except the
    highest.
    3351 The number of equations and dependent variables in
    the transformed system will be equal to the sum of the indices
    of order of the highest derivatives. This will, therefore, in
    general be the number of constants introduced in integrating
    those equations. If, after integrating , all the new variables
    be eliminated , there will remain n equations in the original
    variables and the above-named constants. These equations
    form the complete solution .
    In practice, such reduction is unnecessary. The following
    are methods frequently adopted :
    3352 RULE I. — Differentiate until by elimination of a vari.
    able and its derivatives an equation of a higher order in one
    dependent variable only is obtained .
    3353 Rule II. — Employ indeterminate multipliers.
    3354 Ex. ( 1) : 2z = ax + by, Y2 = a'x + b' y.
    By Rule I.,differentiating twice for t and eliminating y and yz ,we obtain
    241 - ( a + b') azt + (ab' - a' ) x = 0 ,
    which may be solved by (3239).
    Otherwise by Rule II., exactly as in (3344),we find
    a 'm + (a - b') m - b = 0,
    and for the exact differential
    (x +my) 2 = (a + ma') (x +my) ,
    the solution of which , by (3239), is
    @ + my = Cyev(a +ma')t + 0 ,e - v (a +ma’)t
    in duplicate with the two values of m .

    3355 Ex. ( 2 ) : 271 – 2ayz + bx = 0 , yz + 2ax: + by = 0 .


    Differentiate, and eliminate y , yt, Yzt ; thus
    24 + 2 (2a + b) 222 + 6*x = 0,
    and solve by (3239). Otherwise assume
    æ = cos at + n sin at, y = ncos at - & sin at,
    and the given equations reduce to
    $ze = - (a ' + b ) , Mze = - (a ' + b ) n,
    which are solved in (3257) . [Boole, p. 311.
    PARTIAL DIFFERENTIAL EQUATIONS. 499

    3356 Ex. (3 ) .— Let u = 0 , v = 0, w = 0 be three equations in c, y, z, t,


    involving derivatives of t up to 234, Y6t, 274.
    To obtain an equation between a and t. Differentiate each equation
    6 + 7 = 13 times, producing 3 + 13x3 = 42 equations involving derivatives
    of t up to 2161, Y190, 720t. Between these 42 equations eliminate y, Yts ... Y19 ,
    %, 2 , ... % 20t, in all 41 quantities, and an equation of the 16th order in x and t
    is the result . [De Morgan ,

    3357 If a number of equations involve the quantities x , Xat,


    West, & c., Yı, Ysı, Yst, & c., all in the first degree ,these quantities
    may be eliminated by assuming
    x = L sin pt, y = M cos pt.

    3358 If there be n linear homogeneous equations in n vari.


    ables x , y , % , ... and their derivatives of the 2nd order only ,
    ns
    the equations may be solved by putting
    x = L sin pt, y = M sin pt, % = N sin pt, & c .
    3359 Ex. : 224 = ax + by, yz = gæ + fy .
    Putting x = L sin pt, y = M sin pt,
    (a + p ') L + M = 02 lap,
    gL + ( f + p ) M = 0S? " g f + p1
    p and the ratios L : M are thus found.
    Suppose L = - kb and M = k (p + a ),
    then x = - kb sin pt, y = k ( p * + a ) sin pt,
    and k and t are arbitrary constants.

    PARTIAL DIFFERENTIAL EQUATIONS.

    3380 An equation is termed a general primitive or a com


    plete primitive of a partial differential equation , according as
    the latter is obtained from it by eliminating arbitrary functions
    or arbitrary constants, as illustrated in (3150 – 7 ).
    LINEAR FIRST ORDER P. D . EQUATIONS
    3381 To form the P . D . equation from the primitive
    U = ¢ ( ) , where u and v are functions of x , y , % .
    500 DIFFER ENTIAL EQUATIONS.

    RULE. — Differentiate for x and y in turn, and eliminate


    $ ' ( v ). See (3054 ) .
    Otherwise. — Differentiate the equations ura, v = b ; thus
    u,dx + u ,dy + u,dz = 0 ,
    vydx + vydy + v,dz = 0.
    Therefore dx = dye de where P = da((yzx) , de.
    Then the P . D . equation will be
    PzZ + Qz, = R .
    PROOF . - Since z is a function of x and y, 2 .dx + z ,dy = dz. But dx = kP,
    dy = kq, dz = kR , therefore kPzz + kQ2y = kR .

    3382 Ex. - The general eqnation of a conical surface drawn through the
    point (a , b, c) is y - b
    C - a
    the form of o being arbitrary.
    Considering x as a function of two independent variables a and y , differ
    entiate for æ and y in turn , and eliminate $ as in (3154) . The result is the
    partial differential equation
    (x - a ) 2 : + (y — 6 ) , + - c = 0.
    3383 To obtain the complete primitive ; that is, to solve
    the P . D . equation , Pz: + , = R ,
    P , Q , R being either functions of x , y, z or constants.
    Rule. - Solve the equations

    Let the two integrals obtained be u = a, v = b ;


    then u = $ (v )
    will be the complete primitive.
    Propositions (3381) and (3383) extended to any number
    of variables.
    3384 To form the partial differential equation from the
    primitive $ (u , v , ... W ) = 0 ..................... ( 1),
    where u , v, ... w are n given functions of n independent vari.
    ables x, y , ... % and one dependent t.
    PARTIAL DIFFERENTIAL EQUATIONS. 501

    RULE. — Differentiate for all the variables Mthus,


    US,
    Pu du + pdv + ... + øvdw = 0 ...............(2 ).
    Therefore, since p is arbitrary , du ,dv ... dw must separately
    vanish , giving rise to the n equations
    du = u_dx + u,dy + ... + u4dt = 0,
    dv = v,dx + v,dy + ... + vtdt = 0 ,
    dw = w _dx + w ,dy + ... + w dt = 0.
    Solving these for the ratios,by (583),we get
    s = d ... . ........(3),
    P , Q .. . R , S being functions of the variables or else constants .
    Now , t being a function of all the rest,
    tedx + tydy + ... + t,dz = dt ............... (4 ),
    therefore, by (3) and (4 ), the partial differential equation
    required is
    3385 Pt + Qt, + ... + Rt, = S .
    3386 If u , V ... w be n functions of n variables , x , y ... t, the
    condition of interdependence of the functions or existence of
    some relation expressed by equation (1) is J (u , v ... W ) = 0 .
    (see 1606) ; that is, the eliminant of equations (2) must vanish .

    3387 Conversely, to integrate the partial differential equa


    tion Pt, + Qt, + ... + Rt, = S .................. (1).
    RULE. — Solve the system of ordinary equations
    = & c. – dz _ dt
    = & c. = RES ...... (2),
    and let the integrals obtained be u = a, v = b , ... w = k ;
    then ^ (u , v, ... W ) = 0 will be the complete primitive.
    If P , Q ... R , S are linear functions of the variables, the
    integrals of equations (2 ) can always be found by the method of
    (3346).
    502 DIFFERENTIAL EQUATIONS.

    NOTE. — Suppose, in equation (1), that any coefficients P , Q


    vanish ; then , by (2 ), dx = 0 , dy = 0 , and therefore the cor
    responding integrals are x = a , y = b. The complete primi
    tive thus becomes
    $ (X , Y, U , V ... W ) = 0 .

    3389 When only one independent variable occurs in the


    derivatives of the partial differential equation , the equation
    may be integrated as though the others were constant, adding
    functions of the remaining variables for the constants of
    integration .

    3390 (Ex. 1): 42


    da = Integrating for æ as though ywere con
    stant, the complete primitive is
    % = y sin - 2 + (y ) .

    Some equations are reducible to the above class by a transformation .


    Thus :
    3391 Ex. (2 ) : zy = * + y?. Put zx = u,
    therefore U ,, = x + + y?, therefore u = % = 2 *y + $ y8 + ( 2x ),
    therefore z = fx®y + {xy' + S¢ ( ) dx + 4 (y ),
    or % = } ( x* y + xy') + x (x ) + ų ( y ).

    3392 Ex. (3 ): (x –a ) zz + (y — b)2, = C— %.


    Solving by (3283), dx = dy – dz .
    x -a7-67-
    The integrals are
    log (y — b ) – log (x — a ) = log 01. or y - b = c. z C = '.
    . log (x – c ) - log (x — a ) = log C ' S ? 2 - a ' X - a

    therefore 46 = 9 (2 -4 ) isthe complete primitive.


    a

    For the converse process in respect of the same equation, see (3382) .
    3393 Ex. (4 ). — To find the surface which cuts orthogonally all the
    spheres whose equations (varying a ) are
    . ..................... (1).
    x ? + y + z — 2ax = 0 .......
    Let g (x, y, z) = 0 be the surface . Then
    ( c − a) + + + + = $2 0
    by the condition of normals at right angles. Substitute the value of a from
    (1 ), and divide by Oz ; thus,
    (x* — y: — zº)zz + 2xyz, = 2zx.
    PARTIAL DIFFERENTIAL EQUATIONS. 503

    By(3383), 3+*+x + =
    dy z dz gives \ = c for one integral.
    Substituting y = cz , we then have
    I da
    23 - ( + 1)
    which, being a homogeneous equation in x and z ,may be solved by putting
    z = vx (3186 ). The resulting integral is y + ? = 0. Hence the com
    plete primitive is tº+ 4*+** = ( .) and the equation ofthesurface sought.
    3394 Ex. (5 ). — To find an integrating factor of the equation
    (x *y – 2y ) dx + (xy* — 2x *) dy = 0 ...... ...... ... ( 1 ).
    Assuming z for that factor, the condition (Mz), = ( Nz) . (3087) pro
    duces the P . D . equation
    (wy* — 2x ) 2 + (2y* — x*y) , = 9 (28 — y*)....................(2).
    The system of ordinary equations (3283) is
    - dx dy - dz
    g° – 2x+ 2y - dºg 9 ( dº - gº) =
    The first of these equations is identical with (1) (and such an agreement
    always occurs). Its integral is
    Also y dx + x dy
    ask – –ay 9(a = )
    which reduces to 3dx + 3dy + dz = 0 ;
    and thus the second integral is mºyez = c .
    Hence the complete primitive and integrating factor is
    11x

    Any linear P . D . equation may be written as a homogeneous


    equation with one additional variable ; thus, equation (3387)
    may be written
    3395 Pur + Qu , + ... + Ru, = Sue
    SIMULTANEOUS LINEAR FIRST ORDER P . D . EQUATIONS.
    3396 PROP. I. - The solution of such cquations may be made
    to depend upon a system of ordinary lst order differential
    504 DIFFERENTIAL EQUATIONS.

    equations having a number of variables exceeding by more than


    one the number of equations.
    Let there be n equations reduced to the homogeneous form
    (3395 ) involving one dependent variable P and n + m inde
    pendent. Select n of the latter, X , y ... 2 , and let the remaining
    m be , n ... . From the n equations find P .x, P , ... P , in
    terms of Pt, P , ... Ps, and arrange the results as under :
    Pr +aPf+ b, P , ... + k , P = 0 )
    P , + d , P + b , P , . .. + k , P = 0 ..... ( 1).
    Pr+ anPx+bPr ... + 1, Ps = 0 )
    Multiply these equations by 11, 12,...d respectively,and add ;
    thus ,

    1,8a +AP,...+ 29Pc + 5(2a)Po+ 9 (AL)P....+ ( ) Pę... =(2 )0.


    From this, as in (3387), we have the auxiliary system
    zaka) = 2 .. 2(C2) . (3),
    and ,by eliminating 11, do ...
    dę – a dx — a dy .. . - a , dz = 0 )
    dn - b ,dx — bądy ... — b, dz = 0 !
    . .. . .. . ..
    dų – kądx — Kądy ... — k , dz = 0 )
    Then , if u = a , v = b , & c . be the integrals of (4 ), they
    will be values of P satisfying the equivalent system (1 ) , and
    the integral of that system will be F (u , v, ...) = 0 .

    3397 PROP. II. — To integrate a system of linear 1st order


    P . D . equations.
    Let = adx + bd , ... + kda,
    so that AP = 0 represents a homogeneous linear P . D . equa
    tion of the 1st order .
    RULE. — “ Reduce the equations to the homogeneous form ( 1 ) ;
    express the result symbolically by
    4P = 0 , 4 , P = 0 , ... A , P = 0 ,
    PARTIAL DIFFERENTIAL EQUATIONS. 505

    and examine whether the condition


    (4:4; — 4;4;) P = 0
    is identically satisfied for every pair of equations of the system .
    If it be so, the equations of the auxiliary system ( Prop. I.) will
    be reducible to the form of exact differential equations, and
    their integrals being ura, v = b , w = C, ..., the complete
    value of P will be F (u , v , w , ... ), the form of F being
    arbitrary .
    “ If the condition be not identically satisfied , its application
    will give rise to one or more new partial differential equations.
    Combine any one of these with the previous reduced system , and
    again reduce in the same way.
    “ With the new reduced system proceed as before,and continue
    this method of reduction and derivation until either a system
    of P . D . equations arises, between every two of which the above
    condition is identically satisfied , or , which is the only possible
    alternative, the system P = 0 , P , = 0 , .. . appears. In the
    former case, the system of ordinary equations corresponding to
    the final system of P . D . equations, will admit of reduction to
    the exact form , and the general value of P will emerge from
    their integrals as above. In the latter case , the given system
    can only be satisfied by supposing Pa constant."

    3398 “ Ex. : Px + (t + xy + xz) P + ( + – 3. ) P , = 0,


    P, + (xzł + y — ay) P2 + (zt - Y) P = 0 .
    Representing these in the form AP = 0, A , P = 0 , it will be found that
    (4 , 4, – 4 ,4 ,) P = 0 becomes, after rejecting an algebraic factor,aPz + P = 0 ,
    and the three equations prepared in themanner explained in the Rule will
    be found to be
    Px + (3x + t) P, = 0 , P , + yP: = 0, Pc + xP, = 0 .
    No other equations are derivable from these. We conclude that there is but
    one final integral.
    " To obtain it, eliminate Px, P , P , from the above system combined with
    Pedx + P ,dy + P dz + P /dt = 0,
    and equate to zero the coefficient of P , in the result. We find
    dz - (t + 3x”) dx - ydy - ædt = 0 ,
    the integral of which is - « t — 2 — Žy: = c.
    “ An arbitrary function of the first member of this equation is the general
    value of P ." [Boole, Sup., Ch. xxv.
    For Jacobi's researches in the same subject, see Crelle's Journal, Vol. Ix.
    3 T
    TIAL S
    506 DIFFEREN EQUATION .

    NON- LINEAR FIRST ORDER PARTIAL DIFFERENTIAL


    EQUATIONS.
    3399 Type F (x , y, z , 2 , Zy) = 0 ..................... (1).
    CHARPITS'S SOLUTION . — Writing p , q instead of zz and Zy,
    assume the equations
    = du = do_ = dp_ ... ..... (2).
    lp 91 - ľP9p9x + pq:
    Find a value of p from these by integration , and the corres .
    ponding value of q from the given equation , and substitute int
    the equation
    dz = pdx + qdy... ... ........( 3),
    and integrate by (3322) to obtain the final integral.

    P , = qx ed ,thisPc,onand
    inExpress he bhypothesis
    sta fqxunonctitthe
    PROOF. - Since dz = pdx + qdy, we have, by the condition of integrability,
    ypor that z is a function of 2 , y ;
    p a function of x , y , % ; q a function of x , y , z , P ; considering x constant when
    finding Py, and y as constant when finding 12. Equating the values of P , and
    2 . so obtained, the result is the equation
    Apr+ Bp , + Cp: = D ,
    in which A , B , C , D stand for – 2 , 1, 9 - 99p q + P1z.
    Hence, to solve this equation , we have, by (3387), the system of ordinary
    equations ( 2).
    3400 NOTE. — More than one value of p obtained from equations (2 ) may
    give rise to more than one complete primitive.
    The first two of equations (2) taken together involve equation ( 3).

    DERIVATION OF THE GENERAL PRIMITIVE AND SINGULAR


    SOLUTION FROM THE COMPLETE PRIMITIVE .
    RưLE . — Let the complete primitive of a P . D . equation of
    the 1st order be
    z = f (x, y , a, b )........................(1).
    3401 The general primitive is obtained by eliminating a
    between z = f {x, y, a, 9 (a)} and f. = .............(2),
    the form of o being specified at pleasure.
    3402 The singular solution is obtained by eliminating a and
    b between the complete primitive and the equations
    fx = 0 , f , = 0 .. . ... .. ... ...( 3).
    PARTIAL DIFFERENTIAL EQUATIONS. 507

    Proof. — By varying a and b in ( 1) ,


    p = fr + faQx + fobx, q = fy + faa, + f7b ,.
    Therefore, reasoning as in (3171), wemust have
    frax + fobe = 0 and faay + f b, = 0 .................. (3),
    therefore either fa = 0, fo = 0 , leading to the singular solution ; or , elimi
    nating fas for arb,, a ,,bx = 0 ,
    and therefore, by (3167) , b = q (u ). Multiply equations ( 3 ) by dæ , dy re
    spectively, and add , thus fada + fodb = 0 . Substitute b = º (a ) in this and
    in (1), and the equations (2 ) are the result.

    SINGULAR SOLUTION DERIVED FROM THE DIFFERENTIAL


    EQUATION.
    3403 RULE. — Eliminate p and q from the differential equa
    tion by means of the equations
    2p = 0 , 2q = 0 .
    PROOF. — Let the D . E . be z = f ( x , y , p , q), and the C . P . = F (x , y , a , 6 ).
    Now p and q being implicit functions of a and b , we have, from the first
    equation , Za = % ,Pa + % qla 40 = 2ppo + %990.
    Hence the conditions za = 0 , % = 0 in ( 3 ) involve, and are equivalent to ,
    2, = 0 , 2q = 0.

    3404 All possible solutions of a P . D . equation of the 1st


    order are represented by the complete primitive , the general
    primitive, and the singular solution . [Boole, p. 343.

    3405 To connect any given solution with the complete


    primitive.
    Letz = F (x , y, a , b ) be the complete primitive, and
    z = $ (x , y ) some other solution .
    Determine the values of a and b which satisfy the three
    equations F = , Fn = $ x F, = py
    If these values are constant, the solution is a particular
    case of the complete primitive ; if they are variable so that
    one is a function of the other, the solution is a particular case
    of the general primitive ; if they are variable and unconnected ,
    the solution is a singular solution .

    3406 CoR . – Any two solutions springing from different


    complete primitives are equivalent.
    508 DIFFERENTI EQUATIONS.
    AL

    3407 Ex.: z = px + qy + p9 ...


    By (3299) , dos = dy= f= d .......... ....(2),
    and we have
    q =* * ; A = = 2 = 6 + )
    C = q-2 – 2pz + yz - pix. = 9x + p9, = P + P = 0.
    (p + y)? p + y
    Hence (9 ) hecome
    omes
    s ( p + y ) , » - dy =_ ( p + y ): dz _ dp .
    xy + z 2pz + yz - p*x
    ii dp = 0 , p = a ; .:. q = 2 - lee Substit utina
    Ax. Substituting in dž = pdx + qdy,
    aty
    dz = adx + 2 — ax du .. . . . . . .
    aty
    By (3322), making % constant, 4adx + dy = 0,
    % - ax aty
    therefore – log ( - ax ) + log (a + y ) = 0 (2 ) ......... . ... .. . ( 4 ) .
    Differentiate for x , y, z , and equate with ( 3), thus ¢ (2 ) = 0, therefore
    0 (z ) = constant (say - log b ) ; therefore, by (4 ) , z = ax + by + ab, the C . P .
    of ( 1) .
    3408 To find a singular solution by (3402), we must eliminate a and b
    between q = 0 , 2x = 0 ; that is, a + b = 0 and yta = 0 ,
    therefore % = - xy - xy + xy = - XY
    is the singular solution.
    To find the general primitive by (3401), eliminate a between the two
    equations % = ax + ( y + a ) • (a ) and a + (y + a ) $' (a ) + o (a ) = 0 .

    NON -LINEAR FIRST ORDER P . D . EQUATIONS WITH MORE


    THAN TWO INDEPENDENT VARIABLES.
    3409 PROP. — To find the complete primitive of the differ
    ential equation
    F (X1,X,...Wns , P1,P2...Pn) = 0 ............(1),
    re
    where ps by p:= d. de.
    3410 RULE . — Form the linear P . D . equation in o denoted by
    s SdF dF do dF (dø . do 12 _= 00 ,
    4: { \dx, Prdz ) dp, dp, ldx, T Prdz ) ) -
    the summation extending from r = 1 to r = n . From the
    auxiliary system (3387) n - 1 integrals
    0i = a1, 0 , = 22, ... 0n - 1 = An - 1
    SECOND ORDER P . D . EQUATIONS. 509

    may be obtained. From these equations, together with ( 1 ), find


    P1, P2 ... Pn in terms of X1, X , ... Xn, substitute the values in
    dz = pıdx + p ,dx, ... + p ,dxn,
    and the integral of this last equation will furnish the solution
    required in the form
    f (x1, X , .. . Xn, 2 , 21, 22 .. . an) = 0 .
    [Boole, Diff. Eq., Ch . xiv ., and Sup., Ch. xxvii.

    SECOND ORDER P . D . EQUATIONS.

    3420 Type F (x , y , % , % x ,my,72xy @ xy,m2v) = 0 .


    The derivatives x, y, 72.x2 + xys 22, are briefly denoted by
    P , q , r, s, t respectively .
    z being a function of the two independent variables x and
    y , the following values are of frequent use
    3421 dz = pdx + qdy; dp = rdx + sdy ; dq = sdx + tdy.
    If u be any function of x , y , and % , the complete deriva
    tives of u are indicated by brackets, thus
    3422 (2x) = ur + puz, (u ) = 1 , + quz.

    A linear 2nd order P . D . equation is of the type


    3423 Rr + Ss + Tt = V ... ..... .
    in which R , S , T , V are functions of x , y , % , p , q .

    PROPOSITION . — Any P . D . equation of the 2nd order which


    has a first integral of the form u = f ( u ) , where u and v involve
    X , Y , Z , P , 9, is of the form
    3424 Rr + Ss + Tt+ U (rt — sº) = V ...............(2),
    where R , S , T , U , V are functions of x , y , % , p , q , and
    3425 U = U , V , — U ,Up ... ...... ... ......... ... (3) .

    Proof. — Differentiate u = f ( v ) for x and y separately, considering x, y , z ,


    P , q all involved in u and v, and eliminate f ' ( v ). The result is equation ( 2 ) ,
    with the values
    510 DIFFERENTIAL EQUATIONS.

    3426uR = u ,(0,) — (U.,) %pe u T = vg(14x) — (v.x)Uq,


    pS = v) (1.,) — (vx)un — v , ( ,) + (0,) Wq,
    AU = U »V — U ,'ne u V = (1. )(4x) – (un)( ,),
    with the notation (3422) , u being an undetermined constant.
    3427 CoR. — The condition to be fulfilled in order that
    equation ( 1) may have a first integral of the form u = f (x ) is
    un ” , - U , V , = 0 .

    SOLUTION BY MONGE'S METHOD OF


    3428 Rr + Ss + Tt = V .
    RULE . - Write the two equations
    Rdy — Sdx dy + Tdx = 0 .................. ( 1 ),
    Rdpdy - Vdx dy + Tdqdx = 0 ...............(2 ).
    Resolve ( 1) into its factors, producing the two equations
    dy - m ,dx = 0 and dy - m ,dx = 0.
    From dy = m , dx and equation ( 2 ) combined , if necessary ,
    with dz = pdx + qdy, find two ist integrals u = a , v = b ;
    then u = f (v ) will be one Ist integral of the given equation .
    Similarly from dy = m , dx find another 1st integral.
    3429 The final 2nd integral may be found from one of the
    1st integrals by Lagrange's method (3383).
    3430 Otherwise , determine p and q in terms of a , y, z from
    the two 1st integrals ; substitute in dz = pdx + qdy, and then
    integrate by (3322) to obtain the final integral.

    3431 If equation ( 1 ) is a perfect square, there will be only


    one 1st integral, and Lagrange's method only is applicable.
    Proof. — By (3427) we may put ug = mup, vo = mvp ; and also
    dz = pdx + qdý (3321) in the complete derivatives
    (du) = u , da + wydy + u ,dz + u ,dp + u , dq = 0, (dv) = & c . = 0 ;
    : . by (3422) (1x)dx + (14 ) dy + u , (dp + mdq) = 0 ......... (3 ).
    (2x) dx + (vy) dy + v, (dp + mdq) = 01"
    Solving these equations for the ratios dx :dy :dp +mdq, we obtain at once
    2 – dy + mdx – mdu – dp +mdq ........... ( 4 ),
    R - T
    with the values of R, S, T, V in (3426).
    SECOND ORDER P. D . EQUATIONS. 511

    Equations ( 1) and ( 2 ) are the result of eliminating m from (4 ) . These


    two equations with dz = pdx + qdy suffice to determine a first integral of
    (3428) when it exists in the form u = f (v ).

    3432 Ex. (i.) : 9 ( 1 + q ) r - (p + a + 2pq) s + p (1 + p ) t = 0.


    Solving the quadratic equation (1), we find
    pdx + qdy = 0, or (1 + p)dx + (1 + 2) dy = 0 ...............(5).
    First, dz = pdx + qdy = 0 , ii % = A .
    Monge's equation (2 ) is 9 ( 1 + 9) dp dy + p ( 1 + p ) dq dx = 0 ,
    dp - dq ; and, integrating , 1 + P = B .
    which, by pdx = - qdy, " gives
    & i1 + p i + q amu,meraung 1 + 9
    Hence a first integral is 1 + P = $ (2 )...... ............(6 ).
    1 + 9
    Next, taking the second equation of (5 ) with
    dz = pdx + qdy, dx + dy + dz = 0 , s. x + y + z = C .
    Also, by (5 ) , equation ( 2) now reduces to qdp = pdq , and by integration,
    p = qD ; therefore the other first integral is p = q4 ( x + y + z ).
    For the final integral integrate p - 94 = 0 ; i.e., 7x – 47,, = 0 , by (3383 ) ;
    da
    .. z = , and dx -= lx + dy + dz
    dy
    . . dx =- -
    * (x + y + z) – Ū - 1 - 4 (x + y + z )
    :: x = ( - (x4 (+x y+ +y z)+ x ) .= f '(x + y + z ) + B .
    J1
    Hence the second integral is a: – f (x + y + z ) = F (-).

    3433 Ex. (ii.) : 2, - aʻra, = 0.


    (i.) Here, in (3428), R = 1, S = 0 , T = - a , V = 0 ; therefore (1)
    and (2 ) become dy' – a'dx’ = 0, dpdy - a *dq dx = 0 .
    From ( 1 ) dy tadx = 0 , giving ytax = c, and converting (2 ) into
    dp + adq = 0 , which gives p + aq = c' ; therefore a first integral is
    p + aq = ( y + ax) ... ....... .......... ....... ( 3 ) .
    Similarly, from ( 1), dy a dx = 0 gives rise to another first integral
    p - aq = y (y - ax ) ... ......... (4 ) .
    Eliminating p and q by means of ( 3) and (4) from dz = pdx + qdy, we find
    dz = (2a)-' {$ (y + ax ) (dy + a dx) – 4 (y — ax) (dy - adx) },
    therefore, by integrating, z = ( y + ax ) + 4 (y - ax ).
    For the symbolic solution of the same equation , see (3566).

    SOLUTION OF THE P . D . EQUATION .


    3434 Rr + S8 + Tt + U (rt - 3") = V ... . .. ... ... ..
    Letmı, m , be the roots of the quadratic equation
    3435 m – Sm + RT + UV = 0 . .. ... . . (2).
    512 DIFFERENTIAL EQUATIONS .

    Let u , = l , V = b , and U2 = a ', 1 , = b ' be respectively


    the solutions of the two systems of ordinary differential
    equations.
    3436 Udp = mzdy – Tdx ). Udp = mdy - T dx )
    Udq = mydx — Rdy (3 ) , Udq = m _dx – Rdy (4 ).
    dz = p dx + qdy) dz = p dx + qdy)
    Then the first integrals of (1) will be
    U = f (u ), = f (v2).
    To obtain a second integral :
    3437 1st. — When mi, m , are unequal, assign any particular
    forms to fi and f2, then substitute the values of p and q , found
    from these equations in terms of x and y , in dz = pdx + qdy,
    which integrate . Otherwise, assign the form of one only of
    the functions fi f2, involving an arbitrary constant C , solve
    for p and q , and integrate dz = pdx + qdy, adding an arbitrary
    function of C for the constant of integration .

    3438 2ndly . — When m1, m , are equal, and therefore, by (2),


    S ? = 4 (RT + UV) ...................... (5 ).
    Equations (3) and (4 ) coincide, and, since m = is,
    reduce to
    3439 Udp = isdy — Tdx .....................(6 ),
    Udq = iS dx - R dy .................. ... (7 ),
    dz = p dx + q dy .....................(8 ).
    Here p , = , and therefore the last equation is integrable
    if the values of p and q, obtained by integrating (6 ) and ( 7 ),
    be substituted in it. Let u = a , v = be the integrals of (6 )
    and (7); and let z = $ (x,y,a ,b, c) ..... ( 9)
    be the integral obtained from (8).
    The general integral is found by making the parameters
    a , b , c vary subject to two conditions b = f (a ), c = F (a ) ;
    that is , by differentiating
    aling

    z = " { x, y, a , f (a ), F (a ) }
    for a , and eliminating a .
    3440 The general integral therefore represents the envelope
    of the surface whose equation is ( 9 ) .
    SECOND ORDER P. D. EQUATIONS. 513

    PROOF. — (Boole, Sup., p . 147.) Assuming a 1st integral of the form


    u = f (v ), eliminate u and v from equations (3 126 ) by multiplying (i.) by
    (uz) un, (ii.) by (wy) Up, (iv .) by (uz) (u ), ( v.) by upup, and adding. Again ,
    eliminate u and v by multiplying (i.) by (uz)?, ( ii.) by (u,) , (iii.) by (uz)(uy),
    (v.) by (ur) Up + ( ) Ug, and adding. The two resulting equations are
    R (Uz)ug + T (1,) Un – U (uz)(Un) + VU,V , = 0 ..(10).
    R (uz)' + S (1x) ( ,) + T (12)* + V {(uz) u + (u ,) un} = 0 )
    Multiply the 2nd of these by m , divide by V , and add to the 1st equation ;
    the result is expressible in two factors either as (11) or (12),
    {R (14) + m2 (un ) + Vup}{m , (ur) + T (1, ) + Vuq} = 0 ......... (11),
    { R (Ux) + m2(1 . ) + Vur} {m2 (ur) + T (1 , ) + Vur} = 0 ......... (12),
    M1, m , being the roots of the quadratic ( 2 ). By equating to zero one factor
    of (11) and one of ( 12 ), we have four systems of two linear lst order P . D .
    equations. Taking each system in turn with the equations
    (uz) + rug + su , = 0,
    (U,) + su , + tug = 0,
    and eliminating (u ) , ( u ), U7, Uq, we have the de.
    is
    I m V
    terminant annexed for the case in which the 1stt me T O P
    factor of (11) and the 2nd of ( 12) are equated to = 0,
    0 g s
    zero. In this case , and also when the 2nd factor 1 8 t !
    of ( 11) and the 1st of ( 12 ) are chosen . trans. 0
    posing my, m in the determinant, the eliminant is equivalent to
    V {Rr+ Ss + Tt+ U (rt-— s )— V } = 0,
    having regard to the values of m , m , and m + m , from (2).
    When the 1st factor of both ( 11) and (12) is taken, the 2nd order P . D .
    equation produced by the elimination is
    Vt- R (rt — ) = 0,
    and when the 2nd factor of each is taken ,the elimination produces
    Vr - T (rt - s*) = 0.
    Hence the hypothesis of a 1st integral of (1), of the form u = f (v ),
    involves the satisfying one or other of the systems of two simultaneous equa
    tions, (13) or (14 ), below :
    R (17) + m (un) + V1, = 0 ){ . .. ( 13). R (1x) + m2(1 ) + Vu, = 0 )
    m , (uz) + T (Un) + Vuq = 0 ) " my (14) + T (1.,) + Vuq = 0 5 ****
    Now multiply the 2nd equation of (13) by 1 and add it to the 1st.
    In the result, collect the coefficients of Wz, Uy, Up, Ug Uz. The Lagrangean
    system of auxiliary equations (3387) will then be found to be
    _ dx = dy _ = dp = dq = dz
    R + Amam + AT - R p + m ,9 + 1 ( Tg + m , n ) Ū*
    Eliminating 1 , equations (3 ) are produced. Treating equations (14) in.
    the same way, equations (4) are produced.
    3
    514 DIFFERENTIAL EQUATIONS.

    POISSON 'S EQUATION .


    3441 P = (rt - 2)" Q ,
    where P is a function of p , 9 , 1 , s, t, and homogeneous in r , s,
    t ; Q is a function of x , y , z , and derivatives of % , which does
    not become infinite when rt — s2 vanishes, and n is positive.
    RULE . — Assume q = p (p ) and express s and t in termsof qp
    and r ; thus, rt - s : vanishes, and the left side becomes a func

    Solve for a 1st integral in terms of p and q , and integrate


    again for the final solution .
    Proof. — s = q; = 4pPx = qor ; t = 9 , = WP, = dir ;
    therefore rt — gʻ = 0. Also P is of the form (r, s, t)" = 3 " (1, 9p, 97.)" .
    Hence the equation takes the form (1 , 2 , 0;) " = 0.

    LAPLACE'S REDUCTION OF THE EQUATION .


    3442 Rr + Ss + Tt + Pp + Q9 + Zx = U ............ (1 ),
    where R , S , T, P , Q , 2 , U are functions of a and y only .
    Let two integrals of Monge's equation (3428)
    Rdy. — Sdxdy + Tdx = 0
    be ♡ (@ , y) = a , + (x, y) = b.
    Assume & = º ( ll, y ), n = { (x , y ).

    we have
    P = = + " < ; 1 = , = 26, + 2,9, ; :
    p = 222 = 2252 + 2 59 x1.x + 227 + 6822 + 2,024 (1701)
    t = 2y = 2245 + 27 n $,9, + 221?; +382, + , 12y ;
    s = "xy = 22:58, + 7.2. 9.x 9, + en (Ex1, + 8,9x) + z< 5xy + 2, wy•
    The transformed equation is of the form
    zien + Lzz + M2, + Nz = V .................. (2),
    where L , M , N , V are functions of & and n . This equation
    may be written in the form
    (ds+ M ) (d , + L ) + (N - LM – Ly) z = V .........(3 ).
    If N - LM - I = 0 ... .................. ( 4),
    we shall have
    (d : + M )a = V with ( 1, + L ) z = ,
    LAW OF RECIPROCITY . 515

    and
    from
    the solution by a double application of (3210) is obtained
    is e-Siln{$ (E) + (zefLandn},
    a= (w)+ ſveswat .
    By symmetry, equation (1) is also solvable, if
    N - LM – M , = 0

    But if neither of these conditions is found to hold , find %


    in terms of z ' from ( 3 ) . It will be of the form
    % = Azx + Bz' + C ,
    where A , B , C contain Ğ and n . Substitute this for % in
    (d , + L ) % = z' , and the result is of the form
    then + L'zę + M 'za + N 'z' = V '.
    The same conditions of integrability, if fulfilled for this equa
    tion , will lead to a solution of ( 1 ), and, if not fulfilled , the
    transformation may be repeated until one of the equations,
    similar to (4 ) or (5 ), is satisfied.

    3444 CoR. — The solution of the equation


    en + azy + b2n + abz = V
    an + be

    is z = e -an 6 (E) + e=6$ 4 (n) + e -an-of lean+8€ Vdn dĘ.


    3445 For the solution of equation ( 2 ), when L , M , V contain also z , see
    Prof. Tanner, Proc. Lond. Math . Soc., Vol. viii., p . 159.

    LAW OF RECIPROCITY . [ Boole, ch. xv .

    3446 Let a differential equation of the 1st order be


    $ (x , y, z , p, q) = 0 . . . . . . . . . . . . . . . . . . . . ( 1 ) .
    Let the result of interchanging x and p , y and q , and of
    changing z into px + qy — % , be
    $ (p, q,px + qy — , x , y) = 0...............(2);
    then , if z = { ( , y ) be the solution of either ( 1) or (2), the
    516 DIFFERENTIAL EQUATIONS.

    solution of the other will be obtained by eliminating & and in


    between the equations
    x = d4 ($, n ), y = d .y ($,n), < = { r + ny — ų (s, n ).
    3447 Ex. — Let : = pq ...... ( 1), pr + ay - z = zy ...... (2),
    be the two reciprocal equations.
    The integralof (2)is = = xy + zj (% ), :: 4 (3n) = En +zj ( ).
    E, n have now to be eliminated between
    = - (?) + (!) y = +f (z) = n... (3).
    Each form assigned toj gives a particular integral of (1).
    the equations ( 3) become r = ntb, y = & ta, z = in,
    and the elimination produces z = (2 - 6 ) (y - a ).
    3448 In an equation of the 2nd order, the reciprocal equa
    tion is formed by making the changes in (3416 ) , and , in
    addition , changing
    20 into t
    p into i s into - into
    rt - 3 ? rt .82 rt

    then , if the 2nd integral of either equation be z = * (x , y) , that


    of the other will be found by the same rule .

    3449 The above transformation makes any equation of the


    form ( p , q ) rty ( p , q ) stx ( p , q ) t = 0
    dependent for solution upon one of the form
    x (x, y) r - 4 (x, y) s + $ (x, y) t = 0.
    3450 And , in the sameway , an equation of the form
    Rr + Ss + Tt = V (rt — 3°)
    is dependent for solution upon one of the form
    Rr + Ss + Tt = V .
    See De Morgan, Camb. Phil. Traus., Vol. VIII.
    SYMBOLIC METHODS.

    FUNDAMENTAL FORMULÆ .
    Q denoting a function of 0,
    3470 (do - m ) - ' Q = eme ſe -me Qdo.
    SYMBOLIC .METHODS. 517

    PROOF. — The right is the value of y in the solution of doy - my = Q by


    (3210 ) . But this equation is expressed symbolically by (d . - m ) y = Q ( see
    1492) , therefore y = (do - m )” Q .
    Let @ = €9, therefore do = æd, and æd0 = dx . Hence
    (3470) may be written
    3471 (vd : — m )--Q = 2m [v-m-1Qdx.
    3472 Cor . (do - m ) -²0 = Cemo,
    3473 or (.vdo -m )-10 = Cx”.
    Let F ( m ) denote a rational integral function of m ; then ,
    since de eme = memo , d2e emo = maeme, & c ., the operation de is
    always replaced by the operation mx. Hence, in all cases,
    3474 F (do) emo = emoF (m ).
    3475 F (do) emo Q = eme F (do+ m ) Q .
    Formula (2161) is a particular case of this theorem . rem

    3476 eme F (do) Q = F (do - m ) emo Q .

    Also ,by (3474–6 ),


    3477 F (m ) = e-m® F (do) emo.
    3478 F (do + m ) Q = e-me F (do) emoQ .
    3479 F (d .) Q = e -me F (do - m ) emeQ .
    To the last six formulæ correspond
    3480 F (adv) mm = x" F (m ).
    3481 F (xdx)2MQ = x " F (xdx + m ) Q .
    3482 8" F (xdx) Q = F (xd, - m ) x ” Q .
    3483 F (m ) = x -mF (xdx) xm.
    3484 F (xde + m ) Q = x -mF (xdx) 2 " Q .
    3485 F (xdx) Q = x -mF (xd, - m ).x " Q .
    If U = a + bx + cx ? + & c., then , by (3480),
    3486 F (xdx) U = F (0) a + F ( 1)bx + F ( 2)cx? + & c.
    518 DIFFERENTIAL EQUATIONS.

    3487 F -1(.vd ,) U = F -1(0 ) a + F -1 (1 )bx + F -1 (2)cx? + & c.


    3488 F(xdx, ydy, zd .,...).x"y"=".... = F (m ,n, p ...)x+"y"z"...
    3489x "Una = do(d. - 1) (d . — 2) ... (do- n + 1) u ,
    or, more succinctly , writing D for des
    D (D - 1) ... (D - n + 1) u or D " u ( 2452).
    3490 Otherwise x " Unæ = xd ;(vd: – 1) ... (xd , - n + 1) u .
    Proof. — As in (1770 ). Otherwise, by Induction , differentiating again,
    and remembering that X , = x .

    NOTE. - In the symbolic solution of differential equations,


    we may either employ the operator rd , directly , or the
    operator de after substituting e for a . Formulæ (3480 -5 ) or
    (3474 - 9 ) will be required accordingly .
    3491 { $ ( D ) ero} " Q
    = enre$ ( D + nr) $ ( D + n - 1 .r ) ... $ ( D + r ) Q
    = $ (D ) $ (D — r) $ (D - 2r)... ¢ { D - (n - 1)r } enre Q .
    Proof.- By repeated application of (3475) or (3476 ).

    For ready reference , formulæ (1520, ' 21) are reprinted


    here .
    3492 f (x + h) = ehd:f (x ).
    3493 f(x + h, y + k) = etd,tkd,f(x,y).
    Let do taqx + aza ? ... taną" = f (x ),
    then , denoting do by D ,
    3494 f ( D )uv = uf( D ) v + uof ( D )v + 19f" (D )v + & c.,
    where f'( D ) means that D is to be written for x after differ
    entiating f (x ).
    PROOF. — Expand uv, D . uv, Dº.uv ... D " .uv by Leibnitz's theorem (1460) ;
    multiply the equations respectively by do, aj, a , ... an, and add the results.

    3495 uf (D ) v = f (D ).uv - f”(D )u.v + +5f" D .1290 — & c.


    SYMBOLIC METHODS. 519

    Proof. — Expand wv,,UV29, UV3, ... Uvno by theorem (1472), and proceed as
    in the last .

    3496 F (dze)sin mx = F (- mº)sin mx . OS

    A more general theorem is


    3497 F ( )(uim + u -im) = F ( - mº) (Win + u _im),
    where u and 7 have the meanings assigned below (3499),
    and i = V - 1.

    Theorem . - If $ and denote any algebraic functions of x


    and y , it may be shown, by (3474 ) and (3475 ), that
    3498 % (da + y ) + (x ) = $ (d , + x) 4 (y).
    3499 Let u , or, more definitely , un = ( , y , 2 , ... )" , represent
    a homogeneous function of the nth degree in severable vari
    ables, and let
    3500 = xdx + yd , + zd , + & c .
    Then, by (3480),
    3501 mu = nu , r’u = n'u , nu = n'u, & c.
    3502 Hence F (7 ) u = F (n ) u .

    REDUCTION OF F (T ) TO f ( ).
    3503 Let u be any implicit function of the variables, and
    let 7 = + T2, where 7 , operates only upon a as contained in
    1 , and 7t, only upon x as contained in all , & c . after repetitions
    of the operation n . Then
    3504 Tu = Tu, mu = (1 - 1) TU ,
    3506 tu = (1 - r + 1) ... (17 — 2) ( - 1) TU .
    PROOF. — Tu = (1 - 72) u = mu,
    since , has here no subject to operate upon .
    tu = (1 ^ ^ ,) Tu = (1 - 1) TU ,
    for,nu being of the 1st degree, 7 , and 1 are equivalent as operators. In the
    next step, # , and 2 are equivalent, and so on.

    CoR . When u is a homogeneous function, we have, by


    520 DIFFERENTIAL EQUATIONS.

    (3501), au = nºil,therefore r and n are equivalent operators


    upon u . Hence (3506) may be written
    3507 ou = (n —r + 1)...(1 – 2)(n - 1)nu = n4"}u,
    which is Euler 's theorem of homogeneous functions (1625 ),
    since in that theorem the operator is confined to u .

    3508 As an illustration , let au = (xd , + yd ,) u = 7 , 0,


    then u = ( 1 + 2ưu , + y^ 2 ) 4, TỶu = ( 1 + 2) = = = + T TM.
    Here Tu= (xd , + yd )) (xd, + yd.,) u ,
    the operation being confined to x and y in the second factor (3503), and there
    fore producing (ed. + yd.,) u merely .
    Hence a'u = (a’dz + 2xydxy + y*dzy + xd, + ydy) u, which proves ( 3505).

    If U = U . + 4 , + 1 + ..., a series of homogeneous functions


    of dimensions 0 , 1, 2 , ..., then , by (3502),
    3509 F (T ) U = F (0) u , + F (1) u , + F (2 )uzt ...,
    3510 F -1 (T ) U = F -1 (0 ) u , + F -1 (1) u , + F -1( 2)uz+ ...
    3511 Ex. 1 : a" U = w , + au ta’ug + ...,
    3512 a - * U = Uita- lu , ta- 2u2 + ...

    Ex. 2 : If u have the meaning in (3499),


    3513
    223
    F (*)e* = F (0)1+ F(n)u + F (2n) "* + F(31)7. .3 + &c.,
    and similarly for the inverse operation F - 1 ( ) .
    PROOF. - By (3502) applied to the expansion of the subject by ( 150).
    3514 C (n , m ) un = EXPdpwy 'd, "drz.
    m ! p ! q! ! ... "
    where p + q + r + ... = m , and p ! = 1. 2 ... p.
    Proof. — Equate coefficients of am in the expansion of
    (1 + a )" U = (1 + a) = (1 + a)" ,(1 + a)" .... U ,
    reducing by (3490) .
    3515 The general symbolic solution of the equation
    F (de) u = Q is
    u = F -1(d.) Q + F-1(d.) 0, by (1488 – 90).
    SYMBOLIC METHODS. 521

    3516 The solution of the equation (3238), viz.,


    ynx + a,y(n-1)x + ... + A(n-1)Yx + any = Q .........(1),
    where Q is a function of 2 , is most readily obtained by the
    symbolic method . Thus mi, m2, ... mn being the roots of the
    auxiliary equation in (3239), and A , B , C ... N the numerators
    can
    of the partial fractions into which (m " tamn-1 + ... tan)- Can
    be resolved , the complete primitive will be
    3517
    y = { A (d - m ) - + B (d : - m ,)-2... + N (d . - mm)- } (Q + 0),
    where (d : — m ) - Q = em fe-mx Qdx, (3470 )
    and the whole operation upon zero produces, by (3472), for
    the complementary term ,
    3518 Cemx + Cjemx... + Cjem,d.
    PROOF. — Equation ( 1) may be written
    (dnv + a, din -1)2 + azdın -2)x ... + an) y = Q ,
    (do - m ) (d . - mg) ... (do -mn) y = Q ,
    : . by (3515), y = {(d : – m ) (de - m .) ... (do - mn) } - (Q + 0 ),
    which, by partial fractions, is converted into the formula above.

    If r of the roots mi, m .2, ... are each = : m , those roots give
    rise in (3517) to a single term of the form
    3519 (A + Bd,+ Cd2 ... + Rd.,x) ema | e-m* Q.
    PROOF. - By (1918 ), the r roots equal to m will produce
    { A '(d , - m ) -" + B'(d. - m )->+1 ... + R ' (d m )- } Q ;
    or ( A + Bdx + Cd2e ... + Rd-x) ( do - m ) “" Q .
    3520 But, by (3470), (d.– m )-?Q = (d.– m )"tema e-mxQdx
    = em ſ{e=ne me fe=utQdx}dx = emel e-*tQda, and so on.
    3521 Ex. (1) : 72 – 42 - 572 - 3 = Q .
    Here me - mº 5m - 3 = (m - 3) (m + 1 )”,
    and 1 - 1 1
    " (m – 3)(m + 1): – 16 (m — 3) 16 (m + 1) 4 (m + 1)”
    therefore y = (d , – 3)-'Q - 1 (d ,+ 1)-+Q - 1 (d, + 1)-4Q
    = iteBe Je -33 Qda - iotaje** Qdx – ļe c. (3520)
    3 x
    522 DIFFERENTIAL EQUATIONS.

    3522 Ex.(2): U2x + a’u = Q ,


    therefore u = (d.2x + uº) -' Q .
    Here (mº + a )- = (2ia ) -' {(m - ia ) -' - (m + ia ) -'},
    therefore u = (2ia )-' { (d . - ia)-' Q - (d , + ia )-' Q }
    = (2ia)-'{eiar ſe-iax Qdx - e-iar ſeiar Qdx } (3470)
    = a-lsin ax Scosax Q dx - a- cosax ſsin ax Q da ,
    by the exponential values (766 -7).
    3523 Cor . 1. — The solution of uz + a’ u = ( is
    U = A cos ax + B sin ax.

    3524 Cor. 2. — The solution of Uzz - a?u = 0 is


    u = Aear + Be-ax.
    Change a into ia in the fifth line of (3522), and put Q = 0 .

    3525 When Q is a function whose derivatives of the nth and


    higher orders vanish , proceed as in the following example .
    Ex. (3 ): Uzta’u = (1 +2 )?,
    therefore u = (d2x + a ) - ( 1 + x )' + (dze + a ) - ' 0
    = (a-? - a- d2x + a-6d1e - & c.)(1 + 2x + x*) + (dzę ta’)-10
    = a - ? (1 + x ) ? — 24 - + A cos ax + B sin ax,
    the last two terms by (3523).

    Exceptional Case of the Inverse Process.


    3526 Ex. (4 ): 12x + a’u = cos nx ,
    . . u = (dz + a ) - ( cos nx + 0 ) = 1 (d., ta’) -? (eine + e -inx + 0)
    = } (eint te-inx) ( - nº + a®) - + A cosax + B sin ax by (3474 ) and (3523)
    = cosnx (a' — n ?)-' + & c.
    Now, if n = a, the first term becomes infinite. In such cases proceed
    as follows :
    COS NX - COS (10
    Put A = A' - (a ' — n ) -', and find the value of cos ai - r ? when
    n = a. By ( 1580) it is = ***2a .
    Thus the solution is

    u = & sin ax ,
    " + A ' cos ax + B sin ax .
    2a
    The same result is obtained by making Q = cos ax in the solution of (3522).
    For another example, see (3559 ).
    SYMBOLIC METHODS. 523

    3527 Ex. (5 ): 42 – 9yz + 20y = x®eke,


    therefore y = { (d . - 4 ) (d , - 5 ) } -? x ?el? + { (d . - 4 ) (d . - 5 ) } - 0
    = ett {(d . - 1)(d. - 2)}-\x? + Ae + Best (3475, 3517,3472)
    Now (m2 — 3m + 2 ) - - = _ 3m - mº) -
    2

    Hence the solution becomes


    = + {1+ 3m7m*+ (3m *m*)*+&c.}
    y = *t { } + d . + ]d28 + & c. } x2 + Ae** + Bear
    = { 2012 ,+ 3x + 72 +, Aet? + Best.
    30 .
    = 2T 2 ' 4 )

    3528 Ex. (6 ) : (d. - a )"u = eam,


    therefore u = (d. –a)"# = eur(d.)-*1(3476) = c«* 479 + 1 ). (2149)
    3529 Ex. ( 7) : (d . + a )' y = sin mx,
    therefore y = (( x + a ) - sinmx + (d . + a ) -30
    = (d . - a ) ? (d21 - a ) -? sin me te-ax (dx) -2 0 [by ( 3478) with Q = 0 ]
    = ( - m - a ) - (d - a ) sin mie te-ax (Ax + B ) (by 3496 )
    = (mº + a ”) –? ( – mº sin mc — 2am cos mx + a’ sin mæ ) + e -ar (Ax + B ).

    REDUCTION OF AN INTEGRAL OF THE Nth ORDER.


    71 - 2

    3530 SQ= »– 11 {.:+ſądx =(n=1)****SQxdx


    12 . 00

    N - 3

    + C(n,2)2***| Qu*dx...ESQx*- dx,


    where n - 1 ! = 1 . 2 ... N .
    Proof.— By (3489) d„ Q = e -no(d.- n + 1)(d. - n + 2) ...d,Q .........(1),
    therefore d -nQ = {(d . - n + 1) (d. - n + 2) ...do}-4eme Q
    =nu{(d.-n + 1)-'-(n -1)(d.-n+ 2)
    + C (n , 2 ) (d — + 3) -- - & c.} e" Q (3517 )
    = {e10-1•(1.)-'60– (n - 1)en– (.)-1e* + &c.} Q.
    Then replace e by x .

    The equation
    3531 ax" Ymx + bx ”ynx + & c . = A + BX + Cx2 + & c. = Q
    524 DIFFERENTIAL EQUATIONS .

    may, by (3489), be transformed into


    {a (xdx) + (val .)\" + & c.} y = Q or F (xdx) y = Q .
    The solution is then obtained from
    3532 y = F -1(cd.») Q + F - 1(æd. ) 0 .
    The value of the 1st part is given in (3487).
    3533 If a , b , y, & c . are the roots of F (m ) = 0 , the second
    part gives rise to the arbitrary terms
    C .Va + C228 + & c .
    3534 If a root a is repeated » times, the corresponding
    terms are
    2a{ C, (log x)=-1 + C ,(log x)"-9 + ... + C ;} .
    Proof. — The partial fractions into which F - 1(xd ,) 0 can be resolved , as
    in (3517), are of the type C (all , - m -' 0, m being a root of F (x ) = 0 . But
    (xd, - m ) - ' 0 = Cx " ( 3473 ), O being an arbitrary constant.
    For a root m repeated » times, the typical fraction is C (xd , – m ) - ), P
    being less than r. Now
    (xdı— m ) Cam (log x)P -1 = (d. - m ) Cews7p-1 = em®(d ) COP- (3475) = 0,
    therefore (axdo - m )-P0 = Cæ" (log æ) -!

    The equation
    3535 ayme +byno + &c. = f(eº, sin 0, cos )
    is reducible to the form of (3531) byx = e® ; or, substituting
    from (768), it may be written m
    o
    o
    m

    F (do) y = I ( Amemo),
    and the solution will take the form
    3536 y = EAmeme F - (m ) + F -1 (d ) 0 , .
    for the last term of which the forms in (3533 – 4) are to be
    substituted with a changed to e® .

    3537 Ex. (1) : ZºY2 = ax" + bæ"


    æd : (xd, - 1) (xd1 - 2) y = axm + bæ ” ,
    :: y = {xd , (æd , - 1) ( cdr - 2 )} - (ax " + bx") + {æd, (ædz - 1) (xd2 - 2 ) } - 0
    ахт bu
    = m (m - 1) (m - 2) + n (n - 1)(n - 2) + A + Bx + Cz”,
    by (3180) and (3533) . A result evident by direct integration.
    SYMBOLIC METHODS 525

    3538 Ex. (2) : r *Y2x + 3.xyz + y = (1 - x ) ? By ( 3490)


    {æd (æd - 1 ) + 3xd , + 1 } y = (xdz + 1) ’ y = 1 + 2x + 3x2 + & c.,
    :: y = (xdx + 1)-?(1 + 2.0 + 3x² + ) = (0 + 1)-² + 2 (1 + 1)-2x + 3 (2 + 1)=2x2 +
    (3480) = 1+ + + + &c.+ 4 log2 + B = - log (1 –2) + &c.
    3539 Ex. (3) : 42x + (4x — 1 ) yz + (4x2 – 2x + 2) y = 0 .
    Let a = d , + 2x. Then the equation may be written
    * (1 - 1) y = 0, . . y = {+ (1 - 1)} -'0 = (1 - 1)-'0 — -10.
    Let ( - 1)-10 = u , . ( - 1) u = 0, or uz + (2x - 1) u = 0, :. u = Mex"-4 .
    3540 The solution of a P . D . equation of the type
    at " > + 67 * + & c . = uz + uz + & c .,
    where U7, U2, & c . are homogeneous functions of the 1st , 2nd
    degrees, & c . in a , y , and 1 = xdx + yd , (3503), is analogous to
    (3531), and is obtained from that solution by substituting
    11, U2, & c . for Bx, Ca’ , & c. ; and , for such terms as Cæ", an
    arbitrary homogeneous function of x and y of the same degree.

    3541 Solution of F (Tu = Q ,


    where F (T ) = q * + A471 –1 + A2759 - + An,
    and Q = Untuitu , + & c .,
    a series of homogeneous functions of x, y , 7 , ... of the respec
    tive dimensions 0 , 1 , 2, & c .
    Here u = F -1 (7 ) Q + F -1 ( ) 0 .
    3542 The value of the 1st term is given in (3510) . For the
    general value of the last term (see Proof of 3533) , let
    F (m ) = 0 haye r roots = m ; then
    3543 C (r — m )-P0 = C { u (log x )p -1 + v (log x )? -?... + w } ,
    where u , v , ... w are arbitrary functions of the variables all of
    the degree m .

    3544 CoR. - (1 - m )-10 = (x ,y, ...)",


    that is , a single homogeneous function of the variables of the
    degree m (1620 ).
    526 DIFFERENTIAL EQUATIONS.

    3545 Ex. : 2 *222 + 2xyzzy + y *22, - a (azz tyz, ) + az = Umtune


    Um, Un being homogeneous functions of the mth and nth degrees . The equation
    may be written ( - a7 + a ) z = 4m + un ;
    or,by (3505), (7 - a )(1 - 1) x = Um + Une
    therefore % = { (I - a ) ( - 1) }- ' (Um + un) + { (T - a ) (1 - 1)} - ' 0
    = - m it - un + 0 , + U ,.
    (m - a )(m - 1) ' (n - a ) (n - 1)
    The first two terms by formula (3502) ; the last two terms are arbitrary
    functions of the degrees a and 1 respectively, and result from formula (3543)
    by taking p = l and m = a and 1.

    3546 To reduce a P . D . equation, when possible, to the


    symbolic form
    (II" + A,11"-1 + A,II"-2...+ An)u = Q .........(1),
    where I = Md. + Nd., + & c.,
    and Q , M , N , & c. are any functions of the independent
    variables .
    Consider the case of two independent variables,
    (Md, + Nd,)”u = M²u2x + 2 MNU.y + N21124
    + (MM2 + NM,)w : + (MN, + NN,)Wy... (2).
    Here the form of II is obtainable from the right by con
    sidering the terms involving the highest derivatives only , for
    these terms are algebraically equivalent to (M1, + Nd,) .
    The reduction being effected , and the equation being
    brought to the form of (1) ; then , if the auxiliary equation
    3547 m " + A ,m " - 1 + A2m " -9... + An = 0 ......... ... ( 3)
    have its roots a , b , ... all unequal, the solution of (1 ) will be
    of the form
    3548 u = (11 — a)-+ Q + (II — )- Q + &c.............(4).
    The terms on the right involve the solution of a series of linear
    first order P . D . equations, the first of which is
    3549 Mux + Nu, + ... — au = Q ,
    and the rest involve b, c, & c .
    If equal or imaginary roots occur in the auxiliary equation ,
    we may proceed as in the following example .
    SYMBOLIC JETHODS. 527

    3550 Ex. :
    ( 1 + x2)* 22.7 — 4xy ( 1 + w *) % xy + 4.c?y 22, + 2x (1 + x?) % + 2y (« ? - 1 ) % , ta’ = 0 .
    Here II = ( 1 + x?) d – 2xy dy, and the equation becomes ( II° + a ”) z = 0 .
    Let the variables x, y be now changed to Ę, n, so that II = dy. Therefore,
    since II (E) = 1, II (E) = (1 + x *) &z - 2xyx, = 1.
    1 dx dy
    Therefore, by (3383), i1 + z2
    - 2xy
    from which , by separating the variables and integrating,we obtain
    wʻy + y = A . .. . .. .. . .. . .. . . .. . .. . .. . .. .. . .. . .. . ( 1 ) ,
    and, by (1436 ), & = tan -' x + B ......... ... ..................... (2 ).
    Also , since II (n ) = ne = 0 , (1 + w*) n . - 2xyn, = 0 .
    - dx dy dn
    Therefore
    1 + x2 – 2cy – .
    the solution of which is equation ( 1) . Thus
    & = tan - and n = xạy + y .
    The transformed equation is now (lq + a ") z = 0 ,
    and the solution, by (3523), 1is8
    z = 0 (n ) cos aš + 4 (n ) sin a£,
    arbitrary functions of the variable, which is not explicitly involved, being
    substituted for the constants (3389). Therefore finally,
    x = $ (x y + y ) cos ( a tan - 'x ) + 4 (xảy + y ) sin (a tan -' x ).

    MISCELLANEOUS EXAMPLES.
    3551 Uzx + U2y + 122 = 0 .
    Put d2, + dng = a’. Thus Uzx + a’u = 0, the solution of which, by (3523),
    . is u = q ( y, z) cos ax + 4 ( y, z ) sin ax,
    arbitrary functions of y and Z being put for the constants A and B . Expand
    the sine and cosine by (764 - 5) ; replace a' by its operative equivalent, and,
    in the expansion of sin ax, put ay (y, z) = x ( y,z ) ; thus
    u = $ (3,4)– (1.,+dwu)$ (3,2)+ y(12,+d..)®(y,z) =&c.
    + zx(1,2) (1.99+du»)x ( ) ) + (1.,+d.w) x(1),z)–&c.
    [See ( 3626 ) for another solution.

    3552 wx + u , + u , = xy% .
    Here u = (dx + d , + da) -- (ryz + 0 )
    = {d -o - d -22 (d, + d2) + d -& (d , + d .) — ...} (xyz + 0).
    Operating upon xyz, we get
    u = læ yz - * * * (z + y) + 1gx*,
    528 DIFFERENTIAL EQUATIONS.

    the rest vanishing. For symmetry, take frd of the sum of three such
    expressions; thus
    u = } { 11 (2 * + y * + 24) — ; (aøy + xy + yºz + yz + x*x + zv ) + zxyz(x + y + z )}.
    Operating upon zero , we have, in the first place, d -x0 = (yz)
    instead of a constant, therefore d -270) = xq ( yz ), & c.
    The result is
    { 1 - « (d , + d2) + æ (d ,, + d .) ? ...} (yz ) = e -+(( y+d='p (yz) = ° (y - 2 , 2 — 2 )
    (3493) the complementary term .

    3553 Otherwise, putting d, + d = D , we have, by (3478 ),


    (dz + D ) - xyz = e -xDd _terD xyz = e - & D d .z{ x (1 + x ) ( + 2 )}, (3493)
    = e-xD { ’yz + 128 (y + z ) + *secx *}
    = $ x* ( y - )(2 - x ) + 12 % (y + z 2x) + 12 *,
    which agrees with the former solution.

    3554 aux + buy + cu , = xyz.


    Substitute x = aš, y = bn, x = ct, and the equation becomes
    Ug + wn + ug = abcns,
    which is solved in (3552).

    The samemethods furnish the solution of


    3555
    6666 ) au, + bu, + cu , = @ " y"z".
    3556 xz, + yx, = 2xyda' — .
    Put z = a sin v ,
    :: az = a cos v.tv , :. qv = 2xy, isz = a sin (xy + c).

    3557 axux + byu , + czu , - nu = 0 .


    Put x = 0, y = n , % = 50 ; 1 1 1
    :: Eugtnu + šuz-nu = 0, :. by (3544 ) u = (ză, y ,z )".

    3558 The solution of any P . D . equation of the type


    F (xdx, ydy,zd,, ...) u = £Ax" y"% ! ...
    is,by (3488) and (3557),
    u = _ Aw" y" _
    F (m ,n,p,...) + F (xdx, yd ,, zd.,...) ”.
    SYMBOLIC METHODS. 529

    3559 Ex. : Xux + yu , - au = Qme


    where Qm = ( , y)" (1620 ).
    Here u = (m - a ) -'Qm + Uq. When a = m , this solution becomes inde
    terminate. In that case, as in (3526 ), assume
    Un= Ve - m Rou- a . u = 2,0– 2a + Vo.
    m - a
    Differentiate for a ,by (1580), putting Qa first in the form
    $ {«°F ( %) +245( )} ;
    thus u = } Qm (log x + log y ) + Vm .

    Similarly , the solution of


    3560 XUx + yu , tzuz - mu = Qm
    u = jQm (log æ + log y + log z) + V m .

    3561 XU , + yuy + zu , = c.
    The solution, by (3560) , is
    u = {c (log x + log y + log z ) + Vo.

    3562 % 25 — 2azry + a+22y = 0 or (d . - ad,) % = 0.


    % = (d : - adv) -' 0) = (dx - ady) - ° (y) eardy ( 3472)
    by putting ad , for m and y (y ) for C . The second operation produces , by
    (3476 ), z = eux du { xq (y ) + 4 ( y) } = æq ( y + ax) + + ( y + ax ). (3492 )

    3563 amz — yºzzy + x % , - yz, = 0 .


    This reduces to (æd, + yd,) (ad: - ydy) z = 0.
    Here * = æd + ydyn and m = 0 in (3544 ),
    therefore * = ( ,y) + (2, 7 ),
    the second term being obtained by substituting y - l = y', and so converting
    the second factor into (« d + y'd , ). The above may also be written

    u = F ( -) +f(wy),
    F and f being integral algebraic functions.

    3564 22x - a - Z2y + 2ab % , + 2a+b % , = 0 .


    Patting y = an, this equation is equivalent to
    (d . - dn + 2ab) (d , + dn) 2 = 0 ;
    3 y
    530 DIFFERENTIAL EQUATIONS.

    putting x = log a' and n = log r', this gives, by (3544),


    z = (e", e -n) -200 + (e", e")" = e-2018 0 (e*+") + ų (ext-")
    = e -2abx F (y + ax) + f (y - ax ),
    the functions being algebraic and integral.

    3565 127 — a’uz, = + (x, y).


    :: u = (d», — a*d,,)~?0 (x, y) (3515)
    = (2ad,)-'{(1;- ad,)-+- (d.+ ad,)-1} $ (x,y) (3470)
    = (2ad,)- {cardyſe-azdyo (x,y) da - e-arly ſeardy (x, y)dx} (3470)
    = (2a)" } { ,(2, y + ax) – D.( , y — ax) } dy,
    since qazdy (x, y) = ° ( , y + ax ) (3492).
    Here 0, (x,y ) = ( ° (x ,y — ax )dx + + (y),
    0,(x,y) = 1° (x, y + ax )dx + x (y).
    3566 If q (x , y ) = 0 , the solution therefore becomes
    u = 4 ,(y + ax ) + xı (y — ax ) [Boole,ch. 16.
    For the solution in this case by Monge's method, see
    (3433) .

    3567 % : - az, = em cos ny.


    z = (d, — ad,)-lem cos ny = eurd,ſe-axdyem?cos nydo (3470)
    = eart,ſem cosn (y - ax )dx (3492),and this by Parts, or by (1999), is
    = eurdyemv {m cos n (y — ax ) — an sinn (y — ax )} (m * + a’n )-' + eardyo (y)
    :. z = emt {m cosny - an sin ny } (mº+ a’n”)-' + ° (y + ax ), by (3492 ).
    3568 % - a 2 = 0.
    % = (de- ad.2x) -' 0) = eatdur ( ), by (3472),
    $ (x ) taking the place of the constant C .
    Therefore z = 9 ( ) + at02 + ja*t*9w + & c. (3492 )
    Otherwise, to obtain z in powers of x ,we have, putting b? = a - ,
    22 — b*zt = 0 ,
    :. * = {(d.+ bd})(0,+ ba})}-40 = ehsdiq (t) + e+bxdı (t) (3518);
    then expand by (150).
    3569 22x + 22, = cos nx cosmy.
    SYMBOLIC METHODS. 531

    z = ( d.2x + d2y) - cos nx cosmy.


    Treating day and cosmy as constants, we have, by (3526 ), putting d. , for a’,
    z = cos nx (d2y — nº) -- cosmy + A cos ax + B sin ax, or by (3496 ),
    = cosnæ cos my ( - mº —nº)-' + ( ) cos (xd,) + 4 (y ) sin (æd,),
    A and B becoming • (y ) and + (y ).

    3570 224 + 226 + 224 + a +z = cos (mø + ny).


    Therefore ( tia ) (n - ia ) z = } {ei(mø+n4)te-i(mø+mv)},
    where = = d . + d.y. Therefore, by ( 3510), with x = e*, y = et,
    sr ei(mø+ nyu) e -i(mø+Nvo] ) .
    + (e®, ev ) -ia + (es, evia,
    la - ( m + n ) ' a - (m + n )" S
    = = cos(m + mik? + (e ,ex)-ia + (e“,er)ia.
    3571 Prop. I. - To transform a linear differential equation
    of the form
    (a +bx + cx . .) Une + (a' + b'x + c'a ...) U(n-1) + & c. = Q ...(1)
    into the symbolical form
    f ( D ) u + fi (D ) eºu + f; ( D ) e2u + & c. = T ...... (2 ),
    where Q is a function of x , T a function of 0, x = eº and
    D = de.
    Multiply the equation by x ” ; then the 1st term on the left
    becomes , by ( 3489),
    (a + be® + ce? + ...) D ( D - 1) ... ( D - n + 1) u .
    This reduces, by the repeated application of formula (3476 )
    with the notation of (2451) , to
    3572 aD(n)u + b (D - 1)(n)eu + c (D - 2)(n) e2®u + & c.
    The other terms admit of similar reductions.

    3573 Conversely , to bring back an equation from the sym


    bolic form (2 ) to the ordinary form (1), employ formula (3475)
    so as to transfer em to the left of the operative symbol.

    3574 Ex. : 22(x* + 7x4 , + 5u ) = eo{ D (D - 1 ) + 7D + 5 } u


    = €29 ( D ? + 6D + 5 ) u = e? ( D + 1) ( D + 5 ) u
    = ( 1 - 1)( D + 3 ) e2®4 (3476 ).
    532 DIFFERENTIAL EQUATIONS.

    For the converse reduction , the steps must be retraced, employing (3475).
    See also example (3578 ) .

    3575 Prop. II.— To solve the equation


    utad ( D ) eºu + azd ( D ) $ ( D - 1) 228 u ...
    + a $ ( D ) $ ( D - 1) ... ( D - n + 1) eno u = U ,
    where U is a function of 0.
    By (3491)
    ( D) $ (D - 1) ...0 (D - n + 1) eneu = {0 (D ) eo}" u.
    Putting p*u for this, the equation becomes
    (1 +as + agpa... tang" ) u = U .
    3576 Therefore
    u = {41(1 – )- + A (1 – 120)-... + 1, (1 – 9ne)-1} U ,
    where 91, 92 ... Yn are the roots of the equation
    q " + 0 ,qn- 1 + a291- 2 ... tan = 0 ,
    and A, = 7
    * * ( - ) (9r - 72) ... (9- - 9n)
    The solution willthen be expressed by
    u = Aju, + A U , ... + AnUne
    where U , is given by the solution of the equation
    3577 Ur - 9 ,$ ( D )eu , = U .
    3578 Ex.:
    (x2 + 5x3 + 6x“) Uz + (4x + 25x + 362") u , + (2 + 20x + 362") u = 20x®.
    Putting x = e", and transforming by ( 3489),
    (1 + 5e® + 682 ) D ( D - 1 ) u + (4 + 25e® + 36e ) Du + ( 2 + 20e® + 36e e) u = 20e3 .
    The first term = D (D - 1 ) u + 5 ( D - 1 ) ( D — 2) eºu + 6 (D — 2) ( D - 3) em u
    by applying (3476 ). The other terms similarly ; thus, after rearrangement,
    ( D + 1) ( D + 2 ) u + 5 ( D + 1)’ e®u +6D ( D + 1) e% u = 20e% .
    Operating upon this with { (D + 1 ) ( D + 2) } -\, we get
    u + 5 D + l e .uto 20039
    D + 2°
    (3 + 1) (3 + 2) = ell , by (3474 ) ;
    or (1 + 5p + 6pº) u = es , if p = ( D + 1)( D + 2) -? e° ;
    therefore u = { 3 ( 1 + 3p) -- — 2 ( 1 + 2p) -1} 13 = 3y - 22,
    if y = (1 + 3p) - €39 and z = (1 + 2p) ?<3 .
    SYMBOLIC METHODS. 533

    Hence ( 1 + 3p ) y = e30 ory + 3 ( D + 1 ) ( D + 2 ) -le®y = 230;


    therefore ( D + 2) y + 3 ( D + 1) e®y = 23 ( 3 + 2), by (3474 ),
    or ( D + 2) y + 3e' (D + 2) y = 5e39, by (3475) ;
    that is , ( x + 3x ) Yz + 2 ( 1 + 3x ) y = 5x®.
    Similarly ( + 2°) ( + 2 ( 1 + 2a ) a = 5c”.
    Solve these by (3210 ),and substitute in u = 3y – 2z.

    3579 Prop . III. — To transform the equation


    u + $ ( D ) erou = U into v + $ ( D + n ) er® v = V ,
    put u = eno ny and U = eno V .
    PROOF. — By (3474), because $ ( D ) e(n+r)• v = eno o ( D + n ) emov .

    3580 PROP. IV . – To transform the equation

    u + Ⓡ (D ) ero u = U into v + y (D ) enov = V ,


    put u = P ( v and U = P (D) v,
    3581 where p. ° (D ) – ° (D ) ° (D — Y) ° ( D — 2r)...
    try ( D ) * (D ) * ( D - r ) * ( D — 2r)
    PROOF. - Put u = f ( D ) v in the 1st equation , and ero f ( D ) v = f ( D - r) enov
    (3476 ). After operating with f -? ( D ) it becomes
    0 + $ ( D ) f (D - r ) f-' ( D ) e*ov = f-1 (DU,
    therefore 0 ( D ) f ( D - r ) f -' ( D ) = y ( D ) by hypothesis ;
    therefore f(D) = (D) f(D - r) = P (D) $ (D < ») f(D – 2r),
    and so in inf. Also U = f ( D ) V .

    3582 To make any elementary factor x ( D ) of $ (D ) be


    come, in the transformed equation , x ( D + nr), where r is an
    integer ; take 4 ( D ) = x (DÎnr) xi(D ). See example (3589).

    3583 To make any factor of ° (D) of the form


    2 x (DEnr)
    disappear in the transformed equation, take + ( D ) = x1( D ),
    where xı(D ), in each case, denotes the remaining factors of
    ( D ). See example (3591) .
    534 DIFFERENTIAL EQUATIONS.

    3584 In the application of Proposition IV ., differentia


    tion or integration will be the last operation according as
    P, 9 (3581) has its factors , after reduction , in the
    numerator or denominator, and therefore according as 4 ( D )
    is formed by algebraically diminishing or increasing the several
    factors of $ ( D ) . However, by first employing Proposition III.,
    the given equation may frequently be so prepared that the
    final operation with Prop. IV . shall be differentiation only .
    See example (1 ).
    For further investigation , see Boole's Diff. Eq., Ch . 17, and Supplement,
    p . 187.

    3585 To reduce an equation of the homogeneous class


    (3531) to a binomial equation of the same order of the form
    Ynx + y = X .
    The general theory of such solutions is as follows. Let
    the given equation be
    u + q {(D + a )( D + a.) ...( D + a,)} -le®u = U ... (1),
    A1, A2, ... an being in descending order of magnitude. Putting
    u = e -a18 v, by Prop. III.,
    v + q { D ( D - az - az) ... ( D – a , –an) }-4en0v = e«® U ... (2).
    To transform these factors , regarded as ¢ ( D ), by Prop. IV .
    into 4 ( D ) = D ( D - 1 ) ... ( D - n + 1 ) , we convert D into Dtrn
    (3582), r being an integer.
    Hence for the pth factor wemust have
    D + rn — az + ap = D - p + 1,
    3586 and therefore a , - a , = rn + p - 1 ................ (3) .
    If this relation holds for each of the constants a , . .. an,
    equation ( 1) is reducible to the form
    3587 y + q { D ( D - 1) ... ( D - n + 1)} - enøy = Y ...... (4 ),
    which , by (3489), is equivalent to Ynx + qy = Ynx = X .
    y being found in terms of x from the last equation , and,
    v being = P. ,( y (3580), the solution will result from
    L
    3588 ur e-ab
    øp+ " ((DD -- 1a,) (+D a ?)
    — 2 ) ... ( D - n + 1) .
    ... (D - a , tan) :
    while U and Y are connected by the same relation as u and y .
    SYMBOLIC METHODS. 535

    3589 Ex. 1 : Given æ°Uz + 18.rºuzx + 84xur + 964 + 3x”u = 0 .


    Putting x = e® and employing (3489), this becomes
    { D ( D - 1) ( D — 2 ) + 18D ( D - 1) + 84D + 96 } u + 343 u = 0,
    or ( D + 8 ) ( D + 4 ) ( D + 3 ) u + 3e3'u = 0 ,
    therefore u + 3 { ( D + 8 ) ( D + 4 ) ( D + 3)} -% e% u = 0 ..................(1) .
    Employing Prop. III., put u = e -8v,
    therefore ( 3476) v + 3 { D ( D — 4 ) ( D - 5 ) } - e30 v = 0 ........
    To transform this by Prop. IV. into
    we bave
    y + 3 {D (D - 1) (D - 2)}-le® y = 0 ....................(3),
    . O ( D ) _ D ( D - 1) ( D — 2 ) ( D - 3 ) ( D — 4 ) ( D — 5 ) ... = ( 1 - 1) 01 - 2) .
    84 (D ) D (D — 4 )(D - 5)( D - 3)(D - 7)(D - 8) ...
    . . v = (D - 1) ( D - 2 ) y, . u = e- (D - 1) ( D - 2 ) y .........(4 ),
    and the solution is obtained by differentiation only, performed on the value
    of y as obtained by the solution of ( 3), that equation being equivalent to
    D (D - 1) ( D — 2 ) y + 3e3º y = 0, or, by ( 3489), 4x + 3y = 0.
    If, however, Prop. IV .were used to pass directly from (1) to (3), we
    should have
    Dº ( D ) _ D (D - 1) (D - 2)( D - 3) (D — 4 )(D - 5) ...
    +(D) ( D + 8 )(D + 4 )( D + 3)( D + 5) (D + 1) D ...
    ( D + 8) ( D + 5 ) (D + 4 )(D + 3)( D + 2) ( D + 1)'
    and equation (4 ) would involve integrations of y as high as D -6y .

    3590 NOTE. - By the literal application of Rule IV ., the right side of


    equation (3) ought to be V = { ( D - 1 ) ( D - 2 ) } -10 ; but no such term is
    required when the original and transformed equations are of the same order,
    for in such cases the arbitrary constants introduced by the operation upon
    zero disappear with the terms containing them in the final differentiation .
    The result is the sameas if the operation upon zero had not been performed.
    In the following example, V has to be retained .

    3591 Ex. 2 : (x – 23) Uz + (2 - 12x ) Uz - 30xu = 0 ... .............. ... ... (1 ).


    Multiply by x , transform by ( 3489), and remove e2 to the right of each
    function of D by (3476 ), thus
    ( D + 4 ) ( D + 3 ) c2u = 0 .........................(2 ).
    D ( D + 1)
    Transform this by Prop. IV . into
    v- D1980= V ..........................(8).
    We have u = P, D +40 = (D + 4)(D + 2)v,
    V = { (D + 4 ) ( D + 2 ) }- 10 = Ae-20 + Be- 4 (3518 ) .
    The operation upon zero is required in this example (see 3590 ), because (3 )
    536 DIFFERENTIAL EQUATIONS.

    is of a lower order than (2) ; but only one term of the result need be retained ,
    because only one additional constant is wanted . Hence (3 ) becomes
    ( D + 1) v - ( D + 3 ) e2 v = ( D + 1 ) Ae- 2 = – Ae-20.
    Changing again to x , this equation becomes
    (28 — Qº) v, - 4 « *v + A = 0 .
    The value of v obtained from this by (3210 ) will contain two arbitrary con .
    stants . The solution of (1 ) will then be given by
    u = ( D + 4) (D + 2) v.

    3592 Ex. 3 : U22 - n (n + 1)x -?u — q?u = 0 , [Boole, p. 424.


    n being an integer.
    Multiplying by w and employing (3489), this becomes
    u - q° {( D + n ) ( D - n - 1) } -% e u = 0 .
    This is changed by Prop. III. into
    v - q* { D ( D - 2n - 1) } -1200 = 0, with u = e-noy ,
    and this, by Prop. IV ., into
    y - 2° { D ( D - 1) }-% e2 y = 0 or yze - *y = 0 (3489).
    y being found from this by (3524 ), we then have
    - D - 1
    * D - 2n -Gy
    u = e - no P ,
    19= e -n ( D - 1 ) ^2y = 22- " (xdx - 1) "? y .
    But, by (3484), F (xd.— m ) = 2" F (æda)x-",
    i u = ~ =" ® (ædz) x - 1.28 (æd .) 2 - 3 ... 22 - (æd,) x - 3n+ 1 y ,
    or u = x =(n+1) (x®d )" 23–2n +ly
    = x -* - ' (2®dx)" x –2n +1 (Ae?? + Be+9+) (3525).
    This may be evaluated by substituting % = x - % (See Educ. Times Reprint,
    Vol. xvii., p. 77.)

    3593 Ex. 4 : 124 - a uz, - n (n + 1) 2 -2 u = 0.


    The solution is derived from that of Example (2 ), by putting q = ad,,
    and arbitrary functions of y after the exponentials instead of A and B ; thus
    u = -* -* (x®d,)” æ –2n +1 { eard , q (y) te ard of (y) }
    = x -n- 1 (x8d -)" –20 + {9 (y + ax ) + + (y + ax )}, by (3492).
    [Boole, p . 425.

    3594 (1 + ax") Uzx + axu, In’u = 0 .


    To solve this equation or its symbolical equivalent obtained
    by (3489), viz .,
    3595 „ 1 a (D - 2)* + nºe? u = 0.
    ut
    U Ta D ( D - 1)
    ſ dx
    Substitute t = J V (1 + a2") in the solution of Wz + n u = 0 , by (3523 -4 ).
    SYMBOLIO METHODS. 537

    3596 Similarly , to solve the equation


    ( * + a ) cºup + ( 2 * a ) cu, nºu = 0,
    or,the same in its symbolical form ,
    3597 W + ( D - 1) ( D - 2) 2 u. = 0 .
    T aD ? + n?
    ſ dx
    Substitute t = Je1 v (v + a ) in the solution of Uz En u = 0 .
    (3596 ) is obtainable from (3593) by changing 0 into - 0 .

    3598 Pfaff's equation ,


    (a + bw *) x*uz: + (c + ex *) xu, + (f + gw*) u = Q .
    When Q = 0, the symbolical form becomes
    u. +, 6b (( b
    D - n ) ( D - 1 - 1) + e ( D - n ) + 9 ho
    aD ( D - 1) + cD + f
    If n be not = 2, substitute 20' = no, and therefore 2de = nder.
    3599 Thus (D - a ) ( D — a ) e20' = 0 ....... ............. (2) .
    a ( D - B .) ( D - B ,)
    wbere a , a, are the roots of the equation
    b ( ina - n ) (Ina - n - 1) + e (fna - n ) + g = 0 .........
    and Bu B , are the roots of
    a inß ( inß - 1) + c inß + f = 0 .
    Four cases occor
    3600 I. - If az - a, and B . - B, are odd integers, (2) can be reduced by
    . ( D - a ) ( D - 4, - 1) 220'y = 0 .
    Prop . IV . (3581) to the form utº!
    Ta ( D - B.) (D - B , - 1)
    and then resolved into two equations of the first order.
    3601 II. - If any one of the four quantities a , - B1, 2 , - 22, a , - Bi, (1, - B ,
    is an even integer, ( 2 ) can be reduced by Prop. IV . to an equation of the
    first order .
    3602 III. - If B , - 6 , and a , tuz - B - B , are both odd integers, then, by
    Props. III. and IV ., (2 ) is reducible to (3595 ) .
    3603 IV . - If a, - ag and a , tag- B4 - B , are both odd integers, ( 2 ) is
    reducible in like manner to (3597 ) . [ Boole, p . 428 .
    NOTE . — The integers may be either positive or negative, and when even
    may be zero.

    32
    538 DIFFERENTIAL EQUATIONS.

    SOLUTION OF LINEAR DIFFERENTIAL EQUATIONS


    BY SERIES .

    3604 CASE I. - Solution of the linear differential equation


    f. ( D ) u - fi (D ) eru = 0 or f.(xdx) u - fi (xdx) w "u = 0,
    in which f. ( D ), fi ( D ) are polynomial expressions of the form
    2o + a , D + a , D ... taxD" and . ( D ) = ( D - a ) ( D - 6) (D - c ) ....
    3605 Let ” (D ) = fi( D ) - fo (D ), and let
    3606 0 (a ) = 1 + + (a + r ) x * + ° (a + 2r) • (a + r) 2220
    + + (a + 3 ) + (a + 2 ) + (a + r) ºr + & c.
    Then the solution will be
    3607 u = Ava (a ) + Bx' $ (6 ) + Cx°° (c) + & c .
    Proof
    OF
    . — Operating with f -' (D ) and writing p for y (D ) ero,
    u - pu = f ' ( D ) 0 = Aeas + Bebe + & c . (3518)
    Therefore, by (3515 ), u = ( 1 - P )- (Aca® + Bebe + ...)
    = ( 1 + p + pº + ...) Aeaº + ( 1 + p + p + ...) Bebe + & c .
    Now in each term substitute for p" the value in (3491) , and remove D by
    formula ( 3474 ) .

    Case II. - Solution of


    3608 f.(D)u + fi(D) eºu + f2(D)e 'u ... + fn (D ) enou = 0
    ...... (1)
    where fo (D) = (D — a)(D — b)( D — c) ...
    Let ¥ (a) = 1 + F; (a + 1) x + F , (a + 2 )x2 + & c .,
    where the coefficients Fi(a + 1) or v , F , (a + 2) or va, & c. are
    determined in succession by the formula
    3609 f.(m ) vm + fi(m ) ºm –1... + fr (m ) Um -n = 0 and v = 1
    m - 1 . . in

    ......(2)
    The solution will then be expressed by
    3610 u = Axa¥ (a) + Bx"¥ (6) + C¥°¥ (c) + & c.
    Proof. — From ( 1)
    u = {1 + 0.(D ) e'... + ®, (D ) ene}--f: '(D ) .0 ...............(3),
    where Pr(D ) = fr ( D ) = f. (D ).
    Here fo ?( D ) 0 = {( D - a )( D - 6 ) ... } -10 = Aeao + Bebe + ... (3518 ) ;
    and { 1 + 4, (D ) eº ... + Pn ( D ) eno }- ' = 1 + F, ( D) eº + F , (D ) 62° + ...
    SOLUTION OF LINEAR DIFFERENTIAL EQUATIONS. 539

    To determine F , F ,, & c ., operate upon each side with { 1 + ( D ) e® + & c. } ,


    and equate coefficients of powers of e ; thus formula ( 2 ) is obtained . (3 )
    now becomes
    u = { 1 + F, ( D ) cº + F,( D ) 2 + ... } (Aca® + Bebe + .... . ...... (4 ).
    Multiply out ; apply (3474), and put æ for e®.

    3611 Ex.: 2°42 (a + b - 1)æus + abu — qu’u = 0,


    or, by (3489) , ( D - a ) ( D - b ) 4 — qe2® u = 0.
    Here fo ( D ) = ( D - a ) ( D - 6 ), fi ( D ) = 0, $2 ( D ) = - 9 .
    Therefore ( 2 ) becomes (m - a )(m — b ) vm = qum -2
    therefore Fi, Fs, & c. vanish , and F. (a) = 1,
    _ qF , (a )
    F , (a + 2) =
    (a + 2 - a )(a + 2 – 0) 2 (a + 2 - 6 )
    9F , (a + 2 )
    F . (a + 4 ) = (a + 4 - a ) ( a + 4 - 6 ) 4 . 2 ( a + 4 - 6) (a + 2 – 6 ) .
    are:
    Therefore ¥ (a ) = 1 +' 2 (a + b - 2 2 (a - b + 4 ) (a - b + 2 )
    Similarly we find F , (6 + 2 ), F.(6 + 2), & c., and thence ¥ (b ); and, sub
    stituting in ( 3610 ), we have
    u
    1. Agra +2 Aq*xa+ 4
    + 2 (a +-6 — 2) * 4.2 (a − 6 + 4 )(a − 6 + 2) * **
    Bqw6+2 Bq*xb+ 4
    + Brot od
    6° + 2 (b = a + 2) + 4.2 (b - a + 4 )(0 – a + 2) +...
    3612 The solution is arrived at more quickly by formula (3607). We
    have
    ° (D) = (D-a}(D -b)'
    :: 9 (2 + 2) = 2(at2- 1) (a + 4) = 4 (a + 4 - 0)» & c.,
    producing the same series by the value of 0 (a ). Similarly with • (b ).

    3613 When fo ( D ) has r factors each = D - a , the corres


    ponding part of the value of u in equation (4 ) will produce
    3614 A , + A , log x + A ; (log x )?... + Ar-1(log x )=-1,
    where the coefficients A , A1, ... are each of the form Im

    Coxa + C224+1 + C2x4+ + ...


    3615 But if any one of the quantities F , (a + r ) = 0 (3608),
    then Cr = 0 also.
    540 DIFFERENTIAL EQUATIONS.

    Proof. - f= ' ( D ) now contains a term of the form


    *** (co +40 ... + 6282-1) = {@ºv, say.
    The corresponding part of u in (4 ) is
    { 1 + F,(D ) eº + ... }
    = { eme tel@ +1)°F ( D + a + 1) + ere+2)®F, (D + a + 2) + ... Jo by (3475 ).
    Expand each function F by Taylor's theorem in powers of D ,operate upon v,
    and arrange the result according to powers of 0 .

    In practice , proceed as in the following example.


    3616 Ex.: XUzr + ux + qʻxu = 0 .
    Multiplying by x and changing by (3489), this becomes
    D 'u + q*e2® u = 0 . D 'u = 0 gives u = A + BO.
    Substitute this value and operate with D , considering A and B as variables,
    and equate to zero the coefficients of the powers of 0 ; thus
    D' A + q*e2º A + 2DB = 0, D ' B + 9 e2B = 0.
    Then change D into m , and erº A into am -r, to obtain the relations
    miam + q*am -2 + 2mbm = 0 ; m 'bm + q*bm -3 = 0 ,
    which determine the constants successively in terms of Q , and be (which are
    arbitrary ) in the equation
    u = ag + agwoo + * + ... + log (bo + 673+ 672* + ...),
    which thus becomes the solution sought. [ Boole, Diff. Eq., p . 439.

    SOLUTION BY DEFINITE INTEGRALS. *

    3617 Laplace's method .— The solution of the equation


    xº (dz) u + 4 (dz) u = 0 .................. (1)
    u = c [ {e***) ede(pt)-1} dt ............... (2),
    the limits being determined by
    estuser endt = 0 ............... .... (3 ).
    PROOF. — Assume u = e t, and substitute in (1) , patting o (d .) ett
    = q (t) ezt (3474), thus
    (20*9°(t)Tat+ſ4*4(+) Tdt = 0.
    . This method of solution is merely indicated here, and the reader is referred to Boolo's
    Dif . Eq., Ch. xviii., for a complete investigation.
    SOLUTION OF DEFINITE INTEGRALS. 541

    Integrating the first term by parts, this becomes


    ****(t)T ext[de{$ (t) T} -4 (1)T) dt = 0 .........(4),
    an equation which is satisfied by equating each term to zero. The second
    term thus produces a value of T by integration by (3209) , and this value
    substituted in the first term , and in the value of u , gives the results ( 3 )
    and (2 ) .

    3618 Ex. (1) : « Uz: + aur - q*xu = ( * ... ................. (4 ).


    Here o (do) = 0,8 - 0 , 4 (da) = ada, y (t) = ť – q , 4 (t) = at. Hence
    (2) and ( 3) become
    u = 0 ( e* (6 — )* 'dt; ent(7 – 9") = 0;
    a being positive, the limits are t = = 9, and, putting t = q cos 0, we find
    u = 0 <=cososine-10de ........................(5).
    3619 The solution in series by (3608) is as follows. Equations ( 1) and
    (2 ) of that article are in this case
    D (D + a - 1) u — q*e2®u = 0 and m (m + a - 1 ) Um - q*0m -2 = 0.
    Thus, a in (3608) = 0, and b = 1 - a. Therefore (3610) becomes
    v = 4 {1+ zake 15+ 2.412.497%( +3)+ &c.}
    +Bs {1+ 2leta + 2.4634 (5–a)+&c.}.....(C).
    Both series are convergent by (239 ii.).

    The results deduced by Boole are these


    3620 (5 ) is equivalent to the particular integral represented by the first
    series of (6 ).

    3621 A second particular integral, by assuming u = ell-ajov, is found to


    be, when 2 - a is positive,
    u = C,xl-e Jo"eracoso sind-code ...................(7)
    3622 When a lies between 0 and 2,the complete integral is
    u = 0,ſcazcºsino-1020+ 0, -0flerez coosinº-code.......(8).
    • The method by definite integrals is elucidated by Boole chiefly in the solution of this
    important equation .
    542 DIFFERENTIAL EQUATIONS.

    3623 But, if a = 1, the solution becomes


    u = 1*erocomo { A + Blog (x sin°6)} do ... ....(9).
    3624 If a does not lie between 0 and 2 , then, if a be negative, put
    a = á — 2n , and replace the first term of (8 ) by
    C (æd,+ a'- 1)(xde+ a' — 3) ... (æd ,+a'— 2n + 1) ["exscoso sin@=1d0 ...(10),
    the transformation being effected by (3580).
    3625 And if a be positive and > 2, put u = ell –a)ev = x1-av. This con
    verts a into 2 - a , a negative quantity , and the case is reduced to the last
    one.

    3626 Ex. (2 ). — To solve by this method the P . D . equation


    Wax + U2y + U2 = 0 ........................ (1 )
    when r = (22 + y?).
    Eliminating x and y, (1) becomes
    ray + 4 + ruz = 0 ..... ....... (2 ).
    Now the solution of this equation is number ( 9 ) of Example ( 1 ), if we
    change x into r, q into ide, and A and B into arbitrary functions of % . We
    thus obtain

    u = 1*ercosoid:{P(x) + 4 (2) log (rsin’8)} do............(3),


    or, by (3492),
    u = cos * (z + ir cos 0) log (r sin . 6) d0......(4 ).
    See ( 3551) for another solution.
    3627 If u be the potential of an attracting mass at an external point,
    and if u = F (x ) when r = 0 ; then, since log r = 00 , 4 (2 ) must vanish ;
    therefore
    F (-) = *°(2)d6= #$(2).
    Hence(4) reduces to u = ? ("F {u+ir cose} do..
    Parseval's Theorem .
    3628 If, for all values ofU ,
    A + Bu + Cu? + ... = $ (u )
    and A ' + B 'r -1 + CP2 - + ... = * (u ) ......... ... ( 1),
    DIFFERENTIAL RESOLVENTS. 543
    then

    AA'+ BB +...=21"[*(**)*(*)+$(= )4( - )]do.


    PROOF. – Form the product of equations ( 1), and in it put u = eie and e -io
    separately, and add the results. Multiply by do, integrate from 0 to 7 , and
    divide by 27 .

    P . D . EQUATIONS WITH MORE THAN TWO


    INDEPENDENT VARIABLES.

    3629 By means of Fourier's theorem (2742), the solution


    of the equation
    194 — ha(lize + 12y + 122) = 0
    may be deduced by a general method in the form
    A +

    u = (1 + 2 ) TTTTTTei(4+ Bht) ys(a,b,c)da dbdcdădudv,


    the limits of each integration being – o to on , and the func
    tion 4 being arbitrary and different in the two terms arising
    from the operator (1 + dt) .
    Boole, Ch . xviii., and more fully in Cauchy's Exercice d 'Analyse Mathé
    matique, Tom . I., pp. 53 et 178 .

    3630 Poisson's solution of the sameequation in the form of


    a double integral is
    C # 127
    u = (1 + dt) t sin & (x + htsingsinn, y + htsinĞ cos n, % + ht cosE)dedn
    with the same latitude in the interpretation of 4 .
    [Gregory's Examples, p . 504 .

    DIFFERENTIAL RESOLVENTS OF ALGEBRAIC


    EQUATIONS.

    3631 THEOREM I. (Boole). — “ If y ,Y2... Yn are the n roots


    of the equation 11

    y — ayn-1 + 1 = 0 ..... ......... ..... (1),


    and if the mth power ofany one of these roots be represented
    Owe e es
    544 DIFFERENTIAL EQUATIONS.

    by U , and if a = eº, then u as a function of 0 satisfies the


    differential equation

    - -(6. D + m. –1)* "(0 - -1)[D" ]="e u = 0,


    and the complete integral of the same will be
    u = Cym +Cym + ... + Cnym.
    3632 “ Cor. I. — If m = - 1 and if n be > 2 , the differential
    equation o
    r
    O

    D6 -12 – 1n (. -10-
    n 1 -1) ="ex®u = 0
    has for its general integral
    u = Ciyi? Czyż ... + Cn-19n-
    Y ,Y ... Yn–1 being any n – 1 roots of (1).
    “ If O be changed into – 0 , and therefore D into - D , the
    above results are modified as follows :
    3633 “ Cor. II. — The differential equation
    (n) [ in - 1 m ) (n - 1) m7- 1
    U - (D " I enou = 0
    In

    has for its complete integral


    u = C141 + Czym ... + Cnymi
    Yu, Ya ... Yn being the roots of the equation
    ay " — yn- ita = 0 ...

    3634 “ Cor. III.— The differential equation


    u -n (0 –2)6*-» [(* 7+ D + * ]*e u = 0,
    supposing n > 2 has for its complete integral
    u = C1y1 + Czyż+... + Cm-147 1,
    91, Ya ... Yn-1 being any n - 1 roots of (2).
    DIFFERENTIAL RESOLVENTS. 545

    “ Theorem II. (Harley).


    3635 “ The differential equation
    af(vd.)”u -("7".ed.+ –1) *( ...- *. 1)","u
    = 0
    ne mth power of any root of the equation
    y " — xy " - " ta = 0 ,
    u being considered as a function of x .
    3636 “ Cor.- The differential equation
    *(" :" d_- ) ”(xd) ºw
    -(n - 1)" ( xd. - * - 1)"x"u = 0
    is satisfied by the mth power of any root of the equation
    y" — ny" -" + (n - 1) x = 0 ."
    [ Boole, Diff. Eq., Sup. 191-- 199.
    3637 See also Boole, Phil. Trans., 1864 ; Harley, Proc. of the Lit. and
    Phil. Soc. of Manchester, Vol. II. ; Rawson , Proc . of the Lond . Math . Soc .,
    Vol. 9 .

    4 A
    CALCULUS OF FINITE
    DIFFERENCES.
    -- -

    INTRODUCTION .

    3701 In this branch of pure mathematics a function ( )


    is denoted by Ux, and ( x + h ) consequently by 1 .ethe The
    increment h is commonly unity . If Ax denotes the increment
    h , and All the consequent increase in the value of Ux, we have
    3702 Aux = Ux + 0.x — Ux•

    3703 When Ax diminishes without limit, the value of


    + AC

    que, or " r+ , ", is dus


    3704 The repetition of the operation A is indicated as
    follows :
    AAU, = A ’ux, JA’u , = A'uz, and so on.
    3705 Ex. : Let uz = x*,
    x = 1 2 3 4 5
    x = 1 4 9 16 25
    Az* = 3 5 7 9 ...
    A ’r? = 2 2 2 ... ...

    FORMULÆ FOR FIRST AND 7th DIFFERENCES.


    If Uz = ax" + bain - 1 + cx " - 2 + & c.,
    3706 A ”un = an (n - 1) ... (n - r + 1) an-* + czan-r -1 + & c.,
    (

    3707 A Uz = an (n - 1) ... 3 . 2 .1.


    FORMULÆ FOR FIRST AND nth DIFFERENCES. 547

    3708 Hence the nth difference of a rational integral func


    tion of the nth degree is constant.
    3709 So also A " q" = 1. 2 . 3 ... n .
    3710 NOTATION . - Factorial terms are denoted as follows :
    UxUx-1 ... Ux- m + 1 = um),
    3711 - su - m )
    UzUx+ 1 • Ux +m - 1
    3712 Thus ~ (~ — 1)...(x –m + 1) = qım),
    3713 Sal -m ).
    æ (x + 1) ... (a + m - 1 )
    Hence m , m !, and m (m) are equivalent symbols.
    3714 According to (2452), 2 (m ) would here be denoted by a . . The
    suffix, however, being omitted , it may be understood that the common differ
    ence of the factors is always – 1.

    3715 Av(m ) = mx(m - 1), A ” v(m) = m (n)qim -n), A ” x (m) = m (m),


    and, if m < n , A *a (m ) = 0 , since Ac = 0 if c = constant.
    3718 Axl-m) = -mvl-m-1), A" vl-m) = (- m )(n)wol m-n).
    3720 Aum) = (Ux+1 - Uz- m +1) ukm -1),
    Auk-m ) = (uz - Ux+m )uk-m -1).
    3722 Ex. :
    A (ax + b)(m) = am (ax + b)/m -1), A (ax + b)(-m) = -am (ac+ 6)(-m-1).
    (m - 1) W + 1

    3724 Alogus = log {1+ 44 }, Alogusm -1) = log2


    Ur - m + 1

    3726 Sa® = (a − 1)a", A" ** = (a” – 1)"a**.


    3728 4"cus(ax+b)= (2 sing)"con {ax +b+ n(Q+ )}.
    Proor.–
    OF.
    A sin (ax + b) = sin (ax + a + b) – sin (ax + b )
    = 2 sin sin(ax+b + at.").
    That is, A is equivalent to adding a7 * to the angle and multiplying the
    sine by 2 sin
    548 CALCULUS OF FINITE DIFFERENCES.

    3729 Conversely , the same formula holds if the sign of n


    be changed throughout.

    EXPANSION BY FACTORIALS.
    3730 If A " (0) denote the value of Aạo(@ ) when x = 0,
    then $ (x) = $ (0) + 44 (0) x + 4°0 (0)2:02 + 420 603 2049 + &c.
    3731 If Ax = h instead of unity , the same expansion holds
    good if for A " ° (0 ) wewrite ( A ” ° (X ) = h ")x=0 ; that is, making
    x = 0 after reduction .
    Proof. — Assume 0 (x ) = ao + a ,x + az2(?) + 2z2 (3) + & c.
    Compute Aç (2 ), A *P (x ), & c., and put x = 0 to determine do, aq, Ag, & c .

    GENERATING FUNCTIONS.
    3732 If unta be the general term in the expansion of $ ( t) ,
    then o (t) is called the generating function of ur or ( t) = Gug
    Ex. : (1 - t)- 2 = G (x + 1), for æ + 1 is the coefficient of fit in the
    expansion .
    3733 Gur= $(1), Guzti = P(0),...Gusto = 0(0)
    C + 1

    3734 Gau; = (1, -1)*(),... GAⓇU. = (1, -1)468).


    PROOF. GAUX = Guz+1 - Guz, & c.

    THE OPERATIONS E, A, AND do


    3735 E denotes the operation of increasing x by unity,
    Eu, = Ux+1 = uz + au = (1 + 1 )Uz.
    The symbols E and A both follow the laws of distribution ,
    commutation , and repetition (1488 – 90 ) .

    3736 E = l + A = ex ore.*
    PROOF. - Eur = Uz+1 = uz + d. Ux + 1d2zUzt duruz + & c .

    = (1 + dz + 1d2x + o d3x + & c.) uz = e" U z.


    By ( 1520 ) , Ax being unity.
    * The letter d is reserved as a symbol of differentiation only, and the suffix attached to
    it indicates the independent variable . See (1487) .
    FORMULÆ FOR FIRST AND nth DIFFERENCES. 549

    3737 Hence A = e - l and D = log E .


    3739 Consistently with (3735) E -1 denotes the diminishing
    a by unity ; thus E -10 = Ux-1.
    For Eug-1 = Uxe : : Ux-1 = E -'u go

    Uz+ n in termsof u , and successive differences.


    3740 Vx+u = ur + 1Au, + C (n , 2 ) A ’u , + C (n , 3 ) A ’ux + & c.
    Proof. — (i.) By induction, or (ii.) by generating functions, or (iii.) by
    the symbolic law :
    (ii.) Gwin = (*)**(+) = {1+ (1 -1) }">(t).
    Expand by the Binomial theorem , and apply (3734 ).
    (iii.) Ww+n = E " u : = (1 + A )"U z .
    Apply the laws in (3735 ) by expanding the binomial and distributing the
    operation upon Uz.

    Conversely to express A "un in terms of urg Ux+1, Ux+ 2, & c .


    3741 A ” u = Uz+n — nux+n -1 + C (1 , 2) Ux+n -2 ... ( - 1)" U z.
    PROOF. - Apur = ( E - 1 )" 12 (3736 ).
    Expand, and apply (3735) as before. Putting x = 0, we also have
    3742 A "U , = U , – nun-1 + Cn, Un-2 ... ( - 1)" Uo.
    3743 A " " = (x + n )" — n (x + n - 1)"
    + C (n , 2 ) (x + n — 2)" — & c.
    3744 A ”Om = nm — n (n - 1)" + C (n , 2)(n - 2)"
    - C (n , 3 ) (n - 3)” + & c .
    3745 Ex. : By (3717 ) A ”On = n ! Hence a proof of theorem (285 ) is
    obtained .

    3746 A " U _Y2 = (EE' — 1) " u V.x,


    where E operates only upon u , and E ' only upon vr.
    Proof. AuxVx = Ux+1Vx+1 — U , V , = Eur. E 'v - 4 ,44 = (EE – 1) u ,Vr.

    Applications of (3746 ) .
    3747 Ex. ( 1 ) : A ”UxVx = ( - 1 )" ( 1 - EL”) ” U Vxo
    Expand the binomial, and operate upon the subjects uz Vz; thus
    3748 A" uzv. = ( - 1)" { UzVn – Nuz+]Vz+ + C (1, 2)Uz+2Vx+2 = & c. }.
    550 CALCULUS OF FINITE DIFFERENCES.

    3749 Ex. (2) : To expand a* sin æ by successive differences of sin x.


    A"aʼsin x = {E (1+ 4°)- 1}" a"sin« = { A + EA?}"a"sinæ
    = { 1" + nar-1EA' + C (n, 2) Aº-? E A + & c. } aʼsin x
    = A ” a ”. sin x + nan-1q ++? A sin x + C (1 , 2) A ? -? q*+ 2A’ sinx + & c .
    = a* { (a – 1)" sinæ + n (n - 1)"-1aA sinæ + C (n,2)(a − 1)n-?a?Aºsinx + & c.},
    by ( 3727 ), while A " sin æ is known from ( 3728).

    3750 Ex. ( 3 ) : To expand A ” U , V , in differences of U , and v , alone : put


    E = 1 + A , E = 1 + 4 ' in (3746 ), thus
    A " U ,V = ( A + A' + 44')" UzVz,
    which must be expanded .

    A ” u , in differential coefficients of us. .

    3751 - Apu, = dºu: + A ;8 +! + And +Pu: + & c.


    Proof. A”ur = (6 = - 1 )" u , (3737).
    Expand by ( 150) and ( 125 ) as if d , were a quantitative symbol. See also
    (3761) .

    3752 dan = {log (1 + 0)}"u.


    The expansion by (155) and (125) will present a series of
    ascending differences of u .
    Proof.- eur = 1 + 4, i d, = log (1+ 1).
    3753 Ex.: If n = 1, du = Au - 4*4 + 4°0 –Au + &c.
    If C be a constant,
    3754 $ ( D ) C = $ (A ) C = $ (0) C and $ ( E ) C = $ (1) C ,
    Since every term of $ (D ), or of 0 (A ) C , operating upon C , produces 0;
    and every term of ° ( E ) operating upon C produces C .

    HERSCHEL'S THEOREM .
    3757 $ (et) = 4 ( E ) eº.t
    3758
    INTERPOLATION . 551

    Proof. — Let • (e") = 4 , + Aje' ... + Anent


    = 4 .80.4 + A ,Ecº.t ... + AZE "e0.t = ( A , + A , E ... + A , E , ) €0.
    = ° (E)€0.6 = (E ) { 1+ 0.t + 0 * + &c.},
    and ° (E ) 1 = $ (1) by (3756).
    A THEOREM CONJUGATE TO MACLAURIN ’S (1507).
    3759 $ (t) = $ (D ) ell.t
    3760 = $ (0) + $ (d,) 0.t+ $ (d.) O . * , + & c.
    PROOF. 0 (t) = q (log e') = 9 (log E ) eº. (3757)
    = 4 (D )60-t (3738) = 4 (D) { 1 + 0.4 + + & c.} ,
    and 0 ( D ) 1 = 0 (0 ) (3754 ).

    n being a positive integer ,


    3761 + 2 + 2
    _ d "u A ” On +1 qu +lu A” On+ 2 dn +?
    AU = dan 71.2 ... (n + 1) dwu + 1 + 1 .2 ... (n + 2 ) dxn +2 T
    Proof. —By (3758 ), putting $ (e') = ( – 1)",
    (6 - 1)" = ( E - 1)" 0 .6 + ( E - 1)" 0%. . + & c. = A ”0.t + A” O? ,1 2
    + & c.

    Put t = dr, and employ (3736 ) and (3737).

    INTERPOLATION .

    Approximate value of u , in terms of n particular equi


    distant values.
    3762 If ur is an integral algebraic function of the degree
    n - 1 , A ” Uz vanishes, and therefore by making x = 0 , and
    writing x for n in (3740) ,
    3763 U , = 4 ,+ xAu, + Cx,24 %u ... + Cc,n –14" -1Ug.
    This is formula (265). The given values are wo, Auo, A ’Uo,
    & c., corresponding to a , b , c , . ..
    3764 For an application of the formula to the problem of interpolation,
    see (267 ), in which example x = 1:54 and uz = log 72:54 .
    552 CALCULUS OF FINITE DIFFERENCES .

    3765 When the term to be interpolated is one of a set of


    equidistant terms, employ (3741) . A " llz = 0 , as in (3762) ;
    therefore
    3766 Un — nun- 1 + Cn,2Wn- 2 - Cn,3Un- 3 ... + ( - 1)" u , = 0 .

    3767 Ex.: From sin 0 , sin 30°, sin 45', and sin 60°, to deduce the value
    of sine 15°.
    The formula gives sin 0 – 4 sin 15° + 6 sin 30° — 4 sin 45° + sin 60° = 0 ,
    or - 4 sin 15° + 3 — 2 / 2 + 1V3 = 0 ,
    from which sin 15° = (6 – 4 / 2 + 73) = - 2594 .
    The true value is ·2588 ; the error :0006.

    LAGRANGE'S INTERPOLATION FORMULA .


    3768 Let a , b, c, ... ki be n values of x , not equidistant, for
    which the values of Uz are known ; then generally
    3769
    „ ( x - 6 ) ( x - c )... ( r - k ) . ( P - a )( x -- c) ... ( r - k )
    Un = una
    “ (a b)(a – c)...(a – k) "*(6 — a)(6 — c) ... (6 — k)
    ....... .... 1+U2
    ,, ( r — a )( x — b )(x — c)...
    " * (k — a ) (k - b ) (k — c)...
    Proof. — Assume Uz = A (2 - 6) (« — c) ... ( 2 — 1 )
    + B ( x — a )( x - c) ... (x — k ) + (x a ) (x — b ) (x — d ) ... ( 1 — k ) + & c.,
    and determine A, B , C , & c. by making x = a , b, c, & c., in turn.

    If the values of a ,b, c, ... k are 0 , 1, 2, ... 1 - 1, (3769)


    reduces to
    3770 X (ix - 1 ) . .. ( x - n + 2 )
    Uz = Un - 1 - 1 . 2 . 3 . .. ( n - 1 )
    Lux (x - 1) ... (x - n + 3 )(x - n + 1)
    - 2

    1 . 1 . 2 . 3 . .. (1 - 2 )
    ..tun- 2 V ( x - 1 ) ... ( r - n + 4 ) (x - n + 2)(x - n + 1) _ & c., or
    2 . 1 . 1 . 2 . 3 . .. ( 1 - 3 )
    3771 Vin ) ( U - 1. Cn - 1,1 Un Cn - 1, 2Un -3
    U = .

    (X - n + 1 X - n + 2 ' x - n + 3
    MECHANICAL QUADRATURE. 553

    MECHANICAL QUADRATURE .
    The area of a curve whose equation is y = u , in terms of
    n + 1 equidistant ordinates u ,ug...Un, is approximately
    3172 nu+ au +(* = *:)* % +(** –-*+n )
    + * –3 + 1 –31*)
    +(**»– 2n*+ 35m*–5976°+12nº)SI
    +62– 1579+1773 –25m + 278 – 603)
    Proof. — The area is = | Uzda . Take the value of uz in terms of
    Up, U ,... Un-1 from (3763) and integrate.

    3773 When n= 2, fvdv =" +paytus


    3774 n =3, Sv.de = 3(u+ 31 + 3ues+ ),
    3775 n = 4, (*w dx = 14(u + u.) + 64(101
    45 + .) + 210,
    37176

    n = 6, Su_dx = {u+ us+un+us+5(0,+us)+6u;}.


    In the last formula , which is due to Mr. Weddle , the co
    efficient of Afu is taken as % instead of 40, its true value.
    These results are obtained from (3772) by substituting for
    each A its value from (3742).

    COTES'S AND GAUSS’S FORMULÆ . ·


    3777 These give the area of the curve directly in terms of
    те MS

    fixed abscissæ .
    They are obtained by integrating Lagrange's value of '
    (3769– 71), and are fully discussed in articles (2995– 7).
    4 B
    554 CALCULUS OF FINITE DIFFERENCES.

    LAPLACE 'S FORMULA .

    3778 lude=" + ,+
    (Au,– Auc)+ (1°U.– A u0)— &c.,
    the coefficients being those in the expansion of
    t {log (1 + 1)}"?
    Proof.— .ws = d.{10g( +2)}w., by (3736),
    = d. {1+ 2 - - 1 * +&c.}w..
    Hence, putting Wz = Aw z
    Svzdo = " " - ATP + -&c.,
    Puzda = " + 43- 4 + * –&c.,
    and so on ; then add together the n equations.

    3779 Formula (3778 ) contains Aun, A ’ un, & c ., which cannot


    be found from U , U . .. Un.
    The following formula does not involve differences higher
    than Aun- 1
    3780 u_dx = " +03,+ ....+
    - Aw,-1 -A4, - 2, - - A ) - c.
    E
    - A -1
    PROOF. — In the proof of (3778), change log (1 + A into E
    s log ( 1 + 1 ) * log ( 1 - AE -1)
    and put E -'w , = W3-1 (3739) after expansion, and proceed as before.

    SUMMATION OF SERIES.
    3781 Definition : Eur = untua+ 1 + un+ 2... + uz - 1.
    3782 Theorem : { u = A - lu : + 0 ,
    where C is constant for all the assigned values of x .
    SUMMATION OF SERIES. 555

    PROOF. — Let (w ) be such that Ap (a ) = Uz, then 0 (x ) = A -'Uze


    therefore ua = q (a + 1) - ° ( a ). Write thus, and add together the values of
    Ua, Uq+1, ... Uz -1. Therefore, by (3781), Euz = (a ) - ° (a ) = A -'uz - q (a ),
    and º (a ) is constant with respect to .
    Taken between the limits x = a , x = b - 1, we have the
    notation ,
    x = 6- 1

    3783 Eur or tu £u – Eu, = A -103- A - ua.


    Functions integrable in finite terms :
    3784 Class I. 8.0m) = 2** *" + C .
    (ax + 6 ) (m + 1)
    3785 (ax + b ) m
    a (m + 1) + C .
    3786 Class II. By (3718 ),
    Exl-m ) = amii + C . and notation (3711).

    (ax + b) ( -m + 1)
    3787 E (ax + b)(=m) = C – la a (m - 1)

    Formulæ (3785 ) and (3786 ) are equivalent to the rules


    ( 269) and (271) . They are the direct results of theorem
    (3782).
    3788 Eu” = 1 [By (3726).

    Class III. — If be a rational integral function ,


    3789 24+ = u, = {x + Cx24 + CX,34?+ ...}uq.
    Proof. - By (3735) and (3736 ),
    Wa
    Ug + Wand ... + Wa+2- 1 = (1 + E + E ... + Ex-1)va = E - Tua = ( + 4 ) *
    = the expansion above.
    ...+P***)va= E= « =(1+4)* 1.4
    3790 The formula has been given at ( 266) and an example of its appli
    cation . The series there summed is 1 + 5 + 15 + 35 + 70 + 126 + to 100 terms.
    The function uz which gives rise to these terms is found by (3763) to be
    Uz = (x * + 10x® + 352 ? + 50x + 24 ) - 24 .
    556 CALCULUS OF FINITE DIFFERENCES.

    3791 If this function be presented as u ,, and tu , be required , we


    first find u = 1 , 1 , = 5 , u , = 15, & c . ; then the differences Auto, A 'Wo, ... Aʻu , =
    1, 4 ,6 , 4 , 1 , and then , by (3789), the required sum , as in the example re
    ferred to.

    3792 For another example, let 2 * 2 = 1 + 28 ... + nö be required .


    Here A. = 3x2 + 3x + 1, A ?w = 6x + 6 , 19x = 6,
    therefore A0 = 1, A²O3 = 6, 9°0 $ = 6 ........................ (1 ),
    we may now be expressed in factorials, and the summation may then be
    effected by (3784 ). First, by (3730 ),
    svø = + 3x (x - 1) + x (x - 1) ( x - 2 ) ;
    therefore, by ( 3784 ), 3728 = (n + 1)* (3783),
    mXo" zec” == n (n + 1) + 3 (n + 1)3n (n - 1) + (n + 1) n (n - 1)(n - 2) = n*(n + 1)".
    3793 Otherwise, by ( 3789), taking a = 0, we have
    Ug = a , W , = 0, Aug = 1, Air , = 6, Aļu, = 6 , as above.
    Therefore
    tv,="(n=1)+6n(1=1)(
    1 . 2 . 3x -2)+6n(n=1)(n=2
    1 . 2 . 3 . 4)(1–3)= n*(
    n=1”,
    therefore, changing n into n + 1, "^, = (n + 1)**..
    3794 Class IV . — When the general term of a series is a
    rational fraction of the form
    A + Bx + Cx' + where u , = ax + b ,
    Uz! + 1 ... Ur+ m
    and the degree of the numerator is not higher than x + m - 2 ;
    resolve the numerator into
    A ' + B'uz + C' u,Ux+1 + ... + D 'uqWz+1Xz+2 ...Uz+ m -2,
    by (3730) . The fraction then separates into a series of frac
    tions with constant numerators which can be summed by
    (3787).
    3795 If the factors Uz... Ug+m are not consecutive, introduce
    the missing ones in the denominator and numerator, and then
    resolve the fraction as in the foregoing rule .

    3796 Ex. : To sum the series 14


    rostozat to n terms.
    SUMMATION OF SERIES. 557

    The nth term is - (n + 1) (n + 2 )


    - n (n + 3 ) n (n + 1) (n + 2) (n + 3)
    n (n + 1 ) + 2n + 2 2
    n (n + 1)(n + 2)(n + 3) (n + 2)(n + 3) * (n + 1)(n + 2)(n + 3)
    2
    *'
    n (n(+ 1 ) (n + 2) (n + 3 )
    The sum of n terms is, therefore, by the rule (271),
    2 1 / 2
    3 (n + 2) (n + 3 ) / 3 (2.3 (n + 1) (n + 2 ) (n + 3)
    11 3 n² + 12n + 11
    18 3 (n + 1) (n + 2) (n + 3)
    If the form in ( 3787) is used, the total constant part is determined finally
    by making n = 0 , which gives C :

    3797 Theorem . f ( E ) a *b (x ) = a*f (a E ) $ (x ),


    f being an algebraic function .
    Proof. — Let a = em, then the left
    = f ( E ) emIp (x ) = f ( e" ) em * q ( ) (3736 ) = emoj ( eD +m ) 9 (w ) ( 3475)
    = arf (aE ) • (a ).

    Class V . - If e a rational integral function ,


    3798

    £a*$(a)= *(~)- a 44(2)+ ( 41) 44(a)—&c.}.


    The upper limit is understood to be x - 1, and a constant is to
    be added , (3781–2 ).
    Proof. — £a*ø (« ) = A -!a*$ ( ) = (E - 1)-'a®! (x ) = a* (a E - 1) - ° (x )
    (3797) = a*{a(1+ a) –1 }"$(x) = (1+ )**(*).
    Then expand the binomial.

    Ea +p(x) in successive derivatives of $(x).


    3799 £a*$(x ) = {1 + 4,$'(x) + 4 ; 4"(x)+ &c.},
    - 1
    laE - 11 - 1
    where „ =
    where 4 = (la -=1)*:
    1 ) =(1+ 4)*0
    PROOF. — By (3757), 4 ( P ) = ( E ) eº.D ; therefore (see last proof)
    a* (a E - 1)-40 (Q ) (putting E = e") = at (a E - 1)-120.0° (a )
    com I (1+ ) "{1+0.D + .2D*+&c.}$( ).
    558 CALCULUS OF FIN ITE DIFFERENCES.

    3800 Ex. : To sum the series 2 . 1 + 4 . 8 + 8 .27 + 16 . 64 + to n terms.


    We require 2*x + 2*2 = 2* + 2 (x* — 21.2° + 2°A ? * — 23A %2*)
    = 2*x + 2 { x9 – 2 (32° + 3x + 1) + 4 (6x + 6) - 8 .6
    = 2* { 2x* — 6x9 + 18x — 26 } .

    3801 If A -nu , be known for all integral values of n , and if


    Vi be rational and integral,
    Eu,v. = { 'uz.Vr-2 - & ’uz.Av.-2 + £ }ur.A ?vr-3 - & c.
    Proof. EuxVx = (EE' - 1) - 4x2x = (AE' + 4 ') -° UxVx (3746) and (3736 ).
    Expand the binomial operator, observing (3738) .
    3802
    { " u20 : = u , { "vc - nau, { n + 1vx+1 + Cm,2A %u2n+ 20x +2 - & c .
    Proof. L "urVr = (A ' + AE') -° UxVx, as in (3801),
    = A -»vUr — n ( 1 -1 -1E ) v ,Aur + C2,9 ( A -» -? E ?) v , A ’u , - & c.,
    producing the above by ( 3735) and (3782).
    Observe that, in (3801) and (3802), two forms are obtainable in each case
    by expanding the binomial operator from either end of the series.

    3803 Ex. : To sum the series sin a + 22 sin 2a + 3 sin 3a + to a terms.


    The sum is = aż sin ax + Exº sin « x . Taking u, = sin ax and r': = x*, we
    know A - " sin ax, by (3729) ; therefore (3801) gives
    Exº sin ax = (2 sin ļa)- sin {ax - } (a + 7 ) } (x - 1)
    - (2 sin þa)-2sin { ax - (a + )} (2x – 3) + (2 sin þa)-'sin {ax – } (a + 7 )} 2.

    APPROXIMATE SUMMATION .
    3820 The most useful formula is the following

    31 .= c+fu_de+ : + Bar delte +&c.


    C U , , 1 du , 1 d’u , , 1 du,
    = c +fude " + 12 dx 720 dxi + 30 -an pur
    1 đºur
    1209600 am + 47900160 am – &c.
    Proof. — Su , = (e” – 1)-'07. Expand by (1539) with D in the place
    of x .
    APPROXIMATE SUMMATION. 559

    Ex. 1 : The value of Xx? at (2939) is given at once by the formula.


    3821 Ex.2: To sum theseries 1+ + ...+ approximately,
    1 + 1 = + +loga - . - + 1202 –de.
    Put x = 10 to determine the constant; thus
    1
    + 07 00 + & c .,
    1+ 1 + ...+71b == C+
    C + log 10 +
    log10+ 0– 121200+&o,
    from which C = ·577215 , and the required sum is
    -577215+logæ+ - 12 + 1200t–&c.
    3822 Ex.3: 1+ k + s + 1 +&c.,
    +

    * OB
    -
    1

    1 1
    -

    12 12 20 + 12
    +

    The convergent part of this series, consisting of the first five terms, is an
    approximation to the sum of all the terms.
    3823 A much nearer approximation is obtained in this and analogous
    cases by starting with the summation formula at a more advanced term .
    1 1 1 1
    E .g .: 1 + 53 + 3 + 4 + 5 -
    _ 2035 1 1 , 3B, 5B, 16
    - 1728 + 2 . 50 + 2 .53 + 2 . š: 2 58 + & c .
    _ 203511 1 : 1 1 1
    1728 + 50 + 250 + 2500187500 + & c.
    The converging partnow consists of a far greater number of terms than before,
    and the convergence at first is much more rapid .

    3824 Ex. 4 : The series for log r (@ + 1) at (2773) can be obtained by


    the above formula when x is an integer. For, in that case,
    log (x + 1) = log 1 + log 2 + log 3 ... + log x = log æ + log x,
    and (3820 ) gives the expansion in question , the constant being determined
    by making æ infinite.

    3825 Formula (3820) may also be used to find (w_dx by the


    process of summation , and thus answers the purpose of
    Laplace's formula (3778).
    ES
    560 CALCULUS OF FINITE DIFFERENC .

    " u , in a series of derivatives of uz.


    3826 Lemma.
    n - 1 ! (et - 1) -` = ( – 1)n - :(de + n - 1) (d , + n - 2)
    ... (d4 + 1) {e - 1} -1.
    PROOF. — Putvn for n - 1 ! ( e' — 1) -". Then
    Vn+1 = - (d , + n ) vn = (d + n ) (d , + 1 - 1) vr-1.
    Y "uzmay now be developed .

    3827 Ex. - To develope L® urg (Boole, p. 97)


    2 (€*– 1)- = (dle+ 2)(d,+ 1) { - 1 } --
    = (dx+3ds+ 2){+ - ] + 4,4+ 4, +&c.}
    with A2y = 0, and
    - +
    Agr+1 = ( - 1 )*B21+1 ( 2r + t8
    -
    T (24 ,+ 34, - 1)
    + *{(r + 2)(-+ 1) 4, + 3 (r + 1) Ar.1 + 24,} +".
    Therefore, changing t into d ,,we Bget
    et

    I'v = 51w,do– ſfw.de+ſuçlo us+20 dies –&c.


    3828 snu, in a series of derivatives of u - n.
    Letxe" cosec” x = 1 - Cyma + C * &c., then
    Itu = D-t{1+ c:(2 ) + c (9 )*+ &c.}u,
    Boole, p . 98.

    3829
    $ 0) - (1)+ $(2) — & c. = 7 {1 - 2 + 4 - &c.} $ (0).
    By this formula , a series of the given type may often be trans
    formed into one much more convergent.
    1 1
    PROOF. — The left = - 100 2 +1 (0 ) 2 1 + 1AP "
    the expansion of which is the series on the right.

    3830 Ex.– Tosum 1 – 1 + 1 - 1 +&c. Summing the first six terms,


    it becomes 32 + 1 - 11 + & c. Tak1 ing $(0)2 = (0+ 7)-','
    2.3
    1 1
    7 87& c. = 2 7 + 2 .7 .8 + 4 .7 .8. 9 + 8.7. 8.9. 10 Tac.
    The sum after six terms converges rapidly by this formula, and more rapidly
    than if the formula had been applied to the series from its commencement.
    PLANE COORDINATE GEOMETRY.

    SYSTEMS OF COORDINATES .
    CARTESIAN COORDINATES.
    4001 In this system (Fig . 1)* the position of a point P in a
    plane is determined by its distances from two fixed straight
    lines OX , OY, called axes of coordinates. These distances
    are measured parallel to the axes. They are the abscissa PM
    or ON denoted by x , and the ordinate PN denoted by y. The
    axes may be rectangular or oblique. The abscissa w is
    reckoned positive or negative according to the position of P
    to the right or left of the y axis , and the ordinate y is positive
    or negative according as P lies above or below the x axis
    conformably to the rules (607, ’8 ).
    4002 These coordinates are called rectangular or oblique
    according as the axes of reference are or are not at right
    angles .

    POLAR COORDINATES.
    4003 The polar coordinates of P (Fig . 1) are 1 , the radius
    vector, and 0, the inclination of 1 to 0X, the initial line,
    measured as in Plane Trigonometry (609).

    4004 To change rectangular into polar coordinates, employ


    the equations x = r cos 0, y = r sin 0 .
    4005 To change polar into rectangular coordinates, employ
    r = Vx + y*, = tan- ( 2 ).
    * See the end of the volume.
    40
    562 PLANE COORDINATE GEOMETRY.

    TRILINEAR COORDINATES.
    4006 The trilinear coordinates of a point P (Fig . 2 ) are
    a , ß , y , its perpendicular distances from three fixed lines which
    form the triangle of reference, ABC , hereafter called the trigon .
    These coordinates are always connected by the relation
    4007 aa + bB + cy = 2 ,
    4008 or a sin A + ß sin B + ysin C = constant,
    where a, b, c are the sides of the trigon, and E is twice its
    area .

    4009 If a , y are the Cartesian coordinates of the point


    aßy, the equations connecting them with the trilinear
    coordinates are, by (4094 ),
    a = x cos a ty sin a - Pi,
    B = x cos B + y sin ß - P2,
    y = x cos y + y sin y - P3.
    4010 Here a has two significations. On the left, it is the
    length of the perpendicular from the point in question upon
    the side AB of the trigon . On the right, it is the inclination
    of that perpendicular to the x axis of Cartesian coordinates.
    Similarly B and y .

    4011 The angles a , b , y are connected with the angles


    A , B , C by the equations
    y - B = 1 - A, a -y = r - B, a - B = m + C,
    only two of which are independent.
    4012 P1, P2, P3 are the perpendiculars from the origin upon
    the sides of the triangle ABC .

    AREAL COORDINATES .
    If A , B , C (Fig . 2 ) be the trigon as before , the areal co
    ordinates a', ß ', y ' of the point P are
    4013 a' = a = PBC B B _ PCA Y PAB
    C
    PAB B '= y'= P3 . ABC
    SYSTEMS OF COORDINATES. 563

    The equation connecting the coordinates is now


    4014 a ' + B ' + ' = 1.
    4015 To convert any homogeneous trilinear equation into
    the corresponding areal equation .
    4016 Substitute đa = 2a , 68 = 28, c = 2y .
    Also any relation between the coefficients l, m , n in the
    equation of a right line in trilinears will be adapted to areals
    by substituting la , mb, nc for l, m , n . Similarly for a , b, c,
    f , g , h , in the general equation of a conic (4656 ) , substitute
    aa?, 66°, cc , fbc, gca , hab.
    In either the trilinear or areal systems, a point is deter
    mined if the ratios only of the coordinates are known .
    Thus, if a : ß : y = P : Q : R , then , with trilinear co
    ordinates,
    ΡΣ P.
    4017 a = PLOIt ; and, with areal, a = PD.+ Q + R

    TANGENTIAL COORDINATES.
    4019 In this system the position of a straight line is deter
    mined by coordinates, and the position of a point by an
    equation . If la + mb + ny = ( be the trilinear equation of a
    straight line EDF (Fig. 3 ) ; then , making a , ß , y constant,
    and l, m , n variable , the equation becomes the tangential
    equation of the point 0 (a , b , Y ) ; whilst l, m , n are the co
    ordinates of some right line passing through that point.
    Let i , u , v (Fig . 3 ) be the perpendiculars from A , B , C
    upon EDF , and let P1, P2, P3 be the perpendiculars from A , B ,
    C upon the opposite sides of the trigon ; then , by (4624 ) , we
    have
    4020 RX = lpi, Ru = mp2, Rv = np3,
    where R = / (1 +mº+ n ? — 2mn cos A – 2nl cos B - 21m cos C ).
    Hence the equation of the point 0 becomes mes

    4021
    d a ,
    Natur Y = 0 or i sin 0, -, sin 0, + , sin 03 = 0,
    р р. Ря PiP2 P3
    where pı = 0A, 0 = 2 BOC, & c., and 2ABOC = P2P3sin 0.
    564 PLANE COORDINATE GEOMETRY.

    Formula (4021) shows that,when the perpendiculars 1 , H , v


    are taken for the coordinates of the line, the coefficients be
    come the areal coordinates of the point referred to the same
    trigon .
    4023 Any homogeneous equation in l, m , n as tangential
    coordinates is expressed in termsof 1 , M , v by substituting for
    Iman . . , respectively . By (4020).
    Pi' P2 P3
    4024 An equation in 1,u , v of a degree higher than the first
    represents a curve such that i , u , v are always the perpen
    diculars upon the tangent. The curve must therefore be the
    envelope of the line (a , a , v ).

    TWO-POINT INTERCEPT COORDINATES.


    Let 1 = AD , u = BE (Fig . 4 ) be variable distances from
    two fixed points A , B measured along two fixed parallel lines,
    then
    4025 al + buto = 0
    is the equation of a fixed point ( through which the line DE
    always passes. This may easily be proved directly , but we
    shall show that it is a particular case of the system of three
    point tangential coordinates.

    Let one of the vertices (C ) of the trigon in that system be at infinity


    (Fig . 3) . Then equation (4022) becomes
    sin 0, nu sin 0 , + sin COE sin 0, = 0 .
    . Pi P3
    For v :Pz = sin COE always. Divide by sin COE ; then 1 - sin COE = AD,
    & c., and the equation becomes
    sin (, AD + sin 0, AE + sin 0 , = 0 .
    P
    The only variables are AD and AE. Calling these 1 and y , the equation
    may be written al + by + c = 0 ,
    the form taken by ax + bue c'v = 0 when v = 0 and c' vanishes.
    - - -
    ONE-POINT INTERCEPT COORDINATES.
    4026 Let a , b be the Cartesian coordinates of the point 0
    (Fig . 5 ) ; and let the reciprocals of the intercepts on the axes
    SYSTEMS OF COORDINATES. 565

    of any line DOE passing through O be & = Am n =


    Then , by (4053),
    4027 af + bn = 1
    is the equation of the point 0 , the variables being £, n .
    This is a case of the system of three-point tangential coordinates in which
    two of the vertices ( B , C ) of the trigon are at infinity. Equation (4022)
    now becomes ^ sins + sin BOD sin 0, + sin COE sin 0 , = 0 ,
    Pi
    sino, , sin , sin 0g = 0 .
    P AD ' * AE = 0,
    which is of the form a£ + bn = 1.

    TANGENTIAL RECTANGULAR COORDINATES.


    4028 This name has been given to the system last described
    when the two fixed lines are at right angles (Fig. 6 ).
    The coordinates &, n , which are defined as the reciprocals
    of the intercepts of the line they determine, have now also the
    following values.
    4029 Let x , y be the rectangular coordinates of the pole of
    the line in question with respect to a circle whose centre is
    the origin and whose radius is k ; then

    $= and
    since x .OM = y .ON = k ?; for M , N are the poles of y = 0,
    it = 0 .

    4030 The equation of a point P on NM whose rectangular


    coordinates are OR = a , OS = b , is
    af + bn = 1, by (4053),
    this equation being satisfied by the coordinates of all lines
    passing through that point.
    4031 In all these systems an equation of a higher degree
    in g, n represents a curve the coordinates of whose tangents
    satisfy the equation .
    ANALYTICAL CONICS
    IN

    CARTESIAN COORDINATES.

    LENGTHS AND AREAS .

    Coordinates of the point dividing in the ratio 1 : n the


    right line which joins the two points xy , a 'y'.
    _ nx ' + n 'x _ ny' + n 'y
    4032 Š = -
    ntn ' n + n'

    Proof.— (Fig. 7.) < = x + AC = x + * (c' — x). Similarly for n.

    4033 If n = n', $ = 274, n = 971


    4034 Length of the line joining the points ay, x'y
    = V (x - )? + (y - y )?.
    The same with oblique axes
    4035 V (X — X')? + (y - y )? + 2 (x — X ') (y - y ) cos w .
    PROOF. — By ( Fig. 7 ), Euc. I. 47, and (702) .

    Area A of a triangle in terms


    ms of the coordinates of its

    angular points # 191, X , Y2, xzY3.


    4036 A = } { x192— X ,Y . + X, 93 — X3y2 + x3y2 – x y } .
    Proof. — (Fig . 8.) By considering the three trapezoids formed by Y ., Yu,Ys
    and the sides of the triangle, we have
    A = } (y .+ y )(x ,— ) + } (Ya + y ) (23 — X.) - ; ( + y )(13— 2,).
    LENGTHS AND AREAS. 567

    Area of the triangle contained by the & axis and the lines
    y = mx + 61, y = 228 + 62, (4052)
    4037 A = _ (6 — c )2 – ( B ,C , B ,C ,) 2
    (4056 )
    "* 2 (m , - m .) 2B, B , ( A , B , - A , B )
    Proof. — (Fig . 9.) Area = } ( - c ) p , and p is found from
    pm - pm , = 67 - Cg. The sign of the area is not regarded .

    Cor. — Area of the triangle contained by the lines


    y = m2 + 1, y = mya + ca, y = Mgư + cs,
    4038 A = { } (6 —c )? + (C2 - C3) + (C3 — c )” ).
    ( mi - m2 m2 — mzmz - mi )
    4039 - {c; (m2 - m3) + c2 (mz— m ,) + c3(m , — m ,)}
    2 (m , - m2)(m , - m3)(mz– m )
    2040 ( B ,C , , B , C )2 ,) + & c.
    = 2B , B , ( A ,B , - AA ,B
    , B ,).
    4041 Square of Determinant ( A , B , C3)
    - 2 ( A , B , – A ,B3)( A , B , – A , B ,)( A , B , - A B3)*
    PROOF. — (Fig. 10.) ABC = AEF + CDE - BED. Employ (4037).

    Area of Polygon of n sides.


    First in terms of the coordinates of the angular points
    41919 & 242, ... XnYn .
    4042 2A = (2 ,3 - 0 , ) + (4 ,2/3 - 232 ) + ... + (2 , - ,3 ,)
    = x1(yz- yn) + X ) (93 - y ) + ...... + x » (yı - Yn -1).
    Secondly, when the equations to the sides are given , as in
    (4037 ).
    4043 2A = (c.— c) + (62 — c)* + ... + (c. — c)”.
    m , mi
    mi- ma ' m2- M3
    4044 Also three values similar to (4039, ²40, ’41).
    Proof.—By(4367),adding the componenttriangles.
    4047 Each expression for the area of a triangle or polygon
    will be adapted to oblique axes by multiplying by sin w .
    568 CARTESIAN ANALYTICAL CONICS.

    TRANSFORMATION OF COORDINATES.
    4048 To transform the origin to the point hk.
    Put x = x' + h , y = y' + k .
    To transform to rectangular axes inclined at an angle 0
    to the original axes.
    4049 Put
    x = a ' cos 6 – y' sin 0, y = y' cos 0 + x'sin 0. (Fig. 11.)
    Proof. — Consider a 'as coso and y' as sing. Then x = cos (° + ) and
    y = sin (0 + ) (627 , '9) .

    Generally (Fig . 12) , let w be the angle between the original


    axes ; and let the new axes of x and y make angles a and ß
    respectively with the old axis of x .
    4050 Put x sinw = x ' sin (w - a ) + y' sin (w – B )
    and y sin w = x ' sin a + y ' sin ß .
    oc Proof. - - ( Fig .. 12.)
    -Roofi(Fig 12.) The coordinates of P referred to the old axes being
    OC = x , PC = y , and referred to the new axes, OM = a , PM = y , we have,
    by projecting OCP and OMP at right angles first to CP and then to OC,
    CD = MF - ME, PN = ML + PK ,
    which are equivalent to the above equations.

    To change Rectangular coordinates into Polar, nk being


    the pole 0 , a the inclination of the initial line to the x axis
    (Fig . 13), and sy the point P .
    4051 Put x = h + r cos (@ + a ), y = ktr sin (0 + a ).

    THE RIGHT LINE .

    EQUATIONS OF THE RIGHT LINE.


    4052 y = mx + c ... .. . .. . . .. ... .. . .. .. ... .. (

    4053 ...( 2 ) ,
    ä + * = 1.
    THE RIGHT LINE . 569

    4054 x cosaty sin a = p ... ...... . ...(3).


    4055 Ax + By + C = 0 .. .... .. .......... ... (4 ).
    PROOF. - (Fig . 14 .) Let AB be the line. Take any point P upon it,
    coordinates ON = x , PN = y . Then , in (1 ); m = tan 7, where 0 = BAX ,
    the inclination to the X axis ; therefore ma = - OC , and c is the intercept
    OB . In ( 2 ), a , b are the intercepts 0A, OB. In (3 ) , p = 0s, the per
    pendicular from 0 upon the line ; a = LAOS.
    p = OR + LP = x cos a ty sin a .
    (4 ) is the general equation.

    4056 m = tan0 = - 4 = - = - cot a.


    B
    4060 sin 0 =
    SI
    A , cos
    COS 0 =
    ✓ A + B. VA + B2
    4062 p = c sin a = ✓1
    - + m= = - - A + B2*

    Oblique Axes.
    Equations (4052 , 53, ²55) hold for oblique axes, but
    (4054 ) must be written
    4065 x cos a ty cos ß = p . (Fig . 14 )

    4066 tan 0 = m sin w - A sin w


    1 + m cos w A cos w - B
    w being the angle between the axes.
    PROOF. — From m = sin 0 = sin (W — 0).

    csin w Csino
    4068 = _
    V1 + 2m cos w + m V A + B — 2A B cosa
    PROOF.--- From p = csin (w — 1 ) and (4166).

    The equations of two lines being given in the forms (4052)


    or (4055 ) , the angle , p , between them is given by
    m - m' AB'
    — A 'B
    4070 tan = i mm or AA + BB"
    Proof. - (Fig . 15 ) tang = tan (0) — 0 ') . Expand by (632) .
    4 D
    570 CARTESIAN ANALYTICAL CONICS.

    To oblique axes :
    ( m - m ') sin w
    4072 tan =
    1 + (m + m ') cos w + mm '
    Proof. - As in the last, employing (4066).

    Equation of a line passing through a'y':


    y - y' = m (x — x '), (Fig . 8)
    4073
    4074 or y - mx = y' — mx ',
    4075 or Ax + By = Ax' + By'.
    Proof. – From Figure (13), m being = tan 0 .

    Condition of parallelism of two lines :


    4076 m = m ', or AB ' = A ' B .
    Hence the equations differ by a constant.

    Condition of perpendicularity :
    4078 mm ' = - 1 or AA' + BB' = 0 . (4070 )

    The same to oblique axes :


    4080 + (m + m ') cos w +mm ' = 0 . (4072)
    4081 or AA' + BB' = (AB' + A ' B ) cosw,
    4082 or m ' = - 1 + m cos w.
    m + cos w

    A line passing through the points xıYı, X ,Y2 :


    y - yı — Yi Y2 = m .
    4083 X — X1 X1 X2
    Proof. — ( Fig. 16.) By the similar right-angled triangles PCA,ADB.

    4084 Or y = mx + 192 — # 991,


    X - X2

    4085 or (x - x )(y - y ) = (x — X»)(y - y ).


    PROOF. - This equation represents a straight line because it is of the first
    degree ; and the coordinates of each of the given points satisfy the equation.
    THE RIGHT LINE. 571

    A line passing through a'y and perpendicular to a given


    line (m ) :
    4086 y — y' = - (x - x ). (4073, '78)
    4087 or Bx — Ay = Bx' — Ay'.
    The two lines passing through a 'y and making an angle
    B ( = tan -? m ) with a given line (m ):
    4088
    9000 yy,
    X – X = mi1 + m- ,Mm , and mimm
    +m2 (4073,'70)
    1 - ,

    A line passing through hk and dividing the line which


    joins x1yı and 2.392 in the ratio nz : N, :
    4089 y - k _ n (42 - k ) + n2 ( 4 , - k )
    (4073, '32 )
    –h m , ( - 1 ) + ng ( a − h )

    Coordinates of the point of intersection of two lines :


    x = C1 - C2 – B , C , , B ,C ,
    4090
    m2 - mi A ,B, - A, B,
    Cim , —- C. ; A , C , - A ,C
    4092 y = (4116)
    m , - mi AB, – A . B ,

    Length of the perpendicular from a point a'y' upon a


    given line
    4094 = x ' cos a + y 'sin a - p.
    Proof. — Let AB (Fig . 14 ) be the line, and Q the point a'y . Then, by
    (4054) , a 'cos u + y' sin u = OT, the perpendicular from O upon a parallel line
    through Q , and p = OS.

    Otherwise, the same perpendicular


    4095 Ax' + By' + C (4060, '61,'94)
    VA + B

    The same with oblique axes


    = (Ax' + By' + C ) sin w
    4096
    JA’ + B2 — 2AB cosw)'
    obtained in a similar way from (4065 -69) .
    572 CARTESIAN ANALYTICAL CONICS.

    Condition of three lines intersecting in one point :


    4097 cm , — cum , tc,mz - c3mg + czm , - cm3 = 0 .
    The area in 1009 must vanish.

    4098 Otherwise . If certain values of the constants l, m , n


    make the expression
    1(Aqx + By + C ) + m (A2x + Boy + C .) + n (AzX + Bzy + C3)
    vanish identically , the three lines indicated intersect in one
    point.
    Proor. — By (4099 ), for then values of x and y which make (1) and ( 2)
    vanish also make (3 ) vanish .

    A line passing through the point of intersection of the


    lines Ax + By + C = 0 and A 'x + B 'y + C' = 0 is
    4099 Ax + By + C = k ( A ’x + By + C '),
    4100 or 1 (Ax + By + C ) — m (A ’x + B 'y + C ) = 0 ,
    k , l, and m being any constants.
    4101 RULE. — If the equation of a right line contains a third
    variable k in the first degree, the line always passes through a
    fixed point.
    Proof. — For the values of x and y, which satisfy simultaneously the given
    equations, also satisfy (4099), whatever k may be. See (4604).

    4102 If in the equation of a line Ax + By + C = 0 , the co


    efficients A , B , C involve á', y', the coordinates of a point
    which moves along a fixed right line, then the first line passes
    through some fixed point.
    Proof. — By means of the equation of the fixed line, y ' may be eliminated ,
    and ac' then remains a third variable in the first degree (4101).

    4103 To find the point in which the line Ax + By + C inter


    sects the line joining the points xy, a' y'; substitute
    Ax + By + C for n , and Ax' + By' + C for n ' in (4032).
    Proof. — By (4095), since the segments intercepted are in the ratio of the
    perpendiculars from xy, x'y ' upon the line Ax + By + C .
    THE RIGHT LINE. 573

    Equations of the line with l, m for direction -ratios, hk a


    fixed point on the line, and p the distance of the variable
    point xy from hk .
    4104
    sin (w - 0 sin 0
    4105 where 1 = (Oblique)
    sin w sin w '
    4106 or l = cos , m = sin 0. (Rectangular)

    Polar Equation of a Straight Line.


    4107 r cos (0 – a) = p .
    ( Fig . 17.) Here p is the perpendicular to the line from the
    pole 0 , and a is the inclination of p to the initial line 04.

    When the line passes through the pole, the equation is


    4108 0 = constant.

    A line passing through the two points 7,02, 1902.


    4109 rr, sin (0 – 0 ) + rır sin (0, - 0,) + rır sin (0, – 0) = 0.
    Proof. - (Fig. 18.) APOA + AOB - POB = 0 . Then by (707).

    EQUATIONS OF TWO OR MORE RIGHT LINES.


    The homogeneous equation of the nth degree,
    4110 x" + pjan-ly + pawn-?y* + ... + pny” = 0 ,
    represents n right lines, real or imaginary , passing through
    the origin .
    For it is resolvable into n factors of the form ( x - ay) , by (405 ).
    For the case of two right lines represented by the general equation of the
    second degree, see (4469).

    Equation of two right lines through the origin :


    4111 ax? + 2hxy + by? = 0.
    574 CARTESIAN ANALYTICAL CONICS.

    If o be the angle between the lines,


    4112 tan $ = V (h® – ab) or 2 sin w ✓ (h? – ab).
    a +b a + b - 2h cos w
    axes
    according as the axes are rectangular or oblique.
    acco C

    Proof. — Assume (y –ma) (y - m .2x ) = 0, and apply (4088).

    Equation of the bisectors of the angle o :


    4113 hx? - (a − b )xy - hy' = 0 .
    Proof. — Let y = ux be a bisector (u = tan 4 ) ; then , since 24 = 0, + 0,,
    24 – m + m = 2h , by (4111) ; and u =
    1 -ml- m ,mga- b'
    The roots ofthis equation are always real.

    GENERAL METHODS.

    APPLICABLE TO ALL EQUATIONS OF PLANE CURVES.


    4114 Let F (x , y ) = 0 ...... (i.) and f (x , y ) = 0 ......(ii.)
    be the equations of two curves
    rve
    of any degree.
    Cu

    4115 To find the intercepts on the x and y axes.


    Put y = 0 in (i.), then x becomes the intercept on the x axis .
    Similarly , put x = 0 for the intercept on the y axis.
    4116 To find the points of intersection of (i.) and (ii.).
    Solve as simultaneous equations. Each pair of values of x
    and y so obtained gives a point of intersection . Imaginary
    values give an imaginary point.
    4117 To determine equation (i.) so that the line may pass
    through certain fixed points , X , Y1, X2Y2, & c.
    Substitute x1y1, X ,Y2, 8c. for xy successively , so forming
    asmany equations as there are points. From these equations
    the constants in (i.) must be determined in terms of Xı,yı, X.,,yo,
    & c.
    4118 The number of arbitrary points cannot exceed the
    number of constants in the equation .
    GENERAL METHODS. 575

    4119 Condition that (i.) and (ii.) may touch .


    At a point of contact two or more points of intersection
    must coincide, and therefore the equation for x or y, obtained as
    in (4116 ), must have two or more equal roots for each point of
    contact. The contact is said to be of the second order when
    there are three coincident points ; of the third order when
    there are four, and so on .

    4120 To find the equation of the tangent at a point x'y' on


    the curve f (a , y) = 0 .
    Form the equation to the secant through twoadjacent points
    Xıyı, x,y, (4083), and determine the limiting value of Xı
    Yo -- Y2
    X2
    when the points coincide by means of the equations f (x1, yı) = 0 ,
    f (x2, y2) = 0 .
    4121 Otherwise vise
    m = dy, by (5101).
    m

    4122 For the equation of the normal, change m of the


    tangent into - т
    (4086 ).
    4123 To express the equation of the tangent, or normal, in
    terms of m and the constants of the curve.
    From the equation of the tangent or normal, the equation to
    the curve,and the equation furnished by the value of m , eliminate
    x ' y ', the coordinates of the point of contact of the tangent.

    THEORY OF POLES AND POLARS.


    4124 Let F (x', y', x , y) = 0 represent the equation to the
    tangent of a curve at the point x 'y'.
    Then F ( x , y , x ', y ') = 0 , the equation obtained by inter
    changing the constants a ', y' with the variables x , y, represents
    the polar of any fixed point a 'y ' not on the curve.
    Let x 41 x,y: ( Fig. 19) be points A , B on the curve, and let the tangents
    at those points intersect in a 'i . Consider the equations
    F (x1, Yv, x , y) = 0 ... (1 ) F (x2, Ya, x , y) = 0 ... (2), F (x , y, a', y ) = 0 ...( 3 ).
    Here ( 1 ), ( 2 ) are the tangents , and ( 3 ) is some straight line or curve
    according to the dimensions of x and y . Also ( 3) passes through the points
    576 CARTESIAN ANALYTICAL CONICS.

    of contact 27, 2 /2, and may therefore be called the curve of contact; or, if a
    right line, the chord of contact of tangents drawn from x', y', i.e., the polar.

    4125 Hence the coordinates of the points of contact of


    tangents from an external point a 'y ' will be determined by
    solving (3 ) and the equation of the curve simultaneously .
    4126 Again , let a'y ( Figs. 20 and 21) be any point P not on the curve.
    Then, from the equations
    F (x ,y ,a,y ) = 0...(4), F (X,Y,29, 43) = 0 ... (5), F (x,y, xq,y ) = 0 ... (6),
    we see that (4 ) is some straight line; that, if azy ; and x, y, are any two points
    upon it , (5 ) and (6 ) are the curves of contact of tangents from those points ;
    and that these curves of contact pass through the point x'y'.
    4127 If the points xxs, x ,y , are taken at A , B , where (4 ) intersects the
    curve, ( 5 ) and (6 ) then become curves touching the given curve at A and B ,
    and passing through a 'y '. We may call these lines the curve tangents
    from a 'y'.

    4128 Lastly, let z'y in (3 ) be a point within the given curve (Fig . 22),
    then the equations
    F (x , y , x ', y') = 0 ... (7), F (2 3,43, X , Y) = 0 ... (8 ), F (x ,44,x , y ) = 0 ... (9 )
    show that ( 7 ) is the locus of a point, the curve tangents from which have
    their chord of contact always passing through a fixed point. When x 'y is
    without the curve, as in Fig . ( 19), the same definition applies to every part
    of the locus ( 3 ) from which tangents can be drawn.

    4129 If the given curve be of a degree higher than the


    second , the line of contact of the tangents from a point is a
    curve, and the line of contact of the curve tangents from a
    point is a straight line ( Figs. 19 and 20 ). A similar converse
    relation is exhibited in Figures (21) and (22 ) .

    If the curve be of the second degree, equations (3 ) and


    (4 ) become identical. The line of contact or the polar is
    always in this case a straight line, and so is the locus ( 7 ) .
    Figures (19) and (20) now become identical, as alsó (21)
    and (22)
    4130 The polar of the point of intersection of two right
    lines with regard to a conic passes through their poles.
    Proof. — As in (4124). Let (1 ) and ( 2) be the two lines, (2 ,4 ), (x,y:)
    their poles, and x'y' their point of intersection.
    GENERAL METHODS. 577

    4131 To find the ratio in which the line joining two given
    points xy, a 'y' is cut by the curve f (x , y ) = 0 .
    Substitute for x and y , the supposed coordinates of the
    point of intersection , the values
    nx ' + n 'x ny' + n 'y by (4032)
    I ntn ntn' '
    and determine the ratio n : n ' from the resulting equation .
    The real roots of this equation correspond to the real points of
    intersection .

    4132 To form the equation of all the tangents that can be


    drawn to the curve from a point a 'y ' .
    Express the condition for equal roots of the equation in
    (4131), and consider ay a variable point.
    4133 To form the equation of the lines drawn from a'y' to
    all the points of intersection of two curves .
    Substitute nx' + n'x , ny' + n 'y for x and y in both curves,
    and eliminate the ratio n : n '.
    PROOF. — Take any other point xy on the line through a 'ij and a point of
    intersection . The ratio n : n (4131) is the same for each curve, and there
    fore may be eliminated .

    4134 To find the length , r = AP or AP ( Fig. 23 ) , of the


    segment intercepted between the point A or a 'y' and the curve
    f ( x , y ) = 0 on a straight line drawn from A at an inclination
    O to the X axis . That is , to form the polar equation with
    a 'y ' for the pole and the initial line parallel to the x axis .
    Substitute for x and y , the assumed coordinates of the point
    of intersection , the values x = ON or ON ', y = PŃ or P 'N ',
    that is, x = x ' + r cos 6, y = y' + r sin 0,
    and determine p from the resulting equation . That is, put
    a = 0 in (4051) .
    The real values of r are the distances of the points of inter
    section from a 'y'.
    4135 When an equation has been obtained for determining
    æ the length of a line , important results may frequently be
    arrived at by applying theorem (406 ) respecting the sum and
    product of the roots.
    4 E
    578 CARTESIAN ANALYTICAL CONICS.

    THE CIRCLE .
    Equation with the centre for origin .
    4136 ** + yº = pi . (Fig . 24.)

    Equations of the tangent at the point P or x'y'.


    4137
    y= y = - (2 —x'). (4120)
    4138 xx' + yy' = r .
    Also, by (4124), the polar of x'y', any point not on the curve.
    4139 y = mx +rv1+ mº; m = m =
    (4123)
    4140 x cos a + y sin a = r,
    a being the inclination to the x axis of the radius to the point
    a 'y'.
    Equation of the circle with a , b for the coordinates of the
    centre Q . (Fig . 24 .)
    4141 (x — a )² + (y - b ) = .
    Tangent at a'y', or Polar,
    4142 (x — a)(x' — a) + (y — b)(y'— b) = r , (4138)
    4143 or (r — a ) cos a + (y — b ) sin a = r ,
    a being the inclination of the radius to the point a'y'.
    General equation of the circle :
    4144 X ? + y + 2gx + 2fy + c = 0 .
    4145 Centre ( - 8 , - f ). Radius ✓ (8° + 8° — c).
    Proof. — By equating coefficients with (4141) .

    Equation of the circle with oblique axes : (Fig.25.)


    4146 (x — a )? + (y — 6)° + 2 (x — a)(y — b) cosw = p (702)
    4147 or x2 + 2xy cos w + yº — 2 (a + b cosw ) x
    - 2 (6 + a cosw ) y
    ta' + 2ab cos w + 6 = .
    THE CIRCLE. 579

    General Equation .
    4148 x? + 2xy cos w + y* + 2gx + 2fy + c = 0 .
    The coordinates of the centre are
    4149 a = f cosw - g , b = g cos w - f
    sinº w sin ’ w

    4150 Radius = V {gº — 2fg coswt.fº - c sin’w}


    sin a
    PROOF. — By equating coefficients with (4147) .

    Polar Equation .
    4151 po? + 12 — 2rl cos (0 - a ) = c', (Fig. 26 )
    4152 or pe - 21 cos a r cos 6 — 21 sin a r sin 0 + 1 - C = 0.
    Proof. --By (702), the coordinates of P being r and 0.

    General form ofthe polar equation :


    4153 22 + 2gr cos 0 + 2fr sin 6 + c = 0 .
    4154 tan a = L .
    a
    1 = Vg +fe.
    Proof. — By equating coefficients with (4152).

    4156 Equation of the circle passing through the three


    points X1Y1 X , Y2 , 23Y3.

    (n* + y )|x , y,11 - ( 47 + y;) x, y,11+ (x + y ) X3 Y: 1 - ( + y ) x y 1


    X2 Y2 1 1X3 Y3 1 x y 1 X 1 yıl
    | X3 Y3 1 IX y 1 X 14. 1 X , Y, 1
    Proof. — Eliminate g , f, and c from (4144 ) by (4117).

    Equation of the chord joining x1y1 XY2, two points on the


    circle x + + y2 = pol :
    4157 x (x1+ x2) + y (y + y2) = X1Xx + yıyz+ ?, (4083,4136)
    4158 or x cos ž (0 + 0 ) + y sin ] (0, + 0,) = r cos } (0, - 0,),
    where r cos , = x1, r sin 0 , = 41, & c.
    580 CARTESIAN ANALYTICAL CONICS.

    4159 NOTE. — The coordinates x , y of a point on the circle 2 ? + y = p may


    often be expressed advantageously in this way in terms of 0 , a single variable.

    4160 Let S = ( r — a)? + (y — ) — p ? = 0)


    be any circle (Fig . 27). Then, if æy be a point P outside the
    circle, S becomes the square of the tangent from P . If æy be
    a point P ' within the circle, S becomes minus the square of
    the ordinate drawn through P ' at right angles to the radius
    through P '.

    CO-AXAL CIRCLES.
    (See also 984 and 1021.)
    4161 If S = xº+ yº + 2gx + 2fy + c = 0 ,
    S' = x ? + y? + 2g'r + 2f"y + c = 0
    be two circles,the equation to the radical axis is
    S - S = 0.

    If x = 0) be taken for the radical axis , the equation to any


    circle (radius 7-) of the system of coaxal circles (1021) is
    4162 x + y* — 2kx + 82 = 0 and k — p ? = + 82,
    + in Figure (1 ), – in Figure (2). Here 8 = IR a constant,
    and k = 10 a variable.

    4164 The polar of x 'y for any circle of the system passes
    through the intersection of
    xx' + yy' + 8 = 0 and x + r ' = 0 .
    Proof. — Its equation is xx' + yy' – k ( 2 + x') + d = 0 (4121). Then by
    (4099) .

    4165 When k = d , then k = ID = ID '. D and D ' are


    Poncelet's limiting points.

    4166 The polar of D with respect to any of the circles


    passes through D ', and vice versâ , by (4164).
    4167 Tangents from any point on the radical axis to all
    •е
    circles of the system are equal (4160, ’61) .
    THE CIRCLE . 581
    _

    4168 The radical axes of three circles, Si, S2, S3, meet in a
    point called their radical centre.
    4169 The reciprocals with respect to the origin D or D ' of
    the system of co -axal circles are all confocal conics (4558 ) .

    The equation of the circle, centre Q , cutting the system of


    circles orthogonally is, putting IQ = h ,
    4170 x2 + yº — 2hy — 82 = 0 . (1230 , 1236 )
    This circle passes through D and D '.

    The common tangents to the two circles


    (x — a)2 + (y — b)2 = p 2 and (x — a')* + (y — b') = '?.
    (See also 1037.)
    The equation for a in (4143) is
    4171 (a - a') cos a + (6 — 6') sin a +rFw' = 0 .
    PROOF. — Assume (4143) in a , b , r, a , and also in a', b', , a as coinciding
    lines. Then tan a = tan a '; therefore a ' = a or ata . Take the difference
    of the two equations.

    The chords of contact are


    4172 (a - a')(x — a ) + (6 — 6 ) (y - 6 ) + r (r Fr') = 0 ,
    4173 (a -a')(x — a')+ (6 —6')(y — ') + r(1 Fr') = 0,
    with – for exterior tangents, + for transverse .
    Proof. — For these are straight lines, and they pass through the points of
    contact of each pair of tangents respectively, by (4171) .

    The centres of similitude 0 , Q are the intersections of the


    external and transverse tangents respectively.
    4174 Coordinates of o , a'r - ar ' b'r - br '
    r - p ! p -p .
    4175 toQ , a'r + ar'
    Coordinates of b'r + br'
    r + po' ptpt
    PROOF. - By equating coefficients in (4172) and (4142), the polar of 0
    or Q .

    4176 The six centres of similitude of three circles lie on


    582 CARTESIAN ANALYTICAL CONICS.

    four straight lines called axes of similitude. See the figure


    of (1046 ).
    Proof. — The coordinates of the three centres of the forms (4174, '75 )
    will in each case satisfy equation (4083 ).

    41717 The equation of the external axis of similitude is , in


    determinant notation (554 ) ,
    ( 11 53) « — ( 1r2az) y + (14b2a3) = 0 .
    PROOF. — By forming the equation of the right line passing through two
    of the centres of similitude whose coordinates are as in (4174 ) .

    4178 The remaining three axes are found by changing in


    turn the signs of 11, 12; 12, 13; 1'3, 14.
    4179 If one of the circles touches the other two , one axis of
    similitude passes through the points of contact .

    4180 The angle 0 , at which the circle F (x, y) = 0 , radius


    po (Fig . 29), intersects the circle whose centre is hk , and
    radius R is given by the equation
    R - 2Rr cos 0 = F (h , k ).
    Proof : 0 = OQP, R ’ — 2Rr cos 0 + m2 = PT" + post = F (h , k ) + m2,
    ( 702) and (4160) .

    4181 Cor . 1. - If the circles are given by the equations


    x + y2 + 29'& + 2f'y + = 0, * + y2 + 29x + 2fy + c = 0 ,
    the equation for cos 0 becomes, since h = - 1, k = - f,
    2Rr' cos 0 = 2gg' + 2ff" — c - c . (4145)
    4182 Cor. 2. — The condition that the two circles may cut
    orthogonally is
    2gg' + 2ff ' - C - (' = 0 .

    4183 CoR. 3 . — By solving three such equations, we can


    find the circle cutting three given circles orthogonally (4186 ).

    4184 Cor. 4 . — The condition that four 8 f 11


    circles may have a common orthogonal Ci g1 f 1 = 0.
    circle is the determinant equation C2 & 2 f. 1
    Cz fi
    THE PARABOLA . 583

    4185 Cor. 5 . — If the circle 22 + 42 + 2Gæ + 2Fy + 0 = 0 cuts


    three other circles at the same angle 0 , we have, by (4081) ,

    ఇరు
    three equations to determine G , F , C . The resulting deter
    minant equation may be written
    | + yº — X - y 11 10 - 8 -y 1
    C & fi 11 | + 2R cos 0 | " fi 1 || = 0 .
    C. gif 1 r, g2 fii
    Cz & z f; 1 | Trg & f 1
    4186 The first determinant , put = 0 , is the orthogonal
    circle (4183) , and the second , expanded , is the axis of
    similitude.
    4187 The locus of the centre of a circle cutting three given
    circles at equal angles is a perpendicular from their radical
    centre on any of the four axes of similitude.
    Proof. — By eliminating R and cos a between three equations, like (4180).
    4188 Each of these four perpendiculars contains the centres
    of two circles touching the three given circles.
    PROOF. — Consider a = 0 or 180°, in (4180 ) .

    To draw the eight circles which touch three given circles ,


    see (946 ) and (1049) .
    4189 The equation of the fourth degree of two of the
    touching circles is
    23 / S, + 31 S. + 12 / S, = 0 ,
    where 23 signifies the length of the common tangent of the
    second and third circles, & c .
    PROOF. - By first showing that, if four circles are all touched by another
    circle, the relation
    4190 12 . 34 + 14 .23 + 31.24 = 0
    will subsist, and then supposing the fourth circle to reduce to a point.

    THE PARABOLA .
    4200 DEF. — A conic is the locus of a point which moves in
    one plane so that its distance from a fixed point S , the focus,
    584 CARTESIAN ANALYTICAL CONICS. -

    is in a constant ratio (e) to its distance from a fixed right line


    XM (the directrix ) .
    When e = unity, the curve is a parabola. (See also p.248,
    e

    et seq.)

    Equation of the Parabola with origin of coordinates at the


    vertex A .
    4201 y' = 4ax.
    Here a = AS, X = AN ,
    y = PN .
    Proof. - Geometrically ,at (1229).
    Analytically, from
    PS* = y * + (x — a )' = PM ' = (a + a )".
    The equations with the N
    origin at Sand X respectively
    are

    4202 y = 4a (x + a ) ,
    y = 4a (x — a ). (4048 )

    Equations of the tangent at a'y':


    4204 (4120)

    4205 yy' = 2a (x + x').


    4206 y = mx + , m = (4123)

    4207 (4204) is also the polar of any point x'y', by (4124 ).


    Its intercepts are - a ' and y '.

    Equations of the norma

    4208 (4122)
    y — y' = - (* —x ).
    4209 y'x + 2ay - (y{x' + 2ay') = 0.
    4210 y = mx – 2am - ams. (4123)
    THE PARABOLA. 585

    Equation of the parabola


    with a diameter and tangent for
    axes of coordinates.
    4211 y = 4a'x ,
    where <
    4212 a' = a cosec? 0 = SP ;
    = PV ; y = QV.
    PROOF. - Geometrically , at (1239). Otherwise, let VQ = - VQ be equal
    roots of opposite signs of the quadratic (4221), V being the point a'y',
    therefore ya or m3 = ( y * — 4ax') cosec 0 = 4a cosec? 0.x ,
    since y* = 4a x abscissa of P .

    4213 Equations (4204 – 10 ) hold good for these axes, with a


    written for a in each .
    For the polar equation of the parabola , see (4336 ).
    4214 Quadratic for ny : N2, the ratio of the segments into
    which the line joining two given points Xıyı, X , Y , is divided by
    the parabola ya — 4ax = 0 ,
    ni (y ; — 4axz) + 2nın, { y 2 — 2a ( x1 + x ) } + n ; (y; - 4axı) = 0 .
    (4131)

    Equation of a pair of tangents from any point x'y' :


    4215 (y' — 4ax')(yº – 4ax ) = {yy — 2a (x + x') }' = 0.
    The condition for equal roots in (4214).
    Quadratics for the coordinates of the points of contact of
    tangents from a 'y ' :
    4216 ax? — (y's — 2ax') x - ax"? = 0 .
    4217 yº — 2yy' + 4ax' = 0 .
    PROOF. — Solve simultaneously the equations of the curve and the polar
    (4205 ) and (4125 ).

    Coordinates of the point of intersection of tangents at x, yı


    and Xaya :
    4218
    * = PM y = 4714
    4 F
    586 CARTESIAN ANALYTICAL CONICS.

    Quadratic for m of the tangent from x'y':


    4220 mʻx' — my' ta = 0. (4206)

    4221 General polar equation of the parabola , or quadratic


    for r , the segment intercepted between a point, a 'y', and the
    curve on a line drawn from that point at an inclination @ to
    the x axis (4134) ,
    po? sinº0 + 2r (y ' sin 6 — 2a cos ) + y2 - 4ax' = 0 .

    Quadratics for the coordinates of the points of intersection


    of the line Aa + By + C and the parabola y = 4ax : (4116)
    4222 A ’ x2 — 2 (2B -a - AC) x + Cº = 0 .
    4223 Ay? + 4Bay + 4Ca = 0 .

    Length of intercepted chord,


    4224 47 {(Bºa – ACa) (A + B2) } = A ? (4034)

    Equation of the secant through 2191, 2942, two points on


    sec

    the parabola :
    4225 y (yz + y2) = yıya+ 4ax, (4083)
    4226 or y (m + m2) = 2a + 2m ,max .

    4227 The subtangent NT = 2x. Fig. of (4201)


    4228 The subnormal NG = 2a .
    PROOF. - Put y = 0 in (4205 ) and (4208).

    4229 The tangent PT? = 4x (a + x ).


    4230 The normal PG² = 4a (a + x ).
    The perpendicular p from the focus upon the tangent at xy :
    4231 p = Va (x + a) = Vaa'. (4212),(4095)
    THE ELLIPSE AND HYPERBOLA. 587

    The part of the normal intercepted by the curve is equal to


    4a (1 + m %) 4a
    4233
    ma = sin 8 cos (4221),(4135)
    4234 The minimum normal = 6aV3 and m = V2.
    Length of a chord through the focus
    40
    4235 = sind A = 4a'. (4212)

    Coordinates of its extremities, with the focus for origin :


    2a cos 0 2a sin 0
    4237 X = y =
    cos 0 Fl cos 0 FI

    Coordinates of its centre :


    2a cos²0
    4239 X =
    sin A , y = 2a cot 0 .

    THE ELLIPSE AND HYPERBOLA.


    (See also p . 233 , et seq.)

    4250 Referring to the definition (4200) ; when e is less than


    unity , the conic is an ellipse ; when greater than unity , an
    hyperbola .

    Equation of the ellipse with the origin of coordinates at X


    and SX = p .

    4251 y ' + (x - p ) = eⓇx?.


    PraOF. — By the definition in (4200).

    i Abscissæ of vertices : (Supply A ' in the following figure.)

    4252 (4115)
    XA = ite XA' = Le
    588 CARTESIAN ANALYTIOAL OONIOS.

    Is C G NS JA X T EA SN

    4254 Focal distances of vertices :


    (4251)
    SA = e SA' =
    4256 SL = l= ep = a (1 – ) = 6 (4251)
    |

    4260 b = a*(1 –ce); = "?


    4262
    cx = per
    4264 Ca = 1 = a.
    4266 cs = = ae. (4252)

    4268 If b = a tan a , then e = seca in the hyperbola.


    Equation with the origin at A :

    4269 yº = . (2ax — x ) Ell.


    4270
    y = (2aa4a ) Hyp.
    Proof. — By (4200), y + (x - SA) ' = € (a + AX ) , & c.

    Equations with the origin at the centre C :


    4271 y = (a −x ) = (1 - e°)(z − x ).
    5
    (4269)
    THE ELLIPSE AND HYPERBOLA . 589

    4273 + = 1.
    Proof. — By (4200), y* + (x + 08)' = ¢ (2 + 0X )', & c.

    4274 PN : QN :: b : a . (4271)

    G NA ? M

    4275 DEF. — QCN is the eccentric angle, d , of the point P .


    w and y in terms of the eccentric angle :
    4276 x = a cos , y = b sin $ . (Ell.)
    4278 x = a sec $ , y = b tan . (Hyp .)

    Five forms of the equation of the tangent or polar of the


    point a 'y ':
    4280 (4120 )

    4281 m + = 1.
    4282 y = mx + Vaʼmạ+ 6°. (4123)

    4283 * cos + y sin = 1. (Ell.) (4276 )

    4284 x sec® _ y tan = 1. (Hyp.) (4278)


    4285 X cos y ty sin y = va cos' y + 6 sinºy,
    y being the inclination of p. (4054) & (4372)
    590 CARTESIAN ANALYTICAL CONIOS .

    Five forms of the equation of the normal at æ'y':


    (4122)
    4286 y =y = (2 ).
    4287 NY - Y = a*–bº, or ho –ky = a*— ,
    where h and k are the intercepts of the tangent.

    4289 "m (a ’.+ –6 62)m ?


    y = mx — Va’ (4123)

    4290 ax sec - by cosec = a — h?. (4276)

    4291 818 - yıy = (xjx' - yıy'), (4352)


    wherexy, is the extremity of the conjugate diameter.

    Intercepts of the tangent or polar on the axes:


    62
    4292 and (4115 ), (4281)
    4292 er anden (4115) ,(4231)
    Intercepts of the normal: (4287)
    a ' - 62 or
    4294 On the æ axis, 22 X .

    4296 On the yaxis, o may or may.


    Focal distances r , r' of a point xy on the curve :
    4298 (a + ex ) in Ell.
    4299 (ex # a ) in Hyp .
    Proof. — From mi = (ae £ « )' + y*, and (4272).

    Perpendiculars from the foci upon the tangent :


    (4095, 4282)
    4300 p = 6v , p = bv
    THE ELLIPSE AND HYPERBOLA . 591

    4302 : (p. 588) sin SPT = l = (4365)

    4306 b = pp'. . (4300)

    Segments of tangent and normal:

    4307 PT = Voy', Pt leyVol. (4202)


    4309 PG = Vvv = 66, PGʻ = Vrv = 40 .(1294)
    • Right Line and Ellipse.
    Quadratic for the ratio nz :Na, in which the line joining
    two given points XY1, Xay, is cut by the ellipse (4131).
    4310
    2

    (* + *: -1)+2 (eo + –1)+ (*+ % –1) = 0.


    Equation ofthe two tangents drawn from t’y':
    4311 ( +% -1) *+ , -1) -( * -1) =0.
    PROOF. - By the condition for equal roots of (4310 ) .
    592 CARTESIAN ANALYTICAL CONICS.

    Quadratic for abscissæ of points of contact of the tangent


    from x'y' :
    4312 x2 Cbºx” + a+y ?) – 2a’b'xx' + a*(b? — Y?) = 0. (4282,4125)
    Quadratic for m of the tangent from xy :
    4313 mº(dº - aº) - 2mg + ự – ° = 0. (4282)

    General polar equation of the ellipse, or quadratic for r,


    the segment intercepted between the point a 'y ' and the curve
    on the right line drawn from that point at an inclination 0 to
    the major axis and x axis of coordinates.
    4314 (a' sin ’0 + cos²6) ju?
    + 2r (a’y' sin 6 + b+x' cos6 ) + (a’y ? + ? rº? — a’62) = 0 .

    4315 Length of intercepted chord = difference of roots.


    4316 Distance to middle point of chord = half sum of roots.
    4317 Rectangle under segments = products of roots.
    CoR. - If two chords be drawn to a conic at two constant
    inclinations to the major axis , the ratio of the rectangles under
    their segments is invariable.
    For, if a 'y be their point of intersection , the ratio in ques.
    tion becomes aʼsinº0 + b2 cos 0 : aʼsin ' 0' + cos²0', which is
    constant if 0 and 6 are constant.
    Locus of centres of parallel chords :
    4318 . a’y sin 6 + b? x cos 0 = 0 . (4314 )

    Quadratic for abscissæ of points of intersection of the line


    Ax + By + C = 0 and the ellipse V+ + a’y — a^l2 = 0 . (4116 )
    4319 (Aʼaʼ + B ?62) x2 + 2 A Ca’ r + Cʼaʼ – B a 'b2 = 0 .
    4320 x = – ACaº + Baby Aʼaʼ + B+C — C *
    A ’a ' + B -62
    4321 For the ordinates transpose A , B and a , b .
    THE ELLIPSE AND HYPERBOLA. 593

    Length of intercepted chord :


    4322 2ab / (A ’ + B*)(Aʼaʼ + B6° – C2) (4034 )
    A ? q ? + Bºb

    Hence the condition that the line may touch the ellipse is
    4323 A ?q ? + Bºb = C .

    The chord through two pointsXY1, Xayz is


    4324 * 0 7es) y(9479 )= content + 1;
    or, denoting the points by their eccentric angles a,b, the
    chord joining aß is

    4325 cos at + 6 sin at} = cosas


    The coordinates of the pole of the chord or intersection of
    tangents at241,242 (or aß as above) .
    4326 < = XY2+ x2yı — a’ ( - 42) – cos } (a + B )
    y + y2 Xayı— X1Y2 - cos } (a – B )
    4329 = XY2+ x241 – 6*(x1— X ) – 6 sin }(a + b).
    Xi + X2 8 192 - Xzy . cosi (a - B )*

    The following relations also subsist


    a’62 a sin a sin ß b cos a cos B
    4332
    2004 za
    6? x? + a’y2 = a ' — 202 = 62 —
    _ b (sin a + sin B ) _ a (cos a + cos B )
    2y 2x

    " which are of use in finding the locus of (x, y ) when a, ß are
    connected by some fixed equation .”
    (Wolstenholme's Problems, p . 116 .)
    4 G
    594 CARTESIAN ANALYTICAL CONIOS.

    4334 If a, b , y, 8 are the eccentric angles of the feet of the


    four normals drawn to an ellipse from a point xy, then
    a + B + y + 8 = 34 or 51 .
    PROOF. - Equation (4290) gives the following biquadratic in z = tand,
    byze + 2 (ax ta' — 6 %) 20 + 2 (ax - a * + 64) z — by = 0.
    Let a, b, c, d be the roots. Eliminate d from ab + ac + & c. = 0 and abcd = - 1
    (406 ). Thus ab + bc + ca = I + Itl + +ab .; , from which , since a = tanja ,
    bc ' ca
    & c., we get sin (B + y) + sin (y + a ) + sin (a + B ) = 0 ;
    and, since 1 - (ab + ac + & c.) + abcd = 0 ,
    tan } (a + B + + 0) = 0 , ii a + B + y + 8 = 32 or 57.

    4335 The points on the curve where it is met by the


    normals drawn from a fixed point a 'y ' are determined by the
    intersections of the curve and the hyperbola
    aⓇx'y - b*y'x = cºxy. (4287 )

    POLAR EQUATIONS OF THE CONIC .


    The focus S being the pole (Fig . of 4201),the equation of
    any conic is
    4336 r (1 + e cos 6) = 1,
    O being measured from A , the nearest vertex .
    For the parabola , put e = l .
    Proof. —
    r = SP ; 0 = ASP ; 1 = SL ; p = e (SX + SN ) (4200) = l + er cos 0.

    The secant through two points , P , P ', on the curve, whose


    angular coordinates are a + ß and a - ß (Fig . 28 ), is
    4337 r {e cos 6 + sec B cos (a - 6) } = l.
    PROOF. — Let ASQ = a, PSQ = PSQ = ß . .
    Analytically. Take (4109) for the equation of PP . Eliminate r, and r ,
    by (4336 ), and substitute 2a for 0, + 0 , and 2B for 0, - 0 .
    Geometrically . Let PP cut the directrix in Z ; then QSZ is a right angle,
    by (1166 ) . Take C any point in PP' ; SC = r ; ASC = 0 . Draw CD, CE ,
    CF, CG parallel to SL, SP, SQ, SX, and DH parallel to XL. Then
    l = SL = SH + AL.
    STI - SL SD = er cosé. CE – SP – SL - HL - HL
    :: HL = CE = rsin CSF secß = r cos (a — ) sec ß ,
    :: l = er cos ß + r secß cos (a – 0 ).
    THE ELLIPSE AND HYPERBOLA. 595

    The equation of the tangent at the point a is, conse


    quently,
    4338 r {e cos6 + cos (a – O)} = l.
    A Focal Chord.
    21
    4339 Length
    l - e cos' (4336)
    Coordinates of the extremities, the centre C being the
    origin :
    Laeco
    ( s0 I sin 0
    4340
    1 + e cos 8 ' 971 e cos 0°
    4342 The lines joining the extremities of two focal chords
    meet in the directrix . [By (4337 )

    Polar equation with vertex for pole :


    4343 Joe(1 e cos?6) = 21 cos 6. (4200)

    Polar equation with the centre for pole :


    4344 gol (a sinº 6 + 62 cos 6) = a?b ,
    4345 or ✓ (1 - e cos? 8) = b .
    Proof.- By (4273). Otherwise, by (4314),with x' = y' = 0.

    CONJUGATE DIAMETERS.

    B
    1

    TAN T A R

    Equation of the ellipse referred to conjugate diameters for


    coordinate axes :
    4346 + = 1,
    596 CARTESIAN ANALYTICAL CONICS.

    where
    a’ 62 =
    a 72
    4347 a '? =
    aʼsin ’atb cos a ' a ' sinº B + 6 cos B *
    Here a' = CD, b' = CP, a is the angle DCR , and ß the angle
    PCR .

    Proof. — Apply (4050) to the equation (4273), putting w = and


    tan a tan B = - ,by (4351).
    When a' = b', a + b = 1 , and equation (4346) becomes
    4349 x ? + y = a '? = i ( a + ).

    Let the coordinates of D be a ', \', and those of P 2 , y ; the


    equation of the diameter CP conjugate to CD is
    4350 + = 0.
    4351 tan a tanB or mm' = - . (4318)
    æy in terms of a'y', & c.
    4352 x '. Ell.
    vx == - hi y= . Ell.
    4354
    xx == , y = x'. HY D.
    Hyp.
    Proor.– Solve (4350) with (4273).
    4356 x = dR, - , a = pN. (4274,4352)
    4358 x2 + x’? = a’, y ? + yº = b . Ell. (4352)
    4360 X — ”? = a , y - y' = b . Hyp . (4354)
    4362 a’ + = a'? + 62. Ell. (4358)
    4363 a² - = a '? - 6 '? Hyp . (4360)

    4364 a'? = + ex?. (4271, '61)


    4365 6'2 = a - e? x ? = rr'. (4298)
    THE ELLIPSE AND HYPERBOLA . 597

    The' perpendicular from the centre upon the tangent at


    æy is given by
    4366 (4281,4064)

    The area of the parallelogram PCDL (Fig. of 4307) is


    4367 pa' = ab = a'b' sin w ,
    where p = PF, a' = CD, b' = CP, w = LPCD.
    Proof. – From (4366 ), and (4352), and a® = " + ya.

    Other values of pa:


    4369 a%
    (4362)
    a + 62 - 772

    4371 p = a' sinº 6 + 62 cos? 0 . . .


    Proof. — From (4344,'67), putting r = a . .
    4372 p = a cos’y + bºsinºy, (4371)
    y being the inclination of p .
    4373 på = a' (1 - e'sinʼy). (4372,4260)
    Equations to the tangents at P and P', the coordinates of
    D being a ', y':
    4374 xy - ya' = £ab (4073) m = y .

    DETERMINATION OF VARIOUS ANGLES.


    4375 Fig . p. 595. (4356 )

    4377 tan PCD = - (4070, 4352– 3)


    where c = v (a? — 62) = CS.
    те
    598 CARTESIAN ANALYTICAL CONICS.

    1 te cos 0 (407700,, 42
    (40 425566., 49
    4336 )
    4378 tan (SPT) = e sin ,
    where 0 = PST. [See figure on page 588.
    If 4 be the inclination of the tangent to the x axis,
    4380 tany =
    bex e + cos 0 (4280 )
    4380 tany = -being
    a 'y ons
    sin 0 (4280)
    Proof: p = 0+ SPT. Then by (631) and (4379).
    (652, 4378)
    4382 tan SPS = 2*
    262
    4383 tan APA = tan CP
    4383 tan APA'= - ae- y tan CPG = egy
    If OP, OP'are tangents to an ellipse,
    4385 cos POP' = COʻ - a' - 63
    OS . OS
    PROOF. - By figure and construction of (1180), POP' = MOS . Therefore
    CO DOP - OS* + 08" - SM _ 200° + 2CS2 — 4a' - en
    208. OS 208 . OS

    If x', y' are the coordinates of O ,


    4386 tan POP' = 2V ( *x ? + a’y ? — a’b )
    x + y - a - 64
    PROOF. - By (4311), taking termsof the second degree for the two parallel
    lines through the origin and tan ø from (4112 ).

    It is worthy of remark that the substitutions (4276 – 8 )


    may also be usefully employed when the axes of reference are
    conjugate diameters : though , in that case, the geometrical
    signification of o no longer exists .
    THE HYPERBOLA . - 599

    THE HYPERBOLA
    REFERRED TO THE ASYMPTOTES.

    4387 xy = (a + 62).
    Proof. — By (4273) and (4050 ). Here x = CK , y = PK .

    Equations of the tangent at P , (u', y').


    4388 Xy' + x'y = } (a’+ 6°). (4120)
    4389 y = mx + Vm (a’ + 6°). (4123 )

    4390

    4391 Intercepts on the axes Cl = 2x', CL = 24'.


    THE RECTANGULAR HYPERBOLA.
    4392 Here a = b , e = v2 ; and the equation with the
    ordinary axes is
    4393 23 = a. (4273)
    4394 Tangent xx' — yy' = a . (4281)
    600 OARTESIAN ANALYTICAL CONICS.

    Equation with the asymptotes for axes :


    4395 2xy = a . (4387)
    4396 Tangent xy' + x'y = a '. (4388 )

    THE GENERAL EQUATION .


    The general equation of the second degree is
    4400 ax ? + 2hxy + by® + 2gæ + 2fy + c = 0 ,
    4401 or ax® + by + cz2 + 2fyz + 2gzx + 2hxy = 0, with z = 1.
    The equation will be denoted by u or + (x , y) = 0 .
    THE ELLIPSE AND HYPERBOLA.
    When the general equation (4400 ) , taken to rectangular
    axes of coordinates, represents a central conic , the coordinates
    of the centre, Oʻ (Fig . 30), are
    4402 ab - h = 7 (4665)
    Proof. — By changing the origin to the point x'y' and equating the new
    g and f each to zero (4048 ). .
    For the case in which ab = h ', see (4430 ) .

    4404 The transformed equation is ax? + 2hxy + byº + c' = 0 ,


    4405 where c = ax'? + 2hx'y' + by'? + 2gx' + 2fy' + c.
    4406 = gx' + fý + c.
    abc + 2fgh - af2 - bgol - cha ( 4466 )
    4407 ab — h
    The inclination 6 of the principal axis of the conic to the
    & axis is given by
    4408 tan 28
    20 = 2h
    -6
    PROOF. — (Fig . 30.) By turning the axes in (4404 ) through the angle 8
    (4049) and equating the new h to zero.
    THE GENERAL EQUATION . 601

    The transformed equation now becomes


    4409 a'x * + b'y? + 6 = 0,
    4410 in which a' = } {a + 6 + V 4h + (a − b)"} ,
    in

    4411 b = } {a + 6 – V4k’ + (a − b)"},


    a' and b' are found from the two equations
    4412 a' + b = a + b, a'b' = ab - k”. (See (4418).
    The semi-axes and excentricity are
    11 - YAS

    4414 V - V - , and e= (1-6 ).(1273)(1261)


    For the coordinates of the foci, see (5008).

    4416 Note . — If 6 be the acute angle determined by equation


    (1408),we have tochoose between 0 and 0 + 5 for the inclina
    tion in question , since tan 20 is also equal to tan (20 + ).
    RULE .* — For the ellipse , the inclination of the major axis
    to the x axis of coordinates will be the acute angle 0 or 0 + ,
    according as h and c' have the same or different signs. For the
    hyperbola , read “ different or the same."
    PROOF.— Let the transformed equation (4409 ) be written in terms of the
    semi-axes p , q ; thus q *x ? + pʻy : = p q ', representing an ellipse. Now turn
    the axes back again through the angle – 7, and we get
    : (q* cos' 0 + på sin ) x - (p * — q°) sin 20 xy + (qº sin 0 + p² cos? ) y = pʻq .
    Comparing this with the identical equation (4404), ax? + 2hxy + by: = - 0 ,
    we have ( p - q°) sin 20 = - 2h, p ?q = - ;
    : sin 20 = 2 people Hence ® is <
    when h and c' have the same sign, p being > q. A similar investigation
    applies to the hyperbola by changing the sign of q ?.

    • This rule and the demonstration of it are due to Mr. George Heppel, M .A ., of
    Hammersmith .
    4
    602 CARTESIAN ANALYTICAL CONICS.

    INVARIANTS OF THE CONIC.


    44117 Transformation of the origin of coordinates alone
    does not alter the values of a, h, or b , whether the axes be
    rectangular or oblique. This is seen in (4404).
    When the axes are rectangular, turning each through an
    angle 0 does not affect the values of
    4418 ab - hº, a + b , gol + f , or c.
    When the axes are oblique (inclination w ), transformation
    · in any manner does not affect the values of the expressions
    4422 ab - h” and a + b - 2h cosw
    sin ' w sinºw
    These theorems may be proved by actual transformation by the formulæ
    in (4048 -50 ). For othermethods and additional invariants of the conic, see
    (4951).

    4424 If the axes of coordinates are oblique, equation (4400)


    is transformed to the centre in the same way, and equations
    (4402 - 6 ) still hold good . If the final equation referred to axes
    coinciding with those of the conic be
    4425 a 'x ? + b'y ' + 6 = 0 ,
    and o the inclination of the new axis of a to the old one, we
    shall have d' unaltered ,
    4426 tan 20 = 5 2h sin w - a sin 2w
    2h cos w - a cos 2w - b '
    4427
    ' _ a + b - 2h cosw + vQ. K _ a + b - 2h cosw - VQ .
    2 sin ’ w 2 sin w
    where Q = a + 52 + 2ab cos 2w + 4h ( a + b ) cos w + 4h? .
    PROOF. — (4404 ) is now transformed by the substitutions in (4050 ) ,
    putting B = 0 + 90°, and equating the new h to zero to determine tan 28 .
    a 'and b' are most readily found from the invariants in (4422). Thus, putting
    the new h = 0 and the new w = 90°,
    z a + b - 2h cos w ondok _ ab - h *
    a +b = sin ' w
    sin w
    equ ati ons ch
    whi det erm ine a ' and b'.
    THE GENERAL EQUATION. 603

    The eccentricity of the general conic (4400) is given by


    the equation
    4429 (a + b - 2h cosw) _ 4 .
    1 - e? (ab - hº) sin ' w
    PROOF. - By (4415), and the invariants in (4422).

    THE PARABOLA.
    4430 When ab - h2 = 0 , the general equation (4400 ) repre
    sents a parabola .
    For a , y' in (4402) then become infinite and the curve has
    no centre, or the centre may be considered to recede to
    infinity .
    Turn the axes of coordinates at once through an angle 0
    (4049), and in the transformed equation let the new coeffi
    cients be a', 22', b', 20', 2f', c'. Equate h' to zero ; this gives
    (4408 ) again, tan 20 = a -" 6 If o be the acute angle deter
    mined by this equation , we can decide whether 0 or 0 + 1 is
    the angle between the x axis and the axis of the parabola by
    the following rule .
    4431 RULE . — The inclination of the axis of the parabola to
    the x axis of coordinates will be the acute angle 0 if h has the
    opposite sign to that of a or b , and 0 + ìr if it has the same
    sign .
    Proof.-- -Since ab - h * = 0, a and b have the same sign. Let that sign
    be positive, changing signs throughout if it is not. Then, for a point at
    infinity on the curve, x and y will take the same sign when the inclination is
    the acute angle 0, and opposite signs when it is 0 + . But, since
    ax: + by = + 00 , we must have 2hxy = - 00 , the terms of the first degree
    vanishing in comparison. Hence the sign of h determines the angle as stated
    in the rule .

    со
    4432 sin 0 = Vato cos = Vote
    PROOF. – From the value of tan 20 above, o being the acute angle obtained ,
    and from h = ab .

    4434 Also a' = 0 and b ' = atb.


    For a 'b ' = ab - h = 0 , and we ensure that a' and not b ' vanishes by
    (4431) . Also a ' + b ' = a + b (4412) .
    604 CARTESIAN ANALYTICAL CONICS.

    4436 gʻ = g cos 6 + f sin 0 = & Va + fvb.


    ✓ (a + b )
    4438 f' = f cos0 – g sin 0 = Vam
    4440 But if h has the same sign as a and b , change 0 into
    0 + . (4431 )
    PROOF. — By (4418 , 4432– 3).

    The coordinates of the vertex are

    4441 = PREMIO “=
    Obtained by changing the origin to the point a 'y' and equating to zero
    the coefficient of y and the absolute term . The coefficient of v then gives
    the latus rectum of the parabola ; viz. :
    2 & va +fb
    4443 L=
    =
    ✓ (a + b)***
    (4437)

    METHOD WITHOUT TRANSFORMATION OF THE AXES.


    4445 Let the general equation (4400) be solved as a quad
    ratic in y . The result may be exhibited in either of the forms

    4446 y = ar + B + VV ( — 2px + q ),
    4447 y = ar + B + V + {(x - p) + (9 – pp)},
    4448 y = ar +BEVM (x − y)(x – 8),
    4449 where a = - B= - - = " zab.
    hf - bg
    4452
    4452 p="a - h2
    ab or C,(1642) = abax =- h2 or 4
    44549 –pe= 6(abc+ 24th byge–ck") =4,
    4456 y and 8 = p + v (p2 - 9).
    4458 Here y = ax + ß is the equation to the diameter DD
    THE GENERAL EQUATION . 605

    (Fig . 31), y and 8 are the abscissæ of D and D', its extremities,
    the tangents at those points being parallel to the y axis. The
    surd = PN = P ' N when æ = OM . The axes may be rect
    angular or oblique .

    mes
    When ab - h² = 0 , equation (4446 ) becomes
    4459 y = mx + B + Va – 2p’x,
    4460 where p = bg – hf, (' = f - bc.
    4462 In this case, is the abscissa of the extremity of
    the diameter whose equation is y = ax + ß and the curve has
    infinite branches .

    RULES FOR THE ANALYSIS OF THE GENERAL EQUATION.


    First examine the value of ab - h², and , if this is not zero ,
    calculate the numerical value of c' (4407), and proceed as in
    (4400 ) et seq. If ab - h2 is zero, find the values of p ' and q '
    (4459) . The following are the cases that arise .
    4464 ab - hº positive— Locus an ellipse .
    Particular Cases.
    4465 A = 0 — Locus the point x'y'.
    See (4402). For, by (4404), the conjugate axes vanish .
    4466 bA positive — No locus.
    By (4447 –54), since q - pa is then positive.
    4467 h = 0 and a = b - Locus a circle .
    By (4144 ). In other cases proceed as in ( 4400 – 14 ).

    4468 ab - hºnegative — Locus an hyperbola.


    Particular Cases .
    4469 A = ( – Locus two right lines intersecting in the
    point a'y '.
    By (4447), since q - pe then vanishes. In this case solve as in (4447).
    606 CARTESIAN ANALYTICAL CONICS.

    4470 bA negative - Locus the conjugate hyperbola .


    Cus

    4471 a + b = 0 – Locus the rectangular hyperbola.


    By (4414), since a ' = - b'.
    4472 a = b = ( – Locus an hyperbola , with its asymptotes
    xe
    parallel to the coordinate axes . The coordinates of the centre
    are now - and - , by (4402). Transfer the origin to
    are

    the centre, and the equation becomes S

    4473

    In other cases proceed as in (4400 - 14 ).

    4474 ab - h² = 0 Locus a parabola .


    Particular Cases.
    4475 p = 0 – Locus two parallel right lines. By (4459).
    4476 p = q = 0 – Locus two coinciding right lines.
    By (4459).
    4477 p = 0 and q negative — No locus.
    By (4459). In other cases proceed as in (4430 -43).

    Ex. 1. 22° – 2xy + y*+ 3x - y - 1 = 0.


    re
    Here the values ofa,h,b,9,f,c are respectively 2, – 1, 1, 3 - , - 1,
    ab - h2 = 1 ; d' = aA _ abc + 2fghab- af:
    - h?
    – bgº — ch _ (4406 )

    The locus is therefore an ellipse, none of the exceptions (4465– 7) occurring


    here. The coordinates of the centre, by (4402), are
    ~' = hf - b9 = - 1, sý = gk - af - - 1
    Hence the equation transformed to the centre is
    2x* –224+ y2– 2 = 0.
    Turning the axes of coordinates through an angle 0 so that tan 20 = - 2
    (4408 ), we find the new a and b from
    a' + b' = 3, a'b' = l ; (4412)
    THE GENERAL EQUATION. 607

    therefore a ' = ] (3 – ✓5), b = } ( 3 + ✓5 ),


    and the final equation becomes 2 (3 - V5) * + 2 (3 + 75) y = 9.
    The inclination of the major axis to the original x axis of coordinates is
    the acute angle į tan - ( - 2 ) , by the rule in (4416 ) .

    Ex. (2) : 12x* + 60xy + 75y9 — 122 – 8y — 6 = 0 .


    The values of a , h , b, g , f, c are respectively 12 , 30 , 75, - 6 , - 4 , - 6 ,
    ab - h2 = 0 ; p ' = bg - hf = - 330 ; ( = f* — bc. (4460)
    Since p does not vanish (4475– 7), the locus is a parabola . Proceeding,
    therefore , by (4430 –43) , we have

    tan 20 = 2*6 = sin0 = * coso= 28 (4432)


    By the rule (4431), wemust take 0 + jr for the angle, instead of 0 . There
    fore g' = - g sin 0 + f cos 0 = 9°TIV°
    ✓ (a + b ) ✓29 '
    f' = - f sin 6 - 0 cos A = - f / b - gva -
    ✓ (a + b )
    and b' = a + b = 87 (4435).
    Consequently the transformed equation is
    44 64
    87y + ook + -
    ✓ 297 – 6 = 0 .
    The coordinates of the vertex are computed by (4441) , and the final
    equation with the vertex for origin is yü = 799 .

    Ex. (3) : 2 *+ 6xy + 9y* + 52 + 15y + 6 = 0.5 15


    The values ofa, h, b,g, f, care respectively 1, 3,9, 5 , 5 , 6 ,
    ab - h2 = 0 and p ' = bg - hf = 0,
    therefore, by (4475 – 7 ), if there is a locus at all, it consists of two parallel or
    coinciding lines . Solving the equation therefore as a quadratic in y , we
    obtain it in the form (x + 3y + 2 ) (x + 3y + 3) = 0 ,
    the equation of two parallel right lines .

    The equation of the tangent or polar of a'y' is


    4478 Upx + uyy + uz % = 0 or Upx ' + u , y' + ua' = 0 ;
    (4401, 1405) obtained by (4120 ) in the form
    4479 (aa' thy' + g ) x + (hx' + by' + f ) y + gx' + fy' + c = 0 ,
    4480 or (ax + hy + g ) a ' + (hx + by + f ) y + gx + fy + c = 0 ,
    608 CARTESIAN ANALYTICAL CONICS.

    4481 or
    axt' + h (xy' + x'y) + byy'+ g (x + x') + f(y + y) + c = 0.
    When the curve passes through the origin ,the tangent at
    the origin is
    4482 gx + fy = 0. (4479 )

    And the normal at the same point is


    4483 fx - gy = 0 .
    4484 Intercepts of the curve on the axes ,

    4486 Length of normal intercepted between the origin and


    1or

    the chord
    V gé + f2
    atb (4483
    (4483--44)
    )

    Right Line and Conic with the general Equation .


    4487 Quadratic for n : n ', the ratio in which the line joining
    æy, a 'y' is cut by the curve.
    Let the equation of the curve (4400 ) be denoted by
    $ ( x , y ) = 0 , and the equation of the tangent (4479) by
    ( , 4 , a', y ) = 0 ; then the quadratic required will be found,
    by the method of (4131) , to be

    n°° (ex ', y') + 2nn'y ( x, y, x ', y') + n°¢ ( x , y) = 0 .


    The equation of the tangents from x'y' is
    4488 $ (x', y') $ (x, y) = {4 (x,y, x', y)} .
    Proof. — By the condition for equal roots in (4487 ).
    Cor. — The equation of two tangents through the origin is
    4489 Bx — 2Hxy + Cy2 = 0 . (4665)

    The equation of the asymptotes of u (4400) is


    4490 au , + 2hu ,uy + bus = 0 .
    THE GENERAL EQUATION . 609

    The equation of the equi- conjugates of the conic


    ax: + 2hxy + by2 = 1 is
    4491 (a + b)(ax? + 2hxy + by®) = 2 (ab —hº)(x + + y?).
    Proof. — When the conic is ax ' + by’ = 1, the similar equation is
    (a + b ) (ax + by ?) = 2ab (ix2 + y ) or (ax? — by*) = 0 ,
    given by the intersections of the conic and a circle. Transformation of the
    axes then produces the above by the invariants in (4418 ) .

    4492 When the coordinate axes are oblique, the equation


    becomes

    (a − b)(ax* — by®) + 2x (hx +by)(h — a cosw ) + 2y (ax + hy)(h — b cos w ) = 0.

    General polar equation :


    4493 (a cos? 0 + 2h sin 0 cos 0 + 6 sin’ ) po?
    + 2 (g cos 6 + f sin O)r + c = 0.
    Polar equation with (x , y) for the pole : (4134 )

    4494 (a cos 0 + 2h sin 0 cos 6 + 6 sin ’ ) po?


    + 2 {(ax + hy + g) cos 6 + (by + hx + f ) sin 0} r + F (xy) = 0.
    Equation of the line through a'y parallel to the conjugate
    diameter :
    4495 (« — x')(ax' +hy' + 8 ) + (y — y )(ha' + by' + f) = 0 .
    Proof. - By the condition for equal roots of opposite signs (4494 ).

    Equation of the conic with the origin at the extremity of


    the major axis, L being the latus rectum .
    4496 y = Lx - ( 1 - e ) x ? . (4269, '59)

    Equation when the point ab is the focus and


    Ax + By + 0 = 0 the directrix :

    4497 ✓ {(x — a)2+ (7 - 6) } = e 4x + By + C. (4200, 4095)


    V {A + B }
    4 1
    610 CARTESIAN ANALYTICAL CONICS.

    INTERCEPT EQUATION OF A CONIC.


    The equation of a conic passing through four points whose
    intercepts on oblique axes of coordinates are 8, s' and t, t', is

    4498 +2hay+ * - ( + +) - (4 + 7)+ 1 = 0.


    Equation of a conic touching oblique axes in the points
    whose intercepts are s and t :

    4499 + 2hay+ 24 –24+1= 0,


    4500 or (i + 1 - 1) +vxy = 0.
    Comparing with the general equation (4400),we have
    4501 s = - =- = 2h– = 2 hetents
    Perpendicular p from ay, any point on the curve, to the
    chord of contact :
    vs’try sin ’w (4096 , 4500 )
    4505 $ + ť — 2st cosw '

    Equation of the tangent ata’y':


    4507 26 + - 1) 6 + - 1) +v (xy'+.xʻy) = 0.
    4508 The equation of the director-circle is
    (1 + istv ) (x + + y * + 2xy cosw ) - h (x + y cosw ) - k (y + ä сos w ) + hk cos w = 0 .

    The parabola with the same coordinate axes as in (4499):


    4509 ( + 4 -1) = ahoy or V +V1= 1.
    Proof.–From (1500),putting he - (1174),and therefore r =
    THE GENERAL EQUATION . 611
    611

    Equation of the tangent at «', y' :


    4510 Testy + vilu = 1, (4509)
    mst
    4511 or y = mx + (4123)
    4511 or y = mx+ mmastym
    ms + ť = - (4123)
    Equation of the normal at x’y':
    (4122)
    4512 y = mx+ mavie
    4513 Normal through the origin wis = y / t.
    The equations of two diameters are, with any axes,
    4514 _ = 1 and *- = - 1.
    Colt

    PROOF. — Diameter through 0t, by the property OR = RQ , in


    the figure of (4211).

    Coordinates of the focus :


    st st
    10 * = 5 + t + 2st cosw y = 52 + + 2st cos w (300
    Equation of the directrix :
    4518 x ( s + t cos w ) + y (t + s cos w ) = st cosw .
    Proof. — Expand (4509), and form the equation of the polar of the focus
    by (4479) and (4516) .
    When the axes are also rectangular, the latus rectum
    43242
    4519 L - (4095, 4516 - 8 )

    4520 Locus of the centre of the conic which touches the


    axes at the points s0 , 0t :
    tx = sy . (4500, 4402)
    4521 To make the conic pass through a point a'y'; substi
    tute a 'y ' in (4500 ) , and determine v .
    612 CARTESIAN ANALYTICAL CONICS.

    SIMILAR CONICS.

    4522 DEFINITION . - If two radii, drawn from two fixed


    points , maintain a constant ratio and a constant mutual
    inclination , they will describe similar curves.
    4523 If the proportional radii be always parallel , the curves
    are also similarly situated .

    If there be two conics (1) and (2), with equations of the


    form (4400 ) , then

    The condition of their being similar and similarly


    situated is
    h 6
    4524

    Proof.— By (4404), changing to polar coordinates, r : r' = constant.


    The condition of similarity only is
    ( a + b ) _ (a ' + b') ? .
    4525 · - ab = 12 - a'í' : (4418 - 9 )

    or,with oblique axes ,


    AROC (a + b - 2h cos w )? (a ' + 6 - 2h' cos w )
    4526 h - ab h ? - ab
    (4422 – 3 )

    CIRCLE OF CURVATURE .

    CONTACT OF CONICS.
    4527 DEF. — When two points of intersection of two curves
    coincide on a common tangent, the curves have a contact of
    the first order ; when three such points coincide, a contact of
    the second order ; and so on . To osculate, is to have a con
    tact higher than the first.

    4528 The two conics (Fig . 32) whose equations are


    ax' + 2h xy + by + 2g x = 0..................(1),
    a’xº + 2h'xy + b'y? + 20'x = 0 ............... ...(2 ),
    CIRCLE OF CURVATURE. 613

    touch the y axis at the origin , 0 , by (4482). Eliminate the


    third terms from ( 1 ) and ( 2 ) , and we obtain a = 0 , the line
    through two coincident points, and
    4529 (ab’— a'b) x + 2 (hb' — h'b) y – 2 (bg'— b'g) = 0 ,
    the equation to LM ,the line passing through the two remain
    ing points of intersection of ( 1) and ( 2 ). (4099)

    Again , eliminate the last terms from ( 1) and (2), and we


    obtain
    4530 (agʻ – a'g) x2 + 2 (hg' — h'g) xy + (bg' - b'g ) yº = 0,
    the equation of the two lines OL, OM . [ By (4111) and (4099)
    4531 If the points L , M coincide, the conics have contact
    of the first order. The condition for this is that (4530) must
    have equal roots ; therefore
    (ugʻ – a'g )(bg' — b'g ) = (hg' — h'g )”.
    4533 If the conics ( 1 ) and ( 2 ) are to osculate, M must
    coincide with 0 . Therefore, in (4529) , bg' = b' g.
    If in (4532) bg' = bg, the conics have a contact of the
    third order.

    CIRCLE OF CURVATURE .
    (See also 1254 et seq.)
    The radius of curvature at the origin for the conic
    ax? + 2hxy + by® + 2gx = 0,
    the axes of coordinates including an angle w , is
    4534 p = - b sino
    PROOF. — The circle touching the curve at the origin is
    * + 2xy cosw + y? — 2ræ sin w = 0,
    by (4148) , and the geometry of the figure, 2r sin w being the intercept on
    the x axis. The condition of osculating (4533 ) gives the value of p .
    p is positive when the convexity of the curve is towards the y axis.

    Radius of curvature for a central conic at the extremity


    P of a semi-diameter a ', the conjugate being b .
    614 CARTESIAN ANALYTICAL CONICS.

    2 2' 6 '2 72
    4535 pasin w PP (4367)
    Proof. — Take the equation and figure of (4346) ( a' = CP). Transform
    to parallel axes through P . Then by (4534).
    The same in terms of x , y ,the coordinates of the point P .
    4539 = (b*x * + a *yº)
    a ’62
    Proof. — By (5138 ), or from (4538) and the value of b at (4365).
    The coordinates of the centre of curvature O for P , the
    point xy, are
    ee

    4540 ferming n= com where e = x²– .


    PD yn y
    Proof.— (Fig. 83.) From OP = * = pc and CP = " " = BG
    with the values of p, PG, and PG at (4535) and (4309).

    Radius of curvature for the parabola.


    Taking the diameter and tangent through the point for
    axes,
    2a 2a 25p2
    4542 p = sino = sino = SY (Fig . of 4201)

    By (4534 ), and equation (4211) .

    Coordinates of the centre of curvature at wy (rectangular


    axes) :
    4545 $ = 3x + 2a, n = -*
    Proof. – From y - 1 : p = y : PG and p = 2a coseci0 , PG = 2a cosec 8
    and y = 2a cot 0 .

    The evolute of a central conic (Fig. 33) :


    4547 (ax)} + (by )} = (a' — 62) },
    4548 or (aʼm ? + b*y* — c*)$ + 27a²b'c*x*y * = 0 ,
    where c = a ' — 62.
    CONFOCAL CONICS. 615

    PROOF. — Substitute for x , y in the equation of the conic (4273) their


    values in terms of ļ, n from (4540) . Otherwise as in (4958 ), or by the
    method of (5157 ) .
    The curve has cusps at L , H , M , and K .
    The evolute of the parabola :

    4549 a(*) = 1* 324)


    Proof. — As in (4548), from the equations (4201) and (4545).

    CONFOCAL CONICS.

    4550 + = 1 and +*=1


    are confocal conics, if
    a - a ” = 62 — 6'2,
    or the sign of b'? may be changed .
    For the confocal ofthe general conic, see (5007).

    4551 Confocal conics intersect, if at all, at right angles.


    PROOF. — If u , u ' are the two conics in (4550), changing the sign of 6" to
    make the second conic an hyperbola , u - u = 0 will be satisfied at their point
    of intersection ; and (by a? - a^ = 12 + 69) this proves the tangents at that
    point to be at right angles (4078, 4280 ) .
    Otherwise geometrically by (1168) .
    4552 Tangents from a point P on one conic to a confocal
    conie make equal angles with the tangent at P . [Proofat (1291)

    4553 The locus of the pole of the line Ax + By + C with


    respect to a series of confocal conics in which a - 62 = , is
    the right line perpendicular to the given one,
    BCx - ACy + ABX = 0 .
    Proof. — The pole of the line for any of the conics being wy; Aa’ = - C'x
    and Bb% = - Cy (4292) ; also a - 6° = 1. Eliminate a ’ and / ?.
    4554 COR . - If the given line touch one of the conics, the
    locus is the normal at the point of contact.
    616 CARTESIAN ANALYTICAL CONICS.

    4555 Graves' Theorem . — The two tangents drawn to an


    ellipse from a point on a confocal ellipse together exceed the
    intercepted arc by a constant quantity.
    Proof. — (Fig . 132.) Let P , P be consecutive points on the confocal from
    which the tangents are drawn. Let fall the perpendiculars PN, P ' N '. From
    (1291) , it follows that ZPP ' N = P 'PN', and therefore P ' N = PN'. The
    increment in the sum of the tangents in passing from P to P ' is
    RR' — QQ' + P ' N - PN' = RR - QQ'.
    But this is also the increment in the arc QR, which proves the theorem .
    4556 If the tangents are drawn from a confocal hyperbola ,
    as in (Fig . 133 ), the difference of the tangents PQ , PR is
    equal to the difference of the arcs QT, RT.
    The proof is quite similar to the foregoing.

    4557 At the intersection of two confocal conics, the centre


    of curvature of either is the pole of its tangent with respect
    to the other.
    Proof. — Take + = 1 (i.) and * - = 1 (ii.)for confocalconics .
    At the point of intersection , a " = " and y ' = - (where c = a ’– 6%) ;
    by a — a” = b’ + b'?. The coordinates of the centre of curvature of a'y' in
    (i.) are a ” = 2 , 4" = - % (4540- 1). The polar of this point with
    respect to (ii.) will be + = 1. Substitute the values of 2” ,y ";and
    we see, by the values of x ', y', that this is also the tangent of (i.) at P .
    4558 A system of coaxal circles (4161), reciprocated with
    respect to one of the limiting points D or D', becomes a
    system of confocal conics.
    PROOF. — The origin D is one common focus of the reciprocal conics, by
    (4844 ). The polar of D with respect to any of the circles is the same line,
    by (4166). D and its polar (both fixed ) reciprocate (1858 ) into the line at
    infinity and its polar, which is the centre of the conic. The centre and one
    focus being the same for all, the conics are confocal.
    ANALYTICAL CONICS IN

    TRILINEAR COORDINATES.

    THE RIGHT LINE .

    For a description of this system of coordinates, see (4006 ) .


    The square of the distance between two points aßy, a 'B ' y is ,
    with the notation of (4008 ),
    4601

    4602
    = a a cosA (a – a')' + b cos B (B - B )* + c cos C (y - 7 ) }.
    Proof. — Let P, Q be the points. By drawing the coordinates 13y , B'y',
    it is easily seen, by (702), that
    PQP = [(B - B ') + (y - 7 ) + 2 ( – 6 ') (y - 7') cos A ] cosec’A ......(1).
    Now , by (4007 ), a (a - a ') + 6 (13 — ') + ( - 4 ) = 0,
    from which 6 (B - B ')? = - a (a - a') ( - B ') - c (B - B') (y - y') ,
    and a similar expression for c (y - 7 ' ! Substitute these values of the square
    terms in ( 1), reducing by (702).
    Coordinates of the point which divides the straight line
    joining the points apy, a'B ' y in the ratio 1 : m :
    latma B + mß ly + my. By (4032).
    4603

    ABC being the triangle of reference , and a = 0, B = 0,


    y = 0 the equations of its sides, the equation of a line passing
    through the intersection of the lines a = 0 , B = 0 is
    4604 la - mß = 0 or a - kB = 0 .
    618 TRILINEAR ANALYTICAL CONICS.

    PROOF. — For this is the locus of a point whose coordinates a , ß are in the
    constant ratio m : 1 or k (4099) .
    When I and m have the same sign, the line divides the external angle C
    of the triangle ABC ; when of opposite sign, the internal angle C .

    The general equation of a straight line is


    4605 la +mB + ny = 0 ,
    and it may be referred to as the line (l, m , n ).
    Proof : la tmp = 0 is any line through the point C, and ( la + mp) + ny
    = 0) is any line through the intersection of the former line and the line y = 0
    (4604) , and therefore any line whatever according to the values of the arbi
    trary constants l, m , n .

    The same straight line in Cartesian coordinates is


    4606 (l cosatm cos B + n cos y ) x
    + (l sin a tm sin ß + n sin y) y - (lpa + mpat nps) = 0.
    Proof. — By substituting the values of a , b , y at (4009).
    Or, if the equations of the sides of ABC are given in the
    form 4 ,2 + By + C = 0 , & c ., the line becomes

    4607 (1A , + mA, + n A3) x + (1B , + mB, + n B3) y


    + IC , + mCz + nCz = 0 .
    Proof. — By (4095), the denominators like vedi + B ;) being included in
    the constants l, m , 1 .

    4608 If u = 0, v = 0 , w = 0 are the general equations of


    the lines a , ß , y , then it is obvious that lu + mv = 0 is, like
    (4604 ) , a line passing through the intersection of u and v, and
    lu + mv + nw = 0 represents any straight line whatever.
    To make an equation such as a = p ( a constant) homo.
    geneous in a , b , y ; multiply by the equation & = da + bB + cy
    (4007) , thus
    (ap - ) a + brß + opy = 0 ,
    which is of the same form as ( 1605 ).
    THE RIGHT LINE. 619

    4610 The point of intersection of the lines


    la + mB + ny = 0 and l'a + m 'ß + n'y = 0
    is determined by the ratios
    anamnen Brie mi i'm and (4017).
    The values of a, ß, y are therefore
    41
    4611 E (mn' — m 'n ) & (nľ — n'l) E (lm ' — lm )
    D D
    whereke
    D = a (mn' — m 'n ) + b (nľ' — n 'l) + (Im ' — lm ).
    Proof. — By (4017), or by solving the three equations
    aa + 66 + cy = E, la + mp + ny = 0, l'a + m 'ß + n 'y = 0 .

    The equation
    4612 aa + bB + ry = 0 or a sin A + B sin B + y sin C = 0
    represents a straight line at infinity.
    Proof. — The coordinates of its intersection with any other line
    la + mß + ny = 0 are infinite by (4611) .
    4613 NOTE : aa + b3 + cy = 2 , a quantity not zero. The equation
    aa + bß + cy = 0 is therefore in itself impossible , and so is a line infinitely
    distant. The two conceptions are, however, together consistent; the one
    involves the other. And if, in the equation la + mß + ny = 0 , the ratios
    l : m : n approach the values a : 6 : 4, the line it represents recedes to an
    unlimited distance from the trigon .
    4614 The equation corresponding to (4612) in Cartesian coordinates is
    Ox + Oy + C = 0 , the intercepts on the axes being both infinite. Cartesian
    coordinates may therefore be regarded as trilinear with the x and y axes for
    two sides of the trigou and the other side at an infinite distance.

    4615 The condition that three points a Bir


    a ßiri, azß272, azß : 73 may lie on the same an B Y = 0.
    straight line is the determinant equation ,
    Proof. — For it is the eliminant of the three simultaneous equations,
    la , + mß, + , = 0, lag + mß, + nyo lag + mß + nys = 0 . (583)
    620 TRILINEAR ANALYTICAL COVICS.

    4616 COR. -- The above is also the equation of a straight line


    passing through two of the fixed points if the third point be
    considered variable .
    46117 Similarly , the condition that the three following
    straight lines may pass through the same point, is the deter
    minant equation on the right,
    la + miß + ry = 0
    lya + m ,ß + nzy = 0
    lza + mzB + ngy = 0 , Ils mz ng

    4618 The condition of parallelism of the two straight lines


    latmB + ng = 0 , I m n 1
    l'a + m 'ß + n'y = 0 , | mn | = 0,
    is the determinant equation la C |

    Proof.— By taking the line at infinity (4612) for the third line in (4617).
    4619 Otherwise the equations of two parallel lines differ by
    a constant (4076 ). Thus
    la +mB + ny + k (aa +bB + ry ) = 0 (4007)
    or (l + ka ) a + (m + kb) B + (n + ke) y = 0
    represents any line parallel to la + mB + ny = 0 by varying
    the value of k .

    The condition of perpendicularity of the two lines in


    (4618) is
    4620 il' + mm ' + nn' — (mn' + m 'n ) cos A - (nľ + n 'l) cos B
    – (lm ' + lm ) cos C = 0,
    4621 or l' (l — m cos C — n cos B ) + m ' (m - n cos A - l cos C )
    + n ' (n - 1 cos B - m cos A ) = 0 .
    PROOF. — Transform the two equations into Cartesians, by (4606 ), and
    apply the test A A ' + BB' = 0 (4078) , remembering that
    cos (ß - y ) = - cos A , & c. ( 4011) .
    THE RIGHT LINE . 621

    When the second line is AB or y = 0 , the condition is


    4622 n = m cos A + lcos B .
    It also appears, by (4676), that (4620) is the condition that
    the two lines may be conjugate with respect to the conic
    whose tangential equation is
    4623 P + mº+ n ? – 2mn cos A — 2nl cos B - 21m cos C = 0.
    The length of the perpendicular from a point a'B 'y to the
    line la + mß + ny = 0 :
    la ' + mß ' + ny
    4624
    vir + mº + na - 2mn cos A - 2nl cos B - 21m cos C }
    Proof. — By (4095 ) the perpendicular is eqnal to the form in (4606 ), with
    20', y ' in the place of x , y, divided by the square root of sum of squares of
    coefficients of x and y . The numerator = la ' + mß' + ny'. The denominator
    reduces by cos (B - ) = - cos A , & c.
    4625 Equation of the same perpendicular :
    a á l - m cos C - n cos B
    B B m - n cos A - l cos C = 0.
    ly ý n - l cos B - m cos A
    PROOF. – This is the eliminant of the three conditional equations
    La + MB + Ny = 0, La' + MB' + Ny' = 0 , and equation (4621).

    4626 Equation of a line drawn through a'B 'y parallel to the


    line (1, m , n ) :
    á ( m - bn
    B B ' an - rl = 0.
    Ty v bl - am
    Proof. – It is the eliminant of the three conditional equations
    la + mß + ny = 0 , la ' + mß' + ny' = 0 , and the equation at (4618 ).

    4627 The tangent of the angle between the lines (1, m , n ) ет

    and (l', m ', n') is


    (mn' - m 'n ) sin A + (nl - n 'l) sin B + (lm ' - I'm ) sin C
    Il' + mm ' + nn' - (mn' m 'n ) cos A - (nľ + n 'l) cos B - (lm ' + lm ) cosc
    PROOF. — By (4071) applied to the transformed equations of the lines,
    ( 4606 ) , observing (4007).
    622 TRILINE AR ANALYTICAL CONICS.

    EQUATIONS OF PARTICULAR LINES AND COORDINATE RATIOS


    OF PARTICULAR POINTS IN THE TRIGON.
    4628 Bisectors of the angles A , B , C :
    B - y = 0, 7 - a = 0, a - B = 0.
    4629 Centre of inscribed circle (or in -centre) * 1 : 1 : 1.
    The coordinates are obtained from their mutual ratios by the formula
    (4017 ).

    4630 Bisectors of the angles A , 1 - B , 7 - C :


    B - y = 0 , yta = 0, a + B = 0.
    Centre of the escribed circle which touches the side a (or
    a ex -circle) - 1 : 1 : 1.

    4631 Bisectors of sides drawn through opposite vertices :


    B sin B = y sin C, y sin C = a sin A, a sin A = B sin B.
    4632 Point of intersection (or mass-centre):
    cosec A : cosec B : cosec C .
    PROOF. - Assume mß - ny = 0, by (4604 ), as the form of the equation of
    a line through A , and determine the ratio m : n from the value of y : 3
    when a = 0 .
    The coordinates of the point of intersection may be found by (4610), or
    thus :
    a : ß = sin B : sin A = cosec A : cosec B ,
    Biy = sin C : sin B = cosec B : cosec C ,
    therefore a : ß : y = cosec A : cosec B : cosec C .

    4633 Perpendiculars to sides drawn through opposite


    vertices :
    B cos B = y cos C, y cos C = a cos A , a cos A = ß cos B .
    4634 Orthocentre : sec A : sec B : sec C .

    * This nomenclature is suggested by Professor Hudson , who proposes tbe following :


    “ In - circle , circum -circle , a ex- circle ... mid -circle for inscribed circle , circumscribed circle ,
    circle escribed to the side a , and nine-point circle ; also in - centre, circum -centre, a ed
    centre , ... mid -centre, for the centres of these circles ; and in -radius, circum -radius, a ez .
    radius, ... mid -radius, for their radii ; central line, for the line on which the circum -centre,
    mid -centre , ortho .centre , and mass-centre lie ; and central length for the distance between
    the circum .centre and the ortho .centre."
    THE RIGHT LINE. 623

    If the Cartesian coordinates of A , B , C be X ,Y1, X2Y2, X343,


    the coordinates of the centre of the inscribed circle are

    4635 X =- axi
    ? + bx , + cXg _ ayı + by , + cy :
    a + b + c u +b + c
    PROOF. - By (4032) . Find the coordinates of D where the bisector of
    the angle A cuts BC in the ratio b : 0 (VI. 3 ), and then the coordinates of E
    where the bisector of B cuts AD in the ratio b + c : a .

    4636 For the coordinates of the centre of the a ex-circle ,


    change the sign of a in the above values of x and y .
    463 The coordinates of the mass -centre are
    x = } (x; + xz+ xz), y = } (yi + ya + y ).
    4638 The coordinates of the orthocentre are obtained from
    the equations of the perpendiculars from xy2,2343, viz .,
    (x1 - x3) x + ( y - yz) y = x , (x1 x3) + yz (yı - ys),
    (x1 - x2) x + ( y1 - y2) y = x3 (x1 - x2) + y: ( yı - y ).
    Perpendicular bisector of the side AB :
    4639 a sin A - B sin B + y sin ( A - B ) = 0,
    4640 or a cos A - B cos B - , sin (A
    ra CO
    B ) = 0,
    4641 or
    a sin B sin C )I cos A
    sin C sin A
    ) cos B = 0 .
    2 sin A ) 2 sin B )
    4642 Centre of circumscribed circle (or circum -centre):
    cos A : cos B : cos C .
    PROOF. – A line through the intersection of y and a sin A - B sin B (4631)
    is of the form a sin A - ß sin B + ny = 0 , and , by (4622 ),
    n = - siu B cos A + sin A cos B = siu ( 4 - B ).
    Otherwise, by (4633) and (4619),
    a cos A - p cos Btk = 0
    is any line perpendicular to AB ; and the constant k is found by giving a : ß
    the value which it has at the centre of AB.
    624 TRILINEAR ANALYTICAL CONICS.
    - - - - --- - - -- - - -

    4643 Centre of the nine-point circle ( or mid -centre) :


    cos (B - C) : cos (C - A ) : cos ( A – B).
    Proof. - By (955 ) the coordinates are the arithmetic means of the corres
    ponding coordinates of the orthocentre and circum -centre. Therefore, by
    (4034,'42) and (4017), a =
    sec A cos A
    ( sin A sec4 + sin B secB + sinC seco ' sin A cosA + sin B cosB + sin C cosCl?
    which reduces to cos (B - C ) x constant.

    4644 Ex. 1 . - In any triangle ABC (Fig. of 955 ) , the mass-centre R , the
    orthocentre 0 , and the circum - centre Q lie on the same straight line ; * for the
    coordinates of these points given at (4632, '34, '42), substituted in (4615),
    give for the value of the determinant
    cosec A ( sec B cos C - cos B sec C) + & c.,
    which vanishes.
    Similarly , by the coordinates in (4643) , it may be shown that the mid .
    centre N lies on the same line.
    Equation of the central line :
    Ex. 2 . — To find the line drawn through the orthocentre and mass
    centre of ABC . The coordinates of these points are given at (4632, 34 ) .
    Substituting in the determinant (4616 ) and reducing , the equation becomes
    a sin 2A sin ( B - C ) + ß sin 2B sin ( C - A ) + y sin 2C sin ( A , B ) = 0.
    Ex. 3. - Similarly , from (4629, '42), the line drawn through the centres
    of the inscribed and circumscribed circles is
    a (cos B - cos C ) + (cos C - cos A ) + y (cos A - cos B ) = 0 .
    Ex. 4.- A parallel to AB drawn through C :
    a sin A + ß sin B = 0.
    For this is a line through aß , by (4604), and the equation differs only by a
    constant from y = 0, for it may be written
    (a sin A + B sin B + y sin C ) - y sin C = 0.
    Ex. 5. — A perpendicular to BC drawn through C is
    a cos C + B = 0.
    For a perpendicular is B cos B - y cos C = 0 (4633) .... . ... ... ( 1),
    and a line through C is of the form la + mß = 0 . Hence, by (4619), the
    constant k ( a sin A + B sin B + y sin C ) must be added to (1 ) so as to elimi
    nate y . Thus
    B sin C cos B + a sin A cos C + ß sin B cos 0 = 0,
    B sin ( B + (') + a sin A cos C = 0 or Bta cos C = 0.

    * The central line. See note to (4629).


    ANHARMONIC RATIO . 625

    ANHARMONIC RATIO .

    For the definition , see (1052).


    4648 The three ratios of that article are the values of the
    ratio k : Ké' in the three following pencils of four lines respec
    tively ,
    a = 0, a - kB = 0, B = 0, a + k'ß = 0.... (i.) (Fig. 34 ),
    a = 0, a - kß = 0, a - k'ß = 0, B = 0... (ii.) (Fig.35),
    a = 0, B = 0, atkß = 0 , a + k'ß = 0 ... (iii.) (Fig . 36 ).
    4649 The anharmonic ratio (i.) becomes harmonic when
    k = k '. Hence the lines a + kk , a - kß form a harmonic pencil
    with the lines a , b , the first dividing the external and the
    second the internal angle between a and. ß (Fig . 37) .
    4650 Similarly , the anharmonic ratio of four lines whose
    equations are
    a -miß = 0, a -maß = 0, a - mzB = 0, a -maß = 0,
    is the fraction (J1 — 42) (M3 — 4)
    (M , MA) (M2 - M3)
    PROOF. — Let OL be the line a = 0 , and
    OR, B = 0 .
    My - M , = difference of perpendiculars from 7 D C / B / dl
    A and B upon OL , divided by p .
    Similarly , Mg - Ma, & c. These differences
    are proportional to the segments AB, CD,
    AD , BC , and p is a common divisor.

    4651 Homographic pencils of lines are those which have the


    same anharmonic ratio . Thus the two pencils
    a -mß, a — Maß , a —ugß , a „miß ,
    and a'-mß', a' — uaß', a' - ugß', a - maß',
    are homographic pencils.
    a .

    4 L
    626 TRILINEAR ANALYTICAL CONICS.

    THE COMPLETE QUADRILATERAL.


    4652 DEF. — Any four right lines together with the three,
    called diagonals, which join the points of intersection , make a
    figure called a complete quadrilateral.
    4653 Let O be any point in the plane of the trigon ABC .
    Draw A0a , BOW, COC, and complete the figure. The equa
    tions of the different lines may be written as under, with the
    aid of proposition (4604), the ratios l : m : n being arbitrary
    and dependent upon the position of O .

    р —

    Aa, mß - ny = 0, AP, mB + ny = 0,
    Bb, ny - la = 0 , BQ, ny tla = 0 ,
    Cc, la - mß = 0 , CR, la + mß = 0 ;
    c, mß + ny - la = 0, OP, mB + ny – 2la = 0,
    ca , ny + la – mß = 0 , OQ , ny tla — 2mß = 0 ,
    ab, la +mß - ny = 0 , OR , la + mß — 2ny = 0,
    PQR , la +mb + ny = 0.
    PROOF. — Aa, Bb, Cc are concurrent by addition. bc is concurrent with
    Bb and B , and with Cc and y , by (4604 ). AP and OP are each concurrent
    with bc and a . PQR is concurrent with each pair of lines bc and a , ca and
    B , ab and y . Similarly for the rest .

    4654 Every pencil of four lines in the above figure (supply


    ing AP, BQ, CR) is a harmonic pencil.
    PROOF. — By the test in (4649), the alternate pairs of equations
    parabeing
    ll the
    sum and difference the oother
    nce ofof the et PQinRevery
    th Ltwo S be case
    BC .in fthe iguthe elogram
    Otherwise by projection. Let PQRS be the quadrilateral, with diagonals
    RP, QS meeting in Č . (Supply the lines AC, BC in the figure.) Taking the
    plane of projection parallel to OAB, the figure projects into the parallelogram
    pqrs ; the points A , B pass to infinity, and therefore the lines AO, BC become
    THE GENERAL EQUATION . 627

    lines harmonically divided by the sides of the parallelogram , the centre ,


    and the points at infinity.

    --
    -
    -
    -
    -
    - -
    -
    -
    -
    -

    4655 Theorem (974 ) inay be proved by taking a , B , y for the lines BC,
    CA, AB, and l'a + mß + ny, la + m ' ß + ny, la + mß + n ' y for bc, ca, ab, the last
    form being deduced from the preceding by the concurrence of Aa , Bb, and Cc.

    THE GENERAL EQUATION OF A CONIC .

    The general equation of the second degree is


    4656 aa’ + 682 + cry? + 2fBy + 2gya + 2haß = 0.
    This equation will be denoted by " (a , b , y) = 0 or u = 0 .
    Equation of the tangent or polar:
    4657 Uma + upB +uy = 0 orua' + upB' + u ,y = 0 ,
    the two forms being equivalent and the notation being that of
    (1405) . The first equation written in full is
    4659
    (aa' + hB + gy ) a + (ha' + b6' + fy ) B + (ga' + fB' + cy) y = 0.
    628 TRILINEAR ANALYTICAL CONIOS.

    Proof. — By the methods in (4120) . Otherwise by (4678 ) ; let aßy


    be on the curve ; then $ ( a , b , y ) = 0 . Next let the point where the line cuts
    the curve move up to aßy. Then the line becomes a tangent and the
    ratio n : n ' vanishes ; the condition for this gives equation ( 4658 ) .
    Cor . — The polars of the vertices of the triangle of refer
    ence are
    4660 aa + hB + gy = 1 , ha + bB + fy = 0, ga + fB + cy = 0 .
    4661 The condition that u may break up into two linear
    factors representing two right lines is, by (4469), A = 0 ,
    where
    4662 A = abc + 2fgh - af" — bgº — ch ”. (4454)
    4663 The general tangential equation a
    of the conic (4656 ) expresses the condi. h b ful
    tion that the line la tuß + vy may touch
    the curve and is the determinant equation
    annexed . The same written in full is
    4664 ( bc - f ) A + (ca - gº) tº + ( ab - hº) về
    + 2 (gh — af) uv + 2 (hf -bg) v1 + 2 (fg - ch ) = 0,
    4665 or Al? + Bu + Cv? + 2Fuv + 2GvX + 2H /p = 0 ;
    writing, as in (1642),
    A = bc - f?, B = ca - g?, C = ab —h”,
    F = gh -af, G = hf -bg, H = fg - ch .
    The tangential equation will be denoted by 0 (1, , v) = 0 or
    U = 0 , to correspond with (4656 ).
    Proof. — The determinant is the eliminant of the equation of the line
    du tuß + vy = 0 , and the three equations obtained by equating 1, M , v to the
    coefficients of a , b , y in (4659 ).
    Otherwise . — Assume la + uß + vy = 0 for the tangent. Substitute the
    value of the ratio ß : y obtained from it in the equation of the curve, and
    express the condition for equal roots (4119) .

    4666 Conversely, if the line la tuß


    + vy has the coefficients i , u , v con | H B F
    nected by the equation of the second G F C wil
    degree U = 0 (4664 ), then the enve
    lope of the line is the conic in the Ta B y
    THE GENERAL EQUATION. 629

    determinant form annexed corresponding to (4663), or in full


    4667 (BC - F2) aº + (CA - G ?) B + (AB - HⓇ)ya
    + 2 (GH - AF) By + 2 (HF - BG ) ya + 2 (FG - CH ) aß = 0 .
    4668 or A (aa’ + 6B + cy? + 2 By + 2gya + 2haß) = 0 .
    PROOF. — Eliminate v from U = 0 and the given line. The result is of
    the form L1? + 2 Ray + Mu? = 0 , and therefore the envelope is LM = R ', by
    (4792). This produces equation (4667). The coefficients are the first
    minors of the reciprocal determinant of Á ( 1643), and therefore, by (585) ,
    are equal to as, ba, & c.
    4669 The condition that U may consist of two linear factors
    is , as in (4661) , D = 0 , where
    4670 D = ABC + 2FGH - AF2 - BG – CH² = A ?. (1643)
    In this case U becomes the equation of two points , since the
    line datuß + vy must pass through one or other of two fixed
    points. See (4913) .
    4671 The coordinates of the pole of λα + μβ + νη are as
    Al + Hu + Gv : HX + Bu + Fv : G2 + Fu + Cv,
    4672 or U : Uu : U ,.
    PROOF. - By (4659 ) we have the equations in thee aa hu
    + hi + gy = kx
    + b6 + fy = kije
    margin , the solution of which gives the ratios of a : B : y .
    + fB + cy = kr.
    B - = _ T eki
    48 & AA + H2 + G , HA + Bu + F, GA + Fu + C, A
    Hence the tangential equation of the pole of l'a + m'ß + v' y,
    i.e ., the condition that datußtvy may pass through the
    pole ; or, in other words, that the two lines may be mutually

    4674 Uxtu Uu' + vUv = 0 or X’ U . + p 'Un tv' U , = 0 ,


    the two formsbeing equivalent, and each
    = AM + Buu ' + Cuv'
    + F (uv' tu'v) + G (vX + v1) + H (Nu' + ' ).

    The coordinates of the centre ao, Bu, yo are in the ratios


    4677 Aa + Hb + Ge : Ha + Bb + Fc : Ga + Fb + Cc,
    where a , b , c are the sides of the trigon .
    630 TRILINEAR ANALYTICAL CONICS.

    Proof. — By (4671), since the centre is the pole of the line at infinity
    aa + b3 + cy = '0 (4612).
    The quadratic for the ratio n : n ' of the segments into
    which the line joining two given points aßy, a' B ' y is divided
    by the conic is, with the notation of (4656 – 7 ),
    4678
    $ (a', B , y') n°+ 2 ( .a' + paß + ®7 ) nn' + $ (a , b , y ) n ” = 0.
    Proof. — By the method of (4131).

    The equation of the pair of tangents at the points where y


    meets the general conic u (4656), is
    4679 auß + 2huqug + bua = 0 .
    Proof.--- The point a 'b ', where y meets the curve, is found from
    aa " + 2ha 'B' + b3 " = 0 [ y = 0 in (4656 ) ]. The tangent at such a point is
    uga ' + ugl ' = 0 (4658) . Eliminate a ', B ”.
    The equation of a pair of tangents from a'B'y is
    4680 $ (a'B'y') $ (aby) = (daa'+ $&B'+ 0,7 )”.
    Proof. — By the condition for equal roots of (4678).
    By actual expansion the equation becomes
    (By? + 06 — 2F3y) a'? + (Ca’ + Ay? - 2Gya ) B "? + (AB* + Ba – 2Hap ) y "
    + 2 ( - Apy + Hya + Guß - Fa®) B ' y
    4681 + 2 (HBy – Bya + Faß - GPP) y 'a'
    + 2 (GBy + F'ya - Caß - Hyº) a'B ' = 0.
    In which either a', B ', y or a , b , y may be the variables, for the forms are
    convertible .

    Otherwise the equation of the two tangents is


    4682 • (By - B 'y , ya' - y'a , aß' - a'ß ) = 0 . (4665)
    Proof. — By substituting By - B 'y , & c. for 1 , M , v in (4664 ), the condition
    that the line joining a 'B 'y ' to any point apy on either tangent (see 4616 )
    should touch the conic is fulfilled . The expansion produces the preceding
    equation (4681).

    The equation of the asymptotes is


    4683 (a, b , y) = $ (Qo, Bor yo) = k£ ... ........ (1),
    where ao, Bo, Yo are the coordinates of the centre .
    THE GENERAL EQUATION . 631

    Otherwise the equation , in a form homogeneous in


    a , b , y , is
    4684 (aa, + bB. + ( yo) $ (a, b ,y) = k (da + bB + fy)?..... (2),
    where a , b , c are the sides of the trigon .
    And , finally, if the tangential equation (4664 ) be denoted
    by ♡ (1 , M , v ) = 0 , the equation of the asymptotes may be
    presented in the form
    4685 ♡ (a ,b ,c) $ (a, b, y) = (da + bB + fy)? .........(3).
    PROOF.- (i.) The asymptotes are identical with a pair of tangents from
    the centre ; therefore, put do Bos y, for a ', f ', v ' in (4680) ; thus
    (a , b , y ) • (ao Bos y ) = k* (aa + b3 + cy )? = k ???. .......... (4),
    since the polar becomes the line at infinity .
    Now ,multiplying the three equations in (4672) by a , b , y respectively ,
    and adding , we obtain q (a , b , y ) = k (1a + uß + vy ), and therefore
    (ao Bo, yo) = k (am + b3 + cy) = k ...... .............. (5),
    since the line at infinity (4612) is the pole of the centre.
    From (4 ) and (5 ), by eliminating k , equation ( 1) is produced ; and by
    dividing (4) by (5) , we get equation ( 2) .
    Again , taking the values of a, b , y from ( 4673), we have
    1a + » B + vy = 0 (A,M,»), and therefore aa,+ 4ß,+ c%. = º (a,b,c).
    By the last equation , (2) is converted into (3). See also (4966).
    Cor. — Since the centre (ao,ßo, yo) is on the asymptotes,
    we have
    4686 € (ao,Bo, yo) = £1 : 0 (a , b , c).
    4687 The semi-axes of the general conic (4656 ) are the
    values of r obtained from the quadratic
    as cos
    A h ,
    os cos B )
    lot moun )
    ( c + [s cos C ),) r = 0 ,,
    a
    ,

    72

    where a , b , c are the sides of the trigon, and


    s = abra - (abe).
    632 TRILINEAR ANALYTICAL CONICS.

    Proof. — The centre being a ,ß . yo, put a - a , = a , B - 6 , = y , y - y. = z.


    aby being a point on the conic, and 1 the radius to it from the centre, we

    ... gue = (r ’a cos A + y l cos B + z'c cos C) ............... (1).


    Also (4656), 9 ( a, b, y ) = q (a, + x, B. + y, yo + z) = 0 .
    Expand and write l, m , n for aa, + hße + gyo, ha, + b2 , + fyo, g ' o + fße + cyou
    The terms in x , y,z become
    lx + y + nz = 1 (a - a,) + & c. = I - S = 0 (4007)............ (2 ),
    and we obtain 0 (x, y,z ) = - º ( ao Bo, y ) = S'A = 0 (a, b, c) (4686) ......(3 ).
    The maximum and minimum values of me and therefore of
    aʻa cos A + y²b cos B + z'c cos 0 ...... .................. (4 )
    are required, subject to the equations (2) and (3 ). By the method of unde
    termined multipliers (1862), the quadratic above is found. — Ferrers's Tril.
    Coord ., Ch. 4 , Art. 18 .

    4688 . The area of the conic = - o & abra


    {0 (a,b,c)}}
    Proof. - If the roots of the quadratic (4687) are Eri!, Eri ”, the area
    will be air, The coefficient of 7 -* reduces by trigonometry to – 8%s", and
    the absolute term is - (a, b, c). Hence the product of the roots is found .

    4689 The conic will be an ellipse , hyperbola, or parabola ,


    according as ♡ (a, b , c ) (4664) is positive , negative, or zero.
    Proof.— The squares of the semi-axes have opposite signs in the hyper
    bola. Therefore the product of the roots of the quadratic (4687) must for
    an hyperbola be negative, and therefore º negative in (4688 ) .
    • ( a, b, c) = 0 makes the curve touch the line at infinity (4664 ), a pro
    perty which distinguishes the parabola .
    The condition that the general conic ( y be a
    rectangular hyperbola is
    4690 a + b + c = 2f cos A + 2g cos B + 2h cos C.
    Proof. — Let the asymptotes be
    la + mß + ny = 0, l'a + m 'ß + n 'y = 0 .
    Forming the product, equating coefficients with (4685), and denoting
    (a, b, c) by P , we get the proportions
    Il m m' nn' _ mn' + m 'n
    =aAỸbº - 6kAỸcó -CAỸ 2 (fé – 6c4)
    nil' + n 'l Im ' + I'm
    2 (99 - 01A) 2 (ho - ab A )
    THE GENERAL EQUATION. 633

    We may therefore substitute these denominators in (4620 ) for the condition


    of perpendicularity of the asymptotes. The result reduces to the equation
    above, by (837 ) .
    For another method , see (5002) .
    4691 The general conic (4656 ) will become a circle when
    the following relation exists between the coefficients :
    b sinºC + c sin ’ B - 2f sinB sinC
    = c sin 'A + a sin ’C — 2g sinC sin A
    = a sin’ B + b sin ’ A - 2h sin A sinB.
    Proof. — Equate coefficients of the equation of the conic (4656 ) with
    those of the circle in (4751).

    4692 The equation of the pair of lines drawn from a point


    a ' B ' y to the points of intersection of the conic and the line
    L = datuß + vy = 0 is, writing L ' for la' + uß' + vy', with the
    notation of (4656 – 7) ,
    L' $ (a, b , y) — 2LL'(daa' + baB + $ , 7 ) + L ’ (a', B , y ) = 0 .
    PROOF. — By the method of (4133).

    4693 The Director-Circle of the conic, that is , the locus


    of intersection of tangents at right angles, is, in Cartesians,
    C (22 + y ') - 2Gx — 2Fy + A + B = 0 .
    PROOF. - Let the equation of a tangent through xy be
    mě - n + (y - mx ) = 0.
    Therefore in the tangential equation (4665) put i = m , u = - 1, v = y - mx,
    and apply the condition , Product of roots of quadratic in m = - 1 (4078) .
    The equation of the same circle in trilinears is
    4694 ( B + C + 2F cos A ) a ' + (C + A + 2G cos B ) B2 + (A + B + 2H cos C)ya
    + 2 ( A cos A - H cos B - G cos O - F ) By
    + 2 ( - I cos A + B cos B - F cos ( – G ) ya
    + 2 ( - G cos A - F cos B + C cos ( - H ) aß = 0 ;
    or , in the form of (4751),
    4695
    (aa + 58 + y) ( B + 0 + 2F cos A (a, b, C) (aßy + bya + caß ).
    abc
    4 M
    634 TRILINEAR ANALYTICAL CONICS.

    Proof. — The equation of a pair of tangents (4681) through a point ay


    in trilinears, when the tangents are at right angles, represents the limiting
    case of a rectangular hyperbola . Therefore the equation referred to must
    have the coefficients of a " , ß ' , & c. connected by the relation in (4690) ,
    which thus becomes the equation of the locus of the point apy ; i.e., the
    director-circle.
    4696 When the general conic is a parabola , C = 0 in (4693)
    and 0 (a , b , c) = 0 in (4695), by (4430) and (4689), and these
    equations then represent the directrix .

    PARTICULAR CONICS .

    4697 A conic circumscribing the quadrilateral, the equa


    tions of whose sides are a = 0 , p = 0 , y = 0 , d = 0 , (Fig . 38)
    ay = kß8.
    Proof. - This is a curve of the second degree , and it passes through the
    points where a meets B and d , and also where y meets ß and d.
    4698 The circumscribing circle is ay = + 38 ; + or - , as
    the origin of coordinates lies without' or within the quadri
    lateral.
    Proof. — Transform (4697) into Cartesians (4009) ; equate coefficients of
    x and y and put the coefficients of xy equal to zero.
    4699 A conic having a and y for tangents and ß for the
    chord of contact : (Fig. 39)
    ay = kpa.
    PROOF. — Make è coincide with ß in (4698) .

    4700 A conic having two common chords a and ß with a


    given conic S : (Fig . 40)
    S = kaß.
    4701 A conic having a common chord of contact a with a
    given conic S : (Fig.41)
    S = ka'.
    4702 COR . If RPQ be drawn always parallel to a given
    line, PN2 oc RP . PQ, by (4317) .
    PARTICULAR CONICS. 635

    4703 A conic having a common tangent T at a point a 'y'


    and a common chord with the conic S :
    S = T (lx + my + nz ).
    4704 A conic osculating S at the point a'y' where T touches
    at one extremity of the common chord 1 (@ — ') + m ( y - ý') :
    (Fig. 43)
    S = T (lx +my- IX' –my).
    4705 A conic having common tangents T , T at common
    points with the conic S : (Fig . 44 )
    S = kTT'.
    4706 A conic having four coincident points with the conic
    S at the point where I touches : (Fig. 45)
    S = kT?

    4707 The conics S + L = 0 , S + M² = 0 , S + Nº = 0 ,


    (Fig 46) having respectively L , M , N for common chords of
    contact with the conic S , will have the six chords of inter
    section
    LEM = 0, MEN = 0, NEL = 0 ,
    passing three and three through the same points .
    Proof. – From ( S + M ) - ( S + N2) = ( M + N ) ( M - N ) , & c.
    By supposing one ormore of the conics to become right lines,various theorems
    may be obtained.

    4709 The diagonals of the inscribed and circumscribed


    quadrilaterals of a conic all pass through the same point and
    form a harmonic pencil.
    PROOF. — (Fig . 47.) By (4707) , or by taking LM = R² and L 'M ' = R '?
    for the equations of the conic by (4784).
    4710 If three conics have a chord common to all, the other
    three chords common to pairs pass through the same point.
    PROOF. — (Fig . 48.) Take S , S + LM , S + LN for the conics , L being the
    chord common to all , then M , N , M - N are the other common chords.
    636 TRILINEAR ANALYTICAL CONICS.

    4711 The hyperbola xy = (0x + 0y + p)


    is of the form (4699), and has for a chord of contact at infinity
    0x + Oy + p = 0 , 2 , y being the tangents from the centre.
    4712 The parabola y' = (0x + Oy + p) x
    has the tangent at infinity Ox + Oy + p = 0 .
    4713 So the general equation of a parabola may be put in
    the form of (4699). Thus
    (ax + By)? + (2gx + 2fy + c)(0x + 0y + 1) = 0.
    Here ax + By is the chord of contact, that is, a diameter ; 2gæ + 2fy + c is
    the finite tangent at its extremity, and 0x + Oy + 1 the tangent at the other
    extremity , supposed at infinity .

    4714 The general conic may be written


    (ax® + 2hxy + by?) + (2gx + 2fy + c) (0x + Oy + 1) = 0 .
    For this is of the form ay + kßè, ở being at infinity.

    4715 The conics S and S - k (0.x + 0y + 1)


    have double contact at infinity , and are similar.
    4716 The parabolas S and S - k
    have a contact of the third order at infinity.
    Proof. - For S and S - (0.0 + Oy + k) ' have the line at infinity for a chord
    of contact ; and, by (4712), this chord of contact is also a tangent to both
    curves .

    4717 All circles are said to pass through the same two
    imaginary points at infinity (see 4918) and through two real
    or imaginary finite points .
    Proof. The general equation of the circle (4144 )may be written
    (x + iy ) (x – iy) + (2gx + 2fy + c )(0x + Oy + 1) = 0 ;
    and this is of the form (4697). Here the lines x + iy intersect Ox + Oy + 1
    in two imaginary points which have been called the circular points at infinity ,
    and 2gx + 2fy + c in two finite points P , Q ; and these points are all situated
    on the locus a * + y + 2gx + 2fy + c = 0.

    4718 Concentric circles touch in four imaginary points at


    infinity.
    PARTICULAR CONIOS . 637

    PROOF. — The centre being the origin , equation (4136 ) may be written
    (a + iy ) ( x - iy ) = (0x + 0y + r ) , which, by (4699), shows that the lines x = iy
    have each double contact with the (supplementary) curve at infinity, and the
    variation of r does not affect this result. Compare (4711) .

    4719 The equation of any conic may be put in the form


    x + + y' = eʼye.
    Here æ = 0 , y = 0 are two sides of the trigon intersecting at
    right angles in the focus ; y = 0 , the third side, is the directrix ,
    and e is the eccentricity.
    The conic becomes a circle when e = 0 and y = 00 , so that
    ey = r , the radius, (4718 ) .
    wa
    4720 Two imaginary tangents drawn through the focus IS

    are, by (4699),
    (x + iy)(x — iy) = 0.
    These tangents are identical with the lines drawn through
    the two circular points at infinity (see 4717 ) . Hence, if two
    tangents be drawn to the conic from each of the circular
    points at infinity , they will intersect in two imaginary points ,
    and also in two real points which are the foci of the conic .
    All confocal conics, therefore, have four imaginary com
    mon tangents, and two opposite vertices of the quadrilateral
    formed by the tangents are the foci of the conics.
    4721 If the axes are oblique, this universal form of the
    equation of the conic becomes
    22 + 2xy cosw + yº = e’ya.
    The two imaginary tangents through the focus must now
    be written
    {x + y (coswti sin w )} {x + y (coswớisin w )} = 0.
    4722 Any two lines including an angle o form , with the
    lines drawn from the two circular points at infinity to their
    point of intersection , a pencil of which the anharmonic ratio
    is ei(1 – 20),
    PROOF. — Take the two lines for sides B , y of the trigon. The equation
    of the other pair of lines to the circular points will be obtained by elimin
    638 TRILINEAR ANALYTICAL CONICS.

    ating a between the equations of the line at infinity and the circum -circle ,
    viz ., da + 6B + cy = 0 and “ + + ? = 0. (4738 )
    α β γ
    The result is B? + 2bycos 0 + y = 0 ;
    or, in factors, (Btein) (B + e-iy) = 0.
    The anharmonic ratio of the pencil formed by the four lines B, B + fiy, .
    yB + e -ion is, by (4648, i.),
    ei : e- ie = - ei26 = ei( -20).
    4723 COR. - If 0 = 1 , the lines are at right angles, and the
    four lines form a harmonic pencil. [ Ferrers' Tril. Coords., Ch. VIII.

    THE CIRCUMSCRIBING CONIC OF THE TRIGON.


    4724 The equation of this conic (Fig. 49) is
    iBy + mya + naß = 0 or ? + * + = 0.
    PROOF. — The equation is of the second degree, and it is satisfied by
    a = 0, B = 0 simultaneously . It therefore passes through the point aß .
    Similarly through Bay and ya .
    The tangents at A , B , and C are
    cols

    m n
    4726 + = = 0 , - +
    +

    Proof. — By writing (4724 ) in the form


    mya + ß (ly + na ) = 0 ,
    lytna = 0 is seen, by (4697) , to be the tangent at ay ; for the intersections
    of a and y , with the curve, now coincide, and 8 (now ly + na) passes through
    the two coincident points.
    4729 The tangent, or polar , of the point a'B' y is ,by (4659),
    (my' +nB ) a + (na' + ly') B + (18 + ma') y = 0 .
    4730 The tangents at A , B , C (Fig. 49) meet the opposite
    sides respectively in P , Q , R on the right line
    a e

    + m8 +' n == 00. By ( 4604).


    v

    4731 The line 9 - passes through (D), the intersection


    of the tangents at A and B .
    PARTICULAR CONICS. 639

    4732 The diameter through the intersection of the tan


    gents at A and B is
    nda - nbß + (la — mb ) y = 0 .
    Proof. — The coordinates of the point of intersection are limi - n , by
    (4726 – 7 ), and the coordinates of the centre of AB are b : a : 0 . The
    diameter passes through these points, and its equation is given by (4616 ).
    4733 The coordinates of the centre of the conic are as
    le as

    1 ( - la + mb + nr) : m (la - mb + ne) : n (la +mb - nr).


    PROOF. — By (4610 ), the point being the intersection of two diameters
    like (4732). Otherwise , by (4677) .

    4734 The secant through (außıyı), (azß272), any two points


    on the conic, and the tangent at the first point are respectively,
    la , mß + ny – 0 and la , mß _ ny – 0 .
    2 ,02 * B .B . 717 . a Biri
    Proof. — The first is a right line, and it is satisfied by a = an, & c., and
    also by a = a ., & c., by (4725) . The second equation is what the first be
    comes when ag = an, & c. For the tangential equation , see (4893) .

    4735 The conic is a parabola when


    l'a’ + m %b + nºr — 2mnbr — 2nlca — 2lmab = 0 ,
    4736 or ✓ (la) + (mb ) + 7 (ne) = 0 .
    PROOF.- Substitute the coordinates of the centre (4733) in da + b3 + cy = 0,
    the equation of the line at infinity (4612) .
    Otherwise, the conic must touch the line at infinity ; therefore put a, l', c.
    for , M , v in (4893).
    4737 The conic is a rectangular hyperbola when
    Icos A + m cos Btn cos C = 0 ,
    and in this case it passes through the orthocentre of the
    triangle.
    Proof. — By (4690 ) , and the coordinates of the orthocentre (4634 ) .

    THE CIRCUMSCRIBING CIRCLE OF THE TRIGON.


    4738 By sin A + ya sin B + aß sin C = 0,
    sin A sinB , sinC _ 0 .
    or
    a + B + 7 .
    640 TRILINEAR ANALYTICAL CONICS.

    Proof. — The values of the ratios 1 : m : n , in (4724 ), may be found geo


    metrically from the equations of the tangents (4726 – 8 ).
    For the coordinates of the centre, see (4642).

    THE INSCRIBED CONIC OF THE TRIGON.


    4739 l'a + m *ß + nºye - 2mnßy - 2nlya - Almaß = 0 .
    4740 or ✓ (la ) + 7 (mB) + 7 (ny) = 0.
    Proof. — (Fig .50.) The first equation may be written
    ny (ny - 2la - 2mB) + (la – mB)* = 0 .
    By (4699 ) this represents a conic of which the lines y and ny - 2la - mß are
    the tangents at F and f, and la - mß the chord of contact. Similarly, it may
    be written so as to shew that a and B touch the conic .
    4741 The three pairs of tangents at F, f, & c., are
    2mB + 2ny - la ! 2ny + 2la -mß ?. 21a + 2mB - ny ?
    and a ľ and ß ľ andy
    and they have their three points of intersection P , Q , R on
    the right line la +mß + ny. By (4604 ).
    4742 The coordinates of the centre of the conic are as
    nb + mc : lc + nd :ma + lb .
    PROOF. - By putting a and B = zero alternately in (4739 ), we find, for
    the coordinates of the points of contact,
    + nmai.
    nb2A 2An
    at D, B = and at E . ,
    na + c
    therefore the equation of the diameter througb ( bisecting DE is, by (4603),
    a B
    nb +mclc + na
    Similarly the diameter bisecting DF is ma
    - + 16 nb + mi
    Therefore the point ofintersection , or centre, is defined by the ratios given
    above .
    Otherwise, by (4677), and the values in (4665), writing for a, b,c,f, g, h
    the coefficients in (4739) .

    4743 The secant through aßiya, a,2272 any two points on


    the curve.
    avl(v Biyz + V Boyd)+ B / m (V7,azt vyza.)
    +yn(Vaße + vaß.) = 0 .
    PARTICULAR CONICS. 641

    Proof. — Put a Bir , for aßy, and shew that the expression vanishes by
    (4740) .
    4744 The tangent at the point a Bıyı :
    av +Bvm +rVnt = 0.
    Proof. — Put a, = Qy, & c., in (4743), and divide by 2 / (a Bıyı).
    4745 The equation of the polar must be obtained from
    (4739) by means of (4659) .
    4746 The conic is a parabola when
    á + m + " = 0.
    Proof . — Similar to that of (4736 ).

    THE INSCRIBED CIRCLE OF THE TRIGON .

    4747 aºcos*4 + 8° cos*f + yºcosto


    - 28ycos cos g–24acosa cos –2a cos'Acos:
    4748 or cos q vateos. VB+ cos.@ Vy= 0.
    4749 The a -escribed circle : (4629)

    Proof. – At the point of contact where y = 0 , we have, in (4740), geo


    metrically , r being the radius of the circle,
    1:m = ß :a = r cot Asin A :rcot sin B = + cos”șA :cos'}B;
    + for the inscribed ; – for the escribed circle and - - B instead of B.
    4750 The tangent at a'B'y', by (4744), is
    cos + cos + cos a 1 = 0.
    The polar is obtained as in (4745 ).
    GENERAL EQUATION OF THE CIRCLE.
    4751 (la +mB + ny) (a sin A + B sin B + y sin C )
    tk (By sin A + ya sin B + aß sin C ) = 0 .
    4 N
    642 TRILINEAR ANALYTICAL CONICS.

    PROOF.— The second term is the circumscribing circle (4738 ), and the
    first is linear by (4609) ; therefore the whole represents a circle . By varying
    k , a system of circles is obtained whose radical axis (4161) is the line
    la +mb + ny, the circumscribing circle being one of the system .
    4752 If l'atm 'ß + n ' y be the radical axis of a second system
    of circles represented by a similar equation , the radical axis
    of any two circles of the two systems defined by k and k'
    will be
    k' (la +mB + ny) — k (l'a + m 'ß + n 'y ) = 0 .
    Proof. — By eliminating the term
    Bysin A + ya sin B + aß sin C .

    4753 To find the coefficient of x2 + y2 in the circle when


    only the trilinear equation is given .
    RULE . — Make a , ß , y the coordinates of a point from which
    the length of the tangent is known , and divide by the square of
    that length ; or, if the point be within the circle , substitute
    “ half the shortest chord Through the point” for “ the tangent.”
    PROOF. - If S = 0 be the equation of the circle , and m the required co
    efficient ; then, for a point not on the curve, Sum = square of tangent or
    semi-chord, by (4160 ) .

    THE NINE- POINT CIRCLE.

    4754 a’ sin 2 A + B sin 2B + ge sin 2C


    -- 2 (By sin A + ya sin B + aß sin C ) = 0.
    PARTICULAR CONICS . 643

    PROOF. — The equation represents a circle becanse it may be expressed in


    the form
    ( a cos A + B cos B + y cos C ) ( a sin A + ß sin B + y sin 0 )
    - 2 (By sin A + ya sin B + aß sin C) = 0.
    See Proof of (4751) . Now , when a = 0, the equation becomes
    ( B sin B - y sin C ) (ß cos B - y cos C ) = 0 ,
    which shews, by (4631, '3), that the circle bisects BC and passes through D ,
    the foot of the perpendicular from A .

    47540 The equation of the nine-point circle in Cartesian


    coordinates, with the side BC and perpendicular on it from A
    for x and y axes respectively , is
    2 + yº — R sin ( B - C) x — R cos ( B - C ) y = 0,
    where R is the radius of the circum -circle.

    THE TRIPLICATE -RATIO CIRCLE .


    47546 * Let the point S (Fig. 165) be chosen , so that its
    trilinear coordinates are proportional to the sides of the trigon .
    Draw lines through S parallel to the sides, then the circle in
    question passes through the six points of intersection, and the
    intercepted chords are in the triplicate -ratio of the sides.
    [ The following abbreviations are used , a , b, c, and not a, b, c, being in
    this article written for the sides of the trigon ABC . ]
    K = a + b + cº; 1 = ✓ (b cº + cʻa’ + a ?b ) ; A = ABC ;
    u smiw= 2 BFD = DE'F', & c.; 0 = DFD' = DE D', & c. .
    By hypothesis, a = = 1 = (4007) = .............(1),
    BF
    BD
    therefore BF . BF' = BD .BD', therefore F , F', D , D ' are concyclic.
    If AS, BS, CS produced meet the opposite sides in l, m , n ,
    Bn
    An
    – ab sin BCn – 73
    sin AC
    aa – 721
    a by ( 1 ) . .. . .. .. . .. . .. . . .. ( 2 ).

    * The theorems of ( 1 to 36 ) are for the most part due to Mr. R . Tucker, M . A .
    The original articles will be found'in The Quarterly Journalof Pure and Applied Mathematics,
    Vol. xix ., No , 76 , and Vol. xx ., Nos. 77 and 78.
    Other and similar investigations have been made by MM . Lemoine and Taylor and
    Prof. Neuberg , Mathesis, 1881, 1882, 1884.
    644 TRILINEAR ANALYTICAL CONICS.

    SF = BDD == sinY B =- 2ca


    Koin B (1) = 1. Similarly BF’ = 2*,&c....(3).
    DD’= DP sin d = content ... (4).
    BD' = BD + DD' = a (09/ 04), &c...............(5).
    FD = ✓ ( BD? + BF — 2BD. BF cos B ) = ( 2) and (5 ) ......(6 ) .
    Hence DEF and D ' E ' F ' are triangles similar to ABC, and they are equal
    to each other because ESF = E 'SF = E 'SF , & c. ( Euc. I. 37.)

    DF' = 7 (BD’ + BF"? – 2BD.BF'cos B) = abc............(7). . . .

    Hence DF ' = FE' = ED'.


    D'F = a BDP= *( K **)&o.............(8). . . . .

    BFP+ FD — BD _ a' + 62 + 62 K
    COSw =
    = "" 2BFTIDEP
    . FD = **** ***(5 & 6) = .. .....(9).
    21

    sinw = V (1- 2 ) = A (708)... ......(10).


    cos0 = cos(A –w), &c.= a*cos A + bc............. (11).
    AFE' + BDF ' + CED' = AF
    = . AE ' sin AA +, & c. = 4A = DEF , by (6 ) ... (12).

    Or, geometrically , by Euclid I. 37.

    Radius of T . R .circle, p = uR , by (6 ) (R = circum -radius) ...... (1:


    The trilinear equation of the T. R . circle is
    abc(a*+ 32+y?) = abe(aa + bB + cy)} +a+By+ bºya+ c*aß.....(14),
    or (bº +cº)a? +( + a”)f* +(a®+ bº)y = {(a?+ 6°)(a®+ c^) + B*c*} Bx
    + { (bº+ cº)(b +a?)+ c?a?} x + {(cº+ a )( + 6 )+ a*b*} af ...(15). ab

    Obtained by substituting the trilinear coordinates of D , E , F , through


    which points the circle passes, in (4751), to determine the ratios 1 : m : 1
    and k . The coordinates of D are
    o a ( a’ + 62) sinc ac sin B
    K TK :
    Similarly those of E and F .
    PARTICULAR CONICS. 645

    THE SEVEN-POINT CIRCLE.*


    4754c Let lines be drawn through A , B , C (Fig . 165 )
    parallel to the sides of the triangles DEF, D ' E ' F ', as in the
    figure, intersecting each other in P , P ', L , M , N . Let Q be
    the circum -centre ; then the seven points P , P ', L , M , N , Q , S
    all lie on the circumference of a circle concentric with the
    T . R . circle. ( 16 )
    The proof depends on Euclid III. 21, and the similar triangles DEF,
    D ' E ' F'.

    The radius p of the seven - point circle is


    2PP' sin 2w ............... ( 17),
    p = V (KP–31°) = 111 - 3 tan’w ( ( 18 ) ,

    obtained from pa = pP+8D° — 2pSD cos(B -TDD').


    Expand and substitute cos TDD' = D Er by (3)and (5),
    cos A - 6 ' + c - ?
    sin TDD' = cos 0 (11), cos B =
    < ca 2bc

    3p+ø = F,by(17)and(13); $ = Volt ,by(17)and(9)...{(29):


    COS W

    The trilinear equation of the seven -point circle is ..


    abc (a’ + B + y?) = aºßy + bºya + c°aß .................. (21),
    or a ßy+ bya +caß = (bca + caß +aby)(aa + 18 + cy).........(22).
    If the coordinates of P are av, B1, Yu, and those of Paí, Bí, ví; then
    ajaí = B , Bi = yır ..... .... ........... (23) .
    The equation of STQ is, by (4615 ),
    a sin (B - C) + B sin (C — A)+ ysin(A – B..............(24).
    And the equation of PP' is
    ö (a –b°c)+ (09– 1°C°)+ (60–a*b*) = 0... ....(25).
    The point S has been called the Symmedian point of the
    triangle. It has also this property. The line joining themid
    * This circle was discovered by M . H . Brocard , and has been called “ The Brocard Circle,"
    the points P , P" being called the Brocard points.
    646 TRILINEAR ANALYTICAL CONICS.

    point of any side to the mid -point of the perpendicular on that


    side passes through S .
    PROOF. - Let X , Y, Z (Fig. 166 ) be the feet of the perpendiculars ; x, y, z
    the mid - points of the same, and X ', Y ', Z ' the mid -points of the sides. Now
    the trilinear coordinates of X ', S , and æ in order are proportional to
    0 , 0, b l . This determinant vanishes ;
    a, b, c . therefore the three points are on
    1, cos C, cos B the same right line, by (4615).
    That the three lines X 'x , Y 'y, Z'z are concurrent appears at once by (970 ),
    since CX = 2Y'x , & c.
    The Symmedian point may also be defined as the intersec.
    tion of the three lines drawn from A , B , C to the corresponding
    vertices of the triangle formed by tangents to the circum
    circle at A , B , C .
    Let Ba, CB , Ay be taken = CX , AY, BZ respectively.
    Then Aa, BB , Cy meet in a point £ , by (976 ) , and this point
    by similarity of figure is the Symmedian point of the triangle
    formed by lines through A , B, C parallel to the sides BC,
    CA , AB .
    If the sides of X ' Y ' Z ' be bisected , similar reasoning shews
    that o , the Symmedian point of the triangle X ' Y ' Z ' , lies
    on SE
    It can also be shewn that, if A ' B ' C ' be any triangle having
    its sides parallel to those of ABC and its vertices on SA , SB ,
    SC, the sides of the two triangles intersect in six points on a
    circle whose centre lies midway between the circum -centres
    of the same triangles . When A ' B ' C ' shrinks to the point S ,
    the circle becomes the T . R . circle.

    A more general theorem respecting the triangle and circle


    is the following
    Take ABC any triangle, and let DD 'EE'FF' be the points in order, in
    which any circle cuts the sides.
    Let BD = pc, CE = qa, AF = rb ?
    CD ' = p'b, AE ' = 'c. BACA .................. (26 ) .
    From BD .BD' = BF. BF ', & nc.,os
    Euclid III. 35, we can write three equations
    which are satisfied by the values
    p = r' = tac, q = p' = tab, r = q' = tbc ............(27),
    and from these equations it appears that
    DF = oc; D' F ' = oa, & c., where o = v (t \ _ + K + 1) ...... (28 ),
    so that DEF and D 'E 'F ' are both similar to ABC .
    SELF-CONJUGATE TRIANGLE. 647

    Also DF = tabc, therefore DF = FE' = ED'. (29 ).


    From sin BED tac sin B we
    - can obtain

    cot BFD = coto = F ? -2


    ... (30)
    The radius of the circle = OR .. . ........ .......(31),
    and the coordinates of its centre are
    a = R {cos A++(Ka?-«*= *=c }. Similarly B and y......(32).
    The equation of the circle is
    aBy +bya + caß = t(aa + bB + cy) {abc (1 – ta’) + & c.} ...... (33),
    or tabc {a (1 - ta?) + B? (1 — tb?) + Y'(1 – te?) }
    = aßy { ( 1 – tb?)(1 — te?) + tb?c* } + & c.......(34).
    When t = 0, o = 1 and the circle is the circum -circle ............... (35) .
    When tK = 1, o = t1 = and the circle is the T. R . circle ...... (36).

    CONIC AND SELF -CONJUGATE TRIANGLE .


    When the sides of the trigon are the polars of the opposite
    vertices, the general equation of the conic takes the formm
    4755 Pa? + m B2 - nºy = 0.
    Proof. — (Fig. 51.) The equation may be written in any one of the
    three ways,
    Pa ’ = (ny + mB) (ny - mß), m ?32 = (ny tla )(ny - la ),
    n *y* = (la + imß) (la - imp).
    Hence, by (4699), a or BC is the chord of contact of the ttangents
    nyt mß (AQ, AS) he refore from
    drawn y or AAB, and ce a,f
    in biß Hisenthe ame of
    chord of co ntac ofof tthe
    contact he
    tangents ny + la (BR , BP) drawn from B . Hence a, 6 are the polarsof A , B
    respectively ; and therefore y or AB is the polar of C (4130 ). Also y may
    be considered to be the chord of contact of the imaginary tangents latimß
    drawn from C .
    4756 If the points of intersection of a and ß with the conic
    be joined , the equations of the sides of the quadrilateral so
    formed are
    QR , la + mb + ny = 0 , SP, la + mB - ny = 0 ,
    PQ, - la + mb + ny = 0 , RS, la - mb + ny = 0 .
    Hence QR , SP and PQ , RS intersect on the line y in A '
    and B '.
    648 TRILINEAR ANALYTICAL CONICS.

    4757 Each pencil of four lines in the diagram is a harmonic


    pencil, by the test in (4649) .
    4758 The triangle A' B'C is also self-conjugate with regard
    to the conic .
    Proof. — The equations of its sides CB', CA', A 'B 'are
    la - mß = 0, la + mp = 0, y = 0.
    Denote these by a', B ', y , and put a , ß in (4755) in terms of a ', ß . The
    equation referred to A 'B ' C thus becomes a' + ß " - 2n 'y' = 0 , which is of the
    same form as (4755) .
    4759 It is clear that the triangles AQS and BPR , formed
    by a pair of tangents and the chord of contact in each case,
    are also self -conjugate .

    4760 Taking A' B ' C for the trigon , and denoting the sides
    by a , ß , y , the equations of the sides RS, PQ , QR , SP of the
    quadrilateral become respectively
    nyŁla = 0, mB + ny = 0.
    Ex. - As an example of (4611 ), we may find the coordinates of P from
    the equations
    da + b3 + cy = 2 ) . ca = Elm : (amn + bnl + clm )
    0 + mß - ny = 0 from which 3 B = Emn (amn + bnl + clm )
    - la + 0 + ny = 0 ) ( y = Enl + (amn + bnl + clm ).
    To obtain the coordinates of Q , R , and S , change the signs of m , n , and I
    respectively .

    ON LINES PASSING THROUGH IMAGINARY POINTS.


    4761 LEMMA I. — The right line passing through two con
    jugate imaginary points is real, and is identical with the line
    passing through the points obtained by substituting unity for
    V - 1 in the given coordinates.
    Proof. — Let (a + ia', b + ib') be one of the imaginary points, and therefore
    ( a - ia ', b - ib ') the conjugate point. The equation of the line passing through
    them is, by (4083) and reducing , b' x - a' y ta'b - ab' = 0, which is real.
    But this is also the line obtained by taking for the coordinates of the
    points ( a + a', b + b ') and ( a - a', b - 6 ').

    LEMMA II. - If P , S and Q , R are two pairs of conjugate


    imaginary points, the lines PS and QR are real, as has just
    been shown , and, therefore, also their point of intersection is
    SELF -CONJUGATE TRIANGLE. 649

    real. The other pairs of lines PQ , RS and PR , QS are


    imaginary . But the points of intersection of each pair are
    real, and are identical with the points which are obtained by
    substituting unity for ✓ - 1 in the given coordinates, and
    drawing the six lines accordingly .
    Proof. — Let the coordinates of the four points be as under
    P ......atia', b + ib', Q ...... a tia ', B + iß ',
    S ....... a - ia ', 6 - ib', R ...... u - ia ', B - iß '.
    The equations of PR and QS, by (4083 ), are L + iM and L - iM , where
    L = (b - B ) x - ( a - a ) yta ß - ab ta'B' — a 'b ,
    M = (B' + B ') x - (a' - a ') y ta'ß – a'b - af3' + ab'.
    Now the lines L + iM = 0 intersect in the same real point as the lines
    LEM = 0, because the values L = 0 , M = 0 satisfy both equations
    simultaneously. Hence, to determine this point, we have only to take i as
    unity in the given coordinates.

    LEMMA III. — If P , S are real points, and Q , R a pair of


    conjugate imaginary points, the lines PS and QR are both
    real, by Lemma I., and consequently their point of intersec
    tion is real. The remaining pairs of lines PQ , RS and PR, QS
    and their points of intersection are all imaginary . But the
    line joining these two imaginary points of intersection is real,
    and is identical with the line obtained by substituting unity
    for ✓ - 1 in the given coordinates and drawing the six lines
    accordingly . . . .
    PROOF . --Let the coordinates of the four points be as under
    .. P ...... X,Y12 Q ...... a + ia', B + iß ',
    S ...... X ,Y 23 R ...... a - ia', B - iß '.
    Since the coordinates of R are obtained from those of Q by merely changing
    the sign of i, the equations of the four imaginary lines will take the forms
    .. . .. PQ ...... A - iB, SQ ...... C - 3D,
    PR ...... A + iB, SR ...... C + iD.
    Now let the coordinates of the point of intersection of PQ and SR be
    L + iM , L ' + iM ', then will L - iM , I ' - iM ' be the coordinates of the intersec
    tion of PR and SQ , for the equations of this pair of lines are got from those
    of PQ and SR by merely changing the sign of i. The points of intersection
    are therefore conjugate imaginary points, and the line joining them is real,
    by Lemma I. Also, since that line is obtained by writing 1 for i in the co
    ordinates of those points, it will also be obtained by writing 1 for i in the
    original coordinates of Q and R and constructing the figure as before.
    40
    650 TRILINEAR ANALYTICAL CONÍCS.

    4762 To find a common pole and polar of two given conics:


    (i.) If the conics intersect in four real points P , Q , R , S ,
    construct the complete quadrilateral (4652). Then A' B 'C
    ( Fig . 51) is a self -conjugate triangle for each conic , by
    (4758 ), and therefore each vertex and the opposite side form
    a common pole and polar to the conics.
    ( ii.) If the conics do not intersect at all in real points, the
    triangle A ' B' C is still real, by Lemma II. (4761) , and can be
    constructed in the manner shown .
    (iii.) If two of the points ( P , S ) are real, and two ( Q , R )
    imaginary, then , by Lemma III., the vertex A ' and the side
    B 'C are real, and may be constructed , and they form a common
    pole and polar of the given conics.

    Returning to the triangle of reference ABC,


    4763 Let la = ny coso, mß = ny sin q ; then the chord
    joining two points 01, 02 is
    la cos } (61+ $2) + mß sin } (61+ $2) = ny cosi(6 - 2),
    and therefore the tangent at the point o ' is
    4764 la cos $ + mß sin ( = ny.
    4765 Putting 12 = L, m = M , nº = - N, the conic (4755)
    becomes
    La’ + MB+ + Ny = 0 ... ........(1).
    4766 The tangent or polar of a'B'y is
    Laa' + MBB + Nyy = 0 ..................
    4767 Hence the pole of da +MB + vy = 0
    @ ^)..................(3).
    4768 The tangential equation is
    ** * * * + = 0...............(4),
    and this is the condition that the conic (1) may be touched
    by the four lines
    da Fuß + w = 0.
    SELF-CONJUGATE TRIANGLE . 651

    4769 In likemanner ,
    La” + MB2 + Ny? = 0.....................(5 )
    is the condition that (1 ) may pass through the four points
    (a', EB , Ey).
    4770 The locus of the pole of the line da + uB + ry with
    respect to such conics is

    Proof.— By(3), if (a,b,y) be the pole, a = 1,&c., : L = 1.,in (5),


    the equation of condition .
    4771 The locus of the pole of the line la + mb + ny, with
    respect to the conics which touch the four lines da Fußm

    Peoor.–By(3), if (a,b,y) be the pole, a = ? & c., :: L = , &c.,in


    (4 ), the equation of condition.
    4772 The locus of the centre of the conic is given in each
    case (4770, ' 1) by taking the line at infinity
    a sin A + B sin B + y sin C
    for the fixed line,since its pole is the centre .
    4773 Thus the locus of the centre of the conic passing
    through the four points ( a' EB' + y ) is
    a” sin ATB”B _ y2 sin
    sin BT ~
    C = 0.
    а

    4774 The coordinates of the centre of the conic (1) are


    La
    given by a = b = my
    PROOF. — Let the conic cut the side a in the points (OB.7ı), (OB,y ). The
    right line from A bisecting the chord will pass through the centre of the
    conic, and its equation will be B : y = B , + B : 7 + 72. Now B . + B , is the
    sum of the roots of the quadratic in ß obtained by eliminating y and a from
    the equations La® + MB + Ny = 0 , a = 0 , and aa + b3 + cy = . Similarly
    for y , ty, eliminate a and ß . The equation of the diameter through A being
    found, those through B and O are symmetrical with it.
    652 TRILINEAR ANALYTICAL CONICS.

    4775 The condition that the conic ( 1) may be a parabola is

    Proof . — This is, by (4 ), the condition of touching the lineat infinity


    da + b3 + cy = 0.
    4776 The condition that (1) may be a rectangular hyperbola
    is L + M + N = 0 , and in this case the curve passes through
    the centres of the inscribed and escribed circles of the trigon .
    Proof . — By (4690 ), (a, b, c are now L , M , N ). ( 1) is now satisfied by
    a = £ß = £y, the four centres in question .

    4777 Circle referred to a self-conjugate triangle :


    aʼsin 2A + B sin 2B + yº sin 2C = 0.
    Proof. — The line joining A to the centre is 4 ). Therefore
    M N
    b cos B
    o
    C cosa
    the condition of perpendicularity to a by (4622). Similarly
    N I
    c cos C cos Aī,' therefore ( 1) takes the form above.

    IMPORTANT THEOREMS.

    CARNOT'S THEOREM .
    4778 If A , B , C (Fig . 52) are the angles of a triangle, and
    if the opposite sides intersect a conic in the pairs of points
    a , a ' ; b , b' ; C, c' ; then
    Ac.Ac.Ba.Ba'.C6.Cb' = Ab.Ab.Bc.Be'.Ca.Ca'.
    PROOF. — Let a, ß , y be the semi-diameters parallel to BC, CA, AB ; then,
    by (4317 ), Ab . AB' : Ac : Ad' = 62 : 7 *. Compound this with two similar
    ratios.

    4779 Cor. — If the conic touches the sides in a, b, c,then


    Acé. Ba . C 2 = A6%. Bc . Ca'.
    THEOREMS. 653

    4780 The reciprocal of Carnot's theorem is : If A , B , C


    ( Fig . 52) are the sides of a triangle, and if pairs of tangents
    from the opposite angles are a , a' ; b , b'; C, c' ; then
    sin (Ac) sin (Ac') sin (Ba ) sin ( Ba') sin (Cb) sin (C6')
    = sin (Ab) sin (Ab') sin (Bc) sin (Bc') sin (Ca) sin (Ca'),
    where (Ac) signifies the angle between the lines A and c.
    Proof. — Reciprocating the former figure with respect to any origin 0 ,
    let A , B , C (i.e., RQ , QP, PR ) be the polars of the vertices A , B , C . Then,
    by (4130) , Q , R will be the poles of AB, AC ; and b , b', the polars of the
    points b , 6 ', will intersect in R and touch the reciprocal conic. Similarly , c, c'
    will intersect in Q . A , b' are perpendicular to 0A, Ob', and therefore
    LAB' = L AOC', and so of the rest.

    PASCAL ' S THEOREM .


    4781 The opposite sides of a hexagon inscribed to a conic
    meet in three points on the same right line.
    PROOF.— (Fig. 53.) Let a, ß, y, ', B', a' be the consecutive sides of the
    hexagon, and let u be the diagonal joining the points aa' and yy'. The
    equation of the conic is either ay - khu = 0 or a 'g ' - k 'ß 'u = 0 , and, since
    these expressions vanish for all points on the curve, we must have ay - kßu
    = a 'y ' - k 'ß 'u for any values of the coordinates. Therefore ay - a 'r '
    = u (kß - k 'B ') . Therefore the lines a , a ' and also y , y ' meet on the line
    kß - K'B'; and B, B' evidently meet on that line.
    Otherwise , by projecting a hexagon inscribed in a circle with its opposite
    sides parallel upon any plane not parallel to that of the circle . The line at
    infinity, in which the pairs of parallel sides meet, becomes a line in which
    the corresponding sides of a hexagon inscribed in a conic meet at a finite
    distance (1075 et seq.).
    4782 With the same vertices there are sixty different
    hexagons inscribable in any conic , and therefore sixty dif
    ferent Pascal lines corresponding to any six points on a conic.
    PROOF. — Half the number of ways of taking in order five vertices B , C ,
    D , E , F after A is the number of different hexagons that can be drawn, and
    the demonstration in (4781) applies equally to all.

    BRIANCHON 'S THEOREM .


    4783 The three diagonals of a hexagon circumscribed to a
    conic pass through the same point (Fig . 54).
    654 TRILINEAR ANALYTICA CONIOS .
    L

    PROOF. — Let the three conics S + L ', S + M *, S + N ?, in (4707), become


    three pairs of right lines, then the three lines L - M , M - N , N - L become
    the diagonals of a circumscribing hexagon .
    Pascal's and Brianchon's theorems may be obtained, the one from the
    other, by reciprocation (4840) .

    THE CONIC REFERRED TO TWO TANGENTS AND


    THE CHORD OF CONTACT.

    Let L = 0 , M = 0 , R = 0 (Fig . 55) be the sides of the


    trigon ; L , M being tangents and R the chord of contact.
    4784 The equation of the conic is LM = R ?. (4699)
    4785 The lines AP, BP, and CP are respectively
    uL = R , R = M , u'L = M . [By (4604).
    Since the point P on the curve is determined by the value
    of u , it is convenient to call it the point u .
    4788 The points u and - u (P and Q ) are both on the line
    u L = M drawn through C .

    4789 The secant through the points u , u' (P , P') is


    up 'L - (utu ) R + M = 0.
    Proof. — Write it u ( u ' L - R ) - ( u ' R - M ), and, by (4604 ), it passes
    through the point r '. Similarly through p . Otherwise, determine the co
    ordinates of the intersection of uL - R and uR - M , and of u ' L - R and
    M ' R - M by (4610 ), and the equation of the secant by (4016 ).
    4790 COR. — The tangents at the points u and - u ( P , Q )
    are therefore
    u ’LF2uR + M = 0.
    4791 These tangents intersect on R . [Proof by subtraction.
    4792 Theorem . - If the equation of a right line contains an
    indeterminate u in the second degree, it may be written as
    above, and the line must therefore touch the conic LM = R .

    4793 The polar of the point ( I', M ', R ') is


    LM ' - 2RR ' + L ' M = 0.
    TAE CONIC LM = R '. 655

    PROOF. - For utué' and une', in (4789), put the values of the sum and
    product of the roots of u ' L ' - 24 R ' + M ' = 0 (4790) .

    4794 Similarly the polar of the point of intersection of


    aL - R and bR - M is
    abL - 2aR + M = 0 .

    4795 The line CE joining the vertex C to the intersection


    of two tangents at u and j ', or at - u and - u ', is
    uu ' L - M = 0.
    Otherwise, if two tangents meet on any line aL - M , drawn
    through C , the product of their u ’s is equal to a .
    Proof. — -Eliminate R from the equations of the two tangents (4790 ).
    4796 The chords PQ ', P ' Q and the line CE all intersect in
    the same point on R .
    Proof. — The equations of PQ', P 'Q are, by (4789),
    up'L = ( — ') R - M = 0 ,
    and, by addition and subtraction , we obtain up'L - M = 0 (4795 ), or R = 0.
    4797 The lines qu'L + M (CD) and R intersect on the chord
    PP' which joins the points pl, u ' ; or — The extremities of any
    chord passing through the intersection of aL + M and R have
    the product of their u ’s equal to a .

    4798 The chord joining the points u tan , u cot touches a


    conic having the same tangents L , M and chord of contact R .
    Proof. — The equation of the chord is, by (4789),
    p ’L - uR (tan 0 + coto) + M = 0,
    and this touches the conic LM sin ?20 = R ’ at the point u, by (4792 ).
    4799 The tangents at the points u tan o , u coto intersect
    on the conic LM = R ' sin ? 20.
    PROOF. — Write the equations of the two tangents, by (4790 ),and then
    eliminate y .

    4800 Ex. 1. — To find the locus of the vertex of a triangle circumscribing


    a fixed conic and having its other vertices on two fixed right lines.
    Take LM = Rº for the conic (Fig . 56 ) , aL + M , 6L + M for the lines CD ,
    CE . Let one tangent, DE , touch at the point w ; then ,by (4795 ) ,the others,
    656 TRILINEAR ANALYTICAL CONICS.

    PD , PE , will touch at the points and therefore, by (4790) , their


    equations will be
    1 - 24 R +M , 1 - 2 + M.
    Eliminate y , and the locus of P is found to be ( a + b ) ’LM = 4abR .
    [ Salmon , Art. 272.

    4801 Ex. 2 . — To find the envelope of the base of a triangle inscribed in


    a conic, and whose sides pass through fixed points P , Q .
    (Fig . 57.) Take the line through P , Q for R ; LM - R ” for the conic ;aL - M ,
    bL - M for the lines joining Pand Q to the vertex C. Let the sides through P
    and Q meet in the point u on the conic ; then, by (4797), the other extremi
    ties will be at the points - 9 and - 9 , and therefore , by (4789), the
    u

    equation of the base will be abl + ( a + b ) R + u ? M = 0 . By (4792) , this


    line always touches the conic 4ab LM = (a + b )? R ?. [ Ibid .

    4802 Ex. 3 . — To inscribe a triangle in a conic so that its sides may pass
    through three fixed points. (See also 4823.)
    We have to make the base abL + (a + b ) uR + u ’M (4801) pass through
    a third fixed point. Let this point be given by cL = R , DR = M . Elimi
    nating L, M , R , we get ab + (a + b) uctuécd = 0, and since, at the point M ,
    u L = R , u ’ L = M , that point must be on the line abL + ( a + b ) cR + cd M .
    The intersections of this line with the conic give two solutions by two posi
    tions of the vertex. [Ibid.

    RELATED CONICS .

    4803 A conic having double contact with the conics S and


    S' (Fig. 58) is
    M ’E? — 2p ( S + S') + F2 = 0,
    where E , F are common chords of S and S', so that
    S - S' = EF.
    Proof. — The equation may be written in either of the ways
    (u E + F ) = 4uS or (u E - F ) = 4uS',
    showing that uEEF are the chords of contact AB, CD . There are three
    such systems, since there are three pairs of common chords.

    4804 Cor. 1. - A conic touching four given lines A , B , C , D ,


    the diagonals being E , F (Fig . 59) :
    u ’ E ? — 2m (AC + BD ) + F ? = 0.
    Here S = AC and S = BD , two pairs of right lines.
    ANHARMONIC PENCILS. 657

    Otherwise , if L , M , N be the diagonals and L + M + N the


    sides, the conic becomes
    4805 u ’ L — .(L² + Mº- N2) + M ” = 0 .
    For this always touches
    ( L ’ + M ’ – N3)2 --4L’ M or ( L + M + N ) ( M + N - L )( N + L - M ) ( L + M – N ).
    [ Salmon, Art. 287.]

    4806 Cor. 2 . - A conic having double contact with two


    circles C , C is
    M? — 2u (C + C")+ (C — C') = 0.
    4807 The chords of contact become
    u + C - C' = 0 and u - C + C ' = 0 .
    4808 The equation may also be written
    ✓CE / C = VM,
    which signifies that the sum or difference of the tangents
    drawn from any point on the conic to the circles is constant.

    ANHARMONIC PENCILS OF CONICS.

    4809 The anharmonic ratio of the pencil drawn from any


    point on a conic through four fixed points upon it is constant.
    PROOF. - Let the vertices of the quadrilateral in Fig . (38) be denoted by
    A , B , C , D , and let P be the fifth point. Multiplying the equation of the
    conic (4697) by the constants AB , CD, BC, DA, we have
    , AB .CD ABa . CDY PA . PB sin APB . PC. PD sin CPD
    BO . DA – BCB . DAS – PB. PC sin BPC. PD . PA sin DPA
    sin APB. sin CPD
    sin BPC. sin DPA
    Compare ( 1056 ).
    4810 If the fifth point be taken for origin in the system
    (4784, Fig . 55), and if the four lines through it be
    L - M R , L - M2R , L - MER , L - MAR ,
    4 P
    658 TRILINEAR ANALYTICAL CONICS.

    the anharmonic ratio of the pencil is, by (4650),


    mon

    _ (My – M2)(Mhz — Mux).


    (M , – . ) (pl.: — M3)

    4811 Cor. 1. — If four lines through any point, taken for


    the vertex LM , meet the conic in the points Mie Ma, M3, M4, the
    anharmonic ratio of these points , with any fifth point on the
    conic , is equal to that of the points - Mi, — Ma, Mg, Mn, in
    which the same lines again meet the conic .
    4812 Cor. 2 . - The reciprocal theorem is — If from four
    points upon any right line four tangents be drawn to a conic ,
    the anharmonic ratio of the points of section with any fifth
    tangent is equal to the corresponding ratio for the other four
    tangents from the same points.

    4813 The anharmonic ratio of the segments of any tangent


    to a conic made by four fixed tangents is constant.
    PROOF. — Let u , Mv, Ma, Mg, Mo (Fig . 60 ) be the points of contact. The
    anharmonic ratio of the segments is the same as that of the pencil of four
    lines from LM to the points of section ; that is , of uu, L - M , uu , L - M ,
    Mp3L - M , upe,L - M , a pencil homographic (4651) with that in (4810).

    4814 If P , P are the polars of a point with respect to the


    conics S , S', then P + kP will be the polar of the same point
    with respect to the conic S + kS'.
    4815 Hence the polar of a given point with regard to a
    conic passing through four given points (the intersections of
    S and S') always passes through a fixed point, by (4101).
    If Q , Q ' are the polars of another point with respect to
    the same conics, Q + Q is the polar with respect to S + kS'.
    4816 Hence the polars of two points with regard to a
    system of conics through four points form two homographic
    pencils (4651).

    4817 The locus of intersections of corresponding lines of


    two homographic pencils having fixed vertices ( Fig . 61) is a
    conic passing through the vertices ; and , conversely, if the
    conic be given, the pencils will be homographic.
    Proof. — For eliminating k from P + kP = 0, Q + kQk', we get PQ '= P 'Q .
    CONSTRUCTION OF CONICS. 659

    4818 Cor. — The locus of the pole of the line joining the two
    points in (4816 ) is a conic .
    Proof. — For the pole is the intersection of P + kP' and Q + kQ®.
    4819 The right lines joining corresponding points AA', & c .
    (Fig . 62) of two homographic systems of points lying on two
    right lines, envelope a conic .
    PROOF. — This is the reciprocal theorem to (4817) ; or it follows from
    (4813).

    4820 If two conics have double contact (Fig. 63), the an


    harmonic ratio of the points of contact À , B , C , D of any
    four tangents to the inner conic is the same as that of each
    set of four points (a , b , c, d ) or (a ', 6 ', c', d ') in which the
    tangents meet the other conic.
    Proof. - By (4798 ). The us for the points on the latter conic will be
    equal to the u 's of the points of contact multiplied by tan p for one set, and
    by coto for the other, and therefore the ratio (4810 ) will be unaltered.

    4821 Conversely, if three chords of a conic ad', bb', cc' be


    fixed, and a fourth dd moves so that {abcd } = { a'b'c'd '} , then
    dd' envelopes a conic having double contact with the given
    one.
    For theoremson a right line cut in involution by a conic, see (4824 - 8 ).

    CONSTRUCTION OF CONICS.

    THEOREMS AND PROBLEMS.


    4822 If a polygon inscribed to a conic ( Fig . 64) has all its
    sides but one passing through fixed points A , B , . .. Y , the
    remaining side az will envelope a conic having double contact
    with the given one.
    Proof. — Let a , b , ... Z be the vertices of the polygon, and a , a', a" , a '
    four successive positions of a . Then , by (4811),
    { a, a', a", a" } = {b, b', 6”, 6"" } = & c. = {2,x', z", 2"" } .
    Therefore,by (4821), the side az envelopes a conic, &c.
    660 TRILINEAR ANALYTICAL CONICS.

    4823 Poncelet's construction for inscribing in a conic a


    polygon having its n sides passing through n given points.
    Inscribe three polygons, each of n + 1 sides, so that n of
    each may pass through the fixed points, and let the remaining
    sides be aʼz', a' z" , a'"'z'" , denoted in figure (65 ) by AD , CF, EB.
    Let MLN , the line joining the intersections of opposite sides of
    the hexagon ABCDEF (4781), meet the conic in K ; then K
    will be a vertex of the required polygon .
    Proof. — { D .KACE } = { .KDFB }, each pencil passing through
    K , P, N , L ; therefore theanharmonic ratio { KACE } = {KDFB } for any
    vertex on the conic,by (4809) ; i.e., { Ka'a a'' } = { Kz'z" :"" } . But, if az
    be the remaining side of a fourth polygon inscribed like the others, we have
    by (4811),as in (4822), {aa'a" a"" } = {zz'z":"" } . Hence K is the point
    where a and 2 coincide.

    4824 Lemma. — A system of conics passing through four


    fixed points meets any transversal in a system of points in
    involution (1066).
    PROOF. — Let u , u be two conics passing through the four points ; then
    u + ku' will be any other . Take the transversal for x axis, and put y = 0) in
    each conic , and let their equations thus become ax® + 2gx + c = 0 and
    a 'æ * + 2gʻx + 6 = 0 . These determine the points where the transversal
    meets u and u '. It will then meet u + ku' in two points given by
    ax® + 2gx + c + k (a 'w ? + 2gʻx + C ) = 0 , and these points are in involution with
    the former, by (1065) .
    Geometrically (Fig. 66 ),
    {a .AdbA'} = {c. AdbA'} (4809),
    therefore { ACBA'} = {AB'C'A'} = { A’O'B'A }, therefore by (1069).
    4825 Cor. 1. — One of the conics of the system resolves
    itself into the two diagonals ac, bd . Hence the points B , B ',
    C , C are in involution with D , D ', where the transversal cuts
    the diagonals.
    4826 Cor. 2 . - A transversal meets a conic and two tan
    gents in four points in involution , so as to meet the chord of
    contact in one of the foci of the system .
    For, in (Fig . 66 ), if b coincides with c, and a with d, the
    transversalmeets the tangents in C, C', while B , B', D , D', all
    coincide in F (Fig . 67) , one of the foci on the chord of
    contact.
    CONSTRUCTION OF CONICS. 661

    4827 The reciprocal theorem to (4824) is — Pairs of tangents


    from any point to a system of conics touching four fixed lines,
    form a system in involution (4850).

    4828 The condition that + uy + vz may be cut in involu


    tion by three conics is the vanishing of the determinant
    ay by C 2fi 29ı 2h |
    A, H, B, az by Ca 252 292 2h,
    az bz cz 283 293 2hz
    A , H , B, = a 0 0 0 v
    Az H, B Ou 0 v 0 1
    0 0 v ļ 0 1
    where A1, H1, B , belong to the first conic and have the values
    in (4988).
    PROOF. — The quadratic A4ac2 + 2Hxy + By: = 0, obtained in (4987), deter
    mines the pair of points of intersection with the first conic . The similar
    equation for the third conic will have Az = A , + 14,, & c ., if the points are
    all in involution (1065 ). The third equation is therefore derived from the
    other two ; therefore the determinant vanishes, by (583 ).
    By expanding and dividing by » , the second determinant above of the
    sixth order is obtained.

    Newton's Method of Generating a Conic.


    4829 Two constant angles aPb, aQb (Fig .68) move about
    fixed vertices P , Q . If a moves on a fixed right line, b de
    scribes a conic which passes through P and Q .
    Proof. — Taking four positions of a, we have (see 1054),
    { P.66’b"b""} = {P .aa'a' a"" } = {Q .aaʼa"a"'} = {Q .66°6' 6"" }.
    Therefore, by (4817) , the locus of b is a conic.

    Maclaurin 's Method of Generating a Conic.


    4830 The vertex V of a triangle (Fig . 69) , whose sides pass
    through fixed points A , B , C , and whose base angles move on
    fixed lines Oa , Ob , describes a conic passing through A and B .
    Proof. — The pencils of lines through A and B in the figure are both
    homographic with the pencil through 0 , and are therefore homographic with
    each other. Therefore the locus of V is a conic, by (4817) .

    IN VITLY
    4
    662 TRILINEAR ANALYTICAL CONICS.

    Otherwise , let a , b , y be the sides of ABC'; la + mB + ny, l'a + m 'ß + n ' y


    the fixed lines Oa, Ob ; and a = up the moving base ab.
    Then the equations of the sides will be
    (lu + m ) B + ny = 0 , (l'u + m ) a + n'uy = 0 .
    Eliminate y ; then Im ’aß = (mp + ny) (l'a + n'y ), the conic in question, by
    (4697 ) .

    4831 Given five points , to find geometrically any number of


    points on the circumscribing conic , and to find the centre.
    Let A , B , C , D , E (Fig . 70) be the five points . Draw any
    line through A meeting CD in P . Draw PQ through the inter
    section of AB and DE meeting BC in Q ; then QË will meet
    PA in F , a sixth point on the curve, as is evident from Pascal's
    theorem (4781).
    To find the centre, choose AP in the above construction
    parallel to CD , and find two diameters, as in (1252).
    4832 To find the points of contact of a conic with five right
    lines.
    Let ABCDE (Fig . 71) be the pentagon . Join D to the
    intersection of AC and BE . This line will pass through the
    point of contact of AB , and so on . V .

    Proof. — By (4783 ), supposing two sides of the hexagon to becomeone


    straight line.

    4833 To describe a conic, given four points upon it and a


    tangent.
    Let a , a', b , b' (exterior letters in Fig . 52) be the four
    points . Then , if AB is a tangent, c , d coincide, and Carnot's
    theorem (4778 ) gives the ratio Ac? : Bc?. Then by (4831).
    Since there are two values of this ratio , + ( Ac : Bc), two
    conics may be drawn as required .
    4834 To describe a conic, given four tangents and a point.
    Let a, a', b , b' (interior letters in Fig. 52) be the four
    tangents . Then , if Q be the given point on the curve, the lines
    c , c' must coincide in direction, and (4780) gives the ratio
    - sinº (Ac) : sin ? (Be), by which the direction of a fifth tangent
    through Q is determined . Then by (4832). The two values
    + (sin Ac : sin Bc) furnish two solutions.
    Otherwise by (4804), determining u by the coordinates of
    the given point.
    CONSTRUCTION OF CONICS. 663

    4835 To describe a conic, given three points and two


    tangents .
    Let A , A ', A " be the points ( Fig . 67, supplying obvious
    letters). Let the two tangents meet AA' in the points C , C '.
    Find F , F ', the foci of the system AA', CC ' in involution (1066 )
    determining the centre by (985 ). Similarly , find G , Gʻ, the
    foci of a system on the line AA". Then , by (4826 ) , the chord
    of contact of the tangents may be any of the lines FG , FGʻ,
    F ’G , F' G '. There are accordingly four solutions, and the
    construction of (4831) determines the conic.
    4836 To describe a .conic, given two points and three
    tangents.
    Let AB, BC, CA (Fig. 167) be the tangents, and P, P ' the
    points. Draw a transversal through PP ' meeting the three
    tangents in Q , Q , Q ”. Find F , a focus of the system PP', QQ
    in involution (1066 , 985) ; G a focus for PP ', QQ" , and H for
    PP', QʻQ ” . Construct a triangle with its sides passing through
    F , G , H , and with its vertices L , M , N on BC, ČA , AB ,
    by the method of (4823) , which is equally applicable to a recti
    lineal figure as to a conic . L , M , N will be the points of
    contact. The reason for the construction is contained in
    (4826 ). There will, in general, be four solutions.

    If the conic be a parabola , the foregoing constructions


    can be adapted by considering one tangent at infinity always
    to be given .
    4837 To draw a parabola through four given points a , a ', b , b'.
    This is problem 4833 with the tangent at infinity .
    In figure (52), suppose cc' to coincide and AB to remove to infinity so as
    to become the tangent at c, the opposite vertex at infinity of a parabola , and
    therefore to be perpendicular to the axis. Cc then becomes a diameter of
    the parabola , and Carnot's theorem (4778 ) shows that
    Ca . Ca' _ Ac? Ba _ sin ’ ACc
    Cb . CO' A62 BCasinº BCc'
    since the points C , a , a ', b , b ' are all on the axis of the parabola relatively to
    the infinite distance of AcB. This result , however, is at once obtained from
    equation (4221) , Ca .Ca' : Cb . Cb ' being the ratio of the products of the roots
    of two similar quadratics. Thus a diameter of the parabola can be drawn
    through C by the known ratio of the sines of ACc and BCc.
    Next, describe a circle round three of the given points a , a ', b . By the
    property ( 1263) and the known direction of the axis, the other point in
    which the circle cuts the parabola can be found .
    Five points being known, we can, by Pascal's theorem , as in (4831),
    664 TRILINEAR ANALYTICAL CONICS.

    obtain two parallel chords, and then find P , the extremity of their diameter,
    by the proportion , square of ordinate oc abscissa (1239).
    Lastly , draw the diameter and tangent at P, and then, by equality of
    angles (1224 ), draw a line from P which passes through the focus. By
    obtaining in the same way another pair of parallel chords, a second line
    through the focus is found , thus determining its position .

    4838 To draw a parabola when four tangents are given .


    This is effected by the construction of (4832, Fig . 71) . Let AB, BC, AE ,
    ED be the four tangents, and CD the tangent at infinity . Then any line
    drawn to C will be parallel to BC, and any line to D will be parallel to ED.
    4839 To draw a parabola , given three points and one
    tangent.
    This is effected by the construction of (4835 , Fig. 67) . Let bC" be the
    tangent at 00 ; then the centre of involution 0 must be at C, so that
    CC . CO' = 0 . 00 = CA .CA' = CF , determining F . F ', another point on
    the chord of contact, being found by joining AA" or A ' A " , FF' will be the
    diameter through a , since the other point of contact 1 is at infinity .

    4840 To draw a parabola , given one point and three


    tangents.
    This is the case of (4834) , in which one of the given tangents b ' is at
    infinity. R must therefore be at infinity , and QR , PR and the tangent b ,
    since they all join R to finite points, must be parallel. The ratio found
    determines another tangent, and the case is reduced to that of (4838 ).
    4841 To draw a parabola , given two points and two
    tangents.
    This is problem (4836 ) . Suppose AC in that construction to be the
    tangent at infinity. F, G , H will be determined as in (4839) by mean
    proportionals. The chords LM , NM will become parallel, since M is at
    infinity ; and we have to draw LN and the parallel lines from L and N to
    pass through F, G , H in their new positions, so that the vertices L , N may
    lie on BC and AB.
    Otherwise by (4509), the intercepts s and i can readily be found from the
    two equations furnished by the given points.

    4842 To describe a conic touching three right lines and


    touching a given conic twice.
    Let AD , CF, EB ( Fig . 65 ) be the three lines as they cut
    the given conic . Join AB, AF, BC , BE, and determine K by
    the Pascal line MLN . K will be one point of contact of the
    two conics, by (4822) and the proof in (4823 ), since AD, CF,
    EB , and the tangent at K are four positions of the “ remaining
    side” in that proposition . The problem is thus reduced to
    RECIPROCAL POLARS. 665

    (4834), since four tangents and K the point of contact of one of


    them are now known .
    4843 To describe a conic touching each of two given conics
    twice, and passing through a given point or touching a given
    line.
    Proceed by (4803) , determining u by the last condition .
    To describe a conic touching the conics S + L ’, + M ”,
    S + N ? (4707) and touching S twice. [ Salmon, Art. 387.

    THE METHOD OF RECIPROCAL POLARS.

    DEF. — The polar reciprocal of a curve is the envelope of


    the polars of all the points on the curve, or it is the locus of
    the poles of all tangents to the curve, taken in each case
    with respect to an arbitrary fixed origin and circle of recipro
    cation .
    4844 Thus, in figure (72) , to the points P , Q , R on one
    curve correspond the tangents qr, rp, and chord of contact pq
    on the reciprocal curve ; and to the points p , q , r correspond
    the tangents QR, RP, and chord PQ .
    The angle between the tangents at P and Q is evidently
    equal to the angle poq, since Op, Oq, Or are respectively per
    pendicular to QR , RP, PQ .

    4845 THEOREM . — The distance of a point from a line is to its


    distance from the origin as the distance of the pole of the line
    from the polar of the point is to its distance from the origin .
    PROOF. — (Fig . 73.) Take O for origin and centre of auxiliary circle, PT
    the polar of c, pt the polar of C, CP perpendicular on polar of c,cp perpen
    dicular on polar of C . Then
    p = 0C .Ot = 0c. OT 2 Therefore, by subtraction, OC .mt = 0c.MT,
    and 00 .Om = Oc. OMS" . or 00.cp = 0c. CP
    that is, CP : 00 :: cp : c0 . Q. E. D.
    CoR . — By making CP constant, we see that the reciprocal
    of a circle is a conic having its focus at the origin and its
    directrix the polar of the circle 's centre.
    4 Q
    666 TRILINEAR ANALYTICAL CONICS.

    GENERAL RULES FOR RECIPROCATING .


    4846 A point becomes the polar of the point, and a right line
    becomes the pole of the line.*
    4847 A line through a fixed point becomes a point on a fixed
    line .
    4848 The intersection of two lines becomes the line which
    joins their poles.
    4849 Lines passing through a fixed point become the same
    number of points on a fixed line, the polar of the point.
    4850 A right line intersecting a curve in n points becomes n
    tangents to the reciprocal curve passing through a fixed point.
    4851 Two lines intersecting on a curve become two points
    whose joining line touches the reciprocal curve.
    4852 Two tangents and the chord of contact become two
    points on the reciprocal curve and the intersection of the tan
    gents at those points.
    4853 A pole and polar of any curve become respectively a
    polar and pole of the reciprocal curve ; and a point of contact and
    tangent become respectively a tangent and its point of contact.
    4854 The locus of a point becomes the envelope of a line.
    4855 An inscribed figure becomes a circumscribed figure.
    4856 Four points connected by six lines or a quadrangle
    become four lines intersecting in six points or a quadrilateral.
    4857 The angle between two lines is equal to the angle sub
    tended at the origin by the corresponding points. (4844 )
    4858 The origin becomes a line at infinity, the polar of the
    origin .
    4859 Two lines through the origin become two points at
    infinity on the polar of the origin .
    4860 Two tangents through the origin to a curve become two
    points at infinity on the reciprocal curve .
    4861 The points of contact of such tangents becomeasymptotes
    of the reciprocal curve.
    4862 The angle between the same tangents is equal to the
    angle between the asymptotes. (4857)
    * That is, with respect to the circle of reciprocation , and so throughout with the excep
    tion of (4853).
    RECIPROCAL POLARS. 667

    4863 According as the tangents from the origin to a conic are


    real or imaginary, the reciprocal curve is an hyperbola or
    ellipse .
    4864 If the origin be taken on the conic, the reciprocal curve
    is a parabola.
    For, by (4860, '1 ), the asymptotes are parallel and at infinity .
    4865 A trilinear equation is converted by reciprocation into
    a tangential equation .
    Thas ay = kße is a conic passing through four of the intersections of
    the lines a , ß , 7 , 8. Reciprocating , we get a tangential equation of the same
    form AC = kBD, and this is a conic touching four of the lines which join
    the points whose tangential equations are A = 0 , B = 0 , C = 0 , D = 0 .
    See (4907).
    4866 The equation of the reciprocal of the conic aạy2 + 2422
    = a with the same origin and axes is
    a®w ? + bʼyº = k ”,
    where k is the radius of the auxiliary circle whose centre is
    the centre of the conic .
    PROOF. — Let p be the perpendicular on the tangent, o its inclination ;
    then kør-? = p² = a * cos 0 + b2 sin 0 (4732) .
    4867 The same when the origin of reciprocation is the
    point ae'y',
    (xx' + yy' + k+)2 = a*x* + b y .
    Proof : k*r -1 = p = Va cos? 6 + b² sin’ 6 — (a 'cos 6 + y' sin 0).
    4868 The reciprocal curve of the general conic (4656 ), the
    auxiliary circle being x4 + yº = k or 2 + + y + z = 0 in tri
    linears, will be symmetrically
    AG + Bn + 06 + 2F95 + 2GC8 + 2H &n = 0 ,
    replacing & by —kº.
    PROOF. - Let En be a point on the reciprocal curve, then the polar of En,
    namely , * & + yn - k * = 0 , must touch the conic, by (4853) . Therefore, by
    (4665) , we must substitute g, n , - for i , u , v in the tangential equation
    All? + & c . = 0 .
    4869 From the reciprocal of a curve with respect to the
    origin of coordinates, to deduce the reciprocal with respect to
    an origin a'y', substitute in the given reciprocal equation
    k*x for x and rx' tyy' + k *
    668 ANALYTICAL CONIOS.

    PROOF. — Let P be the perpendicular from the origin on the tangent and
    PR = k'. The perpendicular from a 'y' is P - a ' cos 0 — y' sin 0,
    ē = - x' cos 6 — ý' sin 0, i ha = *x' + yy' + k .
    " R

    :: R cos 6 = kép cos 0


    xx ' + yy' +

    TANGENTIAL COORDINATES.
    4870 By employing these coordinates, theoremswhich are
    merely the reciprocals of those already deduced in trilinears
    may be proved independently . See (4019) for a description
    of this system .
    The following proposition serves to transform by recipro
    cation the whole system of trilinear coordinates of points and
    equations of right lines and curves, into tangential coordinates
    of right lines and equations of points and curves.

    THEOREM OF TRANSFORMATION .
    4871 Given the trilinear equation of a conic (4656 ), the
    tangential equation of the reciprocal conic in terms of i , u , v ,
    the perpendiculars from three fixed points A ', B ', C ' upon the
    tangent ( Fig . 74 ) will be as follows, O being the origin of
    reciprocation and OA', OB', OC' = p , q, r :
    an? bue? co? , 2fuv , 2gv) , 2hhj = 0 .
    487% ? tötpot är t rp- + T pq
    PROOF. — Let a = 0, B = 0, y = 0 be the sides of the original trigon ABC.
    The poles of these lines will be A ’, B ', C ', the vertices of the trigon for the
    reciprocal curve. Let RS be the polar of a point P on the given conic ;
    a, b , y the perpendiculars from P upon BC, CA, AB ; i.e., the trilinear co
    ordinates of P . Let i, u , v be the perpendiculars from A ', B ', C ' upon RS ;
    i.e., the tangential coordinates of the polar of P referred to A ', B ', C '. Then,
    by (4845), ở = a b õp = oor Substitute these values
    of a, b , y in (4656 ) and divide by OP?.
    4873 The angular relation between the trigons ABC and
    A ' B 'C ' is
    B'OC = 7 - A , C'OB' = 1 - B , A’OB' = 7 - C .
    TANGENTIAL COORDINATES. 669

    4874 If ABC be self-conjugate with regard to the circle of


    reciprocation , it will coincide with A ' B 'C'.
    4875 Now let O be the circum -centre (4629) of A ' B 'C '
    (Fig . 74), then it will be the in -centre of ABC, and ,by (4873),
    2A' = 7 - A , 2B’ = - B , 2C = 7 - C .
    Also p = q = rin (4872 ) , which becomes o (a , u , v ) = 0 , so
    that the conic and its reciprocal are represented by the same
    equation . Consequently any relation in trilinear coordinates
    has its interpretation in tangential coordinates. We have
    then the following rule :
    4876 RULE . — To convert any expression in trilinears into
    tangentials, consider the origin of the former as the in - centre of
    the trigon, change a, ß , y into i, u, v, and interpret the result
    by the rules for reciprocating (4846 –65 ) . If the angles of the
    original trigon are involved , change these by (4875 ) into the
    angles of the reciprocal trigon , of which the origin will now be
    the circum -centre.
    4877 Referring trilinears and tangentials to the same trigon
    ABC, the equation of a point, as shown in (4021), becomes
    B

    enn+ B ++ or = 0;
    4878 or, by multiplying by } },
    BOCA + CO Au + AOBv = 0 . (Fig. 3)
    The equation of a point can generally be obtained directly
    from the figure by means of this formula .

    EQUATIONS IN TANGENTIAL COORDINATES.


    For direct demonstrations of the following theorems, the reader may
    consult Ferrers' Trilinear Coordinates, Chap. VII.
    4879 The point dividing AB in the ratio a : b , that is, the
    intersection with the internal or external bisector of C , is
    al bu = 0 . Centre of AB itu = 0 .
    The point 0 in (4878) is now on the side AB.
    4881 Mass- centre , 1 +utu = 0 . (For BOO = COA = AOB.
    670 ANALYTICAL CONICS.

    4882 In - centre, al + bu tev = 0 . By (4878), for


    + BOC _ COA AOB
    4883 a ex-centre, - a1 + bu + fv = 0.Laur a
    4884 Circum -centre i sin 2 A tu sin 2B + v sin 2C = 0 .
    PROOF. — For BOC = iR ’ sin 2A , & c . in (4878). Otherwise. — By recipro .
    cation (4876 ), a sin A + B sin B + y sin C = 0 is the line at infinity referred
    to the trigon ABC ; therefore
    sin A + p sin B + v sin C = 0
    is the equation of the pole of that line referred to A 'B'C' ; that is,
    dsin 2 A ' + p sin 2B ' + v sin 2C', by (4875).

    4885 Foot of perpendicular from C upon AB,


    i tan A tu tan B = 0 .
    4886 Orthocentre i tan A tu tan B + v tan C = 0 .

    4887 Inscribed conic of ABC, [ Proof below .

    Luv + Mvd + Nage = 0.


    4888 Point of contact with AB,
    MA + Lj = 0 .
    4889 In -circle (4629),
    (s - a ) ur + ($ — b ) vl + ($ - c )dp = 0 .
    4890 Point of contact with AB, (s — b) 1 + (5 - a)n = 0.
    4891 a ex-circle, (s — b ) + ($ — c ) vd - Suv = 0 .
    Proof. — Since the coordinates of AB of the trigon are 0, 0 , v, the equa
    tion of the inscribed conic must be satisfied when any two of the coordinates
    1 , M , v vanish, therefore it must be of the form (4887) . Otherwise by
    reciprocating (4724 ) .
    If the circle touches AB in D ( Fig . 3 ), 1 : - u = AD : BD = 8 - a : 8 - 6
    (Fig . of 709), which proves (4890).
    ( 4888) is the equation of the point of contact, because the line (0 , 0 , v)
    passes through it and also touches the conic (4887) .
    (4889) is the in -circle by (4887) and (4890 ) and what precedes .

    4892 Circumscribed conic , [By (4876 ) applied to (4739, '40 ).


    L’ AP + M p ? + N° — 2M Nur - 2NLvi — 2L Mlu = 0, (4740)
    TANGENTIAL COORDINATES. 671

    4893 or ✓ (LX) + Mu + Nv = 0 .
    4894 Tangent at A , Mu = Nv.
    4895 Circum -circle
    a ‘l ? + b* ? + 4°v° — 2bºc?uv — 2cʻa ’vd – 2a²b \ p = 0 ;
    4896 or 71 +b Vu + i Vu= 0.
    Proor.---By (4876) applied to (4747,’8),and by cos ^ = sin A ’ (4875)
    OF .

    4897 Relation between the coordinates of any right line:


    a' (1 - u ) (^ — ) + b ( — v)(x - 1) + r? (v - 1 )(v - ) = ?
    4898 Coordinates of the line at infinity :
    l = u = v.
    PROOF.--- The trilinear coordinates of the origin and centre of the re
    ciprocal conic are a = ß = y , (4876 ). It is also self-evident,

    4899 The point 11 + mu + nv = 0 will be at infinity when


    1 + m + n = 0.
    PROOF. — By (4876 ), for the line la + mp3 + ny = 0 will pass through the
    origin a = p = y when l + m + n = 0 . .

    4900 A curve will be touched by the line at infinity when


    the sum of the coefficients vanishes .
    PROOF. - By (4876 ), for this is the condition that the origin in trilinears,
    a = ß = y shall be on the curve.

    4901 The equation of the centre of the conic ø (1, p1, v) is


    di + Du + , = 0 ,
    4902 or (a + h + g ) 1 + (h + b + f ) p + (g + f + c) v = 0 .
    PROOF. — The coordinates of the in -centre of ABC (4876 ) are a ' = ß ' = ý ,
    therefore the polar of this point with regard to the conico (a , b , y ) is
    9 + 98 + , = 0 (4658 ). This point and polar reciprocate into a polar and
    point, of which the former, being the reciprocal of the in - centre, or origin , is
    the line at infinity, and therefore the latter is the centre of o (a , M , v ), while
    its equation is as stated .
    672 ANALYTICAL CONICS .

    4903 The equation of the two points in which the line


    (1 ', ', v') cuts the conic is
    $ (^', u', v') $ (1, M , v) = (fix' + Pour' + $, r')?. (4680)
    4904 The coordinates of the asymptotes are found from the
    equations
    (^ , M , v ) = 0 and patou + = 0.
    Proof. These are the conditions that the line (^ , u , v ) should touch the
    curve and also pass through the centre (4901).
    4905 The equation of the two circular points at infinity is
    a' (^ — u )( v) + b ? (M — v) ( - 1) + rº (v _ ^)(v — ) = 0.
    Proof. — Put l' = u' = v' in (4903) to make the line at infinity, and for
    the conic take the in -circle (4889) .

    4906 The general equation of a circle is


    a ’ (1 - )( - v) + b? ( -— »)( - 1 ) + r? (v — _ )(v — )
    = (1 +mu + nv)?. ....(1),
    where 11 + mu + nv = 0 is the equation of the centre.
    Proof. — The general equation of a conic in trilinears may, by (4601), be
    put in the form
    a (B - B.)(y - y.) + b ( y - ys) (a - a .) + c (a - a.)(B B .) = (la + mß + ny) ,
    where la + mo + ny = 0 is the directrix , and a boy, the focus. Now let the
    focus be the in -centre of the trigon, and therefore a , = B . = y = 188 - (709) .
    By this relation and aa + bB + cy = 2 , the equation is expressed as
    a (8 — a ) (a - B ) (a - y ) + & c. = (l'a + m 'ß + n 'y ) ,
    (a - 6 ) (a - y ) cos' 1A + & c. = (l'a + m 'ß + n 'y ) .
    Reciprocating by (4876), this becomes
    (^ - ^ ) (1 - v ) sin ’ A ' + & c. = (?X + mu tnv)?,
    the constant factor introduced on the right being involved in l, m , n ; and
    sin A' = cos JA, by (4875 ). And we know that this is a circle by (4845
    Cor.) , and that the directrix of the conic reciprocates into the centre of the
    circle.
    Otherwise. The left side of ( 1 ) represents the two circular points at
    infinity (4905) , and, if for the right we take the equation of a point, the
    whole represents a conic , as in (4909) , of the form AC = B ”. In this case ,
    A , C , the points of contact of tangents from B , being the circular points, the
    conic must be a circle with B = 0 for its centre.
    TANGENTIAL COORDINATES. 673

    Abridged Notation .
    4907 Let A = 0, B = 0 , C = 0, D = 0 (Fig. 75 ) be the
    tangential equations of the four points of a quadrangle , where
    A said + biu + civ , B = a 1 + 624 + cov, and so on . Then
    the equation of the inscribed conic will be AC = kBD .
    PROOF. — The equation is of the second degree in 1 , u , v ; therefore the
    line (^ , u , v ) touches a conic . The coordinates of one line that touches this
    conic are determined by the equations A = 0, B = 0 . That is, the line
    joining the two points A , B touches the conic, and so of the rest.
    4908 If the points B , D coincide ( Fig . 76 ) , the equation
    becomes AC = kB ? ; and A = 0 , C = 0 are the points of
    contact of tangents from the point B = 0 .

    4909 Referring the conic to the trigon ABC (Fig: 78), and
    taking AC = kºBº for its equation , let a tangent ef be drawn,
    and let Ae : eB = k : m . The equations of the points e and
    f will be
    MA + kB = 0 , mkB + C = 0 .
    Proof. — The first equation corresponds to (4879 ). For the equation of f,
    eliminate A from mA + kB = 0 and AC = k * B *.
    4910 Let e, h (Fig . 77) be two points on AB whose equa
    tions are mA + kB = 0 , m ' A + kB = 0 . The equation of the
    point p , in which tangents from e and h intersect , is
    mm ' A + m + m ') kB + C = 0 .
    Proof. — The equation may be put in the form
    (mA + kB ) (m ' A + KB) = 0 ,
    because k ` B = AC if the line touches the conic. The equation being ofthe first
    degree in A , B , C ,must represent some point. That is , the relation between
    λ, μ , ν involved in it makes the straight line λα + μβ + νη pass through a
    certain point. But the equation is satisfied when mA + kB = 0 , a relation
    which makes the straight line pass through e. Hence a tangent through e
    passes through a certain fixed point. Similarly , by m ' A + kB = 0 , another
    tangent passes through h and the same fixed point.
    4911 COR . — Let m ' = m , then the equation of the point of
    contact of the tangent joining the points ma + kB and mkB + C
    (4909) (e and f, Fig. 78) will be
    m ’ A + 2mkB + C = 0 .

    4912 If in Fig . (78) the trilinear coordinates of the points


    4 R
    674 ANALYTICAL CONICS.

    A , B , C are 81, 91, 21, 22, Y2, 22, 23, 43, 23, the coordinates of the
    point of contact p of the tangent defined by m will be
    mʻx + 2mkx2 + xz, m ’yı+ 2mkya+ Ys, mʻzı + 2mk22 + 23,
    and the tangent at p divides the two fixed tangents in the
    ratios k : m and mk : 1, by (4909).
    4913 Note. — The equation U or 0 (^ , u , v ) = 0 (4665 ) expresses the
    condition that la + uß + vy shall touch a certain conic . When U is about
    to break up into two factors, the minor axis of the conic diminishes (Fig . 79).
    Every tangent that can now be drawn to the conic passes very nearly
    through one end or other of the major axis . Ultimately, when the minor
    axis vanishes, the condition of the line touching the conic becomes the con
    dition of its passing through one or other of two fixed points A , B . In this
    case, U consists of two factors, which , put equal to zero, are the equations
    of those points. The conic has become a straight line, and this line is
    touched at every point by a single tangent.
    4914 If U and U ' (Fig. 80) be two conics in tangential
    coordinates, kU + U ' is then a conic having for a tangent
    every tangent common to U and U ' ; and kU + AB is a conic
    having in common with U the two pairs of tangents drawn
    from the points A , B .
    The conic U ' in this case merges into the line AB, or,
    more strictly , the two points A, B, as explained in (4913).
    4915 If either kU + U ' or kU + AB breaks up into two
    factors, it represents two points which are the opposite ver
    tices of the quadrilateral formed by the four tangents.

    ON THE INTERSECTION OF TWO CONICS .

    INTRODUCTORY THEOREM .
    Geometrical meaning of ✓ ( - 1 ).*
    4916 In a system of rectangular or oblique plane coordinates, let the
    operator V - 1 prefixed to an ordinate y denote the turning of the ordinate
    about its foot as a centre through a right angle in a plane perpendicular to
    the plane of xy . The repetition of this operation will turn the ordinate

    * [ The fiction of imaginary lines and points is not ineradicable from Geometry . The
    theory of Quaternions removes all imaginariness from the symbol V - 1, and, as it appears
    that a partial application of that theory presents the subject of Projection in a much clearer
    light, I have bere introduced the notion of the multiplication of vectors at right angles to
    each other. ]
    ON THE INTERSECTION OF TWO CONICS. 675

    through another right angle in the same plane so as to bring it again into
    the plane of æy. The double operation has converted y into - y . Butthe two
    operations are indicated algebraically by V - 1 . 1 - 1 .y or (V - 1)’y = - y,
    which justifies the definition .
    It may be remarked , in passing, that any operation which, being per
    formed twice in succession upon a quantity, changes its sign, offers a con
    sistent interpretation of themultiplier V - 1.
    4917 With this additional operator, borrowed from the Theory of
    Quaternions, equations of plane curves may be made to represent more
    extended loci than formerly . Consider the equation x2 + y ’ = a . For values
    of x < a, we have y = + Vaa ?, and a circle is traced out. For values of
    a > a , we may write y = + i Vx? - a ', where i = V - 1. The ordinate
    V x2 - a ' is turned through a right angle by the vector i, and this part of the
    locus is consequently an equilateral hyperbola having a common axis with
    the circle and a common parameter , but having its plane at right angles to
    that of the circle . Since the foot of each ordinate remains unaltered in posi
    tion , wemay, for convenience, leave the operation indicated by i unperformed
    and draw the hyperbola in the original plane. In such a case, the circle may
    be called the principal, and the hyperbola the supplementary, curve, after
    Poncelet. When the coordinate axes are rectangular, the supplementary
    curve is not altered in any other respect than in that of position by the
    transformation of all its ordinates through a right angle ; but, if the coordi
    nate axes are oblique, there is likewise a change of figure precisely the same
    as that which would be produced by setting each ordinate at right angles to
    its abscissa in the xy plane.
    In the diagrams, the supplementary curve will be shown by a dotted line,
    and the unperformed operation indicated by i must always be borne in mind .
    For, on account of it, there can be no geometrical relations between the
    principal and supplementary curves excepting those which arise from the
    possession of one common axis of coordinates. This law is in agreement
    with the algebraic one which applies to the real and imaginary parts of the
    equation x - ( iy ) ' = a '. When y vanishes, x = a in both curves.
    If either the ellipse b*22 + a’y2 = aʻb? or the hyperbola bʻzi? — aʻy = a ?l ?
    be taken for the principal curve, the other will be the supplementary curve.
    It is evident that, by taking different conjugate diameters for coordinate
    axes, the same conic will have corresponding different supplementary curves.
    The phrase , “ supplementary conic on the diameter DD,” for example, will
    refer to that diameter which forms the common axis of the principal and
    supplementary conic in question.
    4918 Let us now take the circle x² + y2 = a ’ and the right line x = b .
    When bis > a, the line intersects the supplementary right hyperbola in two
    points whose ordinates are + i V6 - a . By increasing b without limit, we
    get a pair of, so - called , imaginary points at infinity . These lie on the
    asymptotes of the hyperbola , and the equation of the asymptotes is
    (a + iy ) (x - iy ) = 0 .
    Wecan now give a geometrical interpretation to the statements in (4720) .
    The two lines drawn from the focus of the conic b ?z? + aʻy = a b ? to the
    “ circular points at infinity ” make angles of 45° with the major axis , and
    they touch the conic in its supplementary hyperbola bạx? - a* (iy)² = aʼvº.
    An independent proof of this is as follows.
    676 ANALYTICAL CONICS.

    Draw a tangent from S (Fig. 81) to the supplementary hyperbola , and


    let x , y be the coordinates of the point of contact P . Then
    =
    vlas - 6 ) : an 1( - a ) =
    Va — b ?) '
    by the value of x. Also
    SN = « –CS = 762 - Va —Ö = 10
    Therefore y = SN , therefore SP makes an angle of 45° with CN .

    The following results are required in the theory of projec


    tion , and are illustrated in figures (82) to (86 ). Two ellipses
    are taken in each case for principal curves, and the supple
    mentary hyperbolas are shown by dotted lines . As the planes
    of the principal and supplementary curves are really at right
    angles, the intersections of the solid lines with the dotted are
    only apparent. The intersections of the solid lines are real
    points, while the intersections of the dotted lines represent
    the imaginary points.
    4919 Two conics may intersect
    (i.) in four real points (Fig .82) ;
    (ii.) in two real and two imaginary points (Fig . 83) ;
    (iii.) in four imaginary points (Fig. 84).
    [When,tuthelatetwo
    [When of theirtakining figures
    d, twohyperbolas the dot(83) cases](84) are similar and
    tedhe and
    similarly situated, two of their points of intersection recede to infinity (Figs.
    85 and 86 ). Hence, and by taking the dotted lines for principal, and the
    solid for supplementary, curves , we also have the cases ]
    (iv .) in two real finite points and two imaginary points at
    infinity';
    (v.) in two imaginary finite points and two imaginary
    points at infinity ;
    (vi.) in two imaginary finite points and two real points at
    infinity';
    (vii.) in two real finite points and two real points at
    infinitý.
    4920 Given two conics not intersecting, or intersecting in
    but two points, to draw the two supplementary curves which
    have a common chord of intersection conjugate to the
    THE METHOD OF PROJECTION . 677

    diameters upon which they are described, or in other words,


    to find the imaginary common chord of the conics.
    Poncelet has shewn by geometrical reasoning (Propriétés des Projectives,
    p . 31) that such a chord must exist. The following is a method of deter
    mining its position
    Let (abcfghVxyl)' = 0 and (a'b'c'f 'g'h 'Qxyl)² = 0 ............ (i.)
    be the equations of the conics C, C (Fig . 89), the coordinate axes being
    rectangular. Suppose PQ to be the common chord sought. Then the
    diameters AB, A ' B ' conjugate to PQ bisect it in D , and the supplementary
    curves on those diameters intersect in the points P , Q . Now , let the coor
    dipate axes be turned through an angle e , so that the y axis may become
    parallel to PQ , and therefore also to the tangents at A , B , A ', B '. This is
    accomplished by substituting for x and y, in equations (i.), the values
    x cos6 - y sin 0 and y cos 0 + æ sin 8.
    Let the transformed equations be denoted by (ABCFGHQxyl)? = 0 and
    ( A ' B 'C ' F 'G ' H 'Tryl) ' = 0 , in which the coefficients are all functions of ,
    excepting c, which is unaltered. Solving each of these equations as a quad
    ratic in y , the solutions take the forms
    y = ax + B + Vu (22 — 2px + q), y = a 'x + B ' + Vu' (x? — 2pʻx + 9') ... (ii.),
    with the values of a , b , M , P , 9 given in (4449 - 53) , if for small letters we
    substitute capitals. Thus, a , P , M , , 9 are obtained in terms of 0 and the
    original coefficients a, h , b , f, g , h .
    Now , the coordinates of D being & = ON , n = DN , we have n = a8 + ß
    and n = a' + B', therefore aš + B = a'5 + B ' ..... ............. ( iii.).
    The surd in equations ( ii.) represents the ordinate of the conic conjugate
    to the diameter AB or A ' B '. For values of a in the diagram > OM and
    < OR , the factor V - 1 appears in this surd , indicating an ordinate of the
    supplementary curve on AB or A' B '. Hence, equating the values of the
    common ordinate PD, we have
    (82 — 2p8 + q ) = " ' (E - 2p '& + 9') .....................(iv.).
    Eliminating & between equations (ii .) and ( iv .), we obtain an equation for
    determining 0 ; which angle being found, we can at once draw the diameters
    AB, A' B'.

    THE METHOD OF PROJECTION.


    4921 PROBLEM. — Given any conic and a right line in its
    plane and any plane of projection , to find a vertex of projec
    tion such that the line may pass to infinity while the conic is
    projected into a hyperbola or ellipse according as the right
    line does or does not intersect the given conic ; and at the
    same time to give any assigned proportion and direction to
    the axes of the projected conic .
    678 ANALYTICAL CONIOS .

    Analysis. Let HCKD be the given conic, and BB the right line, in
    Fig . (87) not intersecting, and in Fig . (88) intersecting the conic . Draw
    HK the diameter of the conic conjugate to BB . Suppose o to be the
    required vertex of projection . Draw any plane ECGD parallel to OBB,
    intersecting the given conic in CD and the line HK in F , and draw the
    plane OHK cutting the former plane in E , F , G and the line BB in A ; and
    let the curve ECGD be the conical projection of HCKD on the plane parallel
    to OBB.
    By similar triangles,
    EF _ 0A na FG _ OA EF . FG - 01
    HF = HA and FK = AK : HF.FK = HA.AK ... ....(1).
    Let a , 6 be the semi-diameters of the given conic parallel to HK and CD ;
    then CF2 62 CF26 ?. HA . AK
    HF.FK - a ' • EF.FG = a .OA ............... (2 ).
    Now , since parallel sections of the cone are similar, if the plane of HCKD
    moves parallel to itself, the ratio on the right remains constant; therefore, by
    (1193), the section ECGK is an ellipse in Fig . (87) and an hyperbola in
    Fig . (88) . Let a , b be the semi-diameters of this ellipse or hyperbola
    parallel to EG and CD, that is, to 0 A and BB ; then , by ( 2 ),
    62 _ B2 HA . AK ,
    = " 02 , :: OA’ = HA..AK ...............(3).
    But 6 HA.AK = AB , where AB in Fig. (88) is the ordinate at A of the
    given conic, but in Fig . (87) the ordinate of the conic supplementary to the
    given one on the diameter conjugate to BB. Therefore
    AOʻ = " AB
    AB? ................................. (4 ) .
    Hence A0, AB are parallel and proportional to a and b . And , since AB
    is given in magnitude and direction, we have two constants at our disposal,
    namely , the ratio of the semi- conjugate diameters a and b and the angle
    between them , or, which is the same thing, the eccentricity and the direction
    of the axes of the ellipse or hyperbola on the plane of projection.
    4922 The construction will be as follows :
    Determine the point A as the intersection of BB with the
    diameter HK conjugate to it. Choose any plane of projection ,
    and in a plane through BB, parallel to it, measure All of the
    length given by equation (3 ) or (4 ), making the angle BAO
    equal to the required angle between a and b . O will be the
    vertex of projection , and any plune LMN parallel to OBB will
    serve for the plane of projection.

    4923 Cor . I. - If AO = AB, the projected curve in Fig. (88


    will in every case be a right hyperbola .
    THE METHOD OF PROJECTION . 679

    4924 Cor . 2 . - If BAO is a right angle , the axes of the pro


    jected ellipse or hyperbola are parallel and proportional to
    AQ and AB. Hence, in this case , the eccentricity of the
    hyperbola will be e = OB : 0A.
    4925 Cor. 3. — If A0 = AB and BAO = a right angle ,
    the ellipse becomes a circle and the right hyperbola in Cor . 1
    has its axes parallel to AO and AB.

    4926 To project a conic so that a given point in its plane


    may become the centre of the projected curve.
    Take for the line BB the polar of the given point, and con
    struct as in (4922) . For , if P be the given point, and BB its
    polar (Fig. 87 or 88), p the projection of P will have its polar
    at infinity , and will therefore be the centre of the projected
    ellipse or hyperbola , according as P is within or without the
    original conic .

    4927 To project two intersecting conics into two similar


    and similarly situated hyperbolas of given eccentricity .
    Take the common chord of the conics for the line BB ( Fig.
    88) , and project each conic as in (4922) , employing the same
    vertex and plane of projection . Then , since the point A and
    the lines AB and ÁO are the same for each projection, corres
    ponding conjugate diameters of the hyperbolas are parallel and
    proportional to AO and AB ; therefore, & c.

    4928 To project two non - intersecting conics into similar


    and similarly situated ellipses of given eccentricity .
    Take the common chord of a certain two of the supple
    mentary curves of the conics (4920 ), in other words, the
    imaginary common chord of the conics, for the line BB, and
    proceed as in (4927).

    4929 To project two conics having a common chord of


    contact into two concentric, similar and similarly situated
    hyperbolas.
    Take the common chord for the line BB, and construct as in
    (4922 ). The common pole of the conics projects into a common
    centre and the common tangents into common asymptotes.
    680 TRILINEAR ANALYTICAL CONICS.

    4930 To project any two conics into concentric conics.


    Find the common pole and polar of the given conics by
    (4762), and take the common polar for the line BB in the
    construction of (4922) . The common pole projects into a
    common centre.
    4931 Ex. 1. — Given two conics having double contact with each other,
    any chord of one which touches the other is cut harmonically at the point of
    contact and where it meets the common chord of contact of the conics.
    [ Salmon 's Conic Sections, Art. 354.
    Let AB be the common chord of contact, PQ the other chord touching
    the inner conic at C and meeting AB produced in D . By (4929 ) , project
    AB, and therefore the point D , to infinity. The conics become similar and
    similarly situated hyperbolas, and C becomes the middle point of PQ (1189 ).
    The theorem is therefore true in this case . Hence, by a converse projection,
    the more general theorem is inferred.

    4932 Ex. 2 . - Given four points on a conic , the locus of the pole of any
    fixed line is a conic passing through the fourth harmonic to the point in
    which this line meets each side of the given quadrilateral. [Ibid., Art. 354.
    Let the fixed line meet a side AB of the quadrilateral in D , and let
    ACBD be in harmonic ratio. Project the fixed line, and therefore the point
    D , to infinity. Obecomes the middle point of AB (1055 ), and the pole of
    the fixed line becomes the centre of the projected conic . Now , it is known
    that the locus of the centre is a conic passing through the middle points of
    the sides of the quadrilateral. Hence , projecting back again , the more
    general theorem is inferred .

    4933 Ex. 3. — If a variable ellipse be described touching two given


    ellipses, while the supplementary hyperbolas of all three have a common
    chord AB conjugate to the diameters upon which they are described ; the
    locus of the pole of AB with respect to the variable ellipse is an hyperbola
    whose supplementary ellipse touches the four lines CA, CB, C ' A , C ' B , where
    C , C ' are the poles of AB with respect to the fixed ellipses.
    (Salmon , Art. 355.)
    PROOF. - Project AB to infinity and the three ellipses into circles. The
    poles P , C , C ' become the centres p, c, c' of the circles. The locus of p is a
    hyperbola whose foci are c, c'. But the lines Ac, Bc now touch the supple
    mentary ellipse of this hyperbola (4918 ). Therefore, projecting back again ,
    we get AC, BC touching the supplementary ellipse of the conic which is the
    locus of P. Similarly , AC', BC touch the sameellipse .

    4934 Any two lines at right angles project into lines which
    cut harmonically the line joining the two fixed points which
    are the projections of the circular points at infinity .
    Proof. This follows from (4723).
    INVARIANTS AND COVARIANTS. 681

    4935 The converse of the above proposition (4931), which is the theorem
    in Art. 356 of Salmon, is not universally true in any real sense. If the lines
    drawn through a given point to the two circular points at infinity form a
    harmonic pencil with two other lines through that point, the latter two are
    not necessarily at right angles, as the theorem assumes.
    The following example from the same article is an illustration of this
    Ex. - Any chord BB ( Fig . 88 ) of a conic HCKD is cut harmonically by
    · any line PKAH through P , the pole of the chord, and the tangent at K .
    The ellipse BKB here projects into a right hyperbola ; B , B project to
    infinity . The harmonic pencil formed by PK and the tangent at K , KB and
    KB projects into a harmonic pencil formed by pk and the tangentat k , kb
    and kb, where b, b are the circular points at infinity : but pk is not at right
    angles to the tangent at k of the right hyperbola . The harmonic ratio of the
    latter pencil can , however, be independently demonstrated, and that of the
    former can then be inferred . (Note that k is G in figure 88 . )
    If we may suppose the ellipse to project into an imaginary circle having
    points at infinity , the imaginary radius of that circle may be supposed to be
    at right angles to the imaginary tangent. The right hyperbola , however, is
    the real projection which takes the place of the circle in this and all similar
    instances ; and it is only in the case of principal axes that the radius is at
    right angles to the tangent.

    INVARIANTS AND COVARIANTS .

    4936 Let u = (abcfgh / xyz)?, u' = (a'b'c'f'gʻh'Xxyz)?


    be two conics as in (4401) with the notation of (1620).
    The three values of k , for which ku + u ' = ( represents two
    right lines, are the roots of the cubic equation
    4937 Ak + Ok? + O ' k + A = 0 ,
    4938 where A = abc + 2fgh - af2 — bgº — ch”,
    4939 © = Aa' + Bb' + Co + 2Ff' + 2Gg' + 2Hh',
    and A = bc - f?, F = gh - af, & c. (4665)
    For the values of A and O ' interchange a with a', b with b',
    & c.
    Proof.— The discriminant of ku + u', which | ka + a', kh + h', kg + g'
    must vanish (4661) , is evidently the determi- | kh + h ', kb + b ', kf + f '
    nant here written, and it is equivalent to the kato'. kf + f'. ke te'
    cubic in question .

    4940 A , 0 , 0 ', and Aare invariants of the conic ku + u '.


    4 s
    682 ANALYTICAL CONICS.

    That is , if the axes of coordinates be transformed in any


    manner, the ratios of the four coefficients in (4937) are
    unaltered .
    Proof. The transformation is effected by a linear substitution , as in
    (1794 ). Let u , u' thus become v, v'. Then kutu' becomes kv + v', and
    k is unaltered . If the equation ku + u' = 0 represents two right lines, it will
    continue to do so after transformation ; but the condition for this is the
    vanishing of the cubic in k ; and k being constant, the ratios of the coeffi .
    cients must be unalterable .

    4941 The equation of the six lines which join the four
    points of intersection of the conics u and u ' is
    Au' – Ou’ u + Oʻu'u? -- A'v ' = 0 .
    Proof. — Eliminate k from (4937) by ku + u ' = 0 .
    4942 The condition that the conics u and u'may touch is
    (OO' — 9A1') = 4 (OP – 340') (Oʻ 3A 'O ),
    4943 or 4A0 ' + 4A 'O ' + 274 ^ A ^2 — 18AA'OO' — OʻOʻ?.
    PROOF. — Two of the four points in (3941) must coincide. Hence two out
    of the three pairs of lines must coincide. The cubic (4937) must therefore
    have two equal roots . Let a , a , ß be the roots ; then the condition is the
    result of eliminating a and ß from the equations
    A (2a + B ) = - 0 , A (a’ + 2a3) = 0 ', Auß = - A ' (406).
    4944 The expression (4943 ) is the last term of the equation
    whose roots are the squares of the differences of the roots of
    the cubic in li, and when it is positive, the cubic in k has two
    imaginary roots ; when it is negative , three real roots ; and
    when it vanishes, two equal roots.
    PROOF. - By (543) or (579). The last term of f (x ) in (543) is now
    = 27 F (a ) F (B ) , a , ß being the roots of 3Ax? + 20x + 8 = 0. When this
    term is positive, f (x ) has a real negative root (409), and therefore F (x ) has
    then two imaginary roots ; for, if ( a − b ) ' = - c, a - b = ic, and a and b are
    both imaginary. When the last term of f (x ) is negative, all the roots of
    f (a ) are positive, and therefore the roots of F (a ) are all real.

    INVARIANTS OF PARTICULAR CONICS.


    4945 When u = ax? + by? + cm2 and u = x2 + y + z ,
    A = abc, © = bc + ca + ab, O ' = a + b + c, A ' = 1.
    INVARIANTS AND COVARIANTS. 683

    4946 When u = (abcfgh /xyz )? and u' = x2 + y2+ ,


    © = A + B + C , Ⓡ ' = a + b + c, A ' = 1.
    4947 When u = **+ y* —22 and u' = (x — a)* + (y – B)? — s ,
    U

    A = - p? A = - s?,
    @ = aº + B2— 2r?— ?, ' = a + B? — p?— 2s”.
    4948 The cubic for k reduces to
    (k + 1) {s’kº + (ro? + 52 - a? - B2) k + pe} = 0 .

    4949 When
    u = b*x*+a+yje –a?l? and u = (x —a)? + (y – B)2– 2°,
    A = - a*b*, © = a’62 {a’ + B2 — a — 62— } ,
    Oʻ = a *B? + 6°a? — aʼb? —22(a’ + 6²), A ' = — goº.
    4950 When u = yje — 4mx and u' = (x — a)? + (y – B)? — 7-2,
    A = - 4m ”, O = - 4m (a + m ), Oʻ = B? — 4ma — p ?, A' = - ;?.

    4951 When u = (abefghXx yz) and u' = xº + 2ạy cos w + y”,


    A , A ' = 0, @ = c (a + b) — fa - gº + 2 (fg - ch ) cosw ,
    O ' = c sinºw .
    Hence the following are invariants of the general conic,
    the inclination of the coordinate axes being w .
    abc + 2fgh - afa – bgº — ch _ A
    4952 csin ' w
    ...... (1),
    4953. c (a + b ) - 12 - gº + 2 ( fg - ch) cos w _
    c sinºw ........(2),
    ab - h2 a + b - 2h cos w
    4954 and ......(4 ).
    sin ' w sin 'w
    For these are what (1 ) and ( 2 ) become when the axes are
    transformed so as to remove f and g .
    684 ANALYTICAL CONICS.

    If the origin be unaltered , c is invariable , and transforma


    tion of the axes will then leave invariable
    4956 2fgh - af- — bg and f + gº — 2fg cos w
    sin ’ w sin ' w
    ea
    as appears by subtracting (3) from (1) and (2) from (4).
    4958 Ex. (i.) — To find the evolute of the conic 6-22 + a ’ya
    = a ^l ?. See also (4547 ) .
    PROOF. -- Denote the conic by u , and by u ' the hyperbola cʻry + b *y 'r - a * r'y
    (4335), which intersects u in the feet of the normals drawn from x 'y '. Two of
    these normals must always coincide if a 'y ' is to be on the evolute . u and u '
    must therefore touch. We have
    A = - a*b , O = 0, O ' = - a ’b (apx? + bʻy — c ), d' = - 2a²bc ry .
    Substitute in (4942), and the equation of the evolute is found to be
    (a’x? + bʼy : — (4) + 27a?l?c*x*y ’ = 0 .

    4959 Ex. (ii.) — Similarly the evolute of the parabola is


    obtained from
    u = y² – 4mx, u' = 2xy + 2 (2m - 2 ') y - 4my',
    A = - 4m², O = 0 , O ' = - 4 (2m — æ ), A ' = 4my,
    producing the equation 27my’ = 4 (x — 2m )". See also (4549) .

    4960 Ex. (iii.) - The locus of the centre of a circle of radius


    R , touching the conic l ?a ? + a y — a l?, is called a parallel to
    the conic. Its equation is
    | Rºck – ĐI”. {c (a + 2 ) + (a 20°) a + (2a - lº gº?
    * + 4a-b? + 64) — 2c* (as — a ’b? + 364) x² + 2c (3a* — a ?b? + 64) ya
    + (at — 6a b? + 664) * * + (6a' — 6a’b? + 64) y* + (6a* – 10a’b* + 664) z’y }
    + R 1 - 2a²b²c4 (a ’ + 62) + 2cº (3a* — a’b? + 64) x2 – 2¢? (a * — a ’b? + 364) y
    : - (6a — 10a²b² + 684) (0 c4 + a’yt) + (4a — 6a*82 – 6a²64 + 469) x?y2
    + 2 (a ’ — 202) bʻx® + 2 (1² — 2a”) a ’ye — 2 (a — a b? + 374) xty
    - 2 (3a ' — aʼb? + 64) a ?y*}
    +(bºx'+a?y?–a’ ":{(2 —c)*+y*} {(x +c)2+y } = 0,
    PROOF. - If the curves in (4949 ) be made to touch , aß will be a point on
    the curve parallel to u at a distance r. Therefore put the values of A , e , e ',
    and A ' in equation (4942) . [ Salmon , p. 325.
    INVARIANTS AND COVARIANTS. 685

    4961 When u of (4936 ) represents two right lines, A'


    vanishes , and
    4962 O' = 0 is the condition that the two lines should
    intersect on u ;
    4963 0 = 0 is the condition that the two lines should be
    conjugate with regard to u .
    Proof. - Transform u ' = 0) into 2xy = 0, so that the axes x , y are the
    right lines. This will not affect the invariants (4940). We now have, by
    (4937 ), A ' = 0 , 0 = 2 ( fg - ch ), O ' = - c.
    c = 0 makes u pass through the origin æy ; fg = ch makes x and y conjugate .
    For in (4671), if ix + py to becomes y = 0 , then = v = 0 , and the pole
    is given by H : B : F. But x = a = 0 at the pole, therefore =
    fg - ch = 0.

    4964 The condition that either of the lines in u should


    touch u is, by (4943),
    OP = 4AO' or AB = 0 ,
    with the above values of O and O'.

    4965 The equation ofthe two tangents to u , when læ tuy tv


    is the chord of contact, is, with the notation of (4665),
    u $ (^ , M , v ) = (lx + py + vz ) A .
    Proof. — The conic of double contact with u , ku + (1x + uy + v ) ? (4699) ,
    must now become two right lines . In (4937) A ' = 0 and oʻ = 0 , therefore
    kA + 0 = 0 . But 0 = 0 (^ , M , v ) . Hence eliminate k .
    4966 COR. — Taking the line at infinity ax + by + cz, we obtain
    the equation of the asymptotes (4685).

    The invariant 0 of the conic kutu vanishes


    4967 (i.) Whenever an inscribed triangle of u' is self-con
    jugate to u .
    4968 ( ii.) Whenever a circumscribed triangle of u is self
    conjugate to u .
    4969 e ' vanishes under similar conditions, transposing u
    and u' in (i.) and (ii.)
    PROOF. - (i.) u becomes ax' + by + cza (4765 ), and f = g = h = 0. There
    fore 0 in (4937) vanishes if a ' rb' = (' = 0 ; i.e., if u is of the form
    f'ya + gʻzx + h 'æy (4724 ).
    686 ANALYTICAL CONICS.

    (ii.) In this case, f'rg' = h ' = 0 and 0 vanishes if bc = f', & c., i.e., if
    the line x = 0 touches u , & c.

    4970 If u , u be two conics, and if 0² = 4A0', any triangle


    inscribed in u ' will circumscribe u , and conversely.
    in
    Proof. — Let u = x + y + zi — 2yz — 2zx — 2xy andd uni' = T2fyz
    hen + 2gzx
    zighi+ 2hxy ,
    both referred to the same triangle, (4739) and (4724 ). Then
    A = - 4, @ = 4 ( f + 9 + h ), o ' = - ( f + g + h)', A' = 2fgh ;
    therefore 0º = 410', a relation independent of the axes of reference (4940) .

    4971 Ex. ( i.) — The locus of the centre of a circle of radius r, circum
    scribing a triangle which circumscribes the conic b ?z ? + a’yº = a ’b?, is
    (zº + yº — a? — b? + r?)? + 4 { b?z? + a’y — a?b? — (a' + b?) } = 0,
    from O ' = 400' and the values in (4949) .
    4972 Ex. (ii.) — The distance between the centres of the inscribed and
    circumscribed circles of a triangle is thus found, by employing the values of
    0 , 0 ', and A in (4947), to be D = ✓ (7 ”? + 2rr') , as in (936 ).

    4973 The tangential equation of the four points of intersec


    tion of the two conics u = 0 , u = 0 is
    V ? = 4UU',
    with themeanings
    4974 U = (AB CFGHXdqv ) ; (4664)
    U ' = (A 'BẮCʻF ’G ’ H 'X1u ») .
    4976 V = ( A ” B ” C" F " G " H "Xuv)?
    4977 A = bc - fº, &c.; A' = b'c —f", & c.,
    as in (4665), and
    4978 A ” = bc' + b'c - 2ff", F" = gh’ + g'h – af” – a'f,
    4979 B ” = ca' + ca 28g', G " = hf' + h'f - bg' — b'g ,
    4980 C ” = ab' + a 'b - 2hh', H " = fg' + f' g - ch' - c'h .
    Proof. — The tangential equation is the condition that da + up + vy may
    pass through one of the four points of intersection of u and u '. The tan .
    gential equation of the conic u + ku ' is obtained by putting a + ka' for a , & c.
    in U ( 4665 ), and is U + kV + k * U ' = 0 . The tangential equation of the
    envelope of the system is V = 4UU ' (4911) . This is the condition that the
    line (a , M , v ) may pass through the consecutive intersections of the conics
    obtained by varying k . But these conics always intersect in the same four
    points. The above is therefore the tangential equation of the four points.
    IN VARIANTS AND COVARIANTS. 687

    4981 The equation of the four common tangents of two


    conics u , u ' is
    F = 4AA’uu',
    where F = (a "b"c"f" :" h" Xaßy)”,
    and a " = BC' + B 'C — 2FF', & c.,
    f" = GH ' + G ' H - AF' – A ' F , & c.,
    as in (4978 –81).
    PROOF. — This is the reciprocal of the last theorem . U + KU ' is a conic
    touching the four common tangents of the conics U and U '. The trilinear
    equation formed from this will, by (4667) , be ua + kF + k *u ' A ' = 0 . The
    envelope of this system of conics is the equation above, which must therefore
    represent the four common tangents .
    The curve F passes through the points of contact of u and u with the
    locus represented by (4981).
    4982 Hence the eight points of contact of the two conics
    with their common tangents lie on the curve F .
    4983 The reciprocal theorem from equation (4973) is —
    The eight tangents at the intersections of the conics envelope
    the conic V .

    4984 F = 0 is the locus of a point from which the tangents


    to the two given conics u , u form a harmonic pencil.
    PROOF. — Putting y = 0 in (4681) , we get a quadratic of the form
    aa’ + 2haß + bB2 = 0 , which determines the two points in which the line y is
    cut by tangents from a ', 6 ', y'. Let the similar quadratic for the second conic
    be a ' a ’ + 2h'aß + b 'ß = 0 . Then , by ( 1064) , ab' + a 'b = 2hh' is the condition
    that the four points may be in harmonic relation . This equation will be
    found to produce F = 0 .
    4985 The actual values of a , h , b , suppressing the accents
    on a', ß ', y , are
    CB + Byº — 2FBy, GBy + Fya - Caß - Hy',
    Ani + Ca' - 2Gya ;
    and similarly for a', h', b', with A written for A, & c.
    4986 If the anharmonic ratio of the pencil of four tangents
    be given, the locus of the vertex will be F = kuu'. If the
    given ratio be infinity or zero, the locus becomes the four
    common tangents in (1981).
    688 ANALYTICAL CONICS.

    4987 V = 0 is the envelope of a conic every tangent of


    which is cut harmonically by the two conics u , a ' ; i.e., the
    equation is the condition that datuß + vy should be cut har
    monically by the two conics .
    PROOF. - Eliminate y between the line (A , M , v ), and the conics u and u
    separately, and let Aa' + 2 Haß + BB* = 0 and A 'a ' + 2H 'ap + B ' ' = 0 stand
    for the resulting equations. Then , by (1064 ), AB ' + A ' B = 2HH ' produces
    the equation V = 0 , which, by (4666 ), is the envelope of a conic .
    4988 The actual values of A , H , B are respectively
    av? + cl — 2gvd , hvi - guv - fvd + chi, cp²tbv — 2 fuv;
    and similarly for A', H ', B', with a' for a, & c.

    4989 F ? = 4AA’uu' is a covariant (1629) of the conics U , u '.


    For the four common tangents are independent of the axes of reference.
    4990 U = 0 and V = 0 (4973) are both contravariants
    (1814 ) of u and u .
    PROOF. — For U = 0 is the condition that datuB + ry = 0 shall touch the
    conic u ; and V = 0 is the condition that the same line shall be cut bar
    monically by u and u '; and if all the equations be transformed by a recipro
    cal substitution (1813, ' 14 ), the right line and the conditions remain
    unaltered.
    4991 Any conic covariant with u and u ' can be expressed in
    terms of u , u ', and F ; and the tangential equation can be
    expressed in terms of U , U ' and V .

    4992 Ex. ( 1). — The polar reciprocal of u with respect to u


    is Ou = F .
    PROOF. — Referring u , u ' to their common self-conjugate triangle,
    u = ax + by* + cz?, u = x + y2+ z?,
    F = a (b + c) x2+ 6 (c + a ) y* + c (a + b) z*.
    The polar of %, n , Ś with respect to u ' is Ex + ny + 87, and the condition that
    this may touch u is bct? + can? + abs = ( (4664 ), or, which is the same
    thing, (bc + ca + ab) (ac2 + y² + z ) = F or Ou ' = F (4945).

    4993 Ex. (2 ).-— The enveloping conic V in (4987) may also


    be written
    Ou' + 'u = F .
    INVARIANTS AND COVARIANTS. 689

    PROOF. — With the sameassumptions as in Ex. (1), V in (4973) becomes


    (b + c ) 1 : + ( c + a ) u + ( a + b ) v2 = 0 . The trilinear equation is, therefore,
    by (4667),
    (c + a )(a + b )aº + (a + b) (b + c) y® + (6 + c) (c + a ) z = 0 ,
    or (bc + ca + ab)(x2 + yº + ) + (a + b + c)( ax ? + by+ + cza) = F .
    4994 Ex. (3). — The condition that F may becometwo right
    lines is AA' (00' — AA ') = 0 .
    PROOF. — Referring to Ex. (1 ), A = bc, B = ca , C = ab, F = G = H = 0,
    A ' = B ’ = (' = l ; therefore, in (4981) , a' = B + C = a ( b + c ), & c. Hence
    the discriminant A of F = abc ( b + c) (c + a )(a + b),
    or abc { (a + b + c)(bc + ca + ab) - abc } = the above, by (4945).
    4995 To reduce the two conics u , u ' to the forms
    . x² + y + x = 0, ax ? + By² + y2 = 0.
    By (4945 ), a , ß , y will be the roots of the cubic
    Ak – Ok? + O'k — A ' = 0 .................. (1),
    and x?, y , za will be found in terms of u , u and F ,by solving
    the three equations at + y2 + 7 = 1 , ax? + By: + y2 = u and (by
    4994 ), a (B + y) aº + B (y + a) yº + y (a + B) z* = F . .... (2).
    4996 Ex. (1) : Given x² + y’ + 2y + 2x + 3 = 0; x2+ 2y* + 4y + 2x + 6 = 0 ;
    to be reduced as above. To compute the invariants , we take
    - a b c f g h
    = 1 1 3 1 1 0 in the first equation .
    • and 2 6 2 1 0 in the second .
    = 1
    therefore в с F G H
    = 2 2 1 - 1 - 1 1 in the first equation .
    and = 8 5 2 - 2 - 2 2 in the second.
    Therefore (4938 , '9 ) A = 1, 0 = 6 , o ' = ll, A ' = 6 . The roots of equa
    tion ( 1 ) are now 1, 2 , 3 . Therefore (2 ) becomes 5X + 8Y? + 92* = F .
    Computing F also by (4981) with the above values of A , B, & c., we get the
    three equations as under, introducing z for the sake of symmetry,
    X ? + Y ? + 7 = x² + y + 3x + 2yz + 2zx ,
    X ? + 2Y + 32° = x + 2y² + 6z + 4y2 + 2zx,
    5X + 8Y? + 97° = 5x2 + 8y + 2222 + 16yz + 10za ,
    The solution gives X = x + 1 , Y = y + 1, 2 = l , and the equątions in the
    forms required are (x + 1) ’ + (y + 1)' + 1 = 0, (x + 1)' + 2 (y + 1)' + 3 = 0 .

    4997 Ex. ( 2). — To find the envelope of the base of a triangle inscribed
    in a conic u so that two of its sides touch 1 .
    4 T
    690 ANALYTICAL CONICS.

    Let u = x + y + z2 - 2yz – 2zx — 2xy — 2hkxy,


    and u ' = 2fyz + 2gzx + 2hæy,
    x and y being the sides touched by u . Then u + ku' will be a conic touched
    by the third side z . By finding the invariants, it appears that o — 440 '
    = 4AA 'k , whence k is determined , and the envelope becomes
    (0° - 4A0') Ú ' + 4AA'u = 0.
    Compare (4970).

    4998 The tangential equation of the two circular points at


    infinity (4717) is
    1P + m ? = 0 .
    PROOF. — This is the condition that dx + uy tv should pass through either
    of those points, since c£iy = c is the general form of such a line.

    4999 U = 0 being the tangential equation of a conic, the


    discriminant of kU + U ' is
    AB + 40k + A'Ok + A ”?.
    PROOF. — The discriminant of kU + U ' is identical in form with (4937 ) ,
    but the capitals and small letters must be interchanged . Let then the dis
    criminant be Ak + ők + Oʻk + Ā = 0 . We have
    J = A ' (4670), 7 = (BC - F ?) A ' + & c. = A 'aA + & c. (4668) = ao'.
    Similarly ē ’ = A ' , Ā = A .

    5000 If o , o be the invariants of any conic U and the pair


    of circular points 1? + m ? (4998) ; then 0 = 0 makes the conic
    a parabola , and O ' = 0 makes it an equilateral hyperbola .
    Proof. — The discriminant of kU + 1 + M is kA?+ k (a + b) A + ab - hº.
    For, as above, Ā = A°; 7 = A 'aA + B 'ba = (a + b) A since A ' = B ' = 1,
    C ' & c. = 0 ; Ē ' = ( A ' B ' - Hº) C = C = ab - h * ; and A = 0 . The rest fol.
    lows from the conditions (4471) and (4474).
    5001 The tangential equation of the circular points is, in
    trilinear notation ( see the note at 5030),
    + mº + v — 2uv cos A – 2vd cos B — 2dge cos C .
    Proof : \' + Me = 0 , in Cartesians, shows that the perpendicular let fall
    from any point whatever upon any line passing through one of the points is
    infinite . Therefore , by (4624 ) .

    5002 The conditions in (4689) and (4690 ) , which make the


    general conic a parabola or equilateral hyperbola , may be
    obtained by forming and e ' for the conic and equation
    (5001) and applying (5000).
    INVARIANTS AND COVARIANTS. 691

    5003 If oʻ? = 40 , the conic passes through one of the


    circular points.
    5004 When u ' in (4984 ) reduces to 12 + u ”, that is, to the
    circular points at infinity, F becomes the locus of intersec
    tion of tangents to u at right angles, and produces the equa
    tions of the director-circle (4693) and (4694 ).

    5005 The tangential equation of a conic confocal with U is


    kU + 12 + u ? = 0 ;
    5006 And if the left side, by varying k , be resolved into
    two factors, it becomes the equation of the foci of the system .
    PROOF. — Since 1? + pel represents the two circular points at infinity (4998 ),
    kU + 1 % + m = 0 , by (4914 ), is the tangential equation of a conic touched by
    the four imaginary tangents of U from those points. But these tangents
    intersect in two pairs in the foci of U (4720) ; and, for the same reason , in
    the foci of kU + 1 + u ?, which must therefore have the same foci.
    If kO + 1 % + u ' consists of two factors, it represents two points which , by
    (4913), are the intersections of the pairs of tangents just named, and are
    therefore the foci.

    5007 The general Cartesian equation of a conic confocal


    with u = 0 (4656 ) is
    k’Au + k {C (x2 + yº) — 2Gx — 2Fy + A + B } + 1 = 0 .
    ProOF. - (5005) must be transformed. Written in full, by (4664 ) , it
    becomes (kA + 1) 12 + (kB + 1 ) u ' + kCv?. Hence, by (4667), the trilinear
    equation will be
    { (kB + 1) kO —KF2} a ’ + & c. = k? (BC - F2) a’ + kCa’ + & c.
    = k ’aAa’ + kCa² + & c., (4668 )
    and so on, finally writing x, y, 1 for a, b , y .

    TO FIND THE FOCI OF THE GENERAL CONIC (4656 ).


    (First Method .)
    5008 Substitute in kU + 12 + res either root of its discriminant
    k ’ A ? + k (a + b ) Atab - h2 = 0 (5000 ) , and it becomes re
    solvable into two factors (Axí tuyítv) (1x, + ry2 + v ). The
    foci are xıyı and xyz, real for one value of k and imaginary
    for the other .
    Proof. — By (5006 ) the two factors represent the two foci, consequently
    the coordinates of the foci are the coefficients of 1 , M , » in those factors.
    692 ANALYTICAL CONICS.

    (Second Method.)
    5009 Let xy be a focus ; then , by (4720), the equation of an
    imaginary tangent through that point is ( - x ) + i (n - y ) = 0
    or & + in - ( x + iy ) = (). Therefore substitute, in the tangential
    equation (4665 ) , the coefficients l = l , uri, v = - (x + iy ) ,
    and equate real and imaginary parts to zero. The resulting
    equations for finding x and y are, with the notation of (4665) ,
    5010 2 (Cx — G ) = A ( a − 6 + { 4h + (a − b)"} ].
    5011 2 (Cy - F ) = A [6 – a + v {4h + (a − b)2} ].
    5012 If the conic is a parabola, C = 1, and the coordinates
    of the focus are given by
    (F2+ G2) x = FH + 1 ( A – B ) G ,
    (F2 + GP) y = GH - } (A – B ) F .
    5013 Ex. — To find the foci of 2x2 + 2xy + 2y + 2x = 0 . By the first
    method , we have
    a, b , c, f, g , h , from which A = - 2 . The quad
    = 2, 2, 0 , 0, 1, 1 | ratic for k is
    and A , B, C, F, G , H k‘ A ’ + 4kA + 3 = ( 2k - 3 )(2k - 1) = 0 ,
    = 0 , - 1, 3, 1, - 2, 0 therefore k = or .
    Taking , kU + 1² + M = ( - 48+ 31° + 24v — 4v1 ) + 18 + pi = 0 ,
    or 212– 120d M ? + 9y2 + 6uv = 0 .
    Solving for 2, this is thrown into the factors
    ( 21 + x72 – 3 (2 + V2) v } { 21 –4V 2 – 3 (2 – V2) » } .
    Therefore the coordinates of the foci, after rationalizing the fractions, are
    2 - 12 V2 - 1 2 + 12
    V2 + 1 2
    3 and
    3 33
    5014 Otherwise, by the second method , equations (5010, ' 1 ) become, in
    this instance , (3x + 2) ² = + 2 , (3y - 1) ' = £2, the solution of which pro
    duces the same values of x and y .

    5015 When the axes are oblique, the coordinates x, y of a


    focus are found from the equations
    {C (x + y cosw) — F cosw — G }? = {A (VT?– 4J + 2a - 1)
    (Cy - F ) sin ’w = } 4 (VT2 — 4J - 2a + I),
    where I and J are the invariants (4955) and (4954) respec
    a
    INVARIANTS AND COVARIANTS. 693

    tively . The equations may be solved for x' = x + y cosw and


    y ' = y sin w , which are the rectangular coordinates of the focus
    with the same origin and x axis .
    PROOF. — Following the method of (5009) , the imaginary tangent through
    the focus is, by (4721) , & - & +- (n - y ) (cos w + i sin w ). The two equations
    obtained from the tangential equation are, writing Aa for BC - F" , & c .
    (4668 ),
    X - Y ’ = - A (a + b - 2h cosw — 2a sin ’w ), XY = A (h sinw - a sin w cosw ) ;
    where X = ( (x + y cosw) – F cosw — G and Y = (Cy - F ) sinw.
    5016 If the equation of the conic to oblique axes be
    ax' + 2hxy + by + c = 0 ,
    the equationsfor determining the foci reduce to
    y (x + y cosw) _ x(y + x cosw ) – 0
    a cos w - h - b cos w h- ab - h2

    5017 The condition that the line læ + uy + vz may touch the


    conic u + (\'x + r'y + víz) is
    U + $ (qur' — u'v, va' - v 1 , dpí' — ' ) = 0 . (4656, 4936,'74)
    5018 or (A + U ') U = II , (4938 )
    where 211 = X' U + uu + v' U ,. (4674 )
    PROOF. - Put a + 1 " for a , & c. in U of (4664). The second form follows
    from the first through the identity
    Ap (ur' - u 'v, & c.) = UU' – IT?.
    5019 Otherwise, let P ' = Ux + 1 , 4 + 4 ,1% , the polar of
    a ', y', (4659 ), then the condition that P ' may touch u + P ' ?
    becomes , in terms of the coordinates of the poles,
    5020 ( 1 + u ") u' = $xx"' + bvy" + $ 2,2" . (See 4657).
    PROOF. — If we put uz , Uy', Ug , from (4659), for \ , u , v in U to obtain the
    condition of touching, the result is Au ' ; and similar substitutions made in
    II give A (P " + & c .), therefore (5018) becomes (1 + " " ) u' = (Px-ce" + & c.).

    5021 The condition that the conics


    u + ( ' + u'y + víz)?, u + (\" x + u " y + v" z)
    may touch each other is
    ( A + U )( 1 + U”) = ( A + II)”. (4938-74)
    694 ANALYTICAL CONICS.

    PROOF. — Make one of the common chords


    (1'x + p 'y + vʻz) + (1 " x + p " y + v" z )
    touch either conic by substituting X X " for 1 , & c . in (5018 ). The result
    is ( A + U ') ( U ' + 211 + U " ) = ( U 'EII) , which reduces to the form above.
    5022 The condition , in terms of the coordinates of the poles
    of the two lines, is found from the last , as in (5019), and is
    (1 + u')( 1 + u " ) = { 1 + ( 6xx" + bwy" + $23") } .

    5023 The Jacobian, J, of three conics u , v , w , is the locus


    us

    of a point whose polars with respect to the conics all meet in


    a point. Its equation is
    Q7X + hiy + 81% , 0 , x + hay + g2% , Aza + hy + 83 %
    ha + by + fiz, hqx + b2y + f % , h3x + by + f; = 0.
    gix tfiy + c % , 82x + fay + C2 % , g3x + foy + Cy !
    Proof. — The equation is the eliminant of the equations of the three
    polars passing through a point Enl, viz., u75 + u ,n + uzb = 0 , 0 5 + 0 ,1 + v ,5 = 0 ,
    W .6 + w,n + w . = 0. See (4657) and ( 1600).

    5024 The equation of a 1 a B ? ya B y ya a B


    conic passing through five | α β γ βγ, για, α, β,
    points aißiyo, a ,ß272, & c . is
    the determinant equation a Bi yi B2y. Y .Az aß, = 0.
    annexed ; and the equation aš Bš yả BsY: Y:Az azBs
    of a conic touching five | aš Bi yi Biyo yo as aße
    right lines dyMyvi, 12/12 V2, & c . aB B : : Y Az azB :
    is the same in form , 1, u, v
    taking the place of a , ß , y .
    PROOF. — The determinant is the eliminant of six equations of the type
    ( 4656 ) in the one case and (4665 ) in the other. By (583).
    5025 If three conics have a common self-conjugate triangle,
    their Jacobian is three right lines .
    Proof. — The Jacobian of a,22 + bayº + cz?, azãº+ bay’ + com?, (ngd ? + b3y + esz?
    is, by (5023), xyz = 0.
    For the condition that three conics may have a conimon point, see
    Salmon 's Conics, 6th edit., Art. 389a , and Proc. Lond . Math . Soc., Vol. IV .,
    p. 404 , J. J. Walker, M .A .
    INVARIANTS AND COVARIANTS. 695

    5026 _ A system of two conics has four covariant forms


    u , u ', F , J , connected by the equation
    J = F3 - F2 (Ou' - O ’u ) + F (A 'Ou? + AO ’u '2)
    + Fuu' (OOʻ — 344') – AA”?u8 – A’A %u'3
    + A 'u ’ u ' (2A0 ' — ©2) + Au' u (2A ' Q - O '?).
    Proof. – Form the Jacobian of u , u ', and F . This will be the equation of
    the sides of the common self-conjugate triangle (4992, 5025). Compare the
    result with that obtained by the method of (4995).
    5027 By parity of reasoning , there are four contravariant
    forms U , U ' V , T where I is the tangential equivalent of J ,
    and represents the vertices of the self-conjugate triangle . Its
    square is expressed in terms of U , U ', V and the invariants
    precisely as JP is expressed in (5026 ) .
    5028 The locus of the centre of a conic which always
    touches four given lines is a right line.
    Proof. — Let U = 0 , U ' = 0 be the tangential equations of two fixed
    conics, each touching the four lines ; then , by (4914 ), Ū + kU ' = 0) is another
    conic also touching the four lines. The coordinates of its centre will be
    G + kG ' F + kF '.
    C +k wa4 C + kC"is by by (4402). The point is thus seen, by (4032), to lie
    on the line joining the centres of the two fixed conics and to divide that line
    in the ratio kc' : C .
    5029 To find the locus of the focus of a conic touching four
    given lines.
    In the equations (5010, '1) for determining the coordinates of the focus,
    write A + kĂ ' for A , & c., and eliminate k . The result in general is a cubic
    curve. If , be parabolas, 8 + kE' is a parabola having three tangents in
    common with and X . If C = ( ' = 0 the locus becomes a circle . If the
    conics be concentric, they touch four sides of a parallelogram , and the locus
    is a rectangular hyperbola .
    Note on Tangential Coordinates.
    5030 It must be borne in mind that a tangential equation in trilinear
    notation (that is , when the variables are the coefficients of a, b , y in the
    tangent line la + mß + ny) will not agree with the equation of the same locus
    expressed in the tangential coordinates 1 , M , v of (4019). Thus, to convert
    equation (5001), which , for distinctness, will now be written
    ľ + m2 + m2 — 2mn cos A - 2nl cos B - 21m cos C = 0
    into tangential coordinates, we must substitute, by (4023), al, bu, cv for
    l, m , n . The equation then becomes
    a’l? + 6 %j ? + ( y2 – 2bc cos Aur - 2ca cos Bvd - 2ab cos Clue = 0 .
    Put 2bc cos A = b + A ’, & c., and the result is the equation as presented
    in (4905) .
    Corrigenda. - In (4678) and (4692) erase the coefficient 2 ; and in (4680) and (4903) supply
    the factor 4 on the left of the equation .
    THEORY OF PLANE CURVES.

    TANGENT AND NORMAL.

    5100 Let P (Fig .90) be a point on the curve AP ; PT,PN,


    PG , the tangent, ordinate, and normal intercepted by the a
    axis of coordinates. See definitions in (1160 ). Let ZPTX

    5101 tany = y by(1403); siny = y cosy = doce


    5104 Sub-tangent NT = yxy, Sub-normal NG = yy..
    5106 PT = y V (1 + x ), PT' = x V (1 + y ).
    5108 PG = y V (1 + y ), PG’ = xv (1 + x ).
    Let OP = r (Fig. 91), u = r , AOP = 0, OPT = \;
    Arc AP = 8. Then , by infinitesimals,
    COS

    5110 sinø = r que cos element tan$ = r um


    5113 (dx)2 + (dy) = (ds) , s, = v (1 + y ).
    5114
    tan y = ro sin 6 + r cos 6 (1768)
    ro cos 0 - r sin 0

    5115 Interceptsof Normal OG = ran , OG = r (Fig . 90)


    TANGENT AND NORMAL. 697

    PROOF : OG = " sin OPG _ rcos


    = 'sin PGN = sin NPT = " "die
    5116 8 = v (n + r ), Sn = (1 + r20 ).
    Proof. — By 10, = sino and tan o = r0, (5110 ).

    EQUATIONS OF THE TANGENT AND NORMAL .


    The equation of the curve being y = f (x ) or u = q (x, y)
    = 0 , the equation of the tangent at ay is
    5118 n - y = 22 (5 — X), (4120)
    5119 or $y: — = xyz - y,
    5120 or Eur + nu , = Xux + yuy. (1708 )
    5121 If q (x , y ) = vu + un -it ... + vo, where vn is a homogeneous function
    of x and y of the nth degree, the constant part forming the right member of
    equation (5120 ) takes the value
    - V1 -1 - 209 -2 — ... - (n - 1) v, - nvo.
    By Euler's theorem ( 1621) and (x , y) = 0 .
    The equation of the normal at æy is
    5122 (4122)
    n-y = - dy (5 —x),
    5123 or Exy + n = XX , ty,
    5124 or $u , - nu = xu, - yuz. (1708)

    POLAR EQUATIONS OF THE TANGENT AND NORMAL.


    Let r , o be the coordinates of P (Fig . 91), and R , O those
    of S , any point on the tangent at P ; and let u = p -?, U = R - ,
    p = 0 - 0 ; the polar equation of the tangent at P will be
    5125 R' ~= d . (r sin o), or U = u cosr + u , sin t. .
    The polar equation of the normal is

    5127 R = 2. arcose or U = u cost — u’9, sin 7.


    U COST : IT

    Proof. – From 7R . = OP sin OSP


    os _ sin ( @ —p -) , and from tan g = rill,
    OPS _ sin sin
    (5112). Similarly for the normal.
    4 U
    698 THEORY OF PLANE CURVES .

    Let OY = p be the perpendicular from the pole upon the


    tangent, then
    5129 p = r sin $ = (u? + u;) * (5112)

    5131 XYZ - Y xºxt ? (4064 & 5119 , '20)


    ✓ (1 + y ) ✓

    OS,drawn at right angles to p to meet the tangent, is


    called the polar sub-tangent.
    5133 Polar sub-tangent = r20 (5112)

    RADIUS OF CURVATURE AND EVOLUTE .

    Let g , n be the centre of curvature for a point xy on the


    curve, and p the radius of curvature ; then
    5134 (x - &)²+ (y - n)' = p .................. (1).
    5135 (x — $) + (y - n) Y: = 0 ..................
    1 + 4: + (y - n ) Y2x = 0 .................. (3 ).
    Proof. — (2) and (3 ) are obtained from ( 1) by differentiating for 2, con
    sideringą, n constants .

    The following are different values of p :


    ' e

    5137 = (1 + yx ) _ ( 83 + $ 3)
    Y2 * 26 , , - $ 2003 – 22, %
    51 39
    5139 = (x + ) = 72, Rayle
    5141 = JehtyA) = =
    5144
    5146 = $y = p + P 24 = Por p
    Proofs. — For (5137), eliminate x - 5 and y - n between equations (1),
    (2 ), and (3 ) .
    (5138) is obtained from the preceding value by substituting for Y . and
    RADIUS OF OURVATURE AND EVOLUTE . 699

    Y2x the values (1708, '9 ). The equation of the curve is here supposed to be
    of the form o (x , y) = 0 .
    For (5139) ; change the variable to t. For (5140 ) ; make t = s.
    For (5141- 3) ; let PQ = QR = ds ( Fig . 92) be equal consecutive elements
    of the curve. Draw the normals at P , Q , R , and the tangents at P and Q to
    meet the normals at Q and R in T and S . Then, if PN be drawn parallel
    and equal to QS, the point N will ultimately fall on the normal Q0. Now
    the difference of the projections of PT and PN upon OX is equal to the pro
    jection of TN . Projection of PT = dsx, ; that of PN or QS = ds ( x, + 228 ds)
    (1500) ; therefore the difference = ds22 ds = TN cos a . But TN : ds =
    ds : p , therefore pxey = cos a . Similarly pyz = sin a .
    For (5144 ) ; change (5137) to r and 0, by (1768, ’9) .
    (5145) is obtained from p = rry = - ' and (5129) ; or change (5144 )
    from r to u by r = u -?.
    (5146.) In Fig. ( 93), PQ = p , PP' = ds, and PQP' = dy .
    (5147.) In Fig . (93), let PQ , P 'Q be consecutive normals ; PT, P' T '
    consecutive tangents ; OT, OT', ON, ON ' perpendiculars from the origin
    upon the tangents and normals. Then, putting p for OT = PN , q for
    PT = ON , and dy for TPT' = PQP', & c ., we have

    and p = PQ = p + QN = p + P24
    (5148.) dp = r cos o dy and cos o = r,. Eliminate cos p .

    5149 DEF. — The evolute of a curve is the locus of its centre


    of curvature. Regarding the evolute as the principal curve ,
    the original curve is called its involute .
    5150 The normal of any curve is a tangent to its evolute.
    PROOF. — By differentiating equation (5135 ) on the hypothesis that & and
    7 are variables dependent upon x, and combining the result with ( 3 ), we
    obtain yang = - 1 .
    In ( Fig . 94 ), the normal at P of the curve AP touches the evolute at Q .
    Otherwise the evolute is the envelope of the normals of the given curve.
    If æy and En are the points P , Q , we have the relations
    5151 dź _ dn _ dp.
    p d $ + ? =p Ydn = dp, § -Xn- y P
    = * o'
    Proof. — Take Qn = dę and ns = dn, then Qs = dp. The projection of
    Qn, ns gives dp in (5151) and proportion gives (5152).
    5153 The evolute and involute are connected by the for
    mulæ below , in which ro', p', s' in the evolute correspond to
    rep, s in the involute.
    5154 p + s' = constant ; p = pe - pº; po ? = pi+ p2 - 2pp.
    700 THEORY OF PLANE CURVES.

    PROOF. — From Fig . (94), dp = £ds', & c., s being the arc RQ measured
    from a fixed point R . Hence, if a string is wrapped upon a given curve, the
    free end describes an involate of the curve. (3155, '6 ) from Fig . (93).

    515 To obtain the equation of the evolute ; eliminate & and


    y from equations (5135, '6 ) and the equation of the curve.
    5158 To obtain the polar equation of the evolute ; eliminate
    r and p from (5156) and (5157) and the given equation of the
    curve r = F (p ).

    5159 Ex. — To find the evolute of the catenary y = (e te ). Here


    ys= (e = V(19=49); y = 2(ečte > = ;sothat equations
    (5135,’6) become
    ( – )+(3–») V(20) = 0 and 1 + +(y =») = 0.
    From these we find (7 * — 4c?). Substituting in the
    equation of the curve, we obtain the required equation in Ğ and n.

    INVERSE PROBLEM AND INTRINSIC EQUATION .

    An inverse question occurs when the arc is a given func


    tion of the abscissa, says = ( ) ; the equation of the curve
    in rectangular coordinates will then be
    5160 y = S ✓ (8 – 1) dx . [ From (5113 ).
    - -- -

    5161 The intrinsic equation of a curve is an equation inde


    pendent of coordinate axes. Let y = $ (2 ) be the ordinary
    equation, taking for origin a point 0 on the curve (Fig. 95 ),
    and the tangent at O for x axis. Let s = arc OP, and t the
    inclination of the tangent at P ; then the intrinsic equation of
    the curve is
    5162 s = S sec yox ,dy ;
    where x is found from tant = $ (w ).
    ASYMPTOTES. 701

    To obtain the Cartesian equation from the intrinsic equa


    tion :
    5163 Let s = F ( 4) be the intrinsic equation . Eliminate *
    between this and the equations
    x = S cosyds, y = S sin y ds.
    5165 The intrinsic equation of the evolute obtained from
    the intrinsic equation of the curve, s = F (V ) , is

    = l, a constant. (5154)

    5166 The intrinsic equation of the involute obtained from


    s' = F (x ), the equation of the curve, is
    s = S{l - F ( )} dų .
    For dit is the same for both curves (Fig. 94), 4 only differing
    by jr , and s = $ pdy .

    ASYMPTOTES.

    5167 DEF . — An asymptote of a curve is a straight line or


    curve which the former continually approaches but never
    reaches. ( Vide 1185) .

    GENERAL RULES FOR RECTILINEAR ASYMPTOTES.


    5168 RULE I . - Ascertain if y , has a limiting value when
    x = 00 . If it has, find the intercept on the x or y axis, that
    is, x - yx , or y - xy . (5104 ).
    There will be an asymptote parallel to the y axis when y , is
    infinite, and the x intercept finite , or one parallel to the x aris
    when y , is zero and the y intercept finite.
    5169 RULE II. — When the equation of the curve consists of
    homogeneous functions of x and y , of the mth, nth, fc. degrees,
    so that it may be written

    2* * (* ) +x*x( % ) + & c. = 0 ...........(1);


    702 THEORY OF PLANE CURVES.

    putux + B for y and expand $ (x + 2), &c., by (1500). Divide


    (1) by x ”,and make x infinite; then 9 (M ) = 0 determines u.
    Next, put this value of u in ( 1 ), divide by xm - 1, and make x
    infinite ; thus B & ' (u ) + x (u ) = 0 determines ß . Should the
    last equation be indeterminate, then
    %B2%" (u) + Bx'(u) + 4 ( ) = 0
    gives two values for B , and so on .
    When n is < m - 1, B = 0 , and when n is > m - 1 , B = 00 .
    5170 RULE III. - If • (x , y ) = 0 be a rational integral equa
    tion , to discover asymptotes parallel to the axes, equate to zero
    the coefficients of the highest powers of x and y, if those co
    efficients contain y or x respectively .
    To find other asymptotes - Substitute ux + B for y in the
    original equation , and arrange according to powers of x . To
    find y , equate to zero the coefficient of the highest power of x .
    To find ß , equate to zero the coefficient of the next power of x ,
    or, if that equation be indeterminate, take the next coefficient in
    order, and so on .

    5171 RULE IV . - If the polar equation of the curve be r = f (0)


    and if r = 00 makes the polar subtangent r20, = c , a finite quan
    tity , there is an asymptote whose equation is rcos (0 - a ) = c ;
    where a + = f - ( 00 ) = the value of 0 of the curve when r is
    infinite.
    5172 Asymptotic curves. -- In these the difference of corres
    ponding ordinates continually diminishes as æ increases.
    As an example, the curves y = ° (cn ) and y = 0 are asymptotic.

    5173 Ex. 1. - To find the asymptotes of the curve


    (a + 3x )( + y ) = 4. 8 ...... ...................... ( 1) .
    The coefficient of y ', a + 3x = 0, gives an asymptote parallel to the y axis.
    Putting y = mx + B , ( 1) becomes
    (a + 3x )(x² + m *w* + 2mp3.x + B ) – 4.x8 = 0 .................. (2 ).
    The coefficient of * *, 3 ( 1 + ?) — 4 = 0 gives u = Substituting this
    value of u in (2), the coefficient of ze becomes 4 + 6% ; and this,equated
    2a
    to zero , gives B = F 3 ./-4 35. Hence the equations of two more asymptotes
    are 3y / 3 = = (3x — 2a ).
    SINGULARITIES OF CURVES. 703

    Ex. 2 . - To find an asymptote of the curve r cos 0 = a cos 20. Here


    , do a’ cos 20
    dr a cos 20 sin 0 - 2a sin 20 cos 0
    When r = 0 , 0 = it, and 70, = - a. Hence the equation of the asymp- .
    tote is r cos 0 = - a .

    SINGULARITIES OF CURVES.
    5174 Concavity and Convexity . – A curve is reckoned convex
    or concave towards the axis of æ according as yyz is positive
    or negative.

    POINTS OF INFLEXION .
    5175 A point of inflexion (Fig .96) exists where the tangent
    has a limiting position , and therefore where yx takes a maxi
    mum or minimum value.
    5176 Hence Yzx must vanish and change sign, as in ( 1832).
    5177 Or, more generally , an even number of consecutive
    derivatives of y = $ ( ) must vanish , and the curve will pass
    from positive to negative, or from negative to positive, with
    respect to the axis of a , according as the next derivative is
    negative or positive. [See ( 1833).

    MULTIPLE POINTS.
    5178 A multiple point, known also as a node or crunode,
    exists when yc has more than one value, as at B (Fig . 98). If
    ♡ (@ , y ) = 0 be the curve, px and 0 ,, must both vanish , by
    (1713). Then , by (1704 ) , two values of yu determining a
    double point, will be given by the quadratic
    $29? + 24x9x + 2 = 0 ............ ... ( 1).
    5179 If Pzp $2u2 Paxy also vanish ; then, by (1705) , three
    values of yx, determining a triple point, will be obtained from
    the cubic
    psy? + 302392 + 3 y2rY: + $ e = 0 ..... ...... (2).
    5180 Generally , when all the derivatives of ¢ of an order
    an
    704 THEORY OF PLANE CURVES.

    less than n vanish , the equation for determining y: (put = z)


    may be written
    (zd, + d .)" ¢ ( ', y) = 0 .
    Proof.— Let ab be the multiple point. Then, by (1512),
    © (a + h,6 + k) = (hd. + kdo)" (x, y)
    + terms of higher order which vanish when h and k are small.
    And h
    = - 4 , = dx
    44 in thelimit.

    CUSPS.
    5181 When two branches of a curve have a common tan
    gent at a point, but do not pass through the point, they form
    a cusp , termed also a spinode or stationary point.
    5182 In the first species, or ceratoid cusp ( Fig. 100 ), the
    two values of yu have opposite signs.
    5183 In the second species, or ramphoid cusp (Fig. 101),
    they have the same sign .

    CONJUGATE POINTS.
    5184 A conjugate point, or acnode, is an isolated point whose
    coordinates satisfy the equation of the curve. A necessary
    condition for the existence of a conjugate point is that de and
    $, must both vanish.
    Proof. — For the tangent at such a point may have any direction , there.
    fore 2x is indeterminate (1713 ).

    5185 There are four species of the triple point according as


    it is formed by the union of
    (i.) three crunodes,as in (Fig. 102) ;
    ; (ii.) two .crunodes and a cusp, as in (Fig . 103) ;
    (iii.) a crunode and two cusps, as in (Fig . 104) ;
    (iv .) when only one real tangent exists at the point.
    5186 Ex. — The equation y2 = (x — a )(x — b)(x — c )* when
    a < b < c represents a curve, such as that drawn in (Fig. 97).
    - , * Salmon 's Higher Plane Curves, Arts. 39, 40.
    UM

    SINGULARITIES OF CURVES. 705

    When b = c the curve takes the form in (Fig . 98). But


    if, instead , b = a , the oval shrinks into a point A ( Fig . 99 ).
    If a = b = c the point A becomes a cusp , as in (Fig . 100 ). .

    A geometricalmethod of investigating singular points.


    5187 Describe an elementary circle of radius r round the
    point x , y on the curve • (x , y ) = 0 , intersecting the curve in
    the point æth, y + k . Let h = r cos 0 , k = r sin 0. Expand
    $ (a + h , y + k ) = 0 by (1512), and put $ x = K sin y, P , =
    K cos y. We thus obtain
    K sin (y + 0) + ($ 20 cos?0 + 20ysin 0 cose + º2, sin?0) + b = 0
    R being put for the rest of the expansion .........(1),
    According as the quadratic in tano,
    $ 2x + 29xy tan0 + P2, tan 0 = 0 ,
    has real, equal, or imaginary roots ; i .e., according as
    P v - $ 22024 is positive, zero , or negative, xy will be a crunode ,
    a cusp, or an acnode. By examining the sign of R , the
    species of cusp and character of the curvature may be deter
    mined .
    Figures (105 ) and (106 ), according as R and $ 2 have
    opposite or like signs, show the nature of a crunode ; and
    figures (107) and (108) show a cusp.
    PROOF. – At an ordinary point the circle cuts the curve at the two points
    given by 0 = - , 0 = - y . But, if px and P , both vanish, there is a
    singular point. Writing A , B , C for P233 Cry ,P24 , equation (1) now becomes
    C cos'o { tan’0 + 2B tan 6 + A } + 2R = 0...... .
    (i.) If Bº > AC, this may be put in the form
    C cos 0 ( tan 0 – tan a ) (tan 0 – tan |
    and the points of intersection with the circle are given by 0 = a , B , ata,
    and a + ß . (Figs. 105 and 106 .)
    (ii.) When B = AC, we may write equation (1)
    C cos? 0 (tan 8 - tan a ) + 2 = 0 .
    If R and C bave opposite signs, there is a cusp with a for the inclination of
    the tangent (Fig . 107). So also , if R and C have the same sign, the inclina
    tion and direction being ata ( Fig . 108 ) . The cusps exist in this case
    because R changes its sign when # is added to 0 , R being a homogeneous
    function of the third degree in sin 6 and cos 0.
    (iii.) If B * < AO, there are no real points of intersection, and therefore xy
    is an acnode.
    4 x
    706 THEORY OF PLANE CURVES.

    CONTACT OF CURVES.

    5188 A contact of the nth order exists between two curves


    when n successive derivatives, Yr , ... Yne or ro, ... Ire, corres
    pond. The curves cross at the point if n be even . No curve
    can pass between them which has a contact of a lower order
    with either.
    Ex. — The curve y = ° (20) has a contact of the nth order, at the point
    where a = a, with the curve y = 0 (a ) + (x — a ) '(a ) + ... + n !
    9 * (a@ )) ..

    5189 CoR . - If the curve y = f (x ) has n parameters, they


    may be determined so that the curve shall have a contact of
    the (n — 1) th order with y = ¢ (a ).
    À contact of the first order between two curves implies a
    common tangent, and a contact of the second order a common
    radius of curvature.

    Conic of closest contact with a given curve.


    5190 Lemma. - In a central conic (Fig . of 1195 ),
    tan CPG =
    Proof. - Putting PCT = 0 , CPT = 4 , CP = r, CD = R , we have, by
    (1211), pe + R = a + b?, : rr, = - RR .. .................. (i.).
    Also Rr sin o = ab, by (1194), : : Rr20. = ab, by (5110) ......... (ii.) .
    Now tan CPG = -cotø = - (5112) = . = by (i.) and (ii.).
    But p= (4538), : = R' = tan CPG .
    5191 To find the conic having a contact of the fourth order
    with a given curve at a given point P .
    If Ö be the conic 's centre, the radius p = OP, and the
    angle v between r and the normalare found from the equations
    COS Y
    AV

    tanv = copr = 1
    and these determine the conic.
    ENVELOPES. 707

    Proof. - In Fig . 93, let O be the centre of the conic and P the point of
    contact. The five disposable constants of the general equation of a conic
    will be determined by the following five data : two coordinates of 0 , a com
    mon point P , a common tangent at P , and the same radius of curvature PQ .
    Since v = POT, do = POP', dy = TOT', and ds = PP', we have, in
    passing from P to P ', dv = P 'OT' – POT = dy - do. Now rdo = ds cos v,
    therefore cos V = dy — dv = 1 _ ob
    dv ;. .and tan v has been found in the
    lemma.

    The squares of the semi-axes of the same conic are the


    roots of the equation
    (9 + 62 — 3ac): x ? - 9aº ( 18 + 262— 3ac)(9 + 62 — 3ac) x + 729a*
    = 0,

    a , b, c being written for p, Pos P2 . The eccentricity is found


    from 9 (e - 2 (18 + 262 — 3ac)
    e1 - 9 + 62 _ 3ac
    Also the equation of the conic referred to the tangent and
    normal at the point is

    Ax® +2Bxy + Cy = 2y,


    where A = 5, B= - O= +
    Ed. Times, Math . Reprint, Vol. XXI., p . 87, where the demonstrations by
    Prof. Wolstenholmewill be found .

    ENVELOPES.
    5192 An envelope of a curve is the locus of the ultimate
    intersections of the different curves of the same species, got
    by varying continuously a parameter of the curve ; and the
    envelope touches all the intersecting curves so obtained .
    5193 Rule . - If F (x , y , a ) = 0 be a curve having the para
    meter a , the envelope is the curve obtained by eliminating a
    between the equations
    F (x , y, a ) = 0 and dqF (x , y , a ) = 0 .
    708 THEORY OF PLANE CURVES.

    PROOF.— Let a change to a th. The coordinates of the point of intersec


    tion of F (x , y , a ) = 0 and F (ic, y, a + h ) = 0 satisfy the equation
    F (x, y, a + h ) – F (x , y, a) = 0, i.e., dF (2, y, a ) = 0. (1404 )
    da

    5194 If F (x, y , a , b, c, ...) = 0 be the equation of a curve


    having n parameters a , b , c, ... connected by n - 1 equations,
    then , by varying the parameters, a series of intersecting
    curves may be obtained . The envelope of these curves will
    be found by differentiating all the equations with respect to
    a , b , c, & c ., and eliminating da, db , ... and a , b , ...
    5195 Ex. - In ( 2) of (5135), we have the equation of the normal of a
    curve at a given point wy ; &, , being the variable coordinates, and x , y the
    parameters connected by the equation of the curve F (x , y ) = 0 . By differ
    entiating for x and y , (5136 ) is found, and the elimination as directed in
    (5157) produces the equation of the evolutewhicb, by (5194), is the envelope
    of the curve.

    INTEGRALS OF CURVES AND AREAS .

    FORMULÆ FOR THE LENGTH OF AN ARC S.

    5196 s = ſds = Sv (1+93)dx = (v1 +dě,dy (5118)


    5200 = $ V(x +y )dt = ſv( *+r%)de (5116)
    5201 = Sv(2*8 +1)dr = Svon (5111)

    5203 Legendre's formula,s= pu+Spely.


    5204 The whole contour of a closed curye = pdy .
    PROOF. — In figure (93 ), let P, P ' be an element ds of the curve ; PT, P' T '
    tangents, and OT, OT theperpendiculars upon them from the origin ; OT = p ,
    PT = q. Then ds + P ' T ' - PT = TL, i.e., ds + dq = pdy ; therefore sta
    = Spdy. But qdy = - dp ; therefore s = pr + Spdy. Also, in integrating
    all round the curve, P ' T - PT taken for
    r2
    every point vanishes in the summa
    tion,or dq = 0. Therefore ds = ( pdiy.
    INVERSE CURVES. 709

    FORMULÆ FOR PLANE AREAS.


    5205 If y = ° (w ) be the equation of a curve, the area
    bounded by the curve, two ordinates (x = a , x = b ), and the x
    axis , is, as in (1902),
    A = S**(x)dx.
    5206 With polar coordinates the area included between two
    radii (@ = a , 0 = B ) and the curve is

    A = +Spa0 =+( pds= 15 procent


    Proof. — From figure (91) and the elemental area OPP'.
    5209 The area bounded by two circles of radii a , b, and the
    two curves 0 = ° (r), 0 = * (r) (Fig. 109).
    Oru (c)
    A = N°|
    Ja J $ (r )
    rdrdo = r { (r) – 4 (1)} dr.
    Here r { * (r) - ° (r )} dr is the elemental area between the
    dotted circumferences.
    5211 The area bounded by two radii of curvature, the curve,
    and its evolute (Fig. 110 ).
    A = * ſp*dy = ſpds.
    Proof. — From figure (93) and the elemental area QPP'.

    INVERSE CURVES.

    The following results may be added to those given in Arts. (1000 - 15) .
    5212 Let 7 , P be corresponding radii of a curve and its
    inverse, so that ror = k ; s, s' corresponding arcs, and 0 , $
    the angles between the radius and tangents, then

    as and $ = $'.
    Proof. — Let PQ be the element of arc ds, P' Q ' the element ds', and o
    the origin .
    Then OP .OP = OQ .OQ', therefore OPQ, OQ'P ' are similar triangles;
    therefore PQ : P ' Q ' :: OP : OQ ' = rir'; also ZOPQ = OQ ' P '.
    710 THEORY OF PLANE CURVES.

    5214 If p, p' be the radii of curvature,


    pp8+ = 2sin $.
    Proof. – From p = rsino, p = r' sin , we have
    p = k , therefore andit = ku7°2- 2017..............(i.).
    Also
    Also , therefore interes ... .......... (ii.).
    o
    therefore

    Now é = p (5148 ) , therefo by (i.) and (ii.).


    do p' dr dr

    Therefore = 2 sin p .

    5215 To find the equation of the inverse of a curve in


    rectangular coordinates, substitute
    kx2
    Xt2 and
    * * *
    for x and y in the equation ofthe given curve. 1 Curve .

    5216 The inverse of the algebraic curve


    Un + Un- 1 + Wm- + ... + Uy tuo = 0 ,
    where un is a homogeneous function of the nth degree, will be
    kanun + k (n -1)un-1 ( x2 + y ) + k ?(n –2)Un-2 (x2+ y ) + ...
    ... + 90( 22 + y ?)" = 0 .
    5217 The inverse of the conic wz + un + wo = 0 is
    k*uz + k*uj (x+ + y ) + uo (w * + ya) = 0 .
    5218 If the origin be on the curve, this equation becomes
    k’uz + u ; ( x + y ) = 0 .
    5219 The angle will also be unaltered in any curve,
    g = f (0 ) , if the inversion be effected by putting
    p = krn and 0 = no'.
    PROOF. —
    tang' = r’O (5112) = r’Orr- = p kny'n-1 = kr’ = r0, = tan .
    PEDAL QURVES. 711

    PEDAL CURVES.

    5220 The locus of the foot of the perpendicular from the


    origin upon the tangent is called a pedal curve. The pedal of
    the pedal curve is called the second pedal, and so on . Re
    versing the order, the envelope of the right lines drawn from
    each point of a curve at right angles to the radius vector is
    called the first negative pedal, and so on .
    5221 The pedal and the reciprocal polar are inverse curves
    (1000, 4844 .)

    AREA OF A PEDAL CURVE .


    5222 Let C , P , Q be the respective areas of a closed curve,
    the pedal of the curve, and the pedal of the evolute ; then
    P - Q = C, P + Q = } \ rºdų , 2P = C + } |r dy .
    PROOF. - With figure (93) and the notation of (5204),we have, by (5206),
    P = } $pdy , Q = 1 \ qdy ; therefore P + Q = . s (pº + q*) dy = i sredy.
    Also, taking two consecutive positions of the triangle OPT = A , we get
    OPT - OP' T' = dA = 8C + 0Q - SP . Therefore, integrating all round ,
    A = () = C + Q - P .

    5225 Steiner's Theorem . — If P be the area of the pedal of a


    closed curve when the pole is the origin , and P the area of
    the pedal when the pole is the point xy,
    P - P = 36 (2* + y*) —ax —by,
    where a = p cos de and b = "p sin 0do;
    0 being the inclination of p.
    Proor. — (Fig . 111.) Let LM be a tangent, S the point xy, perpendiculars
    OM = p and SR = p . Draw SN perpendicular to OM , and let' ON = Pı;
    then P' = 1[p*dy = 1/(P –P.)?d4 = [pd4 +1ſp;dy= [op,duy
    | p (æ cos 0 + y sin 8 ) do, by (4094 ), and d0 = dy.

    And "pidų = twice the area of the circlewhose diameter is OS.


    712 THEORY OF PLANE CURVES.

    5226 Cor. 1. - If P be given , the locus of xy is a circle


    whose equation is (5225 ), and the centre of this circle is the
    same for all values of P ', the coordinates of the centre being
    a and .
    5227 CoR. 2. — Let Q be the fixed centre referred to, and
    let QS = c. Let P " be the area of the pedal whose origin is
    Q ; then P - P = c.
    For a and b must vanish in (5225 ) when the origin is at the
    centre Q , and 22 + yº then = c .
    5228 Cor. 3. — Hence P" is the minimum value of P'. *

    ROULETTES.
    5229 DEF. — A Roulette is the locus of a point rigidly con
    nected with a curve which rolls upon a fixed right line or
    curve .

    AREA OF A ROULETTE.
    5230 When a closed curve rolls upon a right line, the area
    generated in one revolution by the normal to the roulette at
    the generating point is twice the area of the pedal of the
    rolling curve with respect to the generating point.
    PROOF. — (Fig . 112.) Let P be the point of contact of the rolling curve
    and fixed straight line, Q the point which generates the roulette . Let R be
    a consecutive point, and when R comes into contact with the straight line,
    let P 'Q ' be the position of RQ. Then PQ is a normal to the roulette at Q ,
    and P is the instantaneous centre of rotation . Draw QN , QS perpendiculars
    on the tangents at P and R . The elemental area PQQ' P ', included between
    the two normals QP, Q ' P ', is ultimately equal to PQR + QRQ . But PQR
    = dC , an element of the area of the curve swept over by the radius vector
    QP or r round the pole Q ; and QRQ = iridit ; therefore, whole area of
    roulette = C + 1 dy = 2P , by (5224 ).
    5231 Hence, by (5228), there is one point in any closed
    curve for which the area of the corresponding roulette is a
    * For a discussion of the pedal curves of an ellipse by the Editor of the Educ. Times and
    others, see Reprint, Vol. 1., p . 23 ; Vol. XVI., p . 77 ; Vol. XVII., p . 92 ; and Vol. xx., p. 106.
    ROULETTES. 713

    minimum . Also the area of the roulette described by any


    other point, distant c from the origin of the minimum roulette ,
    exceeds the area of the latter by # ca.

    5232 When the line rolled upon is a curve, the whole area
    generated in one revolution of the rolling curve becomes
    27

    C+1 * *(1+ % )dly,


    where p , é are the radii of curvature of the rolling and fixed
    curves , and C is the area of the former.
    PROOF. — (Fig . 113.) Instead of the angle dy , we now have the sum of
    the angles of contingence at P of the rolling curve and fixed curve, viz.,
    dif +d = dy (1+ aute = d4 (1+ ),
    since pdy = ds = p'dy ', by (5146).

    LENGTH OF THE ARC OF A ROULETTE.


    5233 If o and be corresponding arcs of the roulette and
    the pedal whose origin is the generating point ; then , when
    the fixed line is straight, o = ; and when it is a curve,
    5234
    - Sao = {(1+ 2 )at.
    PROOF. — (Fig . 112.) Let R be the point which has just left the straight
    line , Q the generating point, N , S consecutive points on the pedal curve .
    Draw the circle circumscribing RQNS, of which RQ == r is a diameter, and
    let the diameter which bisects NS meet the circle in K . Then , when the
    points P , R , P ' coincide, KN and RQ are diameters, and SRN = SPN = dų
    = QRQ '; therefore SN or ds = rdy = QQ' or do. When the fixed line is a .
    curve, do = rdy( 1+ %),as in (5232).
    RADIUS OF CURVATURE OF A ROULETTE .
    5235 Let a (Fig . 113) be the angle between the generating
    line r and the normal at the point of contact ; e , é the radii
    of curvature of the fixed and rolling curves, and R the radius
    of curvature of the roulette ; then ,
    72
    R =
    ppo cosar
    pt
    4 y
    714 THEORY OF PLANE CURVES.

    Proof. — Let consecutive normals of the roulette meet in 0 ; then


    OQ = R , PQ = 1, MPT = a .
    R - _ PM _ ds cossą,
    a and do = r (dy + dy' s , ds
    R QQ do p
    from which R is obtained . If the curvature of the roulette is convex to
    wards P ( Fig . 114 ), wemust write Rtr instead of R - r above.

    5236 The curvature is convex towards P when R is posi


    tive, that is, when the carried point Q falls within the circle
    whose diameter measured on the normal of the rolling curve
    = _PR_,. When Q falls without this circle, the curvature is
    - otp
    concave ; and when Q falls upon the circumference, the point
    is one of inflexion . The circle has for this reason been called
    the circle of inflexions.
    5237 In figure ( 163) let PA = P, PB = p , PQ = r , OQ = R ,
    as in (5235). Draw ÝCD, the circle of inflexions, with its
    diameter PC = paper , and therefore PD = pPlte_ncosa. From
    these values and proportion it follows that BC : BP : BA and
    QD : QP : Q0. Also, if the circle on diameter PE = PC be
    drawn, AE : AP : AB and OF : OP : OQ .
    5238 A simple construction for the centre of curvature of
    the roulette is the following . (Fig . 164, with letters as in
    5237.) At P draw a perpendicular to PQ to meet QB in N .
    Join NA, which will meet QP produced in 0 , the required
    point.
    PROOF.-- From equation (5235), assuming 0 to be the centre of curvature,
    we can deduce the relation (BA : AP ) (PO : OQ) (QN : NB) = 1, therefore,
    by (968 ), A , O, N are collinear points .

    THE ENVELOPE OF A CARRIED CURVE.


    5239 When a curve is rigidly connected with a rolling
    curve, it will have an envelope. The path of its point of
    contact with the envelope is a tangent to both curves , and
    therefore the normal, common to the carried curve and its
    envelope, passes through the point of contact P of the rolling
    and fixed curve.
    5240 The centre of curvature of the envelope is obtained as
    follows.
    TRAJECTORIES, 8C. 715

    In Fig . (163), from P draw a normal to the carried curve meeting it in


    Q , and let 8 on PQ be the centre of curvature of the envelope for the point
    Q ; and ( that of the carried curve. Then PS is found from
    pé1 + 1 = cos a ( 5 + po )
    5241 When the envelope is a right line, the centre of curva
    ture lies on the circle of inflexions (5236 ). When the carried
    curve is a right line, the same point lies on the circle PEF
    ( Fig . 163), and if the right line always passes through a fixed
    point, that point lies on the circle PEF.
    5242 If p be the perpendicular from a fixed point upon a
    carried right line whose inclination to a fixed line is t ; the
    radius of curvature of the envelope is p = p + P24, by (5147) .

    INSTANTANEOUS CENTRE.
    5243 When a plane figure moves in any
    e manner in its own
    ny manner wn

    plane, the instantaneous centre of rotation is the intersection


    of the perpendiculars at two points to the directions in which
    the points are moving ; and a line from the instantaneous
    centre to any point of the figure is the normal to the path of
    that point.
    Ex. — Let a triangle ABC slide with its vertices A , B always upon
    the right lines 0A, OB. The perpendiculars at A , B to 0A, OB meet in Q ,
    the instantaneous centre, and QC is the normal at C to the locus of C .
    Since AB and the angle AOB are of constant magnitude, OQ, the
    diameter of the circle circumscribing OAQB, is of constant magnitude.
    Hence the locus of the instantaneous centre Q is a circle of centre 0 and
    radius OQ.

    5244 Holditch's Theorem . - If a chord of a given length LM


    moves completely round a closed curve , the area enclosed
    between the curve and the locus of a point P on the chord is
    equal to acc' where c = LK , C = MK.
    5245 If the ends of LM move on different closed curves
    whose areas are 1 , M , while the area described by K is k , then
    Actuc - CC' .
    K =
    ctc
    PROOF. — (5244). Let the innermost oval in figure (134) be the envelope
    of LM , e its area , and E the point of contact. Let EL = l, EM = m ,
    716 THEORY OF PLANE CURVES.

    EK = k , l + m = a = c + c' ; 0, the inclination of LM . Then, integrating


    in every case from 0 to 27 ,

    : Sr8 = 1 - ) :: 15(1 –mº)do = S(1 –m )do= 1 – m .


    3 [mºd0 = -e ) Also , S (2+ m ) do = wa”,
    :: a fio = ta’ + 1 – 4 ...... (i.). Similarly o ſio= rc* + 1 - K ...... (ii.),
    the last being obtained from S (7 — k?) do = . - K. K is then found by
    eliminating the integral between (i.) and (ii.).
    (5245.) If the curves 1 , u coincide, 1 = u and therefore i - * = acc'.

    TRAJECTORIES.
    5246 DEF. - A trajectory is a curve which cuts according to
    a given law a system of curves obtained by varying a single
    parameter.
    The differential equation of the trajectory which cuts at a
    constant angle ß the system of curves represented by
    $ (x , y , c ) = 0 is obtained by eliminating c between the
    equations
    ou
    $ (x,y,c) = 0 and tan B = 9 , t- $ #, ye'
    the derivatives of ¢ being partial, and Y : referring to the
    trajectory.*
    PROOF. — At a point of intersection we have for the given curve
    m = - 97 • Py, and for the trajectory m ' = yz. Employ (4070) .

    If the trajectory is to be orthogonal, tan ß = 00 , and the


    second equation becomes
    0 , - xyz = 0 .
    Ex. — To find the curve which cuts at a constant angle all right lines
    passing through the origin .
    Let y = cx represent these lines by varying c ; then , writing n for tanß,
    the two equations become y - cx = 0 and n ( 1 + cyz) = Y . - c. Eliminating
    c, xyr - Y = n (YYx + x ) . Divide by x2 + y' and integrate ; thus
    tan - Y = n log ✓ (x2 + y ) + 0 ,

    which is equivalent to r = aeñ,the equation of the logarithmic spiral (5289).


    * For a very full investigation of this problem , see Euler, Novi Com . Petrop., Vol. xIv.,
    p . 46 , XVII., p . 205 ; and Nova Acta Petrop., Vol. 1., p . 3 .
    CAUSTICS, 8c. 717

    CURVES OF PURSUIT.
    5247 DEF. — A curve of pursuit is the locus of a point which
    moves with uniform velocity towards another point while the
    latter describes a known curve also with uniform velocity.
    Let f ( , y ) = ( be the known curve, xy the moving point
    upon it , èn the pursuing point, and n : 1 the ratio of their
    velocities. The differential equation of the path of En is
    obtained by eliminating æ and y between the equations
    f(x, y ) = 0 ......... (i.), y - n = ne(x -— $)... .....(ii.),
    V (x + y ) = n ✓ (1 + n ) ..... ......... (iii.).
    Proof. — (ii.) expresses the fact that ay is always in the tangent of the
    path of En .
    (iii.) follows from 1 : n = V (dz + dn ) : „ (dx? + dy') ; the elements of
    arc described being proportional to the velocities.

    Ex. — The simplest case , being the problem usually presented, is that in
    which the point xy moves in a right line. Let x = a be this line, and let
    the point &n start from the origin when the point ay is on the æ axis. The
    equations (i.), (ii.), (iii.) now become, since Xx = 0 ,
    x = a, y = n + nı(a E), y = ny (1 + n ).
    From the second yg = n2x (a - 5), therefore (a – ) N2x = n ✓ (1 + n?).
    Putting n = P , dp ndę
    ✓ (1 + p ) (a - )
    Integrating by (1928),we find
    log ( p + v1 + p ) = - n log (a - 5) + n log a ,
    so that p and & vanish together at the origin ;
    therefore V1+pe+ p = ( % )", and therefore V1+po- p = ( = )";
    therefore = {l )*-(02)"},
    therefore r = } {c"(n=7)** + 19* } + ,
    the equation of the required locus, the constant being taken so that Ž = n = 0
    together. If, however, n = 1, the integral is
    - 52 — 2aţ _ awa- E
    718 THEORY OF PLANE CURVES.

    CAUSTICS.
    5248 DEF. — If right lines radiating from a point be reflected
    from a given plane curve, the envelope of the reflected rays is
    called the caustic by reflexion of the curve.
    Let • (x , y) = 0 , 4 ( x , y ) = ( be the equations of the
    tangent and normal of the curve, and let hk be the radiant
    point; then the equation of the reflected ray will be
    $ (h,k) x (x, y) + y (h,k ) $ (x ,y) = 0,
    and the envelope obtained by varying the coordinates of the
    point of incidence, as explained in (5194 ), will be the caustic
    of the curve.

    Ex. — To find the caustic by reflexion of the circle x' + yº = p?, the radiant
    point being hk .
    Taking for the tangent and normal, as in (4140), a cos a + y sin a = 1 ,
    and x sin a - y cos a = 0 , the reflected ray is
    (h cos a tk sin a - » ) ( x sin a - y cos a )
    + (h sin a - k cos a ) ( cos a ty sin a - 1 ) = 0 .
    Reducing this to the form
    A cos 2a + B sin 2a + O sin a - D cos a = 0 ,
    and differentiating for a,
    - 2A sin 2a + 2B cos 2a + C cos a + D sin a = 0.
    The result of eliminating a is
    { 4 (h + k *) (x + + y*) — (a + h )? — (y + k )?} = 27 (kx - hy)? (z* + y - h k ),
    the envelope and caustic required.

    5249 Quetelet's Theorem . — The caustic of a curve is the


    evolute of the locus of the image of the radiant point with
    respect to the tangent of the curve.
    Thus, in the Fig . of (1178 ), if S be the radiant point, W is the image in
    the tangent at P . The locus of W is, in this case, a circle, and the evolute
    and caustic reduce to the single point S'.
    Since the distance of the image from the radiant point is
    twice the perpendicular on the tangent, it follows that the
    locus of the image will always be got by substituting 2r for r
    in the polar equation of the pedal,or 2 for r in the polar
    equation of the reciprocal of the given curve with respect to
    the radiant point and a circle of radius k .
    TRANSCENDENTAL AND OTHER CURVES. 719

    TRANSCENDENTAL AND OTHER CURVES:

    THE CYCLOID .* (Fig. 115)


    5250 DEF. - A cycloid is the roulette generated by a circle
    rolling upon a right line, the carried point being on the cir
    cumference . When the carried point is without the circum
    ference, the roulette is called a prolate cycloid ; and ,when it is
    within , a curtate cycloid .
    5251 The equations of the cycloid are
    x = a ( @ + sin 8), y = a (1 - сos 6),
    where 0 is the angle rolled through, and a the radius of the
    generating circle .
    PROOF. — (Fig . 115 .) Let the circle KPT roll upon the line DE , the
    point P meeting the line at D and again at E . Arc KP = KD ; therefore
    arc PT = AK = OT. Also 0 = PCT, the angle rolled through from A , the
    centre of the base ED. Then
    x = 0T + TN = a0 + a sin 0 ; y = PN = a - a cos 0.

    5253 If s be the arc OP and p the radius of curvature at P ,


    s = 2PT = ✓ (Bay), p = 2PK .
    PROOF. — (i.) The element Pp = Bh = 2 (OB - 00) ultimately ; therefore,
    by summation, s = 20B . Also OB = PT = V ( TK . TR ) = v ( ay) .
    (ii.) Let two consecutive normals at P and p intersect in L . Then PL,
    pl are parallel to BA , BA ; therefore PLp is similar to Bai. But Pp = 2Bi ;
    therefore p or PL = 2BA = 2PK .
    5255 CoR . — The locus of L , that is the evolute of the
    cycloid , consists of two half -cycloids as shown in the diagram .

    5256 The area of a cycloid is equal to three times the area


    of the generating circle, and the curye length is four times
    the diameter of the same circle.
    Proof. — (i.) Area PpN = PprR = BbqQ ultimately. Therefore, by
    summation, DE . A0 - cycloid = ma ? But DÊ .A0 = 2a. 2a = 4ma” ; there
    fore cycloid = 37a .
    (ii.) Total curve length = 8a , by (5253).
    * The earliest notice of this curve is to be found in a MSS. by Cardinal de Cusa, 1464
    See Leibnitz, Opera, Vol. m ., p . 96.
    720 THEORY OF PLANE CURVES.

    5257 The intrinsic equation of the cycloid is


    s = 4a sin y .
    Proof : 8 = 2PT = 4a sin PKT, and PKT = PTN = 4 *

    THE COMPANION TO THE CYCLOID .


    5258 This curve is the locus of the point R in Fig . (115).
    Its equation is
    y = a (1 - cos ).
    Proof. — From æ = að and y = a (1 - сos 6).
    5259 The locus of S , the intersection of the tangents at P
    and B , is the involute of the circle ABO.
    Proof : BS = BP = arc OB.

    PROLATE AND CURTATE CYCLOIDS. (5250)


    5260 The equations in every case are
    x = a (8 + m sin 0 ) y = a ( 1 - m cos 6 ).
    The cycloid is prolate when m is > 1 (Fig. 116 ), and curtate
    when m is < 1 (Fig . 117), m being the ratio of CP to the
    radius a .

    EPITROCHOIDS AND HYPOTROCHOIDS. (Fig. 118)


    5262 These curves are the roulettes formed by a circle
    rolling upon the convex or concave circumference respectively
    of a fixed circle, and carrying a generating point either within
    or without the rolling circle .
    The equations of the epitrochoid are

    5263 x = (a + b) cos 6 —mbcosao ,


    5264 y = (a +b) sin 6 - mb sin "700,
    * For other properties, see Pascal, Histoire de la Roulette ; Carlo Dati, History of the
    Cycloid ; Wallis , Traité de Cycloide ; Groningius, Historia Cycloidis, Bibliotheca Univ . ; and
    Lalouère, Geometria promota in septem de Cycloide libris ; Bernoulli, Op ., Vol. 1
    Euler, Comm . Pet., 1766 ; and Legendre, Exercice du Calcul. Int., Tom . II, p . 491.
    TRANSCENDENTAL AND OTHER CURVES. 721

    where a , b are the radii of the fixed and rolling circle


    ( Fig . 118 ) , 0 is the angle OCX , Q is the generating point
    initially in contact with the x axis, and m is the ratio OQ : b .
    The dotted line shows the curve described . For the hypo
    trochoid change the sign of b .
    Proof: x = CN + MQ ; CN = (a + b ) cos 0 ;
    MQ = OQ cos OQM = - 0Q cos (0 + 0), where o = POR and bo = a0.
    5265 The length of the arc of an epitrochoid is
    $ = (a + b) }{ 1+ mº— 2m cos 48}" ,
    Sa.
    which is expressed as an elliptic integral E (li, p) by substi
    tuting a0 = 269 .
    For the arc of a hypotrochoid, change the sign of b.
    Proof : 8 = Ss,do = S V (x* + y:) d0 (5113). Find x, and y, from (5263 –4).

    EPICYCLOIDS AND HYPOCYCLOIDS. ( Fig. 118 )


    5266 For the equations of these curves make m = 1 , in
    (5263, ' 4 ) . P is then the generating point, and the curve is
    shown by a solid line in Figure (118).
    5267 If 4 be the inclination of the tangent at a point P on
    any of these curves ,
    cos 6 — m cosa bo a + 26
    tany =
    sin 0 – m sin a + b = ta n 7 0 , if m = 1.

    5268 Hence, in the epicycloid, y = " # 260,


    and the equation of the tangent is
    a sin + 266 – y cos a 7,258 = (a + 26) sin 0.
    5269 The equation of the normal will be
    x cos" +,206 + y sin " + 260 = a cosmo.
    * Prof. Wolstenholme has investigated these curves considered as the envelopes of a
    chord whose extremities move on a fixed circle with uniform velocities in the ratio m : n or
    in : ( - n ). - Proc. Lond . Math . Soc ., Vol. iv ., p . 321.
    4 z
    722 THEORY OF PLANE CURVES.

    5270 The length of the arc of an epicycloid or hypocycloid


    included between two successive cusps is

    8b(a + b), and the included area is "b*(3a + 26).


    ea

    Proof. — Putting m = 1 into (5265) and að = bø, the length becomes


    25 (a + b) * sin dø = $ (a + b).
    Otherwise by (5234) ; the pedal being the cardioid whose perimeter = 8a
    (5333) .
    (ii.) The area, by (5232),is +6 + * 48° sinº (1 + 5)dø; since, in
    Fig. (118), dy of (5232) = dPOR = dø and r = PR = 26 sin ,

    5271 The evolute of an epicycloid is a similar epicycloid .


    PROOF. — The equation of the tangent referred to an x axis drawn throngh
    the summit of the curve will be (by turning axes through an angle b # + a ),
    x cosa todos 0+ ysinat 2.0 = (a +26)cos 0.
    Comparing this with (5270) , which is the equation of the tangent of the
    evolute , we see that the epicycloid and its evolute are similar curves having
    their parameters in the ratio a + 2b : a ; and that the radius drawn through
    a cusp of either of the curves passes through a summit of the other.
    5272 When b = - ja , the hypocycloid becomes a straight
    line, namely , a diameter of the fixed circle .

    THE CATENARY. (Fig . 119)


    5273 Characteristic. — The perpendicular TP from the foot
    of the ordinate upon the tangent is of a constant length c,and
    therefore equal to 0A, the perpendicular from the origin on
    the tangent at the vertex. c is the parameter of the curve.
    The equation is
    ol

    5274 y = (ešte s). 2

    PROOF : tan PCT =_ da


    " =_
    Pie . . x = c { log (y + Vys - c ) - log c }
    ( 1928) , since x = 0 when y = c. Therefore
    e+= + {y+ V (79—()} therefore e = 1 {y - v (32–cº)}.
    TRANSCENDENTAL AND OTHER CURVES. 723

    5275 If 8 = arc AC, s = (eě - e 8) = CP .


    Proof: s = Sv (1 + y )dx (5197)
    = SV (1+y**.*)dx = 5%dx = g(e8 - 6 ) = 1/(49–cº) = CP.
    5276 The area OACT = cs. (5205)

    5277 The radius of curvature at C = } , and is therefore


    equal to the tangent intercepted by the axis of x.
    Proof: cosy = . .- -sinyko = - Yo,:. = s, = (5146).
    5278 The catenary derives its name from a chain , which ,
    when suspended from its extremities, takes the form of this
    curve.
    For the equation of the evolute of the catenary,see (5159).

    THE TRACTRIX . (Fig . 119 )


    52179 Characteristic . — The length of the tangent intercepted
    by the x axis is constant. This curve is the involute of the
    catenary , being the locus of P in Figure (119) .
    The equation of the tractrix is
    5280 x = c log {c+ (c yº)} - c logy - (0 - y ).
    PROOF. — Let the tangent PT = C, then the differential equation of the
    carve is therefore yx , = - 1o - y . Substitute z = V02 - y ’, and integrate
    by (1937) .
    5281 The area included by the four branches = mcº.
    Proof.–Area = 4ſydx = -46V8 –1jdy= ne",by (1933).
    THE SYNTRACTRIX .

    5282 This curve is the locus of a point Q on the tangent of


    the tractrix in Fig . (119) . Let QT be equal to a given con
    stant length d ; then the equation of the syntractrix will be
    5283 x = c log {d + v (d — y?) } - c log y - ✓ ( — y ).
    724 THEORY OF PLANE CURVES.

    THE LOGARITHMIC CURVE.* (Fig. 120)


    5284 Characteristic. — The subtangent is constant.
    The equation of the curve is either

    5285 y = aer , or x = n log


    where n = NT, the constant subtangent, and a is the intercept
    on the y axis .
    5287 If n be an even integer, y may take negative values.
    The most general form of the equation may perhaps be
    assumed to be
    2

    y = ei(cos21" + isin 277) +


    29

    THE EQUIANGULAR SPIRAL . (Fig . 121)


    5288 Characteristic. - The angle OPS between the tangent
    and radius is constant. The equation of the curve is either

    5289 r = aer or 0 = n log .


    5291 tan p = n , s = r sec 6 ,
    measuring s from the pole.
    Proof. — By (5112) and (5200) .
    5293 Hence the length of the spiralmeasured from the pole
    0 to a point P (Fig . 121) is equal to PS, the intercept on the
    tangent made by the polar subtangent OS.
    5294 The locus of S is a similar spiral, and is also an invo .
    lute of the original curve.
    5295 The pedal curve, which is the locus of Y, is also a
    similar equiangular spiral.
    Proof. — The constancy of the angle o makes the figure OPYS always
    similar to itself. Therefore P , Y, and s describe similar curves. Hence, if
    ST is the tangent to the locus of S, OST = = OPS ; therefore PST is a
    right angle ; therefore the locus of S is an involute of the original spiral. I
    * Originated by James Gregory, Geometrie Pars Universalis, 1668 .
    + See Euler , Anal. Infin ., Vol. 11., p . 290 ; Vincent, Ann. de Gergonne, Vol. xv., p . 1 ;
    Gregory, Camb. Math . Journal, Vol. 1., pp. 231, 264 ; Salmon, Higher Plane Curves, p . 274 .
    I For additional properties, see Bernoulli, Opera, p . 497 .
    TRANSCENDENTAL AND OTHER CURVES. 725

    THE SPIRAL OF ARCHIMEDES.* (Fig. 122 )


    5296 Characteristic. — The distance from the pole is propor
    tional to the angle described . Hence the equation is
    5297 r = al. Also tan p = 0. By (5112).
    5299 The intercept, PQ , on any radius between two succes
    sive convolutions of the spiral, is constant and = 2an .
    5300 The area swept over by any radius is one third of the
    corresponding circular sector of that radius.
    5301 This curve is one of the class the general equation of
    which is
    r = að , with tan p = %

    THE HYPERBOLIC OR RECIPROCAL SPIRAL. (Fig. 123)


    a

    5302 The equation is r = a .


    5303 An asymptote is the line y = a . (5171)

    5304 The spiral is also an asymptote to itself.


    For when the radius is of the first order of smallness , the distance
    between two successive convolutions is of the second order. Hence the
    distance to the pole measured along the curve is infinite .

    The area between the radiants ru, r, is = ża (rı - rə).


    5305 The equation of the Lituus is r = "

    THE INVOLUTE OF THE CIRCLE. (Fig. 124)


    5306 The equation is
    V ( – aº) = a ( + cos-1m ).
    PROOF : % = OPY = cos -1 ✓ ( po2 — aʼ) = BP = arcAB = a (0 + ).

    5307 The pedal of the involute is the spiral of Archimedes.

    * Invented by Conon , B . c. 250.


    726 THEORY OF PLANE CURVES.

    Proof . — Let r', o be the coordinates of Y on the pedal carve. Then


    g' = BP = arc AB = a ( 0' + ). (See 5297).
    5308 The reciprocal of the involute is the hyperbolic spiral.
    Proof. — (Fig. 124.) Let P' on OY correspond to P , and let r', o' be the
    polar coordinates of P'. Then r' = OP'
    But OY = BP = arc AB = a (B'+ ža), :; r =am See (5302).

    THE CISSOID.* (Fig . 125)


    5309 Characteristic . — A line drawn from the end , 0 , of
    a fixed diameter of a circle to the end , Q , of any perpendicular
    ordinate intersects the parallel ordinate equidistant from the
    centre in a point, P , whose locus is the cissoid. The equation
    of the curve is
    5310 y ? (2a - x ) = 218 and dy – (Ga — 2x ) Vx
    dx = 2 / (2a — )* *
    Proof. — By similar triangles, y : x = V (2ax - x *) : 2a — x . Two mean
    proportionals between the radius a and CS are given by the carve, for it
    appears that a ’ : CT2 :: CT : CS, and therefore a : CT : VCS. CT : CS.
    5311 The tangent of the circle at B , the other end of the
    diameter, is an asymptote to both branches of the cissoid.
    5312 The area between the curve and its asymptote is equal
    to three times the area of the circle.
    za
    Proof : In ydx substitute x = 2a sin ’ 0 .

    THE CASSINIAN OR OVAL OF CASSINI. (Fig. 126 )


    5313 Characteristic. -- The product PA . PB of the distances
    of any point on the curve from two fixed points A , B is con
    stant ; the equation is consequently
    : {yº + (a + ) } {yº + (a — a )2} = m *
    or (x2 + y² + a ?)2 — 4aʼx ? = m *,
    where 2a = AB. The equation in polar coordinates is
    704 - 2aⓇp2 cos 20 + a * - m * = 0 .
    * Diocles, A .D. 500.
    TRANSCENDENTAL AND OTHER CURVES. 727

    5314 If a be > m , there are two ovals , as shown in the


    figure. In that case, the last equation shows that if OPP'
    meets the curve in P and P ', we have OP.OP = v ( a * — m *) ;
    and therefore the curve is its own inverse with respect to a
    circle of radius = v (a * - m *).

    5315 O being the centre, the normal PG makes the same


    angle with PB that OP does with PA .
    PROOF. – From (r + dr) (r ' — dr') = m and rp = m % ; therefore rdr' = r'dr
    or 9 : r ' = dr : dr' = sin 0 : sin 6', if 0, 0' be the angles between the normal
    and r, r'. But OP divides APB in a similar way in reverse order.
    5316 Let OP = R , then the normal PG , and the radius of
    curvature at P , are respectively equal to
    mºR 2m ’ RS
    R +a and 3R * + at + m ***

    THE LEMNISCATE . + (Fig. 126 )


    5317 Characteristic. — This curve is what a Cassinian be
    comes when m = a . The above equations then reduce to
    (x2 + y?) = 2a* (wº — y ) and po ? = 2a cos 20.
    5318 The lemniscate is the pedal of the rectangular hyper
    bola , the centre being the pole .
    area
    5319 The area of each loop = a”. (5206 )

    THE CONCHOID . I (Fig . 127)


    5320 Characteristic . - If a radiant from a fixed point o in
    tersects a fixed right line, the directrix , in R , and a constant
    length , RP = b , be measured in either direction along the
    radiant, the locus of P is a conchoid . If OB = a , be the per
    pendicular from 0 upon the directrix, the equation of the
    curve with B for the origin or 0 for the pole is
    5321 X *yº = (a + y ) (62 - y ) or r = a sec 0 + 6 .
    * B . Williamson, M . A ., Educ. Times Math., Vol. xxv ., p . 81.
    + Bernoulli, Opera , p . 609 .
    I Nicomedes, about A. D . 100 .
    728 THEORY OF PLANE CURVES.

    5323 When a < b , there is a loop ; when a = b , a cusp ;


    and when a b , there are two points of inflexion .

    5324 To draw the normal at any point of the curve, erect


    perpendiculars, at R to the directrix , and at 0 to OP. They
    will meet in S the instantaneous centre, and SP will be the
    normal at P (5242) .
    5325 To trisect a given angle BON by means of this curve,
    make AB = 20N , and draw the conchoid , thus determining
    Q ; then AON = 3A0Q .
    PROOF. — Bisect QT in S ; QT = AB = 20N , therefore SN = SQ = ON ;
    therefore NOS = NSO = 2NQ0 = 2AOQ .

    5326 The total area of the conchoid between two radiants each making
    an angle 0 with OA is
    a ’ tan 0 + 260 + 3a v ( - a ”) or a’ tan 0 + 26²0,
    according as b is or is not > a .
    The area above the directrix ) - + 620.
    between the same radiants
    The area of the loop which exists when bis > a is
    0 -1 a 21l at ( - as) 1 , 101.2 _ )
    a - ✓ ( –a ) '

    THE LIMAÇON.* (Fig. 128 )


    53217 Characteristic. - As in the conchoid , if , instead of the
    fixed line for directrix , we take a fixed circle upon OB as
    diameter. This curve is also the inverse of a conic with
    respect to the focus. The equation , with OB for the initial
    line and axis of æ is
    5328 r = a cos 0 + b or (x2 + y — ax) = 6? ( x2 + yº),
    where a = OB, b = PQ .
    5330 With 6 > a , 0 is a conjugate point.
    With b < a , O is a node. [ For m = a , see (5332).
    5331 The area = a (ja? + 62).
    When a = 2b, the limaçon has been called the trisectric .
    * Blaise Pascal, 1643.
    TRANSCENDENTAL AND OTHER CURVES. 729

    THE VERSIERA.* (Fig . 130)


    (Or Witch of Agnesi.)
    5335 Characteristic . - If upon a diameter OA of a circle as
    base a rectangle of variable altitude be drawn whose diagonal
    cuts the circle in B , the locus of P , the point in which the
    perpendicular from B meets the side parallel to 0A, is the
    curve in question . Its equation is
    5336 xy = 2a / (2ax —— xº),
    where a = 0C the radius.
    5337 There are points of inflexion where x = la .
    The total area is four times the area of the circle .

    THE QUADRATRIX . + (Fig . 131)


    5338 Characteristic . — The curve is the locus of the inter
    section , P , of the radius OD and the ordinate QN , when these
    move uniformly , so that æ : a :: 0 : , where X = ON ,
    a = 0A , and 0 = BOD . The equation is
    a - X
    y = x tan
    a 2)
    5339 The curve effects the quadrature of the circle , for
    OC : OB :: OB : arc ADB .
    Proof : 00 : 0B :: CP : BD. But CP = x in the limit when it is small,
    therefore CP : BD : : a : ADB.

    5340 The area enclosed above the æ axis = 4a’o -1 log 2 .


    Proof.-— In the integral « tan (*. * )dx substitute - (a — ~) = 2ay,
    and integrate ſy tan y dy by parts, using (1940). The integrated terms
    produce log cost - log cost at the limit , which vanishes though of
    the form 00 – 00 . The remaining integral is log cos y dy, and will be found
    at ( 2635 ) .

    THE CARTESIAN OVAL . (Fig . 134)


    5341 Characteristic . — The sum or difference of certain fixed
    multiples of the distances of a point P on the curve from two
    * Donna Maria Agnesi, Instituzioni Analitiche, 1748, Art. 238. † Dinostratus, 370 B.C.
    5 A
    730 THEORY OF PLANE CURVES.

    fixed points A , B , called the foci, is constant. The equations


    of the inner and outer ovals are respectively
    5342 mri + lr, = ncz, mri - Ir, = ncs,
    where ri = AP, r. = BP, cz = AB, and n > m > l.
    5343 To draw thecurve, put --т = "u and nos
    m
    = -a ; therefore r, ur, = a,
    where a is > AB and u < 1 ( 1 ) . Describe the circle centre A , and radius
    AR = a . Draw any radiant AQ, and let P , Q be the points in which it cuts
    the ovals, then, by. (1) ,
    5344 PR = uPB and QR = MQB............ .. .... ... ... (2 ).
    Hence , by (932), we can draw the circle which will cut AR in the required
    points P , Q . Thus any number of points on the oval may be found .
    5345 By (2 ) and Eac. vi. 3, it follows that the chord RBr bisects the
    angle PBR .
    Draw Ap through r , and let PB, QB produced meet Ar in p and q . The
    triangles PBR , qВr are similar, therefore gr = uqB ; therefore q is on the
    inner oval. Similarly p is on the outer oval. By Euc, vi. B ., PB .QB
    = PR .QR + BR’ ; therefore, by ( 2 ), ( 1 - ") PB . QB = BR”. Combining
    this with PB : Bq = BR : Br, from similar triangles , we get

    5346 BQ . Bq = BR.
    1-
    Br _ ai
    .......................(3).
    1 - pa

    5347 Draw QC to make _ BQC = BAq; therefore, A , Q , C, 4


    being concyclic , we have, by (3 ),
    a' -
    BQ . Bq = AB.BC = 1 - u? "
    .....(4).
    Hence C can be found if a, u , and the points A , B are
    given . C is the third focus of the ovals, and the equation of
    either oval may be referred to any two of the three foci.
    Putting BC = 41, AC = C , AB = cg, the equation between l, m , n is
    obtained from (4 ) thus : czci (1 – u ?) = a* - cś; therefore cg (cs + C ) = a ' + uGcg.
    But cz + a = ca, a = nos, u = , and theresult is
    m

    5348 Pc + n 'cz = méc, or lBC + m ’CA + n 'AB = 0 ...(5 ),


    where CA = - AC.
    Putting ru, ru, rz for PA, PB, PC, the equations of the
    curves are as follows
    TRANSCENDENTAL AND OTHER CURVES. 731

    Inner Oval. Outer Oval.


    5349 mri + lry = ncg ... (6 ), mri - Ir, = nez ... (7),
    5351 nr. + lrg = mcg ... (8 ), nri - lrg = mc2 ... ( 9) ,
    5353 mrz - nr, = lc ... (10), nr, - mr3 = lc ...(11).
    That (6 ) and ( 7 ) are equations of the curve has been shown. To deduce
    the other four, we have 2 APB = AqB = ACQ (5347) ; therefore ACQ ,
    APB are similar triangles. But, by (6 ), MAP + IBP = nAB, therefore
    MAC + ICQ = nAQ or nAQ - 10Q = mÃO, which is equation ( 9 ). Again ,
    ABQ , APC are similar. But, by (7 ) , MAQ - 1BQ = nAB ; therefore
    MAC - ICP = NAP or nAP + ICP = mAC, which is equation (8 ) .
    Equations (10 ) and (11) are obtained by taking (6 ) from (8 ) and (7 )
    from (9 ), and employing (5 ).

    5355 AP . AQ = AB . AC = constant.
    PROOF. — Since A , Q , C , q are concyclic , ZQCA = QA = ABB ; there
    fore P, Q , C , B are concyclic ; therefore AP . AQ = AB . AC = constant (12 ).
    5356 CP.CP' = CA . CB = constant.
    Proof: ZPCB = PQB = Bpq = BCq. Hence, if CP meets the inner
    oval again in P ', CBq, CBP' are similar triangles. Again , because BPC
    = BQC = BA = BAP', the points A , B , P ', P are concyclic ; therefore
    CP. CP ' = CA .CB = constant. Q. E. D .
    ng ide,
    Hence , by maki P, P ' coinc we have the theo rem :
    535 7 The tang ent from the exte rnal focu s to a series of tri
    confocal Cartesians is of constant length , and = ✓ (CB .CA) .
    5358 To draw the tangents to the ovals at P and Q . De
    scribe the circle round PQCB, and produce BR to meet the
    circumference in T ; then TP, TQ are the normals at P
    and Q .
    The proof is obtained from the similar triangles TQR, TBQ , which show
    that sin TQA : sin TQB = l : m , by ( 2) , and from differentiating equation (7 ),
    which produces cm : 41 = 1:m.*
    dsds

    5359 The Semi-cubical parabola yº = ax is the evolute of a


    parabola (4549). The length of its arc measured from the
    origin is
    s = 2% (1+ zar) - 1}.
    * For the length of an arc of a Cartesian oval expressed by Elliptic Functions, see a paper
    by S . Roberts , M . A ., in Proc . Lond. Math . Soc., Vol. v ., p . 6 .
    732 THEORY OF PLANE CURVES.

    5360 The Folium of Descartes , 23– 3axy + y = 0 , has two


    infinite branches, and the asymptote x + y + a = 0 .
    For the lengths of arcs and for areas of conics, see (6015), et seq .

    LINKAGES AND LINKWORK .

    5400 A plane linkage, in its extended sense, consists of a


    series of triangles in the same plane connected by hinges, so
    as to have but one degree of freedom of motion ; that is, if
    any two points of the figure be fixed , and a third point be
    made to move in some path , every other point of the figure
    will, in general, also describe a definite path . With two points
    actually fixed, the linkage is commonly called a piece-work ,
    and if straight bars take the place of the triangles , it is called
    a link -work .

    THE FIVE-BAR LINKAGE.


    5401 Mr. Kempe's fundamental five-bar linkage is shown
    in Figure (135 ). A , B , D ' are fixed pivots indicated by small
    circles. C , D , B ', C ', in the same plane, are moveable pivots
    indicated by dots. The lengths of the bars AB, BC, CÒ , DA
    are denoted by a , b , c , d . The lengths of AB', B ' C ', C ' D ', D ' A
    are proportional to the former, and are equal to ka, kb, kc, kd ,
    respectively. Hence ABCD , AB' C ' D ' are similar quadri
    laterals, and LAD' C ' = ADC. P being any assigned point
    on BC and BP = ) , P must be taken on D 'C " so that
    D'P' = 14
    ab Draw PN, P'N ' perpendiculars to AB. Then ,
    throughout the motion of the linkage in one plane, NN' is a
    constant length .
    Proof: NN' = BD’ – (BN + N ' D'). But BD’ = a - kd, and
    BN+ N'D’= 1 cos B-1cm cos D =
    с CO
    (2abcosB – 2cd cosD)
    z (a + 63 - - d ) (702). Hence
    LINKAGES AND LINKWORK . 733

    5402 NN' = a - kd - (a +6°– 6 – d”).


    5403 Case I. — (Fig. 136.) If ca? = (+?a 12- kd) ab
    _ 2 _ ,729 then
    NN' = BP ; consequently , if the bars PO = BB and P0 =
    P' D' be added , the point 0 will move in the line AB.
    10

    If, in this case, d = ka and b = c , then = b and P coin


    cides with C , P ' with C', and B ' with D , O as before moving
    in the line AB .

    5404 CASE II. — (Fig. 137.) If, in Case I., kd = a and


    a ’ + 62 = c+ + d ?, 1 is indeterminate ; that is , P may then be
    taken anywhere on BC . D ' coincides with B , and NN ' = 0 .
    PP is now always perpendicular to AB. If the bars PO ,
    P 'O be added , of lengths such that PO? — P ' O ? = PB — P ' B ?,
    O will move in the line AB. If, on the other side of PP', bars
    PO = PB and P ' O ' = PB be attached , then I will move in a
    perpendicular to AB through B .

    5405 CASE III. — (Fig . 138.) If, in Case I., kd = a , b = d ,


    and c = - a , the figure ABCD is termed a contra-parallelogram .
    BP = , is indeterminate, BO' = kc = - " and BP' = - .
    Hence BC" and BP' are measured in a reversed direction ;
    PP is always perpendicular to AB, and if any two equal bars
    PO, P 'O are added, 0 will move in the line ÅB .
    5406 If three or more similar contra-parallelograms be
    added to the linkage in this way, as in Figure (139) , having
    the common pivot B and the bars BA, BC , BE, BG in geo
    metrical progression ; then , if the bars BA , BG are set to any
    angle , the other bars will divide that angle into three or more
    equal parts .
    5407 If, in Figure (138 ), AD be fixed and DC describe an
    angle ADC, then B ' C ' describes an equal angle in the opposite
    direction . Mr. Kempe termssuch an arrangement a reversor,
    and the linkage in Figure (139) a multiplicator. With the aid
    TTY

    ( CRIV
    734 THEORY OF PLANE CURVES.

    of these, and with a translator (Fig . 140 ), for moving a bar


    AB anywhere parallel to itself, he shows that any plane curve
    of the nth degree may, theoretically , be constructed by link
    work .*

    5408 CASE IV . - (Fig . 141.) If, in the original linkage


    ( Fig . 135) kd = a , D ' coincides with B . Then , if the bars
    RPO, RP'd be added by pivots at P , P ', and R ; and if
    OP = PR = BP and OP = P ' R = BP ; the points 0 , 0
    will move in perpendiculars to AB. For by projecting the
    equal lines upon AB, we get NL = BN ' and BN = N ' I ' ,
    therefore BL = BI' = NN ' = a constant, by (5402).

    5409 CASE V . - (Fig . 142.) Make ka = d and 1 = b . Then


    B ' coincides with D , P with C , and P with C . Replace D ' C ' ,
    C ' D by the bars DK , KD' equal and parallel to the former.
    Also add the bars CO = DK and OK = CD . Draw the per
    pendiculars from 0 , C and Cº to AB. Then by projection ,
    NL = N ' D ' ; therefore BL = BN + NL = BN + N ' D ' = BD
    - NN ' = constant. Hence the point 0 will move perpendi
    cularly to AB.
    5410 CASE VI. — ( Fig . 143.) In the last case take k = 1.
    Therefore d = a , D ' coincides with B , BK = BC , and CDKO
    is a rhombus. This is Peaucellier 's linkage.

    5411 CASE VII. — (Fig. 144.) In the fundamental linkage


    ( Fig . 135 ), transfer the fixed pivots from A , B to P , S , adding
    the bar SA , so that PBSA shall be a parallelogram . Then ,
    since NN' is constant (5 -402), the point P ' will move perpen
    dicularly to the fixed line PS.
    5412 Join AC cutting PS in U , and draw UV parallel to AD .
    Then UV : AD = PU : AB = CP : CB = constant ; there
    fore PU and UV are constant lengths. Hence it follows that
    the parallelism of AB to itself may be secured by a fixed pivot
    at Ū and a bar UV instead of the pivot S and bar SA .
    5413 In Case VII. (Fig. 144), with fixed pivots P and s
    * Proc. of the Lond. Math . Soc., Vol. vir., p. 213 .
    LINKAGES AND LINKWORK. . 735

    and bar SA, make b = a , d = c, ka = d , 1 = b . Then B ' coin


    cides with D , N ' with N , P with C and L , and P ' with C' ;
    and we have Figure 145 . DC , DC are equal, and they are
    equally inclined to AB or CS ; because, in similar quadri
    laterals, it is obvious that AB and CD and the homologous
    sides DC and AD include equal angles. Therefore CC is
    perpendicular to CS, and C moves in that perpendicular only.

    5414 If two equal linkages like that in (5413 ), Figure (145),


    but with the bars AS, CS removed , be joined at D (Fig. 146 )
    and constructed so that CD ) , yDC' form two rigid bars, then
    AB, aß will always be in one straight line. Let A , B be made
    fixed pivots , then , while C describes a circle , the motion of
    the bar aß will be that of a carpenter's plane.
    5415 On the other hand , if the linkage of Figure (145 ), with
    AS and CS removed as before, be united to a similar inverted
    linkage (Fig . 147), with DC , DC common , then , with fixed
    pivots A , B , D , the motion of the bar aß will be that of a lift ,
    directly to and from AB.
    5416 The crossing of the links may be obviated by the
    arrangement in Figure (148). Here the bars C'B , C' D , C' D '
    are removed , and the bars FD , FE, FG added in parallel ruler
    fashion .

    5417 CASE VIII. — (Fig . 149.) In Case VII., substitute the


    pivot U and the bar UV for S and SA . Make d = a , and
    therefore k = 1. Then b' = b and c' = 0 , making BCDC a
    contra- parallelogram ; D ' coincides with B , and B' with D .
    The bars AB, AD are now superfluous. Take BP = d ; then
    BP = 1 ; ;therefore PP is parallel to CC', therefore to BD ,
    therefore to PV (5412) ; therefore V , P , P are always in one
    right line. P ', as in Case VII., moves perpendicularly to PU
    and AB . This arrangement is Hart' s five-bar linkage.

    5418 When a point P (Fig . 152) moves in a right line PS,


    it is easy to connect to P a linkage which will make another
    point move in any other given line we please in the same
    736 THEORY OF PLANE CURVES.

    plane. Let QR be such a line cutting PS in Q . Make Q a


    fixed pivot , and let OQ, OP, OR be equal bars on a free pivot
    0 . Then , if the angle POR be kept constant by the tie-bar
    PR, PQR , being one half of POR (Euc. III. 21), will also be
    constant, and therefore,while P describes one line, R describes
    the other.
    If the bar PO carries a plane along with it , every point in
    that plane on the circumference of the circle PQR will move
    in a right line passing through Q .

    THE SIX - BAR INVERTOR .*


    5419 If in the linkwork (5410, Fig . 143) the bar AD be
    removed , and D be made to describe any curve, 0 will describe
    the inverse curve , just as, when D described a circle , O moved
    in a right line which is the inverse of a circle.
    PROOF. — Let BOD and CK intersect in E . Then BO . OD = BE - OE
    = BOP — 002 = a constant called the modulus of the cell.

    THE EIGHT-BAR DOUBLE INVERTOR .


    5420 Two jointed rhombi (Fig. 150) having a common
    diameter AB form a double Peaucellier cell termed positive
    or negative according as Por Q is made the fulcrum . We
    have PQ . PR = PQ .QS = AP2 – AQ', the constant modulus
    of the cell.

    THE FOUR-BAR DOUBLE INVERTOR.


    5421 If, on the bars of a contra - parallelogram ABCD
    (Fig. 151) four points p, q, r , s be taken in a line parallel to
    AC or BD , then in every deformation of the linkage, the
    points p , q , r , s will lie in a right line parallel to AC ; and
    Papr = pq . qs = a constant modulus. Thus, if p be a ful
    crum and r describes a curve, q will describe the inverse
    curve. If q be the fulcrum , p will describe the inverse curve.
    PROOF. — Let Ap = mAB, therefore pq = mBD , and pr = (1 - m ) AC,
    therefore pq.pr = m ( 1 - m ) AC .BD = m ( 1 - m ) (AD — AB ) = constant.
    * Since the curve described is the inverse and not the polar reciprocal of the guiding curve,
    it seems better to call this linkage an invertor rather than a reciprocator .
    LINKAGES AND LINKWORK . 737

    THE QUADRUPLANE, OR VERSOR INVERTOR.


    5422 Let the bars of the contra-parallelogram invertor
    (5421, Fig . 151) carry planes, and let P , Q , R , S be points in
    the planes similarly situated with respect to the bars which
    contain p , q , r , s respectively , so that _ PAp = QAq and
    AP : Ap = AQ : Aq ; and similarly at 0 . Then , if P be the
    fulcrum and R traces a curve, Q will trace the inverse curve
    and the angle QPR will be constant.
    PROOF. — Let PA = nAB and PB = n 'AB , therefore, by similar triangles,
    PAQ, BAD, PQ = nBD. Also, by the triangles PBR, ÅBC, PR = n'A0 ;
    therefore PQ. PR = nn' AC . BD LA
    = PAB
    .FB ((AV
    AD AD
    - AB )),, aa constant.
    constant.
    Again , the inclination of PQ to BD = that of AP to AB , which is con
    stant. Similarly , by the triangles PBR, ABC , the inclination of PR to AC
    = that of BR to BO , which is also constant ; therefore QPR, the sum of
    these two inclinations, is a constant angle.

    THE PENTOGRAPH, OR PROPORTIONATOR.


    5423 Let ABCD (Fig . 153) be a jointed parallelogram , A, B
    fixed pivots, q a tracer placed at any assigned point in BO
    produced ; then a pencil at p will evidently reproduce any
    figure traced by q diminished in linear proportions in the ratio
    of Bq to BC.

    THE PLAGIOGRAPH , OR VERSOR PROPORTIONATOR .


    5424 In the same figure, make an angle qBQ = pDP,
    BQ = By, and DP = Dp , and let a tracer Q and pencil P be
    rigidly connected to the arms BC and DC. Then P will pro
    duce a similar reduced figure as before , but no longer similarly
    situated . It will be turned round through an angle QBq.
    This is Prof. Sylvester's Plagiograph .
    PROOF. — Let BC = k . Bq; therefore AD = kBQ, DP = kAB, and LABQ
    = PDA ; therefore (Euc. vi. 6 ) AP = KAQ. Also PAQ is a constant angle,
    for PAQ = BAD - BAQ - PAD = BAD - BAQ - BQA = BAD - ( + - ABQ )
    = BAD - * + ABO + QBq = QBq.

    THE ISOKLINOSTAT,* OR ANGLE-DIVIDER .


    5425 This linkage (Fig. 154) accomplishes the division of
    an angle into any desired number of equal parts. The dia
    * Invented and so named by Prof. Sylvester.
    5 B
    738 THEORY OF PLANE CURVES.

    gram shows the trisection of an angle by it. A number of


    equal bars are hinged together end to end, and also pivoted
    on their centres to the same number of equal bars which
    radiate, fan -like, from a common pivot. The alternate radial
    bars make equal angles with each other.
    The same thing is accomplished in a different way by
    Kempe's Multiplicator (5406, Fig . 139).

    A LINKAGE FOR DRAWING AN ELLIPSE.


    5426 In the arrangement of (5413, Fig . 145) the locus of
    any point P , on DC', excepting D and C ', is an ellipse.
    Proof. — Take CS, CC” for and y axes ; P the point xy ; SCD = 0,
    and therefore CDC ' = 20 ; PD = h . Then we have æ = (c - h ) cos0,
    y = (c + h ) sin 0, therefore 7
    (c - h )? ' (c + h )?z = 1 is the equation of the locus.
    Any point on a plane carried by DC also describes an ellipse round C ; bat
    if the point lies on a circle whose centre is D and radius DO, the ellipse be
    comes a right line passing through C , as appears from (5418) .

    A LINKAGE FOR DRAWING A LIMAÇON , AND ALSO A


    BICIRCULAR QUARTIC .*
    5427 ( Fig . 155 .) Let four bars AP', AQ ', BC , CD be
    pivoted at A , B , C , D , and let AB = BC = BQ = a ; AD
    = DC = DP ' = b . Take a fulcrum F on BC , a tracer at P ,
    and a follower at Q , so that PQ is parallel to BD. Let FP = p,
    FQ = r ; then, if P traces out a circle passing through F , Q
    will describe a limaçon .
    PROOF.— Let BQ = ma, therefore PD = mb ; r = 2m . BN , p = m + 1) . DN
    + ( 1 - m ) BN . Also BN ? — DN = a ' - 6". Eliminate BN and DN , and the
    equation between r and p is
    + (1 – m ) re - mpº = m ( m + 1 )^(a − b ) = k .
    If P describes the circle p = c cos 0, Q describes the locus
    p2 + (1 - m ) cr cos 0 - me cos 0 = km, .
    which is the inverse of a conic, that is, a limaçon (5327).
    If C be made the fulcrum , the equation reduces to pl - p* = 4 (a ’ — ).
    5428 With the same fulcrum F , drawing FH parallel to
    AC, if a tracer at H describes the circle, then a follower at K
    on CD will trace out a bicircular quartic.
    * W . Woolsey Johnson, Mess. of Math ., Vol. v., p. 159.
    LINKAGES AND LINKWORK. 739

    PROOF. — Draw FL, LK parallel to BA , AD . Let FH = P, FK = r,


    CK = B , CF = a = nFB, and therefore CL = np. Now
    2 ( a? + 3 ) = + pº + d
    Therefore, if I moves on the circle p = c cos 6, K will describe the curve
    Joll + nºc?r cos’0 – 2 (a ’+ ß")på+ ( a’ - B2) = 0,
    or (z?+ y?)2+ (n*c* — 20²— 232) 22— 2 (a’ + %) y’ + (a — 62)² = 0.

    A LINKAGE FOR SOLVING A CUBIC EQUATION.*


    5429 Let the three-bar linkwork (Fig . 156 ) have the bars
    AB, DC produced to cross each other. Let AB = AD = a ,
    BC = b , CD = c ; and let b and c be adjustable lengths.
    Suppose w — qx + p = 0 a given cubic equation .
    Make c = } )(2+ ), 6 = / 9- ); then deform the
    quadrilateral until EC = CD ; DE will then be equal to a real
    root of the cubic.
    PROOF : cos E = ac
    2 + c _ 7622 (x + a )? + 4c? — a
    20:2 4c (x + a )
    from which 208 — 2 (cʻ + b ) x + 2a (c* — 6²) = 0.
    Equate coefficients with the given cubic.t

    ON THREE -BAR MOTION IN A PLANE.


    5430 If a triangle ABC ( Fig. 157) be connected by the
    bars AO, BOʻ to the fulcra Ò , O , the locus of C is called a
    three-bar curve.
    0A, O ' B meet in Q , the instantaneous centre of rotation of
    the triangle , since QA ,QB are perpendicular to themovements
    of A and B respectively. Therefore CQ is the normal to the
    locus of C .
    5431 If a triangle similar to ABC be placed upon 00
    (homologous to AB), the circum -circle of the triangle will
    pass through the node, and the vertices of the triangle are
    called the foci of the curve .
    * M . Saint Loup, Comptes Rendus, 1874.
    + The foregoing account of linkages is taken chiefly from a paper by A . B . Kempe,
    F . R . S ., in the Proc. of the Royal Soc . for 1875 , Vol. xxiii. Other results by the same author
    will be found in the Proc . of the Lond . Math . Soc., Vol. ix ., p . 133 ; and by H . Hart, M . A .,
    ibid ., Vol. vi., p . 137, and Vol. VIII., p. 286. See also The Messenger of Mathematics, Vol. v.
    740 THEORY OF PLANE CURVES.

    Figures (158) and (159) exhibit different varieties of the


    curve according to the relative proportionsbetween the lengths
    of the bars.*

    MECHANICAL CALCULATORS.

    The Mechanical Integrator. f


    5450 This instrument computes not only the area of any
    closed plane curve, but the moment and also the moment of
    inertia of the area about a fixed line. The principle of its
    action is shown in Figure ( 160). OP is a bar carrying a
    tracer at P , and a roller A at some point of its length. The
    end O is constrained to move in the fixed line ON . When
    the tracer P moves round a closed curve, the length OP mul
    tiplied by the entire advance recorded by the roller is equal to
    the area of the curve.
    PROOF. — Let the motion of the tracer from P to a consecutive point Q be
    decomposed into PP' and P 'R parallel and perpendicular to ON . Let
    OP = a and PON = 0 . When the pointer moves from P to P ', the roll
    accomplished is PP ' sin 0 . The roll due to the motion from P ' to Q will be
    neutralized by the exactly equal and opposite roll in the motion of the pointer
    from q to p ', since the bar will there bave again the same inclination. Con
    sequently the product of the entire roll and the length a is equal to the sum
    of such termsas aPP' sin 0. But this is the area OPP' O ' = NPP' N '. The
    algebraic addition of such rectangles gives the entire area , and the instrn
    ment effects this, for the area SN is subtracted , by the motion of the roller,
    from the area QN which is added .

    5451 The instrument itself is shown in Figure (161). A


    frame moving parallel to OX by means of the guide BB
    carries two equal horizontal wheels geared to a central wheel
    which has two circumferences, such that its rate of angular
    motion is half that of the lower wheel and one third of that of
    the upper. The latter wheels carry two rollers, M and I, on
    horizontal axles ; and the middle wheel carries an arm OP, a
    pointer at P , and a roller A . In the initial position , the
    * The curve is a tricircular trinodal sextic, and is completely discussed by S . Roberts ,
    F . R . S ., and Prof. Cayley, in the Proc . of the Lond . Math . Soc., Vol. VII., pp . 14, 136 .
    + Invented and manufactured by Mr. J. Amsler-Laffon, of Schaffhausen. The demon
    strations (which in clearness and elegance cannot be surpassed) of the action of this instru
    ment, and of the Planimeter which follows, were communicated to the author by Mr. J.
    Macfarlane Gray, of the Board of Trade,
    MECHANICAL CALCULATORS. 741

    rollers A and I are parallel, while M is at right angles to A .


    The frame is thus supported above the paper on the three
    rollers ; and if the arm OP be moved through an angle AOA' ,
    the axles of the rollers M and I will describe twice and three
    times that angle respectively. Putting OP = a as above, and
    A , M , and I for the linear circumferential advances recorded
    by the three equal rollers respectively , we have the following
    results
    I. — The area traced out by the pointer P = Aa.
    II. — Themoment of the area about OX = M .
    III.— The moment of inertia about OX = (3A + I)
    PROOF. — I. Since O moves in the line0X, while the pointer P moves round
    a curve, the roller A will, as shown above, make the rolling Eh sin 0, where
    h = PP' in Figure (160 ) , and the area of the curve = axh sin 0 or a x roll.
    II. The moment of the area about OX

    = 3(an einox a sinº) = 24 –Eh cos 26).


    Now Sh vanisheswhen P returns to the starting point, and - Eh cos 20 is the
    roll recorded by M . For, when OP makes an angle 0 with OX , the axis of
    M willmake an angle - (90° + 20 ) with OX . In this position ,while P makes a
    parallel movement h , the roll produced thereby in M will be — h sin ( 90° + 20)
    = - h cos 20. There roll of M = moment of area.
    III. Lastly, the moment ofa inertia of the area about OX
    Ay ? sind A
    3 ) h sin 0 - h sin 30) .
    Now , when OP makes an angle 0 with OX, the axis of I makes – 30 ; there.
    fore – Eh sin 30 is the entire roll of I. Hence the moment of inertia

    = ex roll of A + 1 x rollof I.
    4

    The Planimeter. (Fig . 162)


    5452 This instrument * is a simpler form of area computer.
    O is a fixed pivot ; 0A, AP are two rods having a free pivot
    at A ; C is the roller, and P the pointer. The area of a closed
    curve traced by the pointer is equal to the total roll multiplied
    by the length AP .
    * Like The Integrator, the invention of Mr. Amsler .
    742 THEORY OF PLANE CURVES.

    PROOF. — Decompose the elementary motion PQ of the pointer into PP',


    effected with a constant radius OP, and P 'Q along the radius OP', and so all
    round the curve. The roll of C accomplished while P moves from P ' to Q
    will be neutralized by the equal contrary roll when P moves from q to p ' on
    the radius Op' = OP. Thus the total roll recorded will be the sum of the
    rolls due to the movements PP', QQ', & c .
    Draw OB perpendicular to AP, and, when P comes to R , let B ' be the
    altered position of B . The area PQSR = } (OP – ORP) w , where w = POQ.
    But OP = 0AP + PA — 2PA . AC — 2PA. BC (Euc. 11. 13) ; therefore , since
    BC is the only varying length on the right, we have PQSR = PA (BC - BC ) w .
    But BCw is the roll of C due to the angular motion w of the rigid frame OAP,
    and the subtraction of the area OSR from OPQ is effected by the instrument,
    since when the pointer moves from S to R the direction of the roll must be
    reversed. Hence the total area = PAX the total recorded roll.

    APPENDIX ON BIANGULAR COORDINATES.*

    5453 In the figure of (1178 ),the biangular coordinates of a


    point P are defined to be 0 = PSS and p = PS' S , or
    a = cot 8 and ß = cotp .
    5454 The equation of a right line YY' is
    aa + bB = 1 ,
    where a = cot SYS and b = cot SY 'S '.
    Proof. - Supplying the ordinate PN in the figure and denoting the angle
    S 'SY by y , the equation is obtained from CN cos y + PN siny = p the per
    pendicular on the tangent, SS ' sin y = YY' and SS cos i = SY - SY'.

    5455 coty = a b.
    5456 Equation of a line through C : a - B = const.
    5457 Equation of the line at infinity : a + B = 0.
    5458 Let SS' = C , then the distance between two points
    aßı, azß, is

    =¢ {(1,78.-278)+ (,40 - , )}.


    * Quarterly Journalof Mathematics, Vols. 9 and 13 ; W . Walton, M .A.
    APPENDIX ON BIANGULAR COORDINATES. 743

    5459 The equation of a line through the two points is

    5460 The length of the perpendicular from a'ß' upon the


    line da toß = 1 is
    P = _ c_ aa' + bB - 1
    a' +B✓{(a − b) + 1 }"
    5461 Cor.— The perpendiculars from the poles S, S are
    therefore
    bc
    SY =
    S {(a − b)2 + 1 }' S / {(a − b)? + 1}
    5463 When the point a'ß ' is on SS at a distance h from S ,
    ne (a − b ) h + bc
    v {(a − b)2 + 1}
    With two lines aa + bB = 1, a'a + b'ß = 1, the condition
    5464 of parallelism is a - b = a' — 6',
    5465 of perpendicularity (a − b)(a' — b') + 1 = 0 .
    5466 The equation of the line bisecting the angle between
    the same lines is
    aa + bB - 1 a 'a + 6B - 1
    ✓ {(a — b) + 1} " > {(a' — b')? + 1 }
    5467 The equation of the tangent at a point a'ß ' on the
    curve F (a , b ) = 0 is
    (a - a') Fc + (B - B ) Fg = 0 .
    5468 And the equation of the normal is
    - a- a B - B
    (a'B — 1) F$ + (1 + a'?) F = ( 'B - 1) F. + (1 + B2) Fş
    5469 The equation of a circle through S, S' is
    aß — 1 = m (a + B ),
    where m = cot SPS the angle of the segment.
    744 THEORY OF PLANE CURVES.

    5470 If C be the centre, the equation becomes O

    aß = 1.
    5471 And, in this case, the equations of the tangent and
    normal at a'ß ' are respectively

    + = 2 and a - B = a' - B .
    5472 The equation of the radical axis of two circles whose
    centres are S , S', and radii a , b, is
    (c' — aº + ) a = (cº + a? — 62) B .
    Proof. — By equating the tangents from af to the two circles, their
    lengths being respectively
    c" ( 1 + a ?)
    (a + B «?
    ) –a and *(1+ ) –b, by (5458).
    5473 To find the equation of the asymptotes of a curve
    when they exist ,
    Eliminate a and ß between the equations of the line at
    infinity a + B = 0,
    the curve F (a, b ) = 0 ,
    and the tangent (a - a') Fo + (B - B ) Fp = 0 .
    Ex. — The hyperbola a ' + BP = mºhas, for the equation of its asymptotes,
    a - B = + mv2.
    SOLID COORDINATE GEOMETRY.

    SYSTEMS OF COORDINATES.

    CARTESIAN OR THREE -PLANE COORDINATES.


    5501 The position of a point P in this system (Fig. 168) is
    determined by its distances, x = PA , y = PB, % = PC , from
    three fixed planes YOZ, ZOX , XOY, the distances being
    measured parallel to the mutual intersections OX , OY, OZ of
    the planes, which intersections constitute the axes of coordi
    nates. The point P is referred to as the point xyz, and in
    the drawing X , Y , % are all reckoned positive, ZoX being the
    plane of the paper and P being situated in front of it, to the
    right of YOZ and above XOY. If P be taken on the other
    side of any of these planes , its coordinate distance from that
    plane is reckoned negative.

    FOUR-PLANE COORDINATES.
    5502 In this system the position of a point is determined
    by four coordinates a , B , 7 , 8, which are its perpendicular
    distances from four fixed planes constituting a tetrahedron of
    reference. The system is in Solid Geometry precisely what
    trilinear coordinates are in Plane. The relation between the
    coordinates of a point corresponding to (4007) in trilinears is
    5503 Aa + BB + Cy + D8 = 3V,
    where A , B , C, D are the areas of the faces of the tetrahedron
    of reference, and V is its volume.

    TETRAHEDRAL COORDINATES.
    5504 In this system the coordinates of a point are the
    volumes of the pyramids of which the point is the vertex and
    50
    746 SOLID GEOMETRY.

    the faces of the tetrahedron of reference the bases : viz ., Aa,


    { BB , žcy, jD8. The relation between them is
    5505 a' + B + y' + 8 = V .

    POLAR COORDINATES.
    5506 Let O be the origin (Fig . 168), XOZ the plane of
    reference in rectangular coordinates, then the polar coordi
    nates of a point P are r , 0, 0 , such that r = 0P , 0 = _ POZ,
    and p = _ XOC between the planes of XOZ and POZ.

    THE RIGHT LINE.

    5507 The coordinates of the point dividing in a given ratio


    the distance between two given points are as in (4032), with a
    similar value for the third coordinate 6 .
    5508 The distance P , Q between the two points xyz, é'y'z' is
    PQ = v {(x — X )?+ (y - y) + (3 - ')?}. (Euc.I.47).
    5509 The same with oblique axes, the angles between the
    axes being 1, u , v.
    PQ = {(x — x )? + (y - 7 ) + (3 ^ 2 ) + 2 (y - y ) (z ') cosa
    + 2 (3 -7)(x — x ) cosu + 2(x — x')(y - y ) cos v}. (By702).
    5510 The same in polar coordinates,the given points being
    r0° , r'o'o',
    PQ = ~ [pP+ po ? — 2rr'{cos 6 cos ® + sin 8 sin 8 cos ($ - ')} ].
    Proof. — Let P , Q be the points, O the origin . Describe a sphere cutting
    OP, OQ in B , C and the z axis in A ; then, by (702), PQ = 0P + 00
    - 20P . OQ cos POQ and cos POQ, or cos a in the spherical triangle ABC, is
    given by formala (882), since b = 0, c = 0', and A = 0 - 0 .

    DIRECTION RATIOS.
    5511 Through any point Q on a right line QP (Fig. 169),
    draw lines QL, QM , QN parallel to the axes, and through any
    other point P on the line draw planes parallel to the coordi
    THE RIGHT LINE. 747

    nate planes cutting the lines just drawn in L , M , N ; then the


    direction ratios of the line OP are
    5512 QP : m = Q
    ÖMD, N
    nn = Q :
    D : T OP

    The angles PQL, PQM , PQN are denoted by a , b , y ; and the


    angles YOZ, ZOX, XOY between the axes by 1, u , v .
    5513 When 1 , u , v are right angles , the axes are called
    rectangular, and the direction - ratios are called direction
    cosines , being in that case severally equal to cos a , cos ß ,
    COS y .
    5514 When L , M , N are the direction -ratios (or numbers
    proportional to them ) of a line which passes through a point
    abc, the line may be referred to as the line ( LMN, abc), or, if
    direction only is concerned , merely the line LMN.

    EQUATIONS BETWEEN THE CONSTANTS OF A LINE.


    5515 The relation between the constants of a line with
    rectangular axes is
    Cºmº + mº = 1 ;
    and with oblique axes, it is
    5516 I cosatm cosBtn cos y = 1 .
    Proof.— The first by (Euc.1.47). The second by projecting the bent
    line QLCP (Fig . 169) upon PQ, thus PQ = QL cos a + LC cos B + CP cos y ,
    and QL = PQ .1, & c., by (5512 ).
    5517 Also, when the axes are oblique,
    cos a = 1 + m cos vt n cosu ,
    cos B = m + n cos 1 + I cos v ,
    cos y = nt I cos p + m cos ..
    PROOF. - By projecting QP in figure ( 169 ) and the bent line QLCP upon
    each axis in turn, and equating results ; thus PQ cos a = QL + LC cos ß
    + CP cos y, applying (5512) .

    5518 The relation between l, m , n and 1 , u , v is


    1? + m + nº + 2mn cos . + 2nl cos p + 2lm cos v = 1.
    Proof. — By eliminating cosa, cos B, cos y between (5516 ) and (5517).
    748 SOLID GEOMETRY .

    5519 The relation between cosa, cosß , cos y and 1, M , v is


    cos'a sin d + cos' ß sinºu + cos y sin ’ v
    + 2 cos ß cos y (cos j cosvớcos 1)
    + 2 cos y cos a (cos y cos . — cos )
    + 2 cos a cos ß (cos à cos j - cos v )
    = 1 - cos - cos' p - cos' v + 2 cos i cos cos v .
    Proof. — By eliminating l,m , n between the four equations in (5516 ) and
    (5517) .

    5520 The angle 0 between two right lines Imn, l'm 'n ', the
    axes being rectangular:
    cos 0 = ll + mm ' + nn '.
    PROOF. - In Figure ( 169 ), let QP be a segment of the line Imn. The
    projection of QP upon the line l'm 'n ' will be QP cos 0. And this will also
    be equal to the projection of the bent line QLCP, upon l'm 'n ', for, if
    planes be drawn through Q , L , C , and P , at right angles to the second line
    i'm 'n ', the segment on that line intercepted between the first and last plane
    will be = QP cos 0 , and the three segments which compose this will be
    severally equal to QL . , LC . m ', CP .n ', the projections of QL, LC, CP.
    Then , by (5512), QL = QP .1, & c.
    5521 sin 0 = (mn' — m 'n )? + (nľ – n'l)? + (Im ' — I'm ) .
    Proof. – From
    1 - cos' 0 = (1 +mº+ nº) (1+2 + m " + n '?) - (ll' + mm ' + nn')? (5515 , ' 20 ).
    5522 With oblique axes,
    cos 0 = ll’ + mm ' + nn' + (mn' + m 'n ) cosa
    + (nľ + n'l) cos p + (Im ' +lm ) cos v.
    PROOF. — As in (5520 ), substituting from (5517) the values of cosa, & c .

    EQUATIONS OF THE RIGHT LINE.


    X - a y - 6 % - C - a - 6b %
    5523 -
    M . N m 7
    n

    Here abc is a datum point on the line, and if r be put for the
    value of each of the fractions, r is the distance to a variable
    point xyz. L , M , N are proportional to the direction ratios of
    THE RIGHT LINE. 749

    the line, which ratios must therefore have the values


    L M
    5524 l = m =
    ✓ ( LP + Mº + N ✓ ( L ? + MP + N ) '
    N
    n = V(Lº+ Mº+ N9)
    5525 NOTE. — Instead of a , b , c in the equation we may use
    kL + a , kM + 6, kl + c , where k is an arbitrary constant.

    5526 The equations of a line may also be written in the


    forms x = dta, y = uz + B .

    5527 These are the equations of the traces on the planes of


    xz and yz, and are equivalent to

    5528 If the line is determined as the intersection of the two


    planes Ax + By + Cz = D and A 'x + B 'y + C 'z = D', we may
    write equations (5523) by taking
    L = BC - B'C, M = CA' — C' A , N = AB' — A 'B ,
    a _ DB' —V D'B ,
    = b0 == DA' N— D'A., eC == 0.
    0.

    PROOF. – Eliminate z between the equations of the planes, then the


    reciprocals of the coefficients of a and y will be L and M .

    5529 The projection of the line joining the points xyz and
    abc upon the line lmn is
    1(x — a )+ m (y — b) + n (3 — c).
    5530 Hence, when the line passes through abc, the square
    of the perpendicular from xyz upon it is equal to
    (ar — a)” + (y —b) + (z — c)2— {1(x — a) + m (7 - 6) + n (z — c)}”.
    5531 Condition of parallelism of two lines LMN, I'M 'N ':
    L : L = M : M ' = N : N ',
    750 SOLID GEOMETRY.

    5532 Condition of perpendicularity :


    LL' + MM ' + NN' = 0. (5520 )
    5533 Condition of the intersection of the lines (LMN, abe )
    and ( L ' M ' N ', a'b'c') (5514 ) :
    (a - a') (MN' – M 'N ) + (6 — 6') (NL' — N 'L )
    + (c - c ) (LM ' — L ' M ) = 0 .
    Proof. — Eliminate x , y , z between the equations

    Iq = +* = * = r and * IC = = * = r,
    by subtracting in pairs, and then eliminate r and r'.

    5534 The shortest distance between the same lines is


    (a - a') (MN' — M ' N ) + (6 — 6 ') (NL ' – N ’ L ) + ( c - c ) (LM ' - LM )
    ✓ { (MN' — M 'N ) 2 + (NL' – N ' L ) + ( LM ' - L' M )" }
    Proof. - Assume 1, u , v for the dir -cos. of the shortest distance. Then ,
    by projecting the line joining abc, a 'b' c' upon the shortest distance, we get
    p = (a - a ') + (b - b ') u + (c - C') v. Also, by (5520 ), LA + Mu + Nv = 0 and
    IX + M 'u + N ' v = 0 , giving the ratios 1 : u : y = MN ' — M ' N : NL - N ' L
    : LM ' – L ' M ; and (5524 ) then gives the values of A , M , v .

    5535 The equation of the line of shortest distance between


    the lines (lmn, abc) and (l'm 'n ', a'b 'c') is given by the inter
    section of the two planes
    1 ( x - a ) + m (y — b) + n (2 - y) = " Fu coso . .. .. (i. ) ,
    sin 0

    l( –a )+ m'(y =0 )+n'(x− y) = " coso...(ii.),


    sinº 0
    where u = l (a' — a ) + m (b’ — b) + n (c' — c),
    u' = l (a – a') + m ' (6 — 6') + n' (c - c ),
    and cos 0 = ll' + mm ' + nn'.
    PROOF. — ( Fig. 170.) Let O be the point xyz on the line of shortest dis
    tance AB ; P , Q the points abc, a 'b ' c' on the given lines AP, BQ. Draw BR
    and PR parallel to AP and AB ; RT perpendicular to BQ ; and QN, TV per
    pendicular to BR. Then _ RBQ = 0, RN = U , QT = u ', therefore NI
    = ucos 0 and RM = RN + NM = u + u ' cos 6, and in the right-angled tri.
    angle RTB, RM cosec 0 = RB, the projection of OP upon AP, that is, the
    THE RIGHT LINE. 751

    left member of equation (i.). Similarly for equation (ii.) . It should be


    observed that (i.) and (ii.) represent planes through AB respectively per
    pendicular to the given lines AP and BQ .
    5536 Otherwise, the line of shortest distance is the inter.
    section of the two planes whose equations are
    l' (x - a ) + m ' ( y - b ) + n ' (2 - 0)
    2 =- cos 6
    1 (x — a ) + m (y — 6 ) + n (z — c)
    _ 1 (x - a ') + m (y — 6') + n (2 - 0 )
    l' (x - a') + m ' (y — 6') + n' (2 - 0 )
    For these equations state that cos 0 is the ratio of the projections of OP
    or of OQ upon the given lines, and this fact is apparent from the figure.

    5537 Equations of the line passing through the two points


    abc, a 'b 'c' :
    X - a y - b 2 - 0
    a - a - 6 - 6 -

    5538 A line passing through the point abc and intersecting


    at right angles the line Imn :

    where L = lm (6 — 6') + nl (c - c ) – (mº + nº)(a - a'),


    and symmetrical values exist for M and N .
    Proof.— The condition of perpendicularity to lmn is
    Ll + Mm + Nn = 0 ; (5520)
    and the condition of intersecting the line is
    (a - a') (Mn - mN ) + (6 — 6') (Nl - nL) + (c - 0') (Lm - IM ) = 0 .
    These equations determine the ratios L : M : N .

    5539 Equations of the line passing through the point abc ,


    parallel to the plane Læ + My + Nz = D , and intersecting the
    line (l'm 'n ', a'b'c') :

    where l, m , n are found, as in the last, from


    Ll + Mm + Nn = 0 , and
    (a - a')(mn' — m 'n ) + (6 — 6')(nl - n 'l) + (c - c')(Im ' — I'm ) = 0 .
    752 SOLID GEOMETRY.

    5540 Equations of the bisector of the angle between the


    two lines lumini, lzm ng:
    X - _ y =
    litlemitm ni + ng
    PROOF . — Taking the intersection of the lines for origin , let 6,91 , x ,yaz , be
    points on the given lines equidistant from the origin ; then , if xyz be a point
    on the bisector midway between the former points, x = } ( , tag), & c .
    (4033) ; and the direction -cosines of a line through the origin are propor
    tional to the coordinates.

    5541 The equations of a right line in four plane coordinates


    are
    aI = M = = R. ........... (i.),
    where aßyd is a variable point, and a'B' y'8' a fixed point on
    the line. The relation between L , M , N , R is
    5542 AL + BM + CN + DR = 0 . .. ... . ..... .. (ii.) .
    Proof. – For, since equation (5503) holds for abyd and also for a 'B 'yo,
    we have A (a - a') + B (B - B') + C (y - 7 ) + D (8 - 5 ) = 0 .
    Substitute from (i.) a - a' = rL, B - B ' = rm , & c.
    5543 In tetrahedral coordinates the same equation (i.) sub
    sists, but the relation between L , M , N , R becomes , by
    changing Aa into a , & c.,
    5544 L + M + N + R = 0.

    THE PLANE.

    5545 General equation of a plane:


    Ax + By + Cx + D = 0 .
    5546 Equation in terms of the intercepts on the axes :

    * +* + = 1.
    5547 Equation in terms of p , the perpendicular from the
    origin upon the plane, and l, m , n , the direction -cosines of p :
    lx + my + nz = p.
    THE PLANE . 753

    PROOF. — If P be any point xyz upon the plane,and 0 the origin , the pro
    jection of OP upon the normal through O is p itself ; but this projection is
    lx + my + nz, as in (5520) .
    5548 The values of l, m , n , p for the general equation
    (5545) are
    A.
    I
    l= (A + B + C )' &c., p = 1 ( A² + B + C?)
    Proof. — Similar to that for (4060 - 2 ) : by equating coefficients in (5545)
    and (5547) and employing 12 + mº + n ? = 1.

    5550 The equation of a plane in four-plane coordinates is


    la + mB + ny + r8 = 0 ,
    with T :m : n :r = Qy : B. : 21 : 8 .
    1. P1 P2 P3 P4
    where an, Bi, Yu, &, are the perpendiculars upon the plane from
    A , B , C , D , the vertices of the tetrahedron of reference , and
    Pu, P2, P3, P4 are the perpendiculars from the same points upon
    the opposite faces of the tetrahedron .
    Proov. - Put y = : = 0 for the point where the plane cuts an edge of the
    tetrahedron, and then determine the ratio 1 : m by proportion .
    See Frost and Wolstenholme, Art. 81.
    5551 The equation of a plane in tetrahedral coordinates is
    also of the form in (5550 ) , but the ratios are, in that case ,
    lim : n : r = an: B. : 41 : 8 .
    The relation between the three -plane and four-plane coor
    dinates is a = p - lv - my - nz.

    5552 The equation of a plane in polar coordinates is


    ro {cos6 cos 0 + sin 8 sin 8 cos ($ - $ )} = p .
    Proof. Here p is the perpendicular from the origin on the plane, and
    p . 0 ', the polar coordinates of the foot of the perpendicular. Then , if 4 is
    the angle between p and r , we have p = r cosy and cos y from (882 ).

    5553 The angle 8 between two planes


    lx + my + nz = p and l'x + m 'y + n 'z = p
    5 D
    754 SOLID GEOMETRY.

    is given by formula (5520) , and the conditions of parallelism


    and perpendicularity by (5531) and (5532) , since the mutual
    inclination of the planes is the same as that of their normals .

    5554 The length of the perpendicular from the point x'yz


    upon the plane Ax + By + Cz + D = 0 is

    + 4Ax + By + Cx + PD
    + 7 (A’ + B++ C ) = p - lx'-my - nz'.
    Proof.- As in (4094 ).
    5556 The same in oblique coordinates
    _ (Ax + By + Cx + DD ) =_ p x cos a - y cos B - % cosy,

    where p is found from (5519) by putting A , B , C for p cos a,


    p cos , p cos y. This gives
    SA’ sin + Bº sin ’ u + C sin ’ v + 2BC (cos u cos v - cos 1) ?
    _ 7 + 2CA ( cos v cos i - cosu ) + 2AB (cos cosu - cos v ) )
    1 - cos' I - cos'u - cos* » + 2 cos i cos j cos v

    5559 The distance of the point d'y'z from the plane


    Ax + By + Cz + D = 0 , measured in the direction lmn, the axes
    being oblique:
    p =_ Ax' + By' + Cz' + D
    Al + Bm + Cn
    Proof. — By determining r from the simultaneous equations of the line
    and the plane, viz.,
    * 1 = * m= * = = = = = r and Ax + By + C++ D = 0.
    71

    Otherwise ,by dividing the perpendicular from x'y'z' (5554 ) by the cosine of
    its inclination to Imn, viz. Al + Bm + Cn
    VIL , ✓ (A? + B ? + C%)

    EQUATIONS OF PLANES UNDER GIVEN CONDITIONS.


    5560 A plane passing through the point abc and perpen
    dicular to the direction Imn :
    1(x — a) + m (y — b) + n (z — c) = 0.
    THE PLANE . 755

    5561 A plane passing through two points abc,a'l'c':

    IX
    + %+ C = 0,
    5562 with 1 +uto = 0 .
    ProOF. - By eliminating n between the equations
    1 (x - a ) + m (y - b ) + n (2 — c ) = 0 , l ( a - a ') + m (b — b ') + n (c - c') = 0 ,
    and altering the arbitrary constant.

    5563 A plane passing through the point of intersection of


    the three planes u = 0 , v = 0 , w = 0 :
    lu + mv + nw = 0 .
    5564 A plane passing through the line of intersection of the
    two planes u = 0 , v = 0 :
    lu + mv = 0 .
    5565 A plane passing through the two points given by
    u = 0 , v = 0 , w = 0 and u = a , v = b , w = c :
    lu + mu + nw = 0 with lat mb + nc = 0 .

    5566 The equation of a plane passing X y % 1


    through the three points X1Y171, X2Y222, X39373 X 1 yı > 1
    = 0.
    or A , B , C , is given by the determinant X 2 Y2 % , 1
    annexed , in which the coefficients of x , y , z
    | Q'; Y ; % z 1
    represent twice the projections of the
    area ABC upon the coordinate planes .
    PROOF. — The determinant is the eliminant of Ax + By + C2 = 1, and
    three similar equations. Expanded it becomes
    ® ( yzs - sz + /g=1 - 317g + ghĩx -2y= ) + g ( & c.) + = ( c.) + 21 /yzs - & c. = 0.
    Hence, by (4036 ), we see that the coefficients are twice the projections of
    ABC , as stated .

    5567 The sum of squares of the coefficients is equal to


    four times the square of the area ABC.
    Proof.-- For, if l, m , n are the dir-cos. of the plane,and ABC = S , the
    coefficients are = 281, 28m , 2Sn , by projection .

    5568 The determinant (21, 42, 23) , that is, the absolute term
    in equation (5566 ), represents six times the volume of the
    tetrahedron OABC, where 0 is the origin .
    756 SOLID GEOMETRY.

    Proof. — Writing the equation of the plane ABC , Ax + By + Cz + D = 0 ,


    we have for the perpendicular from the origin , disregarding sign,
    D
    p = JA + B + C2) = 28 (5567),
    therefore D = 2pS = 6 x the tetrahedron OABC.
    5569 If xyz be a fourth point, P , not in the plane of ABC,
    the determinant in (5566 ) represents six times the volume of
    the tetrahedron PABC.
    PROOF. — By the last theorem the four component determinants represent
    six times (OBCP + OCAP + 0ABP + OABC) for an origin O within the
    tetrahedron .

    5570 A plane passing through the points abc, a'b'c', and


    parallel to the direction Imn :
    X -a 4 - 6 % - 0
    a - a 6 - C-
    = 0.
    i mn
    a- a b— c—
    Proof. — Eliminate 1 , M , v between the equations (5561 – 2) and
    it
    a - a - 5 - 6 * c" > , = 0 , the condition of perpendicularity between Imn
    and the normal of the plane (5561).

    5571 A plane passing through the point abc and parallel


    to the lines Imn , i'm 'n' :
    X - a
    y —6 m m = 0.
    % -On n '
    PROOF. — The equation is of the form 1 (2 - a ) + (7 - 6 ) + v (2 - c ) = 0,
    and the conditions of perpendicularity between the normal ofthe plane and the
    given lines are ld + mu tnv = 0, 18 + m 'u + n 'v = 0 . Form the eliminant
    of the three equations.

    5572 A plane equidistant from the two right lines (abc, lmn)
    and (a 'b 'c', l'in 'n ') :
    X — } (a + a') ľ
    y - 1 (6 + 6 ) m m = 0.
    - 1 (c + c ) n n' By (5571).
    .X
    TRANSFORMATION OF COORDINATES. 757

    5573 A plane passing through the line (abc, lmn) and per
    pendicular to the plane l'x + m 'y + n 'z = pi.
    The equation is that in (5571).
    For proof, assume 1 , H , v for dir-cos. of the normal of the required plane,
    and write the conditions that the plane may pass through abc and that the
    normal may be perpendicular to the given line and to the normal of the
    given plane.

    TRANSFORMATION OF COORDINATES.

    5574 To change any axes of reference to new axes parallel


    to the old ones :
    Let the coordinates of the new origin referred to the old
    axes be a , b , c ; xyz and x'y'z', the same point referred to the
    old and new axes respectively ; then
    x = x ' ta, y = y' + b , % = + c.
    5575 To change rectangular axes of reference to new
    rectangular axes with the same origin :
    Let ox , OY, OZ be the original axes, and OX', OY', OZ'
    the new ones ,
    I myn, the dir -cos. of OX ' referred to OX , OY, OZ ,
    12 m₂na do . oy do . do .
    13mgng do . OZ do . do.
    xyz, End the same point referred to the old and new axes
    respectively . Then the equations of transformation are
    5576 x = l, $ + l2 n + lz $ ...... ............ (i.),
    v = m & m ,ntm Ś .................. (ii.),
    % = nı $ + nın + ngŚ ..................(iii.).
    And the nine constants are connected by the six equations
    5579 li + mi+ ni = 1 ...(iv.), lzls + m .mz+ Nznz = 0 ... (vii.),
    li + mi + n ; = 1 ... (v.), izli + mzma + nzn , = 0 ... ( viii.),
    Vs + mi+ n ; = 1 ...(vi.), 112 + mmy + nin , = 0 ... (ix.),
    so that three constants are independent,
    758 SOLID GEOMETRY.

    Proof. — By (5515 ) and (5532), since OX', OY', OZ'are mutually at


    right angles .

    5578 Therelations(iv.to ix.)may also be expressed thus


    m ,
    mıng - ming = nyls - n,1, = 1;m -- Izm . = + 1 .. ... (X.),
    mg No - + 1
    1
    = E . ..
    mzn , - ming ngli - nils lgm , - limg
    Iz mg Ny
    Ng _ = + 1 . .. . .. (xii.).
    ming - mzn nilz - neli lym , - lom
    Obtained by eliminating the third term from any two of equations
    (V11. - 1x .) . Also, by squaring each fraction in (x .) and adding numerators
    and denominators, we get
    +m +
    ( + m + 1)(n + m + n ) - (0,1 + m + n1 ]) = 1, by (5577).
    5579 If the transformation above is rotational, that is, if it
    can be effected by a rotation about a fixed axis , the position
    of that axis and the angle of rotation 0 are found from the
    equations 2 cos 0 = li + me + ng- 1,
    cos' a cos' B cos y
    5580
    ma + ng - 1, -1 n3 + 1, -m2- 1 - 17 + m , - ng - 1'
    where a , b , y are the angles which the axis makes with the
    original coordinate axes.
    Proof. — (Fig . 171.) Let the original rectangular axes and the axis of
    rotation cut the surface of a sphere, whose centre is the origin 0 , in the
    points x , y , z , and I respectively . Then, if the altered axes cut the sphere in
    Ě, n , %, we shall have 0 = L XI5 in the spherical triangle ; Ix = IĚ = a ; Iy =
    In = ß ; Iz = I = y , and by (882) applied to the isosceles spherical triangles
    & IĚ, & c., ly = cosxt = cos a + sin a cos 0 , m , = cos yn = cos* ß + sin ’ ß cos ,
    ng = cos zs = cos’ y + sin ' y cos 0. From these cos 0 , cos a , cos ß , and cos y
    are found .

    5581 Transformation of rectangular coordinates to oblique :


    Equations (i. to vi.) apply as before, but (vii. to ix .) no
    longer hold , so that there are now six independent constants.
    TËÈ SPÉERE. 759

    THE SPHERE.

    5582 The equation of a sphere when the point abc is the


    centre and r is the radius,
    (x — a)2 + (y — 6 ) + (3 c) = px.
    5583 The general equation is
    x + y² + x + Ax + By + Cx + D = 0.

    The coordinates of the centre are then - A , -


    and the radius = i✓ (A + Bº + C 2 — 4D).
    Proof. —-By equating coefficients with (5582).
    5584 If xyz be a point not on the sphere, the value of
    (a - a )? + ( y - 6 ) + (2 - 0) — 922 is the product of the segments
    ny right line drawn through xyz to cut the sphe
    Proof.- From Euc. III., 35, 36.

    THE RADICAL PLANE.


    5585 The radical planes of the two sphereswhose equations
    are u = 0 , u' = 0 , is
    - = 0.

    5586 The radical planes of three spheres have a common


    section , and the radical planes of four spheres intersect in the
    same point.
    PROOF. — By adding their equations, and by the principle of ( 4608)
    extended to the equations of planes .

    POLES OF SIMILITUDE.
    5587 DEF. — A pole of similitude is a point such that the
    tangents from it to two spheres are proportional to the radii.
    5588 The external and internal poles of similitude are the
    vertices of the common enveloping cones.
    760 SOLID GEOMETRY.

    5589 The locus of the pole of similitude of two spheres is a


    sphere whose diameter contains the centres and is divided
    harmonically by them .

    CYLINDRICAL AND CONICAL SURFACES .

    5590 DEF. — A conical surface is generated by a right line


    which passes through a fixed point called the vertex and
    moves in any manner.
    5591 If the point be at infinity , the line moves always
    parallel to itself and generates a cylindrical surface.
    5592 Any section of the surface by a plane may be taken
    for the guiding curve.
    5593 To find the equation of a cylindricalor conical surface.
    RulE. — Eliminate xyz from the equations of the guiding
    curve and the equations * I a = mm = D of any generating
    line ; and in the result put for the variable parameters of
    the line their values in terms of x, y , and z .
    5594 Ex. 1. – To find the equation of the cylindrical surface whose
    generating lines have the direction lmn, and whose guiding curve is given
    by b*x * + a*y * = a *b* and z = 0 .
    At the point where the line ma = 42
    m
    = ?ne meets the ellipse, z = 0 ,
    æ = a, y = ß . Therefore ba? + a* f * = a*b ?. Substitute in this, for the
    variable parameters, a, b, a = 2 - 12, B = y , ma; and we get, for the
    cylindrical surface bº (nx — lz )? + a²(ny - mz)' = aʻoʻna.

    5595 A conical surface whose vertex is the origin and


    guiding curve the ellipse b*v* + a ’y? = a’b?, z = C , is

    + - = 0.
    PROOF.--- Here the generating li = - . Atthe point of inter
    CONICOIDS. 761

    section of the line and carve z = c,æ = =mo; by + = a*b*.


    Substitute for the variable parameters lim : n the values æ : Y : % , and the
    result is obtained.

    CIRCULAR SECTIONS.
    5596 RULE. — To find the circular sections of a quadric curve,
    express the equation in the form A ( xº + y² + z² + c ) + & c . = 0 .
    If the remaining terms can be resolved into two factors, the
    circular sections are defined by the intersection of a sphere and
    two planes.
    5597 Generally the two quadrics
    ax + by2 + cz2 + 2fyz + 2gzx + 2hxy = 1
    and (a + 2 )aº + (6 + 1 ) yº + ( + 1 ) 72 + 2fyz + 2gzx + 2hxy = 1
    have the same circular sections.
    Proof. — Let r, p be coincident radii of the two surfaces having Imn for a
    common direction . Then -> = al + bm²+ cn ” + 2fmn + 2gnl + .2hlm and -
    po
    = the same + . . Therefore, if bas a constant value throughout any
    section , p is also constant throughout that section .

    5598 Ex. - An oblique circular cone whose vertex is the point a, 0, b ,


    and guiding curve the circle 2 * + y = 0 ; z = 0 ; is
    (az - bx)? + b*y* = c* (z — ) .
    The equation may be written
    23 ( x2 + y2 + x _ (*) = z {2abæ + (1 ? + c? — a ?) z — 26cº) ,
    and therefore the cone has two series of parallel circular sections, z = k and
    2abx + ( + 0? – a ) z — 2bc² = p (5583). ( Frost and Wolstenholme.)

    CONICOIDS.

    5599 Ders. — A conicoid is a surface every plane section of


    which is a conic .
    The varieties are the ellipsoid, the one-fold and two- fold
    hyperboloids, the elliptic and hyperbolic paraboloids, the
    spheroid of revolution , the cone, and the cylinder .
    5 E
    762 SOLID GEOMETRY.

    In any of the following equations of a conicoid, by making one of the


    variables constant, the equation of a section parallel to a coordinate plane is
    obtained, and the equation of the surface is by that means verified. Thus,
    in the equations of (5600 ) or (5617) , Figs. (172) and (173 ), if z be put
    = ON , we get the equation of the elliptic section RPQ, the semi-axes of
    which are NQ = a v (c - ON ?) and NR = ? (c?– ON ), a,b, c being the
    principal semi-axes of the conicoid ; that is, OA , OB , OC in the figure.

    THE ELLIPSOID .

    5600 The equation referred to the principal axes


    xes
    of the
    figure is
    = 1. (Fig . 172)
    + + 1. (Fig. 172)
    5601 There are two planes of circular section whose
    2 are es Se

    equations are

    **(+ - ) -- (4 - ) = 0,
    with a > b > c.
    Proof : (ak – 1)+ (1 - ) ( – 3) = 0
    is a cone having a common section with the conicoid and a sphere of radius
    r. If the common section be plane, one of the three termsmust vanish
    in order that the rest may be resolved into two factors.
    Since a > b > c, the only possible solution for real factors is got by
    making r = b .
    5602 Sections by planes parallel to the above are also
    circles, and any other sections are ellipses.

    5603 The umbilici of the ellipsoid (see 5777) are the points
    whose coordinates are
    Ja - - 62 16 -
    X = fun 2 y = 0, x = # cy

    Proor.— The points of intersection of the planes (5601) and the ellipsoid
    162 - - ?
    = Eeva
    EaVa - c >z
    (5600) on the xz plane are given by a ' = Hav
    Since, by (5602) the vanishing circular sections are at the points in the rz
    plane conjugate to æ ' and z ', we bave, by (4352) , x = - “ ', z = ' & '.
    CONICOIDS. 763

    5604 If a = b , in (5600), the figure becomes a spheroid , and


    every plane parallel to ay makes a circular section . Hence
    or oblate according as the ellipse is made to revolve about its
    major or minor axis .

    THE HYPERBOLOID .
    5605 The equation ofa one-fold hyperboloid referred to its
    principal axes is
    (Fig . 173 )
    + - = 1.
    5606 The planes of circular section , when a > b > c, are all
    parallel to one or other of the planes whose equations are

    69(1 )-- (4 + 2) = 0.
    PROOF. — As in (5601), putting = a.

    5607 The generating lines of this surface belong to two


    parallel systems (i.) and (ii.) below , with all values of 0 .
    5608
    . (i.) ...(ii.).
    = sin0– Žcos oj *** * = sin 0+ coso$*
    For the coordinates which satisfy either pair of equations,
    ( i.) or (ii.), satisfy also the equation of the surface. The
    equationsmay also be put in the forms
    X — a cosy - b sin o
    5610 a sin = Chaos A =

    5612 If % = 0 , x = a cos 0 and y = b sin 0 . Hence 0 is


    the eccentric angle of the point in which the lines (i.) and (ii.)
    intersect in the xy plane.

    5613 Any two generating lines of opposite systems intersect ,


    but no two of the same system do.
    764 SOLID GEOMETRY.

    5614 If two generating lines of opposite systems be drawn


    through the two points in the principal elliptic section whose
    eccentric angles are 0 + a , 0 - a , a being constant, the coordi
    nates of the point of intersection will be
    x = a cos 0 sec a , y = b sin 0 sec a, % = £ctan a ,
    and the locus of the point, as 0 varies, will be the ellipse
    v2 2
    5
    56115 ci Serika + spa = 1 ; % = Ectan a .
    Proof. — From (i.) and (ii.), putting 6 £a for 6 .*
    5616 The asymptotic cone is the surface given in (5595).
    Proof. - Any plane through the z axis whose equation is y = mx cuts the
    hyperboloid and this cone in an hyperbola and its asymptotes respectively.

    5617 The equation of a two-fold hyperboloid is


    (Fig. 174)

    and the equation of its asymptotic cone is

    5618
    PROOF. — Any plane through the x axis, whose equation is y = mz, cuts
    the hyperboloid and this cone in an hyperbola and its asymptotes respec
    tively .
    There are two surfaces, one the image of the other with regard to the
    plane of yz. One only of these is shown in the diagram .

    5619 The planes of circular section when bis > c are all
    parallel to one or other of the planes whose joint equation is

    **(* + + ) -- (4 - 1 ) = 0.
    Proof. — As in (5601), putting r = – b?.
    5620 If b = c, the figure becomes an hyperboloid of revo
    lution .

    THE PARABOLOID .
    5621 This surface is generated by a parabola which moves
    with its vertex always on another parabola ; the axes of the
    two curves being parallel and their planes at right angles.
    * The surface of a one-fold hyperboloid , as generated by right lines, may frequently be
    seen in the foot- stool or work -basket constructed entirely of straight rods of cane or wicker ,
    CENTRAL QUADRIC SURFACE . 765

    The paraboloid is elliptic or hyperbolic according as the


    axes of the two parabolas extend in the same or opposite
    directions.
    5622 The equation of the elliptic paraboloid is
    * + *= , (Fig . 175 )

    b and c being the latera recta of the two parabolas.


    Proof : QM° = .OM ; PN° = c.QN; .- QM ' + PNP = OM + QN = x.
    F :

    If b = c, the figure becomes the paraboloid of revolution .


    5623 Similarly the equation of the hyperbolic paraboloid is
    = x. (Fig . 176)*
    (Fig.176 )*
    5624 The equations of the generating lines of this surface
    are Í + = m and hot be =
    the upper signs giving one system of generators and the lower
    signs another system .
    5625 The equations of the asymptotic planes are les are

    + Fe = 0.

    CENTRAL QUADRIC SURFACE .

    TANGENT AND DIAMETRAL PLANES.

    5626 Taking the equation of a central quadric + +


    = 1 to include both the ellipsoid and the two hyperboloids
    The curvature of this surface is anticlastic, a sort of curvatura which may be seen in
    the saddle of a mountain ; for instance, on the smooth sward of some parts of the
    Malvern Hills, Worcestershire,
    766 SOLID GEOMETRY.

    according to the signsof b and c’, the equation of the tangent


    plane at xyz is

    en + EE+ = 1. By( 7)
    5627 If p be the length of the perpendicular from the origin
    upon the tangent plane atxyz,

    Proof.— From (5549)applied to (5626).


    5628 The length of the perpendicular let fall from any
    point End upon the tangent plane at xyz is

    P (en + E + -1). (5554& 5627)


    5629 Direction cosines of the normal of the tangent plane
    ataya, la personas ma n=
    Proof. — By (5548) applied to (5626) and the value in (5627).
    5630 If l, m , n are the direction cosines of p ,
    p = lx + my + nz and p = u ’ l? + b m² + cën ”.
    Proof. — (5630) By projecting the three coordinates x , y , z upon p .
    (5631) By substituting the values of x , y, z, obtained from (5029), in
    (5630 ) .

    5632 The equation of the normal at xyz is

    (8–c) i = (n =y) = (5-0)


    since the dir- cos. are the same as those of the tangent plane
    at (5626 ) .
    5633 Each term of the above equations
    = pv {($ — x )? + ( - y)' + (5 — )2}
    or p multiplied into the length of the normal. nor
    CENTRAL QUADRIC SURFACE . 767
    767

    PROOF. — Each term squared =

    Add numerators and denominators, and employ (5627).

    5634 Equation (5631) is the condition that the plane


    lx +my + nz = p may touch the conicoid ; and if p = 0 , we
    have for the condition of the plane læ + my + nz = 0 touching

    the cone +* + = 0,
    5635 aʼl + b m² + cʻn ' = 0 .

    5636 The section of the quadric made by a diametral plane


    conjugate to the diameter through the point xyz has for its

    equation By (5688 ).
    + + 0.
    5637 Hence the relation between the direction cosines of
    two conjugate diameters is

    + mm' + nn'= 0.

    ECCENTRIC VALUES OF THE COORDINATES.


    5638 These are defined to be
    x = ah, y = bu , > = cv, with l’ + ? + y2 = 1.
    5640 } , , v are the dir -cos. of a line called the eccentric
    line ; and & = 1 ) , n = 9M , G = rv are the coordinates of the
    corresponding point upon an auxiliary sphere of radius r.
    5641 The eccentric lines of two conjugate semi-diameters
    are at right angles . By (5637) .

    5642 The sum of the squares of three conjugate semi


    diameters is constant and = a + 62 + ca.
    PE00F. — Let a ', u', .' be the semi-diameters, and a 4 % . X ,Y ,Z2, X2Yză , their
    extremities. Put the eccentric values in the equations x + y; + z = a ", & c.,
    and add. By (5641), 11 + + 1 = 1, & c.
    768 SOLID GEOMETRY.

    5643 The sum of the squares of the reciprocals of the same


    is also constant.
    PROOF. — Put r , cos au, r, cos Bu r, cos y , for X1, 41,% , in the equation of the
    quadric . So for X2, 42, zg and Xz, ,Zg. Divide by ri, ru,rs, and add the results.

    5644 The sum of squares of reciprocals of perpendiculars


    on three conjugate tangent planes is constant.
    Proof. – For each perpendicular take (5627) , and substitute the eccentric
    values as in (5642).
    5645 The sum of the squares of the areas of three con
    jugate parallelograms is constant.
    Proof. — By the constant volume of the parallelopiped P , A , = P , A , = Pz43,
    (5648) and by (5644).

    5646 The sum of the squares of the projections of three


    conjugate semi-diameters upon a fixed line or plane is
    constant.
    Proof. — With the samenotation as in (5642), let (Imn) be the given line.
    Substitute the eccentric values (5638) in (la , + my, + nz,)? + (lx , + myz + 12. )
    + (lx , + myg + nzu)?. In the case of the plane we shall have
    a ” – (læs +my, + nz.)’ + & c.
    5647 CoR. — The extremities of three conjugate semi
    diameters being X Y1 1, X ,Y222, 234323, it follows that , by pro
    jecting upon each axis in turn ,
    2 +a +2 = a ; g + + = 0 ; s7 + = + = = c.
    5648 The parallelopiped contained by three conjugate semi
    diameters is of constant volume = abc.
    Proof. — By (5568), the volume = 1 x, y % = abc 1 Mi vil
    X Y Z 1 , Miro
    | A₂ M3 V3
    by the eccentric values (5638 ). But the last determinant = 1 by
    ( 584, I.) .

    5649 Cor. - If a', u', c' are the semi-conjugate diameters ,


    w the angle between a ' and l', and p the perpendicular from
    the origin upon the tangent plane parallel to a 'b ', the volume
    of the parallelopiped is pa'b' sin w = abc.
    CENTRAL QUADRIC SURFACE . 769

    5650 Hence the area of a central section in the plane of a 'b'


    - abc
    = ma'b' sin w = 740
    P

    5651 Quadratic for the semi-axis of a central section of the


    972 22 2

    quadric 2 + + = 1 made by the plane læ +my + nz = 0 :


    aʼlb m ? c?n ?
    q2 - 22 + 62 --22 + _ po2 = 0 .
    Proof. — The equation is the condition , by (5635), that the plane
    la + my + nz = 0 may touch the cone

    *( - 1) +y (1 - 1) ++ (1 - 1) = 0,
    as in the Proof of (5600). For another method, see (1863).

    5652 When the equation of the quadric is presented in the


    form
    ax® + by® + cz? + 2fyz + 2g +x + 2hxy = 1,
    the quadratic for 72 takes
    the form of the determi
    nant equation annexed . M
    Or, by expanding , and
    writing A ' for the same
    determinant, with the
    fraction erased , the 1 1 m n
    equation becomes
    A'p*+ {(6 + c)2 + (c + a)mº+ (a + b)n? – 2fmn - 2gnl — 2hlm }ju?
    – – mº – ? = 0.
    PROOF. — The equation of the cone of intersection of the sphere and
    quadric now becomes

    (a - 72) +(6- 7)y*+(c- *) *+ 2fyz+ 2gzx + 2høy = 0,


    and the condition of touching (5700) produces the determinant equation .
    5 F
    770 SOLID GEOMETRY.

    5654 To find the axes of a non-central section of the


    quadric + + = 1.
    Let PNQ ( Fig . 177) be the cutting plane. Take a parallel
    central section BOC , axes OB, OC , and draw NP, NQ parallel
    to them . These will be the axes of the section PNQ, and NQ
    ON2 N
    will be found from the equation et arão = 1.

    5655 The area of the same section

    = tøbe(1–
    where p ' and p are the perpendiculars from 0 upon the cutting
    plane and the parallel tangent plane.
    NO ?
    OF
    Proor.— The area = 1NP.NQ = * * *.OB.OC
    area

    = = (1 - ON") 06.00 = abe (1 – ), by (5650).


    SPHERO-CONICS.
    DEF. — A sphero-conic is the curve of intersection of the
    surface of a sphere with any conical surface of the second
    degree whose vertex is the centre of the sphere.
    Properties of cones of the second degree may be investi
    gated by sphero -conics, and are analogous to the properties
    of conics.
    A collection of formulæ will be found at page 562 of Routh' s Rigid
    Dynamics, 3rd edition .

    CONFOCAL QUADRICS.
    5656 Definition.-- The two quadrics whose equations are
    INITION Is are

    + + = 1 and content on = 1,
    are confocal. We shall assume a > b > c .
    5657 As decreases from being large and positive, the
    third axis of the confocal ellipsoid diminishes relatively to the
    CENTRAL QUADRIC SURFACE. 771
    others until = - , when the surface merges into the
    focal ellipse on the xy plane,

    a = 1.
    à still diminishing , a series of one-fold hyperboloids appear
    until 1 = — , when the surface coincides with the focal
    hyperbola on the zx plane,

    a ’ — 62 6² -
    The surface afterwards developes into a series of two-fold
    hyperboloids until i = - a , when it becomes an imaginary
    focal ellipse on the yz plane.

    5658 Through any point xyz three confocal quadrics can be


    drawn according to the three values of 1 furnished by the
    second equation in (5656 ). That equation, cleared of frac
    tions, becomes
    5659 19 + (a ’ + 62 + - X ? — yº — *)
    {b’cº + c’a’ + a’b? — (b’ + (?) a ? - (e* + a ”) yº — (a’ + ) } ^
    + a’b?e? -- b e*.. c°a?yº — a ’bm = 0 .
    These three confocals are respectively an ellipsoid , a one
    fold hyperboloid , and a two- fold hyperboloid . See Figure
    ( 178 ) ; P is the point ayz ; the lines of intersection of the
    ellipsoid with the two hyperboloids are DPE and FPG , and
    the two hyperboloids themselves intersect in HPK .
    PROOF. - Substitute for , successively in (5659) a ’, b?, c , – 00 ; and the
    left member of the equation will be found to take the signs + , - , + , -
    accordingly . Consequently there are real roots between a ’ and b ?, 12 and c?,
    cand – 00 .

    5660 Two confocal quadrics of different species cut each


    other everywhere at right angles.
    Proof. — Let a , b , c ; a', l', c'be the semi-axes of the two quadrics ; then ,
    at the line of intersection of the surfaces, we shall have

    va( en Hy) +v'(1 -4 )+2*(1 - )= 0,


    772 SOLID GEOMETRY.

    which, since a ' – a’ = 6 "2 – = 6" - c = , becomes the condition of per.


    pendicularity of the normals by the values in (5629) . Thus, in (Fig . 178 ),
    the tangents at P to the three lines of intersection of the surfaces are
    mutually at right angles .

    5661 If P be the point of intersection of three quadrics


    abıcı, a ,1 ,C2, azbzcz confocal with the quadric abc ; the squares
    of the semi-axes, de, dz, of the diametral section conjugate to
    Pin the first quadric are ( considering a > 0 , > A3, and writing
    the suffixes in circular order )

    dš = ai - a d = aivas,
    In the second , di = a -aš di = a ; - ai,
    in the third, di = aș - aș di = aš - aš.
    Or, if for a , a ;, a we put a ' + 11, a + 12, a' + 13,the above
    values may be read with 1 in the place of a and the same
    suffixes.
    PROOF. — Pat a : - a =

    are confocal quadrics. Take the difference of the two equations, and we
    obtain , at a common point ay 26 _ ut
    + & c. = 0 . Comparing this with
    (5651), the quadratic for the axes of the section of the quadric by the plane
    lx + my + nz = 0, we see that, if l, m ,n have the values , & c., je is identical
    with r '; the plane is the diametral plane of P ; and the two values of u are
    the squares of its axes. Let d , d, be these values ; then, since there are but
    three confocals, the two values of u must give the remaining confocals, i.e.,
    a; - d ; = a; and a ; - d = a .
    The six axes of the sections are situated as shown in the
    diagram ( Fig . 179 ). Either axis of any of the three sections
    is equal to one of the axes in one of the other sections, but
    the equal axes are not those which coincide. O is supposed
    to be the centre of the conicoids, and the three lines are drawn
    from O parallel to the three tangents at P to the lines of in
    tersection .

    5662 Coordinates of the point of intersection of three con


    focal quadrics in terms of the semi-axes :
    CENTRAL QUADRIC SURFACE . 773

    aſažas , 666
    la - 63)(a - ) y* = Tu - 13) (6 - 0 )
    cic; c;
    (ai – c )(6 — c )
    The denominators may be in terms of any of the confocals
    since ai – = až — = a — bž, & c.
    +
    PROOF. — The equation of a confocal may be written +
    = 1, producing a cubic in a’, the product of whose roots a , a , a gives a?.
    5663 The perpendiculars from the origin upon the tangent
    planes of the three confocal quadrics being P1, P2, P3 :
    azoch
    ni = (q3– 4:30 – a ) p.; = (a?–4;)& - a)
    a

    På = (a}– a )(a - a )
    PROOF. — By (5649), pidzdz = a;bıcı ; then by the values in (5661).

    RECIPROCAL AND ENVELOPING CONES.


    5664 DEF. — A right line drawn through a fixed point always
    perpendicular to the tangent plane of a cone generates the
    reciprocal cone.
    The enveloping cone of a quadric is the locus of all
    tangents to the surface which pass through a fixed point
    called the vertex .
    5665 The equations of a cone and its reciprocal are respec
    tively

    Ax*+ By + C+ = 0.....(i.), and ** + + * = .....(i.).


    Proof. — The equations of the tangent plane of (i.) at any point xyz, and
    of the perpendicular to it from the origin , are
    Amb+ Byn+ C+5 = .......(ii.), and = B = e ... (iv.).
    Eliminate x , y , z between (i.), (iii.), and (iv.).

    5667 The reciprocals of confocal cones are concyclic ; that


    774 SOLID GEOMETRY.

    is, have the same circular section ; and the reciprocals of con
    cyclic cones are confocal.
    Proof. — A series of concyclic cones is given by
    Ax? + By' + Cz’ + 1 (xº+ y’ + + ) = 0
    by varying \ ; and the reciprocal cone
    2
    is
    x3
    47 В + , т Ctī = 0 . (5665 )

    5668 Thereciprocals of the enveloping cones of the series


    of confocal quadrics + t = 1, with fgh for
    the common vertex , P , of the cones, are given by the equation
    a’ l? + b*y* + cºm — (fe + gy + hz)? + 1 ( ? + y* + x ) = 0 .
    Proof. — Let Imn be the direction of the perpendicular p from the origin
    upon the tangent plane drawn from P to the quadric. Equate the ordinary
    value of på at (5631) with that found by projecting OP upon p ; thus
    (a ’ + ) 7+ + (6 + 1) m² + (c + 1) nº = ( fl + gm + hn )?.
    Now p generates with vertex () a cone similar and similarly situated to the
    reciprocal cone with vertex P , and l, m , n are proportional to x , y , z, the
    coordinates of any point on the former cone. Therefore, by transferring the
    origin to P , the equation of the reciprocal cone is as stated.
    5669 Cor . — These reciprocal cones are concyclic ; and
    therefore the enveloping cones are confocal (5667).

    5670 The reciprocal cones in (5668 ) are all coaxal.


    PROOF. - Transform the cone given by the terms in ( 5668) without A to
    its principalaxes ; and its equation becomes Ax? + By? + Cz* = 0 . Now , if
    the whole equation , including terms in 1, be so transformed , a + 1° + 7+ will
    not be altered . Therefore we shall obtain
    (A + 1 ) x' + ( B + 1) y’ + ( C + 1) =+ = 0,
    a series of coaxal cones.

    5671 The axes of the enveloping cone are the three normals
    to the three confocals passing through its vertex.
    Proof. — The enveloping cone becomes the tangent plane at P for a con
    focal through P, and one axis in this case is the normal through P. Also
    this axis is common to all the enveloping cones with the same vertex , by
    (5670 ). But there are three confocals through P (5658 ) , and therefore
    three normals which must be the three axes of the enveloping cone .

    5672 The equation of the enveloping cone of the quadric


    THE GENERAL QUADRIC. 775

    a ? + 1 + wetn+
    04 2=1 isis,, when
    when transformed to its prin
    cipal axes,

    on =0 or +xH ma xH= 0,
    where hede, dz are the values of 1 for the three confocals
    through P , the vertex , and dz, dz are the semi-axes of the
    diametral section of P in the first confocal (5661) .
    PROOF. — Transform equation (5668 ) of the reciprocal of the enveloping
    cone to its principalaxes, as in (5670 ) . Let , .,, 1 be the values of i
    which make the quadric become in turn the three confocal quadrics through
    P . Then the reciprocal ( A + 1) x : + (B + ) y ? + (C + 1) z = 0 must become
    a right line in each case because the enveloping cone becomes a plane.
    Therefore one coefficient of æ*, y', or must vanish. Hence A + l, = 0,
    B + 1 , = 0 , C + 13 = 0 . Therefore the reciprocal cone becomes
    (1 - 1 ,) a * + (^ - ,) y' + (1 - 13) z* = 0 ,
    and therefore the enveloping cone is
    22 y2 + = 0.

    THE GENERAL EQUATION OF A QUADRIC .

    5673 This equation will be referred to as f (x , y , z ) = 0 or


    U = 0 , and , written in full, is
    ax' + by’ + cx? + 2fyz + 2gzx + 2hxy + 2px + 2y + 2rz + d = 0 .
    By introducing a fourth quasi variable t = 1 , the equation
    may be put in the homogeneous form
    5674 ax + by + cxº+ dwº + 2fyz + 2g2x + 2hxy
    + 2pxt + 2qyt + 2rzt = 0 ,
    abbreviated into
    (a , b , c, d , f, g , h , p , q,rXx, y , , t)' = 0,
    as in ( 1620).
    as

    Transforming to an origin a 'y' and coordinate axes


    parallel to the original ones, by substituting a' + Ě , y' tn, a +
    for x, y , and %, the equation becomes, by (1514),
    776 SOLID GEOMETRY.

    5675 a& tbm '+ 252 + 2975 + 2956 + 2h5n


    + & U : + qU , + ŠU, + U = 0,
    where U = f (x , y ,z ) (omitting the accents).
    5676 The quadratic for r , the intercept between the point
    a 'y' and the quadric surface measured on a right line drawn
    from x 'y'=' in the direction lmn, is
    po? (aľ + bmº + cn ’ + 2finn + 2gnl + 2hlm )
    + r (IU +mU, + nU .) + U = 0 .
    Obtained by putting & = rl, n = rm , L = rn in (5674).
    5677 The tangents from any external point to a quadric are
    proportional to the diameters parallel to them .
    Proof. — From ( 5676 ), as in (1215) and (4317) .
    - - -- - - - -
    5678 The equation of the tangent plane at a point xyz on
    the quadric is
    (5 — x ) U . + (n - y) U , + (5 — ) U , = 0
    5679 or $Ur+ nU, + GU, + TU , = 0 ,
    with = and t made equal to unity after differentiating .
    Proof. — From (5676 ) . Since xyz is a point on the surface, one root of
    the quadratic vanishes. In order that the line may now touch the surface,
    the other root must also vanish ; therefore lU . +mU + nU , = 0 . Pat
    rl = & - x , rm = n - y , in = 5 - z ; Ens being now a variable point on the
    line, and therefore on the tangent plane.
    5680 Again , « Uz + yU , + zU , + tU , = 20 , by (1624),
    therefore « Ux + yU , + zU , = - tUts
    which establishes the second form (5679) .

    5681 Equation (5679) also represents the polar plane of


    any point xyz not lying on the quadric surface. Written in
    full it becomes
    Š (ax + hy + gz + p ) orx (as + hn + g $ + p )
    + n (hx + by + fz + q ) + y (hø + on + f6 + )
    + $ (gx + fy + cx + r ) + (g $+ fm + 05 + r )
    + px + ay + rz + d = 0, + pš + an + r5 + d = 0 .
    THE GENERAL QUADRIC . 777

    5683 That is, the forms


    Uc + nU , + GU , + U = 0 and XU + yU, + xU ; + U = 0
    are convertible, U standing for f (x , y , z ) in the first , and for
    f (?, n , %) in the second.
    5685 The intersection of the polar planes of two points is
    called the polar line of the points.
    5686 The polar plane of the vertex is the plane of contact
    of the tangent cone.
    Proof. If ns be the vertex and xyz the point of contact, equation
    (5683) is satisfied. If x , y , z be the variables and ens constant, the second
    form of that equation shows that the points of contact all lie on the polar
    plane of the point ins.

    5687 Every line through the vertex is divided harmonically


    by the quadric and the polar plane.
    Proof. — In equation (5684) put x = $ + Ri, y = n + Rm , : = + Rn to
    determine R , the distance from the vertex to the polar plane. This gives
    - 20
    R = TU + mU, + nU '
    I , employing (5680) .
    Now , if r, r ' are the roots of the quadratic (5676 ), with &, n, Ś written for
    2, y,ā, it appears that r 200+ por = R , which proves the theorem .

    5688 Every line (lmn) drawn through a point xyz parallel


    to the polar plane of that point is bisected at the point, and
    the condition of bisection is
    lUr + mU,, + nU , = 0 .
    Proof. — The equation is the condition for equal roots of opposite signs in
    the quadratic (5676 ). Since l, m , n are the dir . cos. of the line and UX, U ,,
    U , those of the normal of the polar plane (5683), the equation shows that
    the line and the normal are at right angles (5532) .
    5689 The last, when X , Y, % are the variables, is also the
    equation of the diametral plane conjugate to the direction Imn .
    Expanded it becomes

    (al + hm + gn) x + (hl + bm + fn ) y + (gl + fm + cn )


    + pl + om + rn = 0 .
    5 G
    778 SOLID GEOMETRY.

    For the point xyz moves, when x , y , z are variable, so that every diameter
    drawn through it parallel to Imn is bisected by it, and the locus is, by the
    form of the equation , a plane.
    If the origin be at the centre of the quadric, p, q , and r of course vanish.

    5690 The coordinates of the centre of the general quadric


    U = 0 (5673) are

    Proof. — Every line through xyz, the centre, is bisected by it. The condi
    tion for this, in (5688 ), is U = 0 , U , = 0 , and U , = 0 , in order to be inde
    pendent of Imn. The three equations in full are
    la h 9 p la h 91
    ax + hy + gz + p = 0 ) . _ hb q
    hæ + by + fz + q = 0 ; and A ' = I g A = nb
    gx + fy + cz + r = 0 ) cr
    f f cl
    Ip q r d Tg
    Solve by (582 ).

    5691 The quadric transformed to the centre becomes


    ag + bm* + 062 + 2675 + 2988 + 2h6n + 4 = 0.
    PROOF. - By the last theorem , the terms involving g, n , in (5675 ) vanish .
    The value of Ŭ or f ( x, y, z ), when xyz is the centre, appears as follows:
    U = { U2 (5680) = pæ + ay + rz + d = P 2A

    The last equation , being again transformed by turning the


    axes so as to remove the terms involving products of coordi
    dinates, becomes
    5692 axº +Byº + ya + 4 = 0,
    5693 where a, ß , y are the roots of the discriminating cubic
    R – Rº(a + b + c) + R (bc + ca + ab - 49 - g — ) – A = 0,
    or (R - a )(R - 6)( R — c) – ( R - a)f2— (R — b) ge- (R — c)h
    - 2fgh: = 0.
    PROOF. — It has been shown, in (1847 – 9), that the roots of the discrimi.
    nating cubic (multiplied in this case by - > ) are the reciprocals of the
    maximum and minimum values of x + y + x . But such values are evidently
    THE GENERAL QUADRIC . 779

    the squares of the axes of the quadric surface. Let the central equation of
    the surface be + + = 1. Therefore = - 4a, &c., producing
    a 6
    the equation above.

    5694 The equations of the new axis of x referred to the old


    axes of ?, n, $ are
    (F + af ) x = (G + ag ) y = (H + ah) z ;
    and similar equations with ß and y for the y and 2 axes .
    Proof.— When Imn, in (5689), is a principal diameter of the quadric,the
    diametral plane becomes perpendicular to it, and therefore the coefficients of
    x , y , z must be proportional to l, m , n . Putting them equal to R1, Rm , Rn
    respectively , we have the equations
    (a - R ) l + hm + gn = 0 ...... The eliminant of these equations is
    hl + (b - R ) m + fn = 0 ...... (2) { . the discriminating cubic in R al
    gl + fm + (c - R ) n = 0 ...... ( 3 ) ) ready obtained in (5693) .
    From ( 1) and ( 2 ), lim = hf - 9 (b - R ) : gh - f ( a - R ),
    and from (2 ) and ( 3 ), m : n = fg - h (c - R ) : hf - g (b - R ) ;
    therefore (gh - af + Rf) l = (hf - bg + Rg) m = ( fg - ch + Rh ) n ,
    which establish the equations, since x : y : z = limin and F = gh - af,
    & c., as in (4665 ).

    5695 The direction cosines of the axes of the quadric.


    If the discriminating cubic be denoted by • ( R ) = 0 , and
    its roots by a , ß , y ; the direction cosines of the first axis are

    For the second and third axes write ß and y in the place of a .
    PROOF. — Let F + af = L, G + ag = M , H + ah = N .................. (i.),
    (a − b) ( a – c) - 5° = 1, (a - c )(a - a) - gʻ = M , (a – a ) (a − b) - h = v ...(ii.).
    Then the equation • (a ) = 0 may be put in either of the forms
    L = uv, M = v ., N = Au ............ ......... (ii.).
    Now the dir. cos. of the first axis are, by (5694 ), proportional to
    1 1 1
    ī M N = v : Vp :Vv, by (iii.) .
    Their values are, therefore,
    ✓
    Vapp to) VCA ** +v) ( +* + v)"
    But i = - agla and itu turug , 2, by actual differentiation of the
    - da
    cubic in (5693) .
    780 SOLID GEOMETRY.

    5696 Cauchy's proof that the roots of the discriminating cubic (5693)
    are all real will be found at (1850) .

    5697 The equation of the enveloping cone, vertex xyz, of


    the general quadric surface U = 0 (5673) is
    4 (abcfghXlmn)? U = (1U + mU, + nU .)?,
    with Ğ - X, n - y, f — % substituted for l , m , n.
    PROOF. — The generating line through xyz moves so as to touch the
    quadric. Hence the quadratic in r (5676) must have equal roots. The equa
    tion admits of some reduction .

    5698 When U takes the form ax® + byl + cza = 1 , equation


    (5697) becomes
    (al + bmº+ cnº) (ax? + by? + cm + - 1) = (alx + bmy + enz)..
    5699 The condition that the general quadric equation may
    represent a cone is A ' = 0 ; that is , the discriminant of the
    quaternary quadric, (5674 ) or (1644 ), must vanish .
    PROOF. - By (5692). Otherwise A ' = 0) is the eliminant of the four
    equations Uz = 0 , 0 , = 0 , U : = 0 , U = 0 , the condition that equation
    (5675 ) may represent a cone.

    5700 The condition that the plane | a h § 1|


    lx + my + nz = 0 may touch the cone 1 h b fm
    (abcfghayz )2 = 0 is the determinant gfcn
    equation on the right.
    1 т п
    PROOF. — Equate the coefficients l, m , n to those of the tangent plane
    (5681), p , q , r being zero, and xyz the point of contact. A fourth equation
    is lx + my + nz = 0 , which holds at the point of contact . The eliminant of
    the four equations is the determinant above.

    5701 The condition that the a h g p l


    plane lx + my + nz + t = 0 may
    touch the quadric go f c 7 n = 0.
    (abcdfghpqrXayzl )2 = 0 pardt
    (5673) is the determinant equation li m n t
    on the right.
    PROOF. - As in (5700) .
    RECIPROCAL POLARS. 781

    5702 If the origin is at the centre, p = q = r = 0 . In that


    case , transposing the last two rows and last two columns, the
    determinant becomes

    ung
    | a h g 101 = 0 , or, da h g 1 = t | a
    h 6 f m 0 hbf m h bf :
    g fon 0 & fcn g fc
    I m n 0 1 т п
    0 0 0 td

    5703 The condition that the line of intersection of the


    planes
    lx +my + na + t = 0 ... (i.) and l'a + m 'y + n'z + ť = 0 ... (ii.)
    may touch the general quadric (abcdfghpqrXxyz1)2 = 0 , is the
    determinant equation deduced below .
    Multiply equation (i.) by & and (ii.) by n to obtain the plane
    (18 + l'n ) x + (mě+ m 'n ) y + (nĘ + n 'n ) z + t& + t' = 0 ......... (iii.),
    passing through the intersection of (i.) and ( ii.) . The line of intersection
    will touch the quadric if (ii .) coincides with the tangent plane at a point
    xyz, and if xyz be also on ( i.) and (ii.) . Therefore, equating coefficients of
    ( iii.) and the tangent plane at xyz (5681) , we get the six following equa
    tions, the eliminant of which furnishes the required condition ,
    axt hy + gz + pw = 18 + lin ) a h g l .
    hæ + by + fz + qw = mx + m 'n h b f qm m '
    gx + fy + cz + rw = net n 'n gf C n'
    px + qy + rz + dw = tě + t'n ? Pq go dt ť = 0.
    1x + my + nz + tw = 0 I m n t
    l'x + m ' y + n'z + t' w = 0 ľm n

    RECIPROCAL POLARS.

    5704 The method of reciprocal polars explained at page 665


    is equally applicable to geometry of three dimensions.
    Taking poles and polar planes with respect to a sphere of
    reciprocation , we have the following rules analogous to those
    on page 666 .
    782 SOLID GEOMETRY.

    RULES FOR RECIPROCATING .


    5705 A plane becomes a point.
    5706 A plane at infinity becomes the origin .
    5707 Several points on a straight line becomeasmany planes
    passing through another straight line. These lines are called
    reciprocal lines.
    5708 Points lying on a plane become planes passing through
    a point, the pole of the plane.
    5709 Points lying on a surface become planes enveloping the
    reciprocal surface.
    5710 Therefore, by rules (5708) and (5709), the points in
    the intersection of the plane and a surface become planes
    passing through the pole of the plane and enveloped both by
    the reciprocal surface and by its tangent cone.
    5711 When the intersecting plane is at infinity, the vertex
    of the tangent cone is the origin .
    5712 Therefore the asymptotic cone of any surface is
    orthogonal to the tangent cone drawn from the origin to the
    reciprocal surface. The cones are therefore reciprocal
    5713 The reciprocal surface of the quadric is a hyperboloid,
    an ellipsoid , or a paraboloid , according as the origin is without,
    within , or upon the quadric surface.
    5714 The angle subtended at the origin by two points is equal
    to the angle between their corresponding planes.
    5715 The reciprocal of a sphere is a surface of revolution of
    the second order .
    5716 The shortest distance between two reciprocal lines
    passes through the origin .

    5717 The reciprocal surface of the general quadric


    (abcdfghpqr Xxyzl)2 = 0 (5674), the auxiliary sphere being
    à ? + y + z = k *, is
    |a h P Š or, if p = q = r = 1 ,
    h bf d a h g § 1 - k* | a h g |
    g fc $ = 0. hbfn h b f
    p q r d - k & fc $ & fc
    | Š in Š -K | Š in $ 0 |
    THEORY OF TORTUOUS CURVES. 783

    PROOF. — The polar plane of the point Ens with respect to the sphere is
    Ex + ny + šz — k = 0 . This must touch the given surface, and the condition is
    given in (5701) .

    5718 The reciprocal surface of the central quadric


    + + = 1, when the origin of reciprocation is the
    point a'yz , is
    ($a + g + ta –k ) = a + bºmcº ,
    or , with the origin at the centre,
    5719 a > $? + 67' + c^ 42 = k *.
    Proof. — Let p be the perpendicular froin a ' y 'z' upon a tangent plane of
    the quadric, and { n } the point where p produced , intersects the reciprocal
    surface at a distance p from a 'y'z'. Then
    k @p - 1 = p = lx' + my' + nz - / (aʼlº + 6*mº+ c*n ). (5630)
    Multiplying by e produces the desired equation.

    THEORY OF TORTUOUS CURVES.


    5721 DEFINITIONS. — The osculating plane at any point of a
    curve of double curvature, or tortuous curve,* is the plane
    containing either two consecutive tangents or three consecu
    tive points.
    5722 The principalnormal is the normal in the osculating
    plane. The radius of circular curvature coincides with this
    normal in direction .
    5723 The binormal is the normal perpendicular both to the
    tangent and principal normal at the point.
    5724 The osculating circle is the circle of curvature in the
    osculating plane, and its centre, which is the centre of circular
    curvature, is the point in which the osculating plane intersects
    two consecutive normal planes of the curve.
    5725 The angle of contingence , dt, is the angle between two
    consecutive tangents or principalnormals . The angle of torsion ,
    dr, is the angle between two consecutive osculating planes.
    * Otherwise named " space curve.”
    784 SOLID GEOMETRY.

    5726 The rectifying plane at any point on the curve is per


    pendicular to the principal normal; and the intersection of
    two consecutive rectifying planes is the rectifying line and
    axis of the osculating cone.

    5727 The osculating cone is a circular cone touching three


    consecutive osculating planes and having its vertex at their
    point of intersection .
    The rectifying developable is the envelope of the rectifying
    planes, and is so named because the curve, being a geodesic
    on this surface, would become a straight line if the surface
    were developed into a plane.
    5728 The polar developable is the envelope of the normal
    planes, being the locus of the line of intersection of two con
    secutive normal planes. Three consecutive normal planes
    intersect in a pointwhich is the centre of spherical curvature :
    for a sphere having that centre may be described passing
    through four consecutive points of the curve.
    5729 The edge of regression is the locus of the centre of
    spherical curvature .
    5730 The rectifying surface is the surface of centres (5773)
    of the polar developable .
    5731 An evolute of a curve is a geodesic line on the polar
    developable. It is the line in which a free string would lie if
    stretched between two points , one on the curve and one any
    where on the smooth surface of the polar developable.

    5732 In Figure (180) A , A ', A " , A "'' are consecutive points on a curve.
    The normal planes drawn through A and A ' intersect in CE ; those through
    A ' and A " in C ' E ', and those through A " and A " in ( " E " . CE meets C ' E '
    in E , and C ' E ' meets C ' E ' in E '. The principal normals in the normal
    planes are AC , A 'C ', A ' C ” , and these are also the radii of curvature at
    A , A ', A ” , while C , C ', C " are the centres of curvature. LACA' = dy and
    CA' C ' = dr.
    The surface ECC' C ' E ' is the polar developable , CCC" being the locus
    of the centres of curvature, and EE' E ' is the edge of regression .
    EA is the radius and E the centre of spherical curvature for the point
    A . hH , IIH ', H ' H " are elemental chords of an evolute of the curve , Ahl
    being a normal at A , and A 'HH ' a normal at A ', and so on . The first
    normal drawn is arbitrary, but it determines the position of all the rest.
    THEORY OF TORTUOUS CURVES. 785

    PROPERTIES OF A TORTUOUS CURVE.


    5733 The equation of the osculating plane at a point xyz on
    the curve is
    (5 — X) + (n - y)w + (5 - )v = 0 .
    5734 t , u , v are the direction cosines of the binormal, and
    their complete values are
    p (y;22 Yzm.), p (25825 — 725X6), p (XoY2s — X2, ).
    5735 The angle of contingence
    dy = V {(1,73 – .) + (7982, — %299 ) + (x .42 — X2y.)?} ds.
    PROOF. — Let the direction of a tangent be lmn, and that of a consecutive
    tangent 1 + dl, m + dm , n + dn. Since the normal of the plane must be per
    pendicular to both these lines, we shall have, by (5532),
    11 + mu tnv = 0 and (2 + al) 1 + (m + dm ) x + (n + dn ) v = 0,
    therefore 1 :Niv = mdn - ndm : ndl— Idn : ldm -mdl,
    and the denominator in the complete values of 1 , M , v is
    ✓ { (mdn – odm ) + & c. } = sin dy,
    by (5521) ; that is, = dy. Also l, m , n = xo, Ys, 7g and dl = x2,ds, & c.
    Therefore 1 = ( 4.729 Similarly, j and v ; and sv = p, by (5146) .

    5736 The radius of curvature p at a pointxyz.


    4 = x + 4 + x = x + 4že+ *1 —st
    Proof : dy = ✓ { (Y8 28 — 4252 )' + & c.} ds, in (5735),
    therefore yo = V { (x + y; + x ) (mke + ya + z ) - (x8X28 + YoY2s + Z228) }
    = v (xxx + yż, +rže) ; since w + ya + = 1 ;
    and differentiating this equation makes 2gX25 + & c. = 0 .
    Otherwise, geometrically , precisely as in the proof of (5141), we find the
    direction cosines of the principal normal to be
    5737 cos a = PX28, cos ß = PY282 cos y = p 25.
    Therefore p (az + y + x ) = cos ’ a + cos’ B + cos y = 1 .
    The change to the independent variable t is made by (1762) .

    5738 The angle of torsion, in terms of 1, M , v of (5734), is


    dr = 7 (83 + ki+ v.)ds = (4x3 + Myzs + vZ3s) pds
    = v { (uvo - .v )? + (vd - v )? + (Apes - here)?} .
    5 H
    786 SOLID GEOMETRY.

    Proof.— By (5745 ), we have (dr) = (dx)' + (du)' + (dv) ............ (i.),


    which gives the first form . The third reduces to this by the method in
    (5736) . For the second form put u = 48228 — Y287 , & c., then
    - =
    ✓ 1 ds du udk
    u 29 29
    Kdy ), di = K . K . & C.
    Substitute in (i.),reducing by K ' = u ? + v* + wº and KDK = udu + vdv + w dw .

    CURVATURE AND TORTUOSITY.


    Curvature

    sis
    Els
    5739 Radius of cury =

    Radius of torsion Tortuosity

    If to changes sign while passing through the values


    zero or infinity , there is a point of inflected torsion or
    a cuspidal point, respectively. If Tg, without changing sign ,
    passes through zero or infinity , there is a point of suspended
    torsion or infinite torsion respectively.
    If 7, is zero, identically , the curve is plane.

    5740 The radius of spherical curvature,


    R = V (p* + pr).
    Proof. — In Fig. (180) R ' = p + EC and EC = ps by analogy with q = py
    n a plane curve (see proof of 5147) .
    5741 The element of arc of the locus of centres of circular
    curvature is
    ds' = Rdr, and therefore R = $s.
    PROOF. - In Fig . ( 180) ds' = CC' = pdr seco = Rdr.

    5742 The radius of curvature of the edge of regression


    = S% = RR, = p + p2r ,
    S” being the arc of the edge of regression.
    PROOF. - An inspection of Figare ( 180) shows that R and p stand in the
    same relation to the edge of regression that r and p occupy with regard to a
    curve in the standard formula . In fact we may substitute R for r, p for p ,
    THEORY OF TORTUOUS CURVES. 787

    o for 0, - for y , and o remains o . The chosen line of reference AB being


    always parallel to the tangent EC, then AEC = BAE = 0 . Also the angle
    of contingence CEO' = CAC' = dt, by the right angles at C and C '. Ac
    cordingly , we have the formula p = s = rr, = p + P24 from (5146 – 8), and the
    values above corresponding to them .

    5743 A method of estimating the variation in direction of a


    right line whose position is given as depending upon the form
    of a tortuous curve at every point.
    Let x , y , z be the direction cosines of the line referred to a
    fixed principal normal, tangent, and binormal of the curve
    (x , y, % may either be constants with respect to the varying
    principal normal, tangent, and binormal, or they may be func
    tions of the angle between the binormal and the spherical
    radius ].
    5744. The complete changes in x , y, z , with respect to the
    fixed origin and axes , will be
    dx = dx + yd4 — zdt,
    dy = dy — xdy,
    8z = dz + xdı.
    PROOF. - In Figure (180 ) AC, AB are the fixed axes of x and z. Let a
    line AL of unit length bedrawn always parallel to the line in question ; then ,
    if x , y , z be the coordinates of L , x , y, z will also be the direction cosines of
    AL, and therefore of the given line.
    Now , suppose A to move to A ', and consequently AL to take the position
    A ' L '. Then the changes in x , y , z will be the changes dx, dy, dz relatively to
    the moving axes, plus the changes due to the rotations dy round the
    binormal and do round the tangent. With the usual notation, we shall have
    dx = dx + wyz -wy, dy = dy + Wgx — W ,ã, &z = dz + w .y - wył,
    with w , = 0 , w , = - di, wg = - dų .

    5745 If dx be the angular change in the direction of the


    right line,
    dx = v {(8x)? + (8y)² + (8x)2} .
    For dx = LL' since AL is a unit length.

    EXAMPLES.
    5746 The angle between two consecutive radii of circular
    curvature being de,
    (de)' = (dy) + (dr)".
    788 SOLID GEOMETRY.

    Proof. — Here, in (5744 ), x = l, y = 0, 2 = 0, therefore dx = 0 , dy = - dų ,


    ôz = dr. Substitute these values in (5745 ).
    5747 The angle , dn, between two consecutive radii of
    spherical curvature , o being the inclination to the binormal,
    (dn)² = (dy .sin )? + (dø - dr)”.
    Proof. - In (5744 ) the direction cosines of R (Fig . 180) are x = sing,
    y = 0 , z = cos ,
    therefore dx = cos º (dø - dr), òy = - dų sin o , èz = - sin q (do - dr).
    Substitute in ( 5745).

    5748 The angle of contingence of the locus of the centres


    of circular curvature,
    (dx) = (dys. cos 6)? + (dø + dr)?.
    Proof. — The dir. cos. of the tangent at C to the locus (Fig . 177) are
    x = coso, y = 0 , z = sin q ;
    therefore dx = - sin o (do + dr), dy = - dy cos , dz = cos 0 (dp + dr).
    Substitute in (5745 ).

    5749 The osculating plane of the same curve has its


    direction cosines in the ratios

    e ino cos4 :- Come tomy home cos* .


    PROOF. — As in the Proof of (5735 ), the dir. cos. of the normal to this
    plane are proportional to yèz - zdy, zòx — « dz, xày - yox. Substitute the
    values in last proof.
    5750 The angle of torsion of the same curve is found from
    (5745) and (5744 ) as above, x , y , z being in this case the dir.
    cos. of the normal of the osculating plane as given in (5749).

    5751 The direction cosines of the rectifying line are

    PROOF. — The rectifying plane at A ' (Fig . 180) is perpendicular to the


    normal A 'C '. Therefore its equation is x — ydy + zdt = 0 . The ultimate
    intersection of this plane with the rectifying plane at A (that is, the plane
    of yz) is the rectifying line. Hence the equation of the latter is ydy = zdr;
    and the dir . cosines reduce to the above by (5746 ) .
    THEORY OF TORTUOUS CURVES. 789

    5752 Cor. — The vertical angle of the osculating cone


    = 2tan- duo
    5753 The angle of torsion of the involute of the curve is
    = V (V + rs) de.
    PROOF.-- This angle is also the angle between two consecutive rectifying
    lines. Therefore, taking the dir. cosines from (5751), we must put in (5744 )

    therefore dx = duis dy da sedr = 0; èy = reade; dr = Yggde.


    5754 The angle of torsion of an evolute of the curve
    = dy sin (a — r ).
    ProOF. — (Fig . 180 .) Let HH ' H " be an evolute of the curve, AH the
    tangent to it in the normal plane of the original curve at A , and let a = CAH ,
    the inclination of AH to the principal normal. At any other point H " of
    the evolute , where its tangent is AWH'H " , let the corresponding angle be
    0 = (" A " H " . Then 0 = a - T, - being the sum of the angles of torsion
    between A and A ” , or the total amount of twist of the osculating plane . Now
    the normal of the osculating plane of the evolute at H ' , is perpendicular to
    HH ' and H ' H " , two consecutive tangents . Therefore its dir. cosines in
    (5744 ) must be
    x = - sin (a - -), y = 0 , x = cos (a - 1) ;
    therefore dx = cos (a — -) dr + 0 - cos (a - 1 ) dr = 0,
    dy = sin (a — ) dų ; dz = sin (a -- -) dr - sin (a – ) dr = 0.
    Hence the angle required = dy = dy sin (a - 7 ).

    5755 Approximate values of the coordinates of a point on a


    tortuous curve near to the origin in terms of the arc , the axes
    of x , y, % being the principal normal, tangent, and binormal,
    and the arc s being measured from the origin :

    *= * - - (1– 2p + pp.)+&c.,
    y = -- + +&c., * =66- A C + )+ &o,
    p and o being respectively the radii of circular curvature and
    torsion .
    790 SOLID GEOMETRY.

    Proof. — By Taylor's theorem (1500), since x , y , z, s are the same as


    dx, dy, dz, ds initially, we have x = xgs + 42, + inges8 + & c ., and similar ex
    pansions for y and z . The dir. cosines of the principal normal at the point
    xyz will be, from (5737),
    cos (-4) = p.2, cos ( +4) = py2, cos (5 -1) = pzza;
    y = dy and - = dt being estimated positive as drawn in Figure (180) for
    positive values of x, y, z.
    Differentiate these equations for s, and in the results put the initial values
    2', = 2, = 0, y = 1, = r = 0, 4, = , -, = 1, &c.,
    to determine the derivatives in the above expansions.

    THE HELIX .
    5756 The helix is a curve traced on a cylinder of radius a ,
    so that its tangent preserves a constant inclination , = n - a,
    to the axis . Taking the axis of the cylinder for the z axis of
    coordinates, the equations of the helix are
    x = a cost, y = a sin 0 , = að tan a .
    5757 The radius of curvature p = a secʻa.
    5758 The radius of torsion o = 2a cosec 2a .
    PROOF. —- p from (5806 ) ; since pı = a , pa = 00 , and 0 = a at every point.
    By (5739) , r = 8 . But dz = ds sin a and adt = dz cos a .

    5759 The helix of closest contact with a given curve may


    be found as follows.
    Determine the constants a and a from equations (5757 - 8 ) , with the
    known values of p and o for the given curve ; then place the helix to have a
    common tangent with the curve at the point, and make the osculating planes
    coincide.

    GENERAL THEORY OF SURFACES.

    5770 DEFINITIONS. — A tangent plane passes through three


    consecutive points on a surface which are not in the same
    right line.
    5771 The normal at any point of a surface is perpendicular
    to the tangent plane.
    THEORY OF SURFACES. 791

    5772 A normal plane is any plane through the normal.


    57173 A line of curvature on a surface is a line along which
    consecutive normals to the surface intersect. At every point
    of a surface there are usually two lines of curvature at right
    angles to each other ; and to these correspond two principal
    radii of curvature. The two lines of curvature coincide with
    the principal axes of the indicatrix at the point. See (5778 ) .
    5774 The surface of centres is the locus of the centres
    of principal curvature. There are two such surfaces, for
    there are two centres on each normal, and the normal is a
    tangent to both surfaces. Either surface may be regarded as
    generated by the evolutes of the lines of principal curvature.
    5775 A geodesic is a line traced on a surface along which
    C

    the osculating plane at every point contains the normal to


    the surface. See (5779).
    5776 The radius of geodesic curvature * of a curve traced
    on a surface is measured by the ratio of the element of arc of
    the curve to the angle between consecutive normal sections
    of the surface drawn through consecutive tangents of the
    curve. Geodesic curvature, being the reciprocal of this, is
    therefore the rate of angular deviation of the normal section
    per unit length of the curve.
    57717 An umbilicus is a point on a surface where a section
    parallel to and close to the tangent plane is a circle ; in other
    words, the indicatrix is a circle.
    For a definition of Indicatrix , see (5795 ).

    5778 In Figure (182) OCD is the normal at 0 to a curved surface ;


    AOA', BOB' are the lines of curvature, therefore the normals to the surface
    at A and O intersect in the centre of curvature radius pi (5773) , and the
    normals at B and 0 , in the centre, radius P2. The normals to the line of
    curvature BOB' at B and 0 , drawn in the osculating plane BOB', intersect in
    K , and those at B ' and O intersect in H . HOD is the angle between the
    osculating plane of the line of curvature and the plane of normal section .
    Similarly for the line of curvature AOA'.
    5779 If POP' be a geodesic, its osculating plane POP' contains OD the
    normal to the surface at 0 , and therefore p = OD, the radius of curvature
    of this section at 0 ; but PE, the normal to the surface at P , does not inter
    sect OD, the consecutive normal at 0 , unless the geodesic coincides with one
    of the lines of curvature, OA or OB. The angle DPE is the angle of torsion
    which vanishes in the latter case.
    * Not to be confounded with the radius of curvature of a geodesic .
    792 SOLID GEOMETRY.

    GENERAL EQUATION OF A SURFACE.


    5780 Let the general equation of a surface be represented
    by ♡ (x , y , z ) = 0 .
    5781 The equations of any tangent at a point xyz are
    W

    $7 = m = n with .+mo,+nø. = 0.
    PROOF. — At an adjacent point x + rl, y + rm , z trn , we have
    Q (x + rl, y + rm , ztrn ) = 0 ,
    therefore, by ( 1514 ), ¢ ( , y , z ) + r (192 +mp, + np.) = 0 ,
    the rest vanishing in the limit. But 9 (x , y, z) = 0, therefore
    10x + moy + ng : = 0 .
    But l, m , n are the direction cosines of the line joining the two points, which
    becomes a tangent in the limit ; and if &ns be any point on this line distante
    from xyz, & - x = pl, n - y = pm , 5 - z = pn, & c.

    5782 The equation of the tangent plane at xyz is


    ($ — x ) $ x + (n - y ) $ , + ($ — ) $ . = 0 .
    Proof. — Eliminate l, m , n from 1px + mpy + n°: = 0 by < - x = pl, & c.,as
    above.

    TANGENT LINE AND CONE AT A SINGULAR POINT.


    5783 If, in the expansion in (5781) by Taylor's theorem , all the deriva
    tives of $ (x , y , z ) of an order up to n inclusive vanish, we have
    pon + 1
    Q (x + rl, y + rm , z + rn ) = $ (x , y ,x ) + (ldz +md,, + nd2)* +10 (x , y,z ) = 0.
    There are in this case n + 2 coincident points at xyz in the direction lmn,
    and since the equation ( ldr + md,, + nd ,) * +10 (x , y , z ) = 0 is of the n + 1***
    degree in l, m , n ; n + 1 tangents to the surface at xyz can, in general, be
    drawn in any given plane through that point. This equation now takes the
    place of the conditional equation in (5781).
    5784 Equation (5782) is now replaced by
    {($ _ x)dx + (n - y) d,+ (5 ^ 2) d. }"$ (x, y,z) = 0,
    the equation of the locus of all tangents at the point xyz, and
    representing a conical surface generated by the motion of
    those tangents .

    5785 The equation of the normalat xyz is


    (5782 )
    THEORY OF SURFACES. 793

    5786 The equation of the tangent at a point a' y' on the


    curve of intersection of the tangent plane at xyz with the
    surface is

    with the two conditions


    16+ + u6, + v0. = 0 , 10x + moy tuon = 0 .
    For these are the conditions of perpendicularity to the normals of the
    tangent planes at xyz and x'y'z' respectively.
    There are three exceptional cases in which the ratios
    d : u : v have more than one set of values ; namely
    5787 1. — When Pixo Pre de vanish simultaneously, there is a
    tangent cone at xyz .
    5788 II. - When pixie Quie pze vanish simultaneously , X 'y'z is
    a singular point on the surface .
    5789 III. — When Pa = Py = Pz. In this case the point
    Par Pur P21
    a 'y' coincides with xyz, and the tangent there meets the curve
    in more than two coincident points, the condition for which is

    (1d. + ud, + vd.)?¢ x,y,z) = 0 ............ (i.),


    with 16c + ud, tvº : = 0 . .. (ii.) .
    These equations furnish two sets of values of the ratios
    diu : v , giving thereby the directions of two inflexional
    tangents (tangents to the curve of intersection ) at xyz , each
    meeting the surface in three coincident points . If all the
    derivatives of an order less than n vanish at xyz, equation (i.)
    will be replaced by (ad, tud, + vd -)" ° (x , y , z ) = 0 , which ,
    together with (ii.), will determine n inflexional tangents at
    the point.

    5790 The polar equation of the tangent plane at the point


    100, 1 , 0 , $' being the variables, is, writing u for 7 - 1,
    u ' = (ucos 6 — u sino) cos 6 + (usin 0 + u ,cos6 ) cos ($' $ ) sin
    + up cosec 0 sin ($ - $ ) sin 8 .
    5 I
    794 SOLID GEOMETRY .

    Proof. — Write the polar equation of the plane through paß , the foot of
    the perpendicular on the plane from the origin ; thus
    pu = cos 0 cos a + sin 0 sin a cos (0 - 3 ) .
    Differentiate for 0 and 0 to find pug and pus, and eliminate p , a , and B . This
    elimination is troublesome.

    5791 The length of the perpendicular from the origin upon


    the tangent plane at xyz ,
    p = XQr + yo, + 20 nc
    (5782, 5549)
    V {$* + $ + $ } < {$ + $ + $}} '
    the second form being the value of p when the equation of
    the surface is ¢ (x , y , z ) = C , a constant, and when • is a
    homogeneous function of the nth degree (1624 ).
    5793 In polar coordinates ,
    = u + u : + uż cosec 0 = " + r + ri cosec*

    Proof. - Add together the values of the squares of pu, pup,and pu , found
    in (5790).
    For a geometrical proof, see Frost and Wolstenholme, Art. (314 ).

    THE INDICATRIX CONIC.


    5795 DEF. — The indicatrix at any point of a surface is the
    curve in which the surface is intersected by a plane drawn
    parallel to the tangent plane at that point and infinitely near
    to it.
    5796 The following abbreviations will be employed
    The derivatives of ♡ (x , y , z ), P2.29 024 , 022, yz , Pzxe Pry .x, y, Pa,
    will be denoted by a, b , c, f, g, h , l, m , n .
    5797 PROP. — The indicatrix at a point xyz of a surface
    0 (x , y, z ) = 0 is the conic in which the elementary quadric
    surface
    5798 I.
    aff+bn®+c8 +2F95+2016+2160 = * VP +mº +n*= N
    is intersected by the tangent plane at xyz, whose equation is
    5799 II. 18 + mq nl + FN = 0.
    THEORY OF SURFACES. 795

    The origin of coordinates is the point xyz in both equa


    tions. R is an indefinitely small radius from the centre of
    the quadric ( I.) to a point End on the indicatrix, and p is the
    radius of curvature of the section of the surface o by a normal
    plane drawn through R ; the ratio R : p being constant for
    all such planes .
    PROOF. — Let 0 , in Fig . (181), be the point xyz on the surface q ; æ + ,
    y + n , z + 5 an adjacent point P . Then
    © (x + Ě, y + n, z + 5) = (x , y , z ) + 16 + mn + nS + } (a ? + ... + 2h &n ) + & c.
    With xyz for origin , draw the quadric surface
    a + bna + 05* + 2fns + 295€ + 2hen = N ............ ....... (i.)
    and the plane 15 + mn + m + 1 = 0 ......... ... ............ (ii.).
    Since &, n, 5 are very small, N is likewise . Also the unwritten terms in the
    above expansion may be neglected in the limit . Hence, any point Ens lying
    on the intersection of the quadric (i.) and the plane (ii.) will also lie on the
    original surface o (x + Ě, y + 1 , 2 + 5) = 0.
    To determine N , we have the perpendicular from xyz upon the plane (ii.) ,
    - N
    P = 2 (12 + m ne + nº) (5549). The radius of curvature of the section of the
    surface o made by a normal plane at O drawn through P being p, we have
    p= , and therefore N = - * * 1 (C + mº+na).
    In the Figure, R = OP, p = OL, and the intersection of (i.) and (ii.) is
    the conic PSQ . Since p is indefinitely small, we may put N = 0 in equa
    tion (ii.) . This amounts to taking the parallel section of the quadric by the
    tangent plane at O instead of the section PSQ. But these two will be equal
    in all respects , since the section of the quadric is a central one.

    5800 If m = 0 , equation (II.) becomes 18 + nx = 0 , and if


    the inclination of the indicatrix plane to the plane of xy be a ,
    tan a = - - . To obtain , in this case, the equation of the
    indicatrix in its own plane, put & = & cos a, = % sin a, and
    n = n', in equation (I.) .
    5801 When none of the three constants l, m , n are zero,
    the quadric ( I .) simplifies as follows
    From (II.) we have l& + mn = - nt and two similar equa
    tions. Square these, and by the results eliminate the terms
    in nă, ZE, Én from (I.), which then becomes
    5802 III. HÉ? + Kn' + L¢ = N ,
    where H = a+ mim (lj–mg-nh), K = 6+ * (mg- nh –lf),
    m

    L = c+ im (nh – lf –mg).
    796 SOLID GEOMETRY.

    This is the equation of another quadric intersecting the


    plane (II.) in the indicatrix .

    5803 The equation of a surface for points near an origin 0


    ( Fig . 182), the normal at O being taken for z axis , is
    2
    (7

    * +
    P1 P2 = 22,
    where p1, p2 are the radii of curvature of the normal sections
    through the x and y axes , and those sections will be proved to
    be the lines of curvature at 0 .
    PROOF. — Let AC = a and BC = b be the semi-axes of the indicatrix conic
    at a small distance z from 0 (5795). The equation of the conic will there.
    fore be but a - = 2p, and L = 2p2, giving the equation
    required .
    Secondly, on a line of curvature, the normal to the surface at a point zyz
    will intersect the x axis (5773) . The condition for this, by (5533) ( with
    xyz for abc, the origin for a'b'ć',
    L, M , N = 42, 4u, da (5785) = 22, 24, - 2, and L', M ', N' = 0,0, 1),
    Pa Pa
    gives xy = 0 , therefore x = 0 or y = 0 on a line of curvatare.
    P2
    Q . E . D.

    5804 If the equation of the surface with the same axes be le axes

    % = ax® + 2hxy + by? + 2fyz + 2gzx + cz? + higher powers,

    then A = Za P: = 2
    Proof.— Put y = 0 and dividebyz, therefore 1 = a **+ 232 + cz+&c.
    When x and z vanish, we have 1 = 2api.

    5805 For a normal section making an angle 0 with AC,


    - = 2 (a cosa 6 + 2h sin 0 cos 6 + 6 sinº 6).
    p

    PROOF. — Turning the axes in (5804) through the angle 8 by (4049), the
    coefficient of a ' becomes a cos 0 + as above.

    5806 Euler's Theorem . - If be the radius of curvature of


    THEORY OF SURFACES.

    any other normal section at 0 ,making an angle ACP = 0 with


    AČ (Fig . 182),
    1 =_ cos' 0 _+sin 8.
    р - р т р.
    Proof. — Let r = CP ; then x = r cos0, y = r sin 0, and g = 2pz, which
    substitute in (5793) .

    5807 COR. — The sum of the curvatures of two normal


    sections at right angles to each other is constant ; or, if p , e
    be the radii of curvature for those sections, and pay po the
    radii for the principal sections,
    - =
    1. Pa Po

    5808 The radius of curvature of a normal section varies as


    the square of the radius of the indicatrix in that section .
    Proof. — From på = 2pz, in Figure ( 182).
    5809 Meunier's Theorem . - The radius of curvature of an
    oblique section of a surface is equal to the radius of curvature
    of the normal section through the same tangent multiplied by
    the cosine of the inclination of the planes.
    Proof.—( Fig. 183.) p = PN , = PA7?, therefore ! = NO = cos q,
    19

    when NO and NC vanish .


    -

    5810 Quadratic for yx at a point on the surface % = ( , y )


    giving the direction of the principal normal sections, and ,
    therefore, of the lines of curvature (notation 1815).
    {pqt - (1 + q°)s} ye + {(1 + p?)t - (1 + q9)r } ya
    + {(1 + p?) s - por } = 0 .
    PROOF. - (i.) The equations of the normals at the consecutive points xyz
    and x + dx, y + dy, z + dz of the surface ¢ (x , y, z ) = 0 are
    & - 2 – 1 - Y = 5 — 2 and { - (x + dx) – - ( y + dy) – $ - (z + dz)
    ቀህ ቀ። - Px + dox Putdo , pet do :
    5811 The condition of intersection is, by (5533),
    dx dy dz 1 Yu P + qyx
    0x Qy 02 = 0, or 1 P -1 = 0,
    | dør døy doel rtsyx sttyx
    798 SOLID GEOMETRY.

    by dividing the first row by dx, and putting zx = 0 & + 0 , Yz, doz = 02 + ry Yo
    & c. The form of o (x , y , z ) being , in this case, o (x , y ) – 2 , 02 becomes – 1 ,
    and do, becomes zero . The determinant equation produces the quadratic.
    (ii.) Otherwise. — Consider En š the point of intersection of consecutive
    normals. The equations of a normal being
    &–2 – 1 - y – 5 or { - x = p (2 - 5) and n - y = 9 (3 - 5) .
    P 9 - 1
    Differentiate both equations for x , considering 5, n, 5 constant and p, q func
    tions of . and y ; the results are
    1 + (r + syz)(z — 5 ) + p ( p + qyr) = 0 and 4x + ( + tyr) (z - 5) + 9 (p + 242) = 0 .
    Eliminate z — s to obtain the quadratic in Yz.
    5812 If the equation of the surface be in the form o (x , y , z ) = 0 , the
    quadratic for Yx may be obtained in the same way. The requisite substitu
    tions in the first determinant are found from 02 + øy9z + Pzza = 0, giving zz ;
    do . = P2x + Pxy Yx + 4.xz7r, & c., and with the notation of (5796 ) the determin .
    ant equation and quadratic for yz becomes
    nyx - (l + my:)
    m 12 = 0.
    | an - gl + (hn - gm ) yz hn - f1 + (bn - fm ) yz gn - cl + (fn - cm ) ya !
    5813 The above determinant, or the corresponding one in (5810 ) , is the
    differential equation of the lines of curvature.

    5814 The radii of curvature of the principal normal sections


    of the surface (x , y , z ) = 0 at a point xyz are given by the
    following quadratic , in which A ' is the bordered determinant
    in (5700 ), and the notation is that of (5796 ) and ( 1620).
    A 'p + {(a + b + c) ( + mº+ n?) – (abcfgh XImn)2} kp - k* = 0 ,
    where l = 72 + mº + n ”.
    PROOF.— The quadratic in (5653) applied to a section of the quadric ( I.)
    (5798) by the plane (II.) , becomes
    AʼR * + { (b + c ) lº + (c + a ) mº+ (a + b ) n° — 2fmn - 2gnl - 2hlm } NR*
    - ( + m + *) N = 0,
    whose roots , being the two values of R ’, are the squares of the principal
    semi-axes of the indicatrix. Put R ' = 0, as in the Proof of (5797).
    5815 Otherwise, the quadratic in (5651) might be applied to a section of
    the quadric (III.) (5802) by the plane ( I.) .
    5816 If the equation of the surface be given in the form
    THEORY OF SURFACES. 799

    z = ° (x ,y), the quadratic becomes [writing, as in (1815),


    P , 9, 1", s, t for %zo 7y 722) xy» Zay ],
    (rt- s*) p - {( 1 + p%) t – 2pqs + (1 + q?) r } kp + k * = 0,
    where l = p? + q2 + 1.
    Otherwise, this equation may be found from the two equations obtained
    in the second proof of (5810 ), by eliminating yz instead of 2 – 5.

    5817 The radius of curyature at a point xyz on the surface


    ♡ (x , y ,z ) = 0 of the normal section whose tangent has the
    direction cosines 1 , M , v is, with the notation of (5796 ) and
    (1620), _ ( + mº + nº)
    "p = (abcfgh Xhuv)?
    Proof. — From equation (I.) (5798 ), sincet, n, s are respectively equal
    to RA, Ry, and Rv.

    5818 The curvature at any point of a surface + (x , y , z ) = 0


    is termed elliptic or synclastic, hyperbolic or anticlastic, and
    parabolic or cylindrical, according as the indicatrix is an
    ellipse, hyperbola , or two parallel right lines, or according as
    the principal curvatures have the same signs, opposite signs,
    or one of them vanishes ; and this will be according as the
    determinant A ', in (5814 ), or se — rt, in (5816 ), is negative,
    positive, or zero .
    PROOF. — The rule follows at once from the consideration that the two
    values of p in the quadratic of (5814 ) must have the same sign in the first
    case, different signs in the second, and that one value must be infinite in the
    third case.

    5819 The condition for an umbilicus is that the indicatrix


    must be a circle ; therefore, either (III.) (5802 ) must be a
    sphere, or, if it be a quadric surface, the plane (II.) must
    make a circular section of it , and therefore either l , m , or n
    must vanish .
    5820 Otherwise, the quadratic in (5814 ) or (5816 ) must
    have equal roots.
    5821 Otherwise ,the conditions for an umbilicus on the sur
    face • (x , y , z) = 0 are the two equations
    bnº + cm - 2fmn cl + an ’ – 2gnl _ am ? + 612 — 2hlm
    m ? tn? na + 12 [ + mº
    800 SOLID GEOMETRY.

    Proof. — The radius of the indicatrix , and therefore also p in (5817), is


    constant for all values of 1, M , v. Now , by (5817) ,
    al? + & c. = * ;
    :: (a- % )1 + (5 - )* + 60- )» + 2juv+2gvd+ 2h14 = 0,
    and l1 + mu tnv = 0 , since Xuv is always tangential, and Imn is normal to
    the surface. As these equations are true for all values of 1 , M , Y, the second
    expression must be a factor of the first . The quotient, by division , is there

    foro (a- " ) * +(0- M)* +6c- M)


    Equating to zero each of the three coefficients of the remainder, and elimi
    nating p , we obtain the above conditions.
    5822 If a common factor of the three fractions in (5821)
    exists, that factor equated to zero is the differential equation
    of a line of spherical curvature at every point of which there
    is an umbilicus. If the fractions are identically equal, the
    surface has an umbilicus at every point, and must therefore
    be a sphere.
    5823 The number of umbilici on a surface of the nth degree
    cannot exceed n ( 10n : — 25n + 16 ) . Salmon , p . 208.

    5824 The condition that the indicatrix may be a rectangular


    hyperbola is
    (a + b + c)(1 + mº + n *) = (abefgh XImn) .
    Proof. — The quadratic in (5814) must have equal roots of opposite
    signs.
    Similarly , when z = q (x ,y) is the equation of the quadric, the condition
    becomes (1 + p +) t - 2pq8 + (1 + 9') r = 0 . (5816 )
    5825 The condition that the indicatrix may become two
    coinciding lines.
    Here equation I. (5798 ) must represent a cone, and the plane (II.)
    must touch it . Hence N = 0 , and , if & be eliminated , the quadratic for the
    ratio & in obtained is
    (an + cf* — 2gnl) ** + 2 (clm - fnl - gmn + hn ?) &n + (bnº + cm2 — 2fmn) n = 0,
    and this must have equal roots .

    CURVATURE OF A SURFACE.
    5826 DEFs.-- Integral curvature of a closed surface is equal
    to the area of that part of the surface of a sphere of unit
    radius which is intercepted by radii drawn parallel to the
    THEORY OF SURFACES. 801

    normals at all points of the given surface. This area also


    measures the solid angle of the cone generated by the radii .
    The curve on the sphere is called the horograph of the curve
    on the original surface. In other words, integral curvature
    of a closed surface is the area of the horograph of its
    boundary
    5827 Average curvature is the integral curvature divided by
    the area of the surface .
    Specific curvature is the average curvature of a small
    element at the point; i.e., last
    P1P2
    = (ds) = የiP2-
    - - -- - - - --
    5828 The last is the usual measure of curvature at a point,
    and its value in coordinates of the point is given by
    1 rt -
    Pip , - - (P + mº+ n?)2 (1 + p : + 93)2 (5796)
    according as ♡ ( , y, z) = 0 or = ° (x, y ) is the form of the
    equation to the surface.
    PROOF. – From the product of roots of the quadratics in (5814 ) and
    (5816 ).
    5829 In a plane curve integral curvature is the plane angle
    contained by the terminal normals , and average curvature is
    the integral curvature divided by the length of the curve.

    5830 Another measure of curvature at a given point of a


    surface is the ratio of the area of the indicatrix to the area of
    the indicatrix cut off by the same plane on a sphere of unit
    radius .which touches the surface internally at the point. This
    measure is = ✓ Papa.
    Proof. — Putting AC = R ,, BC = R ,, in Fig . ( 182 ), and OC = %, the area
    of the indicatrix of the surface is aR , R , at an ellipsoidal point. But
    R = 28 ,3 and R = 20 ,7 , therefore # R , R , = 27 % ✓ (ee ). Also the indicatrix
    of the sphere = 272 since p = P2 = 1 for the sphere.

    5831 The radius of curvature of any normal section at a


    point P of an ellipsoid ( Fig. 184 ) is equal to the square of
    the semi-diameter parallel to the tangent of that section ,
    5 K
    802 SOLID GEOMETRY.

    divided by the perpendicular from P upon the diametral


    plane conjugate to OP.
    Proof. — Let AOB be the plane parallel to the tangent plane at P ;
    0A = d , the semi-diameter in it parallel to the given tangent PT. Draw
    PR perpendicular to OA and PN = p perpendicular to the plane AOB. The
    radius of curvature at P of theelliptic section PA = (4536). Therefore,
    by (5809), the radius of curvature of the normal section through the same
    tangent PT, will be p = d x PR - ď
    De P = PR * PN = D

    5832 The principal radii of curvature at P , viz. Pie pa, are


    found from their sum and product thus : putting y for OP,
    and a , b, c for the semi-axes of the ellipsoid ,
    eto. = a * + bº + c* - . 2. 0. = a *b *c*
    P*
    ProOF. — Let a, 6 be the semi-axes of the section AOB (Fig. 184), then
    a ' + 39 + y = a * + b + c (5642) and paß = abc (5648). By these values
    eliminate a , ß from pi + p, = d' + B * and pipz = app (5831).

    5833 The lines of curvature on a quadric surface are its


    intersections with the confocal quadrics.
    Proof. — Let the quadric and confocal be the ellipsoid and one-fold
    hyperboloid in (Fig. 178 ) intersecting in the line DPE , and let their equa
    tions be, as in (5656),
    + = 1....(i) and on content of =1....(i).
    Atany point P on the line of intersection o, y, z satisfy the three following
    equations :
    First,the differential of (ii.), de + y + z = 0.
    Second,the difference of (i.) and (ii.),
    ar contus + mins to 6* 1) = 0.
    Third, the difference of their differentials,
    də ydy _ t _ 20 % = 0.
    a”(a’ + 1) * 0°(6 + 1) * *(6 + 1) =
    The eliminant of these equations in x, y, z pro | do dy dz /
    duces the determinant equation here annexed , which ,
    by (5811) , is the condition for the intersection of con
    secutive normals . Hence this condition holds for
    every point of the line of intersection of ( i.) and (ii.) .
    THEORY OF SURFACES. 803

    The general method of determining the lines of curvature


    of a surface from the differential equation in (5811) is here
    exemplified in the case of an ellipsoid .
    5834 The determinant just written gives for the differential equation of
    the lines of curvature
    (6 — c*) æ dydz + (c* — a ) y dzdx + (a? — 6°) zdx dy = 0 ......... (i.).
    To solve this, multiply and substitute for z and dz from the equation
    of the quadric. The result is of the form
    Axy yz + ( — Ayº - B ) Y: - xy = 0,
    in which A = COCO, B = a*( * ); or,multiplying by
    A 99=(ayy= — yº) –B44= + (vyys- yº) = 0,
    which is of the form in (3236) . Solving by that method, we find that the
    two equations 992 = a and ayyz - y: = B have the common primitive
    axé- yº = B , which, with the relation Aaß - Ba + = 0, constitutes the
    solution. The result is that the projections of the lines of curvature upon
    the ay plane are a series of conics coaxal with the principal section of the
    ellipsoid , and having their axes a, b varying according to the equation
    a ’ (a² - 6 ) , 62 (B2 — cʻ) = 1 .
    a’(a ? —62) * b’ ( ? – a )
    At an umbilicus y = 0, therefore, equation (i.) becomes [ (62 -- cº) ædz
    + ( a? — ) zd # ] dy = 0. Here dy = 0, being a solution, gives y = 0 = 0,
    showing that the plane of zx contains a line of curvature . The other
    factor, equated to zero , taken with the differential equation of the curve
    c*xdx taʻzdz = 0 , gives the coordinates of the umbilicns, as in (5603 ) .

    OSCULATING PLANE OF A LINE OF CURVATURE .


    5835 Let o be the angle between the osculating plane and
    the normal section through the same line of curvature, ds an
    element of the other line of curvature, and p, p' their radii of
    curvature respectively : then
    tan = dp . pp.

    Proof. — Fig . ( 185) . Let 0A, OB be the lines of curvature ; OP, AP


    consecutive normals along OA ; and OS, BS the same along OB. Also , let
    BQ , CQ be consecutive normals along the line of curvature BC. Then ,
    ultimately , OP = p, OS = p , BQ = p + dp. Also, let QP produced meet the
    oscalating plane of AO in R . Join RO and RA, and draw QN at right angles
    to PS. Since the tangent to A0 at 0 is perpendicular to the plane OBQP
    and that at A to ACQP, it follows that both tangents are perpendicular to
    QP, which must therefore be perpendicular to the osculating plane ARO .
    Hence o or ROP = PQN .
    804 SOLID GEOMETRY.

    Now we= S8 = =p=dę, ::tano = Neamt ultimately.


    5836 At every point on a line of curvature of a central
    conicoid pd is constant, where d is the semi-diameter parallel
    to the tangent at the point and p is the perpendicular from
    the centre upon the tangent plane.
    Proof. — Let the first and third confocals in (5661) be fixed , and there
    fore a , and as constant. Draw the second confocal through the point of
    contact P of the tangent plane (Fig. 178 ). Then , by (5663) , pid , and pad,
    are constant along the line of intersection of the first and third surface ,
    because, by (5661), d = a ; - a > and d = a ; - a

    GEODESIC LINES.
    5837 The equationsofa geodesic on thesurface + (x , y,z) = 0
    are
    PHOTO
    Proof. — The osculating plane of the curve contains the normal to the
    surface (5775 ) ; therefore , by (5737) and (5785 ).
    5838 A geodesic is a line of maximum or minimum distance
    along the surface between two points.
    PROOF. — The curve drawn in the osculating plane from one point to a
    contiguous point is shorter than any other by Meunier 's theorem (5809) ,
    for any oblique section has a shorter radius of curvature and therefore a
    longer arc. A succession ofminimum arcs, however, niay constitute a maxi.
    mum curve distance between the extreme points ; for example, two points on
    a sphere can be joined by either of two arcs of a great circle , the one being
    a minimum and the other a maximum geodesic .
    5839 A surface of revolution such as the terrestrial globe affords a good
    illustration . A meridian and a parallel of latitude drawn through a point
    near the pole are the two lines of curvature at the point. The meridian is
    also a geodesic, but the parallel is evidently not, for its plane does not
    contain the normal to the surface .

    5840 A geodesic is the line in which a string would lie if


    stretched over the convex side of a smooth surface between
    two fixed points .
    Proof. - Any small arc of the string POP' (Fig. 182) is acted upon by
    tensions along the tangents at P and P ', and by the normal reaction of the
    surface at 0 . But these three forces act in the osculating plane (5775 ) ;
    therefore the string will rest in equilibrium on the surface in that plane.
    THEORY OF SURFACES. 805

    CoR. — Two equal geodesics drawn from a point and in


    definitely near to each other are at right angles to the line
    which joins their extremities.

    5841 If a geodesic has a constant inclination to a fixed line,


    the normals along it will be at right angles to that line.
    PROOF. — Let Imn be the fixed line and a the constant angle; then
    lx , + mye + nz, = cos a, and therefore lxs8 +myzo + nz2, = 0.
    Therefore, by (5837) , the principal normal is at right angles to lmn.
    EXAMPLE. — The helix, the axis being the fixed line.
    5842 On any central conicoid pd is constant along a geo
    desic, where p is the perpendicular from the centre upon the
    tangent plane and d is the semi-diameter parallel to the
    tangent of the geodesic.
    PROOF. — ( Fig . 186 .) Let AT, BT be the tangents at the two extremities
    of a small geodesic arc AB, and let the tangent planes at A and B be ADC
    and BCD. AT and BT make equal angles with CD , by the property of
    shortest distance, for if the plane BCD be turned about CD until it coincides
    with the plane ADC, ATB will become a straight line, and therefore
    LATD = BTC = i, say.
    Let w be the angle between the tangent planes ; let the perpendiculars
    upon those planes from A , B be AM = 9 , BN = q', and from the centre of the
    quadric p , p '; and let xyz and a'y'z' be the points A , B . Then
    g = ATsin i sinw, q' = BTsin i sin w , :: 9 : ' = AT : BT...... (i.),
    q=p -1) (5628); d'= p o t + + 1),
    therefore q : q' = p ' : p ................................. (ii.).
    Again , let d , d be the semi-diameters parallel to AT and BT. Then , by
    (5677), AT : BT = d : d ' ; therefore p ' : p = d : ď or pd = p'd ' ; that is,
    pd is constant.

    5843 If a line of curvature be plane, that plane makes a


    constant angle with the tangent plane to the surface.
    PROOF. — Let PQ , QR, RS be equal consecutive elements of the line of
    curvature. The consecutive normals to the surface bisect PQ and QR and
    meet in a point. Therefore they are equally inclined to the plane PQR .
    Similarly the second and third normals are equally inclined to the plane QRS,
    and so on . Hence, if the curve be plane, all the normals are equally inclined
    to its plane. Hence also the following theorem .

    5844 Lancret's Theorem . — The variation in the angle be


    tween the tangent plane and the osculating plane of a line of
    806 SOLID GEOMETRY.

    curvature is equal to the angle between consecutive osculating


    planes.
    5845 CoR . - If a geodesic be either a line of curvature or a
    plane curve , it is both , but a plane line of curvature , as in
    (5839) , is not necessarily a geodesic .

    GEODESIC CURVATURE.
    Theorem . — The square of the curvature at any point of a
    curve traced on a surface is equal to the sum of the squares
    of the normal and geodesic curvatures (5776 ), or
    1 1 1
    5846

    where p ' is the radius of curvature of the normal section and


    p" the radius of geodesic curvature. Also, if o be the angle
    between the plane of normal section and the osculating plane,
    5847 p = p" sin $ = p cost.
    Proof. — Let PQ = QR (Fig. 187) be consecutive elements of any curve
    traced on a surface. Produce PQ to S ,making QS = PQ . Let QT = PQ
    be the consecutive elements of the section of the surface drawn through
    PQS and the normal at Q . Join RS, ST, TR . PQSR is the osculating
    plane of the curve PQR . PQST is the plane of normal section , and there
    fore PQT is a geodesic . QRT is the tangent plane, and STR is a right
    angle.
    Then , putting SQR = dų, SQT = dt', RQT = d4" , RST = Q , we have
    (5776 )

    Therefore 1 = ds.dy ” = RT = sin %.


    - ds .dľ' _ ST
    Also p ds.dų = SR = cos ® , as in (5809 ).
    Thus both theorems are proved . Note that p' is the radius of curvature of
    the geodesic PQT, while p is the radius of geodesic curvature of PQR .

    RADIUS OF TORSION OF A GEODESIC .


    5848 If be the angle between the geodesic and one of the
    lines of curvature ; pu p2 the principal radii of normal curva
    ture, and o the radius of torsion ,
    11 1
    - - sin cos 0.
    = (4P .)sin &cose
    P2 !
    THEORY OF SURFACES. 802

    ProoF. — ( Fig . 182.) Let OP = ds be the geodesic , OA, OB the lines of


    curvature, and 0 = ACP . The angle of torsion dr measures the rotation of
    the normal to the surface round OP = ds. But this angle is equal to the
    sum of the rotations of the normal round OA and OB resolved along ds.
    For, in travelling along each of the lines CN and NP, which are in the direc
    tions of the lines of curvature, the normal rotates only about the other.
    Therefore, if w1, w , be the rotations round 0A, OB, dt = w cos 0 + w , sin 0.
    But w . = _ ds sin 0, w . = dscos 0 : : 1 = dr = ( 1 - 1 ).sin 0 cos 0.
    po o ds lpi Pe

    5849 The product pd has the same value for all geodesics
    which touch the same line of curvature.
    PROOF. - By theorems (5836 ) and (5842), since the product where they
    touch it must be the sameas that for the line of curvature.
    5850 The product pd has the same value for all geodesics
    drawn through any umbilicus on a conicoid .
    PROOF. — The semi-diameter d , in this case, is the radius of a circular
    section, and therefore equal to the mean semi-axis b for all the geodesics ;
    and p is the same for all.

    5851 The geodesics drawn through any point on a conicoid


    to two umbilici make equal angles with either line of curva
    ture through the point.
    Proof. — pd is the same for each geodesic, by the last, and p is the same
    for each ; therefore d is the same, that is, the diameters parallel to the two
    geodesics at the point are equal ; therefore they are equally inclined to each
    axis of their section ; but these axes are parallel to the lines of curvature
    (5803) ; therefore , & c.

    5852 Hence the geodesics joining any point to two opposite


    umbilici lying on the same diameter are continuations of each
    other.
    5853 The sum of the distances of any point on a line of
    curvature from two interior umbilici is constant ; and the
    difference of the distances from one interior and one exterior
    umbilicus is constant.
    PROOF. — Geometrically , as in the analogous theorem for the focal distances
    in a conic, if r , r' are the distances and 7 + dr, p' + dr' the distances for a
    consecutive point on the line of curvature, it follows froin (5851) that
    dr = - dr' for interior ombilici and dr = dr' for exterior ones .
    5854 A system of lines of curvature and the umbilici on a
    808 SOLID GEOMETRY.

    quadric surface has therefore analogous properties with a


    system of confocal conics and their foci in a plane, the geo
    desics corresponding to straight lines.
    5855 In the same way , every surface has a geodesic geo
    metry proper to itself ; spherical trigonometry , for instance,
    being the geodesic geometry of the sphere.

    INVARIANTS.

    INVARIANTS OF A SINGLE FUNCTION .


    5856 The constancy of the ratio RP : p in equation (5798)
    gives rise to the following invariant forms. Since the quadric
    surface I and the tangent plane II are the same for all posi
    tions of the coordinate axes , they have been called respec
    tively the invariable quadric and the invariable plane. As a
    consequence,
    5857 $* + $ + $ .
    is an invariant of $ (x , y, z ).
    Proof. — By (5791), since the perpendicular from the origin upon the
    invariable plane is constant. Also, the coefficients of the discriminating
    cubic (5693) of the invariable quadric will not be altered by transformation
    of axes. Therefore the following are also invariant forms :
    5858 $ 2& + *24 + $ 221
    5859 $2,022 + % 22 2x + $ 27 $2, - 42- 46- piye
    5860 $ 2: $2,02z + 26 ,26ax Pay - $2: $ : $2,022 022 *y.
    5861 A similar theorem applied to a function + (x , y) of
    two variables gives the invariable conic and invariable line ;
    namely,
    & $ 2x + 2&ndry + n*$2, = 1 and $6e + nd, = 1;
    and from these the invariants,
    5863 02 + 45, 24 + $20, $ 2:$20 - y OYS

    5866 x0, - yda, x¢, + y& c.


    QUADRATURE AND CUBATURE. 809

    Proof. — The last two invariants are obtained from the cosine of the
    angle between the invariable line (5862) and the fixed line y - an = 0 ,
    joining the point xy with the origin , or the fixed line < + yn = 0 .

    INVARIANTS OF TWO FUNCTIONS.


    5868 An invariant of the two functions q (x , y) , 4 (x , y ) is
    DuVx + 0 , 47
    PROOF. — Form the cosine of the angle between the invariable lines
    &qz + ndy = 1 and £4x + ny , = 1, observing (5863) .

    Also the two following expressions are invariants,


    5869 $ 2,422 + $24482, - 20.cy,4 xya
    5870 02x42x + $2, 420 + 2°ry,Vizyo
    Proof. — From the invariable couics (5861) of Q and y,we get
    (023 + 1422) & + 2 (0xy + 14xy) En + ( 2y + 14.24) m?
    invariable for any value of . Hence the coefficients of the several powers
    of 1 in the invariant
    (028 + 14 2z) (02y + 14:2y) – (Pry + 14xy)
    are also invariants. This gives (5869) . Subtracting the latter from the
    invariant (02c + 92y) (42x + 42y) produces (5870) .

    INTEGRALS FOR VOLUMES AND SURFACES.

    5871 If V be the volume included between the surface


    % = ° ( , y ) , three rectangular coordinate planes, the cylindri
    cal surface y = % (w ), and the plane x = a , Fig. of (1906 )
    5872 V = | | | dx dy dz = | | zdx dy.

    For the limits and demonstration, see (1906 ).

    5874 The area of the surface q (x,y,z) = 0 or z = f(x,y)


    will be

    s= {S28 + )dvdy5 orL s= fSv1 +4+4)dvdy.


    810 SOLID GEOMETRY.

    Proof. — The area of the element whose projection is dwdy will be


    dx dy sec y , where y is its inclination to the plane of xy, and therefore the
    angle between the normal and the z axis. Therefore
    sec y = ($ + pe + ox) = . = ✓ (1 + z + z;), by (1708 ).

    5875 Let the equation of a surface APB (Fig. 188) in polar


    coordinates be r = f (0, o), and let V be the volume of the
    sector contained by the planes AOB, AOP, including an angle
    = PHC, the given surface APB, and the portion OPB of
    the surface of a right cone whose vertex is 0 , axis 0A, and
    semi- vertical angle 0 = AOB or AOP ; then

    V = 317* sin 0 dodø.


    PROOF. — Through P , any point on the surface, describe a spherical sur
    face PCD , with centre 0 and radius r = OP. The volume of the elemental
    pyramid , vertex 0 , base Pe, = r . Pf. Pg = tr . rdo . r sin dø . Here the
    error of the small portions, like PE, ultimately disappears in the summation,
    since the volume of PE , being equal to zdr .rdo .r sin odo , is of the third
    order of small quantities ; and so in similar instances.

    5876 The area of the same surface APB (Fig . 188) is


    A OD

    S = " rv {(o® + r )sin?6 + r;} dºdø .


    Jo Jo

    Proof. — Let the perpendicular from 0 upon the tangent plane at P to


    the given surface be ON = p . The element of
    area PE = area Pe. = rdo . r sin 0do . - = ! 3 do do .

    Substitute the value of p in (5793).

    SURFACE OF REVOLUTION .
    If y = f (x ) (Fig . 90) be the generating curve, and the x
    axis the axis of revolution , V the volume, and S the surface
    included between the planes x = a , x = b ;

    5877 V = Swy dx, S = S%2wy / (1 +y:)dv.


    PROOF. — The volume of the elemental cylinder of radius y and height dæ
    is tyødx. The element of the surface of revolution is
    2tyds = 2ty sdx = 2ty ✓ (1 + y2) dx, (5113 )
    QUADRATURE AND CUBATURE . 811

    Guldin 's Rules . When the generating curve of a surface


    of revolution is a closed curve, and does not cut the axis of
    revolution , a solid annulus, or ring, is formed .
    5879 RULE I. - The volume of the solid ring is equal to the
    area of the generating curve multiplied by the circumference of
    the circle described by the centroid * of the area .
    5880 RULE II. — The surface of the ring is equal to the
    perimeter of the generating curve multiplied by the circum
    ference described by the centroid of the perimeter .
    PROOF. — Let A be the area of the closed curve, and dA any element of A
    at a distance y from the axis of revolution. The volume generated
    = |2wydA = 24 ydA = 2wųA,
    by the definition of the centroid (5885) , ġ being its distance from the axis.
    Similarly, if P be the perimeter, writing P instead of A .
    Quadrature of surfaces bounded by lines of constant
    gradient.
    5881 Defining the curve ( y) as the locus of a point on the
    given surface at which the normal has the constant inclina
    tion y to the z axis ; let F (y be the projection of the area
    bounded by the curve ( y ) upon the xy plane ; then the area
    itself will be found from the formula ,
    S = ('sec y F" (y) dy.
    10
    PROOF. -- The element of area between two consecutive curves (w ) and
    (y + dy) projected on the wy plane will be dF (~ ) = F ' ( v ) dy ; and, since the
    slope is the samethroughout the curve (y ), this projected element must be
    equal to the corresponding element of the surface multiplied by cos y .
    5882 RULE. — Equate coefficients of the equation of the
    tangent plane with those of 18 + mn + nt = p , and eliminate 1
    and m from 1? + mº + n ” = 1. The result will be an equation in
    x , y and n = cos y, representing the projection of the curve ( y)
    upon the xy plane. From this F ( ) must be found .
    5883 Ex.— Taking the elliptic paraboloid ** + * = 22 ; the tangent
    plane at æyz is signa + - 5 = . Equating coefficients of the last with
    15 + mn + n5 = p, and substituting for Iand m in 1° +mº +n° = 1, we obtain
    for the projection on the xy plane, a + = tanʼy. The area of this ellipse
    * Centre of mass, or gravity .
    812 SOLID GEOMETRY.

    is F (y) = rabtan” y, and therefore F" (Y) = 2nab tan y sec y. Consequently ,
    by (5881), S = 2tab tan y secºydy = {rab (secºy - 1).

    CENTRE OF MASS.

    5884 DEFINITIONS. — The moment of a body with respect to a


    plane is the sum of the products of each element of mass of
    the body and the distance of the element from the plane.
    5885 The distance (denoted by 2 ) of the centre of mass *
    from the same plane is equal to the moment of the body
    divided by its mass .
    5886 Note . - If the body be of uniform density, as is supposed to be the
    case in all the following examples, assume unity for the density , and read
    volume instead of mass in the above definitions.

    The definition gives the following formulæ for the position


    of the centre of mass of a uniform body :
    5887 For a plane curve,
    X =
    Sæds Sx (1 + yx )dx _ ſr cos0 (72 + r )de2 . (5116)
    Sds - 1 / (1 + y:)dx = SV (p?+ r )do
    For y, change x into y and cos 0 into sin 0 ; but observe that in all cases,
    if the body be symmetrical about the axis of a , y vanishes. The formula
    gives the centre of volume of the portion of the curve included between the
    limits of integration .

    For a plane area ,


    R $ xdxdy sxydx
    5890
    SJ dxdySydx
    The area is bounded by the curve, the x axis, and the ordinates 2 = a ,
    x = b , if such be the limits of integration .

    For a plane sectorial area bounded by two radii SP = r ,


    SP' = q (Fig . 28 ) and the curve r = F (0) ;
    * Also called centre of grarity or inertia, and more recently centroid.
    CENTRE OF MASS. 813

    _ 35 poi cos 0d0dr S18 cos Odo


    5892
    Srd@ dr . Sredo
    5894 S ū = SS
    JJ sin dødr 35 po8 sin 0do
    SSrdodr Ś rodo
    The second forms for žē and y give the centroid of an area like SPP'
    (Fig . 28 ). The double integrals applied to that figure require the limits of
    the integration for p to be from 0 to F (0 ), and afterwards for 0, from
    0 , = ASP to 0, = ASP'. But, if applied to the area in (Fig . 109) , the order
    of integration must be reversed, as explained in (5209) .

    For a surface of revolution round the æ axis,


    KOOK - _ J xy ✓ ( 1 + y ) dx _ Spo? sin 0 cos 0 ✓ (702 + r ) do
    Sy V (1 + y:)dx fr sin / (n® + r ) do
    PROOF .— By (5885), for the moment = .2tyds and the area = 2tyds ;
    the second form by (5116 ). If x = a , x = b are the limits of integration, the
    surface is bounded by the parallel planes x = a , x = b ; and in the second
    form , the corresponding values of 0 are the limits detining the same parallel
    planes.
    For any surface,
    5898 - SS x ✓ ( 1 + x + x ) dx dy (5874)
    SS / ( 1 + x + x ) dx dy
    For a solid of revolution round the x axis,
    Gegg _ Sxy’dx _ SS sin 0 cos6d0dr
    Sy'dx – Sý po sin 0dodr
    Proof. — By (5885), for the moment = 1 x . ydx and the volume
    = 1 Ty'dx. The limits as in (5896 ).

    5901 For any solid figure bounded as described in (5871),


    the coordinates of the centroid are given by

    Vo = SSS«dx dydz = Sfazdædy,


    Vğ = SSS ydxdyda = SS yzdvdy,
    Va = SSS xdxdydx = $ SS= dwdy,
    814 SOLID GEOMETRY.

    where V = dx dy dz = | | zdxdy,
    as in (5872– 3 ), and the limits are as defined in ( 1906 ).
    5902 For thewedge shaped solid (OAPB, Fig. 188) defined
    by the polar coordinates r , 0, 0 , as in (5875) ,
    Væ = frasin’0 cos pdøde,
    Vy = ffr*sinº8 sin $døde,
    Vz = r sin 0 cos 0dod¢ ,
    where V = } $\ r*sin 0dodo.
    Proof.-- By (5875 ) ; multiplying the elementary pyramid kosin dod
    separately by the distances of its centroid from the coordinate planes ; viz .,
    är sin 8 cos , r sin 6 sin o , and r cos 0.

    MOMENTS AND PRODUCTS OF INERTIA.

    5903 DEFINITIONS. — The moment of inertia of a body about


    a given right line or axis is the sum of the products of each
    element of mass and the square of its distance from the line.
    5904 The square of the radius of gyration of the body about
    the given line is equal to the moment of inertia of the body
    divided by its mass .
    5905 The moment of inertia of a body with respect to a
    plane or point is the sum of the products of each element of
    ma
    mass and the square of its distance from the plane or point.
    5906 The product of inertia of a body with respect to two
    rectangular coordinate planes is the sum of the products of
    each element of mass and its distances from the two planes.

    5907 Let A , B , C be the moments of inertia of a body


    about three rectangular axes ; A ', B ', C the moments of
    inertia with respect to the three planes of yz, zx , and xy ; and
    MOMENTS OF INERTIA. 815

    F , G , H the products of inertia with respect to the second


    and third planes , the third and first , and the first and second
    respectively. F , G , H are more frequently called the products
    of inertia about the axes of yz , zx , and xy respectively.
    By the definitions we have the values
    5908 A = $ m ( + 8°), F = 2mg/%,
    B = 2m ( sº + zº), G = Emzx,
    C = km(x + y), H = $ mag.
    5914 A' = Emx* = S - A ) where S = A + B + C .
    B’ = Emyº = S - B } > m (x + y + zº),
    C' = Emz' = S - C ) = Emr .

    5920 Theorem I. — The M . I. of a lamina about an axis per


    pendicular to its plane is equal to the sum of the two M . I.
    about any two axes in its plane drawn through the foot of the
    perpendicular axis and at right angles to each other .
    PROOF. — By the definition (5903), and Euc. I. 47.
    5921 Theorem II. - The M . I. of a body about a given axis,
    plane, or point is equal to the M . I . about a parallel axis or
    plane through the centroid , or about the centroid itself
    respectively , plus the M . I. of the whole mass, supposed col.
    lected at the centroid , about the given axis, plane, or point.
    Proof. — In the figure, p . 168, let the given axis be perpendicular to the
    paper at B ; let A be the centroid , and m an element ofmass at C ; then, for
    every thin section of the solid parallel to the paper ,
    M . I. = { m . BC = { m (AC + AB* — 2AB.AD )
    = { m .40* + Em . AB - 2AB . Em .AD.
    But Am . AD = 0, by (5885), since A is the centroid of the body, which
    proves the proposition. Similarly for the plane or point.

    Cor . 1. — Hence, if the M . I. about any axis is known , that


    about any parallel axis can be found without integration . For,
    let I, be the M . I, about a given axis , whose distance from
    the centroid is a , and let I, be the required M . I. about an
    axis whose distance from the centroid is b ; then , by Theorem
    . II., 1, = 1, - m (a? — bº).
    816 SOLID GEOMETRY.

    Cor. II. — The M . I. has the same value for all parallel
    axes at the same distance from the centroid .

    5922 Theorem III. - The product of inertia for two assigned


    axes is equal to the product for two parallel axes through the
    centroid of the body plus the product taken for the whole
    mass collected at the centroid with respect to the assigned
    axes .
    útý be the cos being the code with respect
    PROOF. — Let x = i + ', y = ý tý' be the coordinates of an element of the
    body with respect to the assigned axes ; ž , y being the coordinates of the
    centroid , and x ', y ' the coordinates of the same element with respect to
    parallel axes through the centroid , all axes being parallel to z. Then
    Em xy = Em (ső + a ) (y + y ') = cy & m + Ema'y ' + Emy' +ýmz'
    = žy Em + Ema'y'.
    Since Ema' and Amy' vanish by the definition of the centroid .
    5923 The M . I. of a body with respect to a point is equal
    to the M . I. for any plane through the point plus the M . I.
    about the normal to the plane through the point.
    Proof. – For the origin and yz plane,
    Emx* + &m (yº + z+) = Emr?. (5908,'14 ,'19)
    5924 Given the moments and products of inertia , A , B , C ,
    F , G , H , as above , about three rectangular axes, the moment
    of inertia of the body about a line through the origin , whose
    direction cosines are l, m , n , will be
    I = Al? + Bm + Cn ' — 2Fmn + 2Gnl + 2Hlm .
    Proof. — (Fig . 11.) Let xyz be a point P of the body, OM the line Imn ,
    and PM the perpendicular upon it. Then the M . I. about OM
    = Em (OP - OM ?) = Em { (re? + y + z+){1* + m² + n *) - (læ + my + nz) } (5530)
    producing the above result, by (5908– 13 ).

    ELLIPSOIDS OF INERTIA .
    5925 The equation of the Momental Ellipsoid is
    Ax2 + By + C%8 — 2Fyz – 2Gzx — 2Hxy = Me,
    obtained by putting Im2 = M . M being themass of the body,
    and &* a constant to make the equation homogeneous. Hence the
    square of the radius of the momental ellipsoid for any point varies
    inversely as the moment of inertia of the body about thatradius.
    MOMENTS OF INERTIA . 817

    ne

    5926 If the products of inertia vanish, the axes are calledaxes a

    the principal axes of the body.


    5927 At every point of a body there are always three
    principal rectangular axes.
    PROOF. — These are evidently the principal axes of the momental ellipsoid
    of the point ; for if the coordinate axes be made to coincide with the former,
    F , G , I will vanish.

    5928 The equation of the momental ellipsoid referred to its


    principal axes will be
    Ax2 + By' + Cz = Me“.
    5929 The moment of inertia about a line Imn will now be
    I = Al + Bm ? + Cn?.
    THE ELLIPSOID OF GYRATION .
    5930 The equation of the Ellipsoid of Gyration referred to
    principal axes is

    + +*= 1
    It is the reciprocal surface of the momental ellipsoid (5719),
    and its property is
    5931 The moment of inertia about the perpendicular from
    the origin upon the tangent plane varies as the square of the
    perpendicular.

    5932 For any other rectangular axes through the point, the
    equation of the ellipsoid of gyration is, by (5717),
    A - H –G X 1 = 0 , being the reciprocal surface of
    the momental ellipsoid ,
    -G - F C L ( A , B , C , - F , - G , - HXxyz )?
    1 = M,
    % Ñ | with the radius of the sphere of
    reciprocation = 1 . The equation when expanded becomes
    | A HG
    5933 (BC - F2) x2 + ...+ 2(FG + CH) xy = H B F |
    G F C
    5 M
    818 SOLID GEOMETRY.

    LEGENDRE ' S EQUI-MOMENTAL ELLIPSOID .

    5934 The equation is

    * + + *=
    with the values in (5914 ).
    5935 The mass of this ellipsoid is taken equal to that of
    the body , and it has the same principal moments of inertia .

    THE MOMENTAL ELLIPSOID FOR A PLANE.


    5936 If A ', B ', C ' be the moments of inertia for the three
    coordinate planes, as in (5914), the M . I. for a plane through
    the origin whose dir. cos. are l, m , n , will be
    I' = A 'T + B ’mº + C 'n ’ + 2Fmn + 2Gnl+ 2Hlm
    Proof : I' = Em (lx + my + nz)? = mx”. 1° + & c. = A'T? + & c.

    5937 The momental ellipsoid for this plane will be


    A ’x ? + B 'y' + C'm ? + 2Fyz + 2Gxx + 2Hxy = Me,
    and its property is
    5938 The M . I. for any plane passing through the centre of
    the ellipsoid is equal to the inverse square of the radius per
    pendicular to the plane.

    5939 If zobe a radius of this ellipsoid , and a , b , c its semi


    axes, the M . I. about a

    +*+ +
    Proof. --- (Fig . 11.) M . I. about r, plus M . I. for the plane OM perpen
    dicular to ru
    = {mOP = {mx*+ Emj + Emz* = * + + , by (5938).

    EQUI-MOMENTAL CONE.
    5940 The equation of the equi-momental cone at any point
    of a body, referred to principal axes of the body at the point,
    MOMENTS OF INERTIA . 819

    is (A - 1) x2 + ( B - 1) y ? + ( C - 1) = = 0 ,
    its property being that
    5941 The generating line passes through the given point, and
    moves so that the M . I. about it is a constant = 1.
    PROOF. — Let Imn be the generating line in one position, then
    Al + Bm ? + Cn = I (1? + m² + n°). Therefore, & c.

    5942 Theorem . - If two systems have the same mass, the


    same centroid , principal axes and principalmoments of inertia
    at the centroid , they have equal moments of inertia about any
    right line whatever , and are termed equi-momental. By (5906 )
    and (5929).

    5943 If two bodies are equi-momental, their projections are


    equi-momental.
    PROOF. - If the projection be from the xy plane in the ratio 1 : n , the
    coordinates 2 , 4 , % of a particle become x , y , nz, and the mass m becomes nm .
    The conditions in (5942 ) will then be fulfilled .

    MOMENT OF INERTIA OF A TRIANGLE .


    5944 The M . I. of a triangle ABD (Fig . 190) about a side
    BD , distant p from the opposite vertex A , is

    1 = mpus
    p
    PROOF. — Let BD = a and EF = Y ; I = 1 2 Im ?

    5945 The M . I. of a triangle ABC (Fig . 190) about a


    straight line BD passing through a vertex B , and distant p
    and q from the vertices A and C , is
    I = mpP + pa + q2

    Proof. — By (5944), taking difference of M . I. of the triangles ABD,CBD.


    5946 The M . I. of a triangle ABC about an axis through
    its centroid parallel to BD, is
    1

    1 = mpangti?
    18
    By (5921)
    820 SOLID GEOMETRY.

    5947 Cor. - If the triangle be isosceles, so that p = 9 , the


    last two moments of inertia become

    mps and maps


    18

    5949 The M . I. of the triangle about axes perpendicular to


    ABC through B and through the centroid , respectively , are
    m 3(°+1293)– and m (a +b36 +c ). (5920)
    5951 The M . I. about GF of the trapezoid ACGF (Fig .
    190 ) , is
    mpte
    5952 The moments and products of inertia of a triangle
    about any axes are the same for three equal particles, each
    one-third of the mass of the triangle , placed at the mid -points
    of its sides.
    Proof. — (Fig. 190.) The M . I. of the three particles at the mid -points
    of AB, BC, CA abont BD, any line through a vertex, will be
    {( 9 + + },
    which is equal to that of the triangle, by (5945).

    MOMENTAL ELLIPSE .
    5953 If a , ß be the radii of gyration of a plane area to
    principal axes Ox, Oy, where 0 is the centroid , the equation
    of the momental ellipse for the point 0 will be
    a’r? + Byº = 2a’B .
    5954 Also the area is equi-momental with three equal
    particles , each one one-third of its mass placed anywhere on
    the ellipse so that O may be their centroid .
    Proof. — Let « y, a 'y', a "y " be the coordinates of three equi-momental
    particles : then
    . (x2 + x '? + a " ) = mp? ; 5 (y + y'? + y'' ) = maʼ; xy + a'y' + x " y " = 0 ;
    MOMENTS OF INERTIA. 821

    and the two systems have the same centroid ; therefore


    2 + x ' + " = 0 and y' + y " + y ''' = 0 .
    Eliminating a ", y', xe" , y " between the five equations, we find the equation of
    (5953) for the locus of xy.

    5955 The momental ellipse for the centroid of a triangle is


    the inscribed ellipse touching the sides at their mid -points .
    Proof. — (Fig . 189.) The inscribed ellipse,which touches two sidesat their
    mid -points, also touches the third side at its mid-point, by Carnot's theorem
    (4779) . Now DF is parallel to AC , the tangent at E ; therefore BE, which
    bisects DF, passes through the centre 0 of the ellipse . Similarly , AD
    passes through it ; therefore 0 is the centroid of the triangle.
    Let OE = a ', and let b ' be the semi-diameter parallel to AC ; then
    ON ?
    + FN
    14 : = 1. But ON = , therefore FN : = 362. The M . I. about
    a 62
    OE, by (5954 ), = { m }b'? sin' w = mo , where a , b are the semi-axes.
    Hence the M . I. about OD, OE, OF varies inversely as the squares of those
    lines, and therefore the ellipse in the diagram is a momental ellipse, since it
    has six points which fulfil the requirements .
    5956 The projections of a plane area and its momental
    ellipse form another plane area and its momental ellipse. (5943)

    5957 The M . I. of a tetrahedron ABCD about any plane


    through A is
    no (a* + B + y + By + ya + aß),
    where a, ß, y are the perpendiculars on the plane from B, C ,D .
    5958 The tetrahedron is also equi-momental with four
    particles, each one - twentieth of the mass, placed at the
    vertices, and a particle equal to the remaining mass placed at
    the centroid (5942).
    5959 The equi-momental ellipsoid of a tetrahedron has the
    same centroid , and touches each edge at its middle point.
    PROOF. — By projecting a regular tetrahedron and its equi-momental
    sphere (for the centroid ) of radius = 73 x radius of inscribed sphere .

    5960 To find the point, if it exists , in a given right line at


    which the line is a principal axis, and to find the other prin
    cipal axes at the point.
    822 SOLID GEOMETRY.

    Let O be a datum point in the line. Take this for origin ,


    the given line for axis of z , and OX, OY for the other axes.
    Then , if h be the distance from 0 to the required point 0 ,
    and 0 the angle between OX and the principal axis O ' X ' ,
    Emyz ZM N and tan 20 = T 2H- B '
    5961 h = Σmm

    where A , B , H are the moments and product of inertia about


    OX , OY.
    PROOF. — At the point 0 , 0 , h , Em ( z - h ) x = Em ( z - h ) y = 0 , from
    which h is found ; and the equation for 6 is that for determining the prin
    cipal axes of the elliptic section of the momental ellipsoid , whose equation
    is Ax' + 2Hxy + By? = Me*, as in (4408 ).

    5964 The equality of the two ratios in (5961) is the condi


    tion that the z axis should be a principal axis at some point of
    its length .
    5965 If an axis be a principal axis at more than one point
    of its length , it passes through the centroid of the system ;
    and , conversely , if it be a principal axis at the centroid , it is
    so at every point of its length .
    Proof. - For h must be indeterminate in (5961). Therefore Emyz = 0 ,
    Emy = 0 , Emz = 0 , Emx = 0 .

    5966 The principal axes OʻX', OʻY' are parallel to the


    principal axes of the projection of the body in the original
    plane of xy. By (5962– 3).
    5967 Given the principal axes of a body at its centroid , to
    find the principal axes and moments of inertia at any point in
    the principal plane of xy.
    Let C in the Figure of ( 1171) be the centroid , CX , CY
    principal axes, A , B the M . I. about them , and P the given
    point. Find two points S, S', called foci of inertia ,such that
    the X and Y moments of inertia there are equal, and therefore
    B + m .CS? = A ; giving CS = CS = VA m- B .... (i.).
    ( 1. ) .

    The internal and external bisectors of the angle SPS' will be


    two of the principal axes at P , and the third will be the normal
    to the plane.
    MOMENTS OF INERTIA . 823

    Proof. — The X and Y principalmoments being equal at S , the moment


    about every line through S in this plane is the same. [ For I = Al2 + Bm ?
    + Cn² and n = 0 and A = B , therefore I = A .] Therefore the moments
    about SP and S ' P are equal. Therefore the bisectors PT, PG of the angles
    at P will be principal axes.

    5968 Let SY, S ' Y ' be the perpendiculars on PT, and SZ ,


    S'Z' those upon PG ; then the M . I. about PT and PG will be
    respectively ,
    A + SY.sY = B + m ( SP + SP )?
    2
    ).

    A -mSZ . S'Z = B + m (SP5SP)


    I 2 /
    .
    PROOF. — Draw CR perpendicular to SY. The M . I. about CR (0 = SCR )
    = A cos + B sin (5929) = A - ( A - B ) sin ’0
    = A - m CS? sina 6 (by i.) = A –mSRP.
    Therefore M . I. about PT = A - m SR² + m RY (5921)
    = A + m (RY + SR ) (RY - SR) = A + mSY. S ' Y '
    ( SP + S' P ) ?
    = A + m BC? (1178 ) = B +mAC (by i.) = B + m 2 1 :
    Similarly for the M . I. about PG .

    5969 Hence , if an ellipse or hyperbola be described with


    S , S ' for foci, the tangent and normal at any point of the
    curve are principal axes, and the M . I. about either is constant
    for that curve.
    5970 Similarly , for a point P in any plane through the
    centroid 0 , it may be shown that the same construction will
    give the axes PT, PG about which the product of inertia
    vanishes, OX , OY being the axes at 0 in the given plane about
    which the product of inertia vanishes.

    5971 The condition for the existence of a point in a body


    at which the M . I. about every axis through it shall be the
    same, is
    There must be two principal axes of equal moment at the
    centroid , and the M . I. about each must be less than the third
    principal moment.
    Two such points will then exist situated on the axis of
    unequal moment, and equi-distant from the centroid .
    824 SOLID GEOMETRY.

    5972 Given the principal axes at the centroid of a body and


    the moments of inertia about them , to find the principal axes
    and moments at any other point.
    [ See (5975) for the result.]
    Let A, B , C be the given principalmoments, and let the
    mass of the body be unity. Then the ellipsoid of gyration at
    the centroid 0 , and a quadric confocal with it, will be
    2
    1 and
    * + 3 + == A + I = 1.

    5973 PROP. I. — The M . I. is constant for all tangent planes


    of the confocal, and is equal to the
    M . I. for the origin 01
    0 + 1 =- St
    S + 1.. (5919 )
    PROOF . — Let l, m , n be the dir. cos. of the tangent plane of the confocal,
    p the perpendicular on the plane from 0 . The M . I. for this plane
    = M . I. for a parallel plane through 0 + pº (5921)
    = A 'l + B ’mº+ C'n ? + p (5936)
    = ( S - A ) 1 + ( S - B ) m ’ + ( 8 - C ) n ’ + ( A + 1 ) 72 + ( B + 1 ) m ’ + ( C + 1 ) na
    (5914, 5631) = 8 + 1, which is independent of l, m , n .

    5974 Prop. II. — All these planes are principal planes at


    their points of contact, and if the three confocals be drawn
    through any point P , the tangent planes at P to the confocal
    ellipsoid , two-fold hyperboloid , and one-fold hyperboloid , are
    respectively the principal planes of greatest, least, and mean
    moments of inertia . The normal to the confocal ellipsoid is
    the axis of least moment, and the normal to the two -fold
    hyperboloid is the axis of greatest moment.
    Proof. — Draw any other plane through P . The perpendicular on it
    from 0 is less than the perpendicular on the parallel tangent plane to the
    confocal ellipse, and greater in the case of the two- fold hyperbola . Then ,
    by (5921) .
    The solution of the problem at (5972) is now given by Proposition III .

    5975 PROP. III. — The principal moments of inertia at P are


    OP2 - 17, OP2 - 12, OP2 – 13, and the normals to the three con
    focals at P are the principal axes.
    Proof. — The M . I. about the x axis at P
    = M . I. for the origin P - M . I. for the yz plane
    = S + 0P9 - ( + 1,) = 0P - 1, (5921- 73) .
    MOMENTS OF INERTIA . 825

    5976 The principal moments of inertia above, expressed in


    terms of , of the confocal ellipsoid and dz, dz, its principal
    me
    semi-diameters conjugate to OP, will, by (5661), become
    OP - 1, 0P - 1, + d3, 0P2 - td .

    5977 The condition that the line abc , Imn , referred to prin
    cipal axes at the centroid , may itself be a principal axis at
    some point of its length , is
    a b b c ca
    m m 1
    16 =
    A -BBC
    - C - A P

    Here abc is any point on the line, and if a confocal quadric


    of the ellipsoid of gyration at the centroid be drawn through
    the stated principal point of the given line, p is the perpen
    dicular from the origin upon the tangent plane of the confocal
    at that point.
    1: - 2 - a _ Y - 6 _ % - C
    Proof. — The given line 7 m
    .. ... (i.)

    must be a normalto the confocal minutito = 1..........(ii.).


    py
    Therefore, by (5629), l = >A + X ' " B + X Ot " ......... ... (iii.) .
    Eliminate x , y, z from (i.) by means of (ii .), and from the resulting equa
    tions eliminate p , and the condition above is obtained .
    Also , by ( 5631),
    på = (1 + 1) 72 + (B + 1) m * + (0 + 1) nº = Al + Bm * + Cn* + 1 ... (iv.).
    The principal point xyz is now found by eliminating 1 and p from equa.
    tions (iii.), by means of (iv.) and (5977).

    INTEGRALS FOR MOMENTS OF INERTIA .


    By the definition (5903), the following indefinite integrals
    for moments of inertia are obtained :
    5978 For a plane curve, y = f (x), the M . I. about the x and
    y coordinate axes are
    e
    ſyeds and fads; andtherefore [(x2+ )ds= [rºds
    5 N
    826 SOLID GEOMETRY.

    is the M . I. about an axis perpendicular to both the former


    through the origin (5920 ).
    5980 Observe that ds may be changed into dx, dy, or do
    by the substitution formulæ (5113, ’16 ).

    5981 For a plane area bounded by the coordinate axes, the


    ordinate y and the curve y = f (x ), the M . I. about the x and
    y axes are

    S v dvdy= }\vdx and [ *dvdy= [x+yde.


    5983 And the M . I. about an axis perpendicular to both
    the former drawn through the origin ,

    = [ «*+y)dvdy= [ſrdrdo =uſredo,


    but in the last two integrals the area has the boundaries
    described in (5894 ).

    5986 The M . I. of a solid bounded by three rectangular


    coordinate planes and the surface z = f (x , y ) about the z axis ,
    will be

    Sf (x +y )zdxdy = $ff-*sinº drdodø,


    but in the last integral the solid is bounded as described in
    (5875 ) .
    5988 The volume, which represents the mass in all these
    cases, has already been expressed (5205 , 5871) ; and by
    dividing by the volume, the square of the radius of gyration
    of the solid is found (5904 ).
    PROOFS. - Formulæ (5981 - 3 ) are directly obtainable by geometry from
    figures 90 and 91, and formulæ (5986 – 7) from figures 168 and 188 . The
    transition to polar coordinates may also be effected by the formula of
    (2774 ).

    5989 In expressing moments of inertia , the factor m will stand for the
    mass of the body, and the remaining factor will therefore be the value of the
    square of the radius of gyration.
    PERIMETERS, AREAS, VOLUMES, S'c. 827

    PERIMETERS, AREAS, VOLUMES,


    CENTRES OF MASS , AND MOMENTS OF INERTIA
    OF VARIOUS FIGURES .

    RECTANGULAR LAMINA AND RIGHT SOLID .*


    For a rectangle whose sides are a , b , the moments of inertia
    about the sides , and an axis perpendicular to both where they
    meet, are respectively
    6015

    PROOF :
    Proof: fax*ds =abd= min The third by(5920).
    6018 Hence, for a right solid , whose dimensions are 2a,
    2b, 2c,
    M . I. about the axis of figure 2c = m . “ To .

    ARC OF A CIRCLE.
    6019 Let AB ( Fig . 191) be the arc of a circle whose centre
    is 0 and radius r . Let the angle AOB = 0 ; then
    Length of arc AB = r0. (601)
    6020 Huygens' Approximation. — RULE. — From 8 times the
    chord of half the arc take the chord of the whole arc , and divide
    the remainder by 3 .
    Proof.— The rule gives (16 sin - 2 sin ).
    Expand the sines by (764) as far as 6%, and the result is ro.

    6021 Taking an axis OX through the mid -point of the arc


    with origin 0 , the centroid of the arc is given by (5889)
    2rsin
    ã = 2rsane. Hence for a semi-circle ã = 2
    * For M . I. of a triangle, see (5944-52).
    828 SOLID GEOMETRY.

    6023 Also, for the centroid of DY


    BX, -
    = 2r sin ka

    where a = _ XOB.

    6025 The M . I. of the arc AB about OX and OY are


    mars(1- sing) and maps 1+ sing). (5978)
    6027 M . I. about axes perpendicular to XOY, through O and
    X the mid- point of the arc respectively , are

    kor
    mps and m (5979)
    mre and m.2re(1 –sing ). (5979)
    6029 Cor. — The M . I. of a ) – mr®
    circular ring about a diameter ) - 2

    SECTOR OF CIRCLE. AOB (Fig. 191)


    6030 Area = 0, ā = 4r sim 10. For XOB, = 4r sin?Ja.
    PROOF.-- , are respectively of , in (6021, '3 ) ; since the centroid
    of each elemental sector is distant fr from 0 . Otherwise, by (5893, '5 ).

    6033 The M . I. about OX and OY are


    may (1- sing) and may(1+ sing).
    PROOF. — By (5981- 2 ) ; or integrate (6025 –6 ) for r from 0 to r.

    SEGMENT OF CIRCLE. ABX (Fig. 191)


    6035 Area = (0 - sin 6), = 4r sin jo
    093( - siño) :
    6037 For CBX, į = " (2 – 3 cos a + cosºa).
    3 ( a - sin a cos a)
    Proofs. — From the sector and triangle ; otherwise, the centroid , by
    (5893, 5 ) .
    PERIMETERS, AREAS, VOLUMES, & c. 829

    6038 M . I. about OX and OY, (5981- 2)


    (39 – 4 sin 0 + sin 8 cos8) and (0 – sin 0 cos 6).
    а.

    6040 Cor.— Hence,for asemi-circle,ā =


    6041 Also , the M . I. of a circle about a diameter, and about
    a central axis perpendicular to its plane, are respectively
    (5920)

    THE RIGHT CONE.


    If h be the height, r the radius of the base, and I the slant,
    6043 Curved surface = url. Volume = karh.
    6045 Distance of centroid from vertex = fh .
    6046 M . I. about axis of figure = m for?
    6047 M . I.about cross axes through the vertex and centroid
    respectively.
    m (r2 + 4h ) and m 3% (4r + h ).

    FRUSTUM OF CYLINDER .
    Let o be the inclination of the cutting plane to the base,
    and c the length of the axis intercepted .
    6048 The distance of the centroid from the axis is
    aº tan 0
    4c .

    6049 The M . I. about the axis = m . , being the same as


    that of a cylinder of height c. Hence,by (5921)and the value
    of ã above, the M . I. about any line parallel to the axis can be
    found.
    830 SOLID GEOMETRY.

    SEGMENT OF SPHERICAL SURFACE. (Fig. 191)


    Let O be the origin of coordinates ; OA = p the radius ;
    and OC = x the abscissa of AB the plane of section .
    6050 The curved area of AB = 27r (r - x ) = the area of its
    projection on the enveloping cylinder of the sphere.

    Proor: Area = [245do = 2«r(r —a). (5878)


    6051 For centroid of surface, ū = } (r + x ).
    6052 The M .I.about the axes Ox,OY are
    ņ (279 – rx - xº) and m (4x + rx + x®).
    HEMISPHERICAL SURFACE .
    6054 Area = 2012, (6050 - 1)
    6056 M . I. about OX or OY = m {p . (6052- 3)

    SEGMENT OF SPHERE .
    3 (r + x ) ?
    6057 Volume = ; (2 + x)(r— a) , ã = 4 ( 2r + x )
    6059 M .I.about OX = ( — ~)*(8r* + 9rx + 3x+).
    6060 M . I. about OY
    = (r — x ) (16r9 + 17r*x + 18rx* + 9x ).
    PROOF. — As in (6146– 7) ; or put a = b = c in the results.

    HEMISPHERE.
    6061 Volume = zaro, ā = {r. (6064 )

    Proof. — Vol. = surface (6054) * ,by elemental pyramids having their


    common vertex at the centre of the sphere. Otherwise, make x = 0 in
    (6057 ) .
    PERIMETERS, AREAS, VOLUMES, & c. 831

    6063 M . I. about OX or OY = m fp? (6059 -60)

    SECTOR OF SPHERE .

    6064 Volume = ſorrº(r — h), ã = { (r + h).


    Proof.— Vol. = surface (6050) x 2 = 1 of õ in (6051), since the
    centroid of each elemental pyramid is distant (ths of r from the centre.
    6066 For the M . I. add together the M . I. of the cone and
    segment (6046 , '59).

    THE PARABOLA, y = 4ax. (Fig.of 1220)


    20 .

    6067 Rad.of curv. p = 29P* = 2a (1+ %) (4542)


    6069 Coordinates of centre of curvature
    3x + 2a , (4545)

    6071 Arc AP = s = v (ax + x2) + alog Vx + v (a + x ).


    - ſa
    6072 = a [cot 6 cosec 6 + log cot (30)].
    Proof:
    PROOF : s= 1 (1+ ) dx.
    S = (5197,4206)
    Substitute ( x, and integrate by ( 1931).
    6073 Arc AL = a 2 + a log (1 + 72).
    Centroid of arc AP with above value of s.
    6074 sõ = 22 +a / (2* +ax)+ " log 2x + a + 2 /(xº +ax).
    6075 sý = }{Va (x +a)— a"}.
    6076 For centroid ofare AL, putting x = a ,
    6 / 2 + log (3 + 2 / 2 (2 v2 - 1 ) a
    a,
    8 { V2 + log (1 + 12 V2 + log ( 1 + 2 )
    832 SOLID GEOMETRY .

    Half-segment of parabola ANP.


    6078 Area = žxy, ã = fx , y = ły.
    6081 The M . I. about the x and y axes are
    Xa xes are

    mfax and mix?.

    THE ELLIPSE.
    6083 The equation being bºx ? + aạy = a b >, the length of the
    arc AP (Fig . of 1205), putting $ for the eccentric angle
    of P , is

    s =av(1 –eºcosºb)dø.
    Proof. — In (ds) = (dx ) + (dy) (5113), substitute dx = - a sin odo,
    dy = b cos o do, by (4276), and use (4260).
    6084 The length of the elliptic quadrant AB is
    e 3e* 32.5e6 32. 5². 788
    2 ! 2 !2 3 ! 3 ! 264! 4 ! 28
    Proof. — Expand the binomial surd above, and employ (2454) and (2472).
    Similarly, from (5887 ) and (5978 ) the three following valges are found.

    6085 For the centroid of the same quadrant,


    }e’ _– 11s
    # 1l -– 1e
    ã = 2a et approximately .
    3.pt

    6086 The M .I.about the x and y axes are approximately,


    mb l - te - 62e ma’ 1 - ge - 640*
    2 1 - le - Beet and 2 '1- e - e*

    6088 Fagnani's Theorem . — (Fig. 192.) Let P be any point


    on the ellipse, CY the perpendicular on the tangent at P ;
    ZACY = 0 ; Q the point whose eccentric angle = - 0.
    Then
    6089 PY + AP = aſ V (1 — e sin’ 6)do = BQ;
    and in the hyperbola (Fig. 193)
    6090 PY - AP = aſV (1 — eº sin’8)do.
    PERIMETERS, AREAS, VOLUMES, 8c. 833

    6091 Cor. — The difference between the lengths of the


    no

    infinite curve and asymptote = a [ (1 – e?sin’0) do, where


    curve

    tan a =
    Proofs. — By (5203 ), re

    AP + PY or s+ q = pdø = a | V (1 —e*sin’8)do = BQ, by (6083).


    In the hyperbola we have 9 - s = pdo.

    6092 Draw the tangent at Q and the perpendicular CU upon


    it. Let æ, æ be the abscissæ of P , Q . The following relations
    subsist,

    PY = a = QU, CY.CU = ab, CP +CU° = +6 .


    Proof. — Let q = the eccentric angle of P , and let ACU = 6'. Then
    tano = m = tan 8. (4276 –80 )

    Similarly for Q, tan( –0) = cot 8 = tanb',


    therefore tang = cotoor ø = -0................... (i.).
    The relation therefore between P and Q is reciprocal. Now PY = e*x sin 0
    (4295) and x ' = a sin 0, therefore PY = e = QU, by the reciprocity .
    Again , CU ? = a' cos ' + b * sinº6' (4372) = a ' sinº° + 1° cos ...... (ii.).
    Put o in terms of 8 by the above, and we find
    ?
    CU ” – a co a’b
    a ’ cos 0 + 18 sin CY?
    Lastly, CP + CU` = x + y + a’ sin ’ 0 + b2 cos”o , by (ii), = a ’ + 6² (4276 - 7 ).
    6095 When P coincides with Q ,the point is called “ Fagnani's
    point," CY = v (ab), PY = a - b , and x = aš (a + b) - 4.

    6096 Griffiths' Theorem .* — If an ellipse of eccentricity e ,


    and a hyperbola of eccentricity e - ?, be placed as in the figure
    of 1205 (the circle representing the ellipse), P , p being con
    sidered corresponding points ; then , calling PQ , in (6088), a
    Fagnanian arc, we have the following theorem :
    * J. Griffiths, M . A ., Proc. Lond. Math . Soc., Vol. v., p . 95 .
    5 o
    834 SOLID GEOMETRY.

    The ratio of the difference of two Fagnanian arcs on the


    ellipse to the difference of the two corresponding arcs on the
    hyperbola is equal to the product of e ’ and the four abscissæ
    of the points on the ellipse.

    SECTOR AND SEGMENT OF ELLIPSE .


    6097 The formulæ for the sector and segment of a circle
    may be adapted to the ellipse by writing a for r and multi
    plying linear dimensions parallel to the minor axis by b : a .
    But a will then represent the eccentric angle of the semi-arc,
    and 0 twice that angle . Thus, in the figure of (1205 ) , if
    ACP be the half sector, a = ACp , 0 = 2 ACp .
    Sector of ellipse (2ACP in fig. of 1205) :
    O

    6098 Area = apo 2 = 4a sin


    30 30, y = 46 sinº
    3a £a,(6030-2)
    the last being for the half sector ACP. The M . I. about the
    a and y axes are

    6101 mb (1–sin ) and ma?(1+sing). (6088)


    Segment of ellipse (2ANP in same figure):
    4a sin 10
    6103 Area = 46 ( –sind), õ = ;3 (0 - sin )
    (6035 - 6 )

    6105 For y of the half segment ANP , and for the M . I.


    about the x and y axes, replace r by b in (6037 – 8 ) and by a
    in (6039).
    6108 For the whole ellipse, the area = mab. (6103)
    (6104 )
    6109 For thehalf ellipse, ă =
    6110 The M . I. about the x and y axes,and a third central
    axis perpendicular to both ,

    mba, mar, and m ( +8). (6041- 2 )


    PERIMETERS, AREAS, VOLUMES, 8c. 835

    6113 The area of the ellipse whose equation is


    (abefghXwy1)' = 0, is = 7h or C3
    PROOF. — If a , ß be the semi-axes of the conic , the area taş takes this
    value, by (4414 ) and (4407) .

    6114 Lambert's Theorem . — The area of a focal sector of an


    ellipse, as PSP ' (Fig. 28 ), in terms of $ ; ', the eccentric
    angles of P , P ', is

    20 { - –e (sin 4 – sin )} = "$ {x - x – (sin x - sin x)}.


    In the second value, sin and sin are = } V +
    respectively , where r = SP, p = SP , and c = PP',* a result
    of use in Astronomy.

    THE HYPERBOLA.
    6115 The length of an arc of the hyperbola 6 * — aʼyº = a b2
    and the abscissa of its centroid may be approximated to , as in
    (6084 ) for the arc of an ellipse, by the substitutions from
    (4278 ),
    Sds = aſsec ® V(e*sec*4 –1)dø
    and ſæds = a*ſsec*$ /(e*sec* – 1)dø.
    6117 Landen's Theorem . — This theorem gives any arc of an
    hyperbola in terms of the arcs of two ellipses, as follows :

    ſv (a® + bº + 2ab cos C) dC =


    \ / (a — sin’ A )dA + TV (b? –aºsin’ B )dB + 2a sinB + const.,
    that is — Arc of ellipse whose semi-axes are ath and a - 6
    = Arc of ellipse whose major axis is 2a and eccentricity b : a
    + difference between a right line and the arc of an hyperbola
    whose major axis is b and eccentricity a : b .t
    * Williamson 's Integ. Calc., Art. 137. + Ibid ., Art. 157.
    836 SOLID GEOMETRY.

    6118e Area ANP ( Fig. of 1183) bounded by x,y, and the


    curv

    = a{0V(x+—w*)–a*log®+ V(4x2+ )}. (1981)


    (4271)
    6119 = } { xy —ablog(* + *)}:
    6120 Area of sector between CA, CP and the curve
    = q log (@ + %)
    6121 Area between two ordinates 41, 42, when the asymptotes
    are the coordinate axes

    = . logo
    OF :
    (4387)
    Proor sin24coſyde = y Soportes
    6122 The centroid of ANP, A being the area (6118 ), is
    given by
    * A = (x – ); Ag = ( -de - g).
    Blu

    6124 The M . I.of ANP about the w and y axes are


    (203– 50*x) /(x2–a”)+ ab*log + + v (x* —aº).
    6125 (212–1* ) (x*–a)– a 10g«*+V(72–a)
    a .
    THE ELLIPTIC PARABOLOID .

    6126 Equation, + = 25.


    VIS

    612 Vol.of segment = a (ab)a , 5 = zz.


    6129 M .I.about the axes of x, y, and z respectively,
    m (9*+ ), n ( + ), max
    .
    PERIMETERS, AREAS, VOLUMES, Sc. 837

    6132 The surface S of the same segment may be found from

    s = 4(Jo resJo /(1+ * + x )dx dy. (3874)


    6133 If the surface of the paraboloid be bounded by a
    curve of constant gradient y (5881 ), the area becomes
    S = žrab (sec'y - 1) . (5883)

    THE PARABOLOID OF REVOLUTION.


    6134 Equation, x ? + y ' = 2az or p = 2az.

    1999
    6136 Surface of segment, S = žava {(2z + a): - a }. (5880)
    6137 Volume = maz? = jarm, x = zz. (5887, ’99 )
    6140 M .I. about axis of figure = mama (6131)

    6141 For M . I. about OX and OY put a = b in (6129- 30 ).


    THE ELLIPSOID .
    6142 Equation, + + = 1, semi-axes a, b,c (5600).
    6143 The surface of the segment cut off by the plane whose
    abscissa is x , will be found from

    S = 4
    ( a *b* + 6* (c — a”) xº + a * (c* — bº) y ') dxdy.
    JXJO a * b * - a²b * x * - a *b -y
    Proof. — By (5874) and (5629, '7 ), eliminating z by means of the equa
    tion of the surface .

    6144 The volume of the solid segment and the centroid are
    given by
    V = "h? (2a + x)(a — «2 ), ā =_ 3 (a + x )?
    4 (2a + x ) '
    PROOFS. — Let (Fig. 177) represent one octant of the ellipsoid ; 0A, OB,
    OC being the principal semi-axes. The elemental section
    4PNQ = *NP.NQdx = Va?—2 Va? - x*le.
    Therefore Vol.=moc["(a =4)dx = h' (2@ –3a*a+x) = &e.
    838 SOLID GEOMETRY.

    The moment with respect to the plane of yz

    = Fade['(a*r—x")da = be (a?— -)',


    and division by the volume gives č as above.
    6146 The M . I. of the solid segment about the axis a
    – mbe (b' + c^) (a — x )3 (8a? + 90x + 3x2).
    60a*
    PROOF. — ( Fig. 177.)
    M . 1. = )
    D NONP + N dx (6112) =abc (5 + C ) 6 - 8 ? ) dx = & c .
    4a'

    6147 The M .I.about the axis b :


    = mbio (a—~)*(84° +9ax +3x®)+ 24*–54Ⓡx +3x }.
    OOF :
    (5921)
    Proof : M.1.=S -NP.NQ(^l?+ON+)dx (5921)
    = ||(a?– )*dx+ Tube (2 = )z*ds= &c.
    6148 The volume of the whole ellipsoid = grabe.
    Proof.-- By making x = 0 in (6144 ).
    Otherwise : Let Ens be the point on the auxiliary sphere of radius r cor
    responding to xyz on the ellipsoid . By (5638 - 9 ) , rc = a , ry: = bn, rz = cz .
    Therefore dxdydz = abdždnds = abc game? (6061)

    6149 For the centroid of the semi-ellipsoid x = va (6145)


    6150 The M . I. about the axis a = m (6° + e“). (6146 )

    6151 The volume of a segment cut off by any plane PNQ


    ( Fig. 177) , where OA = d is the semi-conjugate diameter , and
    AN = h , is hº ( 3d - h )
    V = mabc "
    3d
    Proof. — Taking the area of the section from (5655), the volume of the
    segment will be
    tabe sin€ 1"(12 ) dx, where sin =
    0 being the inclination of d to the cutting plane. Integrate, and put
    a = d - h.
    PERIMETERS, AREAS, VOLUMES, & c . 839

    PROLATE SPHEROID .
    Put c = b in equation (6142) of the ellipsoid ; then a will
    be the semi-axis of revolution .
    6152 The surface of the zone between the plane of yz and a
    parallel plane at a distance x is

    S = " b { sin ** + * V (a?–e*x*)}.


    Proof.— By (5878). S = 2nbe Vie –x*)dx.
    a doVI
    Then by (1933). Otherwise,make b = c in (6143),and reduce.
    6153 Cor. — The whole surface = 206 (6 + a sin-te).
    6154 The centroid of the surface of the zone in (6152) is
    Zone

    iven by i = 2mbe { - -2°)"}.


    m

    Proor.–From Så=2rbeIV(* –- ) de.


    6155 The M .I. of the same zone is
    1 Cit

    = mati(1 - )sintex + (1+ - )«V(0°–€*c*).


    6156 And for the whole surface, by making x = a and a

    doubling)M.I.= mub"(1 - 3 )sin-*e+ wb*(1+z ).


    Proor : M.1.= 2+[vº/ (1+ as )dx = 20e[(a* ) 1 . –2*)dx
    = 2nd" V6 =-)de–200 [< 16 – *)de.
    The first integral by (1933). For the second,by Rule VI. 2048,we obtain
    the formula

    6157 [*+/( –z )do = sin-22 + 2 ***V(a"- "),


    а

    in which a mustnow be written for a.


    6158 For the volume, moment of inertia , and abscissa of
    centroid of the solid prolate spheroid , make c = b in (6144 -51),
    a being the axis of revolution .
    840 SOLID GEOMETRY.

    OBLATE SPHEROID .
    6159 Put b = a in the equation (6142 ) of the ellipsoid ;
    then c will be the semi-axis of revolution .
    The surface of the zone between the plane of xy and a
    parallel plane at a distance z, is
    S = ma: V (@ + a*e*r?) + TC log aez + V (c* + a*e*z*).
    Proor.– By(5878). S = 2042 [ V + *)dz. Then by (1931).
    6160 Cor.— The whole surface = 2wu*+ me log te
    6161 The centroid of the surface of the zone in (6159) is
    given by

    PROOF. — As in (6154 ). Z for the surface of half the spheroid is obtained


    in this case by making z = c, but in (6154) put x = a .
    6162 The M . I. of the same zone is

    M.1.= rode (1- de - )=v(e++d’ex")


    + *e*(19:= 3e") loguez +v(@ +a*erze)
    6163 And for the whole surface, by making = c and
    doubling ,
    M .I. = na'(1- 2e3 30°)logº (1 + e).
    ) + "c*(49*
    Proor:M.1.= 2*[< /(1+ )ds =27cle[(e -) + )dz.
    The first integral involved is given at (1931), and the second is obtained in
    the same way as in the Proof of (6155) , giving
    6164 ( x V x + a’dx = 22° + a e (x* + a?) – a' log {w + v (zº + a*)}.
    6165 For the volume, moment of inertia , and abscissa of
    centroid of the solid oblate spheroid , make b = a in (6144 –51),
    c being the axis of revolution .
    JOINT INDEX
    TO THE

    SYNOPSIS
    AND TO TIE

    PAPERS ON PURE MATHEMATICS,


    CONTAINED IN THE UNDERMENTIONED

    BRITISH AND FOREIGN JOURNALS


    AND

    TRANSACTIONS OF SOCIETIES.

    5 P
    “ There is an immense amount of knowledge lying scattered at the present day,
    and almost useless from the difficulty of finding it when wanted .”
    - Professor J. D . Everett.
    KEY TO THE INDEX.
    Prefixed to each title will be found the symbol by which the work is
    referred to in the Index. The words “ with Vol.” or “ with Year,” signify
    that any number following the symbol in the Index denotes , respectively,
    the Volume or Year of the journal. The year is given in all cases in which
    the work consists of more than one series of volumes. In order to connect
    the volumes with the years of publication , a Chronological Table is prefixed
    to the Index ; in which table successive series of numbers in any column
    indicate successive series of volumes of the publication .

    A . with Vol. – Archiv der Mathematik ; 1843 to 1884 ; 70 vols.


    [B . M . C.: P .P. 1580.] *
    Ac. with Vol. - Acta Mathematica, Zeitschrift Journal, herausgegeben
    von G . Mittag -Leffler ; Stockholm , 1882 to 1885 ; 7 vols.
    AJ. with Vol. - American Journal of Mathematics ; Baltimore . Editor :
    J. J. Sylvester, F .R . S .; 1878 to 1885 ; 7 vols. [B . M . C . :
    P . P . 1575 . 6 .]
    An .with Year. - Annali di Scienze Matematiche e Fisiche, compilati da
    Prof. Barnaba Tortolini; Rome, 1850 –57 ; afterwards — Annali
    di Matematiche pura et applica ; Rome, 1858 –65. Series II.,
    Annali diMatematiche pura et applica, compilati da F . Brioschi
    e L . Cremona ; Milan , 1868 – 85 ; 23 vols. in all. [B . M . C .:
    P . P . 1573 and 952. ]
    At.with Year. - Atti della Reale Accademia delle Scienze di Napoli;
    1819 to 1878 ; 15 vols. [ B . M . C . : for 1819 – 55, 8 vols.,
    Acad. 2813 ; for 1863 to 1878, 7 vols., Acad. 96 . ]
    C . with Vol. - Comptes rendus hebdomadaires des séances de l'Académie
    des Sciences ; Paris, 1835 to 1885 ; 100 vols. [ B . M . C . :
    Acad. 424 and R . R . 2099. c. ] +
    CD. with Vol. — Cambridge and Dublin Mathematical Journal. Editor ,
    W . Thomson, B .A .; 1846 to 1854 ; 9 vols. [B . M . C.:
    P. P. 1565.]
    CM .with Vol. — Cambridge Mathematical Journal ; 1839 to 1845 ;
    4 vols. [ B . M . C . : P . P . 1565 .]
    CP.with Vol. - Cambridge Philosophical Transactions; 1822 to 1881 ;
    13 vols. [ B. M . C . : Acad. 3008.]
    * i.e., British Museum Catalogue, Pressmark P. P. 1580.
    † R . R . signifies Reading - Room volumes within reach.
    844 KEY TO THE INDEX.

    E . with Vol. - Educational Times Reprint of Mathematical Questions


    and Solutions, with additional papers; London , half-yearly,
    1863 to 1885 ; 44 vols. Editor : W . J. C . Miller , B .A .
    [ B . M . C . : 2242 .c .]
    G .with Vol. - Giornale diMatematiche ad uso degli studentidelle univer
    sità Italiane, pubblicato per cura del professore G . Battaglini ;
    Naples, 1863-85 ; 23 vols. [ B. M . C . : P . P . 1572.]
    I. with Vol. - Journal of the Institute of Actuaries, or, The Assurance
    Magazine ; London, 1850 –84 ; 24 vols. [B . M . C .: P .P .
    1423.g .g . and 126 . ]
    J. with Vol. — Jourual für die reine und angewandte Mathematik ,
    herausgegeben von A . L . Crelle ; 1826 – 1856 ; and Journal
    als Fortsetzung des von A . L . Crelle gregrundeten Journals von
    C . W . Borchardt ; Berlin , 1856 – 1884 ; 97 vols. [ B . M . C . :
    P .P . 1585 and R. R . 2022. g.]
    JP. with Vol. – Journal de l'Ecole Polytechnique ; Paris, 1796 to
    1884 ; 34 vols. [ B . M . C . : T . C . 1.6 . ]
    L . with Year. - Journal de Mathématiques pures et appliquées, ou
    Recueil mensuel de mémoires sur les diverses parties des
    Mathématiques , publié par Joseph Liouville ; Paris, 1836 to
    1884 ; 49 vols. [ B . M . C . : P . P . 1575 and R . R . 2022. g.]
    LM . with Vol. — London Mathematical Society' s Proceedings ; 1866 to
    1885 ; 16 vols. [ B . M . C : Acad. 4265, 2.]
    M . with Vol. — Mathematische Annalen, in Verbindung mit C .Neumann
    begründet durch R . F . A . Clebsch unter Mitwirkung der
    Herren Prof. P . Gordan, Prof. C . Neumann, Vols. 1 - 9 ; and
    Prof. K . V . Mühl, gegenwärtig herausgegeben von Prof. F .
    Klein und Prof. A . Mayer, Vols. 10 , & c. Leipsig , 1869 – 1885 ;
    25 vols . [B . M . C . : P. P . 1581.6 .]
    Man . with Year. — Manchester Memoirs, or, Memoirs of the Literary
    and Philosophical Society of Manchester ; 1805 to 1884 ;
    23 vols . [ B . M . C . : 255 .d ., 9 - 12, and Acad . 1360 . ]
    Me. with Year. — The Oxford , Cambridge, and Dublin Messenger of
    Mathematics ; 1862 to 1871 ; 5 vols . Continued as — The
    Messenger of Mathematics. Editors : W . A . Whitworth,
    M . A ., C . Taylor, D . D ., R . Pendlebury , M . A ., J . W . L .
    Glaisher, F .R .S .; Cambridge, 1872 to 1885 ; 14 vols.
    [ B . M . C .: P . P . 1565. b . and 463.]
    Mél. with Vol. — Mélanges mathématiques de l'Académie des Sciences
    de Saint Petersburg ; 1849 to 1883 ; 6 vols. [ B. M . C . :
    Acad . 1125/9 . ]
    Mém . with Year. - Mémoires de l'Académie Impériale des Sciences de
    KEY TO THE INDEX. 845
    Saint Petersburg ; 1809 to 1883 ; 51 vols. [B . M . C .:
    1809 - 30 , T . C. 9.a ., 14 – 20 ; 1831 -59, Acad . 1125 /2 ; 1859– 83,
    Acad . 1125 /3 .]
    Mo. with Year. - Monatsbericht der Königlich -Preussichen Akademie
    der Wissenschaften zu Berlin ; 1856 to 1881; continued as
    Sitzungsberichte der Königlich , & c . ; 1881 to 1885 ; 30 vols.
    .. [ B . M . C . : Acad . 855 .]
    N .with Yoar. — Nouvelles Annales de Mathématiques. Journal des
    Candidats aux Ecoles Polytechnique et Normal. Rédigé par
    MM . Terquem et Gerono ; Paris, 1842 to 1882 ; 41 vols.
    [ B . M . C . : P . P . 1544.]
    No.with Year. — Nova Acta Regiæ Societatis Scientiarum Upsaliensis ;
    1775 to 1884 ; 26 vols. [B . M . C . : for 1775 –1850, 14 vols.,
    T. C . 6.6., 13– 19 ; for 1851– 1884, 12 vols., Acad . 1076.]
    P .with Year. — The Philosophical Transactions of the Royal Society of
    London ; 1781 to 1884 ; 104 vols : i.e., vols. 71 to 174.
    [ B . M . C .: R . R . 2021. g. ]
    Pr. with Vol. — Proceedings of the Royal Society of London ; 1800 to
    1885 ; 39 vols. [B . M . C .: for vols. 1 to 23 , Acad . 3025, 21 ;
    for vols. 24, & c., R . R . 2101. d .]
    Q . with Vol.-- Quarterly Journal of Mathematics ; 20 vols. Cambridge,
    1857 – 78 ; Editors : J . J . Sylvester, F .R .S ., N . M . Ferrers,
    F .R . S ., G . G . Stokes, F. R . S ., A . Cayley, F. R . S ., M . Hermite,
    F . R . S . ; 1878 –85, N . M . Ferrers, F . R . S ., A . Cayley, F . R .S .,
    J. W . L .Glaisher, F .R . S . [ B . M . C .: P . P. 1566 and 251.]
    TA . with Vol. - Transactions of the American Philosophical Society ;
    Philadelphia , 1818 – 71 ; 14 vols . [ B . M . C . : Acad . 1830 /3
    and T . C . 18. b . 11. ] .
    TE .with Vol. — Transactions of the Royal Society of Edinburgh ; 1788
    to 1880 ; 29 vols. [B . M . C . : T .C. 15 .b . 1 and R . R . 2099.g. ]
    TI.with Vol. - Transactions of the Royal Irish Academy; 1786 to
    1879 ; 26 vols. [ B . M . C . : Acad . 1540 and R .R . 2099.g . ]
    TN .with Vol. — Transactions of the Royal Society of New South Wales ;
    Sydney, 1867 -83 ; 17 vols. [B . M . C .: Acad . 1971.]
    Z . with Vol. — Zeitschrift für Mathematik und Physik , herausgegeben
    von Dr. O . Schlömilch und Dr. B . Witzschel ; Dr. O . Schlö
    milch, Dr. E . Kahl, und Dr. M . Cantor ; Leipzig , 1856 – 1883.
    28 vols. [ B . M . C .: P . P. 1581.]
    Year. A | Ac AJI An At | CD | CM | CP1 E 1 G JP Year.

    0 : : : : : :: : : : : : : : : : : : : : : ! ! ! !
    1801 801

    ::: : ::::::: : :: :::::::: ::::::::: :::: ::::::::: :::: :::: ::


    : :::
    : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : 606#
    : ::: ::: : ::::::::::::::::::: :

    12
    13

    ::: : :::: :: : ::
    14
    15
    16
    17
    18
    19
    20
    21
    22
    23
    24
    : : : :: : : : : :: : : : : : :

    25
    26
    27
    28
    29
    30
    !!!

    31
    32
    33
    34
    35
    36
    37
    38
    *

    39
    0 10
    41 11- 13
    42 14, 15
    43
    44
    1618,, 1719
    45 20 , 2321
    22.
    !:VOCIE

    46
    W-1111

    24, 25
    ::

    47
    48 10 , 11
    12 , 13
    14 , 15
    28, 3129
    30,
    51 16 , 17 32, 33
    :::: :::::

    52 1820 ,, 21
    19 II. 1
    53
    54 22, 23
    55 24 , 25
    56 26 , 27 42, 43
    !O

    57 28, 29 44 , 45
    58 30 , 31 II. 1
    32, 33 48 , 49
    50, 51
    34, 35
    W -: : : : : : : : : : : : : : : :

    52, 53
    63
    54,56 55 62
    51188583511:*
    :: :::: : ::::: ::::

    64
    65
    58, 59 63 1
    64
    :. voro

    W-:

    9, 10
    oAWN

    68, 69
    70, 71
    11, 12
    13, 14 1 , 72
    72, 73 15, 16
    74, 75
    76 , 77 17, 18
    19, 20 11 76
    78, 79 21, 22 12
    SO , 81 23, 24 13 SO
    25, 26 14
    27,
    29, 28
    30
    15
    16
    31, 32 17
    voi

    !!!!!!!!

    90, 91
    ot

    92, 93
    94, 95
    84
    3 , 4
    5-7
    9698 ,, 9799
    1885 100 43, 44
    An | At 1c | CD CM | CP | E G UJ | JP
    Year. L LM M Me |Man. Mém . Mél. Mo N No P Pr TN 2 Year.

    Over
    91
    92
    1801
    93

    os
    94
    95
    96
    99
    100 10
    110201 11
    12
    103 13
    104
    105 15
    106 . 16
    107 17
    108 18
    109 19
    110 20
    111 21
    112 22
    113 23
    114 24
    115 25
    116
    117 22
    118 28
    119 29
    120 30
    121
    122
    123 33
    124 34
    125 35
    126 36
    127 37
    128 39
    129 39
    130
    131
    132
    133
    134
    135
    11 136
    12
    13
    137
    138
    14 139 49
    140 50
    10 141 51
    10 11 142 52
    12 143 53
    10 11 13 144 54
    20 12 | 145 55
    13 146 56
    - 14 147 57
    15 148
    18 149 59
    19 150 0
    20 151 61
    III. 1 II. 1 152 62
    | 153 63
    154 64
    155
    11 156 66
    12 157 67
    13 158
    1415 159 69
    160 70
    16
    17 161 19 7112
    18 162 20
    19
    12 163 21
    III. 1 13 164 22
    14 165 23 75
    1 166 24 76
    16
    17 168 25,2726
    167
    169 28,3029 18
    19
    20
    170
    171 31,3332
    80
    81
    III. 1 172 82
    84 10
    20,22,2321 34, 35
    36, 37 83
    84
    1885 ... | 16 24, 25 14 38, 39 1885
    L LM M Me Man. Mém, Mel. Mo N No P Pr Q TA TE TI TN Z
    EXPLANATION OF ABBREVIATIONS, & c.
    a . c. = areal coordinates . i. eq . = indeterminate equation .
    alg . = algebraic . imag. = imaginary .
    up . = application . 1. c. m . = lowest common multiple.
    anal. = analytical. num . = numerical.
    ar. p. = arith. progression . 0. C. = oblique coordinates .
    c. C. = Cartesian coordinates . p . c. = polar coordinates.
    cn . = construction , p. d. e. = partial difference equations,
    cond . condition . p. e. = polar equation .
    cury , curvature. perp. = perpendicular..
    d . c. = differential calculus. pl. = plane.
    d. e. = differential equations. pr. - problem .
    d. i. = definite integral. q . c. = quadriplanar coordinates.
    eq. = equation . rad . radius.
    ex . = example or exercise . sd . solid or 3-dimensional.
    ext. = extension . sol. = solution .
    f. = formula , sym . = symmetrically .
    f. d . c. = finite difference calculus. ta . table.
    f. d. e. = finite difference equation . t . c. trilinear coordinates.
    geo . = geometrical. tg. c. = tangential coordinates.
    g. p. = geometrical progression , tg . e. = tangential equation .
    gn . = general. th . = theorem .
    gz. = generalization . tr . = treatise (i.e.,more than 50 pages).
    h . c. f. = highest common factor . transf. = transformation .
    i. c. = integral calculus.
    153. — The suffix means that three articles under the same heading will be found in the volume.
    18 — 21.- Meansthatonearticle on thesubjectwillbe found in each of the four consecutirevolumes.

    References to the Synopsis stand first, and are the numbers of Articles,
    not of Pages. An asterisk (* ) is prefixed where such numbers will be found .
    The unclassified references following a principal title commonly refer to
    papers on the general theory of the subject ; but some papers are occa
    sionally included amongst these of which the titles are too long for inser
    tion , and do not admit of abbreviation .
    Subjects which might well have been included under the same head.
    ing appear sometimes under different ones, for the following reasons :
    Exigencies of space have decided the insertion of the number of the volume
    only of the particular work in question, and a subsequent examination of
    the Index of that volume is required in order to find the page. It, there.
    fore, became desirable not to change the original title of the paper when
    there was danger, by so doing, of making it unrecognizable . When, how
    ever, the same subject appears in two parts of this Index under different
    names, cross references from one to the other are given . Some changes,
    however, have been made when the synonym was perfectly obvious ; for
    instance, when a reference to a journal, published fifty years ago , is found
    under the heading of “ Binary Quantics," the actual title of the article will,
    in all probability, be “ Homogeneous Functions of Two Variables," and so
    in a few other instances.
    INDEX

    Abacus of the Pythagoreans : L .39. Alternants of 4th order, co- factors of :


    Abelian cubics and symmetrical equa. | AJ.7 .
    tions : Q .5. Alternate numbers : LM .10 .
    of class ( - 31) M0.82. Alternating functions : AJ.7 : C .12 ,
    Abelian equations : A .68 : C.95 : J.93 : 22 : J.83 ( Vandermond's ) : Me.
    M .18 : Mo.77 ,92 . 82.
    Abelian functions : see “ Hyper -elliptic Altitudes , determination of: A .12,19 :
    functions." Mém .15 : N .45.
    *Abel's formula for F (x + iy) + F (x - iy ): Amicable numbers : A .70 .
    2705 . Me.73. Anallagmatic curves and surfaces :
    * Abel's theorems: 1572 : C .94 : J.9,24 , 0 .87 : N .64 (quartic surface).
    61,90 : LM .12 : M .8,17 : P .81, Anallagmatic pavements : E .10 .
    83 : Pr.30,34 . Analysis : A .1 : An.50 : 0 .3,11,12 :
    ♡ (@)+57. (y) = \ {wf(y)+ yf(a)} : An. J.f7: P .14 : Q .6 ,7.
    ap to geometry : G .23 : JP.4.
    * Abscissa : 1160. Analytical :- aphorisms: A .5 : J.10 .
    Acceleration : Me.tr 6 . combination theorem : J.11.
    * Algebra : 1 - 380 : A .tr 20. functions : Ac.6 : thsAn.82 : La
    application to geometry : JP .4 . grange, trJP .3.
    foundation , limitations : AJ.6 : CP.7, system of, and series from it: An .
    89 : Q .6 . | tr 84 .
    history of, in Germany : Mo.67,70 . * geometry : 4001 - 6165 : A .2.11.38 :
    Algebraic : — Calculus : N .81. C .6 : JP.9 : L .72 : M .2 : Mém .
    definitions : C .37 . | 13 : 2.9 : tr11,12.
    forms: C .84 ,94 : M .15 . theoremsand problems: A .8,52: J.46 .
    coordination of : J.76 . plane and solid in homogeneous co
    whose Hessian vanishes identically : ordinates : Z .15,16 .
    M .10 . of three dimensions : CM .4 .
    in theory of cubics : M .8. metrics : Q .7 ,8 .
    formulæ : G .12 : Q .5. theorems: A .8. treatise : C .13.
    functions: A .10.31 : J.92 : L .50 .51 : 1 Angles : - conterminal : Me.74.
    M .10 : N .62. of a central conic : 4375 .
    applied to geometry : G .22 : M .7. division into n and n + 1 parts : A .70 .
    number of constants : J.64. of five circles or six spheres : Me.79.
    as partial fractions : Z .9. of two circles : 4180 .
    rationalization of : A .69. problemson : P .1791.
    representation by : J.77 ,78 . two relations between five : A .20 .
    resolution into factors : A .46 . * trisection of : 5325 : A .4,34 : C .2,66 ,
    theorems: J.82 : M .1,6 . 81 : G .15 : Me.72 : N .56 ,76 .
    synthesis : C .162. Anharmonics : LM .2,3.
    Algorithmic geometry : N .57. *Anharmonic
    — 21.
    pencils of conics : 4809
    Algorithm : — re definition of * Anharmonic ratio : 1052,4648 : GM .12.
    J.27. corresponding to roots of a biquad
    of higher analysis : Mo.75 . ratic : N .60.
    of arithmetical functions: G .23. * ofa conic: Q .4: of fourtangents: 4986.
    850 INDEX.

    Anharmonic ratio (continued ) : Arithmometer : I. 16 – 18 .


    of 5 lines in space : Me.76 . Aronhold ,theorems of : gzJ.73.
    of 4 points in a plane : A .1 : C .77. Associated forms : - systems of :
    sextic : LM .2,60 : Q .37,38 . gzAJ.1 : C .86 : CD .6.
    systems: cnM .10 . and spherical harmonics : Me.85.
    * Annuities : 302 : A .2g: Ac.1 : CP .3 : | Astroid of a conic : A .64.
    J.83 : 1. 1 - 24 : P .1788, -89, -91, *Astronomical distances : p 5.
    - 94 , 1800 , - 10 : N .47. * Asymptotes : 5167 : A .p. c 15 ,17 : CM .
    Anticaustic (by refraction )of a parabola : 4 : M .11 : N .68 : thsN .482 and
    N .83,85. 73.
    * Anticlastic surface : 5623,5818. of conics : 1182, 4490, t.c 4683 : tg. c
    Aplanatic lines, lemniscates, caustics, 4904, -66 : Me.71 : Q . 3,8 .
    & c.: thL.50 : N .45. of intersections of quadrics : N .73.
    Apolarity of rational curves : M .21. of imaginary branches of curves :
    Apollonius's problem : A .37 : M .6 . CM .2.
    Approximation : J. 13 : N .66 . Asymptotic : chords : A .12.
    algebraic : J .76 . cone of a quadric : 5616 : E .30, g.e 34 .
    to functionsrepresented by integrals : curves : 5172.
    C .20 %. lines of surfaces : A .60 ,61 : R .84 .
    of several variables : C.70. law of some functions: Mo.65 .
    successive : Mém .38. methods : M .82.
    Apsidal surfaces : Q .16. planes of a paraboloid : 5625.
    Arabs,mathematics of : C .39,60 . planes and surfaces : CD .3.
    Arbitrary constants : C .15 : L .80 : in Atomic theory and graphical represen
    d .e and f.d . e, TI.13. tation of invariants and covari.
    * Arc, area, & c.: quantification of : ants ofbinary quantics : AJ.12.
    1244, 5205,5874,6015 . Attraction : - of confocal ellipsoids:
    relations of : G .16 : C .80,94: L .46. Me.82 .
    Arcs with a rectifiable difference and of ellipsoids : CD.4,9 : J. 12,20,26 ,31 :
    areas with a quadrable difference : JP.15 : L .10,45 : M .10 : N .76 :
    L .46 . Q .2,7,17.
    * Areas: - and volumes in t.c and q.c : of ellipsoidal shell: J.12 : JP.15 : Q .17.
    4688 : Q .2. of paraboloids : L .57.
    * approximation byordinates: 2991 – 7 : of polyhedra : J.66 .
    C.78 : CD.9. of a right line aud of an elliptic arc :
    between 3 lines : 4038 : Me.75. An .59 : CD.3.
    ext. of meaning : CD.5. ofa ring and of elliptic and circular
    Arithmetic : A .5, 18 : L .59. plates : G .21 : 2 .11.
    ancient : C .71 : N .51. of spheroids : J.12 : JP.8 : L.76 :
    Mém .31 : P .(Ivory ) 12.
    degenre, ext. of the notion : C.94.
    higher : J.35. of solids of revolution , & c. : An.56 :
    history of : C .17. CD.2 .
    of Ibn - Esra : trL .41. solid ofmaximum : TE.6 .
    of Nicomaque deGérase : An.57. theorems: Q .4,17.
    Arithmetico-geometric mean : Mo.58 : theory of : L .44,6 .
    Z .20 . * Auxiliary circle : 1160.
    * Arithmetic mean : 91 : CD .6 : of nl Averages : I.7 ,9 .
    quantities : 332 : L .39. * Axes : - ofa conic : gn .eq 4687 : A .30 :
    * Arithmetical progression : 79 : thL . E .36 : G .12 : Q .q .c 4 ; t.c 5,8,15
    39 : Pr.10 . and 20 : Me.a .c 64,71 : N . 43 ,48 ,58 :
    and g. p when n (thenumber of terms) t.cQ. 12.
    is a fraction : A .35 . construction of : 1252 : Me.66 ..
    * *

    when the terms are only known ap cn . from conj. diameters : 1253 :
    proximately : C .96 . A .13,20 : Me.82 : N .67,78 .
    Arithmeticaltheorems: A .10 : C .93,979: | * of a quadric : 5695 : A . 30 : An.77 :
    CD.6 : G .7,18 : L .63 : of l. c. m ., G .9 : J.2,64,82 : N .43,51, cn 68,
    N .57 : Gergonne, C .5 . 69,74 .
    Arithmeticaltheory of algebraic forms: 1 * rectangular, nine direction cosines
    J.92,93. for two systems : 5577 – 8 :
    Arithmographe polychrome: 0 .51,53. L .442.
    INDEX. 851

    *Axis :-975 .of perspective or homology : | Binary cubics — (continued):


    automorphic transf. of : LM .14 .
    * of reflexion : 1007. and quadratic forms : G .21.
    of similitude : 1046,4177. system of two : E .7 : G .17 : LM .13 :
    Axonometry and projective collineation M .7 .
    in space : M .25 : Z .12.21. resultant : Q .6 .
    tables and classification of : A .31.
    Babbage's calculating machine : C .992. transformation by linear substitution :
    Barycentric calculus and right line con J.38 .
    struction : J.28 . Binary forms: An.56,77 : At.65 : G .
    Battement de Monge : L .82. 2,3,10,162: J.74 : M .2,3,20: Q .14 .
    Beltrami's theorem : A .44 . and their covariants, geo .: M .23.
    * Bernoulli's numbers : 1539 : A .3 : ap. to anal. geometry : L .75 .
    AJ.5,7 : An.59,77 : C .549,583,812 : ap. to elliptic functions: AJ.5.
    G .9 : J.20,21,28 ,58,81,84,85 (first ap . to Euler's integrals : C .47.
    62),88,92 : LM .42,7,9 : Me.75 : canonical: J.54 : M .21.
    gzMém.83 : N .76 : Q.6,22. evectant : Q .11 .
    application to series : see “ Series.” geo . interpretation or ap . : C .78 : G .
    and interpolation : C .86 . 17 : M .9 ,22%.
    and their first 250 logarithms: CP. 12. having the same Jacobian : C .94.
    and secant series : A .1,3,35 : C .4,32 . having similar polar forms: M .8 .
    bibliography of : AJ.5 . in a cubic space-curve : J.86 .
    indeterminate representation of : Z . in twoconjugateindeterminates: C .97.
    19. most general case of linear equations
    new theory of : C .83 . in : 0 .99.
    theorems on : E .2,8 : N .77. ( 9) groups of : M .23 .
    Bernoulli's series : 1510 . with related coefficients : M .12 .
    Bessel's functions (see also “ Integrals transference of, when not of a prime
    of circular functions " ) : J.75 : degree : M .21.
    M .3 ,4,9,14,16 : Q .20,21. transformation of: M .4,9.
    representation of arbitrary functions typical representation of: An.68,69.
    by : M .6 . Binary homographics represented by
    squares and products : M .2. points in space, applied to the
    tables : 2 .2 . rotation of a sphere : M .22.
    Bonnet, two formulæ of : G .4. Binary nonics, ground forms: AJ.2 .
    Bicircular quartics : LM .3,99 : P .77 : Binary octics : thC.96 : M .172.
    Pr.25 : Q .192 : TI.24 . Binary quadrics : C .47 : G .3 : J.27 :
    focal conics of : LM .11. L .59,77 : M .15,172.
    withLM
    collinear triple and double foci : construction of, through a symboli
    .12 ,14 . cal formula : 0 .57.
    nodal, mechanical cn . of: LM .3. indeterminate, integral sol.: J.45.
    Bifocal variable system : M .16 . for a negative determinant : No.81 :
    Bilinear forms: J.68.,84,86 : L .74 : C .60 (table) : L .57 : M .172,21,25 .
    Mo.66 ,68 ,74. partition table : AJ.4 .
    congruent transformation of : Mo. representing the samenumbers: L .59.
    74. transformation of: C .41.
    four variables : G .21 : M0.83. with two conjugate indeterminates :
    relation between two and their C .96 .
    quadric and quartic system : M .1. * Binary quantics : 1636 : An.56 : C.52:
    reciprocals : G .22. CD .9.
    reduction of: C .78,92. (2n - 1)-ic, canonical form of : Q .20.
    Bilinear functions : GM .11. derivatives of : Q .15 .
    polynomials : C .77 . discriminant of: 1638 : Q .10.
    trilinear and quadrilinear systems: reduction of : J.36 : L .52 : Q .7 .
    E .5 . transformation of: CM .1 : thE.28.
    Billiards, theory of : L .83. in two polynomials U , V , prime to
    Bimodular congruences : G .21. each other and of the same de
    Binary and ternary quadratics: N .64,65 . gree : N .85.
    * Binary cubics : 1631 : A .17 : C .92 : G .17 : Binary quartics : - and their invari.
    J.27,53,41 : Q .1,11. ants : A .18 :6 .14 : J.41: M .19: Q .7 .
    852 INDEX.
    Binary quartics — (continued) : * Brachistochrone : 3037, 3044 : L .48 :
    condition for perfect square : E .36 . Man .31: Me.80 : Mém .22: N .
    or quintics, with three equal roots or 77 ,80 .
    two pairs of equal roots : P .68 . Brahmins,trigonometricaltables of the:
    and ternary cubic, correlation be TE .4 .
    tween : An.76 . * Brianchon's theorem : 4783 : A.53 : C .
    Binary quintic : G .14 . 82 : CM .4 : G .2 : J .84, 93 ; gzN .
    canonical form for : Q .19 . 82 : 2 .6 .
    Binary sextics : taAJ.4 : C .64,87,963: and analogues : CD .7 : Q .9.
    G .14 : M .2,76 ,77. on a quadric surface : C .98 : E .19.
    syzygies of: AJ.7. on a sphere : A .60.
    Binet's function : L .75 . * Brocard circle and Lemoine's point:
    Binodal quartic with elliptic function 4754c : gzN .85 .
    coordinates : AJ.5. * Burchardt's factor tables p.7 : erra
    * Binomial : coefficients : 283,366 — 7 : tum , A .23 .
    A .1,2 : G .14 : Mém . 24 : N .th60 , * Burmann's theorem , d.c : 1559.
    61,70,85 : Z . 25 .
    sum of selected : Me.85 . Calculating machine : Pr.37.
    equations : 480 : A .10 : At.65 ,68 : Calculus : - algebraic, which includes
    C .10,44 : G .5,,10 : L .57 : LM .11, the calculus of imaginaries and
    12 ,16 . quaternions : C .91.
    irreducible factors of: An .69 . of chemical operations : Pr.25.
    QuP - 1 = 0 : see “ Roots of unity.” of direction and position : AJ.6 : M .
    equivalences to any modulus : C .25 . pr 6 .
    * theorem : 125 — 36 : A .8, geo. 61 : C . of enlargement : AJ.2 .
    45 : CD .7 : CM .3 : G . 12 : J .1,4 ,5 , ofequivalent statements : see Logic."
    28,55 : Me.71 : N .423,47,50,71,78 : of forms ( Invariant theory ): CD .79,
    P .16 ,16 ,95,96 : TI.12. 8,99
    generalization of : J.1 : N .579. of infinitesimals, third branch of :
    * Binormal: 5723. viz ., given y and yz, to find 2 :
    Bipartite functions and determinants : TE .24 .
    LM .16 . of limits, ap. to a system of d.e : C. 156.
    Bipartition , repeated : Mel.4. of Victorius : 2 .16 .
    * Biquadratic : equation (see also of other subjects : see the subject.
    “ Cubic and biquadratic ” ) : 492 Calendar : J.3,9,prs 22 .
    -- 501 : A .1,12,16 ,23,31,39,40,41, Jewish : J.f 28 .
    45,69 : AJ. 1 : CM .1,46 .,47 : G .5 Canal surfaces : A .1,10.
    - 7,13,21 : J.90 : Me.62: N .44,589, Canon-arithmeticus of Jacobi. C .39,63 :
    59,60 ,63,78 ,81,83 : Q .7,28 : Z .6 ,8 , L .54 .
    18 . Cantor's theorem : M .22.
    cond. for two equal roots : E .44 . *Cardioid : bra in 5328, Fig.129 :
    and elliptic functions: C .57 : L .58. A .59,63,68 : LM .4 : Me.64 : N .81.
    reduction to canonical form : G .5. and ellipses : Pr.6 .
    reduction to a reciprocal equation : *Carnot's theorem : 4778.
    L .63. * Cartesian oval : 5341 – 5358 : A .69 : C .
    solution of : A .51,56 : C .49,82 : L . 97 : LM .1,3 ,99 : Me.75 : Q .1 : Me.
    73 : Q . 19 : numerical, C .61:with 74 : Q .12 ,cn15 .
    out eliminating the 2nd term , area of: E .21.
    A .39 : 4 variables, J.27 : trigono eq.with triple focusas origin : E .9,23.
    metrical, A . 19,70 . foci: E .7.
    and sextic eqs. in the theory of functional images : Q .18 .
    conics and quadrics : J.53. mechanically drawn : LM .5,6 : Q .13.
    values which make it a square : perimeter : E .21.
    496 : G .7 : E .22 . rectific . by ellip . functions: 0.80,87 :
    function with four variables : An.59. LM .5 .
    involutions: C .98. through 4 points on a circle : LM .12.
    Biquaternions : AJ.7 : LM .4. with 2 imaginary axial foci : LM .3.
    Biternary forms with contragredient *Cassinian oval: 5313 : Me.77,83 : N .57.
    variables: M .1. analogous surfaces : An.61.
    Borchardt's functions: J.82. radial of : E .26 .
    INDEX. 853

    Cassinoid with n foci, rectif , of : L .48. I Centroid — ( continued) :


    Catacaustic and diacaustic of a sphere : of a frustrum of a pyramid : Me.79 :
    JP .17. N . 769.
    Catalecticant of a binary quantic : E . of oblique frustrum of a cone: E .33.
    37,38. of a perimeter : A .51.
    Catenary : 5273 : Me.64,66,68. of plane curves : 5887 : J.21 : G . 12 .
    by parabolic trigonometry : Pr.8. * of spherical and other areas: 5898 ,
    revolving : E .22. 6051 : J.50 : L .39,429.
    *Cauchy's formula i.c : 2712. * of surface and solid of revolution :
    closed curve theorem : M0.85 . 5896 – 9 : AJ.3 : L .39.
    various formulæ : 0 .2710. of a trapezium : Q .9 .
    * Caustics : 5248 – 9 : Ac.4 : L .46 : P .57 , * of a triangle : 951 : A .52,58 .
    67 : Pr.8,142,15 : Q .1,2,3 cn8,9,12. | *Change of independentvariable : 1760 —
    by successive reflexion from spheres : 1816 : AJ.3 : CM .1 : G .2 : L .40 ,58 :
    J.13 . Q .1,2, 10 : Z .17 .
    identity with pedals : Z .14 . * from a , y to 1, 0 : 1768.
    *

    of a cardioid : Me.83. from x , y , z to r, 0, 0 : 1783.


    * *

    of a circle : 5248 : CD.2 . in a definite multiple integral : 2774


    of a cycloid : A . 30 . 9 : A .22,41.
    of an ellipse, focus at centre : J.44 . in transcendent definite integrals :
    of infinitely thin pencils : J.98 . 0 .23.
    of refraction at a plane surface : N .47. in thetheory of isotropicmeans : C .34 .
    radii of curvature : N .65 . Characteristics: E .5 : .71 : M .6 .
    surfaces and singularities
    Cells of bees : N .56 : Q .2.
    : J.76. of conics : A .1 : C .67,72,76 ,839 : JP.28 :
    LM .9 : M .15 : N .666,71.
    Cell structure : LM .16 . of conics of 5-point contact : E .27.
    Centimetre - gramme -second system : of cubic systems : C .742.
    p .l. of curves and surfaces : C .73 .
    Centrals, theory of: J.243. of quadrics : C .67 : JP .28 : N .68.
    Centre of : - curves and surfaces : L . relation between two characteristics
    46 : Q .10 . in a system of curves of any de
    a circle touching two : A .24 . gree : C .62.
    geometrical figures : A .16 . surface groups defined by two: C.79 :
    p = l , v = 1, C .78.
    harmonic mean : J.3 .
    mean distance of curves and surface : 1 Chart construction : Mél.2 : N .60,78z.
    s
    N .ths 85 : Q .33. Chemico-algebraic theory : AJ.1.
    mean distance of points of contact of Chemico-graphs : AJ.1.
    parallel tangents, ext. of th . : L . Chess board , ths and prs : A .56 : E .34,
    44. 42 ,44 .
    similarity : 947. Chessmen,
    Mél.3.relative value of: E.39:
    # * *

    similitude: 1037.
    similitude of 3 circles : 1046 : 4176 . Chinese arithmetic and algebra : C.51:
    similitude of 2 quadrics each of which N .63.
    circumscribes the same quadric : | * Chord : - - joining two points on a
    J.31. circle : 4157 : A .43,44 : E .22.
    Centres, theory of : J.24 . * of contact for two circles : 4172.
    Centro -baric methods in anal, geometry : Chronology : J. 26 .
    J .52. | * Circle : 4136 – 90 c.c : A .1,3 th 4 ,9,25,27 ,
    * Centroid :- - formulæ for : 5884 — 5902, th47 : C .94 : J.14,17 : Q .19 : TE .
    6015 : JP .26 : L .43. tho .
    and its use in stereometry : A .39. * approximate rectification and quadra
    of common points of two conics : A .3. ture: 6019, & c: A .2,6,geo13,14,
    of circular arc : 6021 : E .13. 43 : J.32 : Me.75 ,85 : N .45,47 :
    of a dice : N .63. Q .4 .
    of frustrums: 6048 , & c.: A .33. arc of : see Circular arc.
    of a gauche curve after development area of segment : 6035 : A .27,39 ,44.
    on a right line, locus of : 0 .88. chord of : 4157 - 8 .
    * *

    of algebraic curves and surfaces : chord of contact : 1017 : 4138, - 72.


    An.68 . coaxal: 1021 — 36 : A .23 .
    *

    of a frustrum of a prism : L .39. configuration of : C .932.


    854 INDEX.
    Circle — (continued): Circle — (continued ):
    * cn . from 3 conditions: 937 : JP.9. * touching 3 circles, cn : 946 , 1049 :
    * Cotes's properties : 821 : A . 11, ext. to A .24 ,26 ,28,35 : An.68 : C .60 : Me.
    ellipse 30 : P .13 : TI.7. 62: N .63,65,66,84 : Q.8.
    cutting three at given angles : 4185 :
    LM .3,5 : N .83 .
    touching the 4 circles which touch
    division of : A .27,37 ,41 : At.19 : E .1,
    the sides of a spherical triangle :
    A .4 .
    31 : G .6 : C .85,93 : J.27 ,54,56 : and triangle, ths and prs : A .30 ,57 :
    N .53,54 . LM .15 : Q .4.
    and theory of numbers: J.30,812,87 : * two; eq. for angle of intersection :
    N .56 . 4180 — 1.
    ths. on sum of sqs. of perpendicu theorems: 984— 1045 : Q.11 : see
    lars, & c.: 1094 - 8. “ Radicalaxis ” and “ Coaxal cir.
    eight circles touching three, cn : cles."
    4189 : Mél. 3 : Q .5 . and two points ; Alhazen 's pr : AJ.4.
    eight through 6 points of intersection Circulants: — finalexpansion of: Me.85.
    of 3 conics : Q .10. of odd order : Q .18.
    equation of : 4136 – 48, p .C 4151 : * Circular : - arc : length , centroid , & c.:
    Pr.27 : TI.26 . 6019.
    general eq.: t.c, 4691,4751 : tg.c, with real tangents : Z .1.
    4906 . graphic rectification and trans
    Euler's th. extended to ellipse : A .51. position of : 2 .2 .
    five-point th . : E .5. cubics, involution of : LM . 1,7 .
    four pairs of circles through 6 points chord of curvature of : E . cn 36 .
    common to 3 circles : Q .9. and elliptic functions in continued
    four points concyclic, condition : A . fractions : CD.4 .
    44 : N .84. functions : 606 : A .17 : J.16 .
    geometry of : A .67 : 2 .24 . points at infinity : see “ Imaginary
    groups of points on : A .14 . es ditto ."
    in tri-metric point-coordinat : Z .27 . relation of Mobius: LM .8 : N .76 : Q.2.
    and in -quadrilateral : 733 : CD.9. segment : arc, chord and area : 6035 :
    lines of equi-different powers in two N .63.
    circles : A .19. Circulating functions: P .18.
    * of curvature: 1254,5134 : A .31,63 : J. *Circum -centre of a triangle : 4642 : tg.
    45 : p.cN .84. eq 4883.
    polar of X'y': 4138 , - 64. * Circum -circle : - - of a triangle : 713 ,
    rectangles of : 2 . 14 . 4738 : tg . eq 4895 : A .51,58.
    ringMe.78of, touching
    .
    two fixed : J.39 : coordinates of centre : 4642.
    hypocycloidic envelope of Ferrers :
    six points th . : Q .8. N .70.
    and self- conj. triangle : A .41. and in -conic : N .79.
    and sphere, geo. : Mo.82 . * of a polygon : 746 – 8 : A . 19.
    system through a point on a plane or * of a quadrilateral : 733: N .79 .
    sphere : G .16 . Circum -cone of a quadric, locus of
    tangents : 4137 - 43,4160 : L .56 : P. 14 . vertex : N .52.
    common to a circle and conic : *Circum -conic :- - of a triangle : 4724 :
    CnA .69. tg. eq 4892: An.57.
    common to two circles : 953,4171 : * of a quadrilateral: 4697 : At.54.
    not

    COA.34. locus of centre : E .1.


    locus of a point, the tangents from Circum cubic of a complete quadri
    *

    which to two circles have a given lateral: G .10 : Q .5.


    ratio : geo.965 — 6 . * Circum - parallelogram of an ellipse :
    * three : 997 — 9,1036 ,1046 - 51,4183 – 7 . 4367.
    * *

    prs.(Gergonne) : 1049 : At. 19 : L .46 . Circum -pentagon of a conic : M .5 : N .67.


    through mid -points of sides of a tria Circum - polygon : - of a circle :
    angle : see Nine -point circle. 746-- -8 : CD .1 : Me.80 : N .66 .
    through 3 points : 4156 ,4738 . of a conic : M .25 .
    through 3 points on a conic : A .2 : of a parabola : CM .2.
    J.39. of a cuspidal cubic : LM .13 : ditto
    touching a conic twice : J.56 : N .65. quartic : LM .14.
    INDEX. 855

    Circum -quadrilateral: — of a circle : ! Combinations — (continued ):


    E .35 : N .48 . * problem or theorem : 105 — 7 : A .21 :
    of two circles : N .67. C .97 : CD.2,4 ,7,82 : 1.5 : J.3,45,56 :
    Circum -rhombus of an equil. triangle : L .38 : Mém .11 : N .53,73 : 2 .15 .
    of Euler and its use in an eq . : L .39 .
    A .45 .
    Circum -triangle : of a conic : N .70. of 1, 2, ... n , each c. having a sum
    * locus of vertex : 4800 : E .35 . pa: G .19,20.
    of a triangle : J.30. of dominoes : An .73 .
    * Cissoid : 5309 – 12 : A .62,69 : N .43,85. of n dice each with p faces : TE .21.
    tangents of : LM .2. of n points in space : L .40 .
    Cissoidal curves : A .56. of observations: L .50 .
    Clairaut's function and equations: Pr. of planes through a system of
    25 . points : N .57.
    Clinant geometry : Pr.10, 112,12,15 . Combinatorial: - products : A .34.
    Closed curves : Me.77 : geo th Q .4 . systems: L .56 .
    and movingchord,Holditch 's th: 5244 : Combinatory analysis : A .2,50,70 : J.11,
    gzMe.78 : Q .2 .
    ext. to surfaces : Me.81. 22 : Mém .50 : N .80 .
    quadrature of : A .61 : N .43 : Crofton 's Commensurable quadratic divisors : N .
    47.
    ths A .55 : C.65,68 : E.36,.
    [ sin Ꮎ , Commensurables : TE .23.
    in
    JJ tt' dedy, t = 0, ť = 0 being the *Commutative law : 1489.
    tangents from xy : LM .2. *Companion to the cycloid : 5258.
    * rectification of : 5204 . Complanation formula : A .48.
    Complementary functions : C .19 : J. 11.
    Closed surfaces : JP.21. Complete functions : C .86 : J.48 .
    *Coaxal circles : 1021 — 36 : 4161 – 70 : Q. Complete numbers : M0.62 .
    52 : reciprocated , 4558. *Complete primitive : 3163 .
    Poncelet's limiting points: 4165 : Complex axes of a quadric : Z .19.
    | thJ.86. Complexes : L .44,47 : M .2,4 .
    Coaxal conics : Q. 10 . of axes of a quadric : N .83.
    Cochleoid ,(a ? + y?)tan -1 } = ary : A .70. in combinations and permutations :
    A .21.
    ** Coefficients : - - detached : 28.
    differential : 1402.
    of 1st and 2nd degrees and linear
    congruences : An.76 : L .51 : M .2 ,
    * indeterminate : 232. 9 : N .85 : trZ .27 .
    * Cogredients : 1653. linear: N .85 : Z .18 : of an in -conic of
    *Collinear and concurrent systems of a quadrilateral: G .21.
    points and lines : 967 : A 69 : G .21. of 2nd degree : G .8,17,18 : cn J.93 :
    Collineation and correlation : M .22 : M .7 : N .726.
    prM .10. of 2nd degree with a centre : L .82 .
    and reciprocity : M .23. of 2nd degree ofright lines which cut
    “ gleichstimmigkeit ” of, in space: two quadrics harmonically : M .23.
    Z .24,28 . quadratic ray. & web -complexes: J.98.
    multiple c. of two triangles : A .2,70. of nth degree, singularities : M .12 .
    of plane figures, ground forms: J.74 . and congruences, spherical of 2nd
    paradox : Z . 28 . degree,their circles and cyclides :
    *Combinations : 94 — 107 : trA .15 : CP.8 : J.99.
    G .18 : J.5,13,21,,34,38,th53 : M .5 : and spherical complexes, ap. to linear
    Z .2 . p . d . e : M .5.
    ap. to determinants : JP .28 . tetrahedral in point space : 2 .22 .
    complete , i.e., with repetitions : C .92 : Complex numbers : A .28 : C .90,99: G .11:
    N .42,74. J.22,35,67,93: L .54,75 ,80 : M .22 :
    compound : Man .79. Mo.70 : Q .4.
    C (n , r ) an integer : 366 : L .42. from the 31st roots of unity : Mo.70.
    C (n , r) when n is fractional: A .70. from the nth roots of unity : J.10 :
    C (n, 10r) = C (n - 1,- - 1) + C (n - 1,r) : M0.70 .
    2. index and base of a power, geo : Z.5.
    C (m + m',p) = =" C (m ,r).C (m ,p —r): primeand from roots of unity : J.35 :
    L .42. taM0.75 .
    856 INDEX .

    Complex numbers - (continued) : Confocal surfaces : Me.66 .


    primeand from the 5th roots of unity:
    Mo.ta 59.
    Conformable figures : A.59 : LM .10 : M .
    19 : Z .17.
    resolution of A " + Ba + C = 0, and Congruences : C .51,88 : th and ap.J. 19 :
    when n = 5 : L .47 . th Me.75 : N .50 : P .61.
    from the roots of unity ; class num . binomial: AJ.3 : C .61 : expon. to base
    bers: J.65 : Mo.61,639,70. 3, Mél.4 .
    in theory of residues of 5th , 8th ,and classification of roots : C .63.
    12th powers : L .43. Cremonian : LM . 14 .
    Complex unities : C .96 ,99 : J.53. irreducible : J .40.
    Klein 's groups : J.50. and irreducible modular functions:
    Complex roots of an algebraical eq : C .6l.
    M .1 : N .443 : of 20% = 1, Mo.57. linear: LM .4 : of circles in space : C .93 .
    Complex variables, functions of : An. numerical : An.60 ; multiplication of,
    59,68,712,82,83: G .3,6 : J.54,73,83 : 61.
    M . 19 : Ž .82,10 . of 1st degree : A .32 .
    especially of integrals of d.e : J. in several unknowns: L .59 .
    75 ,76 . sol. by binomial factorial: Mém .4 .
    Composite functions of a higher order : transformation of modulus : Mél.
    G .2 . 2 : N .59.
    Composite numbers : - for construc with composite modulus : E .30.
    tion of factor tables : A .45 . of 2nd degree : 0 .622 : Mém .31 : re
    groups of : J.78 : LM .8 : Me.79 . duced forms: C .74 .
    “ Compteurs logarithmiques ” : C .40. of 3rd order and class : LM .16 .
    *Concavity and convexity: 5174 . higher, with real prime modulus :
    Concentric circles : LM . 14 . An.83 : J.31,54,99.
    3 quadrics, intersection of : E .39. resultant of systems of linear : C.88.
    * Conchoid : 5320 : A .55 : N .432. and trigonometrical functions : J.19.
    Concomitants of a ternary cubic : 202 + y = 1 (mod . p ): J.19.
    AJ.4 . a = 1 (mod. p ) : J.31.
    * Concurrent lines and collinear points Congruent divisors of a number, no. of:
    967 – 76 : A .69. A .37.
    on cconic and striangle : E .35.
    th.ycli
    coid
    Conical functions : M .18,19.
    * Conc coni : 995 : Q .11. * ConicalLMsurfaces : 5590 : A .63 : Ac.5 :
    * Cone : 1150 — 59,6043 : A .16 : L .61 : Me. | .3 , : M .3 .
    62. through 6 points, locusof vertex : J.92.
    and cylinder, superfices, tr : An.57 . * Conicoids : 5599 A .48 : Q .tg.c9, q.c
    general d .e of : È .18. and t.c 10.
    intersection of two : N .64. 50 - point : Me.66 .
    oblique: eq 5598 : J.2 : Me.80. *Conics : 403245030 : A .19,5 ,17,31, 32,60,
    * * *

    sections of : 1150 — 9. 68 : C .83 : G .1,2,3,21 : J.20,30,32 ,


    and sphere : 5652: thsMe.64. 45,69,86 : M .17,19 : N .42, 133,449,
    superfices of oblique frustrum : J.2. 45 ,71 ,752,82 : P .62 : Q .8, tg .c9 :
    through m points and touching 6 - m Z .18,21,23 .
    lines : LM .4 . anharmonic correspondents, problem
    volume of frustrum : Ac.41 : N . 13. of 5 conics and 5 lines : N .56 .
    Configuration
    J.86 .
    of 16 points and 16 planes : of Apollonius : L .58 .
    angles connected with : 4375 .
    *

    (3, 3) , and unicursal curves : M .21. arcs similar to : N .44.


    *Confocal conics : 4550 — 8, 5007, tg.e areas of (see also “ Sectors " ) : 4688,
    *

    5005 : J.54 : LM ,12,13 : Me.66 ,68 , 6097 – 6121 : N . 46 : t.cQ.2 .


    73 : N .80: Q .10 : TE.24 : Z.3. auxiliary circle : 1160.
    * *

    * Graves' theorem : 4555 : Griffith 's ext. centre : - coordinates of 4102 , 4267,
    LM .15 . t. c 4733 and 4742 : tg.eq of 4901 :
    * tangents of: 4555. thsand prs N .45 .
    *Confocal quadrics : 5656 — 72 : A .3 : CD . | * locus of 4520, 5028.
    4,5,99: G .16 : M .18 : thMe.72: Q .3. * chords of: 4315 ,4322 ; p.c 4337 and
    relation to curves and cones : CD .4 ,9. CD .1 : Me.66 : see also “ Focal.”
    volume bounded by three and the co cutting an ellipse at a given angle :
    ord . planes : A .36 . E .28 .
    INDEX. 857

    Conics : — chords of — (continued ) : Conics : equations of — (continued):


    intersecting : 1214,4317. intercept: 4498.
    moving round an ellipse : A .43,44 : * equations of parabola : 4201 ; t.c
    E .22 . 4775; p.c 4336 .
    * chord of contact : 4124,4281; 4699 — | * general: 130,4713 ; t.c 4656 with
    4721. 4689; tg .c 4974 with 5000 .
    and circle, intersection th : 1263 : ay = kB8 and derived equations : 4697

    *
    A .59 : N .64. - 4719 : Q .4.
    collinear relation to circle : Z .1. * ay = 44kß:? orS + LM1 % == 0 R2: 4699,4784 : N .
    and companion quadric : An.60 : JP.7. , & c . : 4707 .
    conjoint lines of: L .382. La® + MB? + Ny = 0 : 4755,'65 : Me.62.
    conjugate diameters : 1193 — 1213 , equation in p and p : Q .13.
    4346 ; ths 1278 – 85 : CM .1 : L .37 : equi.conjugates, gen .eq : 4491.
    N .42, ths 44 ,69 : Q .3. formulæ : J.39 : N .62 .
    parallelogram on : 1194 : 4367. from oblique cone: L .38.
    relation to ellipse
    conjugate points : Q .8.
    when equal: A .18. general equation ; cond. for a circle :
    4467 : t.C 4691 and Me.68 : Q .2 :
    construction of : 1245 ,4822 : A .28,43 : from eq . of axes : N .57.
    Ecircle
    .29 : Q .4 : N .59,73 : with help of cond. for an ellipse : 4464 ; t.c 4689.
    of curv. A .24 . cond. for a hyperbola : 4468 ; t.c
    from conj. diameters : 1253: A .52. 4689 : A .39.
    contact of: 4527— 33 : A .1,60 : C .78 : cond. for a rectangular hyperbola :
    Pr.34 . 4737 ; t.c 4690 ; tg.c 5000.
    at 2 points : Q .3 : N .74. cond. for a parabola : 4430 ; t.c 4696 ,
    do, with each of 2 conics or circles : 4735 , 4746 and 4775 ; tg .c 5000.
    4803 – 6 : CD .5,6 : E .31,34. cond . for two right lines : 4469,
    4-pointic with a quartic : M .12. 03 . 4475 , t.c 4662.
    5-pointic : 5190 – 1 : C .782: E .5,,23 : Sgeneration
    e of: N .75 : Z .23 .
    J.21 : P .59 . by a moving chord of a circle : A .34.
    *

    with surfaces : C .912 : P.70,74. Maclaurin 's method : 4830 : LM .4 .


    convexity from focal property : N .56. Newton 's method : 4829 .
    criterion of Mobius : J .89. graphic problems: N .80 .
    criterion of species : 4164 — 77; t.c Halley's pr : N .76 .
    4689: 5000. harmonically in - and circum -scribed :
    * curvative : centre and radius of : Q .18 .
    4534 – 49 : geo 1254 - 66 : thsMe. intersecting in 4 points : J.23.
    73 : N .79,85. intersecting a surface in 5 points :
    geo on : 1265 : A .17 . C .63.
    *

    chord of: 1259, -64: Q.6 : locus of with Jacobian = 0 : M .15 .


    *

    mid -point A .70 . limiting cases : 4465 -- 77 : Me.68g.


    as curves in space : tr A .37 : M .64. normals : 1171,sd5629 – 32: A . 16 ,24,32 .
    definitions : 1160. cn 43 ,47 : An. cn 64,78 : C .72,84 :
    degenerate forms: LM .22. J. cn 48 ,56 ,62 : Me.66 : Mél.2 : N .
    * diameters : 1214, - 35 : eq 4458 : cn 70 ,81 : Q .8 : 2 .11,18 ,26 .
    Me.66 : N .65. circle through feet of : N .80 .
    * director circle : geo 1217 , eq 4693 — 5 : cutting off the min . or max . arc or
    0.c E.40: LM .13 : N .79. area : N .44 .
    directrices: 1160 : trA.63 : LM .11. dividin g ellipse most unequally :
    eccentric values of coordinates : 4275 : E .29.
    CM .4 . eccentric angles of the feet of four,
    eccentricity : 1151,4200 . th : 4334.
    elementary formula : G .9 . equations of: 4286 , 4483, 4512.
    * *

    ellipse and hyperbola : 4250 – 96 . intercepts : 4294 : segments ; 4309,


    equations of : 4251,4273 ; p .c 4336 ; 4186 .
    tg .c 4663,4870 : J.2. least distance between two : A .21,38.
    general: 4400 ,4714 ,4719 ; t.c 4755 number of real : J.59 : N .70,72%.
    and 4765 ; tg .c 4664 and 4872 ; number cut by 8 lines in space : J.
    p.c 4493: t. A .51 : CP.4,5 : tuc 68.
    Ĝ .6,7 : Mém .52 : N .43,45,65. (See number
    59 .
    under double conditions : C .
    also “ Conics, general equation.” )
    5 R
    853 INDEX.

    Conica - continued) : 1 Conics : - targerisconsinua :


    octagram : LM .2. two from zy': $sil: 4438
    passing through given points and 44873.), A57 : te w51 - es 1: 21 :
    touching given lines : cn $ 31 A :13
    4 ). do. for parabola : 4215, en 1922 :
    5 points : cn 4531, eq 5124 : A.27 :
    A .9,21,61: An ) : 677
    ratio of lergiks, 1293 .
    quariratic for m : 4513 : parabo's.
    4 foci of a conic : Q .5 . 4 .
    4 points : 1 .66 : Q.2,8 : 2 .9. quadratic for abscissæ of points of
    4 points, envelop of: E.25 : Do. of
    axes, X .79.
    contact : 0312 : parabola , 4 . 10 .
    subtend equal ang.es at focas :
    4 points, locas of pole of a line : 1181, 12: 1 : C3 .2.
    4770). locus of z'y ': A .69 .
    4 points and touching a line : cn segments of : $307 .
    4333 ; A .65. at 4799.
    the points i, tan 0 , Ecoto:
    3 points and touching a circle
    twice : X250. tangent curves of : 2 . 15 .
    3 points and touching a line : Q.6. theorems: 1207 : A4 : J.16 : L . 4 :
    3 points with given focus : cu A .54 . M .3 : X .45 ,483,72,84: Q .4. ite :
    2 points and touching a line : Q .2. by Pascal, Desarques, Carnot, and
    2 points and touching 2 lines, locus Chasles , A .53.
    of centre : G .7 . conic and triangle, Q .5.
    four such conics, th : Q .8 . 3 circles touch a conic in A . B .C and
    1 point and touching 3 lines : Q.8 . all cut it in D ; A ,B ,C ,D are
    parameter of: N .43 . concyclic, J.36 .
    perpendicular from centre on tan : * three : 4707 , 4710 ; in contact, 4603.
    gent : 1195 , 4366 — 73 . Jacobian of : 5023.
    ditto from foci : 1178 , 4300. touching: — a conic and line, cond.:
    of 8 points : J.65. 3017.
    of 9 points : A .43 : G .1. a curve twice : J.45.
    of 9 points and 9 lines : G .7, 8. curves of any order : C .59.
    of 14 points : Me.66. five curves : C .58 ,.
    pencil of : M .19 . four lines : 4804 : locns of centre,
    and polar, Desarques' th : N .64. 4772, 5028 : locus of focus, 5029 :
    pole of chord joining xºv, 8294 : 4326 : N .45 ,67.
    parabola, 4218 . n lines : V .61.
    properties of: 1274 : A .4,25,70 : p .cJ. a group of lines,and having a given
    38 . characteristic and focus: A .49 .
    quadrature of : TE .6. a quintic curve in 5 points, no. of :
    and quadrics : A .30 : L .42 : N .58.,66., N .66 .
    ths 73 : geo interpretation of two circles twice : 4806.
    variables, N .66 . two conies twice : 4803.
    * rectification of : 6071, 6084 : P .2 : two sides of the trigon : 4784
    TI16 : Z .2 . 4808 .
    reduction of u , u to the forms transformation of: G .10.
    ma + y + 7° = 0), ac? + By + yz = 0 : | * two: 4936 – 5030 : N .58.
    4995 . with common chords or tangents :
    series of : A . cn 67 and 68. 4700 — 5 .
    seven points of : C .94 . common elements, cn : A .68 .
    similar : 4522. with common points and tangents :
    six points of : A .62 : J.92. 4701 – 7 : LM . 14 : Z .18 .
    systemsof : C .62,65 : M .6 : N .66 : of 2, | at infinity : 4715 — 6.
    Q .7 : of 4, L .54 . common pole and polar, cn : 4762.
    and orthogonal lines : N .71. condition of touching : 4942, t.c
    and quadrics : 2 .6. 5021.
    tangents or polar: 1167 — 9, 4280 — 5, intersecting in 4 points : 4700 : A .
    4790 : gn .eq 1478 : A .61 : cnCD . 32 : at 00 , A . 16 .
    3 : CM .1, p.eq 4 : N .79 . points of intersection : tg .e 4973 :
    intercepts on the axes : 4292. cn A .69.
    two at the origin , eq : 4489. reciprocal properties : E .29.
    INDEX . 859

    Conics : - two - (continued ) : Continued fractions --(continued ):


    reduction to 202 + y2 + z = 0 and 11 and 1 1 1 1
    ax ? + By + yz? = 0 : 4995 . at at “ a+ b+a+ b+ .
    six chords of, eq : 4941. b b+1 b > a + 1 : Mél.l.
    tangents of,four: eq 4981 : J.75 : Q .3. a + a + l + a + 2 + & c. "
    under 5 conditions :4822 - 43 : L .10,59.
    6 conditions, cn : C .59. 1 1
    7 conditions in space : C .61. æ + + 1 + x + 2 + and
    Conjugate : - functions : apLM .12 : 1 1
    TI.17. 2x + 1 + 2x + 3 + 2x + 5 + '
    lines of surfaces : CD .9.
    points : 1066 ,5184 : in ellipse, A .38.
    point-pair of a conic : 2 , 17 .
    2 + 1 1. and a +?q + 4.,
    at
    : E.36,33.
    ascending = apr.
    algorithm :prA .60 : 2 .21.
    tetrahedrons of a quadric , each ver
    tex being the pole of the opposite i + bP -2 : 168 : J.
    side : 1 .60,83. 69,75.
    triangle of a conic : N .83. do. ap. to solution of trinomial eq :
    Connexes (1, n ) corresponding to de : A .66 .
    A .69 . combinator representation of the ap
    in space : Mél.6 . proximation : A .18 .
    linear : M . 15 . development : Ac.4 : C .87 : Mém .9.
    of 1st order and class in simple invo. for et and log (1 + x ): CM .4 .
    lution : G .20. for e exp. a + bx + ca ? + : C .87.
    of 2nd order and class : G .19.
    analoguein anal. geo.ofspace : M .14.
    Cono-cunei : A .2.
    fora 1-1-- 2 --4++x2--94 -- 22 =9- 1°+ +. .. J.27.28 .
    Constant coefficients, theory : 1.23. for U sin æ + V cos x + W ; U , V, W ,
    Constant functions and their deriva - | polynomials in æ : J.76.
    tives : A .15. for (m + vn ) + p : N .45.
    *Contact and circle of curvature : 4527, for powers of binomials : CM .4 : Mém .
    5188 , 5134 : A .30 . 18 .
    * Contact of curves : - 5188 : cir. of cury . infinite : A .332.
    A .53 : C .32 : J.66 : P .62 : Pr.11 : numerical values : Q .13.
    Q .7 . Eisenstein 's, TE.28 : Wallis's, Mém .
    with a parabola : Z .2. 15 .
    with faisceaux of doubly infinite periodic : A .19 (2 periods), 33 : G . 16 :
    curves : C .83 . J.53 : N .42,43,45,462.
    with surfaces : L.78 : with triangles , reduction of: J.46 .
    M .7. reduction of a square root to a : 195 :
    Contact: - of lines with surfaces : L . A .64 : J.31. -
    37 : Q .1,17 : Z .12. do. of a cube root : A .8 .
    of an implexe with an alg. surface : do. of an nth root : A .64.
    C .84. Contingence, angle of : 5725 .
    of quadrics : LM .5. Continuity , principle of: CP.8 : in rela
    4-pointic on an algebraic surface: M .9. tion to Taylor's and Maclaurin ' s
    of spheres theorems, L .47.
    of surfaces: :JP.2
    J.4 :. JP.15 : P.72,74,76 : * Continuous functions : 1401 : A .15 :
    Q . 12. Ac.5, : C .18 ,20, and discontinuous
    of 3rd orderbetween 2 surfaces: C.749. 40 , of integrals of d .e 23 : TA .7 .
    problem : of Apollonius, A .66 : spheri. Continuous manifoldness of 2 dimen
    cal, At.192. sions : LM .8 .
    transformation : 0 .85 : M .23. * Contragredients : 1813 .
    Contingence angle of : see “ Torsion ." Contraposition : E .29 .
    Continuants : AJ.1 : Me.79. * Contravariants : 1814 : G .12, of 6th deg
    * Continued fractions : 160 – 87 : A .18,33 , 19.
    55 ,69 : An.51 : gzC .99 : CM .4 : th * of two conics : 4990 and 5027.
    E .30 : G .10 ,15 : J.6 ,8,11, and ap , * Convergents : 160 – 87 : gzC .98 : CD .5 :
    18,53,80 : L .50 ,58 ,65 : LM .5 : Me. J.37,57,58 : N .46 .
    77: Mém. ap toi.c 9,13: Mo. 66: N . Convex polygon , intersection of diags :
    42,tr49,56,66 : Q .4 : geo Z .12. N .80.
    860 INDEX.
    *Coordinates, transformation of : 4048 — 1 Correlative or reciprocal pencils : M .12.
    51, 4871, 5574— 81. Correspondence : - principle of : geo
    *Coordinate systems: 4001 - 31, 5453, thAn.71 : extC .78,80,ex83,853 :
    5501 - 6 : G .16 : J.5,45,50 : LM .12 . N .46 .
    ap . to caustics : An.69. of algebraic figures : M .2,8.
    areal: 4013 : Me.80 ,82 , gn eq 81. application to Bezout's th C .81 ; to
    axial: N .844,85 : Q .10. curves, C .72 ; to elimination, N .
    biangular : 5453 — 73 : Q .9,8, 13. 73 ; to evolute and caustic, N .71 .
    bilinear : ap 5341 : A .32. complementary theorem to : 0 .81.
    bipolar : N .82. determination of the class of the en
    bipunctual : AJ.1. velope and of the caustic of a
    Boothian : see tangential. curve : 0 .72.
    determination of the degree of the
    Cartesian : 4001.
    dual inversion of c.c and t.c : Q .17. envelope of a curse or surface of
    central : Z .20 . n parameters with n - 2 relations :
    curvilinear : A .34 : An.57,64,683,70, C .832.
    73 : C .48,54 : CP.12 : G .10, and determination of no . of points of inter
    i.c 15 : L .40,51,82 :Mém .65 : Q .19 : section of 2 curves at a finite
    on a surface in space,
    and any An.692, distance : L .73 : C .75.
    71,73 : including angle, J.58. determination of the number of sola .
    eccentric values of : 4275 , 5638 : CM .4 . tions of n simultaneous algebraic
    elliptic : A .34,40,: J.85 : Q .7 : Z .20 . equations : C .78.
    Fuchsian functions of a parameter : determination of the order of a geo .
    C .92. metrical locus defined by alge.
    hyperelliptic coordinates : J.65. braic conditions : C .82,842.
    one-point intercept : 4026 . forms : M .7 .
    two- point intercept : 4025 . multiple in 2 dimensions: G . 10.
    linear : Z .21. of curves : C .62 : P.68 : Q .15.
    mixed coordinates : A .13. of two planes : LM . 9.
    parabolic : C .50 . of points : LM .2 : C .62 on a curve :
    parallel : N .84,,85 . Q .11 on a conic : M .18 on two
    pedal : Me.66 . surfaces.
    pentahedral : Me.66 . for groups of n points and n rays :
    of a plane curve in space : LM .13. M .12.
    polar : 4003,sd5506 : Me.76 . of two variables (2,2): Q .12.
    polar linear in a plane : Z .21. of 2nd deg . between 2 simply infinite
    quadrilinear : Me.62,64 . systems: An . 71.
    quadriplanar or tetrahedral: sd5502, Correspondent values, method of : P .
    ap A .53 : Q .4 . 1789.
    surface : C .65,81. Corresponding points : - in some in
    * tangential : 4019, 487044915, 5030 : volutions: LM .3.
    Me.81 : Pr.9 : Q .2,8. on two curves : M .3.
    tangential rectangular, or Boothian : on two surfaces : 0 .70 .
    4028. Corresponding surface elements : M .11 .
    tetrahedral point coordinates : Z.8. Cosmography, graphicmethod : N .82,79.
    triaxial : A .64. * Cotes's theorem : circle,821 ; areas, 2995 .
    trigonal : G .14 : Q .9. analogous ths : CM .3 .
    trilinear : 4006 : 8 .39 : Me.62,64 : N . Counter-pedal surfaceof ellipsoid : AJ.4 .
    63: Q .4,5,6 : conversion to tan - | Coupures of functions : C .99.
    gential, 4876 . *Covariants : 1629 ,4936 – 5030 : C .80 ,81,
    trimetrical point: A .67. th 90 : G .1,20 : Q .5,16 : J.47,87 ,
    Coplanation : — Z .11. 90 : thM .5 : N .59, : ap to i.c C .56 .
    of central quadric surfaces : Z.8. binary : G .2.
    of pedal surfaces : 2 .8 . of binary forms: An.58 : C .82,86 ,87 :
    Coresolvents : Q.6,10,14 : non -linear, G . 17 : L .76 ,79 : M . 22 .
    TN .67. of binary quadrics, cubics, and quar
    Correlation of planes : J.70 : An.75,77 : tics : An.65 : J.50 : Q .10 .
    LM .5 ,6 ,8,10. ofbinary quantics : E.31: Q.4,5,17.
    Correlative figures, focal properties: of a binary quartic: J.53: quintics,An.
    LM .3 . 60.
    INDEX. 861

    Covariants — ( continued) : Cubic curves - (continued) :


    and contravariants of a system of 48 coordinates of : P .81 ii.
    simultaneous forms in n varia with cusps : A .68 .
    bles ; to find the number of: C .84 . degeneration of : A .4 : M .13,15.
    of quantics : An.58 : binary, Me.79. derivation of points : LM .2.
    of a septic, irreducible : C .87. with a double point : M .6 : with two,
    sextic : G .19. M .3 .
    of a system of binary cubo-biquadra . with double and single foci: Q.14 .
    tics ; number of irreducible co Geiser's : J.772.
    variants : C .872. generation of : G .11 : J. 36 : M .59,6 :
    of a system of 2 binary quadratic by a conic pencil and a projective
    forms; number of : C .84. ray pencil, Z .23 : linear, J .52.
    of ternary forms: G .19, and higher curves : A .70 .
    * of two conics: 4989, 5026 : of three, mechanical construction of : LM .4.
    Q . 10. number of cubic classes which belong
    Covariants and invariants : An.60 . to a determining quadratic class :
    of binary forms : An.58,59,61,83 : C . A .19 .
    66 ,69 : reciprocity law , C .86 . nodal, tangents of : LM . 12.
    of a binary octic, irreducible system : of third class with 3 single foci : Q .
    C .86 ,93. 14 ,162.
    of a binary quintic : 0 .96 . 18 points on : E .4 .
    of a binary sextic : C .962. inflexion points of : J.38 : M .2,5 : N .
    as criteria of roots of equations: An . 73,th83,85 .
    68 . 12 lines on which they lie in threes :
    Cribrum or sieve of Eratosthenes : N . E .29.
    43,49. rational : A .58 : G .9.
    Criticalfunctions: see“ Seminvariants.” referred to a tetrad of corresponding
    Criticoids and synthetical solution : E . points : Q . 15 .
    9, 26 ,32. represented by elliptic functions :
    * Cubature of solids: see " Volumes.” JP .34.
    Cube numbers, graphic cn of: E .43 : and residual points : E .34 .
    Me.85 : tables to 12000 , J.42 . resolved into 3 right lines : M . 14 .
    Cube root extraction : A .22,64 : J.11: and right lines depending on given
    N .44 ,583 : TI.1. parameters : J.55 .
    * Cube roots, table of (2 to 30): p.6 . 16 cotangential chords of : Q .9.
    Cubes of sums of numbers : N .71. and surfaces : J.89.
    Cube surfaces : Pr. 17. synthetic treatment : Z .21.
    Cubical parabola : Q .6 ,9. tangential of : P.58 : Pr.9.
    metrical properties : M .25 . tangents to : cn E .25 — 7,ths28 : LM .3 :
    * Cubic and biquadratic equations : 483 Q .3 .
    501 : A .45 : No.1780 : An.58 : with a double point or cusp : M .1.
    C .geo41,90 : JP.85 : L .59 : M .3 : forming an involution pencil : LM .
    N .79 : 2 .8 ; Arabic and Indian 132.
    methods, 15 . their intersections with cubics or
    Sturmian constants for : Q .4. conics : C .41: TI.26 .
    mechanical construction : LM .22. through 9 points : C .cn36 ,37 : L .54 :
    Cubic and biquadratic problems: An.70%. two cubics do., CD.6.z.
    Cubic classes which belong to a deter: through 8 points : Q .5.
    mining quadratic class, number through2circular pointsat co : cnZ.14.
    of: A .19. transformations of : C .91.
    Cubic curves : An.73 : AJ.5,15 ,27 : C . | * Cubic equations: 483 — 91: A .1, 3 — 7 ,11,
    37 : thCD .7 : CP.11 : G .1.,2, 14 , 22, 252, 32, 37,41,42,413, prs 47 ,68 :
    23 : J.11,32,34,ths42,63,geo78, 90 : An.55 : C .num46 ,85 : CD .4 ,6 : E .
    L .13,44 ,45 : M .15 : t.cMe.64 : Mo. 35 : G .12,16 : J.27,56 ,90 : LM .8 :
    geo56 : N .50,67,geo723 : P .57 ,58 : M .3 : Mém .26 : N .45,th52,th56 ,64,
    Pr.8,9 : Q .12,5 : Z .geo15 ,22. 66 ,703,75,78,81,84 : TE .24 : TI.7.
    classification of : Q .16 . See also “ Cubic and biquadratic
    and conics which touch them : J.36 . equations."
    coordinates, explicit functions of a and division of angles : A .15 .
    parameter : J .82. in a homographic pr of Chasles : C .54 .
    862 INDEX .

    Cubic equations - (continued) : Curvature - (continued ) :


    irreducible case : A . 30,39,41,42 : No. 1 * at a cusp : 5182 : N .71.
    1799 : AJ.1, : 0 .58 : J.2,7 : L .79 : * at a double point: 5187 : Q .3.
    N .67 : in real values, A .49 : by dual, evolute and involute : Q . 10 .
    continued fractions, A .2. of an evolute of a surface : C .80.
    roots : - condition ofequality : E .30. of higher multiplicity (Riemann ):
    definition of : thA. 31. 2 .20 , 24 .
    geometrical construction of: C .44, || of higher order : 5188 — 91: M .7,16.
    45 : Arab . m .s.s, J.40.. of third order : 0 .26 .
    integral : N .752. of intersection of 2 quadrics : An.63.
    power-sums of : C .54,55 . mean : C .92 .
    squares ofdifferencesof: An.56 : Q .3. || at a multiple point: 5187 : C.68.
    solution of : - Cardan's formula : of orthogonal lines : JP .24 .
    484 : A .14,-10 ,8222: Me.85, parabolic : 5818 .
    Cockle's : CM .2,3. of a plane section of a surface : C .78 :
    trigonometrical : 489 : A .19 : N .61, Z .17 .
    67,71 : TE.5 . spherical : A .25.
    hy differences of roots : J .42 . * Curvature of surfaces : 5818 - 26 : A .4,
    by continued fractions : A .10 ,39. 20 ,41,57 : An .f and ths61,64 : C .
    by logarithms : C .66 . th 25,49,602,ths66,67,68 ,geo74,84 :
    J.1,3,7 ,8 : JP . 13 : L .41,729 : p.c
    by x = + + h : C.60. Me.71 : Mél.3 : Q .12 : 2 .27 .
    mechanical: 5429 or C .79 . average, specific, integral, & c. : 5826
    numerical : 0 .44. - 30 .
    Cubic forms (see also “ Binary cubics”): | axis of curvature of envelope of a dis
    thsJ. 27 . placed plane : C .70 .
    ternary : J.28,29 : JP.31,32. approach of 2 axes of finite neigh
    quaternary : transfJ.58 : P .60 : Pr. 10 : | bouring curves : C .86 .
    division into five, J.78 . circular and spherical: see " Tortuous
    Cubic surfaces : Ac.3,5 : ths An.55 : C . curves.”
    972,98 : G .22 : J.53 ,65,68 ,69,88 ,89 : constant : J.88 : G .3 : mean , C .76 , L .
    M .6 : Mo.56 : N .69: P.69 : Z .20 . 41,53; neg., C .60,M .16 ; pos.,G .20 ;
    total, C .97 %
    classification of : M . 14 . Euler's theorem : gzC .79.
    double sixers of : Q .10 . Gauss's thC.42 : analogy M .21 : Q.
    with 4 double points : M .5 . 16 .
    “ gobbe ” : G . 14, 17, 21. ap . to aneroid barometers : C .86 .
    hyperboloidal projection : G .2. indeterminate : CD .7.
    27 lines of : 0 .52,68 ,70 : J.62 : L .69: and inflexion : trA. 19.
    M .23 : Q .2. integral : 5826 .
    27 lines and 45 triple tangent planes and lines : An.53,59 : L .41.
    of : An.84 : and 36 double sixers mean = zero throughout : M0.66 .
    Q .18 . and pencils of normals : C .70.
    locus of centre of quadric through 8 and orthogonal surfaces : P.73 .
    points : Me.85.2. of revolution : L .41: 2.21,22.
    model of: CP. 12. skew : 2 .26 .
    polar systems : M .20. sphereA .43ofmean curvature ofellipsoid :
    properties of situation : M .8. .
    in quaternions : AJ.2.
    reciprocal
    73.
    of Steiner 's surface : N .72 , * Curvesand(see“ also “ Curves algebraic "
    Curves and surfaces " ) :
    singular points of: P.63. 5100 : A .2,16 ,32 ,66 : An.53,543 :
    C .geo72 ,91 : J.14 ,31,349,63,643,70 :
    triple tangent planes : CD.42. L .38,44 ,ths57 and 61 : M .16 : N .
    Cubo-biquadratic eqs., no. of irreducible p .c61,71,77 ,853 cn from p.c.
    forms: 0 .87. from Abel's functions, p = 2 : M .1.
    * Curvature : 1254 – 8 : 5134,5174 : A .1, Aoust's problem : A .2,66 .
    28 ,43 : J.81 : Me.622,164,72,75 : arcs of, compared with lengths : JP.
    N .th60,69: Q .t.c8,12. 23.
    * circleJ .68of:. 1254 – 5, 5134 : A.cn30,37: of “ allineamento " : G .21.
    analytical method : LM .9, 16 .
    INDEX. 863

    Curves — (continued ) : Curves - (continued) :


    whose arcs and coordinates are con a family of: N .72 .
    nected by a quadratic equation : four, with two common points : Q .9.
    J.62. generation of : geoJ.58,71 : M .18 .
    whose arcs are expressible by elliptic by intersections of given curves :
    or hyperelliptic functions of the Z .14 .
    1st kind : Z .25 . by collinear ray-systems: Z .19.
    argument of points on a plane curve : geometrical : A .37 two laws, C .84 .
    LM . 15 . relation to harmonic axes : C .734.
    bicursal : LM .4,7. “ gobbe” : of zero kind ,G .11 : rational,
    with branches : imaginary, CM .1,Q .7 : G . 9,12 ,
    infinite, Q .3. higher plane : A .70 : L .61,63.
    2 characteristics defining a system homofocal : N .81.
    of algebraic or transcendental defined by intersecting conics : C .37.
    curves : C .78 . intrinsic equation : CP.8 : Q .5.
    least chord through a given point : joining two points : pr L .63.
    A .23. with multiple points : 0 .62: L .69.
    class, diminution of : N .67. · with three of higher degrees, cn
    closed : see “ Closed curve.” An.58.
    of 2-point contact with a pencil of n -tic with m .p of n - 1th order : C.
    curves : M .3. 80 : N .76 .
    of 3-point contactwith a triply infinite network of: C .67 .
    pencil of curves : M .10 : 4 -point pencils of: A .65: of 3rd order, Z .13 .
    do., LM .8. P + P24 = 84 : E .11.
    whose coordinates are functions of a with a constant polarsubtangent : N .
    variable parameter : Me.85 : ellip 62.
    tic, J.64 : N .68. with6 several “ points d'arrêt" : N .
    cutting others in given angles or in 0.
    angles whose bisectors have a in a power-series of sines: J. 3.
    given direction : 0 .58,83. of pursuit : 5247 : C .973 : N .83.
    and derived surfaces : An.59,61. of “ raccordement ” : JP.12.
    derived from an ellipse : A .10 . rational : A .56 : G . th15 ,16 : M .9,18.
    determination from their curvature: generation of : G .12 .
    . P .83,84 . reciprocal of: J.42.
    from property of tangents : A .51. spdu : J.1.
    determination ofthenumber of curves JR :
    ee r which have a contact
    of degrree
    of deg n < mr, with an m -tic, of section : A .43.
    and which satisfyr (r + 3) - n of a series of groups of points, ths :
    other conditions, and similar C .732.
    problems: C .635. with similar evolutes: Me.66.
    defined by a differential equation : C . | * singularities of : 5187 : An.71 : 0 .78 .
    81,90,93, 98 : L .81,82 . 80 : CP .9 : J.64 : JP .7 : L .37,45 :
    do. algebraic and an analogous LM .6 : M .8 – 10., 16 : N .50,80 ,813 :
    space theorem : L .762. Q ,2,7 : higher singularities, J.64 :
    L .70 .
    of the species 1 : C .97 .
    diamet24 ers
    : p = A sin w , N .76 .
    of: L .49: N .71 : and sur sextactic points of : P .65 .
    face , C .60. on surfaces : see “ Surface curves.”
    eq obtained from tangent : N .45. systems of: An.61 : G .13 : Mo.82 :
    whose16 :equations theory , C .632,94 .
    03 = u (uare- 1): y(u= — ”Væ,
    a ) (u A-.14,
    y), and surfaces : A .73 : Ac.7 : L .65 .
    x and y constants, C .93 . tangential polar eq of: Q .1.
    70 = a sin 0 , A .48 : y = 1 (2 ), 2323. theorems or problems: A .pr1,313,prs
    linear functions of the coordinates : 37 and 12 : G .l, : J.1 : M .14 : Q .3.
    N .65 . re arc CP and chords CP, PM , CM
    equidistant, tangents to : cn Z .28 . Mém .10 .
    whose evolute and involute are equal : to describe curves which shall have
    C .84 . equal arcs cut off by a fixed pen .
    extension to space : C.85. cil of lines: Mém .10.
    864 INDEX.

    Curves - (continued ) : Cusps — (continued ):


    re lines drawn at all points of a construction of 8 cusps of 3 quadric
    curve at the same inclination to surfaces when 7 are given : J.26 .
    it : C .74. * keratoid : 5182.
    tracing apparatus: LM .4. * ramphoid : 5183.
    transformation of : CD.8 : LM . 1 : Cyclic : - carves : A .37 : Z .cp2, 26 ,27 .
    scalene Q .13 : M .4, 20 ,21 : of 14 functions : A .69, and hyperbolic 37 .
    ics which cut a quartic in the interchanges (higher algebra ) : Man .
    points of contact of its double 62.
    tangents : J.52. projective groups of points : M . 13,20.
    and transversals : J.47. number of do. in a space transf. :
    under given conditions : P .68. C . 90 .
    Curves algebraic (see also “ Curves and surfaces : Z .14 .
    surfaces ' ) : C .99,ths60 and 80 : systems: C .76 .
    CM .4 : G .1,4,5 : J.12,47,59 : N .50 ,
    Cyclides : N .66 ,70 : Pr.19 : Q .9,12.
    81,cn83. reducible : LM .2.
    represented by arcs of circles : JP.20. and sphero -quartics : P .71.
    with axes of symmetry : N .80 . *Cycloids : 5250 : A .13 : N .52.82.
    of 2nd class and 2nd order : G .1. and trochoids on surface of sphere :
    of 3rd class and 4th order : G .4 . Mém .22 : Q .19.
    of 3rd class and curves of 3rd order : surface of, th of Archimedes : Me.
    J.38 : L .78 . 84.
    of 4th class with a triple and a single Cycloidal curves : Z.9.
    focus : Q .202. Cyclosis in lines : LM .2 .
    of 6th class : Ac. 2 . Cyclotomic functions: C .903:
    of class n and order m , two laws : | *Cylinder, frustrum of : 6048.
    C .85 . circumscribing a torus of revolution :
    common points, a system of : J.54. C .45 .
    generation by right lines : J.42. and cones, intersection by spheres,
    and homothetic conics, ths : J.53 . ths : J.54.
    lemniscatic : An.58 . and hemisphere : P .12.
    manifoldness of : M .10 . Cylindrical functions : A .56 : An.73 :
    mechanical construction of an n -tic : M .5 ,16 : and di M .8 .
    5407 : LM .7 . of 1st and 2nd class : M .1.
    with a mid -point : J.47. J (« ) analogous to the sphericalfunc
    number of points of contact : C .82. tion Pm (cos ) u : M .3.
    number of intersections : C .76 : M . 15 . representing a function of 2 variables :
    projective involution : M .3. M .5 .
    remarkable group of : M .16 . *Cylindrical surfaces : 5591 : LM .33.
    species determined : M .23. quadrature of : A .9.
    symmetrical expression of constants : Cylindroids : At.19,39 : Me.80 : 2 .25.
    Q .5 .
    theorems: two metric, M .11 : Mac. Decimal fractions : approximation
    laurin 's, N .50. by : N .51.
    Curves and surfaces : - M .8 : N .59 : errorC in.40 addition
    : N .56 .
    of non -terminating :
    gnthsC .45 and G .4 : algebraic,
    An.77 ; J.49 and L .55 . repeating: A .16 ,33,56 : G .9 : Me.85 :
    “ arguisiane” : G . 12. Mél.5 : N .42,49,74.
    curves having the same principal 1
    p-
    normals and the surface which where p is oneof the first 1500
    the normals form : C .852. primes : A.3.
    of same degree, a common property : Definite integrals : see “ Integrals."
    G .8 . Deformation : - of conics : 2 .25 .
    satisfying conditions of double con . of a cache-pot . N .81.
    tact : C .89 . of a one-fold hyperboloid : E .30 .
    Curvilinear angles , ths : L.44,45. of surfaces : C .68,70 : G .16 : JP.22 :
    Curvilinear triangles : A .61. : N .45 . L .60.
    Curvital functions : C .60. * De Moivre's theorem : 756 : A .6 , 11.
    Curvo-graph : A . 1. Demonstrations, reduction to simplest
    *Cusps : 5181: Mém .22 : Q .10 . form : C .83.
    INDEX. 865

    Derivation : — applied to geo prs : Determinants - (continued ):


    An.54 . function in analysis for a certain de
    of analytical functions , gz : G .229. terminant of n quantities : C .70.
    of a curve : An.52. gauche (apı = - Qqp ): C .88,892 : CD .9 :
    Derivatives : see “ Differential coeffi J.32,38,50,th55 : L .54.
    cient.” involving 1, & c.: Q .15 , 16,17.
    * Arbogast's : 1536 : CD .6 : 1.12 : L . of lower determinants : J.61.
    82 : extMe.78 : P .61: Pr.11 : Q . of minors of given determinant : C .86 .
    4 ,7 . minor : 554 : G .1.
    Schwarzian : CP.132. multiplication of : 562, 570 : A .14,59 :
    Descriptive Geometry : An. 63 : L.39 : 11.52 .
    N .522,56. number of terms in : LM .10.
    * Detached coefficients : 28. partial: C .97.
    Determinants : 554 : A .44,56 ,65 : apAn . persymmetric : Me.82 .
    57 : J .22,tr512,72 ,73,74 ,89 : 0 .86 : with polynomial elements : Me.85.
    CP.8 : G .1,4 ,8,9,10 : L .84 : LM . of powers : AJ.4 .
    10 : Me.62,78 ,794,83 : N .51,69,ap quadratic forms of: J.53,89 : L.56 :
    70 : Pr.8 : Q .8 : TE .28 : Z . 16 . N .52 .
    and algebraic “ clefs " : C .36 . ditto of negative dets. : J.37 : L .60 :
    of alternate numbers : LM .11. M .22 : Mo.62,75 .
    application to : — algebra and of rational fractions : Me.82.
    geometry , A .51,50,53. resolution into quadratic factors of a
    contact
    60.
    of circles and spheres : N . det. formed from two circulants :
    Me.82.
    cylindrical surfaces : A .58. of the 16 lines joining the vertices of
    equations : Q .19. two tetrahedrons : J.62 .
    geometry : J.403,49 ,77. of sixth order : Me.84 .
    surfaces of revolution : A .582. signs of the terms: 557: E .29 : Me.
    arithmetical: G .23 : LM .10 : Me.78 : 80 .
    Pr.15. skew : Q .8 , 18 .
    of binomial coefficients : Z .24. of squares of distances of points : Q .
    catalogue
    18 .
    of papers and treatises : Q . 11.
    of Cauchy (" aleph ” ): G .172. Sylvester's det.and Euler's resultant:
    combination of : CD .8 . An .59.
    symmetrical : G .1 : J.82 : M .16 : th
    combinatory analysis of : C .86 . Me. 85 : Q .14 ,18 .
    *

    composite : 555 : J.88 ,89. and Lagrange's interpolation : LM .


    compound : 555 : AJ.6 : LM.14 : Me. 13 .
    82.
    in conics : J.89,92 . ap to a pr in geo : Z. 20 .
    with continued fractions : J.69. of nth order and n - 1th power X
    sq. of a similar determinant :
    cubic : G .6 : LM .13 : and higher, 11. AJ.4 .
    cycle of equations : G .11. theorems and problems : AJ.3 : An.
    of definite integrals : L .52 : Z .11. pr60 : G .2,412,62, 12,16 : J.pr66 ,pr
    development of : An.58 : N .85 : in 84 : L .51,54 : M .13 : Me.79 : N .65 :
    binomials, G . 10 : in polynomials, Q .1,pr2, 15 : 2 .7,pr318 .
    G .13,15 : and ap to resultant of
    2 eqs, G .21. of product, L .60 .
    division problems: A .59. unimodular, cn : 2 .21.
    double orthosymmetric : Z .26 .
    and duadic disynthemes : AJ.2 . for verifying a system of d.e : C .23.
    elements of : G .10, 15 . with a diagonal of zeros : Me.73.
    equation in which apa = dop : C .41. Developable cylinders, motion of : Man .
    of even order,analogy between a class 84 .
    of : J.52. Developable surfaces : A .69 : M .18 : Me.
    of figurate numbers : G .9. 77 : Q .6 .
    functional : CD.9 : J.22,69,70 ,77,84 : circumscribing
    15
    given surfaces : Z. 13 ,
    M .1, 18 : Me.80 : Q .1 : of binary .
    forms, C .92 : of a system of func circumscribing 2 quadrics : C .67,ths
    tions, L .51. 54,gz63 : ČD.5.
    5 s
    866 INDEX.

    Developable surfaces - (continued) : | Differential coefficients of nth order


    of a conical screw : A .69. (continued ):
    edge of regression : L .72. 1 — . 1467.
    of (a’ + a%)" : 2860 ';. 1tæ
    of first 7 degrees : J.64.
    through a given curve which develops * (1 v-x2)n-4 (Jacobi): 1471, A.4 ;
    into a circular arc : L .56 . (a? — b*e*), A.3.
    through a gauche curve : C .97.
    mutual : J.19.
    quintic : 0 .54 : CD.5 : G .3.,. 1 : 1469; 1470, J.8;
    of surfaces having principal lines of | * (a + bæ + cx )", 2858 ;
    curvature plane : C .36 . (x?tax + b )-m , A.8.
    Development of tortuous curves : prs tan « : A .12 ; cos”.e, A .9o ; nos
    sin
    ax ,
    Mém . COS

    Diacaustic of a plane : N .75. 1461, N .62.


    Diagonal scales : LM .6. sin
    Diametral curves : CM .2. * 12, 2862 ; sin -la, 2854 — 5 ;
    of constant sectional area , prs: N .43. tan- la , 1468, apN.9.
    *

    Didon , proposition of : C .86 . ear, eary : 1463 – 4 ; par , CM .2.


    *

    Differences : — and differential-quo ear": 2861, A.30 ; eas cos bæ, 1465.
    *

    tients : A .49,53 : N .69. an - log & : 1466 ; etcosa cos (x sin a ),


    equations of mixed : JP.6 : N .85 . 2856 .
    parameters of functions : C .95 .
    * Differential Calculus : 1400 — 1868 : J. of nth order with a = 0 in the result :
    119,12,133,142,152,16 : Me.66, : Q .4. tan -t« , sin -la, (sin -"x)2, 2865 — 9 ;
    reciprocalmethods : CD.7,8.
    * Differential coefficients or differential
    quotients or derivatives (see also
    cosm sin-la,cosm cos '#, 2871—7;
    • Differentiation ” ) : 1402, 1422 — (1 + x2)+ " cos
    sin m" tan -la , 2883 – 7 ;
    46 : Pr.12.
    of algebraic functions : Mél.i. , aPear cos bx, 2889 — 91 ;
    of log x and a *: A .1 ; of cand at: N . sin (æ or y ) + cos e A3
    53.
    of { f (x )} " : M .3 ; of y exp (3*) : A .22. 1 + 2y cos x + ya
    calculated from differentials : AJ.16 . | * Differential equations (D .E .) : p .460,
    of a composite function : 1420 : nth , 3150 - 3637 : A .1,52,67 : AJ.4 :
    *

    G .13. An.50 ,: C .8, 15 ,23,29,42,54,70 ,83 :


    * of exponentials and logarithms: 1422 CM .3 : E .9 : J.1,36 ,58,64,66 ,74 ,75,
    -- 7 : A . 11 : N .50,52,85. 76 ,78 ,86 ,91 : L . 38 ,52 ,56 : LM .4.10 :
    of a function of a function : 1415 : A .9 : M .8 ,12,25 : Man.79 : Me.81: Mém .
    *

    in terms of derivative of inverse 30 : Mo.84 : N .72,80 : Pr.7 : TI.13 :


    function , Mém .57. Z .4 , 16 ,27 .
    of a function of two independent Abel's theorem : J .90.
    variables : 1815 . algebraic : An.79 : C .86 .
    of irrational functions : P .16 . with algebraic integrals : J.847.
    of products whose factors are con approximate solution : C .5.
    secutive terms of a series : Me.31. by equations of differences : L .37.
    ratio to the function at the limit 0 : by Taylor's theorem : 3289.
    J .74 . ofastronomy : C .9 ,29 : P. 4.
    successive or of nth order : 1405 , 1460 asymptotic methods : C .94 : Q .5 .
    – 72 , 2852 — 91 : A .1,4 ,7 : An.57 : Bessel's numerical solution : 7 .25.
    G .18 : M .4 : Z .3. Complete primitive : 3163 – 6 : J.25 .
    of a sum ,product,or quotient : 1411. no. of constants : CP.9 .
    independent repres. of: M .4 . with complex variables : M0.85 .
    of a function of a function : G .13 : of a conic : E .38.
    n = 4 , 1419. continuous and discontinuous integ
    of Cfunctions
    .93.
    of several variables : rals of: 0 .29.
    for a conical pendulum : A .84.
    of a logarithmic function : A .8 . relation between its constants and
    of a product : 1460 , 1472. the constants of a particular
    and summation symbols: J.33,ths32. solution : C .92.
    INDEX. 867

    D . E .- (continued ) : D . E . - (continued) :
    of curves having the same polar sur. | * Parseval's theorem : 3628 .
    face : An.76 . and p .d .e of first order : J.23.
    depression of order by unity : 3262 – 9. particular integrals of: CM .2 : alge
    with different, total integrals : L .84. braic, C .86 .
    of dynamics : C .5,26 ,402 : CD.2 : G .1, relations of the constants : C .93 :
    4 : L .37,49,52,553,72,74 : M .2,17, J.10 : JP.6 .
    25 : Mél.4 : Pr.122 : P .54,55,63. in problem of n bodies : An .83.
    ap. to engineering : JP .4 . of perturbation theory : Mém .83.
    and elliptic functions : L .49. with quadratic integrals : J .99 .
    elliptic : G .19 : M .21. for roots of algebraic equations: P .
    elliptic multiplier : M .21. 64 : Pr. 13 .
    exact : 3187, 32704- 5 : G .12 : C .1,10,11. | * rule for equivalence of two solutions
    of families of surfaces : Me.77 . 3167 .
    with fractional indices : JP .15 . * singular solution of: 3169 – 78, 3301 -
    of functions of elliptic cylinders : M . 6, 3401 – 3 : C.19,94 : CM .2 : JP .
    22. 18 : M .22 : Man .83,84 : Q .12 ,14 .
    general methods : L .81. of sources : AJ.75 .
    generation of : 3150 . of a surface : G .2 .
    geometricalmeaning of : Q .14 . satisfied by the series 1 + 29 + 2q* +
    homogeneous : 3186 , 3234, 3262 - 8 : 2q° + & c. • 2V9 + 24/q° + 2Vq*5 +
    0 .13 : CM .4 : J.86.
    hyperelliptic : J.32,55 : Mo.62.
    & c.: L .49 : J.36.
    satisfyingGauss's function F (a,b ,y ):
    of hypergeometrical series : J.56,573, L .82.
    73. synectic integrals of : C .40.
    integrability of: Z .12 : immediate, and tortuous curves : L .53.
    0 .82. transformation of : An.52 : CD.9 : in
    whose integrals satisfy relations of curvilinear coords : J.85 .
    the form F [ 0 ] = xx Fæ : C .93. D . E . linear : A .28,35,40,41,43,45,46 ,53,
    whose integrals are satisfied by the 59 ,65,69 : Ac.3 : AJ.7 : An.50,85 :
    periodicity modulus of elliptic At.75 : C .73,293,58,73 ,84,88 ,903,913,
    integrals of the first kind : J.83. 922,94 : CD.3,43,9 : CP.9,10 : G .
    integrating factors : pp468 — 471, 15 : J.23,24,25,40,42,55,63,70,76 ,
    3394 : C .682,97. 79,80,81,83,87,88,91,98 : L .38,64 :
    of Pdx + Qdy + Rdz: Q .2. M .5 ,11, 12 : Me.75 : P .48 ,50,51 :
    integration : by Bessel's function : Pr.55,189,193,20 : Q .8 : Z.3,7,9.
    Me.80 . without absolute term , condition of
    by Gamma function : TE.20 . solutions in common : C .95.
    by definite integrals : 3617 — 28 : with algebraic integrals : C .96 ,97 : J.
    C .17 : J.74. 80 ,90 : M .21.
    by differentials ofany index : C .17 : determination of arbitrary constant :
    L .44 . At.65 : Q .192.
    by elimination : CP.9. argument & parameter interchanged
    by elliptic functions : An.79,82 : C . in the integral : J.78 .
    41: JP.21. bibliography of: AJ.7 .
    by separation of operative symbols : with coefficients that are algebraic
    Z .15 . functions ofan independent varia .
    by series : 3604 – 16 : C .10,94 : LM . ble : C .92, 94 .
    6 : Me.79 : Q .19 : TI.7. * with constant coefficients : 3238 – 50 :
    by theta -functions : C .90 . An.64 : CM .1,,2 : E .34 : JP .33 :
    irreducibility of : J.92. L .42 : N .47 ,84 .
    isoperimeters, pr : Mém .50 . with periodic coefficients : C .91,92 :
    of Lame : J. 89. doubly periodic : C .90,,92, 982 : J.
    of light: M .1. ry 90.
    in linear geomet : M .5 . with rational coefficients, algebraic
    of motion : C .55 : of elastic solids, integrals of : C .96 : JP .32,34 .
    Q .13 : of fluids, CP.7 : of a point, with rational coefficients, upon
    Č .26 . whose solution depends the quad
    with integrals “ monochrome and rature of an irrational algebraic
    monogène ” : C .40 . product : 0 . 913, 92 .
    868 INDEX .

    D . E . linear - (continued): D . E . of first order, linear - (continued ) :


    with variable coefficients : C .92 : * Pidx + Pzdy + Q (ædy - y dx ) = 0 ; Pi
    J.66 ,68,76 : L .80 ,81. P , being homogeneous and of the
    which connect a complete function of pth deg . in x , y ; Q homogeneous
    the 1st kind with the modulus : and of the qth deg .: 3212 .
    C .86 . P1, P2, Q different linear functions
    homogeneous: Ac.1 : J.90 : Mo.82. of x , y : C .78,83 : L .45 : J .24 .
    integrating factors of: 0.97,98. * Yz + Py = Q , where P , Q involve a
    integration by Abelian functions, C . only : 3210.
    92 : J.73 ; by finite diffe rence s, Yz + Py = Qy" : 3211.
    Q .1 ; by series, J.76ls. Q : Mém .11.
    which admit of integra whose loga yy + Py +
    rithmic differentials are doubly
    periodic functions : L .78.
    whose particular integrals are the
    ' re P,
    (P + 2Q « + Rx2)2jg ,' whe
    products of those of two given
    Q , R are functions of a : Mém .11.
    linear d .e : A .41. Yzta + by + y2 = 0 : J.25.
    irreducibility of : J .76 . Yrv (m + x) = lyv m - a : A .42.
    Landen 's substitution, geo : J.91. y: = Lf(y): J.9; yz = f (e,y) : An.73 :
    Malmsten 's theorem : J.40 . .55 .
    singular solution : J .73,83,84.
    transformation of : C .91,96 . _ 3y (y + 1) - 4x . 0.88.
    n variables, 1st order : 3320 - 32 : C . X (8y - 1) : 9
    14,15 : G .13 : J.20 ,80 : L .38. Ya + f(x ) sin y + F (« ) cos y + ( ) = 0 :
    n variables, 2nd order : L .37 : 2 varia L .46.
    bles, C .70. * Uztbum = ca " (Riccati's eq .) : 3214 :
    n variables, any order : Mém .13. A .40 : C .11,85 : (m = - 6 ) E .7 :
    Pdx + Qdy + Rdz = 0 : P , Q ,R involving ext28 : JP.14 : L .41 : P .81 : Q.7 ,
    x , y , , 3320 ; geoM .16 ; Z .20 : P . 11,16 .
    Q , R , integral functions of x only, allied eqs : L .51: Me.78 : Q .12.
    Q.19,: P = (ax" + bxr"-1+ & c.) * , sol. by continued fractions:Mém .18.
    Q , R similarly with y and %, Q .202. by definite integrals : J.12.
    * Xdx + Ydy + Zdz + Tdt = 0 : condition transformation of: Me.83.
    of being an exact differential, | * D . E . of first order : 3221 – 36 : A.29 :
    3330 . C .40,45,66 : M . 3.
    ædt+ yda + zdy + tdz = 0 : A.30. two variables : An.76 : J.40 : Mém .
    62 : N .50 : singular solution, J.38 .
    * D . E . of first order, linear : p467 : C .86 : 1 * Clairaut's equation . u = + ( :
    G .13 : algG .18 ; M .23. 3230 : CM .32 : Me.77.
    exact : 3187. integration by second order d.e: A.46 .
    homogeneous : 3186 . homogeneous in & and y : 3234 .
    integration by a particular integral: reduction to a linear form with respect
    0 .86 . to the derivatives of an unknown
    reduction to a continued fraction of a function : C .87.
    fraction which satisfies a : C .98 . related transcendents : Ac.3 .
    separation of variables : 3185 : CM .1. separation of variables : CD .9.
    Mdx + Ndy = (): 3184 : N .74,77. singular solution : 3230 : A .56 ,58 :
    (ax + by + c) da + (a 'x + b'y + c')dy = 0 : CP.9 : J.48 : Me.73,77.
    3205,2471 : L .59. solution by differentiation : 3236 .
    solution by factors : 3222 .
    (ax +64by. + c) da + (a's + by + c )"dy: A . transf. by elliptic coords : J.65 .
    verified by a reciprocal relation be
    it VQ= 0 ; P ,Q being quartics in
    ✓pt tween two systems of values of
    variables : 0 .15.
    a, y : 0.66 : LM .8 : ME.79. dx2 + dy? = ds2 and analogous eqs :
    f(x) dx _ f(y)dy = 0, f (x) of 1stdeg.: L .73.
    F (a ) F (y) - ° Fla ) of5th deg.: ada + hdy = ds : 3287.
    Ć .92. dx2 + dyº + dz? = ds? : L .48.
    · INDEX. 869

    D . E . of first order- (continued ) : D . E .of 2nd order, linear- (continued ):


    dx2 + dy2 + dz2 = (da? + dB2 + dy?): L . sy" + (r + qa )y' + (p + ne + mx?) y = 0 :
    50. A .23 : 2 .8, 9.
    F (x ,ux) = 0 : C.93. * (1 — 2) y" — « y' + q?y = 0 : 3282.
    so ( p ) + yy (p ) = x (p ) : 3226 . (1 + ax?) y" + axy' + q²y = 0 : 3283,3594.
    D . E . of second order, linear : A .29, 2x (1 - x2)y" — °' + n (n + 1)y = 0 : Q .18 .
    32,55,64 : An.63,79,82; : 0 .829,90 , æ (1 - x ) y" + - 4a ) y'ta'ry = 0 :Me.
    91,939,97 : J.51,74 : L .36 : Me.ld : 82 : Q .17.
    M .11 : Mo.64 : Z .5 . (m + x) (n + x) y" + (m _ n )y'
    with algebraic integrals : C .90 : J.81, < A ( m + c)? Ag = 0 : A.42.
    85 : L .76 .
    with doubly periodic coefficients : (m" .c? JP
    + na + p) y" + (qx + r) y' + oy = 0 :
    Ac.2. .13 : 2.4.
    homogeneous : M .22. duy" ++ Bluy
    Any'+ =Buy = 0,my" + Any'
    0 , and uy" + Any' + Bu
    integration by Gauss's series : Z .19.
    transformation of : An.52. = 0 ; with = a + bx + cal and
    y' = a : 3288. p = b + 2c« : A .423.
    Y" = Py : C. 9. do {(« — æ3) yr} - ay = 0 : L .54.
    y" + a y = Q : 3522,'5 : geoMe.66 : / * " + Py2 + Qy + R = 0, P, Q , R being
    Q = cos nx , 3526 : Q = 0,3523 — 4. functions of a : 3280.
    y" = Ay (a + 26x + c«2)-2 : L.44 . Py” + Qy' + Ry = 0 : Ac.1.
    y' = ax + by : 3281. zy"2z + azyyºr + f (y ) = 0 : Me.71.
    æY" = y : 2 .2. D . E . of second order : Ac.l : An .79 :
    (1 Fæ2) y" + my = 0, & c. : CM .32. JP.29 : C .67,69,80,91 : J.90 : L .39 :
    LM ,11, 12,13, 16 : Z .15.
    #2( " + q2) = p (p - 1) y : CM .2. with algebraic integrals : C .82.
    g” = { 1 + (n + 1) kºsmºa} ( Lames derived from linear eq : Me.73.
    eq.) : C .85. with elliptic function coefficients :
    y" = $ o + you + y2d2+ & c., when P. & c. Ac.32.
    are trigonometrical series : 0 .98 . in theneighbourhood of criticalpoints :
    y" = y (e* + e -*)-2: L .46. 0 .87.
    y" + axmy = f (m ) : E .6. polynomials which verify : Ac.6 .
    solution by definite integrals : A .27.
    y" = ay + y (a ): A.45. by factors : C .68.
    y' = f (y) : 3257. by Challis's method, and application
    *

    y" = f (x, y) (Jacobi) : 3285. to Calc. of Variations : A .65,66 .


    *

    y" + py'try = 0 : C.85,90 : Q .19. yy" = {y^2 + 2py2 : L.73.


    Y " = xy' - nxy : A .53. Myy" + Ny'? = f (x ) : N .79.
    ay" + my' + nay = 0 : L .45,78 . y" + Py' + Qy'3 = 0, P, Q functions of
    ay" + y' + Army = 0 : C.39. « : 3276 .
    dºg ” + rey = ( ba + s) g : An.51 : CD. * y" + Py' + Qy'" = 0 : 3278.
    5. * y" + Py'2 + Qy'" = 0 : 3279.
    aty" + 2x3y' + f ( y) = 0 : A .28,30. * Y" + Qy^2 + R = 0 : 3277.
    y" + f (x) y' + F (y) y^2 = 0 : 3281: L .42. Ixx + + 1: + I = 0,where I is Bessel's
    (a + bx ) y" + (c + dx) y' + (e + fx )y = 0 :
    A .58. function : J.56 .
    (a + bx ”)x2y" + (c + ex ") xy' + ( f + gæ ") D . E . of third order, linear: C.882: Q .
    y = Q (Pfaff) : 3598 : J. 2 ,54 : and 7,8,14 : M .24 .
    like eqs., Z .2,3 :with b = 0, A . 38. y' = y'" : JP .15 : U = U3x : C.3.
    = (a − ba ) g” – 2a (2a - ba ) . a27"" — Y = 0 : Z .8.
    + 2 (3a — bæ ) y = 6a2 : A .28,30. g " = 3m2u” + 6m(1 + 2) + 3m(1 + 2)
    wy" + y' + y(x + 4) = 0 : Me.81,84. (u + 1) y : A .42.
    ay" + y' + y (@ - A) = d , cos (a + 1). Me y'' = æ " ( Ax ?y" + B.cy' + Cy) : A .58.
    +1 D . E . third order : An .832: C .982: M .23.
    82 . * D . E . of higher order, linear : 3237 — 50 :
    apy " — 2xy + 2y = x2yf-2: A.28,30. A .65 : C .977 : J.16 : M .4 : Q .18 .
    870 INDEX .

    D . E . of higher order, linear - (cont.) : | D .E ., simultaneous system of : An.69,


    of orders p and m + p, th : C .43. 82,84 : C .10,432,47,92, : CM . 1 : LM .
    4 = xy . - : A .1. 14 : Me.13,80 : Pr.12.
    Yox = f (x ) : 3256 . Hamiltonian : Q .14.
    integration and inversion of the in
    Ynr = "y : L .39. tegrals : C .23.
    Yux = (a + B.e) y : J.10. Jacobi's : CD .3.
    202m Ymr = a" y : A .32. method of multipliers : 3353.
    22mY2mx = y : A .42. number of arbitrary constants : Me.
    KYnz = y : A. 26 . 77 .
    zeyne = ły : by definite integrals : || redaction of
    reduction order : 3350 .
    to a P . D . eq : C .44.
    C .482,49 : J.57 . theorem of Abel : C . 24 .
    Ynr = x " y :bydefinite integrals : J.19. theorem analogous to Lagrange's
    c" * by (2m +1)= = y : by Bessel's func in the Perturbation theory : L .
    tion : M .2. 51.
    Ynz = x " Y + A + BX + Ca2 + ... + Nx" : theorem of a new multiplier : J.27.
    2 .10 . transformation and integration of :
    Yne = Ax" yx + Bw" -"y : A .28,38. L .45.
    me?"Ynx = Axy + By : A .33. * 22 = ax + by and y2 = ca + dy : 3354
    Ynr - xy = yer tabæ" y : by definite in and a similar example.
    tegrals : J.17. * D .E ., simultaneous linear : 3340 - 59 :
    Ynx = Axʻy2r + Beyo + Cy : A .53 : M .3. AJ.4 : C .9,92 : E .5 : N .66 ,84.
    Ynx = Ax" yzx + B . " -"Yr + Ca " -"y : A . Pfaff's method : C .14 : J.2 :
    29,30,33,38. transformation of: J.98 .
    aynx tayin -1)= = bey : J.2 : Z .10. * d.c _ dy =_ dz * : 33.46 : Q.14.
    wynx - (n -1) + mæ?y = 0 : A .40. Ē=
    a /ng + Àg(m-1) = c (22 + u24) : A.86. die - dy dz 2017
    AXYnx + By(n-1)= = x " (Axyz + By): Z.8. P,-« P - P,- YP P3 -2P : 0*
    *

    a2Ymx + q"xy = p (p - 1) Y(m –2)r : J.2. * Xe+ P (ax + by) = Q and y,+ P(ex + dy)
    Ax?y( +2)e + Beyin +1)2+ Cyne = R : 3348 .
    = @ " (axºy2x + bryx + Cy) : A .38. * tæe+ 2(x , y) = t and tye + (x + 5y) = ť :
    Ynz + a1}{n-1} + ...tany = 0 : 3239 : A . 3349.
    40 . equations in X , X2t, * 4!, & c . . Yı, yze, Yst,
    * ditto = f (x ) : 3243, 3516 : a , ... an & c.: 3357.
    functions of x, 3237 : J.39. * homogeneous in æ , y , % ... and their
    n fractional and all lower orders : second derivatives only : 3358 .
    integral : L . 36 . * D .E ., simultaneous first order : 3340
    (d , + a)"y = f (x ): CM .4. 49 : C .43 : J .48 : Pr.62 .
    ditto = ear : 3528 : ditto = sin mx: * D .E ., symbolic methods : 3470 – 3636 :
    3529. ČD.1 : P .61 : Q .3,17 .
    * (p + qx)"Ynxta,(p + qx)" -ly(n- 1)= t ... | * F (ds)u = Q : 3515.
    tany = f (x ) : 3250 : with p = 0,
    C .96 . Uusta? u = Q and similar : 3522.
    *

    Qm +ny(n +n)= + ... + (am + x ) Ymx + ... exceptional case of the inverse pro
    *

    +2y = 0 : A .47. cess : 3526 .


    reduction of an integral of the nth
    *

    qvs - (a + bic ) ynx tem - c + d) y(n -1)- t ... order : 3530.


    + ty = 0 : J .39. ax "Ymx + bæ"Yurt & c. = Q : 3531.
    Yar+myr + nyz + py = 9 : L.44.
    * * *

    ayma+ byno + & c .= f(e*,sind,cos8 ):3535.


    * D . E . of higher order : 3251 – 69. an," 23540
    + b7 }"z + & c. = ui +17+ & c. :
    .
    * You = F (Y(n-1)2): 3258. F (T )u = Q : 3541.
    * Ynx = F (Y (0 -2) 2) : 3260 .
    *

    Pynz + Q = 0, where P, Q are func Reduction to the form (IT" + 4, 01^- + +


    tions of x , y , and the first n - 1 ... + Anu = Q , where II = Mdx +
    derivatives of y : J.31. Ndy + & c. : 3546 .
    INDEX. 871

    D .E., symbolic methods— (continued): | Directrix : - of a conic : 1160 : gn.eq.


    * F (@ dz, ydy, ...) u = ? Ax" y"... : 3558. A .25 : E .36 : Me.80: t.cQ .13.
    to transform (a + bx + ...)Une + (a' + b'x of in -parabola of a triangle : Me.80.
    of a curve : A .20 : J.2 .
    + ...)U (1- 1)+ + +...& c. = Q into the of a parabola : gn .eq E .29.
    symbolic form : and the con of a quadric : N .74,75.
    verse : 3571, 3573. Discontinuity :- in curves : CM .4 :
    * utajd (D ) eout & c. = U : 3575. 2 .26 .
    * to transform uto (D ) eru = U : in fractions : Man .48 .
    3579 – 80 . in maxima and minima : CD . 3.
    Discontinuous functions : A .7 : C .153,
    * to reduce a homog. eq. to the form 28 : G .19 : J.7,10 : LM .6 : Man .
    Ynx + y = X : 3585 . 48 : TI.21.
    * Differential expressions : 1407 : prAn . Discriminant : 1627 , 1638 - 9, 1644 : Ac.
    85 . 1 : J.90 : LM .2 : M .12 : N .59 :
    algebraic : An.79 : M .9, by homog. Pr.14 : Q .10,11.
    coords. of an algebraic d.e of 1st order in 4
    transf. of: J.85 : Q.162: M0.69. variables, and of its complete
    Differential :- formulæ ,theory of: L . primitive : An.84 .
    52 . of alg. eqs., resolution into factors :
    functions, theory of : C .60. M .24 .
    expressions, linear : J .85. of alg . functions : J.91.
    parameters of functions : C .66 ,78 . applied to conics and quadrics : A .
    quadratic forms : An.84 : transfor 58 .
    mation of, A .16 . of binary quantic : An.56 .
    resolvents of alg. egs : 3631 – 7 : An. of a binary sextic : An.68 .
    *

    83: C .91 : LM . 1,9,14 ; M .geo4,18 : of a quartic : N .83.


    Me.75,82: Man .65,84 : Q .6,11. of a ternary quadratic form : Me.68 .
    of y" — nyn-r + (n - 1)x = 0, & c. : * Discriminating cubic : 1849, 5693 : G .
    16 : J.26 ,71.
    3633 – 6 : Man .65 . * proof of real roots : 1850 : A .29.
    of ym + byr + cx = 0 : Q. 17 . Displacement: theory of : N .82.
    of 12 (y4 + ay? + xy) = a ? : Me.82. of plane figures : C .80 : N .73 .
    Differentiants in terms of differences of of an invariable dihedron : Me.85 :
    roots of parent quantities : AJ.1. infinitesimal ditto, C .84 .
    * Differentiation : (see also • Differential of an92 invariable figure : 0 .51,523,66 ,
    coefficients ’) 1402 – 82 : CM .1. , : JP.26 : L .74,75 .
    formulæ : 1411 - 72 : An.59 : CD.2 : of a figure, two of whose points slide
    CM .1 : Pr.9. on two curves : C .82 .
    by the method of “ Rates " : Me.75 . of a solid : L .40 : determination of
    general, i.e.,with any index fractional the normals to the lines or sur
    or imaginary : An.58 : AJ.3 : CD . faces described : C .62.
    3 ,4,5 : J .12 : JP .13 : L .55 : N .84 : of a system of points : C .78 .
    Q .3,4 : TE .14 ,15 : Z .16 : change virtual displacement : J.11.
    of independent variable, JP.15. infinitesimal : of an alg. surface : C .
    70.
    under the signs: 2253 : A .17. of bodies only defined by 4 coordi.
    successive : A .20 : Q .12. nates : C .73.
    when the function becomes infinite : of a parallelogram : C .97.
    C .88. * Distance : - between 2 points : 4034 - 5 ,
    Digits : - calculus of: Sbonimsky's t.c4601 : Q .7 : sd5508, 5510 : some
    th : J.30. relations, G .9.
    frequency of in numbers : AJ.4. correspondences for quadric surfaces :
    origin of: L .39. LM . 16 .
    Diophantine analysis : see “ Partition of a point from a line and from a
    of numbers." plane : A .57.
    Di-polar geometry : Z.27. * of a point from a plane in a given di
    rection : 5559.
    * Direction ratios and cosines : 5511 – 14. | relations : Z .27.
    Directive algebra : N .68. * Distributive law : 1488 .
    872 INDEX .

    Divisibility : C.96 : E.8 : M .39 : N .67,74. / Dual relation


    J. 10 .
    between figures in space :
    of decadic numbers : Z .22.
    ofnumbers of the form 22m + 1: Mél.52. / Duplication of the cube, approx. : Pr.20 .
    of a quotientby the powers of a fac- | *e (see also “ Expansion " ) : 151: N .67,
    torial : 0 . 94 . 682 : geo meaning E .4 : N .55 .
    of (x + y )" + ( - )" + ( - y)" : Me.79 . combinatorial definition of : A .13.
    Division : prJ.47 : prL .56 . incommensurable : 295 : CM .2 : L .40 :
    * abridged : 28 : arithJ.31 : N .452,463, Me.74 .
    52,54,57,81 : algN .42. and a , numericalth : Q.15 .
    by 73 or 47 ; rule for remainder : E .22 . en +bx+cxº+ in fractions : L .80 .
    effected by determinants : 581.
    Fourier's rule : N .52. e z , & c.: CP.6 .
    of planesand spaces : J.1,2 . DrDehr: E .37.
    of a rectilineal figure and of a spheric
    cal poly gon : J .102. * pur : 766 : AJ.7 : in transformations:
    of a right line into equal parts : 950 . CM .4 .
    and transformation of plane figures : ea + ib : A .33 .
    A .42. *Edge of regression : 5729 : tg .eC .71 .
    of trapeziums,pyramids,and spheres: * radius of curvature of : 5742 .
    A. 11. Eisenstein 's theorem : G .16 .
    of triangles: A .11,17. Elastic curve : C .18 ,19 : JP.34 .
    Divisors : - of an integer, number of : Elementary calculation : N .45 .
    374 : 0 .96 : N .68 . * Eliminants : 583, 1626 .
    of integralrationalfunctions :Mém .57. and associated roots : LM .16 .
    Newton 's method of : 459. of two cubics : J .64 .
    of a polynomial with commensurable degree of: G .11, two eqs 12 : J.22,31.
    coefficients : N .75 . * Elimination : 582 — 94 : A .2 : Ac.6 ,7 :
    rational, of 2nd and 3rd degree: N .45 . C .12,87,90 : CD . 3,,6 : CM .3 : G .
    sum of : 377 : Ac.f4 : L .56 ,57. 15 ,17 : J.34,43,60 : JP .4 : L .41,44 :
    sum of powers of: L .58. LM .11 : M .5 ,11 : N .42,457,46 ,80 ,
    of 22 + Ay ? : L .49. 82,83, : Q .7,th12 : 2 .23.
    Double algebra : LM . 15. problems : C .84,97 : J.58 : M .12 : Q .
    Double function, laws of change of 8,11 : in metrical geo, A .63.
    higher order : A .21. * Elimination of a between two equations:
    *Double points : 5178 : n -tic with 586 - 94 : C .12 : CM .2 : J.15 ,27 :
    in (n - 3), C .60 ; with : (n - 1) Mo.81 : N .432,763,77.
    (1 - 2), M .2; Clebsch 's thsof these * by Bezout's method : 586 : A .79 : J.
    quantics , C .84 : n -ticwith t (m - 1) 53 : Me.64 : N .74,79.
    (m - 2) - 2 double points, L .80 . by cross multiplication : CM .1.
    of plane curves in cubic space : 2 . 28 . by the dialytic method : 587 : N .79 .
    *

    in a locus defined by alg . conditions: in geodetic operations : 2 .3.


    C .88 . by h .c.f : 593 : JP.8 .
    *

    of a pencil of curves : An.64. by indeterminate multipliers : CM .1.


    of plane curves in cubic space : Z .28 . * by symmetricalfunctions : 588.
    in the projected intersection of 2 quad degree of the final equation : J.27 :
    rics : N .84. L .41.
    of tortuous curves : M .3. Elimination : - ap to alg .curves : M .4.
    Double relations : A .60 . ap to in -and circum -conics of a poly
    Double tangents : An.51 : J.49 : P .59 : | gon : At.63.
    Q .4 : Z .21. calculation of Sturm 's functions: C .
    of a Cartesian : E .30 . 80 .
    of an n -tic : M .7 : number = in (n - 2 ) of functions : 3153: C .84,87 : Me.73,
    (112 — 9), J.40,63; N .53. 76 .
    of a quartic : C .37 : J.49,55 ,68,72 : M . with linear equations : At.53.
    1 : N .67 : P .61 : Pr. ll : with a with linear differentials : L .36 .
    double point, M .4,6 : reciprocity with n variables : CP.5 .
    of 28 double tangents . resultants, comparison of: J.57 : and
    to the surface of centres of a quadric : interpolation , J.57.
    C .78 . transformation and canonical forms :
    Drilling, shape of hole : Pr.35. CD .6 .
    INDEX. 873

    *Ellipse (see also “ Conics ”) : 1160 , 1 Elliptic functions— (continued ):


    4250 : cn1245 : geoQ .9. of first and second kind : C . 10 .
    theorems: A .30 ,47,prs49 : N .76 : Mc as functionsoftheir amplitude: JP .
    Cullagh's, N .72. 14 .
    eq. r + r = 2a : A .2. representation in a simple form :
    equal chords : tg .eE .22. 2 .21.
    of maximum surface : N .65 . series by which they are computed :
    as the projection of a circle : 4921 : J.16 ,17 .
    N .752. of second kind : J.93.
    rectification of : 6083 – 96 : A .3,22.27 . mechanical representation : Me.75 .
    30 : At.39 : graphic : Mél.l : N . reduction to first kind from same
    43 : TE .4 : 2 .6 : when e is very modulus : A .56 .
    small, TI.9. of second and third kind, expression
    representation by a circle : An.70 . by A function : Z .10.
    quadrature of : 6108 , 6113; t.c4688 : of third kind : C .94,96 : CD.8 : J.14,
    A .46 : of sector 6098 , A .20 : of 47 : LM . 13.
    segment, 6103. addition of: A .47,geo64 : AJ.7 : C .59,
    and triangle : thQ .4 . 78 : J.35,41,44 ,54,88.,90 : LM .13 :
    * Ellipse and hyperbola : 1160,4250 : A . M .17 : Me.80 ,84 : Q .18 : Z .1.
    24 ,28 . of 1st kind, 2 .26 : 3rd kind, Me.81.
    theorems: A .23 : CM .3 : N .85. 2nd kind by q series : Me.83.
    sectors : TE . 14 . applicationraof: C .857,865,892,90,93,946.
    * Ellipsoid : 5600,6132 : A .28 : thCD.2 : to algeb : J .7 .
    prС.20 : L .38 : cnM .20 : P .9 . to arithmetic : C .98 : L .62 .
    centro -surface : CP.12 : LM .3 . to confocal conics : Z .7 .
    cubature of some derived surfaces : to geometry : G .12: J.38 ,53 .
    A .12 . to in - and circum -circles of a poly
    and enveloping cone : 5664 — 72 : Q .6. gon : L .45.
    to rectification : L .45 .
    generation of ( Jacobi) : CD .3.
    of gyration : 5930 : of inertia : 5925 – to spherical conics : Z .22.
    39. to spherical curves and quadrature :
    and plane of constant segment, th : An.50 .
    E . 32. to spherical polygons with in - and
    * of revolution : 5604 : area , N .42. circum -circles : L .46 .
    a locus in space : Q .16 ,17g. to spherical trigonometry : Q .20 .
    * quadrature of : 6143: J.17 : Z .1 : of * approximation to : 2127 — 32 : P .60,62.
    zone, A .22. arithmetical consequences : Ac.5.
    * volume: 6144, – 8 : A .46 . arithmetico-geometric mean : J.58,85,
    Ellipsoidal geometry : A .10 : LM .4 . 89.
    surfaces : G .17 . arg sn a and (arg sn a)?, as def. inte
    *Elliptic functions : 2125 : A .1,12,,16,21, grals : Q .19.
    35 ,48 : trAc.5,62,ths72 : An.61,84 : in complex regions : Z .282.
    C .46 ,506,90, 96 ,97 : CD .2 .3., 5 : CM . development of: 2127 — 32 : J.81 : 1st
    3, : E .23 : G .14 : J.2,34,44,6 ,8 ,16 ,26 , and 2nd kind, C .92 : with respect
    27,302,32,35,37,39,46,72,832 : JP. to themodulus of a (x ), u (a ) and
    25 : L .55 ,56 ,61 : LM .7,10 ,29 : M . their powers, C .86 .
    3,11,12,pr25 : Me.79,80 ,81,82, : development of an imaginary period
    Mo.81,823,832,85 : N .773,784,792 : when the modulus is small
    P.31,34,75,78 : Pr.6 ,9,10,12,232 : enough : An.70.
    Q .11,17 ,19 : Z .22,11,27. differentiation by periods and invari.
    of first kind : A .12,21 : 0 .16 : J.93 : ants : J.92.
    L .43. discriminant of modular equations:
    with complementary moduli exten M .8 ,9.
    sion of a theorem of Lagrange : double substitution : J. 15 .
    An 83 %.
    normal forms of 3rd and 5th de
    gree: M .172
    replaced by one of second kind : J.55.
    A+"am2.K cos+'am 2.K sin 26 T T

    represented
    C .83.
    by gauchebiquadratics: **ſ*a?am 2Kxdx:J.37.
    5 T
    874 INDEX.
    Elliptic functions — ( continued ) : Elliptic functions— (continued) :
    eqs. for the division of : Mo.75 . spherical triangle of: Q .19.
    formulæ : AJ.5 : J. 15 ,36,30 : LM .13 : 1 and spherical trigonometry : Q .17 .
    Me.78 ,80 ,85 : JacobiMél.1 : Q .16 , substitution of lst order : J.34.
    19 : from confocal conics,LM .14 ; and theory of numbers : L .58.
    differential,Me.82 ; for sn , cn ,dn, transformation : An.573,58,60 : Ac.3 :
    of utu + w ,Me.82. C .49,879,180,82 : CD .3,5 : J.3,34,
    Galois' resolvent : M .18 . 35, f55,55,87,88,89 : LM . 9,11 : M .
    geo. problems : M .19. 14 ,19 ,22 : Me.83,tr84 : Pr.27 : Q .
    geo. properties : L .43,45 : P .52,54 . 13 ,20 .
    geo . representation : A .22 ,61 : An.60 , of 1st kind : A . 33 .
    61 : At.53 : C .19,212 : J.63 : L .44, of 1st and 2nd kind as functions of
    78 : in solid geo, M .9 : of 1st kind, the mod : L .402.
    An.53 : C .70 : CD. 1 : JP .28 : L . of 3rd order : J.60 : Me.83.
    43,45, 46 ,78 : of 3rd kind, A .24 . of 7th order, square of mod : J. 12 :
    identities : Me.77. LM . 13.
    imaginary periods : AJ.6. of 11th order : At.5.
    infinite double products, A .14 : with of the orders 11, 13, 17, and 19 : J .
    elliptic functions as quotients, 12 , 16 .
    J. 35 . cubic : C .64 : Q .13.
    inversion of: J.4 : JP.34 : L .69. and division : J.76 : M .25.
    KE' + K ’ E - KK ' = : Me.75 . of a double integral, & c.: Me.75.
    of įK : Me.85 . Hermite's ; tables : J.72.
    modular equations : C .472 : J.58 : Jacobi's : LM .153,16 : J.87.
    LM . 9 ,10 : M . 12. linear: J.91.
    modular functions : A .11, 13 : G .12 : modular, of Abel: ap to geom : C .
    J.72,83 : L .402: M .17,18 : differen 58 : to conics, C .79 .
    tiation for modulus of am , LM . modulus of in a function of the
    13 : expansion in powers ofmodu. quotient of the two periods: An.
    lus, J.11 : formulæ , L .64 : relation 70.
    between the modulus and the pertaining to an even number : J.
    invariantof a binary quartic, Z .18 . 14 .
    multiplication of: C .882 : J.14,39,41, quartic : Q .12.
    74 ,76,81,86,889 : M .8 : M0.57,833: by roots of unity : J.6.
    and division , 2 .72 : formulæ , trA . of rectangular coordinates : LM .15 .
    36 : C .59 : J. 39,48 : complex, M . and of functions in theory of Cate
    25 : Mo.62 : Q .19,20 : mod . - , nary : A .2 .
    J .48 . triple division of and ap. to inflex, of
    periodicity moduli of hyper -elliptic cubics : A .70 .
    integrals as functions of a para Weierstrass'smethod : AJ.6.
    meter : J.71,91. * Emanents : 1654.
    subsidiary , pm (u , k ): LM .15 . Empirical formulæ , calculation of : Me.
    products of powers : Mém .71. 73.
    quadriqnadric curve : M .25 . Engrenages : L .39,40,.
    q-formula for sin am : LM .11. “ Ensembles," theory of: Ac.2,.
    * Envelope : 5192 : A .24,prs56 : 0 .45,86 .
    q-series:and f852:for + 2 coeff's.: p .d.eCM .4 : G .11: M .84 : Me.64,72 : N .
    44,59,68,74 .
    Q .21. application to perspective : A .9.
    reduction of: An.64 : in canonical
    forms, J.53. class of (Chasles), th : 0 .85.
    from ellipse and circle : LM .15 .
    relations: A .67 : J.56 : between E' of a carried curve : 5239.
    (k )and F' (k ) : J.39. of conics, theorems: N .45 .
    representation by power series : J.54 . ofchordsof a conic : N .48 : subtending
    representation of quantities by sin a constant angle at the focus,
    am (utw , k ) : J .45 . CM .3 .
    series: C .95 . of chord of a closed curve : 9.28 : cut
    sn 8u , cn 8u , dn 8u in terms of snu , ting of a constant area, 1.31.
    tables : Pr.33. of curves in space : L .83.
    sn , cn, and dn of uto + w : LM .13. of a curve with n parameters : 194.
    INDEX . 875

    Envelope - (continued): Equations — ( continued ):


    of directrix of a parabola : E .34. 47,59,62,68,91,97,995: CM .3 : CP.
    of geodesics : M .14 ,20. 4 : G .1 : J. 13,16,34 : L .672,69 : M .
    imaginary,of theconjugatesofa plane 14,21 : Me.76 : Mo.79,180 : N .67,
    curve : C .75 . 68,ths55,67, and 80 : P . 1799.
    of pedal line of a triangle : Q .10 : do. (For Binomia ), Biquadratic, Cubic,
    of in -triangle of a circle , Q .8,9. Cubic and biquadratic , Linear,
    of perpendiculars at extremities of Quadratic, Quintic, and Trans
    diameter of an ellipse : N .46 . cendental equations, see those
    of a plane : C .35. headings. Other kinds will be
    of planes which cut a quartic gauche found below .)
    curve of the 2° in 4 points of a Abel's properties : C .91.
    circle : An.71. algorithms for solving : M .3 .
    of planes perpendicular to radiants of whose coefficients are rational func .
    an ellipsoid at the surface : An. tions of a variable : J.74 .
    59 : Pr.9. of degree above the 4th not soluble :
    of plane curves : G .11,12 : singulari. J.83.
    ties of, LM .22. whose degree is a power of a prime:
    of polars of a curve: J.58. An .61 : C .48 : L .68.
    of a quadric : Q .11. derived : 424 — 31: A .22 : in d.c,
    of a right line : N .63,79,83 : Q .13. 1708 – 12.
    cutting two circles harmonically : developments : An.61.
    N .85. differential operators in : LM .14 .
    sliding on two rectangular axes : Eisenstein 's theorem : LM .7.
    N .45 . extension of theory of: C .58 .
    of a Simson line : E .29,34 . fundamental principles or theorems
    of a sphere : C .67 : J.33 : touching 3 A .1,11: 0 .96 ,97: L .39,40 : J.23 .
    spheres, N .60.
    of a surface : CM .1 : M .5 : degree of,
    Galois' theory : C.60 : G .12 : M .18,23.
    of geometry : C.68 : homogeneous,
    N .60. N .64. "
    of a surface of revolution : L .65. generic : Q .4,5.
    of tangent of 2 variable circles : N . Hariot's law of : J.2 : extC .98.
    51. homogeneous, reduction of a princi.
    Enveloping asymptotic chords and pal function which verifies a
    polars : A .14,16 ,17. characteristic homog . eq .: C .134,
    * Enveloping cone : 566472. 149.
    of an n -tic surface : CD.4. identical: J.27 .
    of a quadric : 5697: th of Jacobi, J. impossible : Man.51.
    12 and CD. 3. in geo .mean of roots : N .45 .
    of a twisted hexagon, locus of vertex : in quotients of roots : N .45 .
    A .10 . in sums of the C (n , 2) roots of another
    Enveloping line of class cubic : inyo eq.: N . 43.
    lution th , E .29. insolubility of quintics, & c.: J.1.
    Epicycloids : J.1 : Mém .20 : N .45,46,60: irrational : Man .51.
    TE.24 : thsZ .15 . with integral coefficients :503 : C .24 :
    centre of curvature : N .59 : planeand J.53 : complex, J.53.
    spherical, JP .14 . irreducible : An.51: Mo.80 : of prime
    double generation of : N .69 . degrees, AJ.7.
    reciprocal polar of : geo E .19. linear : see “ Linear equations."
    * Epi- and Hypo-cycloids: 5266 — 72 : LM . | * miscellaneous : 214 .
    4 : 2 .18 . numerical: C .10,123,32,78,81 : G .13 :
    and derived curves : Z .17 . J.10 : L . 36 ,38 ,41,83 .
    tangential properties of : absPr.34 . and commensurable quadratic fac
    * Epi. and Hypo-trochoids : 5262 – 5 : tors : L .45 .
    LM .4 . of nth degree with two real roots :
    Equality and similarity of figures : J.52. C .98 .
    * Equations (see also “ Linear equa from observations : A .21.
    tions " ) : 50 – 67, 211 - 222, 400 with only one positive root: 411 :
    594 : A. 6,18,57,58,60,61,65 ,67: tr C .98.
    Ac.32 : AJ.6 : An.51,549: C .44, of payments : A .34,36 : CD.1 : CM .2.
    876 INDEX.
    Equations - ( continued) : | Equations - (continued) :
    * reciprocal: 466 : A .44 : C .162: of a - pæ + 9 = () : number of real
    quartic, N .66 . roots : 0 .98.
    reduction of: C .97 : CD.6 : to recipro a2n + qa" + p" = 0 and derivatives :
    cal eqs., A . 35 . N .652.
    relation to linear d .e and f. d. e : L .36 . 2 * -1 + -2 + ... + 1 = 0 irreducible
    roots of : see “ Roots of equations." if n be a prime: L .56 .
    simultaneous (see also “ in two or
    three variables ") : 59, 211, 582 : ax20 +* + bæ* ** + ca" + d = 0 : G .14.
    C .25 : LM .6 : thsN .48 and 81 : (x - 1) ! + 1 = : L .56 .
    quadratics, N .60 . (xP - a ") y (x ) = 0 : N .82.
    deducible the one from the other : (1 + x ) (1 + bx) = 0 when æ is small :
    C .22. A . 2.
    of the form 2 " + y + + ,* = a : N .46.
    solution of : 45 ,54,59,211,466 — 533, * in twoA .
    variables : 59– 67,211,217 — 8 :
    20,25 : CM .2 : J.14 : N .473,48,
    582 : A .64 : trAn.52 : C .32,53,62, 63 : Pr.8 : Q .18 .
    643: J.4,272,87 : Mo.56 ,61. of any degree with a variable para
    by approximation : 506 - 533 : A .30 : meter : L .59.
    Ac.4 : C .11.17 ,45 ,60,799,82 : E .4 : implicit : Mém .30 .
    G .8 : J.14 ,22 : Me.68 : N .51,62,783, numerical solution : 2 .20.
    80,84 : No.58 : P.5 : Q .3 : TI.7 :
    2 .23 . 203 + y3 = a and ały + xy2 = b : A .48.
    Horner's method : 533 : P.19. in three variables : gn.sol,60 : A .1,64 :
    * * *

    Lagrange's method : 525 : C .91. N .47 : M .37 : by a cubo-cycloid , C .69.


    Newton -Fourier method : 527 – 8 : | * (y — c)(z - 6 ) = a , sym in x, y, z :
    AJ.4 : G .2 : Me.66 : N .46,56 ,60,
    69,79.
    219.
    y ? + z2 + yz = a ?, & c., sym : 220.
    *

    Weddle's method : 7 .7,8. w ? - yz = a?, & c.,sym ,and æ = cy + b%


    by continued fractions : J.33.
    *

    by definite integrals : Me.81 : P .64 : | & c., sym: 221 – 2.


    Z3. « - yz = það {(1 - y-)(1 — 22)}, & c.,
    byroots
    diminishing the powers of the sym : A .35. '
    : C .41. ax + by + cz = l, a'& + by + cʻz = l',
    by elimination of integers: N .70 . 22+ yº + z = 1, by trigonometry :
    by geometry : C .87. A .6.
    by imaginary values : J.20 . * Equiangular spiral:5288:Me.62,: N .69,70.
    by infinite series : J.33. Equilateral hyperbolic paraboloid and
    by interpolation : 0 .5. derived ray- system : Z .23 .
    by logarithms: C .95 . Equimultiples in proportion : G1.
    the one by the other : C .72 : L .71. Equipollences,method of : N .692,703,737,
    by radicals : C.58 : Q .15 . 743.
    by series : An.57 : C .49,52 : J.6 : Equipotential curves : Me.82 : Pr.24 .
    Mém .33 . Equipotential surfaces : G .20 : geoJ.
    by transcendents : An.63: Q .5. 42 : M .8.
    by trigonometry : 480 : A .1. of ellipsoid : L .822.
    a nonic eq . which has this charac Equivalence of forms: C .88 ,90 : JP.29 .
    teristic : A given rational sym Equivalent representation : 2 .23 .
    metrical function 0 (a,b ) of two Equivalents, theory of : A .44 .
    roots, gives a third root y, such Eratosthenes' crib or sieve : N .43,49 .
    that a = 0 (B , y), B = n (y, a ), Error in final digit of decimals : C .40 :
    y = 0 (a, B ) : J.34 . Me.74 : N .56 .
    symbolic , non -linear : C .22. Errors of observation : A . 18 , 19 ; An.58 :
    systems of: C .67 : G .11,18 : LM .2,8 : C .93 : JP. 13 : N .56 : P .70 : TE.24 .
    Q .11: M .19 : Z .14, 18 (see also Errors of constants : Mo.83.
    " Linear equations." ) * Euclid , enunciations : p. xxi.
    transformation of : C .64 . axiom 11 : J.1 ; 1.47 : new proof, C .60 .
    in one variable : 45 — 58, 214 – 16, 400 | II. 12 and 13 : Me.80 ; VI.7 : Q.11
    - 550 : approx A .20 . new proof, Q .9.
    graphic solution : C .65 . XI., & c., Me.71 : XI.28 : A .10.
    221 +1 - x - k = 0 : An.59 . XII., & c., G .9; criticism on : Q.7,9.
    INDEX. 877

    Euler's algorithms: A .67. Expansions of a function in a series


    * Euler's constant: 2744 : Pr.15 ,16, 18, 19, * of circular functions : 2955 : A .11 :
    20 , Table 27 . CM .3 : J.43 : L .36 : Q . 12 : of
    and Binet's function : C .77 : L .75 . imag . arcs, J.6 .
    Euler's equation : N .72 : integration of coefficients of: gn form , 0 .85 : gn
    it by the lines of curvature of a property , J.41.
    ruled hyberboloid , N .75 . connected with a 2nd order d.e: C .5 :
    Euler's equations of motion solved by L . 36 ,372.
    elliptic integrals : Q .14 . of denominators of convergents :
    Euler's formula for (1 + x )" : L .44 . C .46 : JP .21.
    * Euler 's integrals : 2280 - 2323 : A .41: of exponentials : J .80.
    Ac.1,2 : An.54 : C .9,17,94,95,th96 : * of explicit functions : 1500 — 47.
    J.15,21,,45 : fL .43 : Me.83 : 2 .9. extended class of : C .82 : approxi
    B (1,m ) : 2280 : An.69 : G .9. mating to functions of very large
    r (n ) : see “ Gamma function .” numbers, L .782.
    ap. to series and functions of large of faculties of the variables : Mém .31.
    numbers : JP . 16 . of implicit functions: 1550 — 73 .
    sum formula and quadratic residues : of Jacobian functions : An.82.
    An.52. of Legendre's functions, Xx : An.75 .
    Euler's numbers : AJ.5 : An.77 : C .66 , with limits : C .34 .
    83 : J.79,89 : prsL .44 : Me.78 ,80 . of another function : 1559 : C .95 ,96 .
    Evectant of Hessian of a curye : E . of periodic quantities : C .52,53 : JP .11.
    32 . of powers of the variable : At.57 :
    * Even and odd functions: 1401. 0 .19 : L .46 .
    * Evolute : 5149 — 59, 5165 : An.53,61 : of powers
    882.
    ofa polynomial: C .86 : J.53,
    C.30 : Q.3,11.
    analogous curves : L . 76 . of powers of another function : Mém .
    of a catenary : 5159. 33 : N .74 .
    of a cubic curve : Q .11. within a given interval according to
    of a cycloid : A .30 . the mean values of the function
    of an evolute, in inf. : L .59 : Me.80. and of its successive derivatives
    ofan63ellipse : 4547,4958 : C .84: N .52, in this interval : 0 . 90 .
    2. by Bessel's function : J.67 : M .10,17 :
    and involute in one : L .41. Z .1.
    of the limaçon, rectif. and quadr. of: by binomialtheorem : 125 .
    * * * *

    E .40 . by factorials : 3730.


    of negative focal pedal of a parabola : by generating functions: 3732.
    E .29. * by indeterminate coefficients : 232 ,
    oblique, direct and inverse of differ. 1527 — 34 : A .3.
    ent orders : C .85. by logarithmic method : C .92.
    of a parabola : 4549,4959 : Q.5 : N .65. * by Maclaurin 's th : 1524 .
    rectification and quadrature of : A .4. by a series: C .93,95.
    of surfaces : C .74. Expansion of: - alg. functions: C .89
    of symmetrical bicircular quartics : 2 .45 .
    Q . 18 . Eisenstein 's th : J.45.
    of a tortuous curve : 5731 : A .25. n , alg . functions from n eqs : G .11.
    * *

    angle of torsion of evolute: 5754.


    integrable equations : L .43. (1 + ax) in an integral series: A.65.
    * Evolution : 35 . (1 - x )(1 — 2)(1 - 23)...: C .92 : J.21 :
    E (X ) = integral part of æ : C .50 : L .57 . L .42.
    * Ex-circle of a triangle : 711,953 : ( 1 — «)(1 — ~2)(1 — « )(1 — æ8)...: Me.80.
    4749 : A .54 : thN .60.
    locus of centre, th : Q .9. (1 + x)(1
    43 .
    + 2x ). .(1 + n – 1x): C.25 : J.
    * Expansions of a function in a series (1 + as + bx2 + ... + lan-1)- 1: AJ.6.
    (see also “ Series ” and “ Sum
    mation ") : A .31 : An.7 : thsAJ. {(x — a,)2 + ... + (2 - a )2} -1 : C.952.
    3 and 4 : C .7,13,17,20 : CM .4 : (1 + ax + bac2)" : Q .18.
    J.90 : L . 38,46 ,76 : M .16 : Mél.3 : (1 — 2ax + x2)-1 ;21 +L1 .372.
    Mém .33 : N .82,83 : num , Q .3 :
    2 .2 . (1 — 2ax + a%a2) ? : C.862.
    878 INDEX .

    Expansion of — (continued): Expansion of— (continued):


    (l - ax - bx2)– : J.43. do. when x = logy: 1570 .
    {(1- )(1- ...}": .40. A cot 8 in powers of sin A : Q .6.
    cot -' (n - 1) - cot -> (m + 1) : A .47.
    (a — )" in powers of 2 — 1 : C .86 . differential coefficients by f.d .c and
    (x + y)("): CM .3. the converse : J.16 .
    nth derivative of v (a? — 1222): A.42. elliptic functions : A . 19.
    and of their powers : C .83 : cos ame,
    ( 1 - V (1 4t))" in powers of t,when L .64.
    equations : L .50 .
    1 + 1 (1
    ) : 1565 . exponential functions : N .82.
    Bernoulli's numbers : 1545.
    circular functions : J.24 : Q .5.
    e = limit
    * e*: of (1+x)*:1590:A.3,23:Q.7.
    C.30: N .48.
    an arc in linear functions of sines or * ettle - 11.: 151539, 1543 – 4 .
    tangents of fractions of the arc +1
    in g.p : L .43. day in powers of yeby : 1571.
    powers of arc in powers of sines : J. 11.
    * + : 2931 - 2 , 2945, 2960 — 2 : Me.78 . (ae - 1 - : At.57.
    powers of 1 : Me.78 : 7 ?, 858 : - ,
    Me.83.
    sin 8 and cos 0 in powers of : 764 ,
    : 2962.
    1531 : A .5 ,29 : C . 16 .
    * sin " and cos" in sines or cosines of eexp.(4x + ): Z.3.
    multiple arcs : 772 - 4 : A .24 ,55 :
    C .12 : CD .3 : J.1,5 ,14 : N .71 : TI.7.
    * ear cosbæ : 798.
    e - ar sin and summation of
    * sin no and cos no in powers of sine or J. e cos bædæ,d lit , an
    cosine : 758 , 775 — 79, 1533 : C .82 : the series : J.41.
    CM .2 : Me.76 : Mém .13,15,18 : N . & exp. sin a : 1529.
    732,83 : Q .4 : convergency of the
    *

    series, J.4 . e exp. a sin -l : 1535.


    *

    cosn0 in powers of cos 0: 780 : Q . e exp. log (= + sin y) : 1557.


    *

    12. e exp. - (« ,y,z...) : 0 .582.


    sin -lx : 1528, - 64 : J.25 : N .74 : re a exp .[x exp .[æ exp , & c. : J.28 .
    mainder , Z.1 . fractions : 248.
    cos"ax : A . 11. functions :-- -
    tan x : 1525, 2913, - 17 : A .16 : 0.88 : Al(a ), Aly(« ),Al2(x)(Weierstrass'sfunc
    N .57 . tions) in powers of themodulus :
    * cotæ : 2911, - 16 : C .88 : Q .17. C .822,85,86 : L.799.
    * *

    sec æ : 1526 : A . 16 : C .88 : J.26 : N .57 :


    Q . 17. 22 ! , by Taylor's th : CM .4.
    "
    cosecx : 2914, - 8. * f(x + h ) (see Taylor's th ): 1500 - 9,
    *

    tan -la : 791. 1520 : Abel's th , 1572 ; Stirling,


    ** * *

    tan no : 760 .
    arsin æ in differences of sin æ : 3749. 1516 ; Boole , 1547: AJ.3.
    si00n nx / cosæ " : A .4. * f(a ) (Maclaurin ) : 1507 : 3759.
    COS * f (x + h, y + k ): 1512, 1521.
    (1ếu cos ) - = £ancos 2np : A .21. * f (x,y) : 1516 , 1523 : Me.3.
    (a? + 62 — 2ab cos q)exp. - (m + }):TE.5 . f (0) (Bernoulli): 1510.
    (a + b cos o + c cos d ')-" in cosines of * • (a + bx + ca ? + ...) (Arbogast ) : 1536 :
    multiples of Q and o' : J. 15 . CD.1,6.
    cos k cos-' (cos w + a) : C .15.
    sin
    (0 + 0.+ ... + 0n- 1): A. 34 .
    piz) (Cayley): 1555.
    COS
    | * f {z + æ (y)} in powers of « (La
    y in powers of a when a = siny . grange): 1552 : Laplace 's th,
    796 ,1558 . sin (y + a ): 1556 .
    INDEX. 879

    Expansion of — (continued) : Factorials — (continued ) :


    * f ( y) in powers of v (y) (Burmann ) : reciprocal: 0 . 17.
    1559. treatment by limits : J.39.
    * f {x -'(x)} and y -'(x): 1561 – 3. 1,22,33...n " : Me.78.
    * p (e) (Herschel): 3757. n ! = r (1 + n) : 2290 .
    Uz+n : 3740 ; A "u : 3761. approx. to when n is large: C .9,50 :
    A "Uz and A ”us: 3741 – 2. J.25,27 : L .39.
    * in differential coefficients of u : n ! = n"e- (217 ) (Stirling): Q .15.
    3751. {22p +1p !mp+1)}-): CM .3.
    1 . 3 . 5 ... them
    *
    A "y" and A”O" : 3743 — 4. . 239
    Unz in differences of u : 3752. ... theorem : 339.
    2 .4 .6 ---
    | u ,da in terms of Up,U1,U2, & c.: 3778 . C (n ,r) when n = a + iß : J.43.
    Factors : 1 – 27.
    a function
    19.
    of a complex variable : M . | in analysis of integral functions : M .
    15 .
    a function of a function : AJ.2,3. application to rotations to indicate
    functions of infinitesimals : G .12 . direction : J.23 .
    a function of a rational fraction : At. of composite numbers : 274 : J.11.
    65. complex : C .24 .
    a function of n variables : C .604: J. equal, of integral polynomials : 0.42 :
    66 . L .56 .
    a function of y,y' in ascending powers of an equation : 400 : J.3 : condition
    of x,x ' when y = x + xoly ) and y ' = for a factor of the form " - a",
    z'ta'oly') as in 1552 : J.48. A .55 ,63.
    holomorphic functions: M .21 : by irreducible, of an integral function ac
    arcs of circles, C .94 . cording to a prime modulus p
    implicit functions : 551, 1550 : L .81. C .86 .
    integrals : A .1 : of linear d.e, An.71 : linear, resolution into : N .822.
    of log &, A .4 . of polynomials and geo .ap : J .29,89.
    logarithms : 152 – 9 : N .82. product of an infinite number of : A ,
    59 .
    * log (1 + x),log + ", & c.: 155 — 9.
    log y and log y" in powers of a -1when coscos i cos ...: N.70.
    7 - ay + b = 0 : 1553 — 4. radical, ofnumbers : C.24 ,25.
    * log (a + bx + ca? + ...): 1537. of An2 + By2 + Cz%, th of Lagrange :
    *

    log (1 + 2 a cos x + a?) : 2922. AJ.3 .


    log (1 + n cosa ): 2933. ofæ2- fgy” = + 1 : A .33.
    of (x + y)» — an - y" : thQ.15,163.
    log 2 sin 6 ) : 2927. of w21 — 2x "Y" cos no + yan : 807.
    log 1 (1 + x): 2294, 2773. of x" — 2n cos n0 + r -" : CP.11 : Me.76 .
    higher integrals of log æ : A.4. ( 1 - x) (1 - x2) (1 - 23)... : C .96 .
    numbers : M . 21. * tables of (Burchardt's) p.7 : to 4100,
    a polynomial: 137 : Z . 26 . J.46 .
    a quartic function : A .35 . geo. properties : J .22 .
    Exponents : N .57 : P .1776 . transformation of : A .57.
    reduction for d .i : C .16 . of 100 ...01 : Me.79.
    Exponential, th : 149: N .52. Faculties , analytical : J.7,11,334, 35,40 ,
    functions : P .16 . 44,51.
    replaced by an infinite product : C .99 . coefficients of : A . 9,11 : At.75 .
    Exponentials, successive of Euler : L .45 . divisibility of: A .48.
    numerical, of 2nd order : Mém .38 .
    Factorials : — calculus of: L.57 : N . series : Z .4 .
    60 : Pr.22 : Q .12 : Q .18. * Fagnani's theorem : 6088: A.26 : LM .
    geom17.
    . i.e (1 + x) (1 + ræ) (1 + % )...: C . 5 , 13 ,23 : 2 .1 .
    curves having Fagnanian arcs: LM .11.
    notation : 94 , 2451 : Q .2. stereometric analogy : Z .17.
    880 INDEX.

    Faisceaux: of binary forms having | Finite difference eqs. — (continued ):


    the same Jacobian : C .93 . of the kind Us,y = Ux-y,ety : CM .4.
    of circles : C .76 . linear : AJ.7 : An.50 : At.65 : Q .1.
    of conics : 2 .20 .
    curvature relations: Z .15. first order, constant coefficients :
    formation of : 0 .45 : CM .3. C .8 .
    intersections of: N .72 : degree of the determination of arbitrary con
    resulting curve, J.71. stants : A .27 : At.65 : G .7.
    of lines and surfaces : N .53,83. integration to differences of any
    plane : N .53 : defined by a first order order : J.12.
    d .e , C .86 . with variable coefficients : C .17.
    of tortuous cubics in connection with partial:
    ray-complexes : 2 .19. constant coefficients : C .8 .
    Fan of Sylvester : E .33. linear of 2nd order : C .98 .
    Faure's theorems: G .1,19: and Pain . of physics : C .73.
    vin 's, N .61. Finite differences : - exercises : No.
    Fermat's theorems : - of (NP-1- 1) 44,47.
    + p : 369 : A .32 : AJ.3 : J.8. formulæ : CD.9 : Q.2.
    sum and difference : J.58 .
    of we" + y" = " being insoluble when of functions of zero : TI.17 .
    n is an odd prime, & c : An.57 : Hin , r) value of: Q . 9.
    C .gz84 and 965,91 : J.40,87: TE . integrals : 0 .39,57 : JP.42 : L .44.
    21.
    analogous theorem : J.3 . expressed by definite integrals :
    case of n = 14 : J. 9. An .53.
    and periodic functions: Me.76 . Lety : A.6 : No.44.
    a + y = 0 , x ? + y2 = 02: Mém .26 . inverse method : C .74 : P .7 .
    of thesemicircle : A .27, 30 ,31 : gzA .31. involving 11: Me.78 .
    of powers converted into di: JP .17.
    method of maxima : 0 .502. Fleflecnodal planes of a surface: Q .15.
    Feuerbach's : th of the triangle,Me.84 : Flexure : AJ.2 : Me.2.
    circle, A .59. of ruled surfaces : An.65 .
    Fifteen girl problem : E .34,35 : Q.8,9. of spaces : LM .9.
    “ Fifteen ” puzzle : AJ.2 . of spherical surfaces : Me.77 .
    * Figurate numbers: 289 : A . 5,69. | * Fluctuating
    *Finite differences, calculus of : 3701 — | functions: 2955a : LM .5 :
    3830 : A .13, 18 ,24 ,63: C .70 : J.112, FluentsM :.20P .1786
    : TI.19.
    12,133,142,152,16 : Me.82 : Mél.5 : .
    Mém .13 : N .69 : thsP .16 ,17. of irrational functions: P.16 .
    ap. to complex variability : An.82 : * Focal: 43chords 39.
    of conics : 1226 , 4235,
    ap. to i.eq, An.50 . circle of conics : Mél.2 .
    * first and nth differences : 3706 . distances : 4298 : N .64.
    A ?u = 0 : An.73. pedalproperties
    of a conic ::- N .66of. conics : 1163,
    A "On : 3744 : Q .5,8 ,9 : Herschel's table, | * Focal
    N .54. 1167 - 9 , 1181, 1286 - 8 , 4298 –
    A" ()* + II (m ), table of : CP.13. 4306 , 4336 — 45, 4378, 4382, 4516,
    duo, 8%uo, & c., in a function of Aun, 4550 — 58, 4719 - 21, 5008 - 16 :
    Aļuo, & c. : N .61. CD .7.
    A" u in successive derivatives of u : of curves : CD.7.
    3761 : N .73. of homographic figures : N .71.
    Alº and Bernoulli's numbers : An.59. of a parabola : 1220, 1223 - 6 , 1230 — 4,
    4231, 4235 — 8 : G .22.
    hu:= Auz– auét Biha sus – &c.: | of a quadric surface : An.59 : N .58.
    Focal quadrics of a cyclide: Me.85 .
    Ac.5. Foci: 5.64 : N .42,44,53,85 : Q.2,a.c9.
    A sin x and A cos x : CM . 1. * of conics : 1160 : trA .25,63,64,cn69 :
    Finite difference equations: AJ.4 : An . gz0.22 and L .39 : CP.3 : N .69,74 ,
    59 : CD ,2 : CM .1,3,4 : CP.6 : JP . 78,81,82 : t.cQ .8,13,12 and45: gen .
    6 : L .83 : P .60 : Pr.10. eq, N .48.
    of integrable form : C .54 . analogous points in higher plane
    ofmixed differences : Q .10 . curves : J .10 .
    INDEX. 881

    Foci: - of conics - (continued) : Functional equations--(continued ):


    coordinates of : 4516 . px + qij = q (æfy + yfw ): J.2.
    eq . of : LM .11 : 0.cE.40 . fiy . 0 & + f2y . 020 + & c.
    exterior : N .79 .
    to find them : Q.25: from gn.eq,5008. = fia .pıy + fx . pay + & c.: J.5.
    through four points : N .833. www . fy . Fx + fe . Fy . J 46 .
    of four tangents : 5029 : N .83. x (ay)
    locus, a cubic : M .5 .
    negative : A .649. \'f(a,@)°(x + 8)do = F (x),to find :
    under three conditions : Q .8 . Pr.8.
    ofcurves: C .82; nth class,86 : N .59,79.
    of cones : N .79.
    of differential curve of a parabola : puoautá = &c.: Q.15.
    A .58 .
    of in.conic of an n -tic, locus of : E .21. f(x)= f(sin";"):0.88.
    of lines of curvature of an ellipsoid : Functional images in ellipses : Q .17.
    Z .26. in Cartesian ovals : Q .18 .
    of quadrics : N .42,66,74 ,75,78. Functional powers : Mém .38.
    of quartics : J .56 . symbols : Q .4 .
    of theN .section
    64,70.
    of a quadric by a plane : Functions: A ..28 : AJ.6 : An.79 : C .43 ,
    91 : CP.1 : J.16 ,pr871,74 ,84,87,
    by another quadric : N .47. 91 : L .45 : Me.7 : M0.80,81 : P .15 ,
    of surfaces : 0 .74 : of revolution , N . 16 ,17 ,62 : Pr.112 : prsZ .26 .
    74 . algebraic, alternating, analytical, cir
    *Folium of Descartes : 5360 : N .44. cular , circulating, conjugate, con
    Forms, theory of : M .18 : of higher de. tinuous, curvital, cyclotomic, de
    gree, M0.83: Pr.38 . rived , discontinuous, elliptic, even
    reciprocity principle : An .56 . . and odd, exponential, Fuchsian ,
    Formulæ : G .15,19. gamma, generating, hyperbolic,
    for log 2, & c.: Me.79. implicit, infinite, imaginary, in
    in the Fund . Nova : Me.76 . tegral, irrational, irreducible,
    * of reduction in i.c : 1965 : Me.3. isotropic, iterative, monodrome,
    Four colors problem : AJ. 27. monogenous, monotypical, non
    Four-point problem : E .5,6 ,8,. uniform , periodic, polyhedral,
    Four right lines not 2 and 2 in same quantitative, rational,representa
    plane: J.5. tive, transcendental, trigono
    Fourier - Bessel function : J.69 : M .3. metrical : see the respective
    * Fourier 's formula in i.c : 2726 - 42 : headings.
    CM .3 : J.36 ,69 : L .36 : M .4 : Me. analogous : to algebraic functions :
    73 : Q .8 : gzZ.9. 0 .89.
    ap.to calculation of differentials : J.13. to circular functions: C .84 .
    * Fourier's theorem : 518 : 528 : An.50, to Euler's : C .89 : M .19.
    75 : J.13 : M .19, : Me.77,82,83 . to functional determinants : J.75 .
    ap. to a function of a complex varia sine and cosine: Q .16 .
    ble : M .21. to modular functions : Ac.2 : C .93.
    Fractions: AJ.32: G .9,pr16 : J.88 : L . 10 . connected by a linear eq . : 0 .17 .
    continued ,decimal,partial,vanishing, condition of f (x , y ) being a function
    & c . : see each title. of $ (X , Y ) : A .21.
    number expressible by digits $ n : development of: see “ Expansion.”
    C .96 . defined by d .e : JP .21,28.
    reduction to decimals : A .1,25 . differing very little from zero : L .74.
    transformation into decimals : A .11. errors of geometricians: J.16 .
    * Frullani's formula : 2700 : LM .9. expressed by other functions, remain
    Fuchsian functions : C .922,93,943,95,96 . der : C .98.
    Fuchs's theorem on F (2, 4, :) = 0 : C .99 . fractional: J.8 : the variable being
    Functional equations : CM .3 : J.90 : the root of an equation , N .56 .
    TE . 14 . from functional equations : M .24 .
    f.ox = fæ : C .88 . from Gauss's equation : C .92.
    f .pu = 1 + fx : C .99. with lacunæ : C .96 .
    0 .fc = F . px , to find o : Mém .31. Lagrange, tr : JP.5,7.
    5 U
    882 INDEX.

    Functions - (continued ) : Functions — (continued ) :


    linear : 0 .90 . which relate to the roots of the equa
    with linear transformations inter se : tion of division of a circle or of
    M . 19, 20 . MP - 1 = 0 : J.17.
    whose logarithms are the sums of representation of : C .923 : M .17 : one
    Abel's integrals of the 1st and valued , Z .25 .
    3rd kind : C .92. approximate : Z .3.
    with non -interchangeable periods : by an arbitrary curve : M .22 .
    M .20 ,21,25 . by Bessel's functions : M .6.
    number of values of : C .48 . by definite integrals : Ac.2.
    do. through permuting the variables : by elliptic functions: An.82.
    JP.10.,18 : L .50,60 . by Euler 's sum -formula : J.56 .
    of two variables : Ac.3 : C .90,962. by Fourier 's series : M0.85g.
    made constant by the substitution by graphic methods : A . 2 : imag.,
    of a discontinuous group : C .97. J.55 .
    which arise from the inversion of by infinite products : Z .24.
    the integrals of two functions :
    C .92 . y = etry, constant and r a positive
    whose ratio has a fixed limit : G .5. integer : A .42.
    f(x,y) such that f {zf(x,y)} is sym y = x"etzt: A .52.
    metrical : J.1. reproduced by substitution : C .19 .
    of three variables satisfying the d.e, resolution into factors : Ac.65 : C .19 ,
    AF = 0 : Ac.4. 30 : CP.11 : J.18.
    of three angles, th. re 1st derivatives : satisfying the eq. AF = 1 : C .96 .
    J .48. singularities of : M .19.
    of 4 and 5 letters : L .56 . whose successive derivatives form an
    of 4 , 5, and 6 letters : L .50. arith . prog.: An.71.
    of 7 letters : C .57, 95 . systems of: Mo.78 : of two inter
    of 6 variables which take only 6 diffe. connected , C .98 .
    rent values through their permu. of two systems of quantities , cor
    tation , not including 5 symmetri relative and numerically equal :
    cal permutations : A .68 . C . 98 .
    of n variables : C .21 : Mo.83 : with which are neither rational nor reduci.
    2n systemsof periods, C .97. ble to irrational algebraic ex .
    analogous to sine and cosine : Q . 16 . pressions: C .18g.
    number of values : J.85 .: do. by which are of use in elliptic functions
    permutationofthe variables: C .214. and logarithms : No.58 .
    obtained from the inversion of the which take a given value in a given
    integrals of linear de with position : An .82 .
    rational coefficients : C .90,: J.89. which have noderivative throughout a
    of an analytical point, ths: C .952. certain interval : An.77.
    of a circular area from a given inte which vanish with their variables :
    gral condition : Z .26 . TI. 162.
    of imaginary variables : 0 .32,48 : JP . 20* : An.63.
    21 : L .58,593,603,613,62 : LM .geo {(23)* }* and so on, and the corres
    8.
    of large numbers, approx. : C .203. ponding inverse function : J.42.
    of a real variable, connexion with arising from „ (4 — 2xz + z"): J.2.
    their derivation : M .23,24 .
    of real arguments, classification ac. 0( )=_ axcx ++ db . LM .9.
    cording to their infinitesimal
    variation : J.79. f (u,z ), u being an implicit function
    of the species zero and unity : 0 .95 . of an imaginary variable z : Pr.42.
    of a variable analogous to the poly f (x ), formula of analysis : J.53.
    nomials of Legendre : C .95 . f (x) = 0, y = f (x), th . re o (y): E. 36 .
    allied to Pfaffians: Q .16 . ¥ (x) = d ,log r (a): 2743.
    rationally connected : L .59.
    with _ recurring derivatives : LM .4 : v (a) of Jacobi: J.93.
    TE .24 . Q (w ) : Ac.2.
    INDEX. 883

    Functions— (continued ) : Functions — (continued) :


    Bessel's,I(a)= 3 cos(a cosO)do sv {1 + ax3} dæ : J.92.
    26
    Y (, ) _ E (p , ° ) do
    = 1- + 2.42 - 22.42.62 + Jo v {1 - p2sin ?
    X , Y, & c., such that SXYdo = 0 and
    Cauchy's numbers ; N-kip and that any function can be ex
    panded in the form aX + BY + & c.:
    = IS*- (einte-imp x LM .10.
    (ein — e-iu) du. *Gamma function , T (n ) : 2284 : A .4,6,61:
    integral;
    cosinetaP.70 .
    Ciq =f*cos * dx : An.69 : 0 .35 ,92,96 : J .35 ,82.90,
    ap 57 : L .42,46,52,55 : Q . 9 : 2 .1,
    Dirichlet's function , 25 .
    application of this and other trans
    F(x)= >( ) : 2.27.
    F (x ) = x1 cendents : C .86 .
    of a complex : Me.84 .
    elliptic ;
    2.3

    curve y = f (x ) : 2323.
    deductions : 2286 — 2316 : A . 10 .
    2.3

    [x =»-1F (**){R (x2)}**dx :J.23. derivatives of : Q .6 . .


    Euler's; B (1,m ): 2280. of equidifferent products : J.36 .
    of an infinite product : J.39.
    expon-integral;
    10 : tap. 70.
    Eiq =f dæ: A. = limt. of 4! Un: 2293 : A.30.
    sin ?" w cos(e cosw ) dw : An . * logarithm of : 2294, 2768 : C. 9.
    numerical calculation : 2771.
    * *

    70 . as a definite integral: 2768 .


    E (w ): Mél.6. n , negative : CD . 3 .
    numerical calculation of : 2317.
    G = ftan-lædæ = 1- + -&c. the function y (a ) = dx log 1 ( c) : 2743
    – 70 .
    = •915965,594177 ...: Mém .83. reciprocal of : 2 .25 .
    r (w ) : see “ Gamma function.” reduction of : J .40.
    Jacobi's transformation of: 2284, 2318 : J .57 .
    r (n + 1) = (27 ) e -" * * ( 1 + e) (Stir
    ( ) =(-1) exp -1 (bi): ling): 0.502.
    C.592,60 : L.47,50. *r (m) (1 –m ) = : 2313 : Ac.3.
    Laplace’s Y (*): M .14 .
    19 de
    log-integral ; Liq= lloga *T(*)F (x+ +)... (*+" ?):2316:
    L .55,56 .
    L"(1+x)=2 + - &c. r (x) = P (@ ) + Q (Q ): Ac.2.
    Legendre's X,: see “ Legendre." G .6.
    "& T (x1 ) :
    P,whereſeexp
    Q . 10.“
    ( ) de = "va: Gauche cubics : C .82 : J.60 : N .623.
    3rd class, theorems : J .58 .
    P"(cosy),n = 0 : G .22. number of common chords of two :
    An.70 .
    plate ): 2.14. through five points : N .83.
    through six points, cn : N .66 .
    11(2)= (x*e->dæ; y (2)=d;log 11(2): Gauche curves : C .70,77,90z.
    Mo.82 : thsN .532.
    Q.1. classification of : JP.32.
    sine- integral; Siq = & de : taP. on a cubic surface : C .62.
    70. of a developable surface, singulari
    ties : An.70 .
    Eſe exp ( *) F (z)dz = 0 : C .93. differential invariants of: JP.28 .
    884 INDEX.

    Gauche curves - (continued) : Geodesics — (coutinued) :


    intersection of two surfaces having flexure of: C .66 .
    common multiple points, singu . forms from cn of their polar systems:
    larities of : C .80. Z .247.
    on a one-fold hyperboloid : An.1 : C . geometry : 5855.
    52,53.
    metric properties of, in linear space of
    problems: A .8 : An.65 : CP.6 : J.53.:
    | L. 49 : TN, 73.. .
    n dimensions : M . 19. on a quadric or ellipsoid : C .222 : CD .4 :
    representative curve of the surface of J.19 : L .42,41,44,48,57 : M .35 : N .
    principal normals of : C.86. 76 .
    of the zero species : C .80. Joachimsthal's theorem : J.42.
    Gauche helicoids : rad. of curv. : N .45. and corresponding plane curves
    in perspective : JP.20. C .50 .
    Gauche: - in -polygons of a quadric : and lines of curvature : L .463 : N .
    C .822. 82.
    perspective of algebraic curves : C .80 . pd constant along such : 5842.
    projection : N .65. shortest lines : J.26 .
    quadric : N .67 : and orthogonal tra through an umbilic : 5850 .
    jectory of generatrices : thN .48 . on a right cone : A .69 .
    quartic : A .62 : C .82 : L .70. * radius of torsion of : 5848 : RR 'Pz +
    9 points of, 7 points of a gauche P = 0, Me.75 .
    cubic and 8 associated points : sections : Z .2.
    C .98. sbortest lines on surfaces : 5838 : J.
    unicursal, a class of : C .83. 20 .
    surface : N .61. on a spheroid : J.43 .
    sextic curve : C .76 . triangles : Mo.82.
    surface : JP.17 : L .37,72. best form of: N .55 .
    deformation of : C .57.
    which can be represented by a p .f.d .e
    reduction of arc of a small one: An.
    50.
    of the 2nd order : 0 .61. Geodesy, spherical problems: A .25.,63.
    Gaussian periods of congruent roots representation of one surface upon
    corresponding to circle division : another : An.70 .
    J .53. Geodetica , integration of its eq : An.
    *Gauss'sfunc tion
    function ."
    : see “ Hyp erg eom etr ic 53.
    Gauss's theorems: J. 3. Geography, comparative: A .57.
    * Geometrical conics : 1150 — 1292 (see
    *General methods in anal. geometry : Contents p. xviii) : G .1 : Me.62,
    4114. 64,71,73 : Q .10.
    General numerical solution of any Geometrical: - constructions : LM .2.
    problem : LM .2 . definitions : J.1.
    *Generating functions: 3732 : J.81 : N . dissections and transformations: Me.
    81 : Pr.5 75 .
    and ground-forms of binary quantics drawing : A .23.
    of first ten orders : AJ.23,3. figures, general affinity of : J.12.
    do. of binary 12-ic and of irreducible forms: G .1 : of 2nd species, G .3,4 .
    syzygies of certain quantities: mean : 92 : CD.8 : Pr.29 .
    AJ.4 .
    of some transcendental series : At.55. approx. to by a series of arith , and
    for ternary systems of binary forms, harm . means : N .79 .
    ta : AJ.J. paradoxes : Z .24 .
    *Geodesics : 5775, 5837 — 55 : A .39 : 0 . progression : 83 : A .pr2,6 : G .11 : N .
    402,41,96 ,p .c97 : CD .5 : G . 19 : J. 54.
    50,91 : M .20 : Me.71 : N .45,65 : Q . a property of 1,3,9,27...: A .33.
    1,5 : Z .18 ,26 . proportion, theory of pure : A .62.
    of cubic surfaces, loci: CD.6 . quantities and algebraic eqs: 0 .29.
    curvature : 5846 : C .42,80 : L .12 . reckoning (Abzählenden ) : M .10.
    on an ellipsoid : M .20. relation of the 5th degree : M .2 .
    radius of : 5776 , 5846 . relations, ap of statics : J.21%.
    duals of : Q .12. signs : Z .14 .
    equations of ; 5837. theorems and problems: 920 — 1102,
    INDEX. 885
    Geometrical - (continued) : Groups : AJ.1 : LM .9 : M .13,20,22.
    theorems: J.11 : L.46 : N .743: Law cyclic, in Cremona's transf. of a plane
    son 's, Man .13. An.82.
    method of discovering : J.8. in 82.
    a quadratic transformation : An.
    from a principle in alg. : LM .11.
    problems: At.25 ,32. discontinuous : C .94 .
    transformations : A .32. of linear substitutions: Ac.1.
    and ultra -geometric quantities : C .52, of finite order contained in a group of
    55 . quadratic substitutions : 0 .97 ,98 .
    Geometry : of the Ancients : At.22. of finite order contained in the semi
    comparative, ap . to conics : N .653. cubic groups of Cremona : 0 .99.
    of derivation : An.543. Fuchsian : Ac.1z .
    elementary : 920 — 1102 : J.6,10 : A .2 : formed from a finite number of
    N .623. linear substitutions: C .83.
    principles of : A .40 : C .56 : G .11.14. of interchangeable elements : J .86 ,.
    20 : LM . 16 : Me.62 : 2 .20 . introduction to the theory of : Me.62
    enumerative : Ac. 1. Kleinean : C .93.
    higher : A.20 : No.73 : prs A .55. of many-valued functions: Man.62.
    instinct of construction : N .56 : Q .2 . modular eqs. (Galois ) : M . 14, 18 .
    der Lage : 2 .6. non -modular : Man .652.
    linear : A .27 : ap. to quadrics, M .10. of points G 's on a sextic with 5 double
    linear and metrical: M .5. points : M .8 .
    of masses : JP.21. primitive : C .72,78 ,96 : L .71.
    and mechanics, on their connection : for the first 16 degrees : C .75
    L .78 . degree of, containing a given sub
    organic, of Maclaurin : L .57. stitution : J.79.
    plane, new anal. foundation : M .6 . (P)360 (II)360 of the figure of six linear
    plane65 and solid ,analogies : L .36 : N . complexes of right lines two and
    3. two in involution : An .83 .
    plane and spherical, ths : Mém .15 . principal, classification of : C .73.
    of position : J.50 : Q .1 : analTE .9. of substitutions: C .67,84,94 : M .5 :
    theorems: An.55 : J.31,34,38,412 : isomorphism of, G .16 .
    TE.28. of 168 substitutions and septic equa
    5 points in space : CM .2. tions : M .20 .
    in lieu of proportion : CP.10 . transitive : G .22 : J.83 : N .84.
    of space, abstract : Pr.14 ,18 : aphor- | *Guldin 's theorems : 5879 : Me.85.
    isms, J.242.
    of three dimensions: 5501 – 6165 : N . Harmonic axes : of curves : C .743.
    63. of a system of right lines and planes:
    Glissettes :- problem : Q .11. G .4 .
    centre of curv . : JP.21: rad . of curv., Harmonic centre for a system of 4 points
    L .45 . in relation to a given pole : 2 .20 .
    Golden section : A .4. Harmonic division : - ofa conic : G .10.
    Goniometrical problems: Q .7,15 . of a quadric : G .10 .
    Graphic calculus : 0 .89. * of a right line by a circle and chord of
    Graphs (Clifford's): LM .10. contact : 948.
    application to binary quantics : LM . Harmonic: - hexhedron and octahe.
    172. dron : Z .18 .
    to compound partitions: AJ.96. * pencils and ranges : 933, 4649 : Q.6.
    Grassman's life and works: M .14 . of 4 tangents to two conics,locus of
    Greatestcommon measure:see " Highest vertex : 4984 .
    common factor." points, system of four : 1063 : N .51.
    polar curves : A .50 : M .2.
    Greatheed 's theorem , D .C : CM .1. progression or proportion : 87 : ext
    Grebe's point : A .58 . of th , A . 31,43 ,tr41 : C .43 : Me.82 :
    Green's26 theorem
    .
    , & c.: J.39,44,47 : TE. N .85 : Z .3,14 : sum of, Pr.20.
    divergency of : A .1.
    *Griffith 's theorem (Conics): 6096 . / * section by a quadric and polar plane:
    Ground figures, single and double rela 5687.
    tions ; J.88 . | Harmonics in a triangle : A.57.
    886 INDEX.

    Hermite's o function, linear transf. of : 1 *Homogeneous products, H (n ,r) : 98


    M .3 . 9 : Q.6,9,10.
    * Helix : 5756 : A .64 . * and sumsof powers : 538 : E .39,40.
    conical : N .13,53 : rectif. of : N .45. Homographic division of three tangents
    on a twisted cone: A .16 . to a conic : Mél.2 .
    relation with cycloid : C .51. Homographic figures : threeC .94 : thQ.
    *Hemisphere, volume, & c.: 6061. 3 : N .58,68,pr6l.
    Herpolode of Poinsot : C .99. corresponding points, th : L .45.
    Hesse's surface, & c.: Z .19. focal properties : LM .2,.
    *Hessian : 1630 : J.80 : curve, M .13. relation of roots : N .73.
    covariant of binary quintic form : M . * Homographic : - pencils : 4651.
    21. systems of points : 1058 — 73.
    of a quaternary function : Q . 12 : on quadric scrolls : Q . 9.
    cubic, Q .7. theorem of a conic : N .48,49.
    of a surface: nodes of, J.59 : con transformation : N .70 : of angles, Q .14.
    stant of, M .23. Homography : Me.62 : N .60 : Z .21.
    Hexagon : thN .65. and perspective : N .69.
    Pascal's : see “ Pascal.” and rotations, correspondence of : M .
    in space
    dron
    : J.85,93. 15 .
    * Hexahe : 907. Homological polar reciprocal curves :
    Higher algebra : An.54 : N .552 (Serret) : ET.44 .
    Q .45. * Homology : 975 : G .3,8 : N .44 : E. 24.
    Higher : - analysis : A .25 : G .14 . conic of: C .94.
    arithmetic : J.6 ,9 : N .81. of sets : Q .2.
    geodesy : Z .19 : trZ .13 . of triangles : 975 : Me.73. .
    geometry : A .10 : N .57 : Z .6,17. | * Homothetic conics: 4523: N .64,th68 .
    planes : A .47. with the same centre : C .66 .
    variation of simple integrals : Z .22. *Horograph : 5826 .
    * Highest common factor : 30 : A .3 : M . * Hyperbola : — theorems: A .27,46 :
    7. : N .42,44,45%. CD .1 : N .424.
    of 2 complex numbers : no. of divi. | * with asymptotes for coord . axes :
    sions : L .46 ,48. 4387 : Me.73.
    of 2 polynomials : CM .4. * asymptotic properties : 1182.
    remainder in the process : 0 .42. conjugate : 1160.
    Holditch 's theorem : see “ Closed construction : 1247, 1289.
    curve." eccentric circles : A .44.
    Holomorphic functions : C .99 : G .22. quadrature of, & c. : 6118 : A . 25,26 ,27:
    development in series : C .94 : M .21. N .44 : TI.7 (multiple areas).
    Homalographic projection : N .61. rectangular : 4392 : 2 .26 : under 4
    Homaloidal system , n -tic surface and conditions, A .3 .
    an (n - 1)-ple point : G .13. segment of : 6118 : N .61.
    Homofocal: - conics : thN.49, : loci | * Hyberbolic arc, rectification of: 6115 :
    relating to parallel tangents, C . J.55 : P .2,11,59.
    62,63, * Landen 's theorem : 6117 : LM .11,13..
    quadrics : 0.50 : L .th51,60 : N .th64, | *Hyperbolic functions: 2180 : A .19 : G .
    79 : Pr.332. 15 : Mém .30 : N .64.
    paraboloids: A .35. analogy with the circle : An.51.
    and conjugate surfaces, tr : Z .73. ap. to evolution and solution of eqs. :
    common tangents of : C .22 : L .46 . A .38 .
    quartic surfaces, triple system of, in construction oftables of : J.16 .
    cluding the wave surface : N .85 . generalization of : A . 35 .
    sphero -conics : L .60. | *Hyperboloid : 5605 : J.85 : Me.66 .
    surfaces, and u sin’ i'+ v sin2i" = a? : theorems: geoG .4 : J. 24 ,86 .
    C .22. one-fold : 5605 : of rotation , A .70 :
    Homogeneity of formulæ : 0.96 : thsN . L .39 : M . 18 : N .58 .
    49. parameter of a parabolic section of :
    Homogeneous coordinates : G .13,8 : Z .15. N .75 .
    metrical relation : G .11. two-fold : 5617 : A . 18,ths27.
    Homogeneous functions : see “ Quan conjugate : CD.2.
    tics . " equilateral and of revolution : Ac.5.
    INDEX. 887

    Hyperboloid — (continued) :
    * generating lines of : 5607 .
    Hyper-Fuchsian functions from hyper
    geometric series of two variables :
    and relation to ruled surfaces : Z .23. 0 .99.
    N .72.
    of revolution :projection Hyper -Fuchsian groups : 0 .98 ,.
    Hyperboloidic of a cubic Hyper- Jacobian surfaces and curves :
    “ gobba ” : An .63. LM .9 : P .77 : Pr.26 .
    Hypercycles : C .945. Hyper- geometric functions or series :
    Hyperdeterminants : CD.9 : J.349,42. 291 : A .55,57 : J.152,75 : M .3 : Q .
    16 : Z .8, 26 ,27.
    Hyper- elliptic functions : A. 16 : AJ.53, as continued fractions : 291 — 2 : J.66 .
    7 : An.70 : C .408,622,67,92 ,94 ,97 : of two variables : C .902,91,95 : L .822,
    CD .3 : J.25,27,30,40,47,52,54 ,75 , 84.
    76 ,81,85: L .54: M .3 ,11,13 : Q .15, 19. extension of Riemann's problem :
    of 1st order : J.12,162,35,98. C . 90 .
    containing transcendents of 2nd of nth order : 0.96 : J.71,72 : M .2.
    and 3rd kind : J.82. and Jacobi's polynomials : C .89.
    multiplication of : Ac.3 : M . 17,20. square of : J.3 .
    transformation of: Ac.3 : (p = 2), Hyper-geometric integrals : J.73: Z .22.
    M . 15 .
    transf. of 2nd degree : M .9 : M0.66 . *Hypocycloid : 5266 .
    with 2 cusps : Z .19.
    transf. of 3rd degree : M .1,193. with 3 cusps: J.64 : Me.83 : N .702,75 .
    transf. of 5th degree : M .16 ,17,20. Hypsometric tables of Bessel: Pr.12.
    of 3rd order (p = 4 ) : M .12.
    of 1st order and 3rd kind : J.65,68,88 . * Icosahedron : 907 : M .12,25 : and star
    of 1st and 2nd kind : An.58,: J.93 : dodecahedrons, Z .18.
    in series, M .9. Icosian game: Q .5.
    of 3rd kind, exchangeability of para - | * Imaginary : - quantities : 223 : A .20.
    meter and argument : J.31. 22 : 8 square, AJ.4 : C .18,24,25 ,
    of nth order, algebraic relations: C . 88.9,94 : JP.23 : N .63,64 : P.1,6,31.
    993. ap. to primitive functions of some
    Göpel's relation : An.82. derived functions : N .63.
    addition theory : M .7. conjugates : 223 : modulus of, 227.
    addition th . for 1st order in a system logarithm of : 2214 : LM .2 .
    of confocal quadrics : M .22. curves : Q .7.
    approximation to : P .60,62. exponents : A .6.
    choice of moduli : C .88 . integrals of d . e : C .23.
    division of : C .68,98 : L .43 : M .1. prime factors of complex numbers
    bisection : C .702 : trisection , An.76 : formed from the roots of irreduci
    M .2. ble rational equations : Z .10.
    generalisation of: C .84,98. transformation of coordinates : Q .7 .
    geo. representation : L .78. variables : C.96 , : polygons of, Č .92.
    inversion of : C .99 : J.70 . VVatib in the form ætiy : A .55.
    in logarithmic algebraic functions : tan - ($ + in ) in the form tiy : A .49.
    M .11. ( , y) + iy (x, y) = F (x + iy) , to deter
    and mechanics : J.56 . mine p and y : A .10 .
    periodicity moduli : A .68 : An.70 . geometry : 4916 : A . 32 ,,61 : 0 .61 :
    periodic : J.32 : of the 1st class , LM . CD.7,8 : J.55 ,70 : M .11 : Me.81 :
    12 ; with four periods, An.71. N .70.,722 : TE .16 .
    with quartic curves, 4 tables : M .10. of Lobatschewski: G .5 : J. 17 : N .
    reduction of, to elliptic integrals : 683.
    Ac.4 : C .855,93,99 : J.55 ,76 ,79,86 , of Standt : M .8.
    89 : M .15 : TI.25. use in geometrical drawing : J.1.
    transformation of 2nd order, which , * circular points at infinity : 4717,4918 ,
    applied twice in succession , pro 4935 : tg.eq4998, 5001 : Me.68 : Q .
    duces the duplication : C .88. 3,8,32.
    transformation : M .7,pr13. coordinates : 4761 : C .75 : Man.79 :
    arguments, complex mult. of:
    oftwoM .21. homog.Q .18.
    elements in geometricalconstructions,
    Hyper-elliptic 6- functions, alg. charac and apparent uncertainty there.
    teristics : M .25. from : 2 .12.
    888 INDEX

    Imaginary — (continued) : Indeterminate equations — (continued) :


    * lines through imaginary points : 4761, and congruences : Pr.11.
    4722 — 3 . quadric : J.45 : in n unknowns, N .84.
    problem , Newton - Fourier : AJ.2. quartic : geo.cnL .63 .
    tangents through the focus of a conic : quintic : J. 3.
    4720 — 1 , 5008 : A .22. quadratics in two unknown integers :
    variables, generating polygons of a 2 - ayº = + 1 : A .12,52 : E .23,28 :
    relation between several : JP .30. J.17 ; by trig . : L .64 – 66 : N .78 .
    atiy = " X + iY , and the lemnis de - a y: = b : C .69 : L .37,38 : Mém .
    catic coordinates of the nth order. 28.
    Implexes of surfaces : 0 .802. aa— ayº = + 4 , a = 5 (mod 8) : J.53.
    * Implicit functions : - of one independ. cca + yº = (a ?+ b2)* : C.36 : An .53.
    ent variable, « (xy) ; values of os, ac2 + y2 = 0 : geoA .55.
    p2r, P3. : 1700 - 6 . (n + 4) xi- nya = 4 : N .83.
    the same when + (x ,y ) = 0 is also
    ax? + bx = y*: L .76.
    *

    given : 1718 – 9. ax ? + bx + c = yº : G .7.


    Yr, Y2 , 43, when • (x,y) = 0): 1707—
    2x2+ 2x + 1 = y?: N .78 .
    *

    16 : Ynr, An.58.
    Pr(u,y,z), P21 (1,y,z): 1720 — 1. ax? + bay + cy' = 0 : geoC .9 .
    re? + nxy - ny' = 1 : N .83.
    * *

    Px (x, y , %, $) when 3 eqs. connect


    X , Y , Z , $ : 1723. æ: — y = xy impossible : N .46 .
    * of two independent variables : axº + bxy + cyº + dx + cy + f = 0 : 0 .87.
    quadratics in three unknown integers:
    Yr: when ° ( , y,z) = 0 : 1728. meº + y + z = ( : geoA.55.
    * 0,(e y z), Dar, Pra, when 4 (x , y,z) æ ? + y : = z2 : A.22,33 : E.30 : G . 19 :
    = 0 : 1729 – 32. solution prior to Diophantus,
    * ° (x, y,z, &,n) when 3 eqs. connect C .283.
    X, Y,« , , : 1735. æ? + ay = 72: N .78.
    * of n independent variables : 1737 : ax ? +by: = zº: G .8
    An. 58 .
    transf. into isotropic means and trig . a * + ay: = : N .78.
    series : C .38 . a?Fay: = 42 : N .72 .
    transf. into explicit functions : C .38. æ° + a (x + b)2 = y : N .782.
    defined by an alg. eq . : C .47 . (x2 + ky?)= = ax + bky : J.49.
    determined by the infinitesimal cal (a, b,c, d ,e, f / xyz) = t : Pr.13.
    culus : C .34.
    * Increment : 1484 . 202 + y2 + 1672 = nº: Mél.4 ; = 4n + 1,
    Incommensurable : numbers : JP. L .70.
    15 : N .43. quadratics in four unknown integers :
    limits of numbers : N .81.
    lines : N .44 : in ratio 13 : 1, A .3. me? + y2 + z = ts : C.66 : N .48 ; 2a + 2y*
    * Indeterminate coefficients : 232 , 1527 : + 372 = t, L .69.
    A .3 : J.5. y = z? + +(x + B ) : N .789.
    * Indeterminate equations (see also quadratics in five unknown integers :
    “ Numbers ” and “ Partition " ) :
    188 — 94 : C .10,th78,88 : G .5 : J.9 ,. aea + by + cz? + dt = u ,with the follow
    Mém .44 : N .44,45 ,pr57,592,71,78 , ing values of b , c, d : 1, 1,1 ; 2,3,6 ;
    81,prs81,85 : TE.2. L .45 : 1,1,2 ; 1,1,4 ; 1,1,8 ; 2 ,2,2 ;
    ap. to a geo. problem : Mém .20 : Z .20. 2,4, 8 ; 4 ,4 ,4 ; 3 ,4 ,12 ; L .61 : 1, 2 ,4 ;
    impossible class of : N .63. 1,4 ,8 ; 2,2,4 ; 2,4,4 ; 2,8,8 ; 4 ,4,8 ;
    linear : JP .13 : L41 : N .43 : P.61 : 4,4, 16 ; 4 , 16 ,16 ; 8,8,8 ; 8, 8,16 ;
    Z .19 . 8,16 , 16 ; 16 ,16 , 16 ; L :62 : 2, 3,3 ;
    with 2 unknowns : 188 - 93 : A .3,7 : 3,a ,3a ; L .66 : 1,3,3 ; L .603, 63 :
    J.42 : L .63,69 : Mém .31. 1,1,3 ; 1,2,6 ; 2,2,3 ; 2,4,6 ; 4,4 ,
    with 3 unknowns: 194 : G .2. 12 ; 1,1,12; 2,2,12 ; 1,4,12 ; 1,
    3 ,4 ; 3,4 ,4 ; 4,12,16 ; 3,6 ,6 ; 3,3,3 ;
    with n unknowns : C . 94 : N .52. 3,3,12 ; 3,12,12 ; 12,12,12 ; L .63 :
    Xi + 2. 2 + ... + nan = m : G .1.
    INDEX. 889

    Indeterminate equations — (continued) : | Indeterminate equations — (continued ) :


    1,1,5 ; 2,3,6 ; 5,5,5 ; L .64 : 1,5,5 ; n -tic solution by aly. identities :
    1,6,6 ; 1,9,9 ; 1,n ,n ; 2,1 ,2n ; L .65 , C .873.
    59: with crab , C .42 and L .56 . " + y" = ą" impossible if n > 2 (Fer
    ax2 + by2 + cz2 + dt = u ,with the follow mat's last th .) : A . 26 ,58 : An .64 :
    ingvalues of a , b ,c, d : 2, 2,3 ,4 ;2,3, C .24,89,90 , 98 : J.17.
    3,6 ; 3,3,3,4 ; 1,2,6,6 ; 2,3,4,4 ;
    L .66 : 2,2,3,3 ; L .65 : 3,4,4,4 ; 221 — y ?" = 22" : L .40.
    3 ,4,12,48 ; L .63. aam + by = cz" : N .794.
    m2 + y2 + 2 + t2 = 4u : L .56. pro " = y + 1, impossible : N .50,70,71.
    ax2 + by2 + cz2 + dťa + exy + fzt = u, with 2 - ay ? = % : C . 99 .
    the following values of a , b, c, d , simultaneous :
    e, f : 1,2, - 2,2,1,2 ; L .63 : 1,2,3,3 , x = u ? ; æ + 1 = 2v2; 2x + 1 = 3w2:
    1,3 ; 2,2,3,3,2,3 ; 1,1,6,6,1,6 ; L .64 : N .78 .
    2 ,3, 2,3,2,2 ; 2,5,2,5,2,2 ; L.64 : 1,1, &? ta = y2; y - a = z2: An.55: C .78.
    2,2,0, 2 ; 1,1, 1, 1, 0, 1 ; 1, 1,1, 1,1, 1 ;
    2,2,3,3,2,0 ; 1,1,3,3,1,0 ; L .63 ; 3,5 , 22 + x + 2 = y ?;22 — & — 2 = 72: N .76.
    10, 10 ,0,10 ; 2,3,15, 15,2,0 ; 2,3,3,3, ax + by + c2 = 0 ; Ayz + Bza + Cæy
    2 ,0 ; L.66 . = 0 : A .28.
    Q2 + 2y2 + 2x2 + 3t2 + 2yz = u : L .64. 202 + 42 — 72 = 0 ; 22 - y2 + 2 = 0 ;
    2. 2 + 3y + 372 + 312 + 2yz = U : L .66. — Q2 + y ? + z2 = 0 : E .20.
    m2 + y2 + z? + 2u² + 2uv + 202 + t2 = w : ? + axy + y = 0 ; y? + ayz + z = 0 ;
    L .64 . 72 + aza + 22 = 0 : E .20 ,21.
    w + y2 + 2z2 + 2zł + 2t2 + 3u2 + 3v2 = w :
    L .64. a + y + z = 0 ; ? + y + m2 = 0 ;
    2(z2 + xy + y2) + 3( 2 + t2 + + + v%) = w : 23 + 73 + z3 = 0 : E. 17.
    L .64. six eqs. in nine unknowns : N .50.
    ay + y2 + zt + tu = v : C .622: L .67. exponential,wy = ył : A.6 : Z.23.
    a ” - 61 = 1 : N .57.
    g = a + + ... + : G.7. te. | * Indeterminate forms: 1580 — 93 : A. 26 :
    quadratics in seven unknown inte AJ.1 exponential : J.1 : Me.75 :
    gers :
    2 + ay2 + bz2 + ct? + du ? + Cv2 = w , with
    the following values of a, b, c, d , e : N.48,77:0,4.21;2.1.20;L.41,
    44,4,4,4,4 ; 1,4,4,4,4; 1,1,
    ,4,4 ; 1,2,2,4,4 ; 2,2,4,4 1,1,
    1,4 ,4 ;,4 ;1,1,2, 42:N.46:f( ) when < = 00 , 1592:
    2 ,4 ; 1,1,1,1,4 ; 4 ,4,4,4, 16 ; L .65 : 0°, J.11,12 : 0 exp 0“, J.6.
    1,1,1,1,1 ; 1,1, 1,1,2 ; 1, 1,1,2,2 ; , * with two variables : 1592a .
    1,2 ,2 ,2 ,2 ; 2 ,2 ,2,2 ,2 ; 2, 2,2 ,2,4 ; 3,3 , * Indeterminate multipliers : 213, 1862 ,
    3,3,3: L .64 . 3346 : N .47.
    higher degrees : * Index law : 1490 .
    cubic : AJ. 2. Indian arithmetic, th : L .57 : calculation
    23 = yº + a : N .782,83 : with a = 17 , of sines, N .54 .
    N .77. * Indicatrix : 5795 : C.92 : Me.72: N .74.
    a3 + y3 = uz3 : N .78,,80. * a rect. hyperbola, condition : 5824 .
    23 + 43 + 23 + 23 = 0 : A.49. two coinciding lines : 5825 .
    an umbilicus : 5819 .
    ax + + byt = %2 : C .87,91,94 : N .79 : to determine its axes : L .78 ,82.
    a = 7, b = - 5 , L .79. determination of a surface from the
    x+ 2y4 = ? and similar eqs : L.53. indicatrix : A .59 .
    æ4tax? y2 + yt = 72 : Mém .20. * Indices : 29 : N .765,773,78.
    in relation to conics : N .729.
    ba?y2+ cyt+ dx3y + exy = fz2 :
    ax++C .883. of functions, calculus of : JP. 15 .
    * Induction : 233 : C .39 : G .15 : L .48 .
    5 + 5 = az : L.43. * Inequalities : 330 — 41 : A .1 ,24 .
    7 + y = z , impossible : 0.823: in integrals : Mél.3 : f.d . c.Mém .59.
    L .404. (I n )? > n " : N .60 .
    5 x
    890 INDEX .

    Inequalities — (continued ) : In-polygons— (continued) :


    a " > : A .14 . two stars, one double the other :
    if x? + yº = z2, 2 + y" : A .20. A .61.
    of a circle and conic (Poncelet): G .1.
    geo. mean of n numbers < arith . | * of a conic : 4822: thsN .47 : cnTN .69.
    mean : 332 : N .42. with sides through given points,
    Infinite : equalities : M .10 : prG .22. on : 4823 : An .5lg.
    functions: An.71 : J.54. semi-regular : N .63.
    from gnomonic projection : Me.66 . of a cubic (Steiner) : M .24 .
    linear point-manifoldness : M .15,17, of a curve : Q .7.
    20,212,23 . of a polygon , th : CD.5 .
    point-mass : M .23. of a quadric with sides through given
    products : J. 27 : N .69. points : LM .22.
    value expressed by r functions : In -quadrics : - - of a developable: Q .10 :
    Ac.3 . quartic, An.59.
    exhibiting circular arcs, logarithms 6 of a quadric , 2 touching 4 : An.69.
    and elliptic functions of the 1st * In -quadrilateral : of a circle: 733 :
    kind : J.73 : Ac.4 . A .5 : cn E .21 : area , N .44 : P .14.
    use of in mathematics : C .73. * of a conic : 4709.
    * Infinitesimal calculus : 1407 : M .11,18. of a cubic : N .84.
    Infinitesimal geometry : An.59 : C .82 . In -sphere of a tetrahedron : A .61.
    of a surface, formulæ : G .13. In -spherical quadrilateral : N .49.
    Infinity, points at on alg . surfaces : C . In -square : of a circle : J.32 .
    59. of a quadrilateral : A .6 .
    * Inflexional tangents : 5789 : A .35 . In-triangles : - of a circle : P .71.
    of a cubic curve : E .30 : J . 38 ,58 . with sides through given points :
    Inflexion curves : Z .10 . J.45 : N .44.
    * Inflexion points : 5175 : CM .4 : J.41. of a conic : J.7 : Maccullagh 's th, N .65.
    of cubic curves : J.28 : axis, E .31. with given centroid : G .23.
    Herse's equation : N .81. similar : A . 9.
    Inscribed figures : of a triangle : thsQ.21.
    In-circle: of a quadrilateral,locus of
    centre : A .52 .
    two (Steiner's “ Gegenpunkte "):
    J.62.
    * of a triangle : 709, 953, 1747 – 50 : tg.e In - and circum -circles : - of a poly
    4889 : CM .1. gon : N .45.
    * in -centre : 709, t.c4629, tg .e 1882. distance of centres : A .32.
    In-conic : - of a circle : thJ.91. of a quadrilateral: Fuss's prMél.3.
    four of a conic : prJ.39. * of a triangle : 935 : A .38.
    of a developable quartic : An.59. * distance of centres : 936 , 4972 : eq ,
    of a polygon : M .25. 4644 .
    of a quadric : J.41. In-andN circum -conics: of a pentagon :
    of a quadrilateral : tg.e4907 : N .63 : .782.
    four, N .56 . ofa polygon : J.64,70 : regular, Z.14.
    of a triangle : 4739 — 46 , tg .e4887 : of a quadrilateral: 60 theorems, N .
    A .2 : N .50 ; max, A .8 : Q .2 : lat. 452: N .76 .
    rect., E .34. ' of a self-conjugate triangle : Me.81.
    In -cubic of a pencil of six lines : Q .9. | * of a triangle : 4724, 4739 : An.52 :
    In -hexagon : - -- of a circle : A .22 . G .222.
    * of a conic : 4781 : N .57,82 . In - and circum -heptagons of a conic :
    In -parabola of a triangle : CD.7. A .3.
    In -pentahedron of a cubic : Ac.5 : M .5. In- and circum -pentagon : of a circle:
    * In -polygons:- - of a circle : 746 : CM .1: A .22 ,43 .
    J.35 : N .50. In - and circum - polygons (see also
    regular of 15,30,60,120, & c., sides : “ Regular polygons ") : - of a
    A .62. circle : L .16 : N .80 : P .11 : Q .11 .
    do. 9 and 11 sides : LM . 10. sum of squares of perps., & c. : ths
    do. 17 sides : TI.13. 1099.
    do. four of 30 sides: N .78 . difference of perimeters,ths : N .43z.
    do.5 ,6 and 10 sides, relation : A .40 , of two circles, respectively : C .53 :
    43 ,45,48 . G .21 : L .78 .
    INDEX. 891

    In - and circum -polygons - (continued ) : I Integrals or Integration - (continued) :


    of a conic : A .4 : ellipse An.52 : An. from orthogonal surfaces' theory :
    57 : J.64 : N .57,84. L . 38 .
    of two conics : C .90 . by Pfaff's method : A .47.
    of a curve : C .78. by series : Me.71.
    of a homonymous polygon : A .50 . by substitution : A .18.
    In - and circum -quadrics of a tetra by Tchebychef's method : L .74 : M .5 .
    hedron : eqsN .65 . comparison of transcendents : Me.79 :
    In - and circum -quadrilaterals : - of a Pr.8 .
    circle : A .48. complex , representing products and
    of a conic : 4709 : and pentagon ths, powers of a definite integral : J.
    N .48. 48 .
    In- and circum -spheres : - of a tetra connected with trinomial integrals :
    hedron : N .73. L .55.
    of a regular polygonron : A.32. convergency of : M .13.
    * of a regular polyhed : 910. definite : - applied to Euler's, & c. :
    circum -triangles
    In- and(Castillon's : - of a circle J. 16 .
    pr) : Q .3. with finite differences : J.12 .
    equilateral, of another triangle : Me. from indefinite : J.41,51,52 .
    74 . whose derivatives involve explicit
    and square of an ellipse : A .30 . functions of the same variable :
    * of two conics : 4970 : N .80. 0 . 12.
    23

    envelope of base : 4997. determination of functions under the


    respectively of two conics having a
    common pole and axis : CD.4 . sign S: JP.15.
    * Instantaneous centre : 5243. * difference between a sum and an
    * Interest : 296 - 301 : and insurance, A . integral: 2230 : G .9.
    26 . division into others of smaller inter
    Integrability of functions: An.50,73 : vals : A .4 .
    0 .28 : J.59,79 : JP .17 : L .492. eight rules for definite integration :
    criterion for max. and min . values of 2245 .
    a primitive : An.52. equations for obtaining functions as
    * Integral calculus : 1900 — 2997 : A .ext integrals : J.79.
    18 ,26 : Euler's, A .20 : C .14,42: expressible only by logarithms: An.
    Newton , CD.8 : G .19 : L .47 : Me. 76 .
    72,74 ,75 : Mém .18,36 . extended independently of the con
    paradoxes : C .44 . ception of differentials : A .61.
    theorems, & c. : 2700 — 42 : A .45 : 0 . formula of: A .1 : J.18,19 : M .4 : Me.
    13 : L .geo .ap50,56 : Me.77 : Mém . 76 : Mo.85 : N .85 : failure of f,
    prs15 ,30 . CM .2 .
    Integral functions : C.88,89,98 : G .4,22,: and gamma-function : LM .12 : Z .9 .12 .
    h .c.f of G .2. higher, of composite functions: A .20 .
    with binomial divisors : J.70. with imaginary limits : C .23 : J.37 .
    and continued fractions : An.77 . use of imaginaries in : M .14.
    reciprocal relation of : A .67. inverse method : CM .4 : CP.4,5 : L .78 .
    * Integrals or Integration : 1908 : A .1,2, involving elliptic functions : Pr.29 :
    4,5 ,6 ,10,23 : Ac.1,32,44 : C .90 : Q .19.
    CD.9 : CP.3 : J.2,4,8,17,25 ,36 ,39 , limits of : 2233 – 44 : L .74 .
    61,92 : JP.9,10,11 : L .39 : M .6 ,16 ,multiplication of : Pr.23.
    73,75 : Mém .31: P.14,36 ,37 : Pr. number of linear independent of 1st
    7,39: Q .11,13 : 7 .7,11,15,18.,22 ,23. kind : An.82.
    * approximation to : 2127, 2262, 2991 : of alg . differentials by means of
    A .9 ,14 : C .97 : CM .2 : G . 3 : J .1, logarithms: Mo.57 : An.75 : C .
    16 , 18,37,48 : L .80 . 902 : J.12,24,78 ,79 : Mo.84 : N .81
    * Gauss's : 2997 : C .84 : M .25 . (see “ Integrals " ).
    by the principle of Abel's derivative : I of algebraic surfaces : C .993 : J.26 :
    J.23. · octic, An.52 : cubature, C .80 .
    by differentiating under the sign of * of circular functions : 1938 - 97 : 24:33
    integration : 2258. - 2522 : No.1799 : LM .4 : M .6 :
    by elliptic functions : G .11 : L.46. Mém . 9.
    892 INDEX .

    Integrals or Integration — (continued ) : Integrals or Integration - (continued ) :


    sine and cosine : G .6 : M .11. * transformation of : 2245 – 52 : A .10 :
    * of exponentialand logarithmic func CM .4 : J.f15,22, 36 : L .36 :Mél.3 :
    tions: 2391 ari
    — 2431 : E .17,18 . Q .1.
    * of circular thmic
    , log and expo . variation of arbitrary constant : 2247 :
    nential functions : 2571 - 2643 J.33.
    (see " Integrals "'). whose values are algebraic : J. 10 :
    of a complex function : A .66 : th of JP .14 : L .38 .
    Cauchy, Ac.84.
    of a closed curve : 5204 : C .23 : E .28 : ALGEBRAIC FUNCTIONS. Indefinite :
    2 . 17. unclassified : An.75 : C .903 : J.12 ,24 ,
    of differentials containing the square 78,79 : M0.84 : N .81.
    root of a cubic or biquadratic : simple functions of x2 + a2: 1926 – 37 :
    Me.57.
    * of discontinuous functions : 2252 : 2", A.4 : v (a?— ~2), A.38: 22, + a ?
    C .23 : LM .6 . N .82:2. 1_ (1dem- x2); An.68.
    of dynamics : L .52,55,58 .
    of explicit functions,determination of fractions involving a binomial surd :
    algebraic part of result : An.61. 2008 — 19.
    of functions which become infinite
    between the limits : 2240 : J.20 :
    JP .11 : Q .6. (1 + x) V 1 (222_ 1): Mém .13.
    of infinite relations : M .14.
    of irratio nal alg . curves by loga
    rithms : An.61.
    (x” — a) 7 (x* — b): Q.18.
    of irrational functions : 2110 – 20 : (1 + ) . 2015 : L .80 : Z .8.
    AJ.2 : An.56 : C .322,89 : L.53,64 : 1
    Mém .30 . (1 + x2) 2 + 1) : Mém .10 .
    limits of : 1903, – 6, 2233, 2775. (1 + x )(1 +
    * **

    for quadrature of curves : 5205 – 11 :


    0 .68,70 : circle, J.21,23 : JP .27. (23 + 8) v (~3 — 1): A.S.
    triple integrals : J.59 .
    * of rational functions : 2021 - 32, 2071 and deductions : 2021 - 8.
    - 2103 : L .27 : N .73.
    of rational fractions : 1915 : CD .3 : 20 " (a + b«")P: 2035 — 60 : A .36 : Mém .11.
    Mém .33 ,: N .72.
    of total differentials : C .99g. 1 * 22 + 02: 2061– 5 : J. 36 .
    of transcendental functions : JP .26 .
    of two-membered complete differen - , * 1
    tials : J.54. (x - a )'(x - 3) : A .40; 17: 2007.
    period s of: C .36,38 ,752: G .753: JP.274.
    principal values of infinite definite :
    functions of (a + bx + cx ?): 2071 – 80 :
    2103
    10 – 9 .
    2240 : A .68.
    properties by elliptic coordinates : Vlatbaterija : A .55 .
    L .51. | * functions of (a + bx + cat): 2081 - 5.
    quotient of two di of the form functions of (a + bx * + 6x2 ) : 2086
    Sdxdy...dz: J.67. 2102 .
    reduction to elliptic functions : An. 1 rational algebraic functions of irra .
    60 : LM .12 : Me.77 ,78 . tional ones :
    residues of : JP.27. integrated
    20.
    by rationalizing : 2110
    Riemann's of first kind : An.79 . reducible to elliptic integrals, viz. :
    singular values of: A .11.
    successive : 2148 : L.62 : 2nd order, rational functions of ✓X4,VXz,and
    M .20 and Z .11. vXG:
    summation of : 2250 : J.47 : JP.12.21. where X , is a quartic in : 2121
    *

    tables of definite, by B . de Haan,note - 41.


    on : C .47. F (*,VX): 2121 : LM .8.: L.57.
    *

    and Taylor's theorem : Me.84 . F ( ): 2133– 6 : J.36 : LM .14 .


    *

    theorems: L .48 : P.55 : Q .10,12. vX '


    INDEX. E93

    Integrals :— ALGEBRAIC — (continued): | Integrals :- ALGEBRAIC — (continued):


    ata : 1.64 : Mél.3. * x **{(ax ) } (Cauchy): 2712:
    A .9.
    vo : C.59: CD.1: E.36 : J.10. On

    . E .41.
    1 + 2x cos $ +1 2:: AA .12:
    F (af )V
    (x )X ,:: C.5
    0 .511.
    dedu ctions from this involving in
    : 2141: J.17: Man.79. te grals of the forms
    011 - " man
    ad fs sin ma
    F ( x3): L.57. J. 1 du J. - n : A . 35.
    (1– ) : Me.82 . So sin at
    Other limits :
    1
    1,31 &c., reduced to * F (a)da p F (a)dæ : 2368– 9.
    Jacobi's functions : Q . 18 . J -2V (1 — ?)' J (ce — c)" .
    f (x )
    V (1 — 48) : J .32. MV (a– ba)*qı -1dæ: A.35.
    sundry : CM .2 : L.47. sina d a = f (a ): geoN .85.
    Limits 0 to 1 : J-, 1 — 2x cos a tū2
    Euler 's : see “ Euler 's integrals.”
    :

    * 7?- ? (1 — ) - : 2280 : A .40 : C .16 : CIRCULAR FUNCTIONS. Indefinite:


    * 2
    J.11,173.
    -1 + m - 1
    * sin , sin -le, & c.: 1938 49: sin* , & c.,
    (1 + ) m : 2341 : C.55. 1954 – 7 : cos” x , N .74.
    * similar forms: 234244, 2352, 2356 a + b cos -a :' 1951: J.9 ; (a- + b cose)
    - 67.
    sa-l næna - 1 1958 : Me.80.
    1- : 2367 : A .10 . * products and quotients of sine and
    1- cosine and their powers : 1959
    qem ( 1 — æ )" . L .59. 80, 2066 — 70 : A .49.
    (1 tax ) binom . funcs. of sin and cos: 1982 – 92.
    { 1 + 1QP(1(1+ am
    — 2e)12
    )? p +29 +2 : L .57. ditto of tangent : 1983, 1991.
    + } ( 1 - )" }
    F (sin x, cosa ) : 1994 — 7: A.12 :
    a cos a + b sin ætc' -
    +1 : L .56 — 7.
    (a + be + cæ2)mimri J.19,32.
    1
    (a − b )" (1 - « )l-" * * * (a + 26 cos a + c cos 2« )– : 1993 ;
    20
    Limits 0 to : 2:21 — 2x ” cos n0 + 1'
    F (cos x ) - : 1996.
    13 ,& c.: 2309 — 12, 2345 — 55 : A . 38 : | * (a,+ ,cos x)(az + b,cosx)...& c.
    J.24 : L .41 : Q .12 : 2 .19. a function of sine or cosine in a ra
    1 pidly converging series, and suc
    ( 2 + a2)» : 2364 . cessive integration of it : J.4,15.
    sin "
    Xnx -r : Me.83. (1 - k sin ) V (1 - K2cos ) .
    2a - 1 An, 2n > a > 0 : N .48.
    1 + " + x2n° 21 (sin amx )2n : J.81.
    2a -1
    m ?(1 - 1), & c .: A .16. ✓ { 1 — 12sin ? (a + x ) sin ? } (a — x )} :
    1 + 2 + ... + (x - 1) J .39.
    24- (1 + 2), and 12 similar : A .35. sin " « : A . 17 : with m = 1, G .7 .
    + 2
    894 INDEX .

    Integrals : - CIRCULAR. Indef.- (cont.): | Integrals : - CIRCULAR. O to 00 - (cont.):


    * a sin ka_æ ?
    or sin (x) (Fourier): 2726 — 42: * 2 +m sin bæ : 2579 — 81.
    A .382. cos
    mtabx,also (* cos & dæ: A.10.
    Limits 0 to :
    * sin"a : 2453 – 5, 2458, 2472 : E.29: | * sin " x : 2510 : A.30 : E .26 : Z .5 : n = 2
    n = , E .28. and 3, 2511 — 2.
    tanam -1æ : 2457. * cos qa - cospæ : 2513 – 5 .
    * sin " x cosPx : 2459 — 65 . a or x2
    cosP ( sin
    cos
    ma : 281, 2484 – 92 : L.43. 1 * sin ax sin bx' and similar : 251422.
    cos" - læ sin ux — I and similar : A .59. ' * cos 2ax sinºa 2722 – 5 .
    sina -2 2
    & cos” -2 x sin nx : 2494 . ° (sin ax , cos bæ ), reduction of: J.15.
    m ? and similar : 2496
    a ? cos?x + b2 sinºa ° (sin ax, cos bæ). J.23,25.
    — 2501, 2344 .
    sin re : 2572 : L.49 : Z .7,8.
    COSP-n-læ sin"-1see cosi g pæ ; and
    sinP-2x sin pæ : 2585 – 8 . cosreg: 2573 : A.10,11,59: J.33 : Me.
    * 2 cot ax : A .34 : No.19 ; 72,76 : 2.7,8. -
    c tan m (1 − b ) : 5340. æ sin re: 2575 : A.10 : J.33.
    Limits ( to a : a² + 2
    sin "æ cos” æ : 2459. cos+ ra2) :. .40 : Q .18 .
    (a2
    *

    sin na sm pæ : 2467 — 9.
    Ø (n\" ) sicosn ,
    *

    | sin sin10 nx : 2474 – 82, 2493. - : A .11 : with (se) = 2* ,


    \ cos ) " cos "
    *

    sin ?"x dnt f(cosa ): 2495. L.46 : • (a ) = tan - ca , A. 11.


    *

    cos n (x — a sin æ): L .41 : 0 to 21 , C .39. sin


    cos (am )2
    cos (a + nx) cos (6 + pæ) cos (c + qx ) : C . į and similar: Z .13.
    54 .
    se sin a sin ax
    1 + a2 cos2x-: Pr.25 : with a = 1, 2506 * related integrals :
    (1 + x2)sin be and
    and Q .11. Ac.7.
    ksin x or sin ac sin rx or cos ra 9292 . 1 - cos" 2 : 2.7,8.
    1 - 2a cos a ta? 1 or a sin cx
    sin2næ or cos px : 0 .11 : LM .11 : n = 1,
    (1 — 2a cos x + a2)" ( 1 + x ) (1 - 2a cos cx taon : 2630 - 2.
    L .74 . tan -lax . 2502. 1 ton -t a ton -line
    & F (sin x cos a ) : JP .27 .
    about 250 integrals with limits chiefly 2504 .
    from 0 to , some from 0 to 00 : ba : Osor
    tan -lax - tan - ba
    2505 .
    Pr.252,26 ,27,29,30,31,329,33.
    Limits 0 to 00 :
    Other limits :
    2
    cosa
    $ , cosxº, cos (ax )2: 2507 — 9 : Q . Jo a + b cos x + c sin -a ,' lim . 0 to 2 :
    A .55 .
    12 : 2602. * tan -?x
    sin 12 n
    13 : sios Jo 1 + *** dæ, deduced from (2416 ):
    COS cos 2" " : Me.72. L .69.
    INDEX . 895

    Integrals : - CIRCULAR — (continued ) : 1 Integrals : — LOGARITIMIC . Indef.


    pitan - ? ax dx . 2502 . (continued) :
    Jo æ (1 - x2) " log , to
    [ F ( ) da,where F (x) is a rational (1 — «")ps type
    by of several: Mém . 18 .
    J - se peata
    integral circular function : CM .3 . Limits 0 to 1 :
    Functions."* * log 1: 2284;x*(log 1)", 2291.
    sine-integral, & c.: see “ Functions."
    EXPONENTIAL FUNCTIONS. Indefinite : * involving log x or log (1 + x ): 2391 –
    et, a* : 1924 ; ama+nt, 2004. 2403 , 2416 – 22.
    et { Q (x) + $ '(x )} : 1998. log (1 + x ). 2416 : C.59 : J .6 : L .43,44.
    1+ 2 .
    e exp (1x2): E .34.
    X being a rational integ . funct. of a : | * logæ 08 2. xm 1-Iloga
    taon ?
    & X : J .13 ; ex , Mém .83 ; e exp V (1 — ~ 2)
    ( - 22), Q .1 . and many cases, J.34.
    e exp ( - ) or the same x B2 or 1 log 11 + 2 2403 : L .73.
    ✓ (a + bx2) zemqem ((10 —- 1)" .: 23
    Z.3 ;. 2 " — 1, A.37.
    + (c + d« ) : L .52.
    log æ log x'
    Limits 0 to 00 : about 540 expressions chiefly formed
    * e - kt q 2284 — 91: see “ Gamma from log (1 + « ) or log (1 + x )
    function ." or log (1 + x + x2) or log (1 + x? + a ),
    * e exp (- k&2) : 2425 : evaluation by a joined to a single factor of the
    continued fraction, J.12. form 2m or ( 1 + 2 ) or 1 F 20m
    It 2 "

    * other forms: 2426 — 31, 2595,'8, 2601.


    with integral values of m and n :
    eexp (–22_ a*): 2604 —5: L.56. A .39,40.
    12 : Z.6 ; ditto x æm, L.46 . about 280 expressions, nearly all
    comprised in the form
    e -nos
    en and the same x x : A . 10 . qem (1 — « ") ? (log x)',
    with integral values of m , n , p, q,
    e - ar cé(x +br): Mé1.2 . and t : A .40 .
    e exp (ax"). Ei (+ ax"): Q.18. about 130 expressions of the forms,
    Other limits : " (1 —2 ") (log æ)2or3,
    expon-integral: see “ Functions." mm (1 — æn)& log x, and
    e exp ( -x2) [ = Erfc & (Glaisher )]: ac" (10" ) log x, with integral
    we" - 1
    de Me. 76 . values of m , n , p , q, and r : A .40.
    penkt F(x)dx : C.777 Limits 0 to 00 : 2423 — 4 .
    Other limits :
    *<**eexp(-2-a ):C.12. (1 — * ) : 2408 – 12 : L .73 : Z1.
    10

    F (remi) e - noi: A . 15 .
    Llog + ®,limits0 to 12 – 1: 2415.
    LOGARITHMIC FUNCTIONS. Indefinite : log-integral : see “ Functions."
    log æ : 1950 ; ac (log )+ ", 2003 — 6. CIRCULAR -EXPONENTIAL FUNCTIONS. In
    am log (1 + x), & c.: A.39. definite :
    F (a ) log « : G .12. enx ( sin
    sm )))" bac : 1999 ; et sin " x cos” « ,
    ac"(log æ)"- and a similarform : 2030 — 2 . (cos
    2000 .
    896 INDEX .

    Integrals : - CIRCULAR -EXPONENTIAL Integrals : - CIRCULAR LOGARITHMIC


    (continued ) : (continued) :
    (eit sin x )" - 1: L .74. log son « and log V (1 — k sin ?x), each
    Limits, 0 to 00 : with theabovedenominator : J.92.
    * e -ar sin ra : 2571, 2591 . Limits 0 to :
    log (1 – 2a cos x + a2) : 2620 — 2 : L .38 :
    * e-ar sin bæ : 2583 : 0 to 1,Mém .30 . Q .11.
    * cos rx13log (1 – 2a cos x + az) : 2625 : A.
    *

    e-arm cosins bx : 2577,2589 : J.33 : 2.7. .


    *

    a log sin æ : 2637 ; « log sin ?2, 2638 .


    * 2-42 (sin )"2: 2608 – 11: A.7: Limits 0 to co :
    *

    cos ad log 2cæ : A .11.


    * with limits – to ., 2612. log ( 1 — 2 a cos cx ta ?) . 26 :31.
    par te -ar
    Ce sin mx, & c .: 2593 — 2600. 1 + x2
    *

    Other limits :
    * e-kä сosmx sin "x : 2717 — 20. paa + b cos x dx : A .53.
    *

    * e exp( - a²c2) Cos, 26x, and similar : Joa — b cos


    *

    2614 – 8 : Z.1,10 . (**F (cosnx) log sinde: Z.10.


    * eexp {- (x2+ x2)cos 8} * &c.: cos:(n- 1)x log tan x dx : A. 16 .
    2606 : Q.12.
    Cos X - e - ar 20109
    - : 261 . log (1 - 2 a cos x + x2) dx, and
    similar: J.403.
    CIRCULAR - LOGARITHMIC FUNCTIONS : EXPONENTIAL-LOGARITHMIC FUNCTIONS :
    @ ??(log x )"
    221 — 2a " cosno + 1 ' * Jo e *t - e
    log (x2 + u )dx, & c. : J.38 .
    Limits 0 to 1 : |***ect- log (1 – 2a cosa + a )dx: J. 40 .
    Jo
    COS
    (m log æ ) / log x, and similar : 2641 MISCELLANEOUS THEOREMS :
    — 3. formulæ of Frullani, Poisson , Abel,
    log (1 — 2a cos 0 + x2) & c. & c . A.34. Kummer, Cauchy, & c. : 2700 — 13 .
    | f'(x) = f (b) - f(a): 1901 — 3 : A.16.
    log 1 –æ2e(1cos-x2)9 +13,& 13 similar: L.732. Y © (ax)— ®(bæ)dx :Me.81:n = 0,2700.
    log sinæ log cosa, & c. (lim 5° (x) {x (a)} +" då : 2001 – 2.
    E.22 . p*f(88)de(6 of Taylor's th): LM.13.
    Limits 0 to
    * log sin æ : 2635 : CM .2 : E.23 : Q.12 : | F(x)dx,approx.to: C.97.
    O to nt and similar integrals, A .
    16 .
    55(a — R)dx = Sf (y)dy: CD.4.
    log (1 + ccosa ): 2633 : 0 to 17, 2634. $ .$(x + 1),f(")de, and similar:
    cOS a Me.75,76 ,77 .
    tan x log cosec & : E .27.
    log (1 +12 sinºa ). 1.46 .
    55(4)º()de:L.49
    ✓ (1 - k sin ?x ) Sude ude('vde:AJ.7.
    INDEX. 897

    Integrals : — MISCELLANEOUS TII. - | Interpolation — (continued) :


    (continued ) : by method of least squares : C .373:
    Mém .59 .
    ( (u)ndu : 2.25. by a parabolic curve : 2992 : C .37.
    Stirling 's series : Me.68.
    (u + k)x+2du = (- 1){ [ (2)x+2du: | and summation : 1.11.,12, 14 .
    tables : 1.11z.
    A.38. of values from observation : Mél.2 :
    Pm (a ) Pn(x ) dx (P x = Legendre's tr, Mém .59.
    coeff.): Pr.23. Intersection : - of circles and spheres :
    ſ du L .38 .
    uU4( = log 4 (u ) + C : J.19. * of 2 conics : 4916 : CD .5 ,6 : N .66 .
    of 2 curves : 4116 ,4133 : CM .3 : J .15 :
    (x *'+ (xp) +(a p + +9)"q "do+ 1° 0 = pos integer : L .54 : by rt. lines, Me.80.
    J 37 * of 2 planes : 5528.
    C .78 . of 2 quadrics : C .62 : N .68g.
    of right line and conic : see “ Right
    line.”
    ° (sin 2x)cosa da of successive loci, ths: N .42.
    of surfaces : J.15 : L .54 : by rt. lines ,
    = JO1 + (cos2a) cos a da : A.21: | Me.80 .
    * Invariable line, plane,conic,and quadric :
    L.532. 5856 - 66 .
    $ $(«) sin wråd (log «): M0.85. * Invariants : 1628 : An.542: C .853 : E .42 :
    ta ? $ csoins a y scoins22 & , transf. of : G .1,2 ,15 : J .62,68 ,69 : L .55 ,61,76 :
    M .3,5,17,19 : Me.81 : N .58,59 ,69,
    pe 70 : P .82 : Pr.7 : Q .12 : Z .22."
    J.36. of binary cubics : An .65 .
    integrals deduced from of binary forms: of 8th deg., C .84 :
    "' + Pz' + Qz = 0, and y " + G .2 : M .5 : simultaneous, M .1.
    of higher transformations : J.71.
    ( P + 2R) y' + { Q + R (P + R ) + R '} superior limit to number of irre
    y = 0 : J.18. ducible : C .86.
    J1 (æ ,y) dx : J.61 :M0.61 : Q .7. of a binary quadric : M .3.
    of a binary quantic : 1648 : E .40 :
    * if f(a + iy ) = P + iQ , th : 2710. Me.79 : of two, 1650.
    (Upwards of 8,000 definite integrals have of a binary quartic : M .3 : Q. 10 .
    been collected and arranged in a 4to of a binary quintic, table of irreduci
    volume by D . Bierens de Haan ; ble : AJ.1.
    Leyden, 1867 (B .M .C.: 8532. ff.)] of a bi-ternary quadric : J.57.
    * Integrating factors of d.e : pp.468 — 471, of a conic : 4417 : 4936 – 5030 .
    3394 . of two conics : 4936 : N .75.
    * Integrator mechanical: 5450. of three conics: Q . 10 .
    Intercalation : CM .3 . of particular conics : 4945.
    * Intercepts, to find : 4115. ofa correspondence : G .20.
    * Interest : 296 : N .48,61,64. differential: M .24 : of given order
    Interpolating functions : C .11. and degree belonging to a binary
    * Interpolation : 3762 : A .32,61,62,70 : 10 -ic, č .89.
    AJ. 2 : C .19,48 ,68,92 : J.5 : L .37, of d.e linear : 0 .88 : of 4th order , Ac.
    46 : Me.78 : N .59,76 : Q .7,8 . 3.
    ofalgebraic functions, Abel's th : J.28 . and covariants of f (x2, y2) relative to
    Cauchy's method : A .2 : C .373: L .53. linear transformation : 4 . 17.
    by circular functions : N .85. of linear transformations : M .20 .
    by cubic and quintic equations : C .25 . mutualrelation ofderived in variants :
    formulæ : J.2 : C.992 : Mo.65 . J.85 .
    Lagrange's : 3768 : J.1,84 : N .57,61. of an orthogonal transformation : J.
    Newton's : N .57 ,61,71. 65 : LM .13.
    for odd and even functions : C.99. of a pair of homog. functions : Q .1.
    and mechanical quadrature : 3772 : partial : LM .2.
    A .20. of points,lines,and surfaces : Q.4.
    5 Y
    898 INDEX.

    Invariants - (continued ) : Irrational functions : M .4 : of the 2nd


    of a quadric : J.86 : oftwo, M .24 : Q .6 . degree, C .98 .
    of a quintic : of 12th order, Q .1 : of Irreducible functions with respect to
    18th order, C .92 : J.59. a prime modulus: C.70,90,93: L .
    related to linear equations : C .94 . 732.
    of sixth order : G .19 . Isobaric : — calculus, N .85 : homog.
    skew , of binary quintics, sextics, and functions, G .22 : algorithm , N .84.
    nonics, relations : AJ.1. Isogonal relations: A .60 : Z .18, 20 .
    of ternary forms : G .192. do. represented by a fractional func
    transformation theorem : M .8 : Me.85 . tion of the 2nd degree : M .18.
    Inverse calculus of differences : N .51. representation of x = " - X and
    * Inverse equation of a curve or inverse
    method of tangents : 5160 : No.
    jaX" + b . 2.26 .
    1780 : J.26 : Mém .9, 26 . VcX " + d
    * Inversion : 1000 , 5212 : An.59 : C .94,pr transformation of plane figures : N .69.
    90 : LM .5 : Me.66 : ths N .61 : Pr. Isomerism , pr : AJ.i.
    34: problems by Jacobi, J.89 : Isoperimeters, method of: N .47,74 ,82.
    geo. ths, Q .7 . problems : J. 18 : M . 13 : Mél. 5 .
    formulæ : An.53 : Lagrange 's, J.42,54. triangle with one side constant, and
    of arithmetical identities : G .23 . a vertex at a fixed point : 0 .84.
    * of a curve: 5212 : G .4 : J.14 : Pr.14. Isoptic loci : Pr.37.
    angle between radius and tangent: Isosceles figures : C .87 : JP .30 .
    5212,5219 : E .30. Isotherms, families of : 2 .26 .
    of 2 non-intersecting circles into con Isotropic functions : C .26.,,27 .
    centric circles : E .39 . Iterative functions : L .84.
    of a plane curve : 5212 : G .4 : Pr.14 . * Jacobian : ths 1600 - 9 : AJ.3 : thZ .10 .
    of a quadric : J.52,76 : Q .11. * of 3 conics : 5023 : LM .4 .
    of a system of functions : An.71. formulæ : d.c.1471 : J.842: Mo.84.
    and stereographic projection : E .35. function : one argument, G .2.
    * Involute : 5149, - 53 , - 66. of several variables : Mo.82 .
    * of a circle : 5306 : C .26 : successive, modular eq. of 8th degree : M .15 .
    E .34. and polar opposites : Me.64.
    and evolute in space : CD.6 . sextic equation : Q .18.
    integrals of oblique : C .85. system , multiplicator : M .12.
    of a tortuous curve : 5753. Jacobi-Bernoulli function : J.42.
    * Involution : 1066 : A .55,63: gzAn.59 :
    At.63 : CD .2 : thCP.11 : thE.33 : Kinematics : A .61: AJ.3: G .23 : L.
    G . 10 ,20 : J.63 : N .53,64,65. 53,80 : LM .th17 : N .82, ths 83
    and application to conics : A .4 ,5. and 84 .
    of a circle : Me.66. of plane curves : A .55 : N .82 : caus
    of a cubic space and the resulting tics, Z .23.
    complex : 2 .24 . paradox of Sylvester : Me.78 .
    of higher degrees : C.993: JM .72 : of of plane figures : Mél.26 : N .78,80. :
    3rd and 4th , An.84 : Z .19. of a triangle, N .53. tric
    of numbers, machine for : P .15 . of a point : N .49,82 : barycen
    of n -tic curves : C .87. method , Mél. 5.
    relation between a curve and an n -tic, of sliding and rolling solids : TA.2.
    the latter having a multiple point Kinematic geometry : of space : J.
    of the n - 1th order : C .96 . 90 .
    pencils with problems in conics and of similar plane figures : Z .19.,20,23.
    cubics : N .85 . Knight's move at chess : 0 .31,52,74 :
    of points on a conic : N .82. CD.7 : CM .3 : E .41: N .54 : Q .149.
    of 6 lines in space : C .52. Knots : TE.28 : with 8 crossings, E .33.
    of right lines considered as axes of *Kummer's equation , i.c. : 2706 .
    rotation : C .52. rational integrals of : M .24.
    * systems of points in : 4826, 4828. an analogous eq . : C .99 .
    ditto, marked on a surface : C .99. Kummer's 16 -nodal quartic surface :
    Irrational fractions : decomposition of, C.92 : J.84 .
    J.19 : irreducibility of, Mém .41 : figures of : M .18.
    rationalization of, A .18 . lines of curv. of : J.98 .
    INDEX. 899

    Lamé's equation : An.79 : C .86 ,90,91, | * Limaçon : 5327 : C .98 : N .81.


    ths92. Limited derivation and ap . thereof :
    Lamé's functions : C .87 : J.56,60,62: Z .12 .
    M .18. Limiting coefficients : C .37.
    * Lagrange's theorem (d .c ) : 1552 : Limits : theory of, Me.68 .
    C .60,77 : CD .6 : CM .3 : Mél.2 : of functions oftwo variables : M .11.
    gzC . 96 and Me.85 : gzQ .2 .
    * Lambert's th . of elliptic sector : 6114 :
    of a parabolic sector, A .16 ,33,48 :
    of (1+ 1 )* when x = 00 : L.40:
    Me.78 : Q .15 . 1.85 : Q.5.
    * Landen 's th , of hyperbolic arc: 6117 : Life 59.
    annuities : A.42: cn of tables, P .
    E .21. Linear: - associative algebra : AJ.4 .
    Laplace's coefficients or functions : see
    “ Spherical harmonics." construction : Man .51.
    Laplace's equation : and its analogues , coordinates in space : M .1.
    CD .7 : and quaternions, Q .1. dependency of a function of one vari.
    Laplace's th . (d .c) : 1556 . able : J .55.
    Last multiplier, Jacobi's th . of : L .45 . dependent point systems: J.88 .
    Lateral curves : A .58 . forms : L .84 : with integral coeffi .
    * Latus rectum : 1160. cients , J .86 ,88.
    *Law of reciprocity : N .72 : d .e,3446 . function of n variables : G .14.
    ext. to numbers not prime: 0 .90 . U ? = V ? where U , V are products of n
    *Least divisors, table of,from 1 to 99000 : linear functions of two variables :
    page 7 . CD .5.
    Least remainder (absolute ) of real geometry , th : M . 22 .
    quantities: M0.85. identities between square binary
    Least squares, method of : A .11, 18 ,19 : forms: M .21.
    AJ.1 : C .34 ,37,40 : CP.8,11 : systems, calculus of : JP .25 .
    G .18 : J.26 : L .52,53,67,75 : Linear equations: A .512,70 : Ac.4 : C .
    Me.80,81 : Mél.1,4 : gzZ .18 . 81,7h942 : G .14 : J.30 : JP. 29 : L .f
    Legal algebra (heredity): N .63. 39,66 : M0.842 : N .51,75 ,802 : Z .
    * Legendre's coefficient or function , X .: 152,22.
    2936 : C .47,91 : L .76 , gz79 : analogous to Lamé's : C .98 .
    Me.80 : Pr.27 . with real coefficients : M .6 .
    and complete elliptic integral of 1st similar : N .45.z.
    kind : Me.85 . solution by roots of unity : 0 .25.
    integral
    rth Me.83 of and log integral of: systems of : 582 : A .10,22,52,57 : G .
    15 : C .81,96 : L .58 : N .46..
    product of. any two expressed by a in one unknown : C .90 : G .9.
    series of the same functions: of nth order : J .15 .
    Pr.27. standard solution : 582: Q .19 : gen .
    th , A .12 .
    Legendre's symbol ( ) : Mél.4,5. symbolic solution in connexion with
    the theory of permutations: C . 21.
    Leibnitz 's th. in d .c : 1460 : N .69: a whose number exceeds the number of
    formula ,Mo.68. variables : N .46 .
    * Lemniscate : 5317 : A .55 .cn3 : At.51 ; Lines : - alg . representation of : C .76 .
    ths E .4 : L .46 ,47 : Me.68 : N .45 . “ de faîts et de thalweg " in topo
    chord of contact, cn : Z . 12 . graphy : L .77 .
    division of perimeter : 0 . 17 : L .43 : generated by a moving plane figure :
    into 17 parts, J.75 : irreducibility C .86 .
    & c. of the partition equation , of greatest slope : A .29 : and with
    J. 394. vertical osculating planes, C .73.
    tangents of: J.14 : cnZ.12. loxodromic : J.11.
    Lemniscatic geometry : Z .212: coordi. six coordinates of : CP.11.
    nates, Z .12 ; of nth order, J.83. * Lines of curvature : 5773 : A .34,37 :
    Lemniscatic function : biquadratic An.53 : C .74;: CD.) : L .46 : M .2,
    theorem , multiplication and 32,76 : N .79 : Q .5 .
    transformation of formula , J.30. of alg . surfaces : 2 .24.
    Lexell's problem : LM .2. * and conics,analogy : 5854 : Me.62.
    900 INDEX.

    Lines of curvature- (continued): Linkage and linkwork — (continued ):


    dividing a surface into squares : C . * the Proportionator : 5423.

    *
    74 : LM .4 : Mo.83. the Quadruplane : 5422.

    * *
    of an equilateral paraboloid : N .84. | * the Reciprocator : 5419.
    of an ellipsoid : .38,48 : An.70 : ths | * the Reversor : 5407.

    * *
    CD.3 : CM .2 ,3,4 : JP .1 : N .81. the Translator : 5407.
    comparison of arcs of,by Abel's th : the Versor-invertor : 5422 .

    *
    An.69. * the Versor-proportionator : 5424 .
    and of its pedal surfaces : Q .12. Lissajons' curves : A .70 : M .8 .
    projection of : 2 .2. * Lituus: 5305 .
    rectification of : An.73. Loci, classification of: C .83,85 : P.78 :
    generation of surfaces by : J.98 : N . Pr.27 .
    77 . Locusof a point: the centre of a circle
    and geodesics of developables : L .59. cutting 3 circles in equalangles :
    near an umbilic : 5822 : A .70 : Q .10. 1 .53 .
    * *

    osculating plane of : 5835 . the centre of collineation between a


    plane, condition for : 5843 : An.68 : quadric surface and a system of
    *

    C .363,422,96 : G .22 : Me.64. spherical surfaces : A .65.


    plane or spherical: An.57 : C .46 : JP. dividing a variable line in a constant
    20 : L .53. ratio : gzAJ.3.
    * of a quadric : 5833 — 4 : C .22.,49 ,51: of intersection of common tangents
    G .11: J.26 : Me.1: N .632: Pr.32 : to a conic and circle : N .63,79.
    TI.14. of intersection of curves: CM .2.
    pd constant along it : 5836 . ditto of two revolving curves : N .64 .
    projected from an umbilic into con on a moving right line : L .49.
    focal Cartesians : E .19 . on a moving curve : Mém .18 .
    quadratic for Yr, giving the direc the product of2 tangents from which ,
    tions : 5810 . to 2 equal circles is constant :
    of two homofocal quadrics : L .45 . An.64.
    of quartie surfaces : C .59 : L .76 . at which 2 given lengths subtend
    of ruled surfaces : C 78 . equal angles : A .68.
    spherical : C .36.,42. . whose sum of distances from 3 lines
    and shortest distance of 2 normals is constant: A . 17 ,46 : from 2
    one of which passes through an lines, N .64 : from lines or planes,
    umbilic : L .55 . A .192,prs and ths31.
    of surface of the 4th class, correlatives whose distances from 2 curves have a
    of cyclides which have the circle constant ratio : An.58 : or satisfy
    double line : C .92.
    infinity for a surfaces
    at tetrahedral a given relation , A .33 symbolic f.
    of the of Lamé, Locus : - of pole of one conic with
    & c. : C .84. respect to another : N .42.
    and triple orthogonal systems: M .3. of remarkable points in a plane tri.
    * Linkage and linkwork : 5400 — 31: E . angle : A .43 .
    28,30 : Me.75 : N .75,78 . of vertex of constant angle touching
    E.34 ;: LM
    3-bar : 5430,4 -piece
    * conjugate 4-bar,
    .9.
    Me.76 . a given curve : N .61.
    of vertex of quadric cone passing
    * for constructing : an ellipse : 5426 . through 6 points : N .63.
    a lemniscate : AJ. 1. * Logarithmic : - curve, 5284 : quadra
    a limaçon : 5427 : Me.76 . ture, N .45.
    *

    æc? and am : AJ.1 and 3 . integral: A.9,,19 : J.17 : Z.6.


    root of a cubic equation : 5429 . numerical determination : A . 11.
    * * * *

    Hart's : 5417 : LM .6 ,82. of a rational differential : J.3.


    Kempe's : 5401 : Pr. 23. parabola : C0 . 7.
    Peaucellier 's : 5410 : E .21 : LM .6 : N . potentials : M .3 ,4,8,13,16 : Z .20.
    82 . rational fractions : A .6 .
    the Fan of Sylvester : E . 33. systems: A .14 .
    the Invertor : 5419. transcendents : P .14 .
    * * *

    the Multiplicator : 5407 . waves: LM .22.


    the Pentograph : 5423. * Logarithms : 142 : P .1792, 1787, 1806,
    * the Plagiograph : 5424 . 17 : TE.26 .
    the Planimeter : 5452. and anti-logarithms, cn : 1.124,24.
    INDEX . 901

    Logarithms- (continued): Malfatti's problem , to inscribe 3 circles


    * calculation of: 688 : A .24,27,42 : LM . in a triangle touching each
    1,5 : Me.74 : N .51 : Pr.312,32 : other : A . 15,16 ,20,55 : J.10,452,76 ,
    TE .6 , 14 : TI.6 ,8. 77,84 ,89 : LM . 7 : M .6 : P .52 : Pr.
    Huyghen 's method : C .662,68, 6 : Q .1 : TE.24 : Z .21.
    and circular functions from definite Malfatti's resolvents of quintic eqs : A .
    integrals : A .65. 45 .
    common or Briggean : A .24 . Malm 's surfaces, th : J.84,88.
    constants in integral : 1.60. Mannheim , two theorems: G .8 .
    of different orders of numbers : L .45 . Martin 's measure of distance : A .19.
    higher theory of: trA .15 . Matrices : E .42 : LM .4,16 : thMe.85 :
    impossible : CM .1.
    natural, or Napierian , or hyperbolic : P.58,66 : Pr.9,14.
    A .25 ,26 , 57. ( cài
    a ,b ) and function f(0) = ax + b.
    of commensurable numbers or of cæ + :
    algebraic irrationals : C .95. Me.80..
    base of: see “ e.” Cayley 's th : LM .16 : Me.85.
    modulus of : 148 : A .3.
    of negative numbers : No.1784 . equation , px = æq : C.992.
    new kind of : J.70 . of 2nd order : linear eq., C .992: quad
    powers of : CM .2. ratic, Q .20 .
    of sum and difference of 2 numbers : of any order : linear eq., C .994.
    A .45 . notation of : J.50.
    with many decimal places : N .67. persymmetrical, th : E .34.
    of 2, 3,5, 7, 10 and e all to 260 decimal product of : G .5,11.
    places : Pr.27 . roots of a unit matrix : C .94.
    of 2, 63,. 20.5, 10 and e to 205 places : Pr. whose terms are linear functions of ze :
    J.50 .
    of primes from 2 to 109 : table viii.,p.6. *Maximum or minimum : 58, 1830 : A .4 ,
    tables of log sines, & c. cn : Q .7. 13,22,35,49,53 ,602,70 : C . 17,24 : J.
    Logic, algebra or calculus of : A .6 : AJ. 48 : JP.25 : Me. 1,geo5,72,76 ,81,
    3,72 : CD .3 : M .12 : Man .71,76 ,823 : 83 : N .43 : Z .13.
    Q .11. problems on : 1835 — 40, 1847 : A .2,
    of equivalent statements : LM .11. geo19,38 ,39 : geoL.42: Mém .11,
    Logic of numbers : AJ.4 . a paradox, N .63.
    Logocyclic curve : Pr.9 : Q .3. arc as a function oftheabscissa :
    of anJ.17.
    Longimetry applied to planimetry : J.52.
    Loto , gam e of : L .42. continuous : 1866.
    Loxodrome : eqA.21 : N .61, : Z .5 . of a definite integral : Z .21.
    of a surface of revolntion : N .74. discontinuity in : CD .3.
    of cylinder and sphere: A .2 . distances between points, lines, and
    of ellipsoid and sphere : A .32. surfaces, geo : At.65.
    of paraboloid of rotation : A . 13. duplication of results : Me.80 .
    Loxodromic triangle upon an oblate ellipse which can pass through 2
    spheroid : A .27 . points and touch 2 right lines :
    Lucas's th : G . 14 : analogous f., G .13. A .14 .
    Ludolphian number : Mo.82. elliptic function method : Mél.5.
    Lunes : J. 21 . of figures in plane and in space : CM .
    *Maclaurin 's th . (d .c) : 1507 : A .12 : 3 : L .41 : J.242 : Z .11.
    CM .32: J.84 : N .70. * functions of one variable : 1830 : ditto.
    symbolic form : CM .4. with an infinity of max. and min .
    Maclaurin -sum -formula : J.12. values : J.63 .
    Magical equation of tangent : Q .6. * functions of 2 variables : 1841 : Mém .
    Magic : - cubes, Q 1.
    .7 : CD . 31 : Q .5,6 : Lagrange's condition ,
    cyclovolute : TA .5 ,9. CM . 2 .
    parallelopiped : A .67. functions of 3 variables : 1852 : CD.1 :
    rectangles : A .65 ,66 . prs 1860 — 5 .
    squares: A .21,57,66 : CD .1 : CM .4 : * functions of n variables : 1862 : L .43:
    E .8 : J.44 : Me.73 : Pr.15,16 : Q . ! Mém .59 : Q .12 : symmetrical,
    6,10,11 : TA.5,9. Mél.2 .
    902 INDEX.

    Maximum or minimum - (continued ): *Measures of length & c. : page 4 .


    of in - and circum - polygon of a circle : exactitude of: 2 .6 : do. with chain ,
    A .29,30 : do. of ellipse,and analo Z .1.
    gous th. for ellipsoid, An.50 . Mechanical calculators : 0 .28 : 1. 16 :
    indeterminates : CM .4. P .85.
    in-polygon (with given sides) of an for “ least squares " : Mél.2 .
    ellipse : A .30 . Mechanical construction of : - curves :
    by interpolation , f : A .25 . M .6 : N .56 .
    method of substitution : A .23 . Cartesian oval : AJ.1.
    of multiple definite integrals : Mél.4 . conics : An.52 : three, N .43.
    planimetrical groups of : A .2 . ellipse: A .65 : Z.1.
    of single integrals between fixed lemniscate : A .3.
    limits : J.54 ,69 : M .25. conformable figures : AJ.2 .
    of the sum of the distances of a point cubic parabola : N .58 .
    from given points, lines, or curves for duplication of roots : A. 48 .
    planes : J.62. (a2 - 22)/ : E .18.
    of the sum of the values of an integral surfaces of 2nd order and class : J.34.
    function and of its derivatives : Mechanical : - division of angles : Q .4.
    L .68 . measurement of angles : A .61.
    solids ofmax. vol. with given surface integrators : 5450 : C .92,94,95 : Pr.244.
    and of min . surface with given for Xdx + Ydy : Me.78.
    vol. : C .63. involution : AJ.4 .
    * ofſF( , y)ds, & c., to find 8 : 3070 — 2. quadrature : 3772: 1 .58,59 : J.6,63
    gzA.66 and C .99.
    * of SS F ( ,y,z) dS, & c.,to find S : 3078 solution of equations: Me.73 : N .67.
    - 80 . linear simultaneous : Pr.28 .
    cubic and biquadratic , graphically :
    Maximum : ellipse touching 4 lines : A .1.
    A .12 . Mensuration of casks : A .20.
    ellipsoid in a tetrahedron : Z .14 . Metamorphic method by reciprocal
    of a factorial function : Me.73. radii : N .54 .
    polyhedron in ellipsoid : A .32 . Metamorphic transformation : N .46 .
    of a product : N .44 . Metrical: - system : E .30 .
    of a sphere, th : N .53. properties of figures, transf. of : N .
    solid of revolution : 3074 . 58 ,59 ,60 : J.4 .
    tetrahedron : - in ellipsoid , A .32 : 1 properties of surfaces : AJ.4.
    whose faces have given areas, C . | *Meunier's theorem : 5809 : gzC .74.
    54,66 : N .623. Minding 's theorem : Quaternion proof:
    volume with a given surface : 3082. LM .10 .
    of sin a : A. 362: of ; x , & c., A .42. Minimum : - - theory of : L.56 : prM .20 .
    angle between two conj. tangents on
    ofax + by + & c.,when æ? + ya + & c. = 1: a positive curved surface : A .69.
    N .46 . area : J.67.
    of circum -polygon : CD.3.
    Mean centre of segments of a line cross of 43.a hexagonal “ alvéole," pr : N .
    ing three others : A .40.
    Mean distance of lines from a point : Z . circum -conic of a quadrilateral : A .
    11. 13 : An.54.
    Mean error of observations: A .25 : C . circum -tetrahedron of an ellipsoid :
    377: TI.22. 2. 25 .
    in trigonometricaland chainmeasure circum -triangle of a conic : 2.28 : of
    an ellipse , Z .25 .
    ments : A .46 : Z .6 . curves on surfaces : J.5 : see “ Geo
    Mean proportionals between two lines : desics.”
    A .31,34 . distance of 2 right lines : G .4 ; of a
    Mean values : C .18,20,23,26,272 : L .67 : point, ths: A .8.
    LM .8 : M .6 ,7 : 2 .3. ellipse through 3 points and ellipsoid
    of a function of one variable : G .16 : through 4 : L .42 .
    of 3 variables, C .29 . ellipsoid, th : Mo.72.
    and probabilities , geo : C.87 : L.79. N . G . F . of a binary septic : AJ.2.
    INDEX. 903
    Minimum : - theory of — (continued) : | * Momental ellipsoid : 5925, 5934 ; for a
    numerical value of a linear function plane, 5936 .
    with integral coefficients of au * Moment of inertia : 5903 : An.63 : At.
    irrational quantity : 0 .53,54 . 43 : M .23.
    perimeter enclosing a given area on a of ellipsoid : 6150 : CD .8 : J. 16 .
    curved surface : J.86 . by geometry of 4 dimensions: Q .16 .
    questions relating to approximation : principalaxes : 5926 , 5967, 5972 : At.
    Mél.2 : Mém .59. 43.
    sum of distances from two points : of a quadrilateral: 5951: Q .11.
    920 – 1. of solid rings of revolution : Q .16 .
    sum of squares of distances of a point * of a tetrahedron : 5957.
    from three right lines : Z.12. * of a triangle : 5944 : Me.4 : Q .6 :
    sum of squares of functions: N .79. polar, N .83.
    Minimum surfaces : - eqA .38 : G .14, * of various laminæ and solids : 6015 –
    22 : J .81,85,87,ext78 : Mo.67 ,72 : 6165 .
    projective, M .14 : metric , M .15 . Monge's theory " des Déblais et des
    algebraic : M .3 : lowest class-number, Remblais ” : LM .14 .
    An .79. Monocyclic systems and related ones :
    not algebraic and containing a succes. J.98 .
    sion of algebraic curves : C .87. Monodromefunctions : C .43 : G .18 .
    arbitrary functionsofthe integral eq. Monogenous functions (Laurent's th ) :
    of : C .40. Ac.42: C .32,43.
    between 2 right lines in space : C .40. Monotypical functions : C.32.
    generation of : L .63. Monothetic equations: C.99.
    representation of by elliptic functions: Mortality : A .39.
    J.99 . “ Mouse-trap " at cards : Q .152.
    of a twisted quadric : At.52. Movements : JP .15.
    limits of (Calc. of Var.) : J.80 : on
    a elliptic and parabolic : JP .30 .
    catenoid, M .2 : determined by groups of : An.692.
    one of the edges of a twisted of a plane figure : thАn.68 : JP .20 ,
    quadrilateral, Mo.65. 28 : LM .3 .
    variation of an invariable system : C .43.
    72. of surface , capacity of : Mo. of a point on an ellipsoid : AJ.1 : J.54.
    Minimum value of relative : JP .19 : of a point, L .63.
    of a right line : C .89 : N .63.
    1 .(A + Ba + Cx2 + & c.)da : N .73. of a solid : JP .21.
    transmission of and the curves result.
    of J V (dx2 + dy? + ...) when the varia ing : JP.3.
    bles are connected by a quadric of “ ahnlich - veränderlicher " and
    equation : J.43. “ affin -veränderlicher " systems:
    Models : LM .39: of ruled surfaces, *Multinomial Z .24 and 19.
    Me.74 . theorem : 137 : Me.62.
    Modular : - equations : An.79 : of 8th Multiple-centres, geo . theory : L .45.
    degree, 59 : 0 .483,493,66 : M .1,2 : | Multiple curves of alg. surfaces : An.73.
    Mo.65 : see also under “ Elliptic * Multiple
    Multiple
    Gauss sums: J.74.
    integrals : 1905, 2825 : A .64 :
    functions." An.52 : 0 .8 ,11 : thsCD . 1 : thE .
    degradation of : M .14. 36 : J.69 : L .39,433,452,46 ,th48,56 :
    factors of integral functions : C .24 . LM .8 : Me.762: 2 .13,3.
    functions and integrals : An.51: J. || double : 2710 , 2734 – 42 : A . 13 : Ac.5 :
    184,193,20,21,23,25 : M .20. An.70 : J.272: G .10 : L .582: Mém .
    indices of polynomials which furnish 30.
    the powers and products of a approximation to : J.6 .
    binomial eq : C .25. Cauchy's theory, ext. of : C .759.
    relations : At.65 . change of order of integration :
    Modulus : - of functions, principal: 2775 : A . 19.
    C .20. expressing an arbitrary function :
    of series : C .17. J.43.
    * of transformations : 1604 : A . 17. reduction of : J.45 : Z .9.
    *Momental ellipse : 5953. residues of : C .75g.
    904 INDEX.

    Multiple integrals,double - (continued ): | Multiple integrals — (continued ) :


    (x ? — yº) dedy volume integral of
    JoJov }(28 — 13)(C2= 22)(1=— y*)(c*— y) }
    = : L.38. (% ) +( ) + ( ) : .14
    SSSe exp (- «? — y2 — 72), "y9z dx dy dz:
    samewith log of numerator: L.50. N .54 .
    (ae' — )(dydz' — dzdy) + sym SSS ... Q (ax: + by® + & c.) copy?...dx dy ...
    JJ V {(x' —~)2 + ...33 with limits ( to op in each case
    = 4m7 : C .66. (Pfaff) : J.28 .
    transf. ofL JJ V (sin ?yd - ody
    sin costo
    dre dy ...
    * {(a — x )2 + (6 — y )2 + ...?"
    J.20 . continuous functions : TI.21.
    1 2-2 cos ix cos je dx dy
    7 ?). Jo v {1 + a? + 2a(ucosa + vcosy)}" do.with n = " and with a numerator
    C . 96 .
    SS F (a + bæ + cy )dx dy : A.37. (a–a)F (.*+ +...): CM.3.
    SS ... F (« , y , 7...) PQ dx dy dz ...where
    SS F' (a + iy )dx dy : J.42 . P = (1 - )* - (1 - 4) 6- ...
    ( F (1 , t, ) at du : 0 .96 . Q yaza+b fa +b+e... : L .59.
    JJ G (2,t,x)" arising from (2604), viz :
    1 ° (ax" +by") čep -1 y2 -1dedy : J.
    Jo Jo
    eexp( -22– )dæ: Pr.42.
    37.
    evaluation : A .5 : by Fourier 's th , *Multiple points : 5178 : CM .2 : thG . 15 :
    CM .4. Me.2 : Q.2 ,6 .
    expansion of : Q .8 . on algebraic curves : An.52 : L .42 :
    Frullanian : LM .15. N .51,59,81, at oo 643.
    limits of : LM . 16 . on two curves having branches in
    reduction of: An.57 : L .41,39. contact : 0 .77 .
    by transf. of coordinates : C . 13 . on a surface : J .28.
    *Multiplication : 28 : J.49 ; abridged, N .
    SÓF (x2+ y’ + ...)© (ax + by + ...)dxdy 79 .
    ... : L .57. by fractions : Me.68.
    of theory ofattraction : CD.7. Multiplicator equations : M .15.
    transformation of : 2774 : A .10 : An . Multiplicity or manifoldness : J.812,86 :
    : Z . 20 .
    53 : No.47 : CM .4 : Mél.2 : Mém . Music :thAc.5
    E .273, 28 : Pr.37.
    38 : Q .4 ,12.
    an indef. double : J .82,10. Nasik squares and cubes : Q .8 ,15 .
    a triple integral: 2774 – 9 : J.45. Navigation , geo . prs. of use in : A . 38 .
    yıdai + ... + yndæ , : LM.11. Negative in geometry : No.1792.
    triple : 2774 : A .30 : J.45. Negative quantities : At.55 : N .443,67 :
    which are unaltered in form by trans TE . 1788.
    formation of the variables : J.15, Nephroid : LM . 10 .
    91. Net surfaces : J.1,2 : M .1 : any order,
    An .64 .
    SSS ...dx dydz ...: Q.23. quadric : J .70,82 : M .11.
    quartic : M . 7 .
    SSS... ?-?y" –1 1...dx dy dz... with and series : C .62 .
    different limiting equations : trigonometrical : Z .14.
    2825 : CM .2: L.51. having a 3-point contact with the
    * some other integrals evaluated by r intersection of two algebraic sur
    functions : 2826 – 34 . faces : G . 9 .
    Newton , autograph m .s .s of : TE .12.
    SSSF (ax + by + cz,a'æ + by + c'z, * Nine-point circle : 954, 4754 : A .41 : E .
    a " x +by + cºx) dx dy dz, limits 7,30,th35,pr39 : G .1,ths4 : Me.64,
    + 00 : A . 30 . 68 : Q .5 – 8 .
    INDEX. 905

    Nine-point circle - (continued ) : | Notation _ (continued ) :


    an analogous circle : A .51. *° (aby) = u : 4656 ; $ (auv) or U , 4665.
    contact with in - and ex -circles : 959 : * Gur: 3732.
    Me.82 : Q .13. * H (n ,r ) : 98.
    and 12-point sphere, analogy : N .63. * J : 1600.
    Nine - point conic of a tetrahedron : N .G .F = numerical generating func
    Me.71. tion .
    Nonions (analogous to Quaternions) : N = b (mod r) signifies that N - b is
    C .97,98 .
    Non -uniform functions : C .88. divisible by r.
    Nodal cones of quadrinodal cubics : Q . * In = n(n) = n ! : 94, 3713.
    10 . at as operator : 3500.
    Node cusps : Q .6 . P (n ,r) or Pro = n ("): 95. Also ,
    Nodes, two-plane and one-plane : M .22. P (n,r) = number of triplets of n
    * Normals : 1160 : 4122 - 3, 5122 : A .13, things, & c .
    53 : LM . 9 : p .eMe.66 : Z .cn2 and 3 .
    of envelopes : Me.80 . ♡ (@) or Z'(c ) = d. log r (x) : 2743.
    * plane of a surface : 5772 . R ,7, 1a : 909 – 13.
    *

    principal: 5722 : condition for being * Sm , Sın,p : 534 ; Sa ,2940.


    normals of a second curve, C .85. * sinh , & c.: 626 ; sinh, & c., 2180.
    of rational space curves : J .74 . * E : 3781 - 3 .
    section of ellipsoid (geodesy ) : A .40 . Un : 3499.
    of a surface : 5771 , 5785 : C .52 : CD .3 :
    CM .2 : L . 39,47 ,72 : M .7 .
    Ši (n) = sum of divisors of n.
    coincident : L .48.
    transformation of a pencil of : C .889.
    la mor är or E (a )=integernext
    *Notation (see also “ Functions " ) :
    A , B ,C , F ,G , H : 1642.
    * A .P ., G .P ., H .P.: 79,83,87.
    * (azb2c3): 554. na =integer next >
    (n) ( ), Jacobi's function (see = rth coeff. of (1 + )".
    “ Functions ” ). < = not less than ; $ = not greater
    a' or a /b: 2451. than.
    st

    at, 1te1t : 160. ( + ) = denominator to be stated after


    wards.
    * * e

    Byn , Bernoulli’s nos.: 1539. ( x ) and ( Ⓡ ) : 1620.


    C (n ,r) or Cn,r: 96. Otherwise algebraic : CP.3.
    C (n ,3) = number of triads of n things, for some developments : C .98.
    & c. continuant = contd . fraction determi.
    C ) = rth coeff.ofnth powerof(1 + x): nant.
    median = bisector of sideof a triangle
    also Jacobi's function (see“ Func
    tions” ).
    drawn from the opposite vertex.
    subfactorial : Me.78 .
    :

    * D : 3489 : dr,dur, & c.: 1405. suggestions : Me.73.


    *Numbers (see also “ Partition of,” and
    * d’y,dx),dy,&c.: 1407. “ Indeterminate equations” ): 349 :
    A .2,16 , 26 ,58,59 : Ac.2 : AJ.4 ,6 :
    * d (uvw ) : 1600. C .f12,43 ,442,454,160 : CM . 1 : G .16 ,
    d (xyz ) 32 : J . 93,39,40 ,, 18 ,77 ; tr ,273,28 and
    * A: 582, 1641, 3701 ; A', 1645. 292 : JP.9 : L.37 – 39,41,45,586,59,
    * E : 902, 3735 . 60 : LM . 4 : Mém .22,24 ; tr (Euler ),
    * e : 150, 1151. 30 : N .443,62,79 : Q .4 : TE.23.
    ap. of algebra, JP .11 ; of r function ,
    e expæ+ 1 or e x+ 19=e*** No.81 ; of infiuitesimal analysis,
    J.19,21.
    *j(x): 400, 1400 ; f-1(a), 430. formulæ : L .642,652.
    * fº (20) fr («): 424, 1405. relation of the theory to i.c : C .82.
    5 Z
    906 INDEX.

    Numbers - (continued ) : Numbers — (continued ) :


    approximation : - to N , E .17 ; to representation ofby forms: C .92 ; by
    functions of large numbers, C .82. infinite products, A .1.
    binomialeqs. with a primemod : C .62. square having prime factors of the
    cube : Q .4 . form 4n + 1 : N .78 .
    cubic binomials : 23 + 43 : 0 .61%. squares of : J.84 : M .13 : Pr.63,7 .
    determined by continued fractions : three in ar.p : N .62.
    LM . 29 . sums depending upon E (x ) : L .602.
    digits, calculus of, th : J.30. sumsof digits : Me.66 : TE.16 .
    digits terminating a power : A .58 : sum and difference of two squares :
    N .46 . ths N .63.
    sums of divisors : 377 : Ac.6 : G .7 :
    Dirichlet's th. (m ) = $(m ): L.57. L .632 : Mél.2 : Mém .50.
    sums ofpowers of (see also " Series " ):
    Dirichlet's f. for class numbers as 276 , 2939 : An.61.65 : thCD .5 :
    positive determinants : L .57. Me.75 : N .42,56 ,70 : Q .8.
    division of : A .26 : J.13 : Mél.3 : Pr. of cubes: An .65 : L .66 ; of the odd
    7 ,10 ; by 7 and 13 , A . 25 ,26 ; by nos., A .64.
    me? Ený?, Mém . 15 : P. 17 ,88 : Q .
    19, 20 . of n primes : N .79,; 4th powers,
    A .54.
    divisors of y2 + Az? when A = 4n + 3 of squares : A .67.
    a prime: J.9. of uneven orders : Mo.57.
    divisors arising from the division of symmetrical functions of: Q .7 .
    the circle : L .60. systems: Z .14 ; history of, by Hum
    4m + 1and 4m + 3 divisors of a num . boldt, J.4.
    ber : LM .15 . theorems: A.7,10 ,20 ,49: An.70 : C .
    factors of: Mém .41. 252,43,83 : CM .2 : G .8 : L .48 ,52 :
    Gauss's form : L .56 . N .75 ; Cauchy's, gzC .53 ; Eisen
    integral quotients and remainders :
    An.52. stein 's, J.27,50,83 : LM .7 : Q .5,6.
    large, analysis of : A .2 : C .2,29. Gauss's on X = - Y"Y : 0.98.
    method with continuous variables : Lagrange's arithmetical : A .47.
    J .41.
    multiples of: C .2. p " + q" in terms of po : N .75 .
    non-pentagonal th : J.31. on 2 + 1 : C .85 ,86 : Me.78 .
    number of integers prime to n in 2, biquadratic character of : C .57,
    m ! = $ ( n) : 1. 57. 66 : L .59.
    on (n + 1)* — " : N .44 ; p: (n ), L .69 .
    odd : A .12: and prime to all squares, on n ' - n (n - 1) + , & c. : 285 : A . 30 .
    C .67. on 2m positive numbers : N .43.
    Pellian equation : prA .49 : LM .15 :
    sol. by ell. functions, Mo.63.
    perfect : 0 .81 : N .79 .
    on P (m ) + E ( m ) P (1) + : Q.3.
    polygonal, Fermat's th . of : P .61. on the greatest product in whole
    polynomialshavingdeterininate prime numbers of given sums: J.57 .
    divisors : C . 98. on an odd sum of 12 squares: L .60.
    powers of, 12 theorems: N .46 . on products of sumsof squares : G .2 .
    prime to and < N : A .3,29 : E .28, : on 4 squares : N .57.
    J.31 : N .45. on 2 , 4, 8, and 16 squares : Q .17.
    prime to and < the product of the on o (a ) + (a ') + , & c. = n , where a ,
    first n primes : A .66. a', & c.,are the divisors of : Cm .3 .
    primewith
    0 .54.
    respect to a given ar.p : Numeration , ancient decimal: C .6 ,84
    Numerical approximations: N .423,53.
    prime to the radix having multiples Numerical functions : L .57 : Me.62.
    made up of repeating digits : simply periodic : AJ.1.
    Me.76 . sums of, approximately : 0.96 .
    products of divisors of : 2 .20 . which express for a negative deter
    quadratic forms of: Mém .53. minant the number of classes of a
    rational linear functions taken with quadratic form , one at least of
    respect to a primemodulus, and whose extremecoefficients is odd :
    connected substitutions : C .483. C .62 .
    INDEX. 907

    Obelisks : A .9,11. | Orthogonals,algebraic system of: C .69.


    Oblique : - bevilled wheels, cn : J.2. * Orthogonal : - circles : 4170, 4182 — 4 ;
    * coordinates : 4050 , 5511 – 9 : fN .54 . of in - and circum -circles of a tri.
    cyclic surface : TI.9. angle, Q.18.
    and osculating circle of a conic : G .22. circle and conic : E .7.
    Octahedron function : Q .16 . coefficient system : A .2,61.
    Octahedron , centroid of : LM . 9. coordinates : C .60 ; curvilinear, JP.
    Octic equations : G .7,10 ; and curves, 26 .
    M .152. conics : N .84 ; families of, A .63.
    Octic surface : G .13 : M .4 : Q .14 . curves : J.35 : N .52,81.
    * Operative or symbolic calculus : 1483, system from logarithmic repre.
    3470 — 3628 : AJ.4 : C .17 : G .202 : sentation : Z .16 .
    J.5 ,59 : LM .123 : Me.82,85 : P .37, | * lines of a triangle : 4633.
    44,60 — 63 : Pr.10.,119,122,134: Q . lines and conics : C .72.
    4 ,5 ,8 . projection : 1087.
    applications : G .19: Me.82. in metrical projective geometry :
    algebraic : TE .14 ; ap . to geometry , GM . 14 .
    CM .1,2 ; , CM .3. of a circle into an ellipse : A .2.
    expansions: Pr.149. of a triangle : E .30 ,31, 36 ,37.
    formulæ : 0 .393. substitution : J .67 : M .13 : Z .24 .
    * index symbol : 1485 : CD.6. surfaces : C .29, spheres 36 ,542,59,72,
    integration : CD.3 : exMe.76 . 79,87,ths17 and 21 : J.84 : JP .17 :
    representation of functions : .43 : L .43,44,46 ,47,63 : N .51 : P .739:
    CD .2. Pr.21 : Q .19.
    semin variant operators : Q .20 . cubic eq . for : C .769.
    on the symbols a ", log x, sin x , cos a , with elliptic coordinates : C.53 :
    sin -'x , cos- tæ : A .9,11. J.62.
    theorems: A .57 : CD.8,: LM .11 : Q . and isothermal: C.84 : JP.18 : L .
    32,15 ,163; from Lagrange's series, 43,49 .
    Q .16 ; from apu - pwu = pu , CD.5. systems: C.67,754: M .7; condition ,
    * Operators : dr, 1405 ;er, 1520 – 1 , Q.9,. J.83 ; quadric, J.76 ; parallel,
    * (do - m )-?, & c.: 3470 – 85 ; gz of 3474, M . 24 .
    C .43. triple : A .55 — 58 : An.63,77,85 :
    C .alg58,67 ; cyclic, G .21,22 : J.84 ;
    * D ( D - 1) ... (D — 12 + 1) : 3489. quartic , 82 : Ž .23.
    a = æd, tyd, + & c.: 3500. trajectories : L .45 : Me.80 : Z .17.
    expansions and formulæ for : of circles : Me.85.
    F (« dz) U , where U = f (x ) = a + bx + of circular sections of an ellipsoid :
    | cc7 + : 3486. L .47 .
    of a moveable plane : Pr.41.
    * f (D) uv, 3494 : uf(D ) v, 3495, with f of a moveable sphere : 0 .42.
    as above. of a surface : Mém . 20 .
    D ' f (xD) U : E .36 . Orthomorphic projection of an ellipsoid
    { 0 (D) ero}" Q : 3491. on a sphere : AJ.3.
    Orthomorphosis of a circle into a para
    ex+DF ( ), & c. : E . 34 . bola : Q .20 .
    = F (x) = x + 1 F ( + 1): E.36. Orthoptic : - lines of a conic : A .57.
    loci of : LM .13 : Pr.37 ; of 3 tangs.
    QF (a,x ): Q.13. to a quadric, E .40.
    surface of a quadric : J.50 .
    2m+//{a vv + b (wi)}, & c.: C.96. Orthotomic circles : Me.64,66 : Q .2 .
    f (x + hD) . 1 : E.39. * Osculating : — circle : 5724 : L .39.
    y (d , + y) pe = º (d , + c)yy : 3498. of conics : A .70 : N .60.
    *

    reduction of F ( TF1) : 3503. of a family of curves : N .70.


    of a parabola , ths : N .66 .
    *

    * F (1) U and F -1(1 ) U : 3509 — 10 . of quadric curves N .43 .


    *

    C (4,m ) un/m !: 3514 . of tortuous curves : N .81.


    *

    transformation of ydr. Udr..., & c : G . | * cone : 5727 : angle of, 5752.


    21. conic : L .39 : triply, A .69 : Z .19.
    Orthocycle : Q .17. of a cubic curve : J.68: : Z .17.
    908 INDEX .

    Osculating - (continued) : Paraboloid - (continued ) :


    curves : Q .11. quadrature of : 6127 : An .55 .

    * *
    helix : N .71. segment of : 6127 – 33 : A .29 .
    line of a surface: C .82: J.81. * hyperbolic : 5623 : A .11.
    parabola : N .81. * of revolution : 6134 .
    paraboloid:
    L .38 .
    JP.15: N .82: ofa quadric , Paradoxes
    16 .
    of De Morgan : J.113,12 , 13,
    * plane : 5721, eq 5733 : and radii of Parallels1,3 :: AMém .8,47.502:
    : At.51 : J.11,73 : Mél.
    Z .21, 22 : Thibaut's
    cury, at a multiple point of a
    gauche curve : An.71 : C .68. proof, A .15.
    of a tortuous curve : C . 96 : J .41,63. | in analytic geometry : A .44.
    sphere : - Mém .20 : N .70 : of curve Parallel curves : J.55,ths32 : LM .3 : Q .
    of intersection of two surfaces, 11 : 2 .5 .
    cn : C .83. closed : A .66 .
    of two curves having a common | * of ellipse : 4960 : A .39 : An.60 : N .442:
    principal normal : LM .16 . Q .12.
    surfaces : C .792, degree of 98 : L .41, Parallel surface : C.54 : LM .12.
    80 : of quadrics, N .60. of surface of elasticity : An.57 .
    Oval of Cassini: see “ Cassinian oval." of ellipsoid : A .39 : Ån.50,60 : E . 17 :
    Oval of Descartes : see “ Cartesian J.93 .
    oval.” Parallelogram with sides through four
    given points : A .39.
    Pangeometry : G .5,15. Parallelogram of Watt : A .8 : L .80 .
    * Pantograph : 5423: Mém .31: TE .13. * Parallelopipeds : on conjugate diame
    Paper currency : A .42. ters : 5648 .
    Pappus, prs in plane geometry : A .38 : diagonals, & c. : CM .1 .
    2 .5 . equality of : A .4 .
    * Parabola : geo. 1220 – 44 : anal.4200 analogues of parallelograms : LM .2 :
    39 : eqCM .2 : N .424,54,70 : geo Me.68.
    CM .4 : Me.71 : cn1249 : thsN .60, on a spherical base : N .45.
    63,71,76,80.2. system of : LM .8 .
    circum - hexagon and triangle : CM .1. Partialdifferences : question in analysis :
    * circle of curvature of : 1260 : A .61. J . 16 .
    chords of : 1239, 4224 . * Partial differential equations (P . D . E .) :
    * *

    two intersecting : 1242 . 3380 — 3445 : 0 .34,11,16 ,78,95,96 :


    determination of vertex and axis : thsCD .3 : J.58 ,80 ,prs26 : JP.7,10,
    N .58 . 11 : L .36 ,80,83 : M .11 : Z .6,8 ,18 .
    eq . deduced from eq. of ellipse : 1219. | * P . D .E ., first order : 3399 - 3410 : A .33,
    focus and directrix : N .49 . tr50 : An.55,69 : C .146,53 ,543 :
    focal chord : 4235 – 9. CD.7 : CM . 1, : J.2,17 : trJP.22 :
    :

    latus rectum : 1222 : Me.75 . L .75 : M .9,11,th20 : Z .22.


    normal, length of: 4233 — 4 . * complete primitive connected with
    plane and spherical : A .60 . any solution : 3405.
    quadrature of : 1244 : A .32. * derivation of the general primitive
    and right line: see “ Right line." and singular solution from the
    segment of : 6078 : A .26 , 29 . complete primitive : 3401.
    solid generated by it : N .42 . derivation of a singular solution from
    sector : E .30 : N .57 : Lambert's th, the differential equation : 3403.
    J.16 . with a general first integral : Me.78 .
    in space: A .3. integration by : - Abelian func
    * tangents : see under “ Conics." tions : C . 94 .
    through 4 points, cn : 4837 : J. 26 . Cauchy's method : C .81.
    * *

    triangle of 3 tangents : 1237 , - 68 : A . Charpit's method : 3399.


    47 : Me.75 . Jacobi's first method : C .79,82 : and
    trigonometry of : CD .8. ap. to Pfaff's pr, J.59.
    * Paraboloid : 5621, 6126 — 41 : N .61 : Q . Jacobi- Hamilton method : M .3.
    13. Lie's method : M .6,8.
    * generating lines of: 5624. Weiler's method : M .9.
    of eight lines : C .84. * law of reciprocity : 3446.
    * elliptic : 5622: A .11 : L .58 : P.96 . and Poisson 's function : 0 . 912.
    INDEX . 909

    P .D . E ., first order - (continued) : P.D .E., second order - (continued ):


    simultaneous: C .68,76 : L .79 : M .4,5. | dz (p sin se) + t + n (n + 1), sin2x = 0 :
    singular solution : 3401 - 3 : J.66 . L .46 .
    systems of : A .56 : M . 11,17 .
    theorem of Jacobi : C .45. 10 (1 + q2) = +(1 + p2) : J.58 .
    3 variables : J .64 .
    n variables: 3409 : A .22 : J.60 : LM . q2r - A2t + 4* nyq= 0: E.22.
    10, 11.
    with coustant coefficients : Mél.5 . d: (pæ)+ d,(qx) = 0 : 2.28.
    integration by calc . of variations : construction of explicitly integrable
    C .14 . equations ofthe form s = z ) (x,y):
    % = px + qy + F (p, q) : Z .5. JP .28.
    cea yozep" q" = A : CM .1. 28" (x) F ' (y) - 7 : C .81.
    (1 + Pit... + PrXdx,d , )" Z = Q : Z.13. {f (a)+ F (y)}2 " :
    * P.D .E ., first order , linear : 3381 — 95 : a dry log 1 + 1 = 0 : L .53.
    reduction to , C .15 : J.81 : Me.78 . 8 + Pp + Qq + 2 = 0 : Me.76 .
    * Pz: + Qz, = R : 3383: extension to 8 + ap + bq tabz = V : 3444.
    n variables, 3384. (ax + by + c)stara + bup = 0 : A.33.
    L (pæ + qy — 2 ) — Mp - Nq + R = 0 : (a + y)2 sta (x + y) p + b (æ + y) q + cz
    0 .83. = 0 : A .33,38.
    a( 2 – z , + 6(zuk - gu ) + c( 2tv - aux) 22 (zs - p9)? + q = F (y) : A .70.
    = 1 : Q .8 .
    Pt: + Qty + ... + Rt, = S : 3387. rt _ 82 : geoQ .2.
    *

    % = _ y * P = (rt - s2)" Q , Poisson's eq. : 3441.


    ✓ (y2 — 2);%
    2) **ry = a2 + y?: 3390 – 1. (1 + r) t + (1 + t)r – 2p78 = 0 : An.532.
    *

    (c — a )% , + (y — 6) Zy = C — % : 3392. 9 (1 + q)r + p (1 + p) t - (p + q + 2pq)


    * 22 + y2 + 2 = 2ax : 3393. s = 0 : 3432.
    simultaneous: 3396 – 7 ; ex. 3398 : 482 + (r — t)2 = 4k2 : approx. integn.,
    J .81. C .74 .
    u = Vy and , = - Vr: J.70. 18 + Bq + y (r,p , q, x, y,z) = 0 : C .93.
    * P.D .E ., second order : 3420 — 45 : A .33: (log -)zy + az = 0 : C.36.
    C .54 ,702,78,98 : JP.tr22 : L .72 : r dr
    M . 15 : Me. 769,77 : Mél. 3 : N .83 : U21 = Rr 2 wb
    P .46 : transf. of, C .97. (2 R72 + A2)
    in two independent variables : trA . L.38.
    54 : trNo.81 : C .92 ; transf. of, 97 : * U2x + 22, + 12: = 0 (see also “ Spherical
    M . 24 .
    in 4 and 5 variables : Mém . 13. d harmonics ”) : 3551, 3626 , J.36 :
    m e t h o
    * Rr + S8 + Tt = V, Monge's method : * Urtu, : Mo. 78 .
    tu: = xyz : 3552.
    *

    3423 : CM .3 : N .76 : Q .6.


    * Rr+ S8 + Tt + U (rt — 82) = V : 3424 , * aur + buy + cuz = xyz, & c.: 3554.
    *

    3434 – 40 : J.61. XU2 tau , — q æu = 0 : 3618 .


    *

    a2 (U2r + U2y + Uzz) = U2 : 3629 : C .7 :


    Rr + S8 + Tt + Pp + Qq + 2% = U :
    *

    3442. LM .72.
    r + t = 0 : A .2 : CM .1 : J.59,73,74 : | integration C .742.
    by change of variables :
    L .43. P.D .E ., third order, two independent
    po+ t + k2% = 0 : M .1. variables : LM .8 : N .83.
    pra?t = 0 : 3433. P .D .E ., fourth order : AAU = 0 : C .69.
    *

    go- a?t = ° ( ,y), & c. : 3565. P .D .E ., any order : No.73 : C .80 ,89 :
    *

    æ (r — aat) = 2np : E .13. M .11,13.


    g - a?y2bt = 0 : E .27,28. sc ?Ene = %ny : 2.7.
    g = q2mt: C .98. two independent variables: C .75 :
    p = q : J.72. CP.8 .
    910 INDEX .

    :
    P . D .E ., any order — (continued)and Partition of numbers---(continued ) :
    any number of functions inde. into ten squares : C .60 : L .66 .
    pendent variables : C .80 . into p squares : 0 .39,90 : L .61 : N .
    and ap. to physics : JP .13. 54.
    of cylinders : Me.77. and an integer : L .57.
    and elliptic functions : J.36 : hyper into Lthe.57.product of two sums of sqs.:
    ell., J.99.
    integration by definite integrals : An. into parts, the sum ofany two to bea
    59 : C .94 : L .54 . sq . : Mém . 9.
    of dynamics : C .5 : J.47. into 2 cabes : L .70.
    Hamiltonian : M .23 : Z . 11. into sum or difference of 2 cubes :
    of heat: L .48 ; of sound, L .38 . AJ.2 .
    integrated in series : C.15,16 . into 4 cubes : N .79.
    of Laplace : G .23. into maximum nth powers : C .95 .
    linear : An.77 : C .13,90 : CD.9,: CM . into 10 triangular numbers : C .62.
    2 : J.69 : JP.122 : L .39. formation of numbers out of cubes :
    J. 14 .
    of orthogonal systems of surfaces : 2 squares whose sum is a sq. : E .20 :
    Ac.4 : C .77. N .50 .
    with periodical coefficients : C .293. 3 squares, the sum of every twobeing
    of physics : L .72,47. a sq. : E . 17.
    P .D .E .,simultaneous : C.92,th 78 : LM .9 : 4 squares , the sum of every two being
    M .23 : 2 .20. a sq . : E . 16 .
    linear : J.65. 3 nos., the sum or diff. of two to be a
    P.D . E., system of: C .13,74,81. sq . : Mém . 18 .
    a ”mmt = x2m2mx: A. 30 ,31, by = eatf(w ). 2 sums of 8 sqs. into 8 sqs.: Me.
    78 .
    anx = %{m +n)y + F (y) + « F2(y) + ... + a sum of 4 sqs. into the product of 2
    wm -?F , (y ) : A .51. sums of 4 sqs. : TI.21.
    Azne+ (d2r + day t ...)"z = 0 : C .94. n nos. whose sum is a sq . and sum of
    dz = & Hdx + Kdy+ Ldp + Mdq + Ndr + sqs. a biquadrate : Ê .18 ,22,24.
    c. : J. 14 . a quadric into a sum of squares : N .
    Partial differentials of re? + 42 : J.11. 80 ,81,
    ns - ms into 3p2 + 92 : N .49.
    a square into a sum of cubes : N .67.
    30 ,66 :
    * Partial fractions : 235, 1915 .2: :A.J.1,5,9, a cube into a sum of cubes : E .22,
    C .46 ,492,783 : CM .1 : G 23 .
    10,11,22,32,50.,: JP .3 : L .46 :Mém . into 4 cubes : N .77 ; into 3 or 4
    9 : N .45 ,,64 ,69 : Q .5 . cubes, A .60 .
    Partition of numbers (see also “ Num . n12 – 9m12 or its double into 2 cabes :
    bers ” and “ Indeterminate eqs." ): N .8i .
    AJ.2,5,6 : An.57,59 : At.65 : 0 .80, 3 nos.whose sum is a cube, sum of
    86,90,91 : CP.8 : J.13,61,85 : M . sqs. a cube, and sum of cubes a
    14 : Man.55 : Me.78,79 : Mém .13, sq . : E .26 .
    geo.ap20 ,44 : Mél.1 : N .69,85 : P . 5 biquadrates whose sum is a square
    50,56 ,58 : Pr.7,8 : Q .12,2,7,15 : Z . E .20 .
    20 ,24 . of n into 1, 2, 3, & c. different num
    by Arbogast's derivatives : L .82. bers : E .34 .
    of complex numbers in Jacobi's th : of pentagonal numbers : C .96 .
    C .96 . a series for the: AJ.6.
    by elliptic and hyper-elliptic func tables, non -unitary : AJ.7 .
    tions : J .13. theorems: AJ.6 : C .40,96 : Me.76 ,80,
    formula of verification : Pr.24 . 83 : pr. symm . functions, G .10 .
    into 2 squares : An.50 ,52,54 : C .87 : J. Partitions : - in theory of alg . forms:
    49: LM .8,9 : N .54 ,78,alg65 : odd G .19 .
    squares, Q .19. number of for n things : E . 10.
    into 3 squares : J.40 : L .59,60. in planes and in space : J.1.
    into 4 squares : C .99 : L .68 : Pr.9 : Q .1. Sylvester' s theorem : Q .4 .
    4 odd, or 2 even and 2 odd : Q .19,20. trihedral of the X -ace and triangular
    into 5 squares : C .97,98 : J.35. of the X -gon : Man .58.
    INDEX. 911

    * Pascal's theorem : 4781: AJ.2 : CD.3 , Periodic functions - ( continued ) :


    4 : CM .4 : J.34 ,41,69,84 ,93 : LM . of 2nd species : M .20.
    8 : Me.72 : N .44 ,52 ,82 : Q .1,4 ,5,9 : of several variables : C .43: J.71.
    Z .6 ,10 . in theory of transcendents : J.11.
    extension of and analogues in space : of 2 variables with 3 or 4 pairs of
    C .82,98 : CD.4 ,5,6 : G .11 : J.37,75 : periods : C .90.
    M . 22 : Me.85.
    ap . to geo . analysis : A .18 .
    with non-periodic in def. integrals :
    C .18 .
    on a sphere : A .60 . Periodicity theory : M .18 .
    Steiner's “ Gegenpunkte " : J.58. Periods : - cyclic, of the quadrature .
    Pascal lines : E .30. of an algebraic curve : 0 .80 ,84.
    Pedal curve: A.35,36,52 : J.48,50 : M .6 : of the exponential et : C .83.
    Me.80 ,81 : Q .11 : Z .52,21. of integrals : see “ Integrals."
    circle and radius of curvature : 0 .84 :
    Z . 14 .
    law of: 0 . 963.
    in reciprocals of primes : Me.73g.
    of a cissoid , vertex for pole : E .1. * Permutations: 94 : Al. : 0 .22: ČD .7 :
    of a conic : A .20 : LM .3 : Z .3. L .39,61: LM .15 :Me.64,66 ,79: N .
    central: A .9 : Me.83 : N .71. 44,71,763,81 : Q .1 : Z .10.
    foci and vector eg . : LM .13 . alternate : L .81.
    negative central: E .20,29 : TI.26 . ap . to differentiation and integration
    negative focal : E . 16 ,17,20 . A .21.
    nth and n - 1th : E .18. of n things : 0. 95 : N .83 ; in groups,
    of evolute of lemniscate : E .30 . L .65 .
    inversion and reciprocation of: E .21. of 3q and 2q letters, 2 and 2 alike:
    of a parabola , focal and vector eqs. : N .74 ,753.
    LM .13. number of values of a function
    rectification of difference of arcs of: through the permutations of its
    2 .3. letters : C .20,21,46 ,47 : L .65 .
    which is its own pedal: L .66 . successive (" battement de Monge " ) :
    Pedal surfaces : A .22,35,36 : M .6 : J.50 : L .82 .
    2 .8 . with star arrangements : Z .23 .
    counter : AJ.5 . * Perpendicular from a point : — upon
    volumes of: C .55 : A .34 : An.63 : J. a line : length of : 4094, t.c4624 :
    62 : Pr.12 . eq4086 , t.c4625 : sd5530.
    Pentagon , ths : A .4 : J.5,56 : N .53. upon tangent of a conic : 4366 — 73.
    * *

    diagonals of : A .57. * upon a plane: 5554 .


    Pentagonal dedecahedron : A .25 . * upon tangent plane of a quadric : 5627.
    * *

    Pentahedron of given volumeandmini * ditto for any surface : 5791 - 3.


    mum surface : L .57. Perpetuants : AJ.72.
    Periodic continued fractions: A .62,68 : * Perspective: 1083: A .692: G .3 : thsL .
    0 .968: J .20,33,53 : N .68 : Z .22 . 37 : Me.75 ,81.
    closed form of : A .62. analytical : A .11.
    representing quadratic roots : A .43 . of coordinate planes : CM .2.
    with numerators not unity : 0 .96 . drawing : 1083 – 6 .
    Periodic functions : A .5 : J.48 : N .67 : of circle and sphere : A .57.
    figures ical:
    gzC.89 : M0.84. isometr CP.1.
    cos — cos 3x + cos 5x : CD.32. oblique parallel : Z .16 .
    doubly : C .329,40,70 ,90 : J.88, : L .54. projection : A .16 ,70.
    of 2ud kind : C .90,98 : gzL.83. relief : A .36 ,70 : N .57.
    of 3rd kind : C .97 . triangles : 974 : E .29 : J.89 : M .22,16 :
    monodromic and monogenous : in a conic , Al.
    C .40. Petersburg problem : A .67 .
    with essential singular points : C .89. Pfaff's problem : A .60 : C .94 : J.61,82 :
    expansion in trig . series : N .78 . M .17 : th of Jacobi, J.57.
    4-ply , of 2 variables : J.13. Pfaffians, ths on : Me.79,81.
    2n -ply, of n variables : Mo.69. Tf (see also “ Expansion of” ) : 0 .95 : E .
    multiply : C .57,58 %. 30 : N .42,45.
    integrals between imaginary limits : calculation of: A .6,18 : E .27 : G .2 :
    A .23 . cnJ. 3 : Me.73,74 : N .50,56 ,66 .
    real kind of : Mo.66,84 . by equivalent surfaces : N .48 .
    912 INDEX.

    - calculation of — (continued ) : Points - ( continued) :


    by isoperimeters : N .46 . four, or lines, ths : CD .8.
    by logarithms: N .56 . at infinity on a quadric : N .65.
    to 200 decimal places : J.27 ; to 208, roots in a closed curve : N .68.
    P .41 ; to 333, A .212; to 400, A .22 ; in space, represented by triplets of
    to 500, A .25 ; to 607, Pr.6 ,11,22. points on a line : LM .2.
    formula for, or values of : A .12 : systems: JP. 9 : M .6 ,25 : N .58 .
    J.17 : L .46 : M .2O. of cubic curves : Z .15 .
    three : coordinates of, N .42 ; pr, A .8 .
    a = 3+ approx.: Me.66 ; * Polar : 1016 , 4124 : A . 28 : J.58 : gzLM .2 :
    Me.64,66 : N .72,79.
    = 2 log i, J.9. of conics : 4762 : thsN .58 .
    of cubic curves : J.89 : L .57 : Mél.5 :
    functions of: p.6 : A .1 : C .56,74. Q .2 .
    - : E.27 ; to 140 places , LM .4 . curves, tangents of : N .43.
    hyperbolic logarithm of : LM .14. developable : 5728.
    powers of 7 and of 7 -1 : LM .8 . inclined : N .59.
    incommensurable : 795. line of two points with respect to a
    series for : Q .12 : TE .14 . quadric : 5685 .
    theorem on ñ and e : Q . 15 . plane : Q.2 : 2 .22.
    of a quadric : 5678, 5687 : An .71 ;
    II (w ) = 1 (1 + x )... {1 + (n - 1)x } : A . 12 : of four, LM .13.
    J.43,67. of a quartic : L .57.
    II («) and imaginary triangles and of 3 right lines : A .1.
    quadrangles : 8 .51. * subtangent : 5133 .
    Piles of balls and shells : N .72. Polar surface : - of a cubic : J.89 ;
    Pinseux's theorem : Mo.84. twisted , Z .23.,24 .
    Plane : J.20,45 : Mém .22 . of a plane : C .60 : N .66 .
    * equations of: 5545, 4.c 5550, p .c 5552 : of a point : N .65.
    2 .1. tetrahedron : J.78 : N .65 : Z.13.
    under given conditions : 5560 — 73. of a triangle : A .59 ; perspective, J.
    * *

    condition for touching a cone : 5700. 65 .


    ditto for a quadric : 5635 ,5701. Pole : of chords joining feet of nor
    cond . for intersection of two planes mals of conic drawn from points
    touching a quadric : 5703. on the evolute: N .60.
    figures,relation between : A .55 : J.52 : | * of the line datuB + vy : 4671.
    M .3 . * of similitude : 5587.
    kinematics of: Q .16 . * Pole and Polar : 1016 , 4124 .
    and line, problems: CD .2 : J. 14. Political arithmetic : trA.36 – 38 .
    motion of : JP.2 : LM .7. Pollock 's geo. theorems: Q .1.
    point-systems : J.77; perspective, Poloids of Poinsot : CD.3.
    Z . 17 . Polyacrons, A- faced : Man .62.
    representing a quadric : N .71. Polydrometry : A .38, 39.
    * Plane coordinate geometry : 4001 - 5473. | *Polygonal numbers : 287 : Pr.10,11,12 ,
    Planimeter : A .58 : Mél.2,3 ; Amsler's, 13 .
    5452, C .77 ; Trunk 's, A .44 ; polar, Polygonometry : thsAn.52 and J.2,47:
    A .51 : N .80 . Mém .30.
    Planimetrical theorems: A .37,60. Polygons (see also “ Regular poly
    Plücker's complex surfaces : M .7. gons ” ) : An .cn53,63 : JP.4,9 : N .
    Plückerian characteristics of a curve 74 : Z .11 ; theorems: A ,1,2 : C .
    discriminant: Q .12. 269 : prsCD .52 : Mél.2 : N .58.
    Plückerian numbers ofenvelopes: 0 .782. * area of: 748 , 4042 : J.24 : N .48,52 .
    Point-pair, absolute on a conic : Q .8 . articulated and pr. of configuration ,
    harmonic to two such : Z .13. tr : An.84 .
    Point-plane system : M .232. centroid : N .77 .
    Points : - in a plane, relation between of circular arcs, cn : A .3 : J.76.
    four : A .2,26 . classification : Q .2.
    tg.eq of two : 4669, 4913. division into triangles : A .1,8 : L .38 ,
    on a circle and on a sphere : N .82. 392 : LM .13 : Pr.8.
    of equal parallel transversals : A .61. of even number of sides : LM .1.
    INDEX. 913
    Polygons - ( continued ) : Potential- (continued) :
    family of : N .83. of cyclides: C .83 : 5.61 ; elliptic, Me.
    maximum with given sides : J.26 . 78 .
    of n + 2n sides, numbers related to :
    A .62.
    cyclic-hyperbolic,
    94.
    tables: J.4,63,72,83,
    of Poncelet,metrical properties : L .79. of ellipsoids : J.98 : Me.84 : Q.14 ;
    semi- regular : JP .24 ; star, A .59 and two homog., J.63,70.
    L .79,80 . elliptic : J .47%.
    sum of angles of: A .52 : N .50 . of elliptic disc, law , 7 -3 : Q .14.
    Polyhedral function (Prepotentials) : Gauss's f. and theory : 2 .84.
    CP.13,. gz . of first and second : L .79.
    * Polyhedrons : 906 : 0 .462,60 — 62 : J.3 : history of : J.86 .
    JP.4,99,15 ,24 : L .66 : Man .55 : N . of homogeneous polyhedra : J.69 .
    83 : P .56 ,57 : Pr.8, 9, 112,12 : Q .7 : Jellett's eq . and ap.: Q .16 .
    Z .11,14 ; theorems : E .39 — 42 : J . Newton 's : M . 11, 13 , 16 .
    18 : N .43 ; F + S = E + 2 : 906 : one- valued : J.64.
    A .24 : E .20 ,27 % p .d .e of: C .90 .
    classification of : C.51,52. Poncelet's ths : Z .3.
    convex : angles of, C .74 ; regular, A . | a related integral: L .45.
    59. of a right solid : J.58.
    diagonals, number of : N .63. of a sphere : Me.81 ; surface of, Me.
    Euler's theorem : J.8,14 : Mém .13 : 83 : Q .12 : 2 .7.
    Z .9 . surfaces : J.54; conicoids, & c., 79.
    minimum surfaces of : A .58 . vector : Me.80.
    maximum : regular , C .612 ; for a given Pothenot's problem of the sphere : A .
    surface, M .2 : Mél.4 . 44 ,47 ,5 +2.
    regular : ellipsoidal, C .27 ; self-conj., Powers : - angular functions, & c.: J.72.
    A .62; star, A .62 : thsC .26.,. and determinants, relation : Z .27.
    symmetrical : J.4 : L .49. of negative quantics : Me.73.
    surface of : A .53 ; volume, J.24 : N .52. of polynomials : JP.15 .
    Polynomials : geometry of, JP.28 : th Power remainders : M .20 .
    Q . 14 . Prepotentials : P .75 .
    determined from its partial differen . Primedivisors of quartics : J.3.
    tial: A .4. Primefactors ofnumbers : J .51: N .71,75 .
    product of two : N . 14. Prime-pairs : Me.79.
    system of : L .56 . * Primes : 319 - 378 : A .2,19 : AJ.7 : fAn.
    oftwovariables analogous to Jacobi's : 69 : C .ths13 ,49,50 ,f63,962: G .5 :
    A .16 . J.th12,th14 ,20 : L .52,54 ,th79 :
    value when the variable varies be LM .2 : M .21 : Me.41: N .46 ,56 : Pr.
    tween given limits : C .98 . 5 : Q.5 : up to 109, M .3.
    Polyzonal curves, VU + V + = 0 : TE . in ar.p : Z .6 .
    25 . calculation of: J.10 : in lst million,
    Porisms: L.59 : P.1798 : Q.11 : TE.3,4,9. M .25 .
    of Euclid : 0 .29,48,56 – 59 : L .55 . in a composite number : C .32.
    distribution of,ap .of recurring series :
    of two circles : Me.84. C .82 .
    Fermat's fourth : A .46 . division of a prime 4n + 1 into sum of
    of in - and circum -polygon : Me.83 : squares : J.50 : ditto of 8n + 3,
    P .6l. 7n + 2 , and 7n + 4 , J. 37 .
    of in - and circum -triangle : LM .6,9 : even number = sum of 2 odd primes
    Q .1. (? ) : E . 10 : N .79.
    Poristic equations : LM .4,5 . Fibonacci's problem : LM .11.
    PoristicLMrelations
    .8 .
    between two conics : of the following forms and theorems
    respecting them : 4k + 1, 4k + 3 ;
    Position , pr. relating to theory of num . L .60 : 6k + 1 ; J.12 : 8k + 1 ; L .613,
    bers : Mél.2. 62 : 8k + 3 ; L .58,60,614,622 : 8k + 5
    Potential: thsAn.82 : C .88 : G .15 : J. or 7 ; L .60,61 : 12k + 5 ; L .61,63 :
    20,32,63,70,th81,85: M .2,3 : N .702: 16k + 3 ; L .61 : 16k + 7 ; L .60 ,61 :
    Z .17. 16k + 11 ; L .60 - 62 : 16k + 13 ; L .
    of a circle : J.76 . 61 : 20k + 3, 20k + 7 ; L .63,64 :
    6 A
    914 INDEX.

    Primes — (continued) : * Principal axes of a body : 5926,'60 ,


    24k + 1, 5, or 7 ; L .614 : 24k + 11 572, 77 : J.5 : JP.15 : L.47: N .46 .
    or 19 ; L .60 : 24k + 13 ; L .61 : Prismatoids: A .39 : volume of, Z .23.
    40k + 3 ; L .61 : 40k + 7 or 23 ; L . Prismoids: A .39.
    603,613 : 40k + 11 or 19 ; L .60 : 1 Prism : volume of, A .6.
    40k + 27 ; L .613 : 120k + 31, 61, 79, * Probability : 309 : A .1,,19,47 : C .65,97 :
    or 109 ; L .61: 168k + 43,67, or 163 ; CP.9 : E .274,30 : G .17 : 1.15 : J.
    L .63: m2 + kn ? with k = 20,36,44 , 262,302,33,34,36,42,50 : L .79: N .
    theorems :: 2.2.
    56 ,or 116 , and n an odd number ; 51,73
    L .65 : 4m2 + 5n2 with m odd ; L . Bernoulli's : LM .5 ; P .62 :
    66 . TE .9,21.
    general formula for : C .63. problems : A .61,64 : CD .6 : E . all the
    of a geo . form , limit of : C .74 . volumes : G .16 : L .37 : Me.4,6 :
    irreducibility of l + æ + ... + cp-1, when Q .9 .
    pis a prime : J.29,67 : L .50 : N .49. de l'aiguille, & c.: L .60.
    law of reciprocity between two, ana in analysis : J.6 .
    logue : J.9 . on decisionsofmajorities : L .38,42.
    logarithms of (2 to 109) : page 6 . duration of life : ČM .4.
    number infinite : 357 : Mo.78 . rouge et noir : J .67.
    number within given limits : A .64 : drawing black and white : I.14 : L.41.
    0 .952 : M .2 : Z .5 . errors in Laplace, p . 279, and Poisson ,
    number of digits in their reciprocals : p . 209 : CP.6 .
    Pr.222,23. games : A . 11 : head and tail ; C .94.
    number in a given quantity : Mo.59. of hypothesis after the event: 324 :
    product of n , th : N .74. Mél.3.
    relative : 349 — 50, 355, - 8 , 373 : J.70 : local : P .68 : ex E .7 .
    L .49. notation : LM .12 .
    tables of : from 108 to 100001699 : of position of double stars : Pr.10 .
    153 of the 10th million : principal term in the expansion of a
    of cube roots of to 31 places : Me.78 . factorial formed from a large
    of sums of reciprocals and their number of factors: C .19.
    powers: Pr.33. random lines : Pr. 16 : E . frequently .
    M = aºl for a prime or composite repeated trials : 317 – 21 : C .94.
    modulu
    totality s: Q .9.given limits : AJ.4 :
    of within statistics : L .38 .
    testimony and judgment : TЕ.21.
    L .52. Products : - continued : Me.772.
    transformation of linear forms of into of differences : thQ.15 .
    quadrat
    on that ic number
    prime forms: ,C .87.
    for which the of 4 consecutive integers : N .62.
    class-number formed from the of inertia : 5906 .
    infinite, convergency of : A .21 : transf.
    ith roots of unity is divisible by of, C .17 .
    : M0.74 . of linear factors : C .9 .
    Primitive roots : C:.64thC.74.
    Primitive numbers : JP.11 : fL .54 : | of n quantities in terms of sums of
    powers : Me.71.
    taJ.45 .
    of binomial eqs.: thsL .40 : N .52. of 2 sums of 4 squares (Euler) : Q .
    of primes 16 , 17 .
    produc t of,: for
    J.49.
    an oddmodulus : J.31. systems of: L .56 .
    of unity : C .92. * Progressions : 79 – 93 : An.64 : C .20 :
    Abeľs theorem : An.56 . G .62,7,11: of higher order, G .122.
    divisors of functions of periods of: with n = a fraction : N .42.
    0 .92. * Projection : 1075,4921 : A .3,6 ,12 : prsJ.
    70 : Q .21,pr13 .
    primes :Jacobi's
    of period, and their: Cresidue
    J.49 : method .70. s , and new geometry : A .1.
    Me.85 . central : G .13 : derivation from or.
    sum of, for an odd modulus : J.31. thogonal, A .62.
    table, for primes below 200 : Mém . centraland parallel, of quadrics into
    . for primes from 3 to 101 : | circles : A .52.
    table38of, of conics : 4921 — 35 : A.66 : J.37,86.
    J. 9. of a cubic surface : M .5 .
    INDEX. 915

    * Projection — (continued ): Pythagorean triangles : taA.1: E .20.


    of curves : J.66 : Z .132: loci of cen - | * Quadratic equations : 45: A.24y: with
    tres, A .6. imaginary coefficients, A .8 .
    on spheres : Q .14 . graphic solution : Me.76 .
    tangents to : cnL.37 . real roots of : J .61.
    of a curved surface on a plane : J.67. solution by continued fractions : L .40 .
    of an ellipsoid on a plane : J.59. by successiveapproximation : N .74.
    of figures in one plane : A .1. Quadratic forms (see also “ Quadric
    gauche : N .66 . functions " ) : trA. 15 : Ac.7 : C .
    ofGauss : Mél.2. 85 : J.27,439,54,56 ,76 ,86 : L .51,,
    map : Mercator's, 1093 : A.50 : G .18 : 73 : M .6,23 : M0.682,743: thsAn .
    JP .24 : N .793. 54 : J.53 : M .20 and 2 .16,19.
    of one line on another : 5529 : on a Dirichlet's method : Mo.64.
    plane, A .6 . having one at least of the extreme
    orthogonal : 1087. coefficients odd : L .67,69.
    plagiographic : A .8. Kronecher's : L .64.
    of ruled quartics : M .2. multiplication of : An.60.
    of shadows : N .56 . number of the genera of: J.56 .
    of skew hyperboloid : Me.75. number which belong to a real deter
    of solids : Man .84 . minant in the theory of complex
    of the sphere : 1090 : AJ.2 . numbers : J .27 .
    stereographic : 1090 : A .30 – 32 : L . odd powers of sq. root of 1 -- 2nU + m2 :
    42,54 . Mél.5 .
    of surfaces on a plane : M .5 . positive : A .11g.
    of surface
    J.70 .
    of tetrahedron on a sphere : | reduction of : J.39 : L .48,56 ,57.
    * of two rectangular lines : 4934 . relation , anal. investigation : Z .14 .
    Projective : correspondence between Quadratic loci, intersection of : AJ.6.
    two planes and two spaces : G .22 . * Quadratrix : 5338 .
    equations of a surface, relation to tangent, cn : N .76 .
    tg .e : E .2. of a curve : C .763.
    figures on a quadric : A .18. Quadrangle : prA .55 : C .95 .
    generation of alg . surfaces : A .1,2 : of of chords and tangents : A .2.
    curve forms, An.14 : J.54 : M .25 : differential relation of sides : Me.77 .
    Z .18 . dualism in themetric relations on the
    geometry : A .8 : An.75 : At.759,78 : sphere and in the plane : 2 .6 .
    G .12 – 14,17 : J.84: M .17 ,18 : ths and groups of conics : A .1.
    N .77 : ths of Cremona , G .133,14 : of two intersecting conics, area :
    thZ .11. LM .8
    loci and envelopes : G .19. metrical and kinematical properties :
    and P .D . eqs.: A .1. C .95 .
    point series : J.91. * Quadrature : 5871 - 83 : A .26 : An.502:
    and perspectivity of higher degree in CD.ths5 , 9 : E .6 : J .34 : L .54 :
    planes : J.42%. Mém.24,41 : N . 12,f55, 54 : N . 75 :
    * Proportion : 68 : A .8 : G .6,13 : TE.4. TI. 1.
    Pseu : G . 10 .: A .2,67 : At.19 : J.
    dosferatheorem approximate : C .95 : N .58 .
    Ptolemy's of the circle : Me.74 : Pr.7,20.
    13 : LM .12. Cote's method : N .56 .
    ext. to ellipse : A .30 : inverse of, A .5 . with equal coefficients, f.: 0 .90 .
    Pure mathematics, address by H . J. S . Gauss's method : A .32: C .84,902:
    Smith , F . R . S . : LM .8 . J.gn55,56 .
    Pyrami36d :: J.3
    triangu lar : A.1,3,21,28,32, from integrals of differentials in two
    : volA .14. variables : M .4.
    vertices of: A .2 . * Laplace's formula , f.d.c : 3778.
    and higher n -drons : prs A .9. of a small geodesic triangle , Gauss's
    and prism sections, collineation, & c.: th : J16,58.
    A .9. by lines of equal slope: 5881.
    Pythagorean theorem : A .11,17,20,24 : quadrics : CD.1 : L .63 : formed by
    gzJ.26 and N .62. intersecting cylinders, An.65.
    sphericalanalogue : A .44 : N .52. sphero-conic : J.14.
    916 INDEX.

    Quadratures — (continued ) : Quadric surfaces : conjugate diame


    which depend upon an extended ters - (continued) :
    class of d .e with rational coeffi. parallelopiped on them : vol. 5648 :
    cients : C .92. sum of squares of areas of its
    *Quadric cones: 5590,5618,54, 97 : N .66. faces, 5645 : do. of reciprocals of
    locus of vertex : C .52. the perpendiculars on its faces,
    throug h six ions
    pointsor: form
    C .52.s : A .139, 38 : 5644 .
    Quad ric funct sum of squares oftheir reciprocals :
    C .44,55 ,78,892,95 : J.47,: JP.28, 5643 : ditto of their projections
    arith 32, positive 99 : algL.74 : on a line or plane, 5646 .
    M0.58 : see also “ Quadratic construction and classification by
    forms." projective figures : A .9.
    bipartite : P.58.
    in coefficients and in indeterminate
    correlation of points and planes on :
    CD .5.
    com xes : J.24.
    ple cubature of : 6126 — 65 : A .14 .
    equivalence of : C .93. diameters : 5677, - 88.
    in n variables : Me.2 : disappearance diametralntplane : 5636 ; gn .eq,5689.
    of products, N .55 . al
    of consta section area : N .43.
    quaternary : An .59 : L .54 : M .5 ,13 : director -sphere of : Q .8 .
    whose det. < 0 , 0 .96 ; and corres duplicate : CD .6.
    ponding groups ofhyperabelians, enveloping cones of : 5664 – 72 , 5697 .
    0 .98 . equation between the coeffs. : J.45 .
    reduction of : C.91,93, 96 : G .5 : to without foci : L .36 .
    sum of squares,Mél.5. focales, a property from the theory
    represented by others: C .93 . of : L .45 .
    transf. of: C .86 : CD.4 : G .1 : LM . 16 : genera l eq .: 5673 : CD.5 : CM .4 : LM .
    24

    N .66 : Q .17. 12, 13 : M . 1 : Me.64.


    reciprocal : J.50 .
    243

    invariability in number of pos. and condition for a cone : 5699 .


    neg . squares : J.53. condition for a sphere : Q .2.
    with two series of variables : C .94 . coefficients : A . 1.
    * Quadric surfaces : 5582 — 5703: A.3,4, generation : 5607 — 24, N .47,75 ; Ja
    12,16 ,45,56 : An.52,geo60 : At.51 : cobi's, J.73.
    C .76 : CD .3 : G .13,14 : J.1,18,38 , homofocal : L .60.
    f42,63,69,89 : JP.6 : L .39,43,50 : 1 indices of points, lines, and planes,
    M .2,23 : geoMe.74 : N .56 ,57,589, theory : N .704.
    593,60,61,77 : Z .5,cn13. intersection of two : 5660 , C .64.: G .
    theorems: An.54 : CD.5 : J.54,85 : 6 : M .15 : Q .10 ; ruled , N .83; with
    L .43 : M0.79 : N .63,64,: 2 .2.4. a sphere, ths, N .64.
    lems paramet er representation of: M . 15 .
    prob .60
    : An.61 : C : J.73 : N . tangent to : LM . 13.
    58 : Q .10 . * intersection of three : 5661, J.73 :
    analogy with conics : Me.72 . TN .69.
    anharmonic section of : G .12. loci from : A .27.
    bifocal chords of : CD .5. normals of: 5629 – 32: C.78: J.73,83:
    * central equations : 5599 – 5672. N .63,78 : Q .8.
    central sections: 5650 : area, 5650 : oblique coordinates : N .82.
    axes, 5651. polar plane of : 5681 – 8 .
    locus of focus : N .66 . Plücher' s method : L .38 .
    non -central sections : 5654. of revolution : 1 .72,81 ; through 5
    *

    centre : N .75 : area, 5655. points, 66 and 79.


    *

    centre, coordinates of: 5690 : A .16 . self-reciprocal: M .25 : Z .22.


    *

    circular sections: 5596 , 5601, 6 ,'19 : sections of: J .74 .


    *

    C .43 : CD . 1 : CM .1 : E .30 : J.47, similarity of two : CD.8 .


    65,71,85 : N .51. with a “ Symptosen -axe ” : A .60,61.
    common enveloping cone : G .6 . system of : Q . 15 ; reduc, and transf.,
    condition for a cone : 5699 L .74 : Z .6 .
    conjugate diameters: 5637.: N .42,61 ; , * tangents : 5677 : G .12.
    * *

    parallel, G .l ; rectangular sys- / * tangent planes : 5626 , - 78 : CM . 1 :


    tem , L .58. cnJ.42 : LM .11 : N .46 .
    INDEX. 917

    Quadric surfaces - (continued ): Quantitative function , transf. of : CD . 3 .


    locus of intersect. of three : LM .15. Quartic equations: see “ Biquadratic ."
    through 9 points or under 9 condi Quartic curves: An.76 ,79 : At.52 : C .37,
    tions : A .17 : CD .4 , : J.24 ,62,68 : ths64,65 ,77 ,98 : CD .5 : G .14, 16 :
    L .58 ,59 : Q .8 : Z .25 . J.59 : M .th1,4 ,7,12 : prN .56 .
    under 8 conditions: C .62. aU + 6BH = 0 : M .1.
    through 12 points : G .17,22. and Abel's integrals : M .11.
    through a twisted quintic curve : E . binodal,mechanical cn . of : E .18 .
    38 . characteristics of a system : C .75 .
    transformation of two, by linear sub chord of contact, eq . : M .17 .
    stitution into two others,in which classification by inf. branches : L .36 .
    the squares only of the variables with cuspidal conics : M .19.
    remain : J.12 . with 3 cusps of 1stkind : An.52 .
    principalaxes : see “ Axes." degenerate forms: LM .2 .
    principal planes of: L .36 : N .67,71 : developable reciprocated : Q .7.
    Z :24 . with a double line : A .2 : 0 .75 .
    and their umbilics : C .54. with a double point : M .19 ; two, C .97. ;
    volume of segment: A .27 ; of oblique three, Q .18 ; three of inflexion,
    frustum , E . 19. N .78 .
    Quadricuspidals : L .70 . with double tangents : J.66 .
    Quadrilateral: A .6,48 : Me.66 : thG .5 : and elliptic functions : J.57,59.
    prsN .43. of 1st kind and intersections with a
    area : N .48 ; as a determinant, N .74 . quadric : An.69 .
    between two tangents to a conic and generation of C .45 : J.44 ; 3rd class,
    the radii to the points of contact : J.66 and Z .18 .
    A .532. 16 inflexion points of 1st species of :
    bisectors, a property : N .75 . tr2 .28. : E .32.
    and circle , geo.: 733 : Q .5. and in -pentagon , th : M .13 .
    * complete : 4652 : A .24 ,69 : At.63 ;mid oblate : C .74 .
    points of 3 diagonals, collinear, parameter representation of : M .13.
    Me.73 . penultimate : Me.72.
    and conic : 4697 : N .75,76. with quadruple foci: Q .18 .
    and in -conic : tg .c 4907 : Q .11. rectification : C .87 .
    cn . with given sides and equal diags. : and residual points : E .342.
    A .5 . and secants : M .12 .
    convex : A .66 ; area, Q .19. singularities of : L .75 : M .14.
    Desargues' theorem : Z .24 . synthetic treatment of: Z .23.
    plane and spherical: A .2 : Z .6 . through which one quadric surface
    and quadrangle : A .1. only can pass : An.61.
    right-angled : A .2,3. trinodal: thE.30 .
    with sines of angles in given ratio : unicursal twisted : LM . 14 .
    A .2 . Quartic surfaces : A . 12 : C .70 : G .11,12 :
    sum of sqs.of sides : th 924.
    th . extended to 3 dimensions : J.56 .
    LM .3,19 : M .1,7,13,18,20 : M0.66 ,
    72 : N .70 ,873 : Q . 10 , 11.
    Quadruplane : LM .14 . containing a series of conics : J.64.
    * Quantics : 1620 : C .47 : J.56 : LM .6 : with a cusp at infinity : LM .14.
    N .48 : Cayley P . 54,56,58,59,61, with double conic : A .2 : M .1,2,4 :
    67,71 and 78 : Pr.72,8,9, 11,15,17 , Mo.68.
    18,23 : thsCD.6 : J.53 : 1 .53 : with eq. Sym . det. = 0 : Q.14 .
    Q .14 ; Cauchy, Pr.42. generated by motion of a conic : J.61.
    derivatives,relation between 1st and Hessian of : Q .15.
    2nd : E .39. and 2 intersecting right lines : M .3.
    derivation from another by linear with 12 nodes : Q .14 .
    substitution : C .42. with 16 nodes : J.65,73,83,84,85 ,86 ,87,
    of differentials : J.70 ,71 : Mo.69. 88 : M0.64 ; principal tangent
    index symbol of: CD .5 . curves of, M .23 and Mo.64.
    integration of a rational : C .97. Steiner 's : C .86 : J.642.
    in linear factors: C .50 : L .ths61 : Q .6 . with a tacnode at infinity at which
    transf. of ax2 + by? + c%2 + dwa,by linear the line at infinity is a multiple
    subst. : J.45 . tangent: LM . 13.
    918 INDEX.
    Quartic surfaces — (continued) : Radii of curvature of a surface : A . 11,
    with triple points : M .24. 55 : Q .12 : Z .8.
    Quaternions : C .86 ,98, : CD .4 : G .20 : 1 principal ones: L .47,82 : M .3 : Me.80 :
    M .11,22 : Me.629,64,81 : P .52 : Q . N .55.
    6 : TE .27,28 : TI. 21. * Radii of curvature of a surface : 5795 –
    ap. to linear complexes and congru 5817 : A .11,55 : Q .12 : Z .8.
    ences : Me.83. ellipsoid : 5831.
    ap. to tangent of parabola : AJ.1g. flexible surface : L .482.
    elimination of aßy from the conditions * principal: 5814 – 6 : L .47,82 : M .3 :
    of integrability of Suadp, & c.: Me.80 : N .55 .
    TE .27. constant : Me.64 .
    equations : C .98 ; linear, 99,: of sur . for an ellipsoid : 5832.
    faces, Joachimstahl's method , equal and of constant sign : C .41 :
    E .43. JP.21 : L .462,50.
    vp op = 0 : TE. 28 : 90 - 90 = 0, Euler 's theorem : 5806 .
    Me.852. one a function of the other : An.65 :
    finite groups : AJ.4. C .84 : J.62.
    f. for quantification of curves, sur product constant : An.57 .
    faces, and solids : AJ.2 . reciprocal of product : An.52.
    geometry of : CD .9. sum constant : An.65 .
    integration ths : Me.85 . sum = twice the normal : C .42.
    transformations: Man.82. * Radius of curvature of a curve : 5134 :
    * Quetelet's curve : 5249. A .cn4,9 ,31,33 : CD .7 : J .2,45 : ths
    Quintic curves : cnM .25 . M .17 : N .62,74 : 9 .c and t.c Q .12 :
    Quintic equations: AJ.62,7 : An .65,682 : Z .3 .
    C .462,48 ,50 ,61,629,73,802,85 : J.59 : absolute : CM .1.
    M .139,14 ,15 : P .64 : Pr.11 : Q.3 : circular : 5736 ; ang. deviation , 5746.
    TI.19 : 2 .4. of conics : 1259 : A .9 : CM .1 : J.30 :
    auxiliary eq. of : Man .15, : P.61 : Q .3. L .36 : Me.66 : Mél.2 : N .45,682.
    condition of transformability into a at a cusp or inflexion point : N .54.
    recurrent form : E .35 .
    in dipolar coordinates : Me.81.
    irreducible : AJ.7 : J.34. of evolutes in succession : N .63.
    of gauche curves : N .66 .
    functions of difference of roots : An. of a geodesic : L .44 : on an ellipsoid ,
    69. An.51.
    reduction of : Q .6. and normal in constant ratio : N .44 .
    resolvents of : 0 .632. of normal section of • (x , y , z) = ( :
    whose roots are functions of a varia 5817 .
    ble : Q .5 . of a parabola : 1261, 4542.
    solution of : An.79 : J.59,87 : N .42 : Q . of polar curves : A .51 : CM .2 .
    2 , 18 . of polar reciprocal : N .67 .
    Descartes'method : A .27 . of projection of a curve : N .61 ; of
    Malfatti's : An.63. contour of orthogonal projection
    in the form of a symmetricaldeter of a surface, C .78.
    minant of four lines : An .70 . of a roulette : 5235 : N .73.
    Quintic surfaces : LM .3.
    having a quintic curve : An.76 .
    transf. of properties by polar recipro
    cals : L .66 .
    Quintics, resolution of: Q .4 . of tortuous curves : Mém .10.
    Radial curves : LM .1 : of ellipse, Q .18 . of circum -sphere of a tetrahedron in
    of conics, catenary , lemniscate, & c. ; terms of the edges : N .74 .
    E .24 . * Radius of gyration : 5904.
    Radiants and diameters of a conic : C. Radii of two circles which touch three
    26 . touching two and two : A .55 .
    *Radical axis (see also “ Coaxal Cir Radius vector of conic : J.30 : N .47 .
    cles " ): 958, 987- 99,4161: LM . Ramifications : E .30,33,40,pr 37,th27.
    2 : Me.66 . sol. by a diophantine eq.: M0.82.
    of symm . circle of a triangle : A .63. Randintegral : J.71.
    of two conics : Q .15 . * Ratio and proportion : 68 ; compound,
    * Radical centre : 997. 74 .
    * Radical plane : 5585. of ap : b : geo.cnN.44 .
    INDEX. 919

    Ratio and Proportion — (continued) : Reciprocal— (continued ):


    of differences of geo .quantities : C .40. transformation, geo.: L .71.
    * limits of: 753. triangle : th Q .1 ; and tetrahedron ,
    * of segments of lines and triangles, Q .1.
    geo. : 929 – 32. Reciprocants : LM .172.
    * of two distances, geo. : 926 — 8 . Reciprocity : anal., A.7 : geo., CD.3.
    Rational : - derivation , cubic curve : * Reciprocity law : 3446 : AJ.1,th 2 : C .
    AJ.33.
    divisor s of 2nd and 3rd degrees : N . 90, : J.289,39 : LM .2 : M0.58 : d.e,
    45 . 3446 and A .33.
    functions, development of : AJ.5. in cubics : M .12.
    Rationalisation of: — alg . fractions : history of : Mo.75 .
    A .13,33,35 . for power residues : C .84 ; quadratic,
    alg . equations: A .13 : CD.8 : J.14 : C .24 ,88 : J.47 : L .47 , and Mo.80 ,
    P . 14 : TI.6 . 84,85 ; cubic, in complex numbers
    alg . functions : A .69. from the cube roots of unity , J.
    .35
    a series of surds (Fermat's pr.) : A . 27, 28 .
    Rational functions : - of n elements : quadratic F2 system of 8th degree :
    M .14. J.82.
    infinite form systems in : M .18. supplementary theorem to : J.44,56.
    Ray systems (see also “ Congruences " ): *Rectangle : M . I. of, 6015.
    J.57 : 9 .60,74 : Me.66 : N .60 ,613, | * Rectangular hyperbola : 4392 : Me.62,
    622: Z .16 . 66 ,72 : N .42,65.
    of 1st order and class and linear pen | * Rectification of curves : 5196 : A .26 :
    cils : J.67,69 : 2 .20 . Ac.5 : An.69 : CR . 95 : CD .9 : G .
    1st and 2nd order : M0.65 : M .15 ,17. 11 : J.14 : L .47 : N .ths53,64.
    2nd order and class : J.92 . approximate : M .4 : Mél.4 .
    3rd order and 2nd class : J.91. by circulararcs : C.77,85 : L.50 :Mém .
    6th order and 2nd class : J.93. 30%.
    2nd class and 16 nodal quartics : J.86 : by elliptic arcs or functions: J.79:
    Mo.64 . Mém .30 .
    complex of 2nd degree and system of by Poncelet's theorem : C .94.
    2 surfaces : M .21. mechanical: Z .16 .
    and refraction theory : Q.142: TI.15 on a surface : Mém .22.
    - 17 . * Rectifying : - developable : 5727.
    forming a group of tangents to a sur * line : 5726 , 451 : N .73 .
    face : Z .18 . * plane : 5726.
    infinite geometry of : 2 .17. * surface : 5730.
    * Reciprocal polars : 4844, 5704 : A .36 : Reflexion : — from a revolving line :
    gzE.24 : J.77 : LM .2 : N .48,49 : TE.28.
    num .fG .21. from plane surfaces : A .60.
    *Reciprocal: of a circle : 4845. from quadric surfaces : J .35 .
    * cones : 5664, 5670. Refraction curve : A .51.
    of a conic : 4866 – 8 .
    of a quadric surface : 5717 — 8 . Règle à calcul: C .58 : N .69.
    radii: M .13. * Regular polygons: 746 : A .21,cn 24,39 :
    relations : J.48,79,90 : M .19,20. L .38 : M .cn6 ,13 : N .42,44 ,47 .
    spiral : 5302. convex : Me.74 .
    *

    surfaces : 5704 — 19: J.79 : M .4,10 : P . eqs. of and division into eqs. of lower
    *

    69. • degrees, tr. : A .46.


    curvature of: L .77. in and circum : N .45 : Q .2.
    degree of : CD.2 : TI.23. in space : Me.75 .
    of Monge : C .42. spherical : N .60,67.
    of quadrics : gn 5705 ; central, 5706. star : J.65 : Me.74 : N .49 .
    of surface of centres o a quadric : funicular : N .49 .
    Q .13 . 5-gon : M .83.
    of the same degree as their primi. 7-gon : A .17 .
    tives : Mo.78 . 7-gon and 13-gon : M .6.
    theorems on conics and quadrics : L . 8- gon : A .6 .
    61. 12-gon : complete, C .95.
    920 INDEX.

    Regular Polygons — (continued ) : Riemann's surface - (continued ):


    17 -gon : 4 .6 : J.cn24,75 : N .74 ; and irrationality of : M .17.
    division of the circle, A .42 . Riemann's function : A .68 : J.83: M .21.
    eqs. for sides and diagonals : A .40. ext. to hyper- geo.-functions of 2 vari
    * Regular polyhedrons : 907 : A . 119,47 : ables : C .9in .
    Pr.34 : Me.66 : Q .15 . A -formula , gz : Ac.3. .
    * relation of angles : 909: Me.74 . * Right-angled triangles : 718 : prs A .2 .
    volumes by determinants : A .57 . with commensurable sides : E .33.
    Related functions : M .25 . Right cone : Me.72,73,75 ,76 .
    Relationship problems : E .35,382. Right line : A .49,57 : fQ . 15 : t.c Me.62.
    Relative motion : N .66 . and circle : ths N .56 .
    Rents : A .40. coordinates of: G .10.
    Representative functions: M .18 . and conic : Q .7 ; cn . for points of sec
    Representative notation : Q .6 . tion , A .59,66 : N .85.
    Reproduction of forms: C .97.
    Reptation : N .54. quadratic
    4319 .
    for abscissæ of the points:
    Residues : A . 26 : C .122,13,32,ap32 ,41,44 ,
    49: CM .1 : J.252,31,89 : L .38 : Mél. tg.e of the points : 4903.
    4 : N .462,Mém 70. condition of touching : 4315,4323,
    ap. to infinite products : C .17. t.c 5017.
    ap . to integrals whose derivatives in drawn from a 'y' across a conic : quad
    volve the roots of alg . eqs. : C .23. ratic for the segments in an el
    ap. to reciprocity law of two primes lipse, 4314 ; parabola , 4221; gen .
    and asymptotes : C .76 . eq ., 4494 ; method, 4134.
    of complex numbers : M0.80. * joining two points, coordinates of
    primes of 5th , 8th , and 12th powers : point dividing the distance in a
    J. 19. given ratio : 4032, t.c 4603 , 5507.
    biquadratic : C .64 : J.28 ,39 : L .67. * tg. eq . of the point: 4879.
    cubic : A .43,63 : C .79 : J.28,32 : L .76 . * joining two points and crossing a
    quadratic : Ac.1 : J.28,71 : Q .1. conic : quadratic for ratio of seg .
    ext. of Gauss's criticisin : M0.76 . ments in an ellipse, 4310 ; para
    of primes, also non -residues : L .42 : bola, 4214 ; gn . eq., 4487 , t.c 4678 ;
    Mél.4 . method , 4131.
    and partition of numbers : J.612. constants, relations between : sd .
    quintic : C .76 : 2 .27. 5515 .
    septic : 0 .80.
    of Sexp.(9exp.9) by division by coordinates of, relation between the:
    primes : A . 35 . 4897 .
    Residuation in a cubic curve: Me.74 . * and curve : 4131 – 5 : ths. in which
    Resultant alg. : M .16 : ext C .583. pairs ofsegments have a constant
    and discriminants and product of dif length , C .836 ; a constant product,
    ferences of roots of eqs., relation : C .822,83, ; a constant ratio, C .83;
    Me.80. ths. in which systems of 3 seg.
    of two equations: J. 30,50,53,54 : M .3 : ments have a constant product,
    P .57 ,68 . C .837.
    of two integral functions: Z .17. crystallography : A .34.
    of n equations : An .56 . equations of : 4052 - 66 , p .c 4107 , t.c
    of covariants : M .4 . N 4605 – 8 ; sd 5523, q.c5541.
    of 3 ternary quadrics : J.572 : N .69. geometry of : A .64 : thsJ.8.
    Reversible symbolic factors: Q .9. * at infinity : 4612 – 4, tg.c 4898 .
    Reversion of angles : LM .6 . cond . for touching a curve : 4900.
    Rhizic curves : Q .11.
    Rhombus : quadrisection by two rect
    pencilsJ.67.of : C .70 : L.72 ; quadruple,
    angular lines : Mém .ii. and plane : prs CD .1 and CM .2 : t.c
    circumscribing an equil. triangle : A . and q .c Q .5 .
    45 . pole of: t. 4671 : tg .e 4674 .
    Riemann's surface: LM.8: M .6,18 ,:
    thsZ .12.
    and quadric : 5676 ; harmonic divi.
    sion , 5687.
    of 3rd species : M .17 . and quadric of revolution : N .82.
    new kind of : M .7,10. six coordinates of: CP.11.
    INDEX. 921

    Right line - (continued ): *Roots ofan equation (see also “ Equa


    system of : At.68 : G .9,10,16 . tions ') : 50,402 : A .14 : CM .2 :
    of 1st degree, G .6 ; of 2nd , G .7. CP.8 : E .36 : J.20,31 : N .42,56 :
    in space : G .113,12 : L .46 . P .1798,37 ,64 : Q .1,5,6 .
    and65.planes, geo . of 2nd kind : At. of a biquadratic, cn : N .44.
    . by parab. and circle : N .87.
    * three,condition of intersection : 4097 : commensurable : 502 : N .45,th57 .

    EX
    t.c 4617 . limits to the number : N .59.
    three points lying on , cond. : 4036 , common : 462 : C .80,88 : N .55,69.
    t.c 4615 . as continued fractions : CM .3.
    through a point: 4073,4088 — 9,4099,
    t.c4608.
    continuity of : N .76 .
    in a converging series : C .23,38 .
    condition : 4101. of cubic : L .44 : N .42.
    and perp . or paral. to a given line : of cubic
    55 .
    and biquadratic : An.55 : L .
    sd5538 - 9 .
    through two points : 4083, sd5537 ; as definite integrals : J.2.
    t.c4616 , 4789 ; p .c4109 : on a y" -- " -1 - 1 = 0 : Me.81 : P .64 .
    conic, equation of, ellipse 4324 , as determinants of the coefficients :
    parab . 4225. A .59,61.
    *

    through four lines in space : A .1 : discrimination of : 409 : A .46 .


    CM .3 : Gergonne's pr. J.2 and equal: 432 _ 47 : CD.5 : E .33 : Mél.1 :
    N .17. P .1782 : Q .9,18 .
    touching a surface : condition , 5786 . with equal differences : G .152.
    existence of : A . 15 : CD.2 : CP.10 :
    quadric : 5703. E.36 : G .2 : J.5,44,88 : LM .1 : Q .
    planes or points through or on given 11 : TI.26 .
    points, lines , or planes, number expanded in power series : J.43 .
    of such : 2 .6 . of the form atıb + uct : N . 15 .
    two : — angle between them : 4112 ; forms for quadrics, cubics, and quar.
    sd5520, 5553 : CP.2 : N .66 . tics : 2 .24 .
    bisector of theangle : 4113 , sd 5540, functions of the roots of another eq.:
    q.c5543. 425,430 ; products in pairs, Q .13.
    cond. of parallelism : 4076 ; t.c4618 ; squares of differences : 541 : ap,N .
    sd5531. 50 : E .10.
    cond. of perp.: 4078 ; t.c4620 : sd as functions of a variable parameter :
    5532. C . 30 .
    cond. of intersecting on a conic , gn . | functions of : similar, L .54 ; relation
    eq : 4962. to coefficients, TE .28 .
    cond. of either touching the conic : geo. cn of : JP.10.'
    4964. in g.p : N .89.
    cond, of intersection : sd5533. in a given ratio : J.10 .
    coordinates : 4090 , t.c4611. * imaginary :408 : geo.cn ,A .15,45 : C .24,
    86 — 88 : JP.11 : L .50 : N .46 ,47,
    shortest distance : sd5534 – 6 . 682 : approx. N .45,53 : Q .9.
    drawn to the points of section of a between given limits : A .21 : L .44.
    right line and conic, eq. of : A .69. Newton 's rule : Me.80 : N .67 : Pr.
    through origin , eq. of: 4111. 13 .
    under given conditions : C .73,74 : Newton -Fourier role : Q . 16 .
    under four, C .68 . Newton -Sylvester rule : 530 : C .
    *Right solid : M . I. of, 6018 : thA.34. 994 : LM .1 : Me.66 : Pr. 14 : Q .9 :
    Rodrique's th . : Me.80 ,842. TI. 24 .
    Rolling cones : L .53. * incommensurable : 506 (see “ Sturm 's
    Rolling and sliding solids : geo thsC .46 . th.")
    Rosettes : N .48. infinite : N .412,45.
    Rotation : CM .3 : LM .32: infinitesimal, in infinite series : A .69 .
    C .78. integral , by Newton 's method of
    of system of lines drawn through divisors : 459 .
    points on a directrix , modulus least : M .9 : TE .28.
    of : C .21. * limits of : 448 : C .58 ,60 ,93 : geo CP .
    Roots of algebraic fractions : N .46 . 12 : N .43,45,592,72,802,81.
    6 B
    922 INDEX .

    Rootsnumber
    of an equation - (continued) :
    between given limits
    Roots of numbers — ( continued) :
    : A.1 : to 25 decimal places : Me.77.
    G .9 : J.52 : L .40. cube root : Horner's method : 37 :
    theofeq. containing only odd powers
    a : N .63.
    A .67.
    of 2 to 28 decimal places : Me.76,78.
    Rolle's th . : 454: AJ.4 : N .44 : ext and sq . root, limit of error : N .48 .
    L .64. fourth root : A .30.
    Rolle, Fourier, and Descartes: A .1. nth root as a fraction : A .46 .
    number satisfying a given condition : * Roots of unity : 475 – 81 : 0 .38 : J .40 :
    C .40 . L .38,54,59 : Me.75 : N .43, : T'E.
    product of differences : Me.80: P .61. 21 : Z .22.
    of a quadratic : 50 — 3. cubic roots, alg . and geo. deductions:
    of a quartic and of a Hessian , rela C .84.
    tion : E .34 . function theory : Z .22.
    of quintics : C.59,60 : LM .14 : T1.18.
    rationalization of : P .1798 ,14 .
    23 roots, composition of number 47 :
    J.55,56 .
    real : A .36,58 : 0 .61: J.50 : JP .10 : * Roulettes: 5229 : Ac.63 : 0 .70 : CP.7 :
    N .50 : P .57. J.65 : L .80,81: N .56 : TE .16 : Z .28.
    of a cubic : N .72 : 2 .2. areas of, and Steiner's transf. : E .35.
    Fourier's th . : N .44 . generated by a circle rolling on a
    developed in a series : L .78 : N .56 . circle : JP .21.
    limits of: J.1 : N .53,79. byline
    focus of ellipse rolling on a right
    series which give the number of: : A .48.
    Z .2 . by centre of curvature of rolling
    to find four : An.55. curve : L .59.
    rule of signs: 416 – 23 : A.34 : C .92, Ruled surfaces : An.68 : CD.8 : G .3:
    982,999 : N .43,46 ,47,67,69,79. J.8 : L .78 : N .61.
    separation of : A .28,70 : J.20 : N .68 , areas of parallel sections : 2 .20 .
    72,74 ,75 ,802. and guiding curve: A .18.
    by differences : N .54 . ofminimum area : L .42.
    for biquadratics : A .47. octic with 4 double conics : C .60 .
    for numerical: C .89,92 : G .6 . P . D . eq . of : Me.77 .
    simultaneous eqs.: C .5. quadric : Me.68.
    squares of differences : 541 : CM .1 : quartic : A .65 ; with 2 double lines,
    N .42,44 : Q .4 . A .65 .
    sums of powers : 534 : E .38 : thJ.9 : quintic : J.67.
    N .53,75 : gzMe.85 : Q .19. represented on a plane : C .853.
    in sums of rational functions of the of species, p = 0 : M .5.
    coefficients : Ac.6 . symm . tetrahedral : C .62.
    surd forms of : CM .3. torsal line : M .17.
    symmetrical functions of : 534: A .16 : transformation of : C .88 .
    AJ.1 : An.549,55 ,60 : C .44,45 : G . * Scales of notation : 342 : J. 1 : L .48,5ry,
    5 ,11 : J.19,54,81 : Me.81 : N .48, 10ry,20ry : Phil. Soc. of Glasgow ,
    50 ,55,66 ,84 : P .57 : Pr.8 : Q .4 : vol. 8.
    TI.25 . Screws : TI.25.
    do. of the common roots of two eqs. : Scrolls : A .53 : CD.7 : CP. 11 : J.20 ,67 :
    N .60 : Z .15 . M .8 : cubic, M .1 : P .63,64,69 :
    do. of differences of roots : C .98. Pr.12,13,16 : Z .cn28 .
    which are the binary products of the condensation of: LM .13.
    roots of two eqs. : An.79. cubic on a quadric surface : Me.85 .
    with a variable parameter : C . 12 . flexure and equilib . of : LM .12 .
    which satisfies a linear d.e of 2nd ruled : A .68 : 2 = mxy", A .55 .
    order : 0 .94 . tangent curves of : M . 12 .
    * Roots of numbers : 108 : A .17,26 ,35 : | * Sections of the cone : 1150.
    C .58 : E .36 : Me.75 : N .61,70. * Sectors and segments of conics and
    square root : 35 : C .93 : N .452,46 ,61, conicoids : 6019 - 6162 : G .1 :
    70. thsZ .1.
    as a continued fraction : 195 : A .6, Secular eq. has real roots : J.88 .
    12,49 : CM .22: L .47 : Me.85 : | * Self-conjugate triangle : 4755, 4967 :
    Mém .10 : TE .5 : Z .17. G .8 : N .67 : Q .5,10.
    INDEX. 923

    Self-conjugate triangle ( continued ) : 1 Series — ( continued ):


    * of 2 conics : Me.77 ; of 3 conics, 5025. a = nb + C (1 , 2) c & c. : J.31.
    and tetrahedron in conics and quad * Tr – m ( n - 1) + C ( 1, 2) ( – 2)" - ...:
    rics : Z .6 .
    Self - enveloping curves and surfaces : 285 ; p = n ,CM .1.
    Z .22. 1 + ax + a (a + 1).azt ...: J.37 ; with
    Self-reciprocal surface: Mo.78. 17ū " to(6 + 1)"?
    * Self-reciprocal triangle : 1020.
    Seminvariants : AJ.7 : E .th42,6 : Q . x = 1, J.2.
    19 - 21.
    criticaland Spencian functions: Q .4 ,6 . (a + nd)k : N.59.
    and symm . functions : AJ.6 .
    Septic equations : Mo.58. deductions from 22n
    * Series : (see also “ Summation " and
    “ Expansions ' ) : 125 – 9 , 149 –
    59, 248 – 95, 756 – 817, 1460, 1471
    ( 200e * = (a+1)"2"n!
    - 2 , 1500 – 73 , 2708 – 9 , 2743 – 60, LM .9.
    2852 – 64 , 2880 , 2911468 , 3781,
    3820 : A .4,52,9,14,18,23,60 : No.39, to. ... and 11 .+ 1 . 1 ... :
    472 : 0 .29,pr92 : CP.9 : G . 10 : J. 2940 – 4 : A .41 : LM .8 : Q .7 : with
    3 ,17 ,34,38,th53 : L .th56 ,81, :Me. n = 1,2...8 ; 2945 ; E . : 2,39 ; G .10 ;
    64 : N .59,th62,70 : Q.3 : 2.15,16 , N .79 ; Z .3. Note that by (2391),
    23.
    1 _ 3127_ 12 & c. _ flog x2
    Useful summations: 5
    care colo

    Jo 1 + x2 "
    onet+...= log : 156.
    + .

    zaten Mém .11:with x = 1,J.5.


    +
    Escle

    So -...= log (1+x): 155.


    col

    :4.61. 8 1 4.41
    1
    +

    +...= }log14 :157. 27 @ 18


    a — 57 + 51 - ...: E.44; with x = 1,
    + +
    +
    0olOMO

    = tan -1« : 791. J.5.


    91
    cC101
    10%.291

    1
    :

    =¢ -1: 150 ç (nn !-a1)x":


    " -1 -
    A.50.
    Color
    +
    +
    reen

    :
    GS

    - + -... = 1- 2 3( - 1)" (a— + 2n2n d


    )(1 — æjn : Q.6.
    +
    1

    2 :E.36.
    +
    +
    :

    + * -...= sinæ: 764. E amm( ): 2709.


    +
    1

    0 _ 1

    to the one +... +04: 1)"nn ++ 111!': A .26.


    3 (- P
    x2
    2: _ x44: 16I
    xb __- .. = 1 -cosa : 765. $ 1 . 3 ... 2n –11
    n ! 2" (2n + 1)5: L.60.
    *

    1P + 2° + ... + P : 2939 : A.65: Me.78. 1 + (9 — 1) (98 – 1) . I


    *

    p = 1,2,3, or 4 : 276 : A .64 : E .34. (9 – 1) ( 99 - 1)


    *

    1P 2P + 3P — ... & 1P - 3P + 59 – ... : (q° - 1)( 2+) - 1)(q® — 1)(98+1 – 1).


    J.7 . (9 - 1) (99 - 1) ( - 1) (qu+1 - 1)
    E nøde" : A .27. + ... : J.32,70.
    (a + n )ox" : N.56.
    3 (an + bræº)ak-n : Z . 15 . - { (^ +1) (473) + ...}:G.11.
    1 _ n _ 3 _ (n — 4) (n — 5 ) _ .
    2 T 2.3 anm + ain " - ' + ... tam '
    924 INDEX .

    Series — (continued ): Series — (continued ):


    sepriat binomial (see “ Binomial theorem " ) :
    ¿R mata ' B a pos. integer < a , analogous series : E . 35 : J.32 : N .
    na + B ' 82 ; with inverse binomial co
    K , = the general term of some efficients, Me.80 .
    recurring series : C . 86 . coefficients independently deter .
    mined : A .18 .
    2 .: A.34. whose coefficients are the sums of
    divisors of the exponents, sq . of
    this series : Me.85 .
    " E ( ): .50. combination : A . 26 .
    convergent: 239 : A .2 ,6 ,8 , 14,26 ,41,67,
    / sin
    COS nk: J.54 ;with k = 2, 69 : No.44 : 0 .10 ,11 ,283,40,43,: J.
    2,32,11,13, 16 ,22,42,45,76 : L .39 —
    2960 — 1, J.8. 42 : M .10,20 – 22 : Me.64 : N .45,
    E sin3(2n + 1) • 1 (2n + 1)4: E.39. 46 ,67,69,70, : P .87 : Z .10 , 11.
    and of d .i with a periodic factor :
    2 (803n°)**f*/nt and
    со L .53.
    power-series : A . 25.
    >(cons»$)*, 18: L.73. representing integrals of d . e : C .402.
    representing functions : M .5,22.
    in Kepler's problems : Ac. 1799.
    * *(auSib»$)/n:2922—3. multiplication of : M .24 .
    and products, condition : M .22.
    sin n2 271P : L.40.
    cos " whose terms are continuous func
    tions of the same variable : C .36 .
    sn sin no:
    no . with constant ultimate differences :
    a + n2 2962 : M .5. Pr.5.2.
    converted into continued fractions :
    E A, cos” o cisno: Z .1. J.32 ,33 : Mém .9 ; into products
    ofan infinite no. of factors, J.12 :
    * 3 SUB (a+ nB): 800: Q.3; L .57 : N .47.
    in cosines of multiple angles : C.44 :
    Ec" (a+ nß), 783 ; Mém . 15 .
    and definite integrals : L .82 : Man.
    2 cm sin (a+nB),788. 46 .
    derived : A .22 ; from tan - 10, A .16 .
    developed in elliptic integrals of Ist
    3f(n)sog ne :J.42: L.52.
    COS and 2nd kind : An.69.
    difference : 264 : A .23,24.
    2 H(m,n) "sign°:3.41. differential transf. and reversal of:
    Pr. 7.
    2 tan : A.44. and differentiations: A .10 : J.36 .
    =f(næ):L.51. 2Anº (n)æn,J.25,28. Dirichlet's f. for 3 0.21: L.
    46 .
    2"} (x + * ): A.22. discontinuous : CP.6 : L .54 : Me.78 ,
    82 : N .85 .
    from [ æ(1– 2)"de: A.47. divergent : A .64 : No.68 : C .17. 20 :
    ČP.8 ,10 : J.11,13,41 : M . 10 : Z .10.
    from ( cos2x log' 11 -+ sisinnex dada.: c.
    G . 10 . division of: AJ.5 .
    vo double : C .63.
    of Abel : C.93: N.85.
    application : thA .48 ; to arith, G .7.
    doubly infinite : CD.6 : M .24.
    ext. of by any parameter : A .48 .
    in a.p : see “ Arithmetical progres. factorial : 268 : Mém .20 : N .67 : TE.
    sion .” 20 .
    analogous series : N .69. of fractions : L.40.
    a .p and g.p combined : A .9 . Fourier's : A .39 : C .91,92,96 : CM .2 :
    with Bernoulli's nos. : An.53 ; and bi. Z .27 .
    nomial coefficients, A .23. of Gauss and Heine: C .73 : G .9.
    INDEX. 925

    Series - (continued ) : Series — (continued) :


    Gregory 's : 791. self-repeating : CP.9.
    harmonic periodic : J.23,25. of spherical functions : An .75 .
    of Hermite, a th . : E .29 . of Sterling, for transformation : J.59.
    from infinite products : Me.733. with terms alternately positive and
    integration of infinite : A .3 . negative : C .64 .
    irrationality of some : J.37. whose terms are the coefficients of
    involving two angles : L .74. the same power of a single vari.
    Klein 's higher : An .71. able in a multiple integral: C .20 .
    of Lagrange : 0 .23,343,522 : L .57 : N . in theory ofnumbers : C .892.
    76 ,8285 : Q .2 ; remainder, C .53. transformation of : C .59 : J.7 ,9 : into
    an analogous series : C.99. a continued fraction , Mém .20,
    of Lambert ; & n : A. 10 : An .68 : 2 .7 ; of sf(se,t) dt and others,
    C .13.
    J.9 : Z .6. in a triangle problem : A .64.
    Laplace's (d.c) : C .68. trigonometrical ( see above) : A.63:
    of Laplace's functions : EYn, C .882; Ac.2 : C .95,97 : M .4 – 6 ,16 ,17,22,
    En Yn, C .44 . 24 : J.71,722: representing an
    in Legendre's function X , : An.76 : arbitrary function between given
    0 .44 . limits, J.4 ; conversion in mul
    of Leibnitz : J.89. tiples of arc, L .51 ; symbolic
    limits of : A .20 : Me.76 ; remainders, transf. of, Q .3 .
    C .34 ; by the method of means, triple : G .9.
    J.13. two infinite,multiplication rule : J.79.
    from log (1 + x), (1 + x)" and it by in * Seven -point circle : 4754c .
    termitting terms in the expan Seven planes problem : N .56 .
    sions : A .21. Sextactic points of plane curves : Pr.
    modular : C .19 . 13, 14 .
    multiple : C. 192; “ regulateur " of, Sextic curves : ax: + by + c = 0, Q.15 ;
    C .44 . mech .cn , LM .2.
    neutral : CP.11. bicursal: LM .7 .
    obtained by inversion from Taylor's and ellipse, pr : J.33.
    series : Mém .11. Sextic developable : Q .7,9.
    of odd numbers : A .64. Sextic equation : C .64 : M .20.
    a paradox : Me.72 . irreducible : J.37 .
    periodic, critical values of : CP.8. solution when the roots are connected
    of polynomials : 0 .96 . by (a - B ) (6 - 7) (c - a) +
    of posterns : G .6 . (a - 6) (B - c ) ( y - a ) = 0 : J.41.
    of powers (see also “ Numbers ' ) : Sextic torse : An.692.
    277 : C .87 : G .2 : cubes, L .64 and Sextinvariant to a quartic and quart
    65 : M .23 : Mo.78 : Q .8 : Z .1. invariant to a sextic : AJ.1.
    approximate fractions : J .90 . * Shortest distance : — between two
    of a binomial : Mém .13. lines : 5534 : A .46 : G .5 : N .49,66 .
    in a convergent cycle, constants between two points on a sphere : A .
    in : M .25. 14 : N .14 ,67,68.
    like numbers : N .71,77. from the centre of a surface : A .63.
    or multiples of 3 : A .272. of a point from a line or plane : N .44 .
    of terms in ar.p : L .46 . Shortest line on a surface : A .23,37,64 ;
    products of contiguous terms of : in spheroidal trigonometry , A .40 .
    Mém .18. Signs: CP.2,11 : J.12 : Me.73; ( = ),
    of reciprocals : Q .8. Me.75 ; ( + ), CD .6,7 : Me.85 : N .
    recurring : 251 : doubly, An,57 : J.33, 48,49.
    38 : Me.66 : Mém .24 ,26 : N .84 : of Similarity :- of curves and solids:
    circles and spheres, N .62 : f, Z .14 . A . 13 .
    represented by rational fractional Similarly varying figures : LM .16 .
    functions : J.30. Simson line of a triangle : E .29.
    reversion of: 551 : J.52,54 : LM .2 : * Simpson 's f. in areas: 2992 : C .78 .
    TI.7. Sines of higher orders : C .914,92 ,; ap. .
    of Schwab : N .59. to d .e, C .903,91.
    926 INDEX .

    Sines, natural, limit of error : N .43z. Space theories - (continued ) :


    Sin 0 < 0 – 00 : geo.N.75. imaginary quantities : Z .23.
    loci (anal.) : C .24.
    Sin-?(x + iy): Q.15. planes (4-dimen .) : A .68.
    plane triangle : A .70.
    Sineand cosine: - extension ofmean . point groups : thsAc.7.
    ing : A .31: C .867. polars and alg . forms: J.84.
    * in factors : 807 : A.27 : J.27 : L.54. potential function : An.82 ,83 .
    of infinity : CP.8 : Me.71 : Q .11. projection ; M .19 ; 4 -dim . into 3-dim ..
    of multiple arcs (see also “ Expan AJ.2 .
    sions ” ) : CM .4 : TI.7. quadric,super lines of (5.dim .) : Q .12.
    of particular angles : 690 ; 30,6°,...to quaternions : CP.13.
    90 ', N .53. regular figures : AJ.3.
    sums of powers : An.l. reversion of a closed surface : AJ.1.
    representation by correlative figures :
    tables42, formation of : 688 : A .66 : N . C .819 .
    7.
    values near 0 and 90° : G .9. simplicissimum of nth order : E .44.
    * of (a + b) : 627 ; A .6,21,36 .
    screws, theory in elliptic space : LM .
    15 , 16 .
    Six-point circle of a triangle : Me.82, 21 coordinates of: LM .10.
    83 . Sphere : geo, C .92 : ths and prs M .4 :
    Six points on a plane or sphere : LM .2 . q .c Me.62.
    Skew surfaces : see “ Scrolls.” and circle : geo , A .57.
    Sliding rule : LM .6 . cn. from 4 conditions : JP .9.
    *Small quantities of second order : 1410. cutting 4 spheres at given angles :
    Smith 's Prize questions, solutions: Me. An .51 : N .83.
    71,723, - 4, 77. cutting a sphere orthogonally and
    Solid angle: A .42; section of, No.19. touching a quadric, locus of cen
    Solid harmonics : Me.80 . tre : TI.26 .
    Solid of revolution : A .60,67. diameters, no. of all imaginable: A. 24 .
    between two ellipsoids : A .2. equation of : 5582 .
    * cubature and quadrature of: 5877 — 5 points of : J.23 : N .84 .
    80 : A .68 : N .42. illumination of: Z .27.
    Space homology : G .20 . kinematics on a : LM .12.
    Space theories : An.70 : LM .14 : P.70 : sector of (eccentric ) : A .65 .
    Z .17, 18. small circle of : Me.85 .
    absolutely real space : G .6 . touching an equal sphere : E .31, 32 ;
    continuous manifoldness of two di as many as possible, A .56 .
    mensions : LM .8 . 4 spheres, pr. : L .46 .
    space of constant curvature : An.69, 4 touching a 5th : At.19.
    73 : J .86 : M . 12. 8 touching 4 planes : E . 19 : N .50 .
    Plücker's “ New geometry of ” : G .8 : 16 touching 4 spheres : J.37 : JP. 10 :
    L .66 : P .11 : 2 .11,12. Me.cn 82 : Mém .10 : N .44,47,65 ,
    Grassmann's “ Ausdehnungslehre " : 66 ,84 : Z .142.
    AJ.1 : CP.13 : M .7,12 : 2 .24 ; ap . * volume, & c. of segment and zone :
    to mechanics, M .12. 6050 : A .3,32,39 : An.57 : P .1.
    non Euclidean or n -dimensional: A . * Spherical : - areas : 902.
    6 ,29,58 : ths64 : AJ.4,5 : An.71 : catenaries : J.33.
    C .75 : G .6 ,10 ,12 ,23 : M .4,5 , -- 7 : class cubics with double foci and
    Me.ths68 ,72 : Pr.37. cyclic arcs : Q .15 .
    3-dim ., J.83; 4-dim ., J.83, M .24 ; 6 . conics : thQ.3 ; and quadrangle, Q.13 ;
    dim ., G .12. homofocal, L .60.
    angles (4-dimen .): A .69.
    areas and volumes : A .69 : CD.7 .
    coordinates : CD.1: CM .1,ap2 ; ho
    mogeneous, G .6 .
    bibliography of : AJ.1,2. curvature : 5728 ,'400'47 : thE.34.
    circle : G .12 ,16 ,18. carves : A . 35,36 : Mém . 10 .
    conics : AJ.5. of 3rd class with 3 single foci : Q.
    curves : C .79 : M .18 . 17.
    Feuerbach 's points : G . 16 . of 4th class with quadruple foci :
    hyperboloid : 2 . 13 . Q .18 .
    INDEX. 927

    * Spherical: - curves — (continued) : | Spherical :- triangle (continued ):


    of 4th order : J.43. and plane triangle : A .l ; of the
    with elliptic function coordinates : chords, A . 33 : An.54 : Z .1.
    J.93. right angled : 881 : A .51 ; solution
    equidistant : An. 71 : J.25 . by a pentagon , A .11.
    and polars : No.63. of very small sides : N .62.
    rectification of: An.54.
    ellipse : t.cQ .8 .
    two,A .2.relations of sides and angles :
    quadrature , & c. : L .45 : N .484,54. * trigonometry : 876 : A.ths2,11,13,28,
    epicycloid : G .12. 37 : J.prs6, 132 : LM . 11 : N .42 :
    excess : Mél.2 : cn ,N .46 : f.Z .6 . Z .16 .
    of a quadrilateral : Me.75 . d .e of circles : Q . 20 .
    figures, division of : J.22. cot a sin b : Me.64 ; mnemonic , 895 :
    geometry : G .4 : J.6 ,8,132,ths15 and CM .3 .
    22 : M .15 : N .48,585,59 : Q .4 : TI.8. derived from plane : A .26 ,27 .
    harmonics or “ Laplace 's functions " : formulæ : 882 : A .5,16 ,24,26 : N .45 ,
    An.68, : C .86 ,99 : CD .1 : CM .2 : 46 ,53 : graphically, A .25: ap. in
    J.26 ,56 ,60,geo 68,70,80,82,90 : L . elliptic functions, A .40.
    45,48 : LM .9 : Me.77,782,85 : P.57 : geodetic reduction of a spherical
    Pr.8,18 : Q .7,: Z .24 . angle : 1 .512.
    analogues of : J.66 : LM .11. Cagnoli's th .: 904 .
    * *

    and connected d .i : Q .192. Gauss's eqs. : 897 : A .13,17 : J.7, 12


    as determinants : Me.77 . LM .3 ,13 .
    and homogeneous functions : CM .2. Legendre 's th.: C .96 : J.44 : L .41 :
    and potential of ellipse and ellip M .1 : N .56 : Z .20 .
    soid : P .79. Llhuillier's th .: 905 : A .20.
    * *

    and ultra-spherical functions : Z .12. Napier's eqs. : 896 : A .3,17 : CM .3 :


    P " (cos y), n = 00 : J .90. LM .3 ,13 .
    Napier's rules : 881.
    | Pium du, &c.: Q.17. supplement to , and geodesy : A .36 .
    * Sphero-conics : 5655a : tg .c,Q .8,99 : Me.3 :
    63° by continued fractions : JP.28 ; 2 :6 ,23 .
    homofocal: C .50.
    w : P .70. mechanical cn : LM .6.
    Sphero -conjugate tangents : An.55 .
    loci in spherical coordinates : TE.12,. Sphero -cyclides : LM .16 .
    oblong : An.52 ; area , J.42 . Spheroidal trigonometry : J.43 : M .22.
    polygonometry : J.2. Spheroidic transformation f. of Bessel:
    polygons in - and circum -scribed to A .53.
    small circles of the sphere, by * Spheroids : 5604,6152 ; cubature, 6158,
    elliptic transcendents : J.5. A .2.
    quadrilateral : A .4 ,40 : th , E .28 : N .45 . | Spieker's point : A .58.
    surface of : A . 349,352. Spiral : A .28 : L .692 : N .79 : Z .14 .
    quartics : foci, Q .21 ; 4-cyclic and * of Archimedes : 5296 : A .65,66 .
    3-focal, LM .12. conical: N .45.
    representation of surfaces : C .68,75, * equiangular : 5288 .
    *

    944,96 : M .13. * hyperbolic : 5302.


    *

    surface represented on a plane:Me.73. * involute : of circle,5306 ; of 4th order,


    *

    triangle : A .9,11,20,ths50 ,65 : E .f30 : C .662.


    J.10,pr28 : JP.2. Squares : J.22.
    ambiguous case : Me.77,852. whose diagonals are chords of given
    angles of, calculated from sides : circles : A .64.
    A .51. maximum with given sides : J.25 .
    by small circles, area : N .53. maximum in a triangle : J.15 .
    and circle : 898 : A .29,33. whose sides and diagonals are
    cos (A + B + C), f. : Me.72. rational: J .37 .
    and differentials of sides and whose sides pass through 4 points :
    angles : A . 10. A .43 .
    and ex-circles : 898 : E .30 . 64 transformed into 65, geo. : Me.77.
    graphic solution : AJ.6. sum of three : G .15 ; of four, L .57 .
    928 INDEX .

    Standards of length : Pr.8,21. Substitutions : - livear - (continued ) :


    Statistics : Z .26 . of a determinant : An.84 .
    Steiner's ths. and prs.: A .53 : J.13,14,16 , and integral : M .24 .
    18,23,25,36 ,71,73 : N .48,562,593,62 : powers and roots of : C .94 .
    Hexagon , A .6 : Ray - systems, reduction of: C .90 : JP .29.
    CD.6 . for reduction of elliptic functions
    Stereographic projection : A .39 : JP. of 1st kind : An.58 .
    16 : L .46 . successive : A .38 .
    Stereograms of surfaces : LM .2. which transform quadric functions
    Stereometry : JP.14 : ths A . 10,31,43,57 ; into others wbich contain only the
    J. 1,5 . squares of the variables : J.57.
    multiplication : J.49. of n letters : geo . for n = 3 ,4 ,5,6 , and
    quadric and cubic eqs. and surfaces : mystic hexagram : An.83.
    J.49g. no. of in a given no. of cycles : A .68 .
    Sternpolygon and sternpolyhedron : A . permutable amongst
    C .21,
    themselves :
    13.
    Stigmatics : see “ Clinant." of equidistant numbers in an integral
    Stewart's
    2, 15 .
    geo.ths.: CM .2 : L.48 : TE. function of a variable : N .51.
    of systems of equations : N .81.
    Striction lines of conicoids : Z .28. of six letters : C .63
    Strophoids : AJ.7 : N .752. a th . of Sylvester : AJ.I.
    * Sturm 's functions : 506 : A .62 : C .36 , which admit of a real inversion : J.73.
    62,68, : G .1,20 : J.48 : L .46 ,48, which do not alter the value of the
    67 : Mo.58,78 : N .43,46,52,54,66, function : C .212.
    67,81 : Q .3. which lower the degree of an eq . in
    and addition th. for elliptic functions two variables and their use in
    of lst kind : Z .17. Abelian integrals : C .15 .
    ext. to simultaneous eqs. : C .35. Sum and difference calculus : A .243.
    and H . C . F . : Pr.6. Sum of squares of lines drawn from a
    and a quartic equation : A .34. point to cut a curve in a given
    and their reciprocal relations: Mo. angle : J.11.
    73 . * Summation of series (see also
    remainders : J.43,48 . “ Series " and “ Expansions " ) :
    tables : P .57 : Pr.8 . 3781 : A .10 – 13,269,30,55,62 : No.
    ap. to transcendental eqs. : J.33 . 84 : C .87,88 : J. 10,31,33 : LM .4 ,
    ap . to transf. of binoinial eqs.: L .42 . 70: Mém .20,30 : P .1782,1784 – 7,
    Subdeterminants of a symm . system : 91, - 98, 1802,46, 47,411, – 19 :
    J .93 : M .82. Pr.14.
    Subfactorial n : Me.78 . approximate : 3820 : J.5 : I.24 .
    Subinvariants = seminvariants to bin . of arcs : A .63.
    ary quantics of unlimited order : | Bernoulli's method : J.31.
    AJ.5. Cauchy's th . : M .4.
    * Subnormal : 1160. C (n , r ) products of a , atb, ... at
    *Subsidiary angles : 726, 749. (n - 1) : Q .18 .
    Substitutions : A .62 : C .ths66 and 67, by definite integrals : A .4,6 ,38 : J.17,
    74,76,79 : G .9, 102,11,14,19 : L .65, 38,42,46 ,74 : Man .46 : Mém .11.
    72 : M .13. of derivatives and integrals : C .443.
    ap. to functions of six or fewer vari by differences : 264 : Mém . 30.
    ables : C .21. by differential formulæ : L .31 : Mém .
    ap . to linear d.e : 0 .78. 15 .
    by approximation, of the ratio of the by + (x ) : 2757.
    variables of a binary quantic to formulæ : A .47 : An.55 : J.30 .
    another function of the same Maclaurin 's, C .869: f (A , D ) DF (n ),
    Q .8 : Lagrange's, J.34 : P .60 :
    degree : C .80. Vandermonde's, L .41.
    canonical forms of : L .72. of terms of a high order : 0 .32.
    and conjugate substitutions: nC.215,22.
    - 3 Kummer's method : C .64 .
    of the form o (r) = e (part- + ar ? ) : Lejeune Dirichlet's : CD.9.
    M .2 . periodic : J.15 .
    linear : C . 98 : J.84 : M .19,202. selected terms: Me.75.
    INDEX . 929

    * Summation of series - (continued) : * Surface or surfaces — (continued):


    of sines and cosines : J.4. 2"а + " + " = 1: A.35 ;
    theorem re | xf(«, k ) dx : 2708. a = b = c = 1, A .21.
    of transcendents in alg .differentials : areas of : G .22: N .52.
    J. 19 . argument of points on : LM .16 .
    trigonometrical and infinite : Me.64. complex : M .5 ; of 4th order and
    Superposition : TE .21,23. class, M .2.
    * of small quantities : 1515. whose coordinates are Abelian func
    * Supplementary : angles : 620. tions of two parameters : C .92 :
    * chords : 120i .
    cones : thN .48 . M .19.
    * curves : 4917 – 20 : Man .54. of corresponding points : M .4 .
    Supplement integrals : J.98 . * cut orthogonally by spheres : 3393 :
    * Surds: 108 : N .47. C .36 .
    quadratic (see “ Roots of numbers.") deficiency of: M .3.
    ♡ (a2 + 62), v (a2 — 62) approximately : definitions : 5770.
    J. 13. " determined from twosurfaces of cen .
    * V (a + b ) : 121 : A .3,13 : J.17,20 : tres : A .68.
    Mém .10 : N .46,48. Dirichlet 's problem : An.71.
    Dupin 's theorem : C .74 : CM .4 : Q .12.
    * x (a vb): 122 ; ” (a + vb), 124. doubly circumscribing an n -tic sur
    * Surface of centres: 5774 : 1 .682: An. face : J.54 .
    57 : 0 .70,71, : E .30 : L .41,58 : of elliptic cone : An.51.
    LM .4 : M .5,16. of equal slope : C .98 : N .65.
    curvatures of the two; relation : C. equation of: gn5780 : A .3 ; for points
    74 ,79 . near origin , 5803.
    the two focal conics of a system of of even order : A .70.
    homofocal quadrics : C .53.. families of : C .70 : Me.72.
    of a quadric : Q .2. flexure of : J.182.
    bitangents of : Q . 13 . Gaussian theory of: LM .12 : N .52.
    model of : M0.62. generation of : C .94,97 : G .9 : J.49, :
    principal axes of : N .48. L .56 ,83 : M .18 .
    Surface curves : A .39,60 : An.54 : C .21, 1 implexes of : C .79,82.
    80, : L .66 : G .19 : J.2 : N .54,84. of minimum area : C .57 : J.8,13 :
    an an alg . surface : An .63 . L .59: Q . 14 .
    on cubic surfaces : M .21. of nth order : cnM .23 ; 2nd , 3rd and
    curvature of : N .65. 4th , Mél.6 .
    on a developable : An.57 ; the oscu of normals : N .59.
    lating plane making a constant whose normals all touch a sphere or
    angle with it , L .47 . conical surface : JP.4 ; do. for
    on an ellipsoid : An.51:CD .3 : geoL .50. surface of revolution , JP .5 .
    groups of rational : M .3. number under 9 conditions : C .62.
    of intersection : M .2 : N .68 ; of 2 octic of zero kind : G .12 .
    quadrics , A . 16 . order determined : Me.83 .
    multiple : G .142 ; singularities of, M .3 . one- sided : A .57.
    on a one-fold hyperbola : CD .4 . parabolic points of : CD .2.
    on quadrics : N .70. and p . d . e : C . 13 : CD .2 : Z .7 .
    rectification of : A .36 . and plane curves : J.54 ,72: M .7.
    relation to their tangents : C .82. and point moving on it : L .76 ,77 .
    on surfaces of revolution : Z .18,28. and point at oo on it : J.65.
    and osculating sphere : C .73 . relation of in Riemann 's sense : M .7 .
    * Surface or surfaces : 5770 : A .14 ,132,41, representation of : J .83.
    59,60 ,62 : An.51 - 3,55 ,60 .,61,65 , on a plane : An.68,71,76 .
    71 : At.57 : C .17,33,37 ,49,ths58 , one upon another : An.77 .
    61,64,67,69,80,86 ,992: G .3,21 : J. of revolution : C .86 : G .4.
    9,ths13 ,58 ,63,64,85 , 98 : JP.19, 24 , areas and volumes : 5877, - 9 : A .48 .
    25,33 : L .44,47,512,60 : M .2,4,7,9.2, of a conic about any axis in space :
    cn19 : Mo.82,833,84 : N .53,652,68 , JP .23 : L .63.
    72 : TI.14 : 2 .74,8 ,20 . of constant mean curvature : L .412.
    6 C
    930 INDEX.
    * Surface or surfaces - (continued ) : Symmetrical- (continued ) :
    meridian of : a lemniscate, G .21 ; of the common sol. of severaleqs. :
    generation of, C .85 . An.58 .
    meridian and contour curves of : multiplication of : Me.85.
    2 .21. of any number of variables : C .82 .
    oblique : Q .6 . simple and complete : M .18,20.
    passing through a given line, tan . tables of : AJ.5 ; of 12-ic, AJ.5.
    gent plane of : N .84. points : of 1st order : A .60 : cnA.64 .
    in perspective : JP.20. of tetrahedrons : A .60.
    of Areg.
    .57.
    polygon about a side, vol.: of a triangle : A .583.
    products : P .61: Pr.1l.
    quadric : A .55 : L .60. with primeroots of unity : Me.85.
    shortest line on : A .38 . tetrahedral surfaces : Z .11.
    superposable : C .80 : N .81. Symmetry ; plane and in space : N .47.
    Riemann's symmetrical; and perio . Synthesis : Ĉ .18.
    dicity modulus of the related * Synthetic division : 28 ; evolution ,
    Abelian integral of the 1st kind : Me.68 .
    Z .28 . * Syntractrix : 5282.
    ruled,with generators part of a linear
    complex : C .84.
    : mathem
    Tables tents, atical; Sect. I., con
    p . xi.
    ruled octic with 5 quartic curves : of Bernoulli's nos., logarithms, & c.,
    C .61. calculation of : J.2 .
    screw , parallel projection of : cnZ.18 . for empirical formulæ : AJ.5.
    section of, homogeneous eq .: LM .15 . in theory of numbers : Q .1.
    self-reciprocal : LM .2 ; quadric, E .36 . of et, e -s, log1004, log 10€ - * : CP.13.
    sextic of first species : M .21. Tac-loci: Me.83.
    singularities of: A .25 : An.79 : CD .7 : Tamisage: LM . 14.
    CM .2 : J.72 : M .9 : N .64 : Q .9. * Tangencies (circles, points, and lines) :
    cubics : M .4 ; and quadrics, A . 17. 937 : Pr.9 : Q .2,8 .
    16 singular points : C .929. Tangential eq . with the intercept and
    solutions by infinites of 3rd order : angle of inclination for coor
    geoC .802. dinates of the line : P .77.
    Steiner's : LM .5,14 : M .5. * Tangent coneat a singular point: 5783.
    touching a plane along a curve : * Tangents : 1160 : CnA .4,33 : J.562,73 :
    CD .3 : CM .2. N .42 : Z .23 .
    touching a fixed surface always : G .20 . and contact-point in loci and en
    transformation of: M .19. velopes : C .85 : cnN .57 .
    and transversal, th .: N .49. conjugate (and Dupin 's th .) : An.60.
    trapezoidal problem : 2 .14 . construction of: M .22 : N .80.
    two series of: prsAn.73 . double : Q .3 .
    web -system : J.82. eq. of; to find it : 4120, - 32 ; Me.66 .
    which cuts the curve of intersection faisceaux of: N .50 .
    of two alg . surfaces in the point cutby lines at a constantangle : A .43 .
    of contact of the stationary oscu locus of intersections of three : LM .
    lating planes : L .63. 15 .
    Surveying : N .50 , : geo. th .A .37. at a multiple point, cn .: JP. 13.
    Symbolic geometry : (Hamilton ), CD . and normals : 5100 ; relations, A .51.
    1 - 4. parallel : N .45 .
    Symbolical language : E .282. segments of: 4307 ; equality of, C .81..
    Symmedian line : E .42 : N .83 – 5 : Q .20 . ofa surface : 5781 ; at singular points :
    * Symmedian point : 4754c. 5783 : M .11,15 .
    Symmetrical : - conics of triangles : * Tangent planes : 5770, – 82 ; p.c5790 :
    A .59. CM .4 : N .45 .
    connections by generating functions : * and surface ; intersection of: 5786
    J.53. – 9 : L .58 .
    expressions : 219. to equidistant surfaces : cnZ .28 .
    figures : J.44 . triple : C .77.
    functions : C .76 : J.69,93,98 : LM .13 : tan 120°: P .8 , 18 .
    Q .20 : Z .4. tan a as a continued fraction : Z . 16 ;
    Brioschi's th .: C .98 . do. tan nx, Me.74.
    INDEX. 931

    tan næ : f. in N .53. Tetrahedron - (continued ) :


    tan - ' (atiy ) : Z .14 . of given surface : J.83.
    * tanh S : 2213. with opposite edges ; equal, N .79 ;
    tantochrone: L .44 : An.53. at right angles, Me.82.
    * Taylor's theorem : 1500, – 20, - 23 : A . with edges touching a sphere : N .74 .
    8,13 : AJ.4 , extl : C .58 : CM .1 : and four spheres : LM .122.
    J.11 : L .37,38,45,58,8264 : M .219, homologous : J.56 .
    23 : Me.72 ,73,75 : Mém .20 : N . and quadric : CD.8 : thN .71 ; 2 quad
    52,63,70,743,79 : TA.7,8 : Z.gz 2 rics, Q .8 .
    and 25. 6 dihedral angles of, eq . : N .46 ,67 .
    analogues of: J.6 . theorems: A.9, 10,31 : N .613,662: Q .3,5.
    convergence of : C .60 ,74 – 6 : J.28 : two : M .19 .
    L .73. volume: 5569 : A .452,57 : LM .2 : N .
    Cox's proof : CD .6 . 67 : Z .11.
    deductions : G .12. volume and surface in c.c and g .c :
    for an imaginary variable : Me.79. A .53 : CD .8 : N .68 .
    kinematic meaning of : J.36 . volume and normal, relation : N .54.
    Tetratops : A .692.
    reduced forms of : 0 .84.
    remainder : 1503: An.59 : C .13 : J.17 : Theoretical value function : J.55 .
    N .60 ,63 : 2 .4 . Theta -functions : Ac.3 : C .ths90 : J.61,
    a transformation of: C .78 . 66 ,74 : M .17 : Me.ap78 : P .80 ,82:
    Terminology : LM . 14 . thsZ .12.
    Ternary bilinear forms : G .21. analogues of : C .93.
    Ternary cubics : AJ.2,3 : C .56,90 : CD .1: addition theory : J.88,89 : LM .13.
    ap. to right line and triangle : A .3.
    J. 39,55 : JP .31,32 : L .58 : M . 1,4 ,9 . argument : J.73 : 4th M .14 : M .16 .
    in factors : An.76 : Q .7. characteristic of : complex AJ.6 : C .
    as four cubes : LM .10 . 882 : J.28 .
    parameter of the canonical transf. : th . of Riemann : J.88 .
    LM .12. as a definite integral : Me.76 .
    reduction to canonical form : C .81. double : LM .9 : Q . transf. 21.
    transformation of : J.63 . and 16 nodal quartic surface : J.83,
    Ternary forms: At.68 : G .1,9,18 . 85 ,87 ,88 .
    order of discriminant : LM .3 . formula of Riemann : J.93.
    with vanishing functional deter Jacobian, num . value : M .7,11.
    minants : M .18 . modular integrals : trAn.52,54 : J.71 .
    Ternary quadric forms: At.652 : C .92, multiplication of: LM .1 : M .17.
    94,96 : G .5,8 : J.20,40,70 ,77 ; transf. quadruple : AJ.62: J.83.
    of, 71,78 and 79 : L .59,77 : P .67. reduction of, from two variables to
    and corresponding hyperfuchsian one : 0 . 94 .
    functions : Ac.5 . representation of : M .6 : Z .11.
    indefinite : J.47; with 2 conjug. inde transf. of: A . 1 : An .79 : J.32 : L .80 :
    terminates, C .68 . M .252.
    representation by a square : An.75 . linear : Me.84 : Q .21.
    simultaneous system : J .80. triple : J.87 .
    table of reduced positive : J.41. in two variables : C.929: J.84 : M .
    Ternary quartic forms: C .563 : An.82 : 14 ,24 .
    M .172,20. Theta -series : constant factors of, J.98 :
    Terquem 's tb. : Q .4 . Me.81.
    Tesselation pr. : LM .2 . n -tuple : J.48.
    Tesseral harmonic analogues : CP .13 .
    Tetradrometry , differential f. : A .34 : Je = a?(c — «)/ {c (c — 2) — 12} : Q.15.
    2 .5 . * Three - bar curves : 5130 : LM .7,9 :
    Tetragon , analogue in space : CD .7. Me.76 .
    Tetrahedroid : J .87. triple generation of : Me.83.
    * Tetrahedron : 907 : A .3 , 16 ,239,51,56 : | Toothed wheels : cnTE.28 .
    J.65 : LM .4 : Me.62,662,82 : Mél. Topology with tables : M .19,24.
    2 : N .pr74,80,81 : Q .5 : geo Z .27. Tore : section of : G .10 : N .59,61,643,652
    determined from coordinates of ver circular, 56 and 65 .
    tices : J.73. and bi-tangent sphere: N .74 .
    932 INDEX.

    Toroid : A .8 : rectif. & c., A .9 : N .44. * Tortuous curves - ( continued ) :


    Toroidal functions: P .81 : Pr.31. singularities of: C .67: J.42.
    Torse : sextic : CP.11,: Q .14. spherical curvature: 5728, - 40 , – 47 :
    circumscribing two quadrics : Me.72. A .19 .
    depending on elliptic functions: Q .14 . triple, and their parallels : A .65.
    and curve: Q .11; and sphere, G .12. * Tractrix : 5279 ; area, E .35 .
    * Torsion : A .19,62,65 : J.60 : angle of, * Trajectory : 52 16 : M0.80 .
    5725 ; of involute, 5753 ; of evo of a displaced line ; oscul. plane, & c.:
    lute, 5754 . C .70 ,76 .
    * inflected , infinite and suspended : of 3 homofocal conics : An.64.
    5739 : N .45 . of meridian of surface of revolution :
    constant : L .422. L .46 .
    * Tortuous curves : 5721 : Ac.2 : C .19,43, surface of points of an invariable
    58 ,cn62 : G .4,5 ,21 : J.16 ,59 ,93 : figure whose displacement is
    JP.2,18 : L .502,th51,522 : M .5, 19 , subject to 4 conditions: C .76 ,77.
    th25 : Mo.82 : 1 .60 : No.79 : Q . of a tortuous curve : L .43 .
    4 ,5,7 . Transcendental: - arithmetic : J.29.
    approximate coordinates of a point curves : 5250 : An.76 : M .22.
    near the origin in terms of the equations : C .5.59 ,72,94 : G .6 ,10 : J.
    arc : 5755. 22 : L . 38 : N .55,56 .
    circular curvature of: 5722 : An.60 : mechanical solution : CP.4 .
    CD.9 : J.60 : JP .152 : Q .6 . separation of roots by " compteurs
    locus of centre of do. : 5741, 5748 logarithmiques " : 0 .44 .
    – 50 . without a root: J.73 .
    with circular and spherical curvature * Transcendental functions (see also
    in a constant ratio : L .51. “ Functions" ): 1401 : A .37 : C .
    cn . upon ruled cubics and quartics : 86 : J.3,th9,20 : L .51.
    0 .53. ofalg. differentials : J.23 : L .44 .
    with coordinates rational functions of arithmetical properties : M .22.
    a parameter : M . 3. classification of : L .37.
    cubic : J.27,802 : M .20 : Z .27 . connected with elliptic : C .17 .
    definitions : 5721. decomposition into factors by calcu .
    determined from relation of curva. lus of residues : C .32.
    ture and torsion : A .65 . and definite integrals : P .57.
    cubics : An.38,59 : C .45 : L .57 : of expansion of : J. 16 .
    3rd class, J.56 ; four tangents, integral : C .94,95 : G .23 : J.98.
    M .13. reduction of : Me.72 .
    and developable surfaces : CD .5 : squares of : Me.72
    L .45 : M .13. theorem of Starm : L .36 ..
    elements of arc : N .73g. whose derivatives are determined by
    generation of : C .94 : L .83; by two cubic eqs. ; summation of the
    pencils of corresponding right same : J .11 .
    lines, G .23. which result from the repeated integ
    with homog . coordinates : Q .3. ration of rational fractions : J.30 .
    the loci of similar osculating ellip . Transformation : - bilinear : M .2 .
    soids ; d .e of same : C .78 . birazionale : G .7 ; of 6th deg. in 3
    with loops: M .18. dimensions, LM .15 .
    with a max .ormin .property: Mém .13. contact : M .8.
    on a one-fold hyperboloid : C .53. * of coordinates : pl.4048 ; cb.5574 - 81:
    with the same polar surface, d.e : A . 13 : A .26 : CM . 1, : J. 2 : JP.7 :
    C .783. N .63.
    quadrics : J.20 . in 3 dimensions : A .13 : Q.2 : J.8.
    quartics : C .543; 1st species, J.93. rectangular into elliptic : Mél.4.
    cn . of two : C .53. Cremona's : M .4.
    intersection of quadrics : C .54. of curves : L .49,50 .
    quintics : 0 .543,58 . of differential equations: Me.82.
    radiiLof.48curvature and torsion :
    : Mém .18 .
    5739 : of equations : A .40 : N .64.; 3 vari
    ables, G .5 .
    and right linemethod : 5743. of a characteristic eq . by a discri.
    sextics, classification of : C .76 . minant : Au .56 .
    INDEX. 933

    Transformation - (continued ): Transitive function (continued ) :


    quadric : CM .2 : N .75 : M .23. reduction to intransitive : 0 .21.
    simultaneous : Q .11. * Transversals : 967 – 74 : A .13,18 ,27,30,
    of figures in a plane : A .4 : N .64,774. 56 : CD .5 : CM . 1 : J.8. : G .22 :
    in space : 0 .94,952,96 : N .79,802. Me.t.c682,75 : N .432,48 : TE.t.c24 .
    double reciprocal by normals to a orthogonal : AJ.3.
    sphere : 0 .56 . of plane alg. curves : Z .19.
    formulæ : J.32. of two points : A .66 .
    of functions: An.50; in an inf. series, parallel: A .13,57.
    A .3. of spherical
    by substitution : Mém .31. A .45 . triangle and quadrangle :
    quadric : Q .2 ; quadric differential, Trees, analytical : AJ.4.
    J.70. Triads of once- paired elements : Q .9.
    (ay - bx)2 + (bz - cy)2 + (cx — az)2 : | * Triangle : 700 ; A .17,19,22,29,33,36 ,43,
    CM . 1,2. cn46 ,61 : G .21 : J.50 : M . geo17 :
    two variables: Mém .11. Me.q .c62 : N .42.,43.
    * angles of: 677 : 738 — 45 : A .65 : P.28 :
    homographic plane : L .61.
    in geometry : C .71 : J.67 : JP .25 . division , A .51,58 : sum , geo960 :
    descriptive : G . 13 : Z .9. C .69,70 : difference , Q .8 .
    * linear : 582, 1794 : CD. 1,6 : CM .3,4 : * area : 707, 4036 — 41 : A .45,57 : CM .2
    M .2 : Mo.84. Mém .13.
    groups of : M .12,16 . bisectors: — of angles : 709, 742,
    932 , 4628 , - 30.
    methods of : C .92 ; which preserve
    an invariable relation between of sides : 738, 922i, 951, 4631 :
    derivatives of the same order, Mém .pr10,13.
    C .82 : L .76 . central line : 957,4644.
    which preserve the lines of curva and circle,
    10 .
    ths.: A .9,40,47,60 : Q.7,8,
    ture : C .92.
    modulus of : 1604. and 3 concentric circles : Me.85.
    orthogonal: 584 of, 1799. of 3 intersecting circles : Q .21.
    for equations dynamics : C .67. circle and parabola : Q .15 .
    plane : M .5 ; and in space, G .16 . and conic : N .70 .
    of powers into binomial coefficients : of constant species : 977.
    * *

    No.75 . construction of : 960.


    of product of n factors : An.51. equilateral, sum of sqs. of distances
    *

    quadratic, of an elliptic differential : of any point from its vertices :


    M .7. 923 : A .69 : gz1094 .
    in rational space : An.73 : LM .3. formed by joining the feet of bisec
    rectangular : LM .14 . tors of a triangle : A .642.
    reciprocal: C .92 : G .17 : N .64,82,83. Gauss's equations for a plane tri.
    of rectilineal space coordinates : J.63. angle : A .5 .
    by series doubly infinite : An.56 . notation : 4629 : E .44.
    space, for representation of alg . sur orthocentre : 952, 4634.
    faces : M .3. pedal line of: Me.83.
    of surfaces : M .21 : N .69: Q. 12 . perpendicular bisectors of sides : 713 ,
    of symbolic functions into isotropic 4639.
    means : C .43. perpendiculars on sides : 952, 4633 :
    Mém .13 .
    of tan - v V cta +' symmet. y , % : and point: Q .5.
    CD .9. and polars : circle, G .11 ; conic, Q.7 .
    of mid -points of sides, t.cQ .8 .
    Tschirnhausen's: An.58: P.62,65 : quadrisection of : Mém .9 .
    rational: A .51,56 .
    Pr.11,14 .
    ext. to quintics and higher : E . 12,4 : | remarkable points of (see also “ In
    J.582. centre," & c.): 955 — 9 : A . four, 47 ;
    unimodular : 1605 . two, 48 ; five, 52; 66,67 : E .28,
    of variables : G .5 : No.78 . 30 ,40 : LM . nine, 14g : N .70,73,83 :
    Transitive function : of 24 quantities : Z .11,15 .
    L .73 . | * of reference in t.c : 4006 .
    doubly, of 5 or 6 variables : C .21,22. 1 and right line: A .59.
    934 INDEX.

    Triangle - (continued ): 1 Underdeterminants of a symm . deter


    scalenity of : Me.66 ,68 . minant (i.e., successive first
    sides of: bisectors : A .17. minors) : J. 91 .
    division of : A .63: N .83. Undeterminants, & c. : A .59..
    containing conjugate poles with Unicursal curves : A .60 : C .78 ,94,96 :
    respect to four conics : An .69. LM .4 .
    cubic eq . for in terms of 4, R , and cubics, ths. re inflexion : E .28 .
    r : J. 20 . quartics : LM .16 : N .84 : Z .28.
    similar : C .792 ; under 3 or 4 condi. surfaces, transf. of: M .66 .
    tions, C .78 . Uniform functions : C .92,94%,95,96 ,
    solution of: 718, 859: A.3,51 : J.44; : with an alg. relation : 0 .91..
    TE .10 . of an anal. point æy : Ac.1 : C .94.
    six - point circle : Q . 4. . two points : C .96 ,97.
    symmetrical properties : A .57 : Pr.11. with “ coupures " : C .96 .
    theorems : A .9, 3,15,35,57,60,613,63, with discontinuities : C .94.
    68 . J.08 71,pr28 : Q .9,t.c7 and 8 , doubly periodic : 0 .94 .
    geol,2 , and 4 . from linear substitutions : M .19,20.
    relating to triangles of same peri. with a line of singular points, decom .
    meter under four other condi. position into factors : 0 . 92 .
    tions : C .84. monogenous : Ac.4 .
    * Triangular : - numbers : 287 : E .30 : | in the neighbourhood of a singular
    J.69 : L .ths63. point : C .89.
    prism : thN .42. of two independent variables : C .94,95.
    pyramid : At.19 : No.pr32.
    Tricircular and tetraspheric geometry :
    *Units of elasticity, electricity, and
    heat : p . 2.
    An.77. * Vanishing fractions : 1582 : M .15 : Q .1.
    * Trigon . in t.c : 4006 . Vanishing groups : CD .2,3,6,7,8.
    * Trigonometry : 600 : A.1,,2,8,11,13,30 : ap. to quantics : CD .2,3,6 .
    G .13 : extM .35. * Variation : 76 : C .17.
    formulæ : A .27,65 : Me.81,gzl : N .77 : of arbitrary constants : L . 38.
    geoQ .5 ,10. calculus of : 3028 - 91 : A .3,prs42 :
    * formulæ : 627,700,823 : A .33 : N .46,80 . An.52g : C .16,50 : CD .3 : CM .2 :
    functions : P.1796 ; in factorials, J.13,pr15,41,54 ,65 ,prs74,82 : JP.
    A .43. 17 : L .41 : M .27,152 : Me.prs72 :
    in binomial factors : L .43. Mo.57 : N .83 : Q.pr10 : and d.e,
    in Žpartial
    .13.
    fractions and products : L .38 and C .49.
    history : N .51.
    functions closely allied to : A .27. immediate integrability : 3090.
    tables, cn : A .1,23.,25 . and infinitesimalanalysis : C.17,40.
    theorems : A .2,50,51 : Q .pr7. relativemax.and min . : 3069 : L .42.
    on product of 4 sines : Me.81. of multiple integrals : J.15,56,f59 :
    Mém . 38 .
    oncos?a(1+sinto )+ (1+ ..:Q.15,16. transformations: J.55..
    two dependent variables : 3051.
    *

    2 sin a cos a sin 0 cos ♡ two independent variables : 3175.


    *

    + sin ? a cost o = 0 : integral, of functions : C .403.


    Trihedron : abouta parabola, locus of parameters : th2714, 3243 : N .77.
    *

    of vertex : N .63. * of second order : 3087 : A .4 : J.55 :


    and quadric : N .71. Q . 14 : Z .23.
    and tetrahedron : A .57. | * of higher order : 3089 : A .27 : 2 .26 .
    Trilinear forms: 0 .92,93. * Versiera or Witch of Agnesi : 5335.
    Trilinear relation of plane systems : * Volumes: - of solids: 5871 – 83 : A.
    J.98 . 31,32,36 : J.34 : N .f57,80.
    * Triplicate-ratio circle : 4754b : E .40,42, approximate : C .95 .
    435,44 . of frustums: A .33 ; of conicoids,
    Twisted surface : see “ Scrolls ." J.44.
    Twist of a bar : Me.80. of right cylinders and cones in abso
    *Umbilics : 5777, 5819 --23 : J.65 : JP. lute geometry : A .59.
    18 : M .9 : Q .3. of surface loci of connected points :
    * of quadrics : 5603,5834 : C. 96 . LM .14 .
    INDEX . 935

    Volumes — (continued ) : Wear of gold coins : E.43.


    and surfaces by curvilinear coor Web surfaces : see “ Net surfaces.”
    dinates: C . 16 . Weierstrass's function inb" cosaӾr
    * Wallis'sJP .28
    formula
    .
    : 2456 : A .39,gz39:
    with a > 1 and b < 1 : G .18 : J.
    Waring's identity : N .49,62 : extMe.85.
    Wave surface : C . 13,47,78 ,ge082,92 :
    63,90 .
    CD .7,8 : CP.6 : L .46 : Me.662,73, expansion in powers of the modulus :
    C .822,85,86 : L .793.
    76 ,78 ,79 : N .63 ,82 : Pr.32 : Q . 2,33, | *Wilson 's theorem : 371 : A .48 : CD .9 :
    4,5 ,9,15.,17. J.8 ,19, 20 : Me.83 : N .43.
    asymptotes : C .97. generalisation : J.31 : Me.64 : Mél.2 :
    and cone : Q .23,26 .
    cubature of : An.61.
    N .45 .
    Wronski's methods : C .92 : L .82,83 .
    generation and on : 0 .902. formula of 1812 : N .74;: Q .th12.
    lines of curvature : An.59 : C .97.
    normals and centres of curvature : Zetafuchsian functions : Ac.5 .
    geoC.64. Zonal conics of tetrazonal quartics :
    umbilics, geo : C .882. Q .10.

    LIBRARY
    OF THE
    UNIVERSITY
    OF CALIFORNIA
    ‫ܕܢܐܬܐ‬
    ‫ܘܝܐ‬
    ‫܂‬ ‫ܢܨ‬
    ‫‪Plate‬‬

    ‫‪.‬‬
    ‫‪.‬‬
    ‫‪1‬‬ ‫܂‬
    Plate 2 .
    16 -- 30

    21.

    Τ' Μ Ν
    - --
    - - - -

    30
    - -
    -
    )
    45
    -
    31
    Ar
    ai.
    A
    Plate4.
    Plate 4

    y 46 -- 5

    4 .

    ✓ M -N
    ( "
    53 Plate
    . 5
    52- 623)

    59
    RETURN TO the circulation desk of any
    University of California Library
    or to the
    NORTHERN REGIONAL LIBRARY FACILITY
    Bldg .400 ,Richmond Field Station
    University of California
    Richmond ,CA 94804-4698

    ALL BOOKS MAY BE RECALLED AFTER 7 DAYS


    • 2 -month loansmay be renewed by calling
    (510 ) 642-6753

    B,CALIF
    FERK
    .OUNIV
    • 1-year loansmay be recharged by bringing
    books to NRLF
    • Renewals and recharges may be made 4
    days prior to due date .

    DUE AS STAMPED BELOW

    MAY 26 2001

    JUL 0 3 2003
    OCT 3 0 2007

    12,000 (11/95)
    YD 07347

    76606
    QAAI
    C3
    Plate 6
    63 - 751
    65 .

    A :

    75 .
    PlPlate7.
    ace7.
    76 - 824

    e
    var :
    ate 8
    8 87 - 88

    W.
    94 -
    TIT - 113 )
    Plate II.

    112 .

    114
    116 - 123

    - --
    Panage 13 124 -132
    129
    Plate 14 133

    на |
    'late 15 .

    142 – 152
    | | 153 - 164
    | 165 -167
    6 168 -177
    182
    -
    178
    9
    19
    Plate
    ‫لا‬
    *
    ‫؟‬ .
    3 3
    19- 18 Z
    Q
    RETURN TO the circulation desk of any
    University ofCalifornia Library
    or to the
    NORTHERN REGIONAL LIBRARY FACILITY
    Bldg. 400 , Richmond Field Station
    University of California
    Richmond, CA 94804- 4698
    ALL BOOKS MAY BE RECALLED AFTER 7 DAYS
    • 2 -month loansmay be renewed by calling
    (510 ) 642-6753
    1-year loansmay be recharged by bringing
    books to NRLF
    • Renewals and rechargesmaybemade 4
    days prior to due date .

    DUE AS STAMPED BELOW

    MAY 26 2001 21995


    AINN
    38
    10'"

    JUL 0 3 2003
    OCT 30 2007

    12,000 (11 / 95 )
    YD 07347

    76606
    QAA1
    c3

    You might also like