This action might not be possible to undo. Are you sure you want to continue?
Cornell ftntuetgUg Babrarg
Strata, SJem 5«>*h
BOUGHT WITH THE INCOME OF THE
SAGE ENDOWMENT FUND
THE GIFT OF
HENRY W. SAGE
1891
(
QA
Cornell University Library
565.H65
Plane algebraic curves,
3 1924 001 544 216
Cornell University Library
The
original of this
book
is in
the Cornell University Library.
There are no known copyright
restrictions in
text.
the United States on the use of the
http://www.archive.org/details/cu31924001544216
PLANE ALGEBRAIC CURVES
OXFOKD UNIVEESITY PEESS
LONDON
TOEONTO
EDINBURGH
MELBOURNE
GLASGOW
CAPE TOWN
NEW YORK
BOMBAY
HUMPHREY MILFORD
PUBLISHER TO THE UNIVERSITY
PLANE ALGEBRAIC CURVES
BY
HAROLD HILTON,
M.A., D.Sc.
PROFESSOR Or MATHEMATICS IN THE UNIVERSITY OF LONDON
AND HEAD OF THE MATHEMATICAL DEPARTMENT OF BEDFORD COLLEGE FORMERLY FELLOW OF MAGDALEN COLLEGE, OXFORD, AND ASSISTANT
MATHEMATICAL LECTURER AT THE UNIVERSITY COLLEGE, BANGOR
OXFORD
AT THE CLARENDON PRESS
1920
&c.PEEFACE THOUGH reader has not the theory of plane algebraic curves still attracts mathematical students. part. all many Salmon's classic treatise supplied desired at the time of edition its that could be appearance. it would be almost impossible to trace the steps by which particular results have reached their present form. a field in which has attracted so many workers. I have not In fact. the English suitable books at his disposal. XX. Teixeira. attempted to give systematic references. But most of the contents and examples are extracted from a very large number of mathematical periodicals. list at ception of the the end of With the exCh. if only to bring some more recent developments within reach of the student. . It seemed therefore as if a new book on the subject might be useful. The solutions mine most even in the case of examples derived from other authors. XVII. was published some and has been long out of print. Basset. XI. of my own are contributions to for the the subject. but the last forty years ago. In some cases I cannot eveli remember whether a . In the preparation of this volume Wieleitner. Loria. result is my own or not and XVIII contain most but Chapters IX. I have made frequent use of the books written by Salmon.
for example. into which any quartic with two unreal biflecnOdes can be projected. No one can really master a branch of mathematics it "himself. work has been published on The reader will doubtless detect other important omissions.vi PREFACE In a book dealing with so wide a subject I can hardly hope to escape the criticism that I have included just that material which happens to interest myself. I have not thorough in cluded any discussion of curves of degree n for special values of n other than 2. except by working at I therefore. and have not attempted to subdivide the all their type by considering possible positions relative to the line at infinity. But on the whole I have tried to cover the limited ground I have selected with reasonable completeness. such as touch How many conies five given conies?' I have treated all curves with the same degree and singularities as forming a single type. I have not seriously dealt with ' problems of enumeration. A dis cussion of quintic curves would be very welcome. I have not given the properties type. setting him on the right track if he in difficulties. Cassinian curves. or 4. as he pleases. The reader I can select from them few or many. At any their rate very little properties. for the long lists of examples. hope that these will be of real assistance to the student. enabling . 8. of ' special plane curves % unless they are representa tive of some general such as. I give hints for solution in most is cases. and have excluded other matter of equal or greater importance. make no apology. but at present the difficulties seem insuperable.
VII. H.. June 1919. . XVIII. PREFACE him him to check the accuracy of his results. #10 and 11. 8 to 10 Ch. are due to friends and pupils who I have made suggestions and pointed out inaccuracies while the book was being written. XII. 7 and 8 3 to 7 XI XX. owe a special debt of gratitude to Miss G. vii and giving a guarantee that the examples are not of undifficulty. Ch. X. reasonable The beginner should not attempt parts which to read the book straight through. IX. Ch. 3 to 12 7 to 10 §§ . Sadd. H. VIII. D. Ch. . Ch. . $ . As a rough guide I recommend the omisa first sion of the following portions on reading VI . 9 to 15 Ch. $$ Ch. Ch. . must also express my gratitude to the Delegates of the University Press for so kindly undertaking the publication of the book. §§ . §§ §§ . 4 and 5 Ch. most. XVII. §§ 6 to*8 Ch. My best thanks XXI. Ch. but should select for himself those he thinks easiest and likely to interest him Ch. XIX. : . $ XVI Ch. who has given me I very valuable lists of corrections required in the MS.
.
4. 3. 3. 1 3 . Examples of curvetracing . III CUBVETRACING 2. 7. The object of curvetracing The method of curvetracing Newton's diagram . Plane perspective Asymptotes Circular points and lines Higher plane curves . 23 25 29 6. 9. Conditions for a double point Points at infinity . 2. Double points Multiple points 3. 5.CONTENTS CHAPTER Coordinates Projection I INTEODUCTOBY PAGE § 1. &c. Theory of equations CHAPTER 1. 4 7 Intersections of two curves 10 12 13 14 16 8. 6. 11. Pencil of curves Tangents Inversion 10. 18 21 2. 37 37 38 39 410. II SINGULAR POINTS Inflexions. 4. 5. Relations between coefficients 33 CHAPTER § 1.
IV PAGE 57 58 59 61 TANGENTIAL EQUATION AND POLAR RECIPROCATION Tangential equation Class of a curve . Tangential equation near a point Common tangents of two curves Polar reciprocal of superlinear branch 83 85 86 CHAPTER § 1. 10. Definition of foci Singular fpci . 5. 5... Tangential equation of any curve Tangential equation of the circular points 5. two curves . and its reciprocal 62 62 63 CHAPTER V § 1... POLAR CURVES of an inflexion 2... VI SUFERLINEAR BRANCHES Expansion of y in terms of x near the origin Intersections of 76 81 2... 3. FOCI . 3.. 6... 69 71 Method of obtaining the foci Inverse and reciprocal of foci 73 74 CHAPTER § 1. Harmonic polar First polar curve Equation of Hessian Intersections of a curve with . 4. 9.. The Steinerian The Cayleyan The Jacobian of three curves . .. . 2. . . 4.. . Equation of polar curves Polar curves of a point on the curve 88 89 92 95 96 98 99 104 107 108 4. 3. its on a cubic Hessian 8.. 3. Polar reciprocation Singularities of a curve . . 2........ .. 4. 7. VII Polar curves ....... Equation of polar reciprocal 7.. .CONTENTS CHAPTER § 1. 6. 8..
Deficiency 8.... 2.. . .116 118 Higher singularities CHAPTER IX QUADRATIC TRANSFORMATION § 1.. . 3... . .. .. Intersections of a curve 11. Definition of quadratic transformation . 8. 5. . 9. . 3.. 6. 7. 5. ..CONTENTS xi CHAPTER VIII plOcker's numbers § 1. Plueker's numbers . .. . 4. . 127 128 129 130 131 12. . . 133 CHAPTER X THE PARAMETER § 1. . 120 128 2. Inversion point * The number 4. . of intersections of two curves at a given 124 125 126 Tangents from a singular point to a curve Latent singularities . . Applications of this transformation . ...115 .. ..... ...146 153 156 7. 2. Pointcoordinates in terms of a parameter Linecoordinates in terms of a parameter Deficiency not negative . . Deficiency PAGE 112 IIS Multiple points with distinct tangents Multiple points with superlinear branches . 137 139 144 .. 6... 5.127 .. Class of a curve . Deficiency unaltered by quadratic transformation Deficiency for higher singularities 10. Coordinates of a point in terms of trigonometric or hyperbolic functions ... with adjoined curves Another transformation . 3. 4. Unicursal curves ... . . .... ..... Curves with unit deficiency ... • .
8. XIII UN1CURSAL CUBICS Types of cubic Geometrical methods applied to unicursal cubics Cuspidal cubics 2.. Theory of residuals 185 186 191 193 194 CHAPTER 1. 4. 5. 9. . Derived curves Evolutes. 3.. of orthopti locus Isoptic locus 174 175 Cissoid .. 4.Xll CONTENTS CHAPTER XI DERIVED CURVES PAGE § 1. Pedal curve . Orthoptic locus 6. numbers .. two nica Intersections of any two curves 8. Plticker's 7. 177 179 182 CHAPTER XII INTERSECTIONS OF CURVES § 1". 3.. 6. 3. Conditions determining a curve 2... 201 201 204 207 .. 10. 2. 161 161 164 166 169 171 4.. Inverse curve . 9. 5.. Cubics through eight points Intersections of .. Conchoid Parallel Curves Other derived curves 11. Nodal cubics . .. 10.. 7.
A standard equation of the cubic with two circuits .....CONTENTS CHAPTER XIV NONSINGULAR CUBICS § 1.8. .. 7. 2. 4. Pencil of tangents from a point on a cubic The cubic {x r y + zf + Qltxyz = •6. . Syzygetic cubics .. 2. Applications of the parameter Another standard equation 8. Any cubic as a Jacobian of three conies .... Coordinates in terms of Weierstrass's function 252 252 254 255 256 257 261 263 . 5... ... 4.. 3.. 4. 5. Symmetry of cubics 7.. Rulerconstruction for cubics Cayleyan of cubic Tangential equation of Cayleyan of cubic .. 5. 3. . A standard equation of the cubic with one circuit Unicursal cubic . Coordinates of any point on a cubic with two circuits Parameters of points on a conic or cubic .... . of the cubic . 6. Jacobian of three conies .. . . Cubics with unit deficiency Circular cubics Foci of circular cubics . 3.. The cubic x? + y 3 + 3 + 6mxyg = 9. xm 2. PAGE 214 217 220 226 227 229 233 237 CHAPTER XV CUBICS AS JACOBIANS § 1. 241 244 245 248 249 CHAPTER XVI USE OF PARAMETER FOR NONSINGULAR CUBICS § 1...
. 6. Nodal quartics Cuspidal quartics Bicircular quartics . 3. • 298 300 304 311 316 319 322 323 325 326 327 327 328 '328 331 15. 5. &c... Quartics with a node and a cusp .. 6... 8. Relations 5. Schlafli's doublesix . Bitangents of nonsingular quartics 4. Quartics with two real nodes Cartesian curves Quartics with two real cusps . Quartics with a rhamphoid cusp CHAPTER XIX NONSINGULAR QUARTICS § 1. 9. Cassinian curves Quartics with two real biflecnodes 10. Quartics with a tacnode .. Types of quartio Geometrical methods .XIV CONTENTS CHAPTER XVII UNICURSAL QUARTICS § 1. 3. 5. 2... Nonsingular quartics • . with two unreal flecnodes with two real flecnodes Quartics with a node and a biflecnode Quartics with a flecnode and another double point 14. Quartics 11. PAGE 264 264 The quartic a/ot? 4.. 7.. + b/y 2 + c/z 2 + 2f/yz + 2g/zx + 2h/xy = Conies connected with a trinodal quartic Tricuspidal quartic . Quartics 12. 6. 2. 8.  . . 273 283 285 289 290 CHAPTER XVIII QUARTICS OF DEFICIENCY ONE OR TWO § 1. 2. 4. complex between complexes The Hessian notation Lines on a cubic surface Steiner's . 7. 13.. 7.. Other quartics with three distinct double points Unicursal quartics with two distinct double points Quartics with a triple point. 3.
.
2 '. references. Ex. for '» — 3'. Ex. Add Gettingen Nachrichten. 12.2n+1 yln+l 22jl +i' f° r ^sn+u Vsn+l) s»+l Page 24. For the solution read Use Ex. § 6. and 'a. The quartic consecutive to Q is supposed noncuspidal. this book was in the press a treatise by S. X. ' ' ' N ment on page 146. Page 369. 11 or Ch. P may be made any point of the nic might be incorrect. Ex. In Ch. Read '2m — 2' • . Page 371. 308 to the list of ' ' ' '. p. While 2. xi (1909). Similarly in Ch. 10. Otherwise the statePage 146. must be less than n. line 33. . Ex. By choosing t properly. . X. In Fig. § 4. 13 it is essential that the nest should lie inside the circle and outside 12. In the theorem at the foot of the page a linear branch is counted as being superlinear of order 1.ERRATA. Ganguli called Lectures on the Theory of Plane Curves has appeared. Omit(l)". § 7. ETC. line 24. Page Page 9. Read ' three real tangents for three tangents '. Preface. I. Page 116. 8. Page 359.
and OM. .' Coordinates. 1) we draw parallels PJ¥. . We If two ON to (Fig. x and y are the Cartesian through a point P PM coordinates of the point P (x. involution. contain x and y units of length. projection. y). shall assume a knowledge of the more elementary portions of the calculus and of pure and analytical geometry including the theory of crossratio. reciprocation. INTRODUCTORY § 1. and inversion but in this introductory chapter we shall remind the reader of some elementary results of which we shall make frequent use. fixed reference lines meeting at an origin 0.1 CHAPTER.
but considered fixed when once chosen. . They are evidently are called the areal coordinates of P. the ratios to the triangle the triangles PBG. For instance. ABC x + y + z= Instead of defining the position of by its areal coordinates.. 2). 1. y = k2 . The new equation of a given curve when fresh homogeneous . z in this We have the relation ca'se general homogeneous coordinates. y. I 1 x = r cos 6.2 COORDINATES If the referenceaxes are perpendicular. y. A Fig. though these are the actual coordinates only if areal coordinates are used. In this k lt k2 k 3 are any constants chosen arbitrarily. PGA. connected by the relation If ^4£C x. y. z we may take any quantities proportional to them when dealing with such homogeneous equationsi. (0. we may use instead any constant multiples of them and write P x = k PBG/ABC. the vertices of the triangle of reference will usually be taken as (1. (0. z = k. 1 . y. 6) with respect to the pole and prime vector y = 0. 2. We call x. We call r and 6 the polar coordinates of the point P(r. z of PAB 1. . We shall mean by the symbols x. PAB/ABC. 1) . ®/h+y/h+ z/ ]h= !> by means of which the equation of any locus may be made homogeneous. y = r sin 6 where r is the distance OP and 6 is the angle between OP and y = 0. PC A/ABC. z the general homogeneous (not necessarily the areal) coordinates. is a fixed triangle of reference (Fig. 0). . unless the contrary is stated. 0. 0. 0). Instead of x. or quantities proportional to them.
. IT' and a fixed point V. Similarly for involution pencils. its projection is . Projection. BD) AB CD/ AD GB = — 1. of a straight line is evidently a straight line. Suppose we have fixed planes n. Then P' is called the projection of P on n'. AD . 3. so that harmonic. where I. ' ' ' b2 . sections of two transversals with the rays of a pencil whose The projection The projection AB . z in the original equation by expressions of the form 1yX + m^ + n^. which is called the projection of c. l 3 iv + m3 y + nz z respectively. P' traces out some locus c'. is harmonic. ' ' ' Fig. . Similarly the projection of a pencil of four concurrent lines In fact. involution range It follows that the projection of the traced out by points P. In particular. l 2 x + m 2 y + n 2 z. PP') is harmonic. if the range (AG. of a range (ABGD) of four collinear points CD/ GB is a range of the same whose crossratio is For the range and its projection are the intercrossratio. P' such that (IJ. y. the pencil and is a pencil of the same crossratio. § 2. Similarly c is the projection of c' on IT. vertex is V. its projection have the intersection a of the planes IT and IT' as a common transversal. V being the vertex of projection'.2 ' 1 PROJECTION 3 coordinates (and triangle of reference) are taken is obtained by replacing x. If P traces out a locus c. Suppose also that P is any point in IT and that VP meets IT' in P'. J are the fixed double points of the involution. is an involution range.
PP' and RR' continue to intersect. It is possible with a given vanishing line to project two angles into angles of given magnitude or to project a given conic into a circle. R with it. In general. If the its approaches projection P lies on v. Q'. Plane Perspective. This is only possible if PP'. and that Q. ' all points at infinity in the plane IT' lie on the straight line which is the projection of the vanishing line v '. P . But P is any point whatever of IT. P projection. since they are not coplanar. QQ'. Q and Q\ R and R' are projections of each other. 2. v'. P Ex. and Q. that case in which the planes are parallel. R are taken as fixed points of IT. Since Q'R' is the projection of QR. are concurrent. even in the say. while IT' is kept fixed. moving in the plane IT. in which the plane of the diagram is perpendicular to the lines v. If two lines through P in the plane n meet v in II and A". therefore. Similarly PP' and QQ'. 1. . Ex. Q. R' continue to be coplanar. Hence QQ' and RR' continue to intersect. at infinity . * Suppose that P and P'.4 PROJECTION I 2 Suppose that the planes through V parallel to II' and II meet IT and IT' in the lines v and v' respectively (see Fig. and P as any other point of IT. 3. [See Ex.] § 3. P' recedes indefinitely while if has no does not meet IT'. two figures! in IT and II' which were originally projections of each other remain projections of each other. QR and Q'R' continue to meet on a. QR and Q'R' meet on a. respect to a conic and between conjugate points or lines is unaltered by projection. v. 2. If the plane IT is turned about a carrying P. Similarly such a statement as a parabola touches the line It is meet We ' projection touches the vanishing line and so on. Similar remarks apply to the line v'. ' means that ' its '. The relation between pole and polar with. if finite points. 1. so that . The vertex V RR' . or (what is the same thing) that the projection of a straight line is a straight line. R. 3. VP point P. when we describe any property we mean a of a curve at an infinitely distant point P' property possessed by its projection at the corresponding point of v it being understood that the property is one and P' were which would be unaltered by projection. Hence when IT is rotated about a. a). the angle HPK projects into an angle equal to HVK. Project any two conjugate pairs of lines through the pole of the vanishing line into perpendicular pairs. convenient to observe the convention that two planes in a straight line. Ex.
since the distance between a and v is constant and is equal to the distance between and v'. 3.13 PLANE PERSPECTIVE turns about the line v'. as is evident from Fig. V Two figures which originally were projections of one another are now two figures in the same plane with the property that the line joining corresponding points of the two figures passes ysy' . Suppose that n turns about a till it coincides with IT.
f(x. is bc/y) = We can always choose the projection so that b — c. then. y. Then putting n cm in (ii) we see that the perpendicular from P' on GB bears a constant ratio to the quotient of the peron GB and pendiculars fro. Now (changing the notation) we may take homogeneous with ABG as triangle of reference. curve. Let Ix the equation of its projection. P P AB so for y/z. If. y. PLANE PERSPECTIVE b(y + c)/y) of I 3 line the is line VP and the joining H' But to the intersection of by definition y) P' with the axis of perspective. + my + n — it distance d from of the point be any line in the first figure. Hence d' (c l 2 2 + n 2 )% = bed (P + mrf/y (ii) of which the Suppose that we take a fixed triangle vertex G coincides with V. coordinates (x. y.. . the axis of x being the vanishing line both for the curve and its projection.y' + b). if f (x. is the homogeneous equation of Conversely. y) is given by The d (I2 + m 5 = Ix + my + n. (x'. y by ax/y. 1) obtained by a projection in which z is vanishing line. P (x. y/z) and GA — = is the vanishing line. = is the equation of the locus of P. is a homogeneous equation of a projection in f(x/z.m B and similarly for the perpendicular from P' on GA. z) of such that x/z is equal to the quotient of the perpendiculars multiplied by any constant we please on GB and from ABG = P A . we get the equation of its projection . 2 ) The distance d' from the corresponding line in the second of the point P' (bx/y. We thus get the useful rule 2 If in the equation of a curve we replace x. and y are the 0. bc/y) is given figure clx + ny + bcm = 2 by d'y (cH + n 2 f = be (I'x + my + n). which z = = = = = = followed by an orthogonal projection. 6 (bx/y. Also the perpendiculars from P' on GB and GA are constant multiples of the Cartesian coordinates of P' referred to and GB as axes of reference. from which (i) at once PR follows. while A and B lie on v. f$x/y.z) is the Cartesian equation of the curve a curve. a /y. and the axis of y being unaltered. f(x. If/ (x. y) is the Cartesian equation of a. x projections of the Cartesian axes of reference.
V. J be the points in one plane corresponding to the circular points of the other plane (§ 5). recedes indefinitely. Asymptotes. and the figures have Fas a centre of similitude. at Fig.] § 4. The axes of reference are IJ and its perpendicular bisector. [Project a to infinity. d/y approaches the limit zero as P approaches Q along the curve. P corresponding to a given point on Given V. the perspective is called harmonic] Ex. [If Q is Given on v. So for the other plane. Hence the point P' of the corresponding curve in the second which is at a distance d' from the projection t' of t. P corresponding to a given point P. the crossratio of (VPRP'J constant. 4. Such a line as t which is the projection of a tangent. v.] Q' is at infinity on VQ. a. struct 2. 1.4 1 ASYMPTOTES 7 in is Ex. 5). [Let I. a. whose point of contact is on the vanishing line (but the tangent and . and a pair of corresponding points Q and P. but t and v do not coincide referred to in Taking t as the line Ix + my + n = (Fig. approaches t' indefinitely as P' figure (the projection of the first). 2. If P and J" are any two corresponding points of two figures plane perspective and VPP meets a in i?. Ex. 5. Q'. (ii) of § 3 and P as any point of the curve. con [P' lies VP and on the line joining construct Q' to the intersection of PQ and a. Now use Ex. Ex.] 3. If the crossratio is —1. The relations connecting the coordinates of any two corresponding points of two planes which have a onetoone correspondence can be put in the form (i) of § 3 by a suitable choice of axes of reference. Suppose that in § 3 a curve in the first figure crosses v Q and t is the tangent at Q.
We shall denote them usually by a> and u> Any line through a circular point. any line parallel to y — + •/(— \)x. 4. infinitely distant points go through the unreal and \/(—l)x. as we may put it. § 5.y) 0. Applying this to the general equation of a circle It is well x* + y 2 + 2gx + 2fy + c — x2 + y* 0. 8 ASYMPTOTES is I 4 vanishing line do not coincide). though most of our work will be concerned with the case n > 2. called an asymptote of the Circular Points and Lines. all circles on y = ± ' Ex. 2. z of degree n (and whose Cartesian equation is therefore obtained by equating to zero a polynomial in x. e. is called a circular line. Ex. known that the lines through the origin parallel to the asymptotes of a conic are given by equating to zero the terms of the second degree in the Cartesian equation of the conic. the properties of conies being supposed too well known to require further investigation. The word higher implies that the degree is greater than the second. y. Hence A curve of degree n meets any straight line in n real or unreal points in general. z) = is the equation of such a curve. y. These points are called the circular points at infinity. If f(x.* ' ' = * note that the number of real points a positive integer see also § 7. another. [With the line joining them as vanishing line project any conic through them into a circle. See § 2. point not on a circular line is at an infinite distance from that line. i. Higher Plane Curves.: . Nevertheless we shall not consider conies necessarily excluded from the definition. where r is zero or .] § 6. is called an algebraic plane curve or higher plane curve of degree n. see that the asymptotes of all circles are parallel to = 0. 1. 1) will be f(vx. 0. the lines joining its intersections with Xx + fiy + vz = to the point (0. Two Any points on a circular line are at zero distance from one Ex. we or. A curve whose homogeneous equation is obtained by equating to zero a polynomial in x. . Project a given pair of points into the circular points. —\x — y. vy. Ex. projected curve. 3. 3. A circular line is perpendicular to itself. Ex. and are therefore n in number in general. We is n — 2r. y of degree n).
3.00. BQn • : Oft Oft independent of the position of 0. Ex. . 5ic.BP... ' curve of degree n ' will often be called for short ' an nic '.. The lines joining «ic in meet a given ratio P P 1( 2 . 2 to the triangle ABO. 4.y')+ {coefficient of a. 2 y . Ex. Qn'' Bi> Ri Sn P1. when taken together. and that the intersections of an infinite number of n tangents lie on the (n— l)ic. The cases n = 1 .. quintic.. .. . 2 . and close to meets the curves in ft and the tangent at Pin T. [Apply Ex.BR. any point to two fixed points A and B Show that the . .. . Two P R P (OQTR). A line curves touch at P. p2 two lines are drawn in fixed directions P„ and ft.. Given all but one of the 3w intersections of three lines'with an construct the remaining intersection. . Show {IrAR.] ' ARn . . 2..e. . take ABC as triangle of reference. 1 or and 2 are known as Menelaus's and Carnot's theorems. is the crossthat the ratio of the curvatures of the two curves at 6. 5. 1 . .. ratio of at Ex. 5.. mic with the sides of wic. the (finite) If the curve meets the line at a point on the vanishing line projections of the curve and line meet in less than points. Then 1 .BQ2 . ." in/(». AR 2 [Use Ex. .. as forming an nic. sides of a triangle ABC meet an nic in ft> ft) ••. .. y) = 0. 2 to the intersections of an any polygon. Show that 0Pj OQ. 3. BP BPn CQ CQ Cft. ft. quartic. OP . show that the n(n — 1) points thus obtained lie on an (n — l)ic. P 2 . [Take Cartesian axes of reference in the fixed directions {oc y ).. Another notation for an nic in common use is G '. 1..'] Ex. OP2 . For instance a cubic may degenerate into a conic and a straight line or into three straight lines. let be 1 .6 1 HIGHER PLANE CURVES 9 n (§ 2). The that . OPn BQ. Q n . Such an nic will be called degenerate. a cubic. n Two or more curves whose degrees have the sum h may be considered... ' Thus a curve of degree Ex. ft. P„ and Q u Q2 . is . A .. Pn . Deduce the fact that the ratio of the curvatures of two curves a point of contact is unaltered by projection. Given a set of n tangents to a conic meeting at \n\n — 1) points and a second set of tangents also meeting at \n(n — 1) points.OP 1 2 OPn = ±f{x'..] Ex. i. curve. = AQ AQn CP BRn CPn .. Oft AP. Extend Ex.. 7. Oft . AP 2 APn through Show is any point not near P. OP. Through any point nic in P. . will be called a 3ic.. meeting a given the ratio is P 2 P.. . and the curve be f(x. 1 . Ex.y)}. Such exceptions will be considered later. 4ic. . 4. independent of the position of 0.
] x= tTz. This eliminant gives the lines joining the point (0.] § 7.. y = (t+T)z.i+1.y sn+1 + zl /1 ^2 . variable line is drawn in a given direction meeting a given where the radius of curvature is p and the tangent makes an angle with the given direction. 4i. ( {f(t). .<p(T)f(T). = xi> 2fr. then („2'»+l_ xx xt Ex. = 0. „n+lfl g .. bn \ homogeneous in x and h h y. zNl and by 1. Intersections of Two Curves. .. the summation extending over the n intersections P of the line and «ic. z 2 . ••X Sn+1 / y1 yi . 0. Let their equations be ~2 a zn + a^z".y = ax + 6. a a a. a.. x.. the equation of the nic being y n — (ax + b)y".. z 3n+1 = 0. Then. *s«+i)» « being homogeneous in z of degree n — 3. Multiplying these equations respectively by 1... then 8. 1) to the intersections of the two curves. (i). the intersections of the tangents whose points of contact are given by f(t)+k<p(t) = 0. 0) where ar and b T denote homogeneous polynomials of the rth degree "in x and y.**) (*sn+ii ys.+an .1 + . z^ 1 we obtain n + fi linear equations in .10 HIGHER PLANE CURVES 16 [Take the conic as the locus of (f. nic at P. = 3 is well known. Two curves of degree n and iV intersect in nN points. in+1 xyzu in If any line meets the curve x + y 1 "+ l + z in+1 2 i)> (*2> jfe.<p(t)} + (tl'). 2 2 2. from which these quantities may be eliminated.1 +I). 1 X.. we have For instance. the line is ^# + py + »» = 0. y.1 + a 2 z n +. a. x Su+1 + y y2 [ . 1). z. where /and <f> are polynomials of degree n in t. lie on the (n — l)ic obtained by eliminating t and T from The case n Ex. .y Sn+1 5"+l)/(X2'1 +l_ v 2 .. (ii). Suppose the equations of the curves are given in homogeneous coordinates. bi \ h = 0. It.... Show that 2cot<£ is constant and 2 (p sin 8 (p) _1 = 0.. if to = 4 and If = 3. whatever h may be. & .. Differentiate this relation twice with respect to x... . for it is readily seen to be The lines are nN~ in number for . z. z2 . 9. xx x2 [If . <t> A [For a given value of x we have 'S..
For other methods see Ch.] and v = are polygon of h If a nondegenerate »ic 2 or else n sides. 1) to those intersections of the curves which do not coincide with (0. . 0.s A r^ and by 1. 0. Higher Algebra. It is impossible for every line to meet a given nic in . = ^ * For a detailed discussion by this method of the intersections of two curves see Segre. 2. = 0. 1) is rbN—kK — Y in general. 1). = where u = and v = 0. n has the symmetry of a regular k. 1). bK are iden.] Ex.* Ex.e. 150. [The only real points on m + (1)* v real wics. . and parallel to any tangent will meet it in n—2 An «ic with unreal equation cannot pass through more than w 2 real points.. Multiplying equations 2 (i) n~k. = 3. Similarly.. z.. For example. N fc It is readily seen that the number of intersections of the curves other than (0.<v (n — k)K+ (N— K) n = nJST— kK in x and y. § 3. if a & and b% have r factors in common. n real points. z2 . z. 1).. b2 A typical term of the determinant just written v *> fc is which is of degree Hence the curves meet in case that ± v. If in this a and bK have a factor in common. we have roots of (ii) occur in conjugate pairs. we say hK of the intersections of the curves coincide with (0. kK+l intersections of the curves coincide at (0.z and eliminating as before. ±b na N n . tically zero. z . a h _ x and b \.. 1. 0. .. we get an equation giving the lines joining (0. xxxvi (1898). In order to observe the convenient convention that in eveiy two curves of degrees n and meet in nN points. the factor is a factor of the first column of the determinant just obtained..x by 1. i. nN—kK points other than (0. a and a x = 0. 1). Suppose that a a v . pp.. 0.b = 0. 0. VI. N "2 '3 "'i a„ 6j &i a3 b2 b3 aA b3 = 0. are the real intersections of u Ex. 3. 0. both curves pass through (0. 0. 1).. For complex '" 1 . ^ . We note is that the number tiN— 2r. .. and Ch. . if n = 4. § 69. kK + r intersections coincide at (0. If a and b are zero. IX. For the general theory of the eliminant (ii) see BDcher. 1). r being of real intersections of the curves zero or a positive integer. Giomale di Matematiche di Battaglini. § 2. 0. " [A line adjacent to real points at most.I 7 INTERSECTIONS OF TWO CURVES is 11 a typical term of the determinant just written The theorem is therefore proved.
[(i) Use polar coordinates. P. assumed that the nic does not pass through the circular points. by analogy with the wellknown 1 case in which n The n2 fixed points through which all curves of the pencil pass are called the basepoints of the pencil..] . ' ' ' ' ' § 8.. and v n passing through the «2 intersections of u The family of curves obtained by taking different values of k is called a pencil' of nics. which touch a given [Take z Ex. (iii) The same is true. Ex. in Ex.] Ex. The sum of the angles which OPlt 0Pa . if we replace Cartesian tangent by Cartesian normal or radius of curvature ' (iv) The centroid of all points on a curve at which the circle of and radius o at an angle a is curvature cuts a circle with centre independent of a and a... The coefficient of se** only involves the coefficients of the — 1 in F. u + = = = = . with the tangents at (iii) Plt P 2 . The number of curves of a pencil of «ics in general 2 (« — 1).. y) = N N Deduce that (i) The centroid of PJf P.. [Consider the intersections of the wic with circles concentric with the polygon. If u degree. the 2 n — 3 in general. (ii) The centroid of all points on an algebraic curve at which the Cartesian tangent is of length a is independent of a. is zero... terms of degrees «.. y) of degrees n and respectively...] 2. v = and k is any constant. 5. If = F (x. we obtain in general an nlf in this equation equation for x of degree nN. .: 12 If INTERSECTIONS OF TWO CURVES I 7 n is not a multiple of h. If is in Ex. = 0. . 4 is independent of a. fixed line is (ii) independent of both a and the position of 0. number [One touching at the basepoint and 2 (n — 2) others. OP2n makes 0PU at with any . Ex. of curves = as the line. They are not necessarily all distinct... 4. 1 the line passes through a basepoint. The coefficient of x degree n in / and degree only involves the coefficients of the terms of _I in F..„. (i) P.. Pencil of Curves. A circle with centre and radius a meets a given Mic at P„ P2 . n — 1 in / and degrees N. (ii) and (iii) follow at once from (i). the reie passes through the circular points. line is 1. are the equations of curves of the nth is a curve of degree lev 0. Show that .. The sum of the tangents of the angles which 0Plt make 2 . The sum of the polar subtange^its of the curve Pu P ••• is zero.] It is N and we eliminate y between two equations f(x.
13 Tangents. touches the curve at P. cubes. PQ and Q (Xx' (x. y) = (»*& + <*rt$ = If the coordinates of ° any point of a curve are given in the form x=f(t). Xz' + fiz) = where j—7 means the result of putting a:' for a. &c. 3'for. <f>y)x + (+f. of dt. y. The line PQ . z') is of degree n. this line goes through the given point and the consecutive point..in^^divides . z) : in the ratio any point on a curve f(x. + fdt). y' for 2/. where A + fi = 1.z) = is any other point. Xy' + py. if + /iy.y = t is 4>(t).z = +{t). The P Ix oz which is therefore the equation of the tangent to the curve is at P.<f>(t + dt). the tangent at where called the parameter of the the point is 0. &c. whose coordinates are (W = + </>'dt. is fi Xy' + fix. second root is zero. This or lies on the curve f(Xx' + fix. Xz' + jjlz). we get the tangent at any point of the curve whose Cartesian coordinates are + Suppose now P (x'. If the equation of the curve in Cartesian coordinates 0. The point X.I 9 TANGENTS § 9. (/ (* + dt). This equation in fi/X gives the ratio in which the curve PQ.Vf)y+(f#ff) * In fact. z') is not on the curve. point. the tangent at (x\ y') is f(x. condition for this is A PQ ° <>y for lies on the curve. One root if is zero. y. Suppose dividing P (x\ y'. y'. y\r(t + dt)) or (f+fdt. neglecting the squares. If we put z — 1.
each conic corresponding to itself in the perspective. such that OP . 1) is in/is zero. n of z If the «ic f(x. circle + y2 = a2 .] Ex. Find the lengths of the tangents and normals from any rJbint to a curve. 2 J = 0. 1). * Or a sphere. u \* + 3u 1 \ 2 li + 3u 3 \n 2 + u 3fi 3 = 0.1 J § 10. Show x that two perpendicular tangents to = a (2cos£ + cos2i).z) = passes through (0. the constant term in zero and the tangent at the origin is obtained by equating to zero the terms of the first degree in /. Ex.3 «. 4. if the loci are not coplanar. z. constant. The tangent at any point P of yx* = a? meets the axes in A. and put the equation of the curve into the form f(r. where u r is of degree r in x. y = 2tr/ (1 + t''). and writing down the condition for equal roots. and they meet the cubic again in six points on a conic.OP' — W.y. 3. y. dr or /= dt 0. the resulting equation that of the tangents from (x'. (See § with O as vertex.BQ are constant. = / Ex. 0. Ex. y'. The tangents meet the two conies again in twelve points on another cubic also touching the tangents.14 TANGENTS we write down 1. in general. we have The tangents from any point + 4m! 3 ) + m 2 2 (4i( z . The two cubics of Ex. 8. Ex. [Putting (i) in the form (n = 3).t) = 0. a. Prove that the ratios AP. 1. if the equation If is (i) in /jl/K has equal roots. z') to the curve. Inversion. ' . Eliminate t from/ = 0. 2. . the locus of P and the locus of P' are said to be inverses of each other with respect to a circle* with centre and radius k or more simply inverses with respect If 0. Ex. the condition for this. being fixed. The axes of reference intercept a constant length on any tangent to x%+y§ = ai. 0.PB. 2 y = a (2 sin? — sin2<) meet on the Ex. and the tangent at (0. The sum of the intercepts made by any tangent is to xi + yi — ai on the axes of reference 5. 3. I 9 will touch the curve.] Ex. 7. 1) and u2 z* + 2xy. 6 are in plane harmonic perspective u 3 (h 2 w 3 ' 6 j^MjKj i/ . P. [Choose the triangle of reference so that = [Take the point as origin. The two conies are u 2 = and 4m m 2 = 3%2 and the other cubic is w 2 m 3 — 6j( m 1 m 2 + 4u 1 3 = 0.) O is (0. the coefficient obtained by equating to zero the terms involving s" _I is Iif(x. y) passes through the origin. P' are collinear points. Ex. and / = 0. B and the curve again at Q. 6. 0. to a cubic are six in number Their points of contact lie on a conic. ' to '. where x = (lt 2 )r/ (l + f).
y). the fact that 0. P. its inverse. Hence two curves cut at the same angle as their inverses. OQ'P' are equal. Let the tangents at and P' to one pair of inverse curves meet in and the tangents to the other pair in K. If now the two inverse figures are coplanar while is taken as the origin of rectangular Cartesian axes and P is the point 2 2 2 2 2 This follows at once (x. inverse is a line through the other circular point. P' are collinear and OP OP' = k 2 from P H PH = P PK . it is evidently its own Also. and the angles HPK. HP' make supplementary angles with = P'K. triangles HPK. or a straight line if The inverse of a straight line is in general a circle through 0. PQ = k 2 P'Q'/ . since the tangents HP.I 10 If INVERSION 15 Q' are points on two inverse curves. from the similar triangles OPQ. Then. Again.0Q'. y) = is the equation of a plane curve. Hence the inverse the inverse of is x + iy k 2 =c and may show is = c(x—iy). Hence. OP': OP'.OQ'. For instance. OQ'P' (Q not necesP'. OF . HP'K are equal. we see that the tangents at P and P' to the inverse curves are coplanar and make supplementary angles with OPP'. P' is (k x/(x +y ). k'y/(x + y )). a plane if We have a similar result a sphere. for three dimensions. that the inverse of a sphere with respect to lies on the given sphere. if the line passes through 0. Hence the angles OPQ. or . k 2 y/(x2 + y 2 )) the equation of the inverse curve. HP'K are congruent. f{¥x/(x 2 + y2). Therefore the P'H and similarly OPP'. P and Q and 0P. 0Q\ which enables us to express lengths in one figure in terms of lengths in the inverse figure. Suppose now that two curves meet in and the two inverse curves in P'. if the line passes through a circular point. .0P'=0Q. is if f (x. the inverse of the circle = is Ic* x2 + y 2 + 2gx + 2fy + c = + 2k 2 (gx +fy) + c(x 2 + y 2 ) = 0. in its of a circle with respect to a point lies on the given circle. Making Q consecutive to P and therefore Q' consecutive to P'. sarily consecutive to PQ Hence P) we have :P'Q'=0P: 0Q' = OP. In fact. plane is a circle. But.
3. According as G 2 + 4iH 3 = a 2 (a 2 a32 is — Qa^a^ + 4a a23 + 4a1 3 a 3 — Sa^a 22 ) negative. [Let Q be a point of the curve near P and Q' its inverse. 1. n= 2. . 5ai + a 5 (a5 5a1 )} ) {a + 25 in the cases (a a. The typical equation of degree n is ~ a xn + »(?! a^.a 4 ) 2 + 16 (a x + a 3 ) (a a2 + a x a^ % + a2 {a (a = 2 ) 0.\. Theory of Equations. This projection is called stereographic] of Ex. I 10 The inverse of a circle with respect to a point not in its plane is a circle.. a3 a.1 + n C2 a2 xn 2 + . [The projection is the inverse of the original figure with respect to V. one real root. For convenience of reference we insert here some wellknown results in the theory of equations. the cubic has three real roots. 3. «2 a. Now let Q approach P. for the intersection of two spheres inverts into the intersection of two spheres. If a curve cuts Ota at P. If a.. 5. 2. .] § 11.a 16 } INVERSION . + 2a a3  +10a 2 +5a4) (a — 5a4) + (a6 + 10a3 + 5 %) (<x6 — 5 a 5 a4 — 2a5 a2 = x) 0.. the inverse with respect to a point P on Oa> is at oi. that circles project into circles and angles are unaltered by the projection. the corresponding tangent at <o of the inverse curve is the inverse of a P. positive. A sphere is projected from any point V on its surface on to Show the diametral plane perpendicular to the radius through V. [Use the fact that all circles through to a circle cut this circle orthogonally. The quartic equation (n 4) has two invariants = I = a ai —4a1 a3 + Sa2 2 1 J= "0 ""1 ""2 «1 a. 4.. • The cubic equation (n = 3) has two invariants = a 2 a3 — 3a a1 a 2 + 2a1s *H = a a2 — a^.. Then «Q' and a'Q are inverse lines. The product of two roots is —1 if — °> a o (ao + 3a2 ) + a s fi ai + az) = 0> {a + 6a 2 + a t ) (a .] two points inverse with respect Ex. Ex. + an = 0. or zero. . Prove that a circle and two inverse points invert into a circle and two inverse points. or two equal roots. a>' are the circular points..
y = yx. 0). 0). (7. y.I 11 THEORY OF EQUATIONS — 17 of the If is the crossratio (a fi) (y — 8)r. j8. J(l + /— 3). y = dx are said to be respectively 'harmonic' or equianharmonic in these two cases. 8 of the quartic. 0). = l. = 0. (8. note the cases = 0. — 8) (y— /3) We If If * 2 3 {(0 + 1) (02) (04)} / = 27 (f4> + l)>J*. y ' . 0) and the pencil of lines = fix. J"=0.(a roots a. 1=0* y • The range of points (a. = ax. the quartic has two equal roots and I 3 = 27 J 2 . and if = § (1 + V^8). —1. (3.
) + . as m m zero. or the line becomes the tangent at the origin (Fig.. &c.. Then = . If the origin lies on the curve. i. Fig. in one point coinciding with the origin (Fig.. 1). 6 + 6 approaches 1 a second root of the equation approaches zero. the equation must be satisfied by x = and y = and therefore a = 0. Inflexions. The equation now becomes = b x + b 1 y + c x2 + 2c 1 xy + c 2 yz + d x 3 + Sd1 xi y + 3d2 xy 2 + d3 y 3 + e x i + The line y = infix meets the curve where = x(b + b 1 m)+x 2 (c + 2c 1 7n + c 2 m 2 ) + x3 (d + Sd1 + 3 2 m2 + d3 m 3) + xi (e + . is made to vary. a second intersection of y tnx with the curve approaches the origin. Suppose that. 2). Suppose that the Cartesian equation of a curve of degree n arranged in ascending powers of x and y is = a + b^ + b^y + c x2 + 2c 1 xy + c2 y 2 + d x 3 + .. CHAPTEE II SINGULAR POINTS § 1.d . since any line through the origin meets the curve . e.. m . One root of this equation in x is always zero as was to be expected.. l. .
3 and 4). The origin is >. * 2 b x + bx y is a factor of c^j? + 2c t xy + c.2 y .ni INFLEXIONS. when 6 + bx fn. called a point of inflexion (or simply an inflexion) in this The tangent at the origin is called an inflexional case. the tangent meets the curve in four points coinciding with the . we have also co + ~ c i m + ct m* = 0. It is evident that 6 + 61 to is a factor of c + 2c 1 what is the same thing. 0. or. 3. Then the tangent at the origin meets the curve at three points coinciding with the origin (Figs. m+c i ri\ : If we have also d + 3dl m+3d2 m2 + cl3 m? C 2  = 0. = = we obtain Suppose that. Fig. namely. ETC. = 0. 19 Eliminating ni between y mx and b„ + bx m. tangent or tangent of threepoint contact. the equation of the tangent at the origin.
The curve crosses the tangent at the point of contact if r is odd. +1 + ... ETC. The follows last statement in the theorem may be proved * as The curve may be written b x + b lV = {n. + tangent at the origin (b is x + b iy) + {b* + b*f (u r + u r+1 + .. y) of the curve on the 1._ ). . contact. point of undulation The tangent is called a tangent of fourpoint is In this case b x + b x y a factor of both . the tangent meets the curve in r points coinciding with the origin.. and is called a tangent of rpoint contact '. + i'. where b x + 6j y = 0.. The II 1 in this case. when x and y are small and approximately in the ratio b x —b But ur evidently does or does not change sign with x and y. + Vl + . but does not cross it if r is even. 4. + v r_ 2 (b 2 + &. ) : . (r — l)*11 degrees. . INFLEXIONS.«)*... If the terms of the first degree are a factor of the terms of the 2 rd 3 rd .. ' Fig. + U. and has therefore the same sign as ur + (6 2 + b^)i. . The perpendicular from any point (x. 1 2 where u and v% are homogeneous of degree k in x and y.+v n ) (l + v + v 2 +.. origin is sometimes called a. 20 origin. according as r is odd or even. .: . The truth of the statement may also be seen geometrically ) = + un f (1 ... the terms of the first e^ + Z^xy + c2 2 : degree equated to zero give the tangent at the origin. 3 2 2 s y and of o] x + 3d1 x y + Sdi xy + d3 y These results can be immediately generalized and we have Ifa curve passes through the origin.
3 for the case r 3). ' ' On the same branch of each curve . if they may be considered as the limiting case of two curves meeting one another at r points * very close to are said to oscuP. .m3 2 + a. and y = infix meets the curve where = x l (c + %cx m + c 2 2 ) + x 3 (d + 3d1 m m+3d m 2 + <£. ) roots of this equation are zero for every value of m. between this equation and y Eliminating the equation of the tangents at the origin m = mx. + 2c^xy + c 2 y 2 + d x 3 + 3d 1 x2 y + 3d2 xy 2 + d3 y a + 6q x + .. . if the radius of curvature of the curve is a maximum or minimum at P. In this case the curve has two real branches through the origin. osculating circle called the circle of curvature '. at each point of a curve there is an late at P. 5). we have c^b'Z^xy + ^y 2 = 0. Such a tangent may be considered as the limiting position of a chord joining the double point to another point of the curve. L Suppose that in = c aj z The curve is now § 1 b and b are zero. line through the origin meets the curve in two points coinciding with the origin. The line y = mx is such a tangent if c + 2c 1 m + e 2 m z = 0. The tangents may be real. whose centre and radius are the centre of curvature and radius of curvaThe circle of curvature has ture of the given curve at Pfourpoint contact with the curve. The definition of ' rpoint contact of a line and a curve may be generalized.4 (e +.II 2 DOUBLE POINTS 21 by considering a tangent of rpoint contact as the limiting position of a line meeting the curve in r points close together (see Fig. Any two curves will have rpoint contact' at P. Curves having threepoint contact at For instance.. * ' . The origin is called a Two Hence every double point..) + . t See treatises on Differential Calculus '. when the origin is called a crunode (Fig.. There are three cases to be discussed. The tangents at the origin are defined as the lines meeting the curve at three points coinciding with the origin.f = ' ' P ' ' ' ' ' § 2. when this point approaches the double point. see next section. .. Double Points.
The origin is a real point of the curve. but there is no real point of the curve adjacent to the origin. e. ' isolated point ' or ' for cusp is obsolete. 7). a factor of As tangent is we show that. ' Fig.. The tangents may be coincident. to a double point which is not a cusp ". when the origin is called a cusp (Fig. if one factor of c x + 2c x xy + c2 y2 is d x s + Sd l x2y + 3dz xy 2 + d3 y 3 the corresponding an inflexional tangent. 7. fiecnode in this case (Fig. The origin is called a 2 . 6. The term spinode for crunode. If both tangents are inflexional tangents.* ' '. when the origin is called an acnode (Fig.22 DOUBLE POINTS 112 The tangents may be unreal. 8). The word node is equivalent to crunode or acnode i. in § 1. ' con . so that the terms of the second degree in the equation of the curve are a factor * Other nomenclatures are ' node ' ' jugate point' for acnode. Fig. 6).
9. 0. If the highest power of z which occurs in ~ the homogeneous . shows that intersections of the curve with the line y = mx equation of a curve of degree n is z n k the point (0. 1) is a multiple point of order lc. 1). if Similarly degrees. 0. . a curve are of degree If the equation of the curve is given in homogeneous coordinates.: : . 0. points (at least) coinciding with the origin. and the tangents at (0. 9). 113 MULTIPLE POINTS is 23 called a biflec of the terms of the third degree. which meet the curve in k+\ points (at least) coinciding with the origin. the origin node (Fig. c 2 are all zero. § 3. . 1) are . In general we have the result If the terms of lowest degree in the Cartesian equation of k. Any line through the origin meets of the curve in k points coinciding with the origin. except the three tangents at the origin obtained by equating to zero the terms These tangents meet the curve at four of lowest degree. Let us suppose the curve is of degree n. Fig. except the k tangents at the origin obtained by equating to zero the terms of lowest degree. fourth. second degree is one of the factors (or both) of the terms of the a factor of the terms of the third. . and passes through the point Exactly as in §§ 1 and 2 the consideration of the (0. If c c 1. . so that the terms of the lowest degree are the terms of the third degree. the origin is called a multiple point order k (a kple point). Multiple Points. a very slight modification is necessary. the origin is called a Any line through the origin meets the curve in triple point three points coinciding with the origin.
[n = 4. crunode at (1. Ex. r = 3. 24 MULTIPLE POINTS II 3 n~ h obtained by equating to zero the terms multiplying z ~ in n ' k z n . i. Ex. 7.] Trace roughly the curve y* + x (a:* . Ex. [Acnode at (1. x y\ a'y i i and trace roughly the curves y> (ax) = x\ 1 xi + yi = axyi . cusp at (0. r = 4. Show that a conic can be and passing through R. Ex. 3 2 Ex. Three tangents of npoint contact of an nic are taken as the Show' that the equation of the nic is of the form the + + (% + cz)" + (cz + ax)" *»«««» + ( ax + sign in the ambiguity being taken if n W = is oV + bnyn + *V odd. = sfun _ r geneous of degree t in x. E. If the terms multiplying z have a factor in common. 7 [The equation has a*" as a factor when we put x — . 4. and PQ meets the curve again at R. = 3.] Ex. 12.z n c r . If n is odd.2. 6 n = 3. triple point. 9. 1. 2 Investigate the nature of the origin ay = x\ + y*)* = ay 1 = x\ a*(x l 9>).] Ex. 4. 8. contact with the curve at P and Q. Trace y* = x t {x + k) when k = 1.] Ex. 3. Trace y + x{xk){x2f Trace y* l i 3. 0). to 10 are special cases. cusp. Ex. triple point (solve for y to trace)..k ~ x . Q. The line joining two real inflexions of a cubic passes through a third real inflexion. 11. drawn osculating the quartic at D. Ex. . S. An «ic has the sides CA. The line through three real collinear undulations of a quartic passes through a fourth real undulation. B of a quartic meets the curve again in D. Show where u t is homothat its equation is of the form xyu^^. The line joining two undulations of a 4ic meets the eurve Show that a conic can be drawn having fourpoint again in P. Ex.2) = 0. biflecnode (turn into polars to trace). r = 3. 5. 2.. either the points are collinear or the three lines joining each to the intersection . Transfer the origin to each point in turn. y. . S.. quadruple point (solve for y to trace). The line joining two inflexions A. 6.] Ex. the three points of contact are collinear. 10. B meet the cuive again in P. when A. The tangents at A. the equation. 5.. 0). [In Ex. An »ic has three tangents having wpolnt contact. Q. undulation. A and B being the points of contact.1 ) {x . 0. If n is even.] Ex. E [n = 4. 1. 0. triple point (turn into polars to trace). (a* 1. z. and either sign if n is even. + xi (x' 2x + k) = (a— 4) s =0 when fc = 1. cusp. 1. or y = 0. CB of the triangle of reference as tangents of rpoint contact. sides of the triangle of reference. 0). !s 3 s [Inflexion. 0. = x\ ay* = x\ (x' + y y = ay{3x y% a"y + x = 2a. the corresponding tangent has (r + 2) point contact with the branch it touches.
2 in Ex. 0. An nic has rpoint contact C. Conditions for a Double Point. Ex. Ex. 20. 13 w = 4. Y) be any point on the curve whose Cartesian equaTransfer the origin to this point. and the is f(x. 15. 17. There are considered as an equation in z. or tion 0=f(X.] Ex. XIX. equation becomes f(x + X. 13. Ex. 6ic. Show that its with each of x = and y equation is of the form = at [See Ch. see Ch.] Ex. y) = 0. An nic has rpoint contact two distinct points. or 7ic lie on a conic. IV. Show that a cubic touches the quartic at the six points in which these two lines meet the curve again. 4. of reference ABC at B Show with the side BC of the triangle that its equation is of the form Ex. and 25 of the tangents at the other two are concurrent.Y) + x¥ +y V \ x DX 2+ y l>XlY +y 37V . The tangents to an nic from the &ple point (0. 18. ? . § 7). . I. [In Ex.II 4 CONDITIONS FOR A DOUBLE POINT = 2. 19. A line is drawn through each of the points of contact of a bitangent of a quartic (a line touching the quartic at two points. &c.] contact of three bitangents of a 5ic. P . Let (X. only the former alternative is possible. What are the corresponding theorems for a 54c and 6ic ? Ex. 19. Enunciate a similar theorem for a sextic having threepoint contact with a line at two points. If the six points of § 6. 1) are found by writing down the condition that the equation of the curve. Ex. 16. § 2. [Use Ch. [Use Ex. should have equal roots. The tangents at three collinear inflexions of a quartic meet the curve again in collinear points. Bt and P2 Show that the conic PiQJjJliItz passes either through P2 or through the harmonic conjugate of P2 with respect to B and C. 11. y + Y) = U. n(n— 1) — k(k+l) such tangents in general. 17. § 4. r=2. the other intersections of the bitangents with the curve are respectively collinear.] Ex. 14. If the tangents are concurrent. quartic at Pt and 2 Q t and Q2 . Discuss the cases n 3. 18. lying on a conic. What are the corresponding theorems for a 2nic having npoint contact with each of three lines at two points for a 6ic with three triple tangents. Three bitangents of a quartic form a triangle ABC whose sides touch the. the base points of a pencil of 3ics.
e. y) of/ with respect to must vanish. according as > ' \oxoy/ Similarly. Xz + /iZ) i. Xy + fiY. Y. let (X.z) be any other point. z) = 0. If (x. = 0. acnode.. Let P (x. oX~~ oY~ oZ~ ' . Z) be any point on the curve of degree n whose homogeneous equation is f(x. y. two roots of this equation in X/fi are If zero for all values of x y z or : . if (x. ^ means the If the result of putting X for x and Y for y ^.&c. and the tangents at the new origin are x )x 2+ ^oXo7 +y oY* Hence. y) J = 0. x and y up the partial derivatives to the order k— 1 inclusive Again. y. This point lies on the curve if f(Xx + ixX. The double point is oy ox a crunode. : P0 in the ratio X. ox' new origin is a double point on the curve. or cusp.26 CONDITIONS FOR A DOUBLE POINT II 4 where in .Z) + ^x(x^ + y^ + z^) ^zix : + 2zx +2xy +Xnf{x y iTJY) + ' z) ' = °" is a double point. and let Q be the point / \x + /j.Y. ^f(X. y) is a double poind on the curve f(x. with a slight change of notation. is < ' ° ox* all oy* a &pie point.X V dividing X + /i /x Xy + fiY Xz + pZ \ X+ X+p ) /J.
These are all satisthe points (0. ~ 2a)(x + 3a)a? = 0. / = yH/JtayYlx* {x + Zaf = we obtain (x + 0. y. Transferring the origin to each of these points in turn. rj. (2a. 1.4 (a. we see that the first two are triple points. s 4 3 i . . £ being current coordi = = 0. Find the double points of x* (i) (ii) (Hi) 2af. z) is a double point on the curve f(x. Hence If (x. acnode. 4ta). We may deduce the result for Cartesian coordinates by putting 2=1.] (y2«)(4»y)V fied at Ex. or coincident nates. Therefore the equation of the tangents at is *X* +y 3P + " *Z* + ys *7*Z + *Z*X +2x ywjT=°= 0. or the result for homogeneous coordinates may r.3«y . since by Euler's theorem on homo Three roots of the equation in \/fi are zero if P lies on a tangent at 0. y. 2. The reality. be deduced from that for Cartesian coordinates by replacing x and y by x/z and y/z. £) geneous functions = 0. unreality. [Equating to zero J J. or ( = 0.0). y (4ay) 2 = 2a.: II 4 It will CONDITIONS FOR A DOUBLE POINT be noticed that 27 implies /(Z. £. and the last is a crunode. (3a. unreal. F. Find the multiple points of .2aV + a* = 0. = 0. 2a). ^4a* 2ays + 4aV + 3«ya =0. 0).4a). Ex. are real. (3a. or cusp. z) The double point the lines is ix ~hz <>y a crunode. or coincidence of the lines is at once determined by considering their intersections with  0. (0. the next two are cusps. xy + 12a (3x + 2y) + 108a = 0. + 3a). according as * lx* + r> y + ^^+^i^^^J^/^teJr .
5). e. Cusp at (1. 0. 2yz(y'1z + yzi 'S.2 2 = a 4 ) ) (ix) (8yx — a) 3 = 2l6axy. 2 (viii) (a. &c. 1 + y 1 .] Node (8<jco s . + 6). [(i) (ii) (iii) Crunodesat Crunodes at (a. +6). Cusp at (3. 0. Cusps at (3a. c2 /a). 0. § 3. 5. an acnode at an acnode at (1. When has / = y z .Wfx. 0). ) . (y z 2 + yz* — 4x* + 2xyz) (1. 0.0). . 0. = 0. For what value of k have the following curves a double point ? (i) (ii) s*x xP = y(yx)(ykx). 4). For these a by z.a). . x 3 + y 3 + z3 + 6 kxyz — 0. k k = = . is 2g 2 x + Bgs z \ = 0. 0).yz 2x3 + 24:xyz)= 0. x +xy 3 + y i — kzy 3 — 2x'*yz — xy*z + y 2 s* = i 0. is Hence or >0. (0. 0). 3. a). i.a )V = (^ . 1. Or we may project the curve as in Ch. 0). ±a) and the circular points. + b/y* + c/s1 + 2f/yz + 2g/sx + 2 h/xy = 0. ° gives 3 + g^xz* + jr s a> = a double poinW? 1 l£ = = §5 iC = yg = 3 1 ^9^ = y +^s^+Sg g s t 3 zi = o. See Ch. a). 5. (a. k = 1 . cusps at (oo. 0). Saw). nodes at (1. &c. . 1.a2)3 + 27 x*yW = 0. (0. latter replace Crunodesat (2a. 0. a crunode at (1.4a. 1) 2 .3^6). (x) x 2f 4:a (x s + y s + 18a 1 xy 21^ = 0.] 0. —a/8). 1. ' . 1). 0. Unreal nodes along x = + y. 0). 3a). X. 1). (0. c 4. (1. 0) are the coefficient of x 1 equated to zero. Ex. (iii) (iv) (v) [(i) (ii) + if + z3 = k(x + y + zf. 3a). 1). 0).. 1). These are only point at ( satisfied if is y = 0. (1. Then the tangents at (1.o 2 a + (y» . (y) (X* ft . (xi) y(a. (viii) Nodes at (0. (2a. (0. Triple point at (1.] the required condition g^^21g If this holds. + 3) = 4(4a 32/)(2a. Ex. (v) Cusps at ( + a. Baa 1 ). (Baa. (x + y + z) 3 + 6 kxyz = 0. 1). nodes at (1. The curve is the fourcusped hypocycloid '. (0. (i) a/x* Find the double points of . 1. 0).28 (iv) CONDITIONS FOR A DOUBLE POINT 4{xl) 3 + (yBx + 2) i ) II 4 = 0. (vi) Cusp at (c." . Fig. oo ). . (ii) (iii) . I. (iv) ' (ix) (x) (xi) at (a. [(i) (ii) (iii) Nodes at Cusps at (0. y 2 = 0. _^(vi) (Baxt/ + 2x3 + c 3 i = 4(ay + x 2 ) i a (vii) (a.' = 27 g^. . 1. 0). (a. there is a double — \g$. which a crunode or acnode as g s <0 Ex. (vii) Nodes at ( + a. 0). (0. (0. 0). 1.
where the to x. 2 = 0. r = node at the intersection [See Ex.. if p approaches . = . the coefficients are made to vary. = are = = and js = an »ic with a are 0. 3 denote partial differentiation with respect [For the cusp. 7. if p p^. 1. 8. 0) . z) should be a node of are fa = 0. = t . = 0. We now apply this result to the theory of plane curves.. 0) . Suppose that in the equation ~ p^xn +p1 xn 1 + p 2 xn 2 + .. If u » 0..If Ex. y.. the tangents at O form an involution. the becomes indefinitely For the equation where xg = 1. (1.8. .. .. If every curve of a pencil has a double point at O. jp«2 + 2g'M« + n. (0.. . Similarly. = 0. are zero. . 0. = 27 ab. Of a pencil of .1. is Po+Pit+P2i*+"+2>n£ n = 0.] Ex [a 6. one of the roots of the equation large. .. 2. r+1 roots of the equa.] have a node in general.. 0. is satisfied [/== Ex. Sab).. — tion in x become very large.= have a . fcple point at P and . «ics 3 (n Ex. Now if p > 0. C = 0. 1. S. 10. are any other curves. up into three lines... b.1 )/.. ^ = e' = = = at P. 1). A node in addition if k has z s <? = 1 or 2.. . one value of £ given by the latter equation tends to zero.] when some of the § 5. In symmetrical form the conditions that (x. S = 0. &c. 2 /= CjS. which gives We leave to the reader the modifications required quantities fu /23 . we get /22 /S3 =/25 2 . y. + C S has a fcple 2 + + C. ) = 0. Points at Infinity._p r all— »0. 6 = 0. ca. . yi=/2 =/3 = o. z. 2 Si = 0. (v) Always a cusp at (0. and that it should be a cusp are 2 3 /= =/ =/ f = JuJs> —/12/13 = /2z/3i ~~fafn = J33J12 ~/31/32 — suffixes 1. 1).Sr = point at P. . C. c = point? (ax + by + cz)xy a double a (be. limit zero. (m2)ics. If 0. For what values of * a.11 5 (jii) POINTS AT INFINITY k 8ft 8 29 (iv) = f an acnode at (1. (»l)'(/i a /i/u/2 ) =/i 2 (^/ii+y/i 2 Then + s/is)/ii(^/2i + 2//2 2 + s/23 .] lines is 9. Hence one value of x given by the former equation becomes very large. The cubic splits . 2 of u and v 0.] Ex. Hence fx //u = /2 //M = /3 //„ = »// (» . +p n = Then.
.y for then nx One = mx =m .x x..+kmn l x71 1 m = mj) 2 mn ) ) + . i. e. m„... Hence any line parallel to y = ra^x meets the curve in one and only one infinite point. degree n in n * m^ The reader if a.. are the factors of the terms of highest degree.. for which p + (% 2 ) (m1 It follows at once that any projection of the curve touches the projection of I at the point where it crosses the vanishing line. is ' . In general y = + c meets the curve once and only once at infinity whatever finite value c may have. but t More accurately. ... (y —m and y we obtain n ~ 1 + bxn . . .. = 0... 3 . = 0. = 0... . Next suppose mx = 2 ^ m3 mi Now y = m^+c meets the curve where ~ pxn 1 + {g + rc + (m1 — 3 ...™" + (q + re + 1 1 . convention every straight line meets a curve of points and so throughout.* The line y <mx meets the curve where ~ ' (m— (m— ). y root of this equation becomes very large... ... ' meets the curve in exactly n — 1 (finite) points it is convenient to keep the.. . sian equation in descending powers of x {y—m^x) Writing Carte (y—m2 x) . while the line at infinity meets the curve twice on y = m^x.. . . § 4). ' ' . if the line as it turns about the origin approaches one of the lines y ="m 1 x.. I is an asymptote (Ch.. (m — m x) (m . + (n—l)Jemj n 2 . This evidently will easily make the necessary modification in the argument a factor of these terms. . + kyn ~ 1 n x) + ax + Axn * + Bxn 3 y + . {my^m^) c 2 }xn ~ 2 + i . m ) .. . {mL — m^c} a. .. First suppose not equal to any one of 2 2 The line y = 1 x + c meets the curve where m m m m . we write p = a + bm + l r= + kni^q = A + Bm 1 + b + .2 y+. y — 2 x. m m .f with the exception of the line I — 3 ) (uij — m„) c = 0. . . ...m 30 m POINTS AT INFINITY .. {p + (m1 — if m 2 ) . where y — m.=0. II 5 its Suppose we have a curve of degree n. . (m—mn —> 0. I.m2 ) .. Let us now discuss a few cases in more detail.) xn ~2 + . if rather The terms of highest degree in the Cartesian equation of a curve equated to zero give the lines joining the origin to the points at infinity on the curve. form up this result in the convenient.. ) We may sum inaccurate... (m— xn + (a + b<m+ .
ay2 = x s has an inflexion.m5 x + c meets the curve at only one infinite point for x every finite value of c. . If y — m^x + c meets the curve at two and only two in =m m m m ^m . {m m„) c 2 meet the 3) 1 curve in three points at infinity. y Suppose now = 2= 3 If the line 1 4 . [Use § 5. these two . however. 6 . to obtain the equation of a projection at/y as in Ch. Then every line parallel to y m^x meets the curve in two points at infinity.. a? 3 Trace these curves. y + ix is a factor of the terms of highest degree in the equaSuch a curve is called a circular curve. . Another method of the curve by replacing x. p 0. 1. infinity at which lines are a pair of parallel asymptotes. so is y — ix. the curve has an inflexion at infinity along y = m^x at which the line at infinity is the tangent. n— = = —m — — = m^ these Hence the curve has a double point at two lines are tangents i. I. If. if point at infinity. § 3. y — m. the curve has a double point at infinity along y x at which the line at infinity 1 is one tangent. bicircwlar and so on. x x where it is a factor of the terms of degree 1 as well as of the terms of degree n in the equation of the curve.. . a3 y = x* has a triple 4 a. If a curve with a real equation passes through one circular For.Y x. and so on. while the two lines y +c for which q + rc + (m l . the curve has a cusp x at which the line at infinity is the tangent. tion. has an undulation. + c meets the curve at two and only two infinite points for all finite vahaes of c but one. y * is by ax/y. If y m. x x + c meets the curve at three and only three infinite points for all but three finite values of c. l —x ) = =a ay = a?y 3 y has a cusp at infinity. 1 If y toj a. Similarly we may discuss the cases in which finite along y =m = =m = m =m =m =m ^m m 1 2 3 t 5 .. and %axy — x 3 + a 3 have crunodes. Ex. . it passes also through the other. (a 2 point. e.] More accurately : the vanishing line touches any projection cf the eurve.. points for every finite value of c. Similarly a curve with a real equation having a node at one Such a curve is called circular point has a node at the other.II 5 POINTS AT INFINITY infinity touches the curve * 31 means that the line at meets y = m. the curve has a triple point at infinity.
7. . the sum is a multiple of 2tt. = and bicircular nic are of the forms + 2/2 )»„. [Use Ch. § 4. if the product is constant. is POINTS AT INFINITY 2.. 9. the terms of Ex.2 + un . 3.. where u 0. Note the cases n =2 or 4. A quartic meets its four asymptotes in lying on a conic. Show that O is the centroid of the intersections of the mic with any line The equations of a circular 8 (a. = 0.32 Ex. [Theconicisg. Ex. (n — r+ 1)* degrees in its equation have respectively ~' the factors (y — mx) r (y — mx) r i (y — mx). See Ch. Conversely..] . I. Ex. [Taking O as origin.. tangents at three collinear points of a cubic meet the curve again in three collinear points '.. Generalize by projection.. . through O. 1.] Ex.] that a cubic meets its three asymptotes in three finite Generalize by projection. are the [The cubic is f uvw+p 0. § 6.2 + «B_i + "«2 + . The sum of the eccentric angles of the intersections ellipse is of a circle and a multiple of 2 v. v 0.= 0. Also y = mx meets this curve in n points whose abscissae and ordinates have zero sum. 4. 6. An wic has n distinct asymptotes all passing through O. and (x* + Z/7X4 + (*' + y*) V. 5. A varying raic passing through n given points at infinity meets a given ellipse in 2m points whose eccentric angles have a constant sum (to within a multiple of 2n)..=(^J in Ex.] Ex. 3. +u = 0. XII. A variable cubic has a cusp and three fixed asymptotes..s + . where u k is homogeneous of degree h in x and y. If an nic has r asymptotes parallel to y the »th (» — l) th . 8. Two whose centroid asymptotes. . the curve meets the line at infinity only at the circular points. eight finite points mx..] Ex. [Any one of the given lines meets the curve in only n — 2 finite points. the curve is u n + it„_ 2 + m„_ 3 + . II 5 The equation of an a l x + bl y nic with n distinct asymptotes . curves with distinct finite asymptotes meet in points is the same as the centroid of the intersections wf the Ex. only... Ex. Show collinear points = = = = — Ex.3 + «W» + . (an x + bn y + cn ) + u n . + c1 = 0. If a jeal uic meets the line at infinity at the circular points any line through a fixed point O meets the curve in n points whose distances from O have a constant product.. = 0... n must be even. a n x + bn y + cn = of the form i (a x + b 1 y + ci )(a i x + b i y + e l ) . If the n points can be paired to form an involution with the points at infinity on the axes of the ellipse as double points.. 10. Show that the cusp lies on the conic touching at their middle points the sides of the triangle formed by the asymptotes._ tv = On projection we get ' the asymptotes and p is the required line.
[The case n =2 is well inown. v = is circular.] § 6. i. The information that the curve has a node (not at a given point) is equivalent to one relation between the coefficients namely the result of eliminating x. y) is stated to be a cusp.+l) whose equation is written down at has no double point. see that a curve ' — '. The information that the curve has a cusp is equivalent. n + 1 of the nth degree.. Hence the. one of whose sides makes an angle OC/n with the fixed direction.e. we have a fourth relation between the coefficients /jv_y = \}>xly) ^...equation has %(n+l) (n + 2) coefficients.. (i) and to two relations found by eliminating x and y from that it should have a A. w If (x. It has no pointsingularities i. ' . y from (i). multiple We 2 relations. namely (i).. multiple points. Show that the locus of P is an «ic whose asymptotes are parallel to the sides of a regular polygon.II 6 RELATIONS BETWEEN COEFFICIENTS 33 u Ex. y). Anautotomic' (not * The definition is therefore not free from objection. ' Similarly the information that a curve has a /cple point at a given point is equivalent to %Jc (k + l) linear relations. The lines joining a point P to n fixed points make angles with a fixed direction whose sum is a (plus a multiple of it). Relations between Coefficients. i/c(A. three of the second degree... The sum of the angles which the n asymptotes of the nics make with a given line is the same. 11. Show that one 0.. . . y) is equivalent to assigning three linear relations between the J coefficients of f(x. e. two terms of the first degree. not exclude aenodes cutting itself) which has been suggested would.ple point to and similarly (ii) . curve of the pencil u + kv = = Ex. /=^ = ^ = *y 7>x .. however. has The information that a curve with equation f(x.* The Cartesian equation of a curve of degree n contains one constant term. 12. or unreal multiple points..?v . Zy lx 2 2 V ( ii). y) = a node at a given point (x. though it has in general 'linesingularities'. Such a curve is called iwnrandom singular.. tangents.
and that no n\c can be found in general satisfying given conditions. the case n = 2. Hence curve of degree conditions. if their number exceeds \n (n + 3). relations in general for it may happen that the § n (n + 3) between the coefficients are inconsistent or not independent. This is legitimate. ' ' . if a curve has 8 nodes. Expressing the conditions for this. y). true in general. But. an infinite number of conies pass through the points for the line through these four points and any line whatever through the fifth point form is . being a linepair. have then We ^(n + 1) (n + 2)l = ±n(n + 3) by arbitrary coefficients. As an easy deduction from the results of this section have we A finite number of curves of degree n can in drawn having 8 nodes and k cusps and satisfying %n(n + 3)82 K other relations. Hence ' One and only one curve of degree n can be found in general passing through %n(n + 3) given points. we have ^n(n + 3) linear relations between the coefficients of the curve's equation. + 3) — 3 8 — 4 k . If three of the points are collinear. if four of the points are collinear. for we did not exclude in the theorem the possibility of the nic splitting up into simpler curves. such a conic.. which determine the ratios of these coefficients uniquely. there are 8 sets of values of x and y satisfying (i). tlte number of For instance. Consider.34 RELATIONS BETWEEN COEFFICIENTS II 6 One of these coefficients may be taken as unity without loss of generality. But such results are not universally true. the conic degenerate. we have 8 relations between the coefficients of / (x. A n is determined in general \n (n + 3) We mean by this that only a finite number of wics can be found to satisfy the given conditions. for instance. and then the theorem is not true. general be other relations is \ n (n If the nodes and cusps are at assigned points. If the conditions are that the curve is to pass through \n (n + 3) assigned points. and must be . We have One and only one conic can be drawn through five points which is We say ' ' .
] Ex. quartic with three given nodes a/u* + b/v* + cluP..x1 ) + cz (z* .1). . where S = 0. [ax {a? f) (x* . 5.z2) {y2 . = = = are given lines through the node.. the a's are arbitrary constants. is Ex. 6 is is the conic aC1 Ci L li + bCs Ci L3i + cC5 Ce LBS = 0. Ex. . 5. 35 applied with due care as is shown by consideration of Ex.1. 1. +6M S„.z 1 ) 1 1 (1. 3. E. 5 to 7. 5 with ft = \ n (n + 3) — 5 r.II 6 RELATIONS BETWEEN COEFFICIENTS . The equation of any nio through r given points can be put in the form S+a 1 S1 + a 2 S2 + . S+a S + a S + l + a^S^ = 0. 1). 1).1). 4. (1. +1). 10. 12. v their intersection. Slt . x = ±y = oo .. .y1 ) + (l1 y + m1 z) {y l z*y + {li z + m 2 x) (z'x'f + il. S^ are given polynomials such that S — 0. The result given multiple points. ...x + m. with nodes at ( + 2. S^ are wics through the r given points. 5. Campbell in Messenger D 2 . where C = through 2. where u Ex. where u v = 0.. and fi = \n(n + %) — r... It is assumed that there is no identical relation of the form =0 S+b. In general one wic can be drawn with a given node and passing through \ (w 2 + 3 n . 3. 7. 4. S1 = 0. 4. Ex. + by (y .8) other points. Ex. 9. &c. but that through the 9 intersections of two given cubics a singly infinite number of cubics can be drawn. Ex.] : * The reader xxi (1892).] Ex. 1. §7. 0).6) other given points while two nics can be drawn with a given cusp and passing through  (m 2 + 8m .1. or of Chap.. 6 and Lu — is the line 12. IV. ... Obtain the equation of any «ic with r given double points and given tangents at those double points. [As in Ex. 17 below. = 0. 6. The' reader may find the general equation of quintics [Use Ex. To be given a &ple point and the tangents at that fcple point equivalent to being given Jfc(fc + 3) linear relations between the coefficients of the curve's equation.. and in = «(» + 3)3n may be extended to cover the case of any [Use § 4.y) {x^tff = 0. may consult a paper by J. S^ = are given Mics with the r given nodes. is The equation of any 1 »ic 2 2 with r given nodes . = — u + lev = is any cubic through Ex. 5. The equation of any 0.] 9 given points . +0^^ = 0. 1. 2. Find the general equation of quintics with nodes at (1.. 8. Show that in general exactly one cubic can be drawn through [If « are the two cubics..* Ex.S1 + b 2 S1 + Ex. 2. 3. (0. 8. (1. w = are the sides of the triangle whose vertices are the nodes.x ) (z2 .] Ex. 0.+ 2f/vw + 2 g/tm + 2 h/uv = 0. (a l The equation of any cubic with a given node is x + b 1 y + z) u? + (a2 x + 62 y + z)y/o + (a s x + 6 3 y + z) and v u 2 = is 0. where S. The equation of any quintic with six given nodes 1.
. 16. Annalen. 18.' 36 RELATIONS BETWEEN COEFFICIENTS is II 6 Ex. 1) as node and the cubic is xy(y — mx) +kz arbitrary constant. 0. n—5. Bull. Ex. the cubics through the seven nodes. de la Soc. y = y'.y. xxv. 13. v. 17. and these five' nodes having double points at the other three nodes. i(x. (0. aU i + 2hUV+hV = 2 V= Q= 12. 14.v. and Ex. 15. nS. For instance. = 0. 1. if P lies on the sextic. Ex. If (x'. p. VII. (1.s — y*) (1. w when the coordinates of P are substituted for x.. II. y) is a centre of the nic f(x. If u = with cusps at u — v = 0. § 10).w) jacobian (Ch. w 3 = v2 is a sextic Ex. Hodgkinson. in x and y Ex. x (1882). 0). If a sextic has nine nodes of which eight are fixed.— w^v) = 0. Soc. 343.] P : : : : . all the partial . and P are {%im^ — jcjit) + &(»». C = is the quartic through a conic through five of the nodes. the cuspidal tangents touching v = 0. Proc. An Mic passes through r 2 J(w + 2) or £(»i+l) (« + 3) as n is even or odd. Math. Note that nine nodes of a sextic cannot be chosen arbitrarily even though such a choice is equivalent to J 6 (6 + 3) = 27 conditions.z) [J = has a node at each of the seven points. 0. w = are cubics through the given nodes. [If «!. The statement 'a curve has an inflexion at a given point' equivalent to two relations between the coefficients in general. when x = x'. xv (1916). &c] cusps ? Ex. If in Ex. z. y.] .. the terms of degrees n — 1. [Taking (0. is a conic and v = a cubic. [The curve comes to selfcoincidence when rotated through 180° about a centre. given points and has a centre Find the . London Math. .] ' (mxy — m 2 a. 598. p.. however. sfcow that a singly infinite family of cubics can be drawn with three given collinear inflexions and a given node. that to be given three collinear inflexions of a cubic is equivalent to five (not six) conditions. v = 0. The. [See Halphen. «>! are the values of «. For instance. m. The statement a curve has a tangent touching at h points equivalent to k — 2 relations between the coefficients.equation of any sextic with eight given nodes CQ. is where ?7=0. the ninth node on a certain 9ic with a triple point at each of the eight fixed nodes.y) derivatives of/ with respect to x. 13 d = 0. i(u. p. all cubics through the seven nodes and of the sextic. [The preceding examples may suggest interesting investigations to the reader. 0). is u s = »2 the most general sextic with six What is the most general septimic with ten given nodes ?. 0) as inflexions. n — 3. de France. 11. The general equation of a sextic with seven given nodes is au? + bt? + cw* + 2fvw + 2gtou+2huv + dJ = where u = 0.«!. a point P on the sextic pass through another point Each such cubic meets the sextic again in points Q and Q' such that all cubics through the nodes and Q pass through Q'. 20. Math.. where k is an is Ex. lies _ . vanish. 19. See Eohn. 1. is Ex. are two cubics through the eight nodes. If the centre is the origin. y of orders n— vanish = 0. n — 5. 162. Hence at their ninth intersection Therefore.] Ex. Show.] r being degree of the locus of the centre. so does the ninth u v iv = «<! % w1 intersection.
the number of its branches. g. The following hints are Find where the curve meets the axes of reference. the curve is symmetrical about the origin. for requires. useful in curvetracing. Or we may wish to obtain only a rough idea of the main features of the curve (e. approximation to the truth. If the equation (i) (ii) contains only terms of odd. finding for this purpose a large number of points on the curve. Or it may imply that the curve is to be drawn with the utmost possible degree of accuracy. where pos. Similarly if only even powers of y occur. and we can spare ourselves the rather considerable expenditure of time which a more detailed tracing usually sible. (iii) Notice if any values of x make y unreal (or vice versa). ' we it shall here mean the third of these will be useful to indicate.' CHAPTER The Object III CURVETRACING § 1. of Curvetracing. If so. &c). or only terms of even degree. and equate to zero . a circle can be traced by means of compasses. how a more accurate diagram may be obtained even if most practical purposes a rough sketch of the curve will suffice. (iv) Equate to zero the terms of lowest degree in the equation to obtain the tangents at the origin. Trace a curve with given Cartesian equation capable of more than one interpretation. the curve is its own reflexion in x = 0. the rectangular Cartesian equation of the curve being supposed given. The Method of Curvetracing. §2. these values of x do not correspond to any real part of the curve. and to draw a sketch of the curve which gives some ' The problem is . which is called a centre of the curve. If only even powers of x occur in its equation. It may mean that a mechanical construction is required for instance. the position of its asymptotes and singular points. By curvetracing alternatives though ' . See if the curve has symmetry.
1. (vii) Find if it is possible to get any number of points on the curve by solving equations of degree 1 or 2 at most. there may be one or more. whose index is equal to the intercept made on the axis of y in Newton's diagram by a line through (a. if necessary. origin to the points at infinity on the curve. §§ 3. A being a numerical coefficient. if the curve is of degree n. Suppose that any term in the equation of a curve is Ax a y&. diagram for the curve. II. Suppose that a line * joining two or more points of Newton's diagram (not parallel to an axis of reference) is such that all other points of the diagram lie to the right and above the line. proceed as in § 4. The above n—r §3. and a. if the points on this line are considered as representing terms of the same order of magnitude. It may sometimes be useful to obtain the tangent at one or more points of the curve. and. where is negative or odd. If * If the curve does not pass through the origin. j8) referred to rectangular Cartesian axes placed as in The points thus obtained are said to form Newton's Fig. x and yl are of the same order of magnitude. Find the asymptotes of the curve. all the other points of the diagram represent terms of higher order of magnitude. the order of the term Ax a y& is the same as that of y0 +a 1/P. all hints are usually more than sufficient to obtain the information necessary for tracing the curve. when x and y are both small or both p Then large. if it is met by a line in r real points. Newton's Diagram. give any fresh information.) Use Newton's diagram to get approximations to the shape of If this the curve near and very far from the origin (§ 3). 5. Suppose now that. is no such line. the terms of highest degree to obtain the lines 'joining the. this requires the solu tion of an equation of degree n — 2 at most. /3 zero or positive integers. If . though occasionally special devices are needed.38 THE METHOD OF CURVETRACING III 2. j8) parallel to the line making intercepts p and q on the axes of x and y. (vi) Find the finite intersections of the asymptotes with the fails to (v) curve : for. there the curve does so pass. Then. (Ch. It is also well to remember that a tracing of an nic cannot be correct. Consider a geometrical representation of the terms of the equation such that Axa y B is represented by the point (a. find on which side of an asymptote the curve approaches it at the two ends.
the terms represented by points on the line give an approximation at infinity. . orioin = the branch through the near the approximates to x2 + 5y = origin. we get an approximation to the curve near the origin. The lines gents are y = the origin to the points at infinity joining jrjr are x = 0. Similarly. 1) gives as approxi. l. and H '* ) 5x + 2y 0. of the Ex. Ex. x + y = 0. Newton's diagram (Fig. Examples of Curvetracing. 1.] §4.Ill 4 EXAMPLES OF CURVETKACING all these 39 we suppress higher terms in the equation of the curve and then divide out by any power of x or y. which is a factor of the remaining terms. Illustrations of the use of Newton's diagram are given in §§48. Enunciate and prove a method similar to that of Newton's diagram for approximating in three dimensions to the shape of an algebraic surface near and far from the origin. The origin is a crunode at which the tanand 5x + 2y = 0. 3. As another example take the curve in § 7. There is no symmetry. we get an approximation to the curve very far from the origin. and far from the origin = x(x2y)(x + y) = 0. in the curve x*y + if — 2xy — x = Newton's diagram gives the approximation y 1 = x near the origin and a? + y = 0. The only new information is that touching y Fig. 2. + bx 3 near the origin. I. Ex. 3 — x y— 2xy 2 + 5xy + 2y* = 2 0. if all points not on the line lie to the left and below the line. The curve meets the axes of reference in no finite point except the origin. In each of the following examples we shall only give details working when such details illustrate methods which have not been used in previous examples. Apply the method to find the shape of a surface in the neighbourhood of a parabolic point. Ex. xy = 1 far from the origin. If a line through two or more points of Newton's diagram is such that all other points lie to the left and above or to the right and below the line. Newton's diagram applies even if the original axes of reference are not rectangular. x = 2y.„ t _ <*rH 2 mations near the origin x + 5 y = 0. For example. [It approximates to the form of the surface z = ay 1. Trace a.
If the equation of the curve had contained terms of lower degree than the second. we should have retained only the terms of highest degree in y in the numerator of the righthand side after putting 2 y for a. The curve evidently meets the asymptotes at their intersections with 3x + 4y = 2 giving three points on the curve. On the righthand side of this equation put in the first approximation —fa.) We have now enough data to trace the curve (Fig. Similarly the other asymptotes are On = x— =— Write x = 1 and x + y =1. We will verify the diagram by finding the side of each asymptote on which the curve approaches it at either end. and we have the second approximation 2y 2. If the equation of the curve had contained terms of degree higher than the third. (See Ch. 2). = .. The curve becomes 2 2 2 . Ex. 3. Thus write the equation of the curve in the there x We know that = 2y. § 5 or we may employ the device just used to find the approximation at the origin. for y and we get the closer approximation 2y + 5x = —^x* as before. II. Write" the curve . Another method is to write the equation of the curve in the form 2y + 5x = (x3 + x 2 y + 2xy2)/y. II. now the equation of the curve in the form (xl){x2y + 2) (x + y— 1) = Zxty + 2.40 EXAMPLES OF CURVETRACING To get an approximation to the branch touching III 4 Tpxit2y + 5x= Y. These form _ 5xy2y 2 x(x + y) the righthand side put in the first approximation x 2y. 0= Y 5xyl8x^ + 9x 7x7 and Newton's diagram now gives the approximation 57+18a:2 = or 5(2y + 5x) + 18x* = 0. may be found as in Ch. we should have retained only the terms of lowest degree in x in the numerator of the righthand side after putting — f a: for y. are asymptotes parallel to x 0. which is the asymptote. where the lefthand side equated to zero is the equation of the asymptotes. § 5. 5x + 2y = 0. x + y = 0.
2 35!/2 + 53M/+2J/ 2 = lies close 0. As stated However. for the 2y — 2 — — 5 for the curve. slightly smaller for the curve than the asymptote. if to the asymptote and to the left of it. the accurate in § 1. we do not usually require this. tote An We get x2y + 2=. y is Similarly.e. . xSxty. The reader may discuss The line x = ty the other asymptotes similarly. tracing of most of the curves given in the examples of §§4 to 9 does not present much difficulty. and 2y — 2 i.Ill 4 EXAMPLES OF CURVETRACING 41 approximation to the curve as it approaches the asympx — 2y + 2 = is % = 2y — 2. large and negative. Hence the curve Pig. 2. the curve lies to the right of the asymptote. meets the curve where (5t + 2) x = (l+t)(2t)' y= " t(l bt + 2 + t)(2t) Putting in any value for t we may obtain any number of points on the curve. and retain only the highest powers of y in the numerator and denominator. Put this value of x into the righthand side to get a closer approximation. and trace the curve accurately. 5 is For a given large positive value of y x asymptote.
2) = 0.y 2 x(x 2 — y2 ) = ix2 — 6xy—4:y2 64 2/ 2 (a.) An approximation at the origin is x + y = — x' the other approximation being x — y = x2 (xi) (a. (1.= 4a:. Fig. + 4)(i/ + l)(a:03) = a:162/12. 2 2 2 . 2y (x x2yy2x+x2 — 4y2 = 0.3)(2a. §2. (iii) Writing the curve y(x + 2) — — x± { — x(x 2 — l)(x + %)}i.2y (x2 .+9y + 12 = 0. 2 — xy + 2y 2 2 2 (a. (1. See Ch. x(2xy) (x + y) + Sy2 = 0. (Ch. See Ch. j^ (ix) (x) (xi) (xii) (xiii) + i6a:'>+a:2 (5a:i2) = o. Fig.+{a:(a: + 2)(a. — xy + y )+xy + y = 0. . 1. .2 x) + x2 (x . (xv) y* — x* = a2xy.2/ + 3)(aj4) + 9a. (vi) and (vii). The curve has the origin as a centre of . xy(x>Sxy + 2y*) + x s ~2x 2y23cy1 + 2y s 3x(x* + y*) + 2x* + 6xy + 6y". we see that the real asymptote is x + 2 = and that the tangent is parallel to x = sX the joints (0. The asymptotes meet the curve on the circle x2 + y 2 — 1. the curve again where y = 0. and meet. aj' + x(x2 — xy + iy2 )+4:y(2x—y) =0. XIV. .+2/3)(a. For real points x must lie between and 1 or between — § and — 1. + y) (x + y ) + 2 x (x .y) = 0.0). (iv) 3y(a: + 4) = 4a. xy (x2 + y2 )+x 3 + y3 = x 2 . 3. . The asymptote # + 2 = meets the curve in no finite point. showing that the asymptotes are x = 0. • (v) (x (vi) (vii) (viii) + y — 2) (a.] origin. : = 5xy+x + 2ij. 2 (ii) (x+l)(y + l)(x2yl){xy + l) = x* + y l. 2. Figs.y 42 EXAMPLES OF CURVETRACING : III 4 Ex. + 242/36 = 0. + 1) (x2 + 2 ) = xy{x+y) + 2x2 Since x + y is a factor of the terms of the third degree as well as of the terms of the second degree. Real asymptote x = 1. 2 2 (xiii) (x +y 2y + l)(x + y + 2) = xSy + 2. See Ch. x + 2y + 2 = 0. Trace the curves (i) (ii) (iii) (iv) (v) . y + 2 = 0. (xii) (a. 4). (f. 4 for Real asymptote x = 4. There is an inflexion on each.+ 1) (x + 2y2){2xy2) + 2 (x + y2) = 0. 'It is an inflexional tangent at infinity.4) 2 . II. branch at the symmetry. [(i) The curve is x{y + 2) (x + 2y + 2) = 4i/. and therefore the curve approaches it on the same side at both ends. 3 = {yx) (y2x). $). Real asymptote x = 1. x=ry.2/ + 4)(a: + «/+l) + 14a. (xiv) (a. XIII. x + y = is an inflexional tangent at the origin.1) . XIII. (ix) (a? (x) \x'+ 1) (y . 1).4)(a. + 8)}4. 9(x + i)y'i + 2ixy + x(x 1 + 2x16) =0. (xiv) x(x + 2y)(2x— y) = 4a. (vi) (vii) (viii) x2y + 2xy2 + 6xy + 2x 2 + 4:X4 y = 0. (xv) Asymptotes x = y.
so that the latter alternative The line x = y + c meets the (See Ch. 43 Ex. § 5. so that the tangents at the infinite double point both coincide with x — y'= 1.c3)2/ + c 2 (c 2 c3) = 0. II.i !: = 0. Fig. 2 curve in only one finite point if (cl) = 0. the 4(cl)V + 2c 2 (2. (xyl) 2 (x + yl) (x + y + 2) = 5x3y2 . Similarly the curve has an infinite double point on x + y = 0. and the curve has two asymptotes coinciding with x — y = 1. Hence the double point is a «usp.Ill 5 EXAMPLES OF CURVETRACING §5. II.) is correct. giving only two finite values of y. Writing the curve in the form = —2. this equation. 3. or else has an infinitely distant double point on x = y. We find that x — y + c meets the curve where Since is — (x — y) 2 (x 2 —y 2 ) (x + 3y) . the tangents at which are x + y = 1 and x + y The curve has these lines as parallel asymptotes.3x2 + 6 xy + y 2 = 0. (as Trace 2 2 2/ ) 2 a factor of the terms of highest degree in curve either has an infinitely distant tangent with its point of contact on x = y. (x' 2 2 tf) 1 (x y' ){x+3y)3x'1 + 6xy + y .
J). an in tangents are parallel 'to parallel to aV = 0. {xyYf = l/2y. (i). ay 9 as' 3/* + 25 = 0.0). (iii) y=0 See Ch. x 2 (2xy)+x 2 + x + y = 0. 9*y + 96asy + 144 (as + y) + 496 = 0. (x 2 l)(x2)y t x3 + 4:y = 0. 100a. x2 (xyfx(xy)(3x + y) = 3x2 8xy + 2y 2 y(xy) 2 = x + y. and that closer approximations to the form of the curve far from the origin are x + yl=2/3y. that the curve meets its asymptotes where they meet 5x = 3y + 2. (6. (3. §4. is flexional (ii) x(x + y)(x+yl) + 2 = 0. namely the end at which y is positive. The other two approximations are interpreted as in § 4.y + 2y (100 + 90*)144a!»+36a! + 469 = 0. 3. 10{x2 4:)(y'i l)+xy + 23x + 26y + 94: = 0. }). (iii) (iv) (v) (vi) (vii) (viii) (ix) (x) (xi) (xii) (xiii) (xi?) (xv) (xvi) (xvii) (xviii) (xix) [(i) y i f + (xi yi )(hyx)3x i 2xy + 9>y l = Q. x 2y 2 2xy(x2y)3x2 + 8xy+3y2 = 0.) Ex. (x l = 0. 2) and 1). xY+12as (3x + 2y) = 0. Ex. X. xy(2xyhx10y) + 50(xyY = 0. = tangent at infinity compare § 4. (1. (f. 3. 225 (yx 2 + l) 2 = (5 + as) (1 + x) (1+2*) (3Bas) (34*) . tof these approximations shows that the curve approaches the asymptote x y + 1 from both sides at one end only. x(xy) 2 2y(xy)+y = 0. (5 as).2). as in §4. II.44 EXAMPLES OF CURVETRACING III 5 we see. = 0. The asymptote a. Solving the equations as a quadratic in x or y we . 2 (iii). Trace the curves (i) : may be put in ~ X _ (t 2 + 2t2)(t 2 + t\) 2 ' y _ ~ (t 2 + 2t2)(t 2 + t3) 2(t 2 + t2) # (ii) x s + 2x 2 y + xy 2 x*xy + 2 xiy i + xy + x + 3y + 3 = 0. The origin is a crunode at which the tangents are The third = 3x2 — 6xy — y 2 See Fig. as = see that the at {xy + l) 2 (x + yl)(x + y2) = x7y + 2. (x + y) x1y 2 + (x2 + 6xyy2 ) xy + 2 (x s + y 3 ) = 0. x 2 (x2 y2 ) + (x+ 1) (2x 2 + xyy 2 ) + x = 0. at (0. xy(x + y) 2 +(x + y)(x2 3xyy 2 )ix 2 + 2y 2 = 0. The coordinates of any point on the curve the form 2(t + t2) (SeeCh. (6. § 4. The asymptotes are Ex. Ex. . x + y + 2=l/2y.
which is the pair of parabolas y 2 = ± F a tacnode in such a case. The curve is symmetrical (x* about the = 5x + v 2. The intersections with the asymptotes and with 2x = 3.8y6. The origin is Hence there are — V . 6. oo ) and (oo Bitangents x = y. Ex. The curve has a crunode at (oo 0). Fig. . 1 . (xvi) Asymptote's (xvii) . See Ch. called ' ' two branches of the curve touching each other and touching the axis of y at the origin. Trace Newton's diagram (Fig. 0). * A tacnode is the point of contact of two ordinary branches of a curve. Fig. (xl)(x+l)(2xy + l) + 3x+l=Q.6. and an oval lying between the asymptote and the axes of reference. 0). x = ±2. XVIII. (xviii) Asymptotes x = ±l. 3x + y = 4. (v) (vi) (vii) (viii) 2)(*yl)« + 8*8y2. (oo 0) and a triple point at It has a branch asymptotic to x + y + 6 = (0.y = ±B. and a triple point at (0. Consider its intersections with y = tx or txy + x + y = 0. Solve for x and y. Solve for y. + x+l)(xy+l)(x + y)=y. oo ). Solve for x and ' «. (x*l)(x2)(y*l)=2x* + x4:y2.* As in § 5 we find the parallel asymptotes y+1 = and y— 2 = 0. §6. y = ±\. The tangents are parallel to «. XVIII. . Compare (ii) T (* + the origin. 4) gives us as the approximation at the origin 4 2/ = 2x\ </2x. (xii) . (xiii) The curve has acnodes at (0. It is sometimes called also a double cusp but this nomenclature is open to objection. See Ch. to (xl)(9 + l)(x+yl)(x+y2) (xi) The asymptotes are x^y* = 0. &c. (5. y{{xyf2} = xy. (xix) Asymptotes x*= 0. . There is an inflexion at (ix) origin and has an inflexion at the origin. (xiv) (x + l)(x + 3)(y+l)(yS) + lSxy + 12x + 6y + 9 = 0. 10) ^) respectively. y = 2 should be found. as we shall see later. Asymptotes xhf = 0. (xv) Asymptotes x2 (25y 2 36) = 0.„ J f i the axes of x and y at ! c Jf and (8.Ill 6 EXAMPLES OF CURVETRACING 45 iv) T he C e h a n0 al ^ymptote. Vx = y. Biflecnodes at (0. oo ). (x 1) (x + 2) (x y l)(xy 3) = lla. III. 2.
and lying above the axis of a.) To get an approximation to the curve far from the origin along x + y = 0. 2 . write its equation in the form since (x (x + y) 2 =x 2 (y + 2)/y2 . parallel to x + y. + »)' = *»(» + 2).: .. § 5. OF CURVETRACING III 6 There are two infinite points on the curve along x + y = 0. (Ch. x on the righthand side and retain the highest power of x in the numerator. This is a parabola with axis for Now put —y Pig. and readily find the side on which the curve approaches either asymptote. 5. See Fig. 5. 46 EXAMPLES. (x (y + 1) it (y. We get as a closer approximation (x + yf =? y. II. + y) 2 is a factor of the terms of the highest degree but x + y = c meets the curve at only one infinite point for all finite values of c.2) meets + yf + (y* + 2y) we see that its Ex. HeDce the curve has an infinitely distant tangent with its point of contact on x + y = 0.— 0. asymptotes where 2x + y = 0. (ii) x*y = x + y". Solving the equation as a quadratic in x we see that for real points on the curve y^ — 2. Writing the eqsation of the curve in the form (2x + y)'= 0. touching y = at the origin. Trace the curves (i) {x + y)(x~yf = %xyy i . /(a.
(xi) Taenode at (oo 0). The tangent at (1. a: 2 (vii) (viii) (ix) (x)  (xi) (xii) (xyf a? (x — y)* + y = 0. + 2y) = 2 (a. % (vi) x (x — yf(x — y — 1) = — y. (xv) Taenode at (oo 0) at which the tangents are y = 0. Asymptotes aj'y 2 = 0. and the other a cusp. Parabolic apApproximations at origin a. x = y + l are the tangents. See Ch.(ya. (xiv). 2 ) 8 = 7(a: + 3)(a: + 2)(a. 2 8a. aV = 16(yl).l)0r4).] . node at (0. — y) = 1/y. but which is an isolated point on the curve. a. 2 (a. oo ). 16(3y + a. y 2 = 4 (y + 1). x and y. cusp at (0. [(i) . Parabolic approximation (y — 2a. a. Parabolic approximation (y — xf = ay Parabolic approximation a. XVIII. Solve for y. oo ) with line at infinity as tangents. Asymptotes at a. x = y. Parabolic approximation (x + y) 2 = 2y.2 y J = 0. Tangents parallel to axes of (iv) an"y — 1.) 3 (a. (xvi) Taenode at (0. Asymptote a. = —2. Asymptote x = a.y + 4) = 2(2y + 3)(y + 2).) 2 (y + a. There is a taenode at infinity at which x = is the tangent. The curve has an infinite cusp. Asymptotes a. (xii) Solve for .0) (v) . — y — 1 = — 1/y. (x) A taenode at the origin.0). ' (xiv) (xv) (xvi) *y = 4(y + 4)(yl). Solve for x and y. 8 + ay = proximation y^ + ax = 0.Ill 6 (iii) EXAMPLES OF CURVETEACING x*y 47 (iv) 4y 2 ( (v) (vi) = f + 2xy+4x. . cusp at (0. meets the curve again at (vii) (1. Asymptote y =J). (xiii) a. 4) respectively. Asymptote y = 1. = 1/y (a. and y* + ax = 0. . 2 . a? a. The approximations at infinity are 2 2 2 a. x = y twice. Sym metrical about (iii) x= 0. —1) and (4. = (a. and a triple point at infinity at •which x = y. Parabolic approximation (x — y) 2 = x. (xiii) Taenode at (oo 0). The asymptote x + y = a is an inflexional tangent at infinity.) 8 + (ya. Asymptotes y 2 (x°. oo). (ii) (x + y + l)(x — yf=x(x + y).. y + x){y2xy = $x!f. oo).) 2 = 6 x. Taenode at (oo. oo ). l (ix) y i (x + y — a)(x + y + a) + a(x + y)(x + xy + ay) = 0.2y). § 14. One curve has a node at (0.— 4) = 0. 2). 3 aajy + a = a^. (viii) Asymptote x = 0. Parabolic approximation a? = y.y + 6) = 9(2y + 3). Parabolic approximation 2y = — x\ Asymptote x + y + 1 = 0. = 2y + l. {yxf = a (y8 + a8 ). 2 2 y (x + y) + ctx(x + yf + a'xy = 0. 8 = ay. Parabolic approximation at ( a: 8 = y. a.
6. 1. . §3. ' The line y tx meets the curve where The tangents x = t a {l±(t* + i)i}.) > parallel to the axes of reference are readily found.48 EXAMPLES OF CURVETRACING §7. 7). we see that x = — y touches the curve at (1. Ex. — 2xs + xys We may verify our approximations at infinity by writing down the equation of a projection as in Ch. II. The curve has an inflexion at infinity with an infinitely distant tangent.  1 1 HI Fig.— 1). Fig. (Ch. Putting in values of t we get any number of points on the curve (Fig. The asymptote is x — 0. III 7 Ex. = y = t*{l + (t* + l)i}. § I 5. Trace 3 2 3 y = 2x + xy Newton's diagram (Fig. taking t = . In particular. IV. 6) gives y 2 = 2x 3 as the approximation at the origin and 2x2 + y 3 = as an approximation at infinity.1. if we replace x and y by x/y and l/y. I. Thus. the equation of the . y 2 7.
at infinity x 2 +y 5 = (ii) Approximations at origin Bay and Bax =y s 2 . 1.1. §. Again. The The tangents are vi) The curve is closed. projected into the axis of y. 2 y (x — y) = ^(x + y). while the line at infinity . . and vice versa. The projection has x = as an ordinary tangent at the origin. and therefore the original curve has the line at infinity as an inflexional tangent. parallel to the axes of J respectively. while the line at infinity is projected into the axis of x. 1 x and y when y = \ (1 + </2) and when x = . .at (oo 0). . (iii) Approximation at origin y = x asymptote y = 1 is tangent at an infinite point of undulation. The reader may apply this process to any of the examples in this chapter. 2 at infinity X + y = 0. .Ill 7 EXAMPLES OF CURVETRACING 49 curve becomes y 2 = 2x 3 y + x. ! a?y — x* + y* = 0. the equation becomes x2 y2 = 2x + y s This is the equation of the projection in which the axis of x is unaltered. [(i) Approximations at origin 0. The . Ex. . and vice versa. replacing x and y by l/x and y/x. . and therefore the original curve has x = as an ordinary asymptote. (ix) y* (x) (xi) (xii) (xiii) (xiv) (xv) (xvi) (xvii) + xy s + x2 y2 + x*y + x 6 = x 2 y 3 xy + y 4 + xy* = X s 2 x 3 {x2y)=%y 2 (yx)l(yi x ). This is the equation of the projection in which the axis of y is unaltered. . 2 2yx 3 + x7 = 0. 2 2 l Bx = 2x y(x+l)+y 2 2 i 3 y + x (x y ) = 0. 2 2 at infinity y = x 3 (iv) Approximation at origin y = x — yy1 at infinity y2 = x3 (v) Approximation at origin x curve is symmetrical about 2y = 1. y 2 2 i 3 y {xy) + x + y* + 2x (xy) + x(x + 2y) = 2 2 a?y 2 = (yx*) (y2x (yBx ). (iv) (v) (vi) (vii) (viii) x 3 + (xyf = y. l 3 y y = x x. The projection as an inflexional tangent at the origin. 3 x* + a (x* + y ) = ScPxy. 0. and 2 2 y2 = x . (i) Trace the curves (ii) (iii) x*8xy' + 2y 4 + y s = 0. 2 3 (x . at infinity x 4 + ay* — 0. y 1 =m =x 3 . x{x + y) = y*y i x 3 (x + y) = y3 + y 2 . y ) .yf + x y = x* + y*. is has x = .
2 and y = Bx 2 at infinity y = Xs and y 2 = —6a. 3 .) (xv) Approximations at origin y = x 2 and y = —Bx 2 at infinity y = —2x* and 2y = 3 (a. namely two coincident parabolas.a2 2 + \y* .  . — 1). 2x s = y. Trace " +1 a. at infinity (x vii) 2 2/ i/ 2 = a. Trace a>6 = a = 6 = c."_1 = ) = 2a8 and 2y = x*\ at infinity Ex.infinity y = x and y* = x*. 2 (x) Approximations at origin x = — y 3 and y = x at infinity x = #3 and y = —x 2 s (xi) Approximations at origin 8(y — x) = x and y + x = 2y 2 at infinity a? + 4y 2 = 0. Trace a. o = c>6. Trace x i 2x s y + 2x 2 y + y 2 Newton's diagram infinity (Fig.2a:.™ + j/ b negative integer. 2. 2 §8. 1). at infinity y = x The asymptote is x+y + l=0 meeting the curve at (0. 2 (a: +# 2n+1 = (2n+ 1) . 3 Any number of points may be found by 2 considering the intersections of the curve with y = fee (Take especially the cases t — § and t = £. 2 (xii) Approximations at origin y = x and x 2 = y s at infinity 2 3 2 a. y 2 = —X s . touches y = at ( — 1. when n is an odd or even positive or integral. Newton's diagram gives (y + x 2f = as the approximation at the origin. III 7 2 s (vii) Approximation at origin y = + x 2 . c.J = «aa. where the tangent is x = 1 which meets the curve again at (1. . V. . 8) gives as = 0. This indicates that the origin is a somewhat complex singularity on the . !) and J). is Trace x n + y n n 2 a 2 x n . at infinity.2 when n is 4. meets curve at The asymptote x — 2y + l = (0. E = 0. 2 An isolated (ix) Approximations at. 3. —1). o>c>6. Approximations at origin «/ + a. (xiv) Approximations at origin y = x 2 and y = 2a. Approximations at origin y 2 = X s and x + y = 2x 2 and asymptote x + y = 2. Approximation at origin y 4 + a. Consider the intersections with y = tx 2 (xvi) Asymptotes + 2y = 2x+l. quadruple point at the origin. . axn y n when n integral. an approximation at Hence the curve has a cusp at infinity with an infinitely distant cuspidal tangent.50 EXAMPLES OF CURVETRACING . .0). — i). and passes through (0. = y and (x — y) = 2a 2 (xiii) Approximations at origin a. Ex. . The line # + 1=0 is an inflexional tangent at (0. . 1) and (— §. (a t + b*)i>c>a>b. 5. The curve has a crunode at the origin. + 2y = — 3a. Ex. (viii) . . Ex. and y s = —2x at infinity y* = — x 2 and (x — y) 2 = .7 = 0. There is an asymptote x — 2y + l = 0. —1). —1) and meets the curve again at (1. .b 2 f = c4 distinguishing the cases <»>6>c. Ex. & . (a4 + 64 )* = c >a > 6. . .
)* From Kepanov = a little horn. The origin is sometimes called in such cases a rhamphoid (ramphoid) cusp * or cusp of the second species '.9). x*2x 3 y + 2x*tj + ij* = possible type and the latter siderable complexity. the ordinary *>cusp being a ceratoid cusp f or a cusp see that for real points ' We ' ' V^ ' ' ' of the first species '. 8. 9. The nomenclature is objectionable. * 0. Hence the approximation at the origin is given only by those parts of the two coincident parabolas for which x is negative (Fig. whereas the former is the cusp of the simplest  Fig. on the curve x must not He between and 2. write its equation in the form y = — x + x ± V(—2x + x 2 3 5 e ). it implies that cusps of the first and second species are comparable. for Pig. is really a singularity of con From fia/upos = beak.Ill 8 EXAMPLES OF CURVETRACING 51 curve. . E 2 . To ascertain more closely the nature of the curve at the origin.
1). nonzero. The radii of curvature of the two branches at an ordinary cusp. I. There is a rhamphoid cusp' at the origin. 5. The curve See Ch. nonzero. (2y + x*y = x.] such that «jj =y. oo ). [See Ch. 3. v.. Note the case in which these radii of curvature are equal and opposite. § 15. (iii) Rhamphoid cusp at (0. + 16). +Un . as 2 2 for which x > 0. 4 points respectively coinciding with the double point. (ii) is seen by solving for The approximations at y..25) = 0.. (viii) There is a singularity at (0. a rhamphoid cusp meets the curve in 3. as in Fig. 4. The (2. approximation at the origin is that part of (f + a: ) = 5 6 meets a. . § 6. a tacnode. The line x + y = 2 touches at (iv) Rhamphoid cusp at (0. oo ) with tangents and the line .at infinity. 2 parabolic [(i) The asymptotes are {x+ 1) (a:l) = 0. i (y + x'*Y = 'ix(x 2x + 2). oo ) whose projection is like a rhamphoid cusp in shape. and 2 «>. (x + tjf + x* (x* (if = 0.. Ex. <x> ). a rhamphoid cusp respectively at the origin.. * [Choose axes of reference Newton's diagram. Also asymptote 4x+16y + b — 0. (2y . Ex. = 0. Asymptotes x* (4y* . Ex.9) 2 2 a? (4j/ x+ 1 = (0. Rhamphoid cusp at (0.. 4(a^+l) 2 = a:(25a.. $) are parallel to ' x = . +un = (tt + «1 2 + %2 «0 +M W3 +M + 1 1 5 . oo).+«„. a tacnode. and equal. a tacnode. 0) (v) (vi) (vii) and (1. 6.] = there Ex. the o = «.. are homogeneous of degree r in x and . If «. 4. EXAMPLES OF CURVETRACING Trace the curves st* III 8 (i) (ii) (iii) (iv) (v) (vi) (vii) (viii) + 2xy <y*) +xy (2xy) + if + x*y = yW. curves 2. Asymptotes Node at Rhamphoid cusp at origin. 5. + u + ut + s 1 2 1 ) = u l (u + u1)+ui + ... and in general unequal. . . XVIII. is [The radius of curvature at the origin of a curve touching y the limit of xl/2y as we approach the origin. y. and there is a approximation at infinity (x + yf = 2x. The ratio of the radii of curvature of the two branches of a curve at a tacnode is unaltered by projection. a rhamphoid cusp are respectively both zero. Then use Ex. infinity are f=x and if = a: s . 2 = ft/ 3 where x = r* (t + l) y = r (t + 1)*. {hy + x>Y = 4:x(xZ){x + h).52 Ex. oo ).] ..a: 2 2 ) + 4a:i/ (y + x') = 0.] Rhamphoid cusp at (0. % = x. have in general an ordinary (keratoid) cusp. 1. The tangents at (. 4(xy + iy=x(9x + lG). \) and (2. The tangent at an ordinary cusp.
XVII. . we see that there is an asymptote parallel to y/x = 1 corresponding to t = 0. (1 + *) (1—2*)/*. To give a tacnode or rhamphoid cusp and the tangent at that point is equivalent to assigning 6 or 7 relations respectively between the coefficients of the equation of a curve. u = v 1 2 becomes a conic of closest contact with the curve at the origin.e. 6. whence t = x = oo or 2+ </2. i. The curve meets the asymptote or £( + /21). t = —1 . and the method of the following example will often suffice. §9If the coordinates of a point on a curve are given in terms y of a parameter t by means of the equations »=/(*). x /y=(l+t)(l2t)/t2 . Making the equations of the curves in Ex. [The origin is called an oscnode Contrast the case of the rhamphoid cusp in Ex. VI. § 8 (iv). As an alternative we may obtain any number of points on the curve by taking various values for t and so trace the curve. x and y are infinite when t = Since y/x = (1 + 1) (1 — 2t). obtain the equation of the curve by eliminating t between these two equations and then trace it by the methods of §§ 4—8. = at the origin and having the same curvature there. 2 and see Ch.. putting s = oo in . If the tangent is not given. The curve only meets the axes at the origin. y = (l + t) 2 (l2t) 2/t. . on putting t = 0. EXAMPLES OF CURVETRACING The curve (M 1 53 0= +»1 2 ) a +(Wl«'i + Ms)(% + V)+V«2+«lM4 + «5+ . we may Ex.] ' '. 7. we obtain x—y = 1 as the asymptote. and then We have y = tx2 and therefore the approximaor \ 2 tions at the origin are y = — x 2 and 2y = x or oo Again.Ill 9 Ex. .. But this is apt to be laborious. = m. 2 and 6 homogeneous by means of u^ = v2 we get the lines joining the origin to the intersections of the curve and u 1 =v1 If we choose the coefficients of » so as to 2 make as many of these lines as possible coincide with u t = 0. Trace x = . +«„ has two ordinary branches touching «. i Moreover. Ex. the number of relations is one les's. . Because x—y = (1+*) (1 — 4* 2). A doubly infinite number of conies meet a curve five times at a rhamphoid cusp. . where (l + *)(l4* 2) = 1. and no conic meets it six times there.
The tangents parallel to x = dec are given by ^ = and are unreal. r (v) x = t + t' y = 1+f. (ix) a: = (1 + * + **)/(* + **). 2/ = *7(l + *). Symmetrical about x = y. 10). (vi)x=*7(l + *). 2. 2y? + x2y(x + l) = xi . the only real tangents parallel to y are the tangents at the origin. (iii) x = (t* + St). y = 3o*/(l +* 3 ). Similarly. and is infinite at an inflexion. (vii) x = (t + iy/t*. 1. 10.54 EXAMPLES OF CURVETRACING see that a? III 9 we + 2y = is the other approximation at infinity. as is seen by changing 1 into 1/t. ordinary cusp (see § Ex. The inflexions are given by xy = xy. y=(t + iy/t\ 3 (viii) x = * (2 . . y = fBt. 6 5 (ii) x = 5e»*7(l + * ). [The radius of curvature is (x? +# a )4 ±(xy—xy). y i (iv) x = (t*2t ). [(i) infinity * Approximations at the origin if = 3 ax and x = Bay. Ex. y = 5o*7(l +* ). y (i) . 4). dots denoting differentiation with respect to t. 2 below. or those given in Ex.*). x = 3 a*Y(l +?). (see Fig.*) = (l + t + t2 )/(l + t). .] Trace the curves . = (* s +l). The real ordinary cusps of a curve are given by values of the parameter * making x and y both zero. As other examples the reader may take the curves in §§ 4 and 5. at x + y + a = 0. It vanishes at an 8. Ex. y = * 4 (2 . = Fig.
= a cos = a cos = a sin St. x + y=a. 2 1 t 1 3 2 . a. = 2y and 2x 3 x =« . For instance. §10. a: Symmetrical about Ex. At infinity approximates to = x (ix) Approximations at origin x* l . 3. at infinity Acnode at origin. and x = y 2 .] Trace the curves (i) (ii) (iii) x x a. y. IV. = asiaBt . of the circle . (iv) x=acos x a. (vii) x3 y l = y* (viii) At origin a rhamphoid cusp. £ . The perpendicular is positive or negative according as the asymptote lies on the right. 1 The asymptotes are given by noting that the polar subThis is readily tangent r 2 5 becomes the perpendicular from the pole on an asymptote.* The case of 5 a = 6 b is shown in Ch. + «+l = 0. Node at = + ^/3. (iii) Approximation at infinity Tangents parallel to axes of y =x x and y where = (inflexion) and t = . drawn by noting how r alters as 6 increased from to 2 7r. <. t. at infinity y = a?. = 6 sinh 3 . * It r may = b cos also be drawn by increasing each radius vector by the constant a. 3 1. cosec2*. 55 f = hax* and x* = 5ay* at infinity = y. (vi) Approximation at origin y* = x at infinity x + y = 1 and » = y. (/ = y asin 2 £. (v) (vi) = a cosh = acos3i <. 2 £ 3 . (v) Approximations at infinity y = 1 anil x = y. cosec2i. cusps at t = + 1. Symmetrical about y = 0. y=l.. Some curves are most readily traced sian into the polar equation.2k (iv) Approximation at origin 2y =9x. = y j/ Asymptotes = 1. 6 £. 4 3 . if 6 has a value a which makes r infinite. or left of a person standing at the pole and looking along the direction 6 = a. Ill 10 (ii) EXAMPLES OF CURVETRACING Approximations at origin Symmetrical about x t . cos2i. Fig. Symmetrical about x = 0. the curve by turning the Carte + y2 bx) 2 becomes in polar coordinates (x 2 = a?(x 2 + y2) r = a + b cos 0. = a sin = a sin y = a cos y 3 6 ?. sin2£.
(a: + if . 2. [The locus of the midpoints of the ordinates of 2 2 5 2 2 t (a. + y ) = a (x* . + y ) x'y = a (a: */ (xi 2 (a: ) ! )'. 2 / + 16y 4 2 ) . 1. Ex.2 + j/2). + 2/ o 2/ ) 2 2 2 2 2 2 2 (4 a. EXAMPLES OF CURVETRACING Trace the curves (»* (a: ! III 10 (i) (ii) (iii) + /)• il = = + y )(*&) 2 = + tf)* a!/ «»(*_. ) . »! a* 2 . Trace the curve (a* + 4y 2) 5 = a 2 {xl . ( vi) (vii) (viii) + y 2 ) s =ai y\ 2 + i/2 )=ay. (iy) (v) a.56 Ex.§x y + y*f.•).] .24a.2 a 2 2 2 2 2 = 4a2 a:2 (a:2 + y (a.f). (x' (3x*. 2 2 2 = a:2 (a.
v mj. Suppose that the triangle of reference of homogeneous coordinates is altered. + m3 y' + n3 z' . If the line Xx + /iy + vz — touches a given curve whose equation in homogeneous coordinates is f(x. and then dropping the dashes. We shall obtain the new equation of a curve f(x..". z) by putting in this equation It is = x l — . v by the same constant.. v) = 0.' = M^' + Mo/jl' + M^'. get the new tangential equation of the curve by putting in the original tangential equation (f> (A. x' ± (i) give on y. v) 0. must hold between A. /x (ii) and then dropping the dashes. j It is homogeneous in A. y.y.. not homogeneous in A.. /i. the tangential equation is the condition that the curve should touch = Xx +liy+\ 0. If the equation of the curve is given in Cartesian coordinates. \i in general. We may call f(x. CHAPTER IV TANGENTIAL EQUATION AND POLAR RECIPROCATION § 1. if we wish to emphasize the difference between it and the tangential equation. (i) n3 being constants .. Tangential Equation... z =i af 3 y = l2 x' + m2 y' + n 2 z'. /i.. y' = L x + M y + N z. for it is not altered if we multiply X. z x x = L x + M y + N z. jx.. jjl. where equations solving for x. T z' = L x + M y + JS z. y. We = A =L X' r + L2 fj. + L3 v' v^NiX' + Nzh' + N^' . z) = the pointequation of the curve. v... v.. This relation is called the tangential equation of the curve. a certain relation. '. 2 2 2 3 3 3 . say 0(A. l x' x + rn 1 y / + n^z'. /*. .z) = 0.
The if line Xx + fiy + vz 0. As a particular case. Class of a Curve. The proof is exactly similar to that used of The common tangents m in CO. the line Xx + fty + vz = becomes (^A + l2 /i + l 3 v) This + (m{X + m2 /i+ m 3 v) y' + (n1 X + n2 p + n3 v) z' = 0. . are found by solving these two equations for X: fi:v. This number is called the class of the curve. is identical with X'x' + fi'y' + v'z' = 0. if X'=l1 X + l2 /i + l 3 v. v) = 0. y'. = passes through the point {x'. /it. m : m * two curves of class and M. straight line has no tangential equation. v) = 0. fi. v'= n^K + n^fi+n^ x' . v) of a curve is homogeneous of degree m.58 TANGENTIAL EQUATION IV 1 stitution given For on changing the homogeneous coordinates by the subby equations (i). v... fi. v .. <j>(X. If the tangential equation <j>(X. v) = O' and $(A. y'. But equations (iii) give equations which proves the result. /i'=m1 X + m 2 /i + m3 v.~z7) to the curve are given by solving for A fi v from A = : : Ax' + fiy' + vz' = 0. (iii). 'Hence^m tangents can be drawn from any point to the curve. The point equation of contact of any tangent having the point X'x + fi'y + y'z = is <f>(X. fi. in A. the tangents from (x'. /jl. tangents can be drawn to the curve in any given direction. /x.. z') Xx' + fiy' + vz' = This is therefore the tangential equation of the point. Eliminating v from these two equations. §7. § 2. ' A line is considered as touching ' a point in the sense of § 1 if it passes through it. They are piM in number. we have an equation of degree in A fi. fi. It is to be noted that the tangential equation of a point is of the first degree.0(X. with the curve whose tangential equation has the tangential equation ^ v) — ¥ ¥ ¥ . (ii) on solving for A.
i. joining its (\x + py)/v) intersections with the line Aa3 + fty + vz = to the vertex (0. 3 TANGENTIAL EQUATION OF ANY CURVE 59 where • ^ If . z) = 0. z') by eliminating (x\ between Xx> + fi y' + „z>=0 and g/A = */? = g/„. = (i) we find the lines f{x. to obtain the pointequation when the tangential is given. . e. z').. means the result of putting A'. by eliminating x/y between (i) and the result of differentiating (i) with respect to x/y. obtain the required tangential equation y'. I. we may call u + kv a (tangential) pencil of curves of the mth class. z') and If 7>x if tty 3/ oz „ to / = at (x\ We therefore y'. An alternative method of finding the tangential equation of is as follows.(\x + fty)/z) = <f>{\. /m. Similarly.IV . 0. z) Suppose (a/. namely the common tangents of u and v 0. we find the condition that equation 0(A. The proof similar to that of Ch. \i. v in ~ <>f &c. y'. (i) are coincident if \x + /iy + vz touches the curve. u and v being homogeneous of in A. To obtain the tangential equation of f(x. we get the required tangential equation. y. if we find the condition that (i) considered as an equation in x/y has equal roots. Hence. I. //. that this last equation should have equal roots considered as an equation in X/fi. y. p. degree The curves obtained by taking different values of k have a common tangents. This line must pass through be identical with the taDgent Xx + jiy + vz — (x''. As in Oh. Tangential Equation of any Curve. v. = = should coincide. z') is the point of contact of the tangent = 0. y'. v) two of the intersections of the curve with a side of the triangle of reference . v is for X. = m m = = § 3. § 8. fjt. f(x.y. Two of the lines represented by. § 9. 1) of the triangle of reference.
If in Ex. best when it can be applied. The product of the perpendiculars from m fixed points on a line is proportional to the cosine of m times the angle which these perpendiculars make with a fixed direction.F>. If in Ex.bg — ch ). the tangential equation vanishes identically.f 60 TANGENTIAL EQUATION OF ANY CUKVE IV 3 the curve are of the tangent c are Again. we gential equation required. 1. I. ! [(i) (ii) 27V + 4V 9/i = 0. (iv) 27aV+ 4 27a4 (X 2 a* + l) 8 = 0. ay 2 z = x{x 2 +y 2 ). 4.* known known have the tan Ex.of . + ^) 2 = (4a 2 \2 + aV) s p+ ». s i/) 27aV« + 4(aX + + 4(X. A = be — 2 F = gh — af. 1 the pointequation is deduced from the tangential in the same way as the tangential from the pointequation. 2 A conic is of degree 2 and class 2. is The tangential equation of ax 2 + by 2 + cz1 + 2fyz + 2gzx + 2hxy = A\ 2 + BJ + <V + 2. * The third method is usually method is best in Ex. (x 2 + y2 ) x% iyi = a (x y = a%z$. (See Ch. Show that the envelope of the line is a curve of class m. the equation at the point is ax + by + cz = 0. and that the tangents from any one of the fixed points are parallel to the sides of the same regular polygon. 3. Ex. if the coordinates of any point of functions of a parameter t. &c. Ex.M ) 3 ( = 0. The second . 1 the original equation is a linepair. we have the 2 2 2 original equation multiplied by (abe + 2/gh . § 9. Ex. (v) (vi) 27 a*\ 2 p 2 18a 2 \p V + 4:a (K + p) v 2 i? = 0. » (iii) = 0. °» where Ex. b. . y"zq = x*+«. where a.. 1. 6. If the original equation is a pair of coincident lines. . functions of t.]  (vii) (\ (viii) a + jiV = a2 kV(XY^pPq9 = F p i/«(p + 2 ) Ex. + 26?i/X + 2//V = .) Eliminating t between a/\ — b/ji = c/v. the tangential equation is the intersection of the lines twice over. 2 2 2 2 ). 3 (x + y) = Xs x(x 2 + y 2 ) = ay 2 a 3 » s = xy (x + y + az). (i) (ii) (iii) (iv) (v) (vi) (vii) (viii) Find the tangential equations of kx 3 = y 2 z. * 5.
the asymptotes of the pencil of wics S+ kS' = sets ot Any one of these asymptotes meets n points with the same centroid. in rectangular Cartesian coordinates the tangential equation of the circular points is A 2 + /* 2 = 0. p) = / and (/> being homogeneous of degrees m and m — 3 respectively in X and ft. + 2{GH'+G'HAFA'F)yz + . ' Take the second conic as the circular points. VIII. 2.. &. The circular points are X 2 + /i 2 + » a 2^i/cos^2eXcosB2Vcos C= for trilinear coordinates. Obtain the equation of the directorcircle of a conic whose equation is given in any coordinates. Write down the tangential equation in trilinear coordinates of a conic with foci (a. e..] from Ch. p) + (\* + n") <£(X. 3. A'\ 2 + 2J> 2 + CV + 2 + 2 (?VX + 2 22> = form an harmonic pencil is F> (BC' + B'C2FF')x>+ . Ex.g.] of the tangents from (x.. yj.y + l perpendicular. y) to y = . for this is the condition that \x+/j. The tangential equation of the circular points is obtained by writing down the condition that any line should be perpendicular to itself. § 1. A curve of class Show that the sum of the angles which points and 3 other points.. _(a 2 j3 y 2 ) and minor axis 2k. + vyj (Xa a + M /3a + vyt ) . for Cartesian axes inclined at Ex. [(Xa t + p/3. 1. . the tangents from any point make with a fixed line is constant to within a multiple of ir. the circle inscribed in the triangle of reference. [The locus of a point such that the tangents from it to the conies ^IX 2 + By? + Gi? + 2Ffiv + 2 Gv\ + 222V = 0. . 4.. The inclination 2 2 is given by /= (X +/* ) (\x + iiy)<j>. that the degree of the envelope §4... The envelope touches the line' at infinity where a curve of the pencil touches it... = 0.IV 4 E is EQUATION OF CIRCULAR POINTS 7" 61 ot class /  T!j e envelo Pe of 2n—l. [It follows S= and S' = in two at each of the 2(nl) points is 4(nl) m general. + . = should be self The circular points are X l! 2X /icosa) + /i2 = an angle a. m— m [The curve is /(X. Tangential Equation of the Circular Points. + . Ex. For instance. =i s (X 2 + 2 2 ^ + » 2^cos^2vXcos£2X M cosC).] Ex. and becomes the directorcircle of the first conic] this ' harmonic locus touches the line at infinity at the circular Ex. 5. ? Deduce the equation of a circle with given centre and radius.
P B PQ R P PQ . * But 1 (x\ y z") lies on the polar reciprocal in this case. the polar reciprocal with respect [The polar of {x'. *x+y + v=0 2y If the condition that a curve (in Cartesian coordinates) should is cj> (X. and this touches <#>(A. the sum of the angles which the common tangents to this curve and any curve of given class make with a fixed line is constant to within a multiple of w. 1 . The envelope of Suppose we have a certain baseconic the polar of any point on a given curve 2 is called the polar reciprocal 2' of 2 with respect to the baseconic. /i.] Ex. ji. Hence 2 is the envelope of the polars of points on 2' or the relation between the curves is a reciprocal one. (ii) The threecusped hypocycloid. a point on 2 and the tangent at this point being the pole and polar with respect to the baseconic of a corresponding tangent to 2' and its point of contact. y') with respect to x"1 + 2y 1= is xx' + y + y' = 0. v) = with respect to the baseconic x 2 + y* + z2 = is <p(x. is equal to the class of its polar recip § 6. 2. if 0(0!'. But these two polars are consecutive tangents to 2' in the limit. pointequation is given. y. y\ z') = 0. [(i) As in Ex. which is always the case . and is the point of contact of either. and Q on 2. Equation of Polar Reciprocal. Hence the polar reciprocal of a curve can be obtained when its tangential equation is known. <j> The polar reciprocal of the curve having tangential equation (X. its intersections with 2 are the poles of the tangents to 2' from the pole of I with respect to the baseconic.] § 5. its when The reader ' Ex. v) to x* + = is <l>(x. z') For the polar of with respect to the baseconic is xx' + yy' + zz' = . Discuss the case m = 3. y'. Hence The degree of a curve rocal. 5. (x'. More generally. Also is in the limit the tangent at to 2. 6. y) = 0. v) = 0. Their Consider two consecutive points polars meet at the pole of with respect to the baseconic. = 0. * '. will notice the close connexion which exists between the algebraic notion of 'tangential equation ' and the geometrical conception of polar reciprocal '. z) = 0. Polar Reciprocation.: 62 EQUATION OF CIRCULAR POINTS IV 4 Ex. The polar reciprocal of a 'Lame Curve' (x/a) n + («/b)n = with respect to {x/Af + (y/Bf = 1 is a Lame Curve. If I is any line. touch 1.
intuitive from a figure. If the baseconic has a real equation. Hence to a node corresponds in the reciprocal curve a tangent with two points of contact. For pictures of bitangents see Ch. 1 Ch. except c. point with ' We shall now ' A . which may its with three. four.f Hence from any point on c two tangents can be drawn to the reciprocal curve coinciding with c unless the point is at I. . 7. VI.* In the same way. Fig. which meets it thrice while of the tangents from any point on i one at C coincides with i. A more rigorous proof is given in Ch. unless P is at C. v) = ax 2 + by 2 + cz 2 + 2fyz + 2gzx + 2hxy = (ax + hy + gz. Fig. Ex. 6. and general a common selfconjugate triangle. its tangents correspond a bitangent of the reciprocal curve and its points of contact. 9. distinct tangents corresponds a tangent distinct points of contact. 2. . The baseconio. inflexional tangent. . be called a triple. quadruple. when three of the tangents coincide with i. X. IV. e. hx + by +fz. and its Reciprocal. i. 4. quad. <f> polar reciprocal have in respect to is The polar reciprocal of with (\. c of the reciprocal To a cusp C of a curve and an inflexional tangent flexion J. &c. its nodes. gx +fy + cz) Singularities of Curve = 0. . C P . f These properties of the cusp are evident from the fact that. 3. . . to a triple. . to a crunode corresponds a real bitangent with two real points of contact. any conic. . . XVIII. for which the properties are at once established. §§ 3.3 . which meets it thrice at I. node is a point of a curve at which there are two tangents. to a cusp and its cuspidal tangent correspond an inflexional tangent and * Sometimes called an ' ideal bitangent. except i. 5. Figs. ruple. consider what corresponds in the polar reciprocal to the singularities of a curve. and to an acnode corresponds a real bitangent with unreal points of contact. § 7. /*. They are almost Similarly for the inflexion. tangent.: IV 7 SINGULARITIES OF CURVE its 63 Ex. by Newton's diagram. when three of the tangents coincide with c while any line through I meets the reciprocal curve once at 7. Such a tangent is called a bitangent. any curve approximates near a cusp to a semicubical parabola ay 2 = a. cusps. Therefore I is an inflexion and c the For the cusp has the properties . 10. ' Hence To a node of a curve and . cuspidal tangent i correspond curve and its in that every line through meets the curve twice at C. 1 Ch.
and two real inflexions. . two of the tangents from to the curve coincide with each tangent at O. a bitangent with real points of contact.. a crunode. The singularities of the limacon are an acnode. If is a multiple point of a curve with distinct tangents. two unreal inflexional tangents.. Ex. The relations between the singularities of a curve and its reciprocal are illustrated by Fig. two unreal cusps (at the circular points). This diagram shows the limacon r = 6 + 5 cos and its reciprocal with respect to the circle 4(r 2 — 3rcos0) 55. bi tangents are shown by the dotted lines. and two real cuspidal tangents. 1.64 SINGULARITIES OF CURVE More generally. . its inflexion. [Compare Ch. If <£ (X. 1. a kple point its gents correspond a kple tangent and = Fig I Ex. Ex. The degree and The two class of the limacon and its reciprocal are both four. v) = is <H[> the tangential equation of a curve. To a triple point with three coincident tangents corresponds in general the tangent at a point of undulation and so for quadruple ' quintuple.] . § 4. the bitangents are given by _ d$ S/L4 _ d<£ u\ 5f Find the inflexions. points. These reciprocate respectively into an ideal bitangent. 2. II. ii. to IV 7 and its k tank points of contact. 3.
5. 0) are [Use § Ex. and i inflexions. If the pointequation of a curve is written down at random. Suppose a curve is subjected to r conditions then its polar reciprocal with respect to any given conic is also subjected to r conditions. and satisfies r other conditions. i cusps. Suppose we are told that the curve ±n(n + 3) = S + 2k + t. the reciprocal has r assigned tangents. has k that only a finite number of curves can be found with these Then (Ch. but we shall see that it has bitangents and inflexions. it has no node or cusp (Ch. t cusps. II. k cusps. 0. 8 bitangents. that the polar reciprocal has degree m. five cusps. The singularities of y"zq reciprocals of each other. we have shown. If a curve has degree n. class m. &c. If the tangential equation of a curve is written down at random. 8 nodes. § 6) properties. class n. triple tangent.IV 8 SINGULARITIES OF CURVE = xv+i at (0. tangential equation of a curve is equivalent to writing down the pointequation of its polar reciprocal with respect to the . 7ic and six cusps. four nodes. ' general.] Ex. Ex. Ex. A sextic cannot have a triple point. 65 and (0. The pointequation of its reciprocal must not be considered as written down at random '. has 8 cusps. 2 baseconic x z + y + z2 = 0. and satisfies r other conditions and also nodes. § 6). has r nodes. in § 7. 5 (viii). A cannot have a quadruple point. For instance. The polar reciprocal is of degree m. and §8. but will In fact the writing down of the have nodes and cusps. II. and k inflexions. 1. is of degree n. the curve will have no bitangent or inflexion.] A quintic cannot have a triple point and three cusps. Note that to be given that a curve has a triple point and three cusps is equivalent to only 4 + 6 conditions. whereas a quintic can satisfy [Its reciprocal would be a quintic with a 20 conditions in general. r bitangents. 7. 3. and only a finite number of . t nodes. . if the given curve is made to pass through r assigned points. . 4. for this reciprocal curve is specialized by the fact that it has been derived by reciprocation from a curve with random pointIn fact the reciprocal has both nodes and cusps in equation. a node. 1) Ex. 5.
(iii) (2». m). and length of major axis. say of degree I. 4). (1. point. given point. Show that the general p curves of the family pass through any given point. 2). = 0. v. 4). 2. 2m). 2). (viii) (1. Find the characteristics of pencil of wics. (vi) (vii)*(l. 2). p). Show that the tangential equation of the family <£(X. (i) 4. (ii) (1. z. deduce in(n + 3)82K 1. 2n2). through (x) [(i) (iii) The circles of curvature (2. = in the parameter a. § 2. (v) (n + m. (iii) Conies through two given points touching a given line at a 3. Ex. (1. (iv) Conies through a given point. 1). 2). (ix) Conies with given centre and eccentricity. m). I) is called the characteristic of the family. (2. 4. 4). = §m(m + 3) — t2l. 3.] characteristic of the polar reciprocal of the family : Ex. (v) Circles (vi) (vii) (viii) touching a given circle with their centres on a given line. Hence m(m+3) = r + 2i + r. of a given parabola. (10. which is an algebraic equation of degree p [(p. touching a given line. § 10. 4). 1) (1. (4. (iv) The family obtained by translating a curve through any distance in a given direction. (v) [(i) Curves parallel to a given curve. See Ch.] Ex. Conies with a given focus. and . (ii) (n. and touching a given line at a given point. 2). a a Conies with axes along given lines and passing through given point. Ex. as r = 4. Conies with given vertices. y. (x) (4. 2. m).66 SINGULARITIES OF CURVE . 6). A (ii) Curves of degree n and class m with a common centre of similitude. XI. Find the characteristics of the following families (i) Conies through r points and touching s lines. 1. (iv) (v) (2. 6). passing given point. a) A singly infinite family of curves has the equation f{x. IV 8 polar reciprocals exist the baseconic being supposed given throughout. and I touch any given line. where r + s = 4. (4. The is (I. /i. (ix) (2. 0. 1). We Ex. (iii) The family obtained by rotating a curve through any angle about a fixed point. (2.] (iv) (n. (ii) Conies touching two given lines at given points. a) is also an algebraic equation in a.
and Ch.] [(6. (1. 5 (vi). 5 to 13 should be omitted on a first reading. is the class of the in general. (iii) Quartics with three given biflecnodes and given tangents at one of the biflecnodes.] Ex. (ii) (iii) in Cubics with a given cusp. intersections with the locus of any line through 0.] 1. ' 7. . (1. Verify by considering conies through four points from or touching four lines. inflexion. W =p+ m' = 2p + M. (1.* Find the characteristics of (i) Cubics with a given cusp. 3). is a family with characteristic v). Find contact of tangents from a given point the Pliicker's I). The families are (a 2 + yi ) = 2 1 y (3a . 6). a (y + ax) 2 = axy 1 + 2a?x i y.2). See § 3. Find the characteristic of the conies having 5point contact with a given cuspidal cubic. Find the characteristics of Cubics with a given node. where k = Consider the is a pple point of the locus. Ex. 5. Cubics with a given cusp. (2. = is an equation of degree in x. cuspidal tangent.] Ex. and given tangents 8. XIV. a 3 a3 = xy (x + y + S az). Ex. and the tangents to the locus. and inflexions. 4). 1). § 3. §§ 5 and 8.] Ex. mx. 3). 3). 4) and ax (3. Ex. while <j> (x. three given collinear inflexions and tangents at two of them. [(1. (ii) Quartics with a given node. at these cusps. (ii) Nodal cubics with. 10.] 6. (2. numbers of the locus of the point of to a family of curves with characteristic (p. two given cusps. (iv) Quartics with three given nodes and four other given points. Ex. f2 . See Ch. 7. The families are 1 zy* = ax 3 z (y + ax) 2 = 2 axy + a?x*y. and inflexional tangent. M * Ex. The complete primitive of the differential equation which is algebraic and of degree v ft in — . and flexional tangent. 3). See Ch. inflexion. envelope of the family. (i) given [(1. §§ 3. Find the characteristics of (i) Lemniscates of Bernouilli with a given node and axis. inflexion. 11. 6. Ex. XIII. 4) and (1. and point of contact. 9. tangent. (i) [(1. XVII. (ii) Cubics with three given collinear inflexions and corresponding inflexional tangents. point. m) (/*. 3). nodal tangents. 6). Find the characteristics of Cubics with nine given inflexions.y ).IV 8 SINGULARITIES OF CURVE 67 Ex. [(1.
The foci of the envelope are wple foci of the locus. I). 13. tangents from m'. m [In Ex. the envelope. 12.] Ex. I). where M is the class of A and B are l(m — l)ple points of the locus and the A and B to the envelope are »»ple tangents of the locus. Find the degree and class of the locus of the intersection of tangents from fixed points A and B to a family of curves of degree n and class with characteristic (p. Find the degree and class of the locus of the foci of a family of curves of degree n and class with characteristic (p. m [n' — l(2m — l). — m(M+p)+2l{m — l). 12 take A and B at the circular points.] .68 SINGULARITIES OF CURVE IV 8 Ex.
+(\. ")• * (x. m has in . If co and a> are the circular points.p. b). (i). any is t curve with the same foci f(\. and the lines Sco./m is f\k . /*. *") = o.ii ») = .. drawn to the curve from co. and ^r(\. general m real and m —m un2 real foci.v) + 4>(X. the tangential equation of the circular — 2.v)=0 is points. and has and 1) arbitrary tangents of coefficients. These tangents meet in the real focus (a. since the line joining two real points is real. m m Hence A ' curve of class /i. can in general be If such a tangent also a tangent.: + CHAPTER V FOCI § 1.ii. and <p has confocal with / where (j> (X. curve whose /=0 0=0 %m(m— /2 = m = = 0./« <P (a. and is x + iy a+ib. and cannot pass through co or co'. v) = m 2m = = = %m(m + 3) — 2m = Jm(m— 1) But (i) evidently touches the common arbitrary coefficients.v).. p.. for no tangent from a can contain more than one real focus.. If /(A.. fi. since this is the number of coefficients in \j/. If the curve is of class m. . real foci. Definition of Foci. the general tangential equation of the real foci have tangential equations /. Sa>' touch a given curve (not at a> and co'). v) = is the tangential equation of the curve. . S is called a focus of the curve. As a special case. then x—iy = = a—ib m tangents m from is to'. so that there There are no more than are real foci. is any curve of class is equivalent For to say that a curve is confocal with / = common tangents of / to saying that it touches the Hence the general tangential equation of a curve 0. the curve being supposed real. 0.
.! ~V.] The sum common . = PM P {xx1 f + (yy. Pis the middle point of the of the line joining two foci of a real line joining two real foci.. If P(x. [Taking rectangular Cartesian axes through and the fixed line parallel to the axis of x.) q ^ q 1 \ as a factor.• 1 and 2 Since the difference of these has X and are equal. If the middle point P is real. of the conic] Ex. e the eccentricity. curve 1. p. is the circular points and i/r = properties of the foci of the general curve are The foci of special (but see the examples below). curves may have interesting geometrical properties. m m ' 1 li + q2 \ m 'i H i ~ Ex. ^{x^.] = + /j? (p 1 +p i ~p i + '»)+(Pi>Pt+Pi. That the foci as defined in this section are the 'foci of the conic. if c and c' are replaced by curves respectively confocal with them. yj a focus. Ex.. Show that a — is 3. [A particular case of Ex. and that the directrix is the chord of contact. to since SP is the perpendicular from e PM. and p x cos a. namely .f = Q touch the conic. The tangents from to a curve make with a fixed line angles whose sum is (X. + y sin a the directrix. 161.y) is any point of a conic. 3. Ex. x. suppose the terms of highest degree in the tangential equations of the curves to be t Show p \ TO p1 \ m V+i'2V tana tan0 == . the equation of the conic is Not many known ' ' ' m= = (x— x^ + (y — i/^ 2 = e 2 (p— x cos oc—y sin. a multiple of n. The tangents to one curve make with any fixed line angles whose sum is a. It is seen at once that the circular lines through S. p. where the directrix. Ex. 2. when 2. The theorem is well known in the case of the angles (to within a multiple of w) which the tangents to two curves c and c' make with any fixed line is not altered. a)*. Tangents are drawn from any point to two confocal curves.•) (g + gs gB + )^{q q2 + qi . 58. [See American Journal Math. 5. and xii. according to the usual definition of foci of a conic may be shown as follows.1 70 DEFINITION OF FOCI V any curve of where class m— <f> = 2. By analogy we may call the chord of contact of the two circular lines through a focus S of any curve the directrix corresponding to S.. that Of — is a multiple of n. and the tangents to the other curve make angles whose sum is 0. and the lines joining to the foci make angles whose sum is 0. 4. The coaxial family of circles through two foci of a curve have two other foci as limiting points.
Singular Foci. The intersection of a tangent at a or vice versa. and that the projection of another focus is near the point and so for a of approach of the projection of a>u' and the curve. the curve has three other sets of four concyclic foci. bb'. the of tangents. Ex. If four foci are concyclic. If the curve passes through a and a/. or touching <b a at a. [A curve of class m touching a>u>' at o>.] Ex. Therefore so do the pencils (abed). the centre of a circle is a singular focus. [A segment through « must have zero length. [The envelope of the family must evidently consist of circular lines.* of foci is less than For instance. (b'a'd'c'). If the two curves of the family Pa?+Qa + R = which pass through any point are always orthogonal. Ex. The projection of a curve nearly passes through the projections and a.] Ex. The envelope of a line of given length with its ends one on each of two given curves has no (ordinary) foci. Find a curve with no. where P. the tangents at m and co'f meet in points which are not usually included among the ordinary foci '. the pencils {abed). §7. &c] § 2. other than moo'. cc'. a> of Ex. The four circles cut orthogonally.] * A point is not called coai'. which can be drawn to m the curve from m and a>. Show that the projections of four of the foci are close together and in the limit coalesce at the projection of a singular focus. y and a is a parameter. focus unless it is finite. t Excluding an ordinary if this is a'. is not counted as either a singular or a ' ' . The projection of a curve nearly touches the projection of a>a. IV. focus. of which 1 are real. and (m— 2k) 2 ordinary foci 2k are real (see Ch. <a' and m — 3 other points. m . if the curve has ordinary contact at one point with mm'. singular or ordinary. it will have (m— l) 2 foci. 4. 7. of which For example. R are polynomials in x. wellknown example is the parabola. Ex. a and m — i other points. that the projection of (m1) foci are close to the projection of a. number If a curve of class touches the line at infinity com'.V2 • SINGULAR FOCI 71 Ex. Show . and having cusps at a and to'. c'. and the circle has no ordinary foci. with a tangent from focus. is less than in the general case so that the number 2 . 2. They will be called singular foci. [If a. If the curve has /cple points at m and m. A m— ' m— 1. The case in which the family consists of conies is familiar. b. it will have in general k 2 singular foci of which k are real. a tangent. 2). b'. then the curves of the family are confocal. d are tangents from &> and a'. c. 6. 3. (a'b'c'd') have the same crossratio. d' tangents from a such that the intersections of aa'. dd' are concyclic. Q. and by their intersections determine the foci.
. parallel to the sides of a and ..] Ex.. (a' 2 /3' 2 ) Show that there exists a curve of class n — 1 with real foci at the former points which touches at its middle point the line joining any two of the latter points. the real singular foci of the locus are the middle points of the segments... Ex.. is 4.. ... 2.... § 2. = 0. 6.... m). . t prove that the centroid of the points (pt lt ft). PB2 x remarquable de coiirbes [In connexion with the above examples see Darboux.—xiy)/(x r — iy. + iy. If <r = kir. For another proof see Ch. + iy.. . is with y Ex..x + iy) = n (x. .. 9. s 2'. A m Bm I . and passing through its own singular foci. 5. . 5. 2 = and 2'= n (\x.. n as singular foci. (0C it j3 2).. the locus of the foci of the pencil 2+&2' = (2m — l)ic having an (m— l)ple point at a and w'.. and Ch..] 2 = [If The lines joining any point of the locus of Ex. 13.) foci of is the same for a 7. 8.x. § 1.] makes = The tangents from the tangential pencil 2 + fc2' = real foci of 2' = to any curve of the touch a curve confocal with 2 = 0. Ex.. y. PA = h . . coincides with the centroid of the points (a'D/3'j)... y'. Discuss the case in which a is an even multiple of \ir. . Ex. 6. Show that the locus of P is a mcircular Zwic. Ex.. XXI. y) to d and h. 6. 11. § 8.. an algebraic equation of degree n with roots are OL\ + Oi\ + iji\ . y) and eiia (x. 2' = is a the mth class. and so for . IV. iy... points. .iy'. If f(x) = . Ex. 10. (x. [See Ex. If o. = 0.. Ex. 2 PBn ^ PBn has A lt A lt The has . 5 to the and 2' = make angles with any fixed line whose sum both curves to within a multiple of n. and the roots of /' (x) »! + j3j (X 2 + i/32 ..is constant. . the angle which the line joining (x.. + iy\ xiy). we have the locus of Ex. ..) are the foci of (r= 1.. Bn _.. 4... If 2 = 0. A n PA 2 PAn = h PB* P£ .is an odd multiple of \n.. . . m fixed line by the tangents from any point of an mic through all its foci latter curve are The asymptotes of the regular polygon. + pyr + 1) + (V + fi2) yjf..] Sur une classe .. PAt .+ (\a 2 + /i/32 + l)1 + AiBu A^B^.. PB m make with a fixed line is <r.. [By § 2 = y r ) and (a/.. PB. 1 if (x. r Writing down the conditions that the should touch 2 + &2' = 0.x + iy). The sum of the angles made with a to a curve of class is constant. locus of such that A P PA. If A lt such that pa i . n as ordinary foci. the locus of P ..] PA m and Ex. An . 5. x + iy) (a/. 72 SINGULAR FOCI V 2 are the tangential equations of two curves of Ex.iy) (x'.] PBlt [The difference of the sums of the angles which PA lf . et de surfaces algibriques. = [The curve is (Xa 1 + ^ 1 subtend angles at P whose sum a. Fixed segments + l).. are fixed .. . and eliminating lines joining (x. we have for the locus w n(x. Blt .. Now use Ex. § 30..
the circular lines through the foci are the tangents at the points given = . b) is foci are the intersections of the asymptotes Ex. ( =  tV 1 eW) = 0. Apply to the case of a conic. p) = of a curve in rectangular Cartesian coordinates. 2\V + 2X x2 / 2 Al2)' + V + 2 fl (iii) + 2\ + + l =0. (1^2. If c = a + ib is such a value. another is xiy = a ~ib. If one such asymptote is x + iy = a + ib. (0.0). 1 + 2^21). 0). If x —f{t). 0). (±1. 2. ) 5a (a* + y*)*2f(a?y*) + *a* =>Q.0). Given the tangential equation $ (X. (0. + 2/ 4^+7 = 0. or otherwise. 1). The distances from the pole of the intersections of a curve with its directrices are given by putting p infinite in its pedal equation. (+. +1). x(a? + y*) = ay\ 2 2 (a. Find the real singular and ordinary i foci of (ii) (iii) (iv) (v) [(i) (ii) 2x{xl + y i) = a(Zx tf).0).obtain the ordinary foci. find an equation giving the distances of the real foci from the origin. 3.Bi 6) are the polar coordinates of a real focus.] Ex. and a real singular focus. r^ 1 ««•'. 0). (j)(r. /.1).] Ex. ±1).1^2). find the value of c such that x + iy = c touches the curve.V3 METHOD OF OBTAINING THE FOCI §3. (i) (ii) Find the real 2 2 foci of 0. !! i! : (o. Now eliminate e ei. (±c. «'?' e~W) and .0). 4. y </>(') is the parametral equation of a curve. (±2. (\ 4\V~3X +1 = 0. [(i) (ii) (iii) (1^2. 73 Method of obtaining the Foci. These asymptotes may be found in the usual manner. 0). Ex. (a. (80. ( (v) ± c«_ a4)/Cj ( ) or (0) ± ( ai4_ c4)j/c ) ag c>a Qr c<a _j Ex.1 e. b) is a real focus. [If (r. (4o. the curve being supposed real. To. 6. {x + y) {a? + y*) +2x {xy) = 0. (1. (0. ( + h 0). a and b being real. . 5. Ex. (a./7. The singular which are circular lines. by putting 1/c for and i/c A lor n in the tangential equation. (3 + 2vV2+l. §72). (i) 1. (iii) (iv) (~ia. The product of the distances between a given real focus of that curve of a tangential pencil which touches the line at infinity and the real foci of any other curve of the pencil is constant.
0). 0) and and S of the limacon . 6. 1. 6 the foci of V 3 x=af + 2bt + x x . a focus of a given curve.] Ex. the lines Oco and 0a>' becomfc the which is points co and <o'. c. if any. If we reciprocate with respect to (i. atp+q . and S'co touch the inverse curve. Sa> and Sea' touch the curve. Ill. the reciprocal of a conic with respect to a focus is a circle. and Sea' respectively § 10). a>' and a node at 0. a quartic having cusps at a>. If I. with respect to a circle whose centre is 0). q being positive integers. e. the reciprocals of the foci are the lines joining the intersections of the reciprocal curve with Oco and Oco'. IX. and the reciprocal of the corresponding directrix is the centre of the If circle. As an example. Ex. Find the singular foci. ij = AP + 1Bt+C. if of the curve. reciprocate with respect to any point 0. Hence The inverse of a curve with a> respect to a focus has cusps at of a conic with and <o. Therefore The inverses of the foci of a curve are the foci of the inverse S'a>' : S The inverses of the lines joining m' to the intersections of the curve with Oco are the tangents at a> to the inverse curve is a focus (see Ch. of the curve in Ex. § 10. 7. (i) METHOD OF OBTAINING THE FOCI Find by means of Ex. § 1). 3. y = + t% 8. p and 3 at /(l (ii) (iii) = atp y = = aty(\ +f). For instance. Hence : The reciprocal of a curve with respect to a focus is a curve through a> and a/. § 9 shows the method of obtaining the tangents at the circular points. The real focus of the limacon is the inverse of S. § 4. [Ch. Hence. and the reciprocal of the corresponding directrix is a singular focus.a2 )/2 6. . S'a> are the inverses of Sa> With respect to a point 0. I. Ex. Inverse and Eeeiproeal of Foci. If S' is the point inverse to S >SV and (Ch. Apply it to Ex. we The singular and ordinary with node foci r = a + hcos6 H ((6 2 are (6.: 74 Ex. 7 (iii). the inverse with respect to real foci S and is a limacon r = a + b cos 6. is Hence curve. or Ch. two tangents at a> to the inverse curve coincide.
] 3. curve. 4 HP2 = 4bSP+2a b\ P i being any point of the [The equation with S as pole is 4V r* . for the points of contact of Ex.+ &cos<9 = 0. If in § 4 S and O are reflexions of each other in the directrix corresponding to S. 2r = <*. the inverse curve has inflexional tangents at a and w' which meet at S'. 1 in the cases &>a>0. There are real inflexions if alb lies between 1 and 2 see Ch. 1. IV.] — = 0. [The equation of Ex. XVII. (ii) a is kept constant and a or 6 respectively vary. 4. . Then we get : ' The tangent to a conic focal distances of the point of contact. parabola. Ex.{a + b cos B)/b sin 8. If a = b. a parameter. 2r2 + ar(3cos 2 02)+2a 2 sin 2 i9 = 0.4br (a? cos 8 + _ h*) + (a? V)*0. [0 is a crunode. r(lfc s ) = a(cos6k). the curve being the inverse with respect to a focus of an hyperbola. For the inflexions 3 ab cos 8 + a 1 + 2 b 2 the bitangent 26 cos 8 + a 0. 6.' similarly other properties of the limacon or cardioid. 9.] 1 In Ex. The inverse of any curve with respect to a singular focus has also a singular focus at 0. Fig.V4 INVERSE AND RECIPROCAL OF FOCI is 75 Ex.] Ex. The angle <j) between the tangent and radius vector of a limacon is a maximum when $ = 8. If P is between OP and any point on the limacon. [tan = . 10.] Ex. and Ch. cusp. ». 5. . 5 is not altered on replacing r by (a 2 &2 ) 2 /46V. a = 6. 1. 8. OSP is twice the angle [Invert with respect to 0. is a family of Trace the limacon of Ex. 7. Ex. the tangent at P. 2.] Ex. a>6>0. acnode. Find the locus of the inflexions and of the points of contact of the bitangent of the family r = a + b cos 8 when (i) b. where fc limacons with a common node and focus. ellipse. Fig. makes equal angles with the Obtain Ex. [(i) (ii) r2 +fercos5 + 2& 2 sin 2 5 = 0. Ex. A limacon is its own inverse with respect to the circle through the node with its centre at the ordinary focus. the curve is called a cardioid.
Take. and a.. (3. m) add m to the righthand side of (i). at any ordinary point of the curve. a. .. It is only at multiple points where two or more tangente coincide that we can have superlinear branches. b. the curve y2x+2x2 3xyy 2 + 4. If we take a single ath root of unity for <o we get a ' partial superlinear branch '. In the neighbourhood of the origin it is possible to express the value of y on any branch of the curve through the origin (not touching x at the origin) in the form = = y . are constants.. . We can also have multiple points with coincident tangents formed by the contact of distinct linear branches see. y) be the equation of an algebraic curve passing through the origin. . Ill. Ch... the branch is called linear. for instance. = ax + b<o&x a + C(a(x a + (i). where 1. Through an ordinary multiple point of order Jc with distinct tangents. y..x* = the method being general. just one linear branch passes through the origin. First suppose the curve has a linear branch through the ' '.Fig. f . . + bx 2 + cx3 + in the For points on a curve near (0. Substituting for y the expansion ax * 0. for example. k linear branches pass.* entire portion of the curve near the origin obtained by putting for co every ath root of unity in turn is called a superlinear is y ax. c. 5. are The positive integers in ascending order of magnitude. origin. f We assume the above results as proved in books on the a a> = branch of order a whose tangent = = Theory of Functions and only give here the practical method of obtaining the expansions such as (i).CHAPTER VI SUPERLINEAR BRANCHES » § 1. S. . If a If we take the origin 1. Expansion of y in terms of x near the Origin. Let f(x.
. Putting x and y zero.. x 2 x 3 . method given below we can expand y in terms of * and thence get y in terms of x. get .* As another example.. .. = We and equate a = 0. which has an inflexion at the. Suppose now that the curve has any singularity at the origin and that y = ax is a tangent at the origin. origin with y = as inflexional tangent. p and q being positive integers prime to one another... . b = to G. Put y ax 2 + bx3 + cx* + . We may therefore confine ourselves to the case in which the curve touches the axis of x at the origin. which sufficiently small.. . Similarly for a curve having rpoint contact with y the origin we obtain an expansion of the form = at axr + bxr+1 +cxr+2 + ..) — (ax + bx 2 + cx3 + . Suppose that Newton's diagram (Ch. we assume legitimate when x we have (ax + bx 2 + ex 3 + .. we obtain a expansion = 2. Ill. . Hence y  Cx3 + (AC+G)x* + .. Y (0.. Suppose that * We = 0.. in this equation. we obtain the values of — .... y) = of the curve put y = YXP.. § 3) gives us as an at the origin terms approximation to the curve touching y = represented by points which lie on a straight line making an angle tan" 1 p/q with x = 0. VI 1 EXPANSION OF y IN TERMS OF x 77 is equation of the curve.. . at the origin. .) 2 + 4x 3 = Equating to zero the coefficients of x.. if we consider X. . x = XI.. = thus get a new curve f(Xi. c=AG+G.. y . 0. and have as the required ..)2x+2x 2 Sx(ax + bx 2 + cx 3 + . —j.. in turn. m) is an intersection Another method is to differentiate the equation of the curve repeatedly. b = 8.... .. YXP) as current coordinates.. Putting ax + y for y in the equation of the curve. zero the coefficients of x 2 x 3 x*. Then the expansion may be written down by Maclaurin's theorem. ~j.. In the equation /(*... y = 2x + 8x 2 + 52x 3 + . consider the curve y= Axy + By 2 + Cx3 + Dx2y + Exy 2 + Fy 3 +Gxi + .. c = 52.. we obtain a curve touching the axis of x at the origin x and y being considered By the as the current coordinates of any point on the curve.
. apply Newton's diagram to this curve as before.. Newton's diagram gives the approximation y 2 origin. we put in the original curve X= y=Y + mXP... .. in particular. ' By an we mean b. . Y= and thence ±a* + ibX±ia?(e + ca + %b2 )X 2 + .. of the original curve.. + ai). . .. into the equation of the curve and equating to zero the co2 3 efficients of X. y= YX\ 2 Y = a + eX + bXY+cX Y 2 + This has linear branches through Putting (0. X= Y= Whence p 1 2 y l = xV(m + bxl + cx% + ... y k y = Exactly similarly for the curve = axk+1 + bxk y + cxk ~ 1 y 2 + . f{Xl. that there is d. m) with 0.... m) we get an expansion rn + bX + cX 2 + .). e. in and thence y in terms of x.. and that through this point there is a linear branch of the curve. so that our 0. and therefore put x The curve becomes 2 — ax s at the =X 2 2 . ordinary cusp between the coefficients a.. 78 EXPANSION OF y IN TERMS OF x VI 1 of the new curve with 0. where w is a primiget a complete superlinear branch But if no linear branch of the new curve goes through its intersection (0.. origin. Y= X X . suppose the curve has an ordinary cusp at the y = 0. ill a> 2 Replacing xl by coxl. . ±ai + BX+GX 2 + .. . the tangent at the cusp being For example... order to expand Y in terms of Substitutions similar to the above must be repeated as often is new curve x=Xl. tive gth root of unity.... we have . ±a^x^ + ^bx 2 ±^ ai(e + ca + ^b 2 )xi + . Y + mXP) = We now X as»necessary.. . Then transferring the origin to (0. we xl.. .. . no relation none of them are zero.. ccPxi. The curve is y 2 = ax 3 + bx 2 y + cxy 2 + dy z + exi + ' .. c.
. expanding by the binomial theorem. = x' 1 y 1 yt and Y= l+aX + bX* + cX + we have a = ±1. i if + 3x*yxy'> + 2x = 0. we may use the preceding method to expand x in terms of y.VI 1 EXPANSION OF ' y IN TERMS OF x 79 origin touching which has an ordinary superlinear branch y = 0* we get an expansion ' of order k at the fc + k l fc +2 k fc +3 k y a> = A<ox + B<o 2 x +Ca> 3 x +. coefficients ..). Ordinary '. y (yx3 Y = x 'y. because there is no special relation between the none of them..Y*X\ ! Putting of to zero coefficients c expansion Ex. [Putting y 3 . s Zx2 yi + xi yxS = Q.. If it does. ' This process is known as reversion of series '.. .. we may expand x in terms of y near the origin by putting tt a = & ^ a a 12 (l + 4^ + B^ + .. 2 = YX\ x = X. b. 6 = X X 3 . . §§ 3 to 12). The arithmetic of the methods described in this and the Devices for saving some of following section is often tedious. and equating coefficients of powers of y on both sides of the resulting identity. 2 2 = + 2/ z)(!/ + a.. Ex. where a<)3<y<. IX.c. In expanding y in terms of x we assumed that the curve did not touch x = at the origin. a of b P gives an expression for x.. have found the expansion Jcth y y = bx + cx a + .] 2. ) ! V(^ * ' a. being any If we root of unity. = Y X*. . (i) is y =x = + Hence the required 2 (I ±x + x2 ± \x*+ . of the following curves near the origin : (ii) (iii) (iv) (v) (vi) (vii) {y~3?f = a. Expand the branches (yx2) 3 =ary..). ..Vyx"i = xi y — 2y*.. we get (Fl) 2 equating 1. for a curve touching y = at the origin. for instance.. being zero. and then obtain the expansion of y in terms of x by reversion of series. 1. the labour will be described under the head of 'quadratic transformation' (Ch. Find the expansion of the branch of the curve (yx2 f near the origin.. Each possible value in the given expression for y in terms of x..
. Hence + 2:£ + £a$ + fa: + a. y y = x (2lx + 0x + . Find the expansion of the branch of the curve (yx2 ) 2 near the origin. = xfy i y [The method of Ex. Ch...).] 1! a..+ a.X80 [(i) (ii) x EXPANSION OF y i i 3 y IN .] 2 Ex. 2 r) c<. y = x% + Sx xi + .. = y (liy + 26y2 + .. 3 4 coefficients. We get Put = y*2y + ^yf + . a.... Find the expansion of the branch of near the origin.V.. = 2. y =xb + 2x3x%+ .:r+ .— 2a? + . X=ff we get 1) after putting 7 = YX\ Y+X .. A Show curve has a rhamphoid cusp at the origin with y that near the origin = as y [Cf.). $•..e = —2.V(B 3 aj5 + .. x x (iv) = 2^4 (l+2y* + 16^1 + . Writing therefore Y=r)g.. Ex. o) 2 2 2 2 2 + ...)...] y = 2 3 s 4 ) .. c = f..V.).. i s Ex...+ aass + w x%& + aslx + ..] ax 2 + &#f + (X s + dxi + . + ... (i) (ii) (iii) (y— 3 ) 2 = ifx™ + yx* 5. i s ^.. B 6 Ex.. + <aa. (ii) x = ±\yi (1 + Ayi + By+ . and compare = 0. expand.. 2 a..(Y+ 2 f approximating to origin. § 8. 6.. TERMS OF e» a.. = yi (1 + Ayi + Byl + . y = x*(l±2x~ix' ±9x + .. 8y = x*Zx Zx* + . where 3 = 1.. 2/ Find x in terms of y 2 3 4 2 if (iv) = 4ic + 8a. a: and compate co and the expansions obtained on replacing yi by (iii) u>y§ and <o 2 2/&. 2 x= x = X...).. i [(i) Put efficients.. y = x?{\±x + \x ±\x + . b = £. We get « ±ly i. Ill.ly±Hy  + . Put then instead y = becomes Y 2 = {Y + 2 f ..4...) ^(l + ^ 7 = l + ag + 6£ + + <2£ 4e£ 6 + ..). $ = .) and y = x {\ + xSx y = a.) and two similar expansions... 3.. = (l +  .)..t (1 +aw + §a. y = = x2 + x* + . X and the curve X X Y =X 2 at the Put in this We get a .. 1 gives a nonlinear branch at (0..... Ex.. tangent. VI 1 (iii) (iv) (v) (vi) (vii) = x. 2a.
We showed in Ch. i. I. § 7)..+% = F(x. y) fo«i) (#««) 1 (yu n) = meet In while = „ in nN .(u n v Jf ) x .2.f (i) One curve has a node or cusp at the origin.(u vN 1 ) x (u 2 Vj) (u 2 v2)..... and the other passes through the origin but does not touch it there.: VI 2 INTERSECTIONS OF TWO CURVES § 2. . G« . (ii)..+bN =(yv )(yv . The number is e. * It is the f Cases 221fi (i).. this ult u2 . e.. 81 Intersections of Two Curves. also that no intersection of the curves other than the origin lies on x = 0. vN are functions of x. I.. 1 that two curves of degrees ••• N o 0) + a2 y n ~* + . ciently small. (v) have been already dealt with in Ch. later on. — y being expansions of y in terms of x for branches of the two curves passing through the origin.. i. We assume that x = meets the curves in finite points only and does not touch either curve. when x is suffi... which will be useful = U{ and y = V. determinant just mentioned multiplied by a constant... Let us consider the following examples.(yv N) = points. § 7. ar b r denote polynomials of the rth degree in x v x v 2 . y»+b 1 y»* + b2 y»* + .. e..2.n.. In this equation ^>(x) is a polynomial of degree nN in x* Suppose we wish to find the number of zero roots... It is evident that the lowest power of x in Uf—Vj is x° unless y uit and y = Vj are (partial) branches through the = origin. x (<!*„*>!) = 0. . of the expansion (§ 1).(u 2 vN) (u n v a ). § 7. where xe is the lowest power of x in (f>(x).. ... un and which we have obtained . the number of intersections of the given curves which coincide with the origin.. n. Hence The number of intersections of two given curves which' coincide with the origin is the index of the product of the lowest powers of x in all possible expressions of the form u^ Vj.N.j = l.. y) =a b y + n a^ I. the product of the lowest powers of x in each of the expressions (ufVj) i = l. f(x. The result of eliminating y between the equations of the two curves may be expressed by equating to zero a certain determinant (Ch. . 2 ).. or in the form (f>(x)= Kfi) (u1 v2 )..
Hence . and the curves meet thrice at the origin.... u 2 = a'x + b'x + 2 v 1 = ax + Bx + .... 2 82 INTERSECTIONS OF In this case TWO CURVES VI w1 = ax + bx2 + ......} {(a'4)a: + ... no two of the Ic + taDgents coinciding...% = a'x + b'x2 + . (iv) One curve has a cusp at the origin. . . Here Uj v2 = axi + ..... (iii) The curves have a node at the origin. the tangents at the node to the two curves being the same. and the curves meet JcK times at the origin.} {(a' —a)x + . ....}... . . . . v3 = Cx 2 + v 3 = — Ax% + or v2 = Bx 2 + .. . and k are of the type U^ The curves Bx2 + Cx s + . In either case the curves meet eight times at the origin. at the origin. . meet k (k + 1) times at the K origin. Ul Here = axi + bx2 +. linear branches through the (v) The curves have k and origin.u2 = axi + bx2 . n(Ui—vA = {(a— a')x+.... — vA are of the type Here k(k— 1) of the factors in ax + bx2 + cx 3 + . * Here II (^ — Vj) has IcK factors each of the type K K ax + bx 2 + cx3 + ..* (vi) The curves have k linear branches through the origin with the same tangents for each curve. .. ..... . . . two tangents at which coincide with the cuspidal tangent.} ..... Hence n(wf «_.}. = Axi + «2 = — axi+. .. Here 2 %! = ax + bx 2 + ... as might have been anticipated. v% — a'x + B'x 2 + . and the curves meet twice at the origin.i\ = Ax 2 + Bx 3 + .. * This result imght have been expected ..) = {(aA)x+ .. while the other has a triple point.. (ii) One curve has a cusp at the origin.v1 = Ax + Bx 2 + .... . and the other passes through the cusp and touches the first curve there. .... for each of the k branches of one curve at the origin meets each of the branches of the other curve once at the origin.. ..} {{bB)x2 + and the curves meet six times {(b'B')x2 +...„ vx = bx + cx2 + ..
Show that they meet q times at 0. 2 and 3 to obtain the conic of origin with x — y + xy + x 3 = 0. Comparing the and the expansion of Ex.(>— 1)} = k'L] § 3. if the expansions of y in terms of as far as the terms in x r . 7. x* in this m = 1.2x s 2 [x + y=y 2 and y — x = (2xy) . Show x for the two curves are identical have (rfl)point contact at the of y in terms of 2 3 2 Ex. Ex. h. meetsit in at the origin to the superlinear branch points coinciding with the origin. 2 and § 1. § 3. IX.... is 3. See also Ch. x2 = Sat/. But a less laborious process •will usually be the following. TANGENTIAL EQUATION NEAR A POINT The tangent /3 83 of § 1 Ex. [See Ex. We may of course find the tangential equation of the curve as in Ch. How many of their intersections coincide at ? Ex. the curves origin. (ii) to find the lines joining the intersections of curve and conic to the origin and to choose m. Ex. 8. [b 3 Find the conies of closest contact at the origin with a 2 x + by 2 + y 3 = and x 3 + y s = 3 axy 2 = (ai x + a2 b2 y)x and if = Sax.] {a?x + by ) 5. 6 = 2. A [Sr {(f. b so that five of them are y x.1) (r+l)/r+(kl) . n having distinct tangents (h a {k — l)ple point such that the tangent to the branch of order rt of the first curve is a tangent to a branch of order rt — 1 of the second. and then employ a method similar to that used in § 1 for pointcoordinates.. . The expansion 2 x i for the conic 2 y= mx + ax + 2hx(ymx) +b (ymxf y Ex. a. ] = Ex. [For the curve y efficients of x. Tangential Equation near a Point.. 4.. = mx + ax + 2ahx + a(4:h + ab)x + 2ah(4:}i = . Another curve has at *'ii ^z. Other methods are (i) to differentiate the equations of curve and conic four times and to identify the values of dy J ' x + x + 2x + 2x + 2x + . . h = 1. 6. 2 3 i 5 closest contact at the co x 2 x3 . IV.] Ex. Find the parabolas of closest contact at the origin with = x + y + xy + x 3 and y = x + x2 . (i) Ex . a—1. VI 3 1. Use Ex..] curve has a &ple point with superlinear branches of order = 2j). Two curves have linear branches touching the same tangent at 0. where p^q. Suppose that it is required to obtain an approximation to the tangential equation of a branch of a curve near a point. curves have linear branches through the origin. + Sab)x* + . 11.. one having ppomt and the other gpoint contact with the tangent. 2  Two that.. G 2 . 3 we have <Py cPy ' d*y ' dx' dx 2 dx 3 dx* for curve and conic at the origin .. § 12.
Transfer the origin to the point The line becomes (h. The above process gives now X= +aZ26Z " = ±aX 3 + cZ 3 . .. + cxi + dx ± 2 3 2 . and obtain the condition by expressing v in terms of X. B. which X is \^". we have on putting . substituting ' e ba>0i a i it is s + ca>y£" + da>s £ a + .... 0).C. in the neighbourhood of a cusp at the origin at which y = is the tangent. Then substitute the value of given by (iii) in the second of equations (ii) and we have v expressed in terms of A.) a(r + fcA0... for i\ and writing £ = Xa = (bPm0Xt3. For example.a /* Comparing with = 1 and X = A0.l + 6Z* + cZ 5 +. T where £ =X 2 .. § 6.a) <»7j7 + Xx + fiy + v + hX = 0... we have = y = ± ax% + 6a. and v by y.. 1 " &(/3a)ffl0X0 + c(ya)a>7X'Y+.. Ex. just as in § 1 we expressed y in terms of x.a ) + d8co s X s .. and we have the pointequation of the branch by § 1 in the form y= The tangent $ 7 ba>Px tt + C(6~'x a + (£. 84 TANGENTIAL EQUATION NEAR A POINT VI Suppose the branch touches y = at the point (h. .a + cy<»7X7«+. s dws xa + is (i) at the point y) of the curve or. . . . in turn... We shall take ft = 1. We should touch the want the condition that \x + fiy + v = branch. IV. Xx + fiy + v + h\ — 0.Q aA0°= (&/3w0X0. . If we now replace X by a. 0).a +.. we get the pointequation of the polar reciprocal of the branch (i) with respect to the baseconic x* + 2y (Ch. J' Substituting in the first of these equations = W * X=AA + BA 2 + CA 3 + (iii) and equating to zero the coefficients of powers of A. 1).a + cya>yXy. 3 .a) a>PXP + c (y . we get A.)xOcy b (j8 .
. 1 = a x + a3 x + ai !c + . find the number of common taDgents by recip Ex. = = &.. coincide with I ? [One or q according as the pointsingularities do not or do coincide. § 7. Show that.. . IV. represent the same linear branch touching y at the origin. we prove that The number of those common in § 2. 1. Ex. involves o 2 a„ . 2.] Ex. . 2. §4. Ex. 7.. Show 1 that. in the first of these equations.. if 3 2 3 and . a.. Ex. i/ Expand v in terms of X for the curves of § i 1. . = = We may also rocation. 2.. 2. +1 a r+! . but not a.. V of this branch with respect to ~ 4 27a* X3 166 ~81tf X4 +  This equation represents a linear branch with an inflexion at the origin. Ex.. r of their common tangents coincide with the tangent at P. (i) to (iii). we get .. Ex. Common Tangents of Two Curves. As tangents to two curves. The polar reciprocal x2 + 2y — is .. Reciprocate and use § 2... Two curves have superlinear branches of orders^ and q(p^q) How many of their common tangents with a common tangent I. is the index of the product of the lowest powers of A in expressions of the form ui — Vj ) v u$ and v Vs being expansions of v in terms of A for branches of the two curves touching the axis of x. [See §2.. _ 166 ...„ whence 4 . X=A\ v _ 2 .. Ex. v b 2 X + b 3 \ + b i X*+ .. and equating to zero the coefficients of A. that to a cusp of a curve corresponds an inflexional tangent of the polar reciprocal. verifying the result of Ch.] . if two linear branches have rpoint contact at P. 1. 2. which coincide with the axis of x. A 2 \ 3 . and § 3.: VI 4 COMMON TANGENTS OF TWO CURVES 85 Putting + B\* + C\ 3 +..X 27 a a 3i^ + 81a 4 ..
of a point of undulation is the tangent to a superlinear branch of the third order. From equations (ii) and (iii) of § 3 we have v + hX expressed in ascending powers of A. deduce at once that The polar reciprocal of a superlinear branch of order a. Ch. and therefore a + a' = /3 = /3'. unless the index of every power of A in the expansion of v + h\ is divisible by a factor of /3 — a. Hence a' = 1 j8' = a + 1 as was to be proved. superlinear branch of order <x at is a whose tangent meets it in /3 points coinciding with superlinear branch of order a. we should obtain 03'O/73' = a//3 and /3*a'>a. j9'>/S.. § 7)... v in the tangential equation of the branch (§ 3). We For taking the tangent to the superlinear branch at the origin as axis of x. + l)point contact with its tangent. the lowest power being A# where A A^ 0( Hence a' /S — a and /S' /8. .8' and . is in general a linear branch having (a. 86 RECIPROCAL OF SUPERLINEAR BRANCH VI §5. This is inconsistent with the preceding. of § 3. y = But the origin is a multiple point of order k. (cf. =  = = = /3a>a'... Therefore j8 of the theorem at the beginning of this section is a + 1. the Cartesian equation of the curve takes the form y'X uk . But considering the original curve to be obtained by a second reciprocation from the reciprocal. so that the tangent to the superlinear branch meets the other branches in k — oc points at the origin. . y for A.. For example. IV. 5 Polar Reciprocal of Superlinear Branch. (3 of § 3. 1 . a = uh + u k+2 + . If the polar reciprocal of a. then a+ «' = /? (i) = /3'./3>/3. It is seen at once by points coinciding with the The polar reciprocal is obtained by putting* a. in which cases (j8 — a)//3 a'/.' at 0' whose tangent meets it in fi' points coinciding with 0'.: . The theorem may also be established directly by the process . Take the superlinear branch § 2 that its tangent meets it in origin. where ur is homogeneous in x and y of degree r. &c. the polar reciprocal of an inflexion is a cuspidal tangent. Putting we get an equation of degree Jc + 1 for x in general. '.
.. Also the number of tangents from a point on the tangent at which coincide with this tangent is equal to the number of intersections of a line through 0' with the reciprocal branch which coincide with 0'. Ex. 10. Ex. = 4. = 4. de France. so that a' = 2. /8' gives us Hence the relation a + a! . p. /3 tangents from coincide with the tangent at 0. ' ' ' ' vi (1877). [For the nomenclature see Halphen. We see that the reciprocal singularity is 2. /3 /3' a rhamphoid of the same Ex. origin with y Suppose a curve has a superlinear branch of order 2 at the = as tangent and expansion y = ax i ±bx* + cx ±dxTz + !> . 2. Bull. If a is the order and — OC the class of a superlinear branch. The origin is Ex. 6). Then a = cusp (§ 1.VI 5 RECIPROCAL OF SUPERLINEAR BRANCH 87 The number of tangents from which coincide with the tangent to a superlinear branch at is equal to the number at of intersections of the reciprocal branch with its tangent which coincide with the singularity 0'.. and j8— a tangents to the branch from any point on the tangent at coincide with the tangent at 0. Write down the order and class of the branches in § 1. Math. . de la Soc.8 = = If a superlinear branch of order ocatO meets its tangent in to the branch j8 points coinciding with 0. the order and class of a branch are respectively the class and order of the reciprocal branch. type as the original. 3. 1. 2 to 5.] Ex.
the curve from a vertex V. meeting a given curve of the The locus of P such that degree in Q x . the hth polar curve {for each value of k) projects into the kth polar curve of the projection of with respect to the projection of the given curve. and so for second polar '.. (n l)th polar . 2..2.. Q2 pQ is OQ.. curve called the third polar curve of with respect to the given and so on. . .. .n.. n 2. If we project of * Or ' ' first polar of for the given curve '. If I is the polar line of 0..i* j) with respect to the given called the second polar curve of curve .n..OQz + . ... is ZOQi.curves are of degrees n 1.j. . is called a ' pole ' of I with We .2.... (n— 3)ic.O Qj OQ k/PQi PQj PQ k = k=l. Let be any fixed point. Through 7ith draw any line OP Qn .* Similarly the locus of 20& is • OQj/PQt .. — — — . . conic. (n— 2)ic. if there is no ambiguity . n— respect to the curve. third. CHAPTER VII POLAR CURVES §1. Polar Curves.j=l.+ OQn_r + . l FQ2 PQ n ) called the first polar curve of with respect to the given curve. the locus of such that P (i. It.j=£ k&i. They are therefore also called 3... second. &c.i ^j) . {n— 2)th. line of with respect to the given curve.PQj P such that = (i. the polar (ml)ic. 1 respectively. shall see in § 2 that the first. .
Equation of Polar Curves. PVQ .Y ' X+/i \z + /iZ\ X+n J .' VII 2 For in the EQUATION OF POLAR CURVES figure..OQk/PQi. cosec VQP __ VO VQP ~ VP OVQ PVQ' • .. PQ k = P'Qk' o. OVQ PVQ and similarly O'Q' _ VO'. be (x. Fig. The locus curve. sin sin P'Q'~ VP'. while the The point Q dividing /Xx'+iiX X + fi \y + /i. z) = OP so that OQ/QP = X/fi is \ P 0. 89 denoting projections of points by dashes. of points a pair of straight lines whose polar conies degenerate into is called the Hessian of the given The theorem just proved shows that The projection of the Hessian of a curve is projection of the curve.P'Q±. VP ' SOQ^OQ. y. we have . be (X. . Hence PQ ~ Therefore. 0'Q{/P'Q(. oosec VP. y. if OQ _ P'Q' ov X vo VO' VP' . Z) and In § 1 let equation of the curve is f(x.PQz. sin sin . z). PQ~ OQ = VO sin sin OVQ . the Hessian of the §2. = o. Y.. i.
Y. Z put for X. and then put 0=1. z.) = = . z respectively. (a...+ V to F— + Z) ly W f J means where Also x. 7i! r V da? <>y <>z/ or by Taylor's theorem.90 If this lies EQUATION OF POLAR CURVES on the curve.e. y/z) (See Ch. Z. of f(x/z. + y y^ + —y j / means the same thing with X. (a: + 2/ + s) z 7' = 2 apqr xPylzr . y. ^y. Xz + fiZ) i. + ±. Y.Xn* /i 2! z(x*+Y?+Z±) f dz/ V ix <>y 2 + . y. = 0. To find the polar curves. We have at once The hth polar curve of an nic is an (n—k)ic. if the equation of the curve is given in Cartesian coordinates as f(x. on the &th polar curve. Hence the equation of the /cth polar curve is If Q lies \ x 3 + Y Ty +z ^)f=°> ™ ( Vx <> <> / Ji / i <s (*rx + yw + 3 \ n ~k **z) f=°> the two equations being of course identical. we find the polar curves by the above method. I. § 3. In this (x~. y) 0. x. the sum of the products of the roots taken k at a time of the equation in X/fi is zero. Xy + fiY. VII 2 f(Xx + /iX.
)'/ 0. If a curve has a curves of any point centre 0. polar of P. If Pand § being (xlt y lt z and {x An alternative method in Ex. If P lies on a given line. so are the asymptotes of the polar curves of any infinitely distant point. If the asymptotes of a curve are parallel to the sides of a regular polygon. The rth polar curve of to the (rs)th polar curve. 1). must be concurrent at 0.raic. if P lies on a given the class is N(n — 1).)*(*£ + ". is the sth polar curve of Ex. 6. Any polar curve of the (x/a)« triangularsymmetric curve + (y/b)« + (z/c) n ' = and the polar with respect triangularsymmetric. is ' ^0 3y*~ ' ^0 ' 3z* ' 2. yt «. (0.. Q lies on the (n— &)th polar curve of P.* the polar curves of at infinity have as a centre. [The polar curves are (*s + Ex. (^ + )^°. is brought into selfcoincidence by rotation through 180° about . the locus of O is an (n — 2)ic. ABC A B B A [Take Ex. 12. Ex. for which z = the polar line of (0. 10. 5. x) P. +un = 0. there is in general one position of and 0. § is node of the (kEx. an . If P lies on the kth. ' Ex. 0. 8. &c. ABC as triangle of reference. More generally. 0.] Ex. EQUATION OF POLAR CURVES The kth polar curves of the 91 vertices of the triangle of reference ^„o 3a* Ex. polar curve of Q for an nie. 1.] ~' . 3. and find the number of tangents from JVic. a node of the (n — k)th nolar of l)th polar of Q. On the polar line of O there are three points whose polar curves have an inflexion at O. 1).VII 2 Ex.° and 2 .. first [Take the curve as z n +s is n 'i u^ + zn 6 us + 5. 5. See Ex..] Ex.. P is a also the rth polar of [It is The &th polar of P with respect to the rth polar of Q for/ = Q with respect to the kth. (ii) If a triangle can be inscribed in the polar conic of O which is * The curve O. (a. If the polar conic of with respect to a given cubic has as a selfconjugate triangle. Ex. the polars of with respect to the polar conies of and C...] 9.).. If and C? are given. 7 is to take P and § P for as vertices of the triangle of reference] 6. 11. the envelope wic is of class of the polar line of P with respect to a given [Take the line as z = n — 1. 4. (i) If the polar conic of O with respect to a given nic has a selfconjugate triangle which is inscribed in a given conic.  +. Ex. is 7.
= If the given curve is ' azn + (b x \y) n x + + (c x2 + 2c 1 xy + c 2 y 2 )zn ' 2 +. we show in the same manner that x + bx y Any an If rple point at is polar cwrve of an rple point of a given curve has with the same tangents as the given curve.+ ^. Math. Folar Curves of a Point on the Curve. 49. 375. p. the kth polar curve of (n— k)l (n — k — 1) v ° ia/ + If b lies is (JI2)l (^ + 2«i^ + «!») «**. and that they are independent of the position of 0. is . ix. 9). and the case of an jtic only meeting the line at infinity at the circular points. a factor of in general an inflexion of the given curve. on the given curve. and are the same for all wics with the same infinite points. [See Salmon's Conic Sections. Suppose the triangle of reference chosen so that Then the &th polar curve for / =' 7> h is :—^ f =0 by § 2. Show that there are n such axial directions in general. a = and the tangent at = 0. de France. Qn •••> OQ1 OQ2 Show that an line is drawn meeting a given nic in 'axial' direction of the line making = P § 3.. is (0. the locus of (of. Through a point Qn any OQn a maximum or a minimum is perpendicular to the polar line of the point at infinity in this direction. Ex. or a triangle can be inscribed to one and circumscribed to the other. See Bull. &c] Ex. 1).. The &th polar curve of P with respect to an «ic having an (n— l)ple point at is an (n—k)ic having an (n— h — l)ple point at 0. de la Soc. . [Use polar coordinates. (i). 0.: z ! 2 92 EQUATION OF POLAR CURVES is * VII selfconjugate with respect to a given conic. more generally. If two of the tangents at to the polar curve coincide (or are perpendicular). the locus of is straight lines through O. The reader may consider the cases in which the conies touch. and similarly The polar curves of a point at which the tangent has" rpoint contact have rpoint contact at with the same tangent. And.] Ex. Discuss the case in which the curve passes through the circular points any number of times. 14. §§ 373. a = and b x + b x y is c x 2 + 2c x xy + c 2 y 2 Hence is an inflexion of all the polar curves. a 2 (n— 2)ic § 9. Qi> . . 13. Note the case n 2. Hence the polar curves of all touch the given curve at 0.
as is seen by taking the origin at the cusp and writing down the equation of the locus.] Ex. (n — /• + 2)th. . 7. which we do not count as a parabola..] is Ex.* + f2) [If a rectangular hyperbola is a (n — 2)ic.y) = un + un  1 + .x^z)zx+{ab)(^y. . Ex.] whose polar conic with respect to a given ( = 2 + r.n x)xy has the triangle of reference as a selfconjugate triangle. 2. 1. where « t is homogeneous of degree k in x and y.. 8. There are conic is a circle. a The nic is a parabola locus of a point whose polar conic with respect to a given is a 2 (n 2)ic. . 1. The locus of a point whose polar conic with respect to a given and the nic has given eccentricity (=£0. x factor. The [n — r)Va. 6 passes through each cusp of the «ic.] Ex. . The number of points on an nic whose polar conic rectangular hyperbola is n (n — 2). The locus of a point (az 8 + ty 2 + cs ). ij. 10. — . Ex. 1. general 2n(n — 2)— 2k. 6. + Mi + m = are rectangular hyperbolas.. 3.. The n asymptotes meet in a point and are parallel to the sides of a regular polygon and so are the tangents at any multiple point. polar curve of Oat is (nl)/(nkl).. the equation of the curve in rectangular Cartesian coordinates is f(x. where k is the [The locus of Ex. the polar of the origin is Wh = Ex. [It is (a» = 11 + brf + c£>) {x 1 + y* + z*) Ex.. asymptote passes through the origin. The number of points on an is in »ic whose polar conic with respect to the curve is a parabola number of cusps.— VII 3 POLAR CURVES OF A POINT ON CURVE 93 Ex. ^Therefore If one common u k = r*> (a cos k 6 + b sin k 6). up and w„_i have a 0. »ic is 4. The polar conic of a cusp is the cuspidal tangent twice over by Ex. (»l) y+(nkl) ax 2 + (.. or 2i) is a 2 (n — 2)ic number of points on an nic whose polar conic with respect to the curve has a given eccentricity is 2n (n — 2). The (n — r+l)th. The ratio of the curvature at of an »ic to the curvature of the kth.. If all the polar conies for f(x..] The polar conic of f) with respect to (6c) (r)ZCy)yz + (ca) {(. polar curves are nonexistent. polar curve of an rple point of an «ie consists of the r tangents at the point. 9. Ex. u n .. the locus is + !{ = 0. [If the curve is = y+axt + . 5. then u k = are lines parallel to the sides of a regular polygon. (n— 2) s points =j£ in the plane of an raic whose polar \<A = 2 <Sx <rZ> t)y* and dxdy = are equations giving the points. Ex. y) = 0.] If two asymptotes meet at the origin.
if ra>2. being the equation of the curve. 11. + a n x parabolas and conversely every curve all of whose conies are parabolas is of this type.] O with respect to a pencil of »ics Ex.. the curve is a (2» 2 — &)ic with a (r— l&)ple point at and B. o. Ex.. then the line cr 2 „ goes through 2n+1 . VII 3 n are Ex. . [Take the line as z = 0.. is circular.. P P P [If u + kv = is the pencil and z = <)x the given line. 0). is 3« <>x 2>v 2iy _ iv in <>y Ex. such that oj is the first polar of Pj for the given curve... . 2^=0 /= Taking r = 2 or Ex. [Take k = nl in Ex. If ?/ S 0.] every polar conic is a circle. . we have the two following examples. The curve is of degree and class n. P . oo ). 16. The tangent at O to that curve of a given pencil which passes through O goes through the intersection of the polar lines of 0#rith respect to any two curves of the pencil.2 „. If every polar cubic or bicircular quartic. The centroid of the intersections of this curve with any other lies on a fixed line.15. 0. the curve is a circular cubic and we Ex. if <»„_! = 0.Yf of a given pencil of mics have a node in general. 17 P%. 16.] Ex. 1. All the polar conies for y — a + a l x + a 2 a? + . 18. P P . 3 (n . the curve is a circle. cr 2 is the first polar of 2 for cr 1 cr 3 is the first polar of s for o3 . The kth polar curves of a point form a pencil of (n — &)ics. As (» — l)ics trace out the same locus. P P . 0) and and (0. If every polar curve of degree r goes through two fixed points and B. 14. 3. 10 such that factor of a. 18. 13. the line at infinity being the tangent. is any point of the line. the first polar curves of If form a pencil travels along the line the basepoints of this pencil of of (n — l)ics. If a line meets a (2» + l)ic in 2 „+1 .. derive a series of curves <r lt <r 2 . 19.. The polar line of such a node is the same for each «ic of the pencil and the node lies on the 2 (»l)ic of Ex. 17. Ex.. is a un .. the locus . Then substitute in ay ay _ lies / ay y The centroid on x — 0.1 at (0. The locus of the poles of a given line with respect to all curves of a pencil of nics is a 2 (» — l)ic passing through the points of contact of those curves of the pencil which touch the line.. — [Taking A and B as (1. . 12. [Choose axes of reference for the curve of Ex.. It has a superlinear branch of order » .] A A Ex. 94 POLAR CURVES OF A POINT ON CURVE .
ill iv .
(iii) x = y. HARMONIC POLAR OF AN INFLEXION Show that the harmonic polar of 3? (0. is the tangent at the point. Prove PF. and y— z = the harmonic 2. 0) is 60a". if any. Ex. Then chords through a point O of a cubic meet the curve again and Q' while they meet the polar conic of O in and fl'.1 + 2 (c x + c x y) 2»" 2 + 3 (d x* + 2d lXy + dtf) n ~ s + Since this meets the given curve at (0. Then the tangent at (0. Ex. Q. 0. (ii) x+y = 1. 1. R.z . Take the curve as (i) in § 3. 6. Any line through an (n — 2)ple point O of an »tic cuts the curve again in Q 1 and Q%. make in [Use the harmonic property of the diagonals of a quadrilateral. 96 Ex. OPQ' are drawn through the inflexion O Show that PF and QQ' meet on the harmonic polar of O. VII 4 — 1. The first polar curve of (1. Hence G is the point of contact of a tangent from 0. 0. QQ'. are concurrent. the first We = . 1) to the given curve is 2/ — and passes through O. Two P and P . Show also that the tangents at P.] Ex. if (OP. OR. « (y*) •=& + <*) (*b* + V)= x y + y' + x + y. Any two chords OPQ. RE R R [As in Ex. = 0. polar is at G.0. 5. 1) (1. O as (1. consider now the intersections with a given curve of polar curve of O with respect to this curve. OS meet the cubic again at four points on a conic also touching the cubic at O. Ex. Take (7(0. 1. Q. . 0)* and any point as B(0. 0. 3. Hence a single intersection of the curve and its first Firstly. First Polar Curve. Q 1 Q2 ) is harmonic. we must have one of various alternatives. 0. [The first polar of O less the inflexional tangents at O. the two chords adjacent.] [The polar conic of (0. 2mx—y—z = polar. . 1) as such a point of intersection. S. * 3.1) (0. we may have a = b = 0. Q. concurrent.] 4.] for for %x* l i of a cubic. . — 1. OQ. A conic touches a cubic at O and cuts it at P. 0) for (x + y + zf + ^hxyz = 0. 0). Show also that the tangents at Pand Q meet on the harmonic polar of O. 0. Find the locus of P. Ex.] § 5. The curve and the first polar curve have' C as an ordinary point and the tangents to the curves at G are distinct. 1) is (2mx—y—z) (y — z) <= 0. Show that OP. 1. 1). 1) for + y3 +s? + Gmxyz = is the same for all values of m. (i) (ii) (iii) [(i) Find the harmonic polar of (0.
1. on a curve.] Find the tangents Ex. 1 + 2~4). if is the class of the given nic We We sum up our first The m (Ch. the reciprocal curve is of degree m. 1) to a?+y = 2z s 3 = hxyz. 1. 9) to x3 + y s 2 3 = . the meets Ex. y^b^ + iz (x + y) = i The (y + 2xyx ) = 0.: : VII 5 FIRST POLAR CURVE 97 Secondly. the given curve may have a cusp at G. It follows at once that. Find the tangents from (1. § 1) to ax(y*s*) + by (z'x^ + cz 7 or Ch. 16. 3a) to ay (a. first polar touches the curve at 0. and thrice at each cusp of the given curve. and has r nodes and i cusps (Ch. e. %xyz. (v) (vi) to = a?.r 2 /) = 0. on a curve must be considered as coinciding with a point the tangent at 0. 1. a) (0. 1) 1 meets the curve where (y^Vxy + x ) contact of the tangents are therefore 2. and the first polar curve has G as an ordinary point. Hence two of the intersections of the curve and its first polar coincide at G. (Jo. 1) is to {a? + y 2 ) z = 2x s i . * 2'ilfi H . 1) to x'+y' + . (i). 8 the number of nodes. the tangent at G to the first polar curve not coinciding with either tangent to the given curve at G. 0. 0. 1. If the curve has t bitangents and i inflexions. e. s 3 (i) "From (1. The given curve has a node at C. i. [The first polar curve of (1. I. we may have a = b = b t = 0. § 7). Taking « = 0as the tangent to the cusp. intersections of the first polar of Hence If lies i. IV. we have a —b = b1 =c = x c2 = 0. For m = n(nl)2S3K m is the number of tangents from 0. n= it m(m— 1) — 2r — 3t (ii). This points of —1 +2i. (. 1. IV. see that the first polar touches the given curve at the cusp. x 3 + ys = a See Ch. the number of with the given curve at points which are not multiple points of the given curve. and k the number of cusps. (iii) (iv) From From From From (11. (1. (ii) From (0. § 2. twice at each node.* Thirdly.twice at Hence two of the tangents from (§ 3). VI. and three of the intersections of the curves coincide at G* results thus polar curve of meets the given curve once at the point of contact of each tangent from to the given curve. is of class n. § 2).
I) to touch at points on the first polar and meet the curve again on § 6. This is a linepair if (X. ay a^ax which degree degree is ay a^ay n in n — 2. (ii) (iii) (iv) (v) (vi) . —3. z) = X IX* "'" y ay* 2£ +2 .«. c*x= {a±(a2 c2 )i} {&±(&2 c2 )4} </. 0. 6 to the case in which tangents are drawn to an »ic from a («3)ple point. Z) lies on ay alF a&a^ 2>x 2>z H= I ay aj/aa. x = at2 y = a? where t = 1. xh/(Bxi 2xy + Sf) = 0. Equation of Hessian. is with respect tof(x.] 2 or . the equation of the Hessian of / x. Heesian we put z — 1. § 9. ay ay 2 <>yiiz = ay as 2 o. 2. Y. (xy)(2xy)(8x2 + xy + Uy') = 0. Ex. y. I. Extend the result of Ch. In § 1 we denned the Hessian of a curve as the locus of points whose polar conic with respect to the given curve The polar conic of (X. . Z) degenerates into a linepair. 3.98 FIRST POLAR CURVE (i) VII 5 [The points of contact are given by xi + 2x*y + 2xf + y l x/y = 0. .1 + a/2. If / 0. each element of the determinant = is of is of Hence The Hessian of an nic is a 3(n — 2)ic.>£. = Ex. z. Y. y. To find the Hessian of a curve whose equation is given in Cartesian coordinates we make the equation homogeneous by Then after finding the writing x/z and y/z for x and y. [The tangents from (0. „JV + 2zx }>Z7>X aFaZ + 2xy ZXIY = 0.
= and K = meet. [Putting i = 1 in § 6 (1) the expression in the first brackets {} is a/a/ ( a 2/ /W ^f dr2 ? 1 dr 7>8 3rd0 ^7>6J (*f'y v/\ 2 I dr/ d<0 j and the expression in the second brackets ^d»Vdfl is { } i Vf{#£ j. but also twice at each intersection /=0and = 0./ 7>y 2 except that / sections of / 2 of coincide (0. z and add the first shall obtain the identity * two rows to the third. . 2 Vda. not only at the inter= and H — 0.VII 7 INTERSECTIONS OE CURVE WITH HESSIAN 9!) add the rows by Multiply the columns of the determinant by x. JJ\ .dj/ 3 2/ _. 0. By its relation Hessian with H= of § 6 the intersections of an %ic / are the same as the intersections of = with /=  Ix ly Ix ~by \ly/ da. Ex. 3/ d 2/ da? d 2/ da. and then multiply the first two columns to the third . For by Euler's theorem on homogeneous functions 3/ da. 1) .1). Intersections of a Curve with (i) its Hessian.„ n 3/ da. 8) = given in polar co ordinates. the intersections of To determine Hessian * H= 3/ as on / = 0. s. H x. Find the Hessian of a curve f{r. *„ di/ 3s dzds H2 . we shall take with any point and for this purpose we may evidently replace the /= and H =0 which by K = 0. or when an approximation to the Hessian is required in the neighbourhood of (0.1 fill 2 2 l dW */V ' 1 J r*\l>rl>6 > 18 J § 7. We i ) a/a/ *h = iniY \h Zx v ' yf ' 2>y %x<sy r^V 7>xf . y. y.r^V \hyJ \ix^  l ^ dj/ 2 This gives an alternative method of calculating the equation of the Hessian which is convenient when the equation of the curve is given in Cartesian coordinates.0.
Summing up the above results we have : The intersections of a curve with its Hessian lie one at each inflexion.(d x + d y)y 1 2 z in 'J +. Hence x2 + c. . = 3n(n2)6S8ic. conic of c is ~ The polar If this is n 2 i. K H The curve and the Hessian intersect eight times at the cusp (Ch. keeping only the highest power of z. k of cusps.. . then. « l ..l)byz + 2c^xy = 0. y K Hence K = . 100 INTERSECTIONS OF CURVE WITH HESSIAN VII 7 Suppose. 0. H= = ~ ~ / = y2 zn 2 + (d x s + 3dx x2 y + 3d2 xy 2 + d z f) zn 3 + .. Suppose now/= has a node at 0. = (and = 0) has a triple point at at which Hence two tangents coincide with the cuspidal tangent of the given curve. 1) and the tangent at on / = = We have byzn ~x /= + (c x2 + 2c t xy + c 2 y 2 )zn 2 + (d x* + 3 d Y x 2y + 3 d2 xy 2 + d 3 y s)+. is a cusp of the given curve at which Suppose now that Then is the tangent. and K = 24.6 +.. is y point is (0.. and the coefficient of z™ 1 a linepair. Ex. VI. In this case = 8( Cl 2 ^c c2) (c Q x2 + 2c 1 xy + c 2 y2 )z Sn . VI.. either c = ~ = = The Hessian meets a curve only at the inflexions ple points of the curve.. K with the same (and 0) has a node at tangents as the given curve. so that 6 = 0. If 8 is the number i of nodes. the curve . . the triangle of reference chosen so that any 0. The curve and Hessian therefore meet six times at the node (Ch. and i of inflexions of the 7iic... or else 6 of the curve.e. (iii)). In the case 6 =£ 0. Ex.. c 2 and multi = 6b2 d1 )y6b 2 d x}z 3n 6 K= {(Sbc 1 + . § 2.2 y 2 + (n. in /is a factor of the coefficient of z / and is a multiple point has an inflexion at 0. § 2. and eight at each cusp of the given curve. = (and therefore the curve and Hence the curve and the Hessian) meet only once at 0. six at each node. (iv)).
But by § 5 m = n(n— 1) — 28— 3k. (x) (xi) (xii) (xiii) a x3 '+3a l x 2 y + 3a 2 xy 1 + a 3 y 3 + 6xyz 2 2 2 y z + x y + z x + 6pxyz = 0. = x3 y + y'6 z + z 3 x = 3 . xyz = h (x + y + z) (ys + zx + xy). 2 [The Hessian is (x3 + y 3 + z3 ) = (2m3 + 1) xyz. z=i x3 g xz g z lc 2 2 3 : l s . — ixy + y2 ).P + *. at the nodes The number of intersections of the Hessian with the curve and cusps may also be obtained in the following manner (due to Cayley). Find the Hessians of the following curves y*>z q : (i) (ii) (iii) = a. (iv) (v) (vi) (vii) (viii) Gz^ + y = (x + y) {x 2 x s = z (x 2 ±y 2 ). 0. {x+y + z)(2yz+2zx + 2xyx2 y2 z2 ) + 2kxyz = 2 2 lie on (x + y + z) (x + ay + a z) (x + a y + az) = 0. meeting the original curve where xyz 0. Eliminating m.2 = 0. 1. x(x 2 Sy 2 ) + a(x 2 + y 2 ) 2 ) = 0. 0. Hence the inflexions are the intersections of the curve with the sides of the triangle of reference. = 0. 0. * (xiv) (yz + [(i) (ii) x2 ) 2 = xy ' flexions (iii) (iv) (v) xp+t 'y2I 2 z2 ^. A and B such that = 3n(n2)A8B K k= 3m(m2)ATBi.] m = Ex. x3 + y 3 z2 x (ix) y 2 = xy(x + y + z). 2. x3 + y3 + z3 = h(x + y + z) 3 . A6 = 3A2B2 = Ex. 3 xy 2 = z(y 2 ± x2 ).VII 7 INTERSECTIONS OF CURVE WITH HESSIAN 101 For the curve and Hessian meet altogether in 3n(n — 2) points. 8. 0. 3x(x2 3y 2 ) + a(x2 + y2) = 0. i n= m(m— 1) — 2t — 3t. = y(y — x)(y — x). \x + y + z) 3 + 6 hxyz = 0. so that there are constants t . ax (y 2 — z 2 ) + by(z2 — x 2 ) + cz (x 2 y 2 ) = 0. or A = 6. B = = 0. The in . from these four equations. the latter equation being derived from the reciprocal curve as in § 5. Find the Hessian of x 3 + ys + z 3 + Smxyz and find its inflexions. The number is evidently independent of n. we have (A6){ni 2n283 K )(n2 283K) + 4:8+6K} + (3A2B2){3n 2 + 6n + A8+(Bl)K} Since this must hold for all values of n. r.
7. 4. The Hessian of the ' triangularsymmetric n ' curve (x/a) n +(y/b) n + (z/c) is = the sides of the triangle of reference n — 2 times. 8 is the tangent at an inflexion O. A curve of odd degree has at least one real inflexion.102 INTERSECTIONS OF CURVE WITH HESSIAN VII 7 2z{x + y + (x + y) (x*4xy + if) = 0. Ex. If a curve has rpoint contact with its tangent at 0. [By Ex.}. Ex. — [The Hessian of yu + v = 0... The number of inflexions of an »ic is even or odd according as the n is even or odd. 8 is the envelope of the (»2)th polar curve of any point on the given line. the 2(n— 2)ic is called the 'poloconic' of the given line..=!v *!/_/_£/ y = a i>x ! 2 (n — l)ic is 5j/ 2 \iix<)y/ the degenerate polar conic of a point on the Hessian is unreal. + . 12 xy* = z (12 g^x 1 + 36 g.i Jix) (x) (xi) (xii) l s i 1 (xiii) (xiv) Ex. In the case n = 3.] Ex. If the locus has a node at given line of Ex. the Hessian has (r— 2)point contact with the same tangent at and the curve and . 6.] l i ) i . = 0. [If /= is the Mio and z ft = the given line. The locus of the points whose polar conic with respect to a given raic touches a given line is a 2 (n — 2)ic. 10.. a x 3 — a 1 x'y — a s xy + a s ys = 2xyz. 8 (yz + a. 2 2 (6 c)(fy+c2)a. while in the other the «ic has no acnode. 8.. Similarly for x n y + y'"z + z n x = 0.] Ex. . 3. the with the given line as one tangent at O. 5. 9. + . z\{(l+. Each side passes through n points of the curve at which the tangent has mpoint contact and passes through the opposite vertex. + . (vii) (viii) .2 ) 3 + 1 6 xy s (yz + a.] Ex. The locus of the inflexions of zx^y varying values of a is 2y* + x > z = 0. S(x s + y") = xy(z — x — y). where the terms of lowest degree in Fare of the (r2)th degree. Ex.] If F>0..xy + g^).k' )xBy}+x{k 4 x*k (l+k >)ry + (lk + k t )y*\ =0. X +y s + z? 9 p*(xz2 + yx' + zif) 3 (18 p s + l)xyz = 0. 8 l y (2x + y)+ax for [The Hessian is x% (2 ax* + 4 xy s + xi yz + 8y i ) = Now is eliminate a. Hessian meet i 2 times at 0. 2 ) + 3 y e = 0. x^z + ifx + z'y = bx^y'z*. The 2 (»2)ic of Ex. 4 each such point counts as (n — 2) inflexions. This (n — 2)ic divides the plane into parts in one of which the nic has no real inflexion or crunode. is of the form yU+ V = 0. = {x^ + y' + z } {a{We^)x + ... (vi). a 6(n — l)ic. where the terms of lowest degree in v are of the rth degree. the p. Ex. The locus of the inflexions of a pencil of nics Explain the case n = 2.1 .
1) to the inflexions of zx^yi + u homogeneous of degree p + q + 1 in a. If p= they become p (p + \)u = xy oxoy . suffixes 1 and 2 denote partial differentiation with respect . Suppose that X. [Eliminate z between the equations of curve and Hessian Ex. ^= 0. 14. With the notation 2 a Ali 2 or use x and y of = Au p \l 22 + X 22 /i. The tangents to a curve at the multiple point (0. where u is The lines joining (0. /'l2 ^M. q. and y.i . Show that the tangents to the Hessian at the same point are in general u and . A fcple point of a curve is in general a (3& — 4)ple point of the Hessian.).VII 7 INTERSECTIONS OF CURVE WITH HESSIAN = 103 Ex.= ' v(. 0.] Ex.! c AM  = »( A V + . v are homogeneous functions of degrees p.a^ + v) + ( w_ l)( 6wA + ^A. 1) are u 0. 9(2+1)* ^S +i>(i' + l)r^ If = 2 i'2^ oxoy q. they become o u v„ = ox 2 0. 12. 13. ix. 0.. Ex. r not less than 2.» = y y ox* oy 1 \oxoy) Ex. u u #u . 14 (v). and the tangents to the curve at are tangents to the Hessian at 0.^ + V.2 X 12 ^ 12 ''1 . where the to x and y. are = 0. 11.
y. Ex. Y..) Ex.. 16.l)Xfts 2a. [If k —2 in Ex. ^) with respect to a curve f(x. Z). &c. or 5..7 104 INTERSECTIONS OF CURVE WITH HESSIAN VII Hence show that the Hessian when expressed ^Sn—Sk2 in (iii) terms of 0= —y1) a (»fc)MflMM + s 3n3kS / fe _ 1 )a {2(fcl)(»fc)«aM.) the curve = by Ex.ji» = 2(21) i^„ + p (p. Eliminate » between the equations of the curve and Hessian. rj. n = 10.. The curvatures of two branches of an wic and its Hessian which touch at a crunode of the wic are in general in opposite directions. &'= 7 k = .. + ^. (iv) In particular the Hessian of degrees (»— 1) and n'rax and y. is using Newton's diagram.l {(fcl) 2 (w^l)(ttS + 2pgMi..] Ex.. In what cases touching a given nic at a fcple point true that each branch of the Hessian is inflexional ? n = [nkZnW + k + Z = 9.1 1 + ax k "'y + + bx w+ . The ratio of the curvatures of the a fcple point branch of an nic through and of the branch of the Hessian touching this branch atOis [If (nk)(kl) is : (nkSnk2 + k + 2). which are derived from (p — 1)\ (v) = x\ n +y\ n .] 15. from the > identities 2(j>l)(sl)X 1 /i 1 =2(gl)pA u +2>(i>l)V 1 i2y'aA ^ 2(pl) (21) A. where u and . 14 (iii) 2/ (x '. + [n(*+8) + (* + l)(*2)]«i IW } 1 .. &c. 17. is called the Steinerian = P . [(i) In the definition of 6Aft substitute A^. ?> are of (*2)«aw. u lie on and on „ {nl)va for m = (n2)uam . . Q = = y(nk)(kl) + (nk3nk s + k + 2)bx* + the branch of the Hessian touching y at the origin . z) of degree n has a double The locus of point at Q (X. The Steinerian. gives » = 10. Suppose that the first polar curve of a point P(g.. the ratio of the curvatures is it is (2 — ») : n. 15. fc = 4.] §8.)2fe(fcl)(wfc)ttaMa . .s (v) The inflexions of zu+v (»l)a«M z + (w2)aM!) = = — *»« MB 2(n2)wa.2 aAll (pV)(q\){\ 1 ^ + \n l ) = q(ql)i1X u +p(p\)'K ll n + 2xya Kll . . is st» +» = 0.
rj. i/. Hence of : If the first polar curve of P has a node at Q. g from equations we have iV aZ 2 */ ay " ay ay azaF aza^ aF 2 ay ay aFaZ ay a^az Hence The locus of given curve of aTal ay zz 2 = o (ii). 112. It may be shown that. 9. a^aF the double points of the first polar curves of a is its Hessian. 183. « we have by Ch. 3 2y ' II. Another proof is given in § 9. §§ 8. yy ^p^=0. (i). ^ ^ (i) ° £> V> 3(%2) 2 . namely.+ VII 8 : THE STEINERIAN down 106 first of the given curve.ay / = + *ZIX^ '«iiF TS Z. 2 az +y aF + aX2 s Q *f 32 ay 2 ay a 2/ rK hY iz' ""'hZiX 1 ay n ~^azaF = are the equations (i). d2 § / aZ 2 ^aTaF + ^aZa* =0 ay i a2 / / ^vi^ry + Tr2 + *dF2>Z' aFaZ 'aF 2 T . the polar conic Q is a linepair meeting at P. see Henrici. Suppose now that corresponding points P' and Q' on the For * Salmon's Higher Algebra. 9. Z. § 4. an extension of the theorems in . i the conditions that the l+ a2 ^ ^ + yy ' = °> should have a double point at Q. London Math.„. ii (1868).. Writing polar curve r of P. Y. Ex.* Eliminating . § 76. since each of the three equations (i) is of degree n — 2 in X. Soc. ^ from the equations + »2 7. The conditions that P should be a node of the polar conic / Z' . pp. F. (i) and replacing £ by a. z we have the equation of the Steinerian. the Steinerian is in general of degree Eliminating . Proc.
^ dX+ & d¥+ ^z e*f + r) s<>x + v ?)Y + hz ' dZ h +  ' using (i). curve* 3(w1)(to2)2<54k. . i (1873). y\ z').¥_ These equations show that the polar line of Q is is the tangent at identical with the line through the points and P'. k are the number of nodes and cusps of the given * For the effect of multiple points of a curve on the degree and claas of the Steinerian. Bull. see Koehler. (+d£) and (X + dX. 1249. nl. P P If the first ofQ is the polar curve of P has a node at Q. polar curves of P and P' go through Q and Q' respectively. This Hessian and first polar curve are of degrees S(n — 2). Z) on the Hessian. Z + dZ) Then. Y. Y+dY. de la Soc. and may be shown to meet twice at each node and four times at each cusp of the original curve (Ch. Hence: The class of the Steinerian is where S. Math. §2). the polar line tangent at P to the locus of P. tangent from (x\ y'. such that (ii) holds and also A X 7>X +V +Z 7>Y *Z~° Therefore the tangents from (x\ y'. z') to the Steinerian are the polar lines of the intersections of the Hessian with the first polar curve of (x'.: 8 106 THE STEINERIAN VII Steinerian and Hessian consecutive to dinates (£ P and Qhave coor + d£.. ¥ . VI. de France. other than the double points of the original curve. since the first respectively. ^ +d Kli + or. z') to the Steinerian is the polar line of a point (X. rj + dr). which Hence to the Steinerian in the limit. pp.
v ^) v / / \ / S2 / a2 /\ Identifying them and eliminating X. tj.^Y) v <^izix~ wix) + y^^ziY. . ¥ Therefore the given line (n — I) 2 intersections of . The two curves (i) touch at this point. we have tions : If two poles of a line coincide. y. . v. (i). Moreover. 0. The tangents at the point to the curves (i) are = < / x idxa2 v d^) + y{ a2 x ^Y.. Suppose that the line Ace + py + vz (X. 107 The Cayley an. ' which are the first polar curves of (to (v. 1). — p. f=azn + hzn Ex. Y.). 2.y*+zni a {d a? + %d1 x y + ^ di xf + ds y*) + Use Ex. Summing up. Z). a node at Q (0. Z) with respect to f(x. . — l) 2 poles with respect to Suppose now that two poles of Xx + fiy + vz qoincide at the point (X. /j. z) = 0. 1. = is the polar line of Then / ¥ a ¥ is . /x J ¥ = v +¥ Zz 7>y . The Gayleyan is the envelope of the tangent at the point of contact of two first polar curves of the given curve. and that the . If the first polar curve of P(l. a Hence : In general a given line has given nie. the Gayleyan (or Pippian ') of the ' given curve. polar conic of Q is to show that Q lies on the Hessian a linepair meeting at P. 1 y + cz 1 n . the corresponding point (£. v between the relaHence thus obtained we get equation (ii) of § 8. 0. Y. Y. Z) lies on the Hessian. 0) l with respect to /= has . 0..— X) and (0. A ^dz the polar line of each of the ¥ = v ^¥ dee and . the coincident poles lie on the Hessian and the line touches the Steinerian at the corresponding point. (X. Ex.VII 9 THE CAYLEYAN § 9. If the first polar curve of envelope of is called the PQ P has a double point at Q. ^) of the Steinerian lies on either of the (identical) tangents by equations (i) of § 8...
a (U. W = dec the curve 7>y iv ix <)x yv yv = iy Zz <>y 0. [The common tangent d^x + d^y = 0. 6.] PQ 4. P' and Q' are consecutive pairs of points on Steinerian at Q'.] is PQ the Steinerian for the degenerate polar conic of Q. Show that O is a node of the Steinerian and that the line is a 4ple tangent of the Cayleyan. [If 7. curve The Hessian and Steinerian of the 'triangularsymmetric' Ex. The Steinerian and Cayleyan touch the inflexional tangents of the given curve. Ex.] and Hessian. V = 0. Ex. 9. Jacobiau of Three Curves. 10. P = 0. 12 (ii) meets the Hessian at the points corresponding to the intersections of the Steinerian with the given conic. The first polar of any point on the tangent at Steinerian touches PQ at Q. are the equations of three curves. W) = 3 (x. y. as it is usually written. P and touches the Hessian at Q. . yw yw Zz or.] Ex. and the Cayleyan into the vertices of this triangle. at The tangent P to the harmonic conjugate is [The tangent at polar conic of P is the polar line of Q. The Q is ' n(nl)asi + 2(nl)bzy + 2cy' i = 0. z) is called the Jaeobian of the three curves.] Ex. [Take O as (0. Ex. [These tangents are the polar lines of the inflexions. Ex. If U= 0. 3.] § 10. v. The locus of § 2. which • naz + by = 0. 0. P to the If PQ Q. 1) and the quartic as s z* + z u 1 + z^xy + ti t = 0. The tangent at Q to the Hessian is the for the tangents to the first polar of at Q. (*/«)» + (y/by + (z/c)« = degenerate into the sides of the triangle of reference. harmonic conjugate of [The tangents are dqX 1 + 2d 1 xy + d2 y l Ex. The polar cubic of with respect to a quartic degenerates into a line and a conic. 8. of 5. 11.108 Ex. THE CAYLEYAN The second polar of P touches the Hessian is VII 9 at Q. PQ and PQ' meet Ex. it touches the Cayleyan at Q.
Hence a first polar curve is only a particular case of a Jacobian. Take the tangent to W= W= 1) and y at (0. 1) is an intersection of J= 0. 0. as = 0. 1) Hence the Jacobian cuts once at in this case. dently of degree is the locus of a point whose polar lines with respect to the three given curves are concurrent. 1) (i) in the following cases. . . and the n\a 0. Hence the Jacobian . if 17 and are straight lines. VII 10 JACOBIAN OF THREE CURVES lines of (x. Y. ^iU „bU Z being current coordinates. the Jacobian is the first polar curve of their intersection with respect to 0. Arranging the determinant just given for Jz in descending powers of z.r ic 0. That curve of the pencil U + kV = there. . Multiplying the columns of the determinant J by x. may generalize this investigation by finding the intersections with a given wic of the Jacobian of an 7i. y.. In § 5 we discussed the intersections of a curve with any first^ polar. 0. It follows that the Jacobian of three curves projects into the Jacobian of their projections. at once verify that. 1) touches W = which passes The Jacobian does not in general touch at (0. If the curves are of degrees n n2 ns the Jacobian is evi1 . = with a curve of the pencil each point of contact of W W= W— . we readily find that the Jacobian passes through (0. an n 2 ic V 0. and taking for U and V the most general expressions of degrees in x and n 2 also arranged in descending powers of z. 0. = V= W= We U= = W= W= <>x Jz = Suppose that (0. z) 109 The polar with respect to the three curves are V IU X. 0. we can express Jz in the form 2>U We %+n 2 + n3 — 3. y. through (0. Then W=b yz 71 '1 + (c x 2 + 2 Cj xy + c 2 y 2 ) z n~ 2 + . 0. z and adding the first two columns to the third.
8. « s ic respectively. V= W Ex. (ii) b  0. Ex. The Jacobian of ofthefamily 4. TT= is also of aVW+bWU* cUV = the locus of the nodes 0. If O is a cusp of a curve. with y = as cuspidal tangent. W ="0 has a cusp at 0. less the nodes of counted twice and the cusps counted thrice. If the curves are of the same degree. 4 to 7 take O as (0. = w2 = n s and ftx = ft 2 = ft3 O is in general a (Sftj— l)ple point of the Jacobian. Q is on the Jacobian. 1. Ex. The number of curves of a pencil of Nics TJ+kV = which = of degree n and class m is in general touch a given curve 2n(Nl) + m. Some important below. the Jacobian of the curves is an (ra. 0. 0. Ex. 1) (0. a Xyple. every node of W= (0. Now (iii) W= b has a node at 1).] 0. of the Jacobian does not vanish identically. Ex. If O is a cusp of = at O. If the two other curves have the same degree. Any three curves of the family all+bV+cW = have the same = 0. the Jacobian touches the third curve at 0. — 3)ic with as a multiple point of order not less than i 1 + & 2 + s — 2. O is a (2 ft t + fts — 2)ple point of the Jacobian at which ft 8 tangents coincide with those of the « 3 ic. This tangents to If = may and = also be proved at U+kV=0 by identifying the any point of inter section. V= and 0. Jyple point of an n^ic. 0. If «. [The points of contact are the (2N + n—S)n intersections of and the Jacobian of £7=0. the Jacobian has a (k— l)ple point at O. [In Ex. Now use equation (i) of § 5. three curves pass thrdugh 0. the Jacobian has a cusp at O with the same tangent as the given curve. 0. V=Q. Ex. touches If O is a &ple point of W W= W= . These results are true only if U and V are perfectly general. the Jacobian of this curve and any other two curves through O has a node at O one of whose branches touches the given curve at O. 1) and the Jacobian passes = twice at Hence the Jacobian meets W 0. through (0. 0. 5. « 2 ic. special cases are discussed in the examples If the first polar curves of P with respect to three given curves are concurrent at Q. U= 0. +«. 1).2. 7. = c = «! = 0. 0. the Jacobian of V= 0. provided the equation Jacobian as the three curves U = 0. If aU+bV+cW = 0. The Jacobian touches y = at Hence the Jacobian meets W = thrice at every cusp of W = Now (0. Ex. W= 0. Ex. If O is ifc " .' 110 JACOBIAN OF THREE CURVES VII 10 JJ+kV 0. If n 1 = n l and fcj = ft. V = 0. 3. ftyple. 0. 6. If two of them are of the same degree. their Jacobian passes through is a node of the Jacobian. + ». 1).] W W=0 W—0 W=0 .
if the nic has ordinary multiple points with distinct tangents at some of which the Nics have ordinary multiple points of assigned order.z) = 0.y. The number Ex.1). The number of curves of a singly infinite family with characteristic (p. less each node basepoint six times. IV> §8. 7 the Jacobian of the raic and two Nics of the pencil meets the wic at the points of contact required. A pencil of Nics has as basepoints points of an nic. of points of contact is therefore (2N+n3)n~6(8 + K )2r.] Ex.] Z) = (m . 1). a) = equation. the . take the node as • Ex.] that the result of Ex. say yjr (X. while r+8 + x = £N(N+3)l. and twice at the other r basepoints of the pencil. and </> (A.S £ fc (& . [By Ex. the points of contact of a curve of the family with the line \x + fiy + vz are the intersections of this line with the curve obtained and = 0. y. n. (IF 7>F iF \ To discuss the intersections of (0. and q curves of the pencil touch the «ie at a point which is not a basepoint. 3. The reader may find . . 12. Show that the of Nics touching the nic at a point other than a basepoint is > (nN)(N»> + 3) + 42)2. Ex. 8. where k is the number of cusps of the mic which are not basepoints of the pencil.VII 10 Ex. The two simplest other cases. 4. . 0. each other node twice. v. 10. class m F=0 [This example is a generalization of Ex. y. 6. [By Ex.] F= and = at » node. 13. Then the points of contact of a curve of the family with are the = with the {l+pn— p)ic intersections of — /= F F= . a. n. v. Any curve of the pencil meets the nic at p points other than the basepoints. x. p. cases are n = 2 and n = 5.Ex. If 24« 2 + 25 is a square number. 11 still holds. [Use Ex. is the family.] [The family whose nodal locus we require is the second family of Ex. 14. I) which touch a given curve of degree n and is in general nl + mp. 4 to Jacobian of the 7 the q points of contact are the intersections of the nic and two Nics with the nic.2) . Show that q = 2p + 2(D.Nics having a (k — l)ple point at each fcple point of the nic Show where 6. the locus of the nodes of 2 those  {y/ (24w2 + 25) — 3}ics which go through the 3» intersections of three nics taken in pairs is the Jacobian of the xics. each other basepoint twice. a) = its tangential If f(x. each cusp basepoint six times. six times at each node and cusp of the wic. 9.1)— k'.1) (« . v and degree I in x. Show that the result of Ex. 9 is true if the wic has multiple points. z. JACOBIAN OF THREE CURVES A an 111 k cusps of number where pencil of Nics has as its basepoints each of the S nodes and wic (n N) and other points of the nic. D = (»1) («2)8k. by eliminating a between where yjr is homogeneous of degree p in X. 11. and each other cusp (k' in number) thrice. For the definition see Ch.
(ii). (v).CHAPTER VIII PLUCKERS NUMBERS §1. cusps. 8. Pliicker's Numbers. or '(i). Then. m. (ii). t. k number of n degree. k. n i (ii) (iii) = m(m — 1) — 2r — 3t. k. Thus. =3m(m2)6T8i. — k = 3 (m — n). t. i are called the Pliicker's numbers of the curve. (ii). we get (vi) and . of which the most useful are perhaps : We = m= = = = = (i) m = n(nl)283K. for instance. class. (iii). assuming (i). the other relations may be deduced. t number of bitangents. (ix) %{nl)(n2)8K = f (m1) (m2)Ti. i. shall use the following notation for any curve throughout the book unless otherwise stated number of nodes. IV. . n2 283 K = m 2 2r3i. 8 number of inflexions. »If take three. if we prove (i). but are connected by various relations. The equations (v) to (ix) do not give any fresh equations by consideration of the reciprocal curve. 8. 2(t — 8) = (m — n) (m + n — 9). = 3«(%2)6<S8/c. (iv) k (v) (vi) (vii) (viii) in(n + 3)82ic= i Jm (m + 3)T2t. i The six quantities n. Hence. n. They are not independent. Of these so on. we get (ix) by subtracting (ii) from (i). (i). we Pliicker's equations only three are independent. The Pliicker's numbers of the reciprocal curve are by Ch. Similarly. § 7 m. (ii) follows at once on consideration of^the reciprocal of the given curve. subtracting (v) from (ix). (iv) follows from (iii). (v).
These intersections lie one at each point of contact of a tangent from 0. proved (i) in Ch. All curves with the same Pliicker's numbers may be said to belong to the same type. a quartic cannot have four double points. These intersections lie one at each inflexion. preferred. by considering the intersections of the first polar curve of any point with the given curve. But we have given an independent proof of (v) in Ch. six at 6ach node. if D and D is zero. X. which the curve. § 8 and we may deduce (iii) from (i). We shall denote %(nl)(n2)8K by the symbol D. § 7.. We proved (iii) in Ch. is called the deficiency (or genus) of cannot in Ch. which we have proved. . For instance. b) is a double point of a cubic. For the line joining two double points meets the curve twice at each. § 3. Again. so expressed. We may deduce (v) from (i). For the conic through four double points and any other point of a curve meets it in nine points at least. and three at each cusp of the given curve. y—b . (ii). Then (ii) follows from the reciprocal curve. and eight at each cusp of the given curve. whereas a straight line meets a cubic in only three points. D = 0. (ii). if We . VIII 2 It DEFICIENCY 113 remains now to establish three of Pliicker's equations. § 5. VII. the line it = t{x—a) meets the cubic again in a point whose coordinates are rational functions of t. by considering the intersections of the Hessian with the given curve. Also if (a. if the coordinates can be We shall prove that. § 2. a cubic cannot have two double points unless degenerates. Also the pencil of conies through three double points of a quartic and any other given point of the curve meets the . two at each node. be rfegative point on the a parameter . Deficiency. (v). IV. that the coordinates of any curve can be expressed rationally in terms of while conversely. It is readily seen that a curve has the same Pliicker's numbers as any projection of the curve but that curves with the same Pliicker's numbers are not necessarily projections of each other. (iii). VII.
&c. = 0. « = x>t and = 4(Z>1). a quintic with two nodes and four cusps. Ex. = £(re + l)(re3)(re 12). 5. D = 0. Enumerate the types of cubic and and Ch. m=8. XXI. If 8 t. and 8 = t. If 2«*=i. Ex. The exceptions are given by 8 = t. The number of cusps cannot exceed the smaller of and l(n2 + 2D) \ {4re3(4« + l)i + 4Z>j. while m > \(n + 2) if n is even. Prove the rest of equations Prove that r (i) to (ix) in § 1. If and [If m > \ + 3) if is odd. i). § 7. For example. and > 0. the quintic with five cusps. whose coordinates are rational functions of the parameter of the pencil. !! Ex. n+m Ex. or a sextic with nine cusps. For a more general definition of deficiency when the curve has multiple points other than ordinary nodes and cusps see Ch. m(m — l) > «. (ii) re = 7. XVII. 1. 6Z)= (rel)(re6). = 9. Ex. then of the same type as its reciprocal.114 DEFICIENCY VIII 2 quartic again in an eighth point. =. L then k 1. =.] < 9. K = n2 + 2D. the number of such types is the integral part of f (re 2 — 5 n + 12). [See Ch. <t 2 as n + m>. For a more general theorem of which this is a special case see Ch.7. the cubic with a cusp.] (i) quartic. the quintic with three nodes and three cusps. m=2(nl)K = l(n+2 + (re re D = and n >4. k = t. » Ex. either n = m. m> ${l+(4n + l)*}.] 11. If Ex. 2D = n + 2. Ex. § 1. k or else the curve is a nonsingular cubic. Equation (vi) of § 1 states that the deficiency of a curve and its reciprocal are the same. the quartic with a node and two cusps. i i. . not all the double points are cusps. (viii). 13. or if k = i. XIII. a quartic with two nodes and a cusp. 2. [Use Ex. = 8 + £(n 1. IX. 10.] . m^ 2(rel). = = [Use § f (vii). (ii) Find the type of curve for which n = m = 8 = K = r = i. T 8 >. 4. § 3. 1! 2re2S3/<)(re 9283K). If 8 = n + m. (i) m = 2(nl)+2D. . If 28 = 8Z> = (re2)(«4). we have re = m.10. If re is given. £ = 0.] i Ex. ' ' Ex. If n>m.K 8 (ii) If = 3 (re2) + 6D2k. If (viii). 3. = \(n\)(n2)DK.] Ex. i Ex. 8 = £(w2)(re3)3Z). If The curve is [(i) Use § 1 (vii). § n = m. 12.8. Ex.6. 9. [For 9.
if the branches are all made to approach 0. The result is plausible for a curve which has k veal branches all nearly passing through has evidently ^k(k — l) crunodes in the neighbourhood of and these coalesce at 0. shall show that We Plucker's equations still multiple point with ^k(k — 1) nodes and each ordinary Jc'ple tangent with k' distinct points of contact as equivalent to J k' (/<. 15. Multiple Points with Distinct Tangents.] . the tangents to the curve and the first polar at being all distinct.: VIII 3 MULTIPLE POINTS R 115 Ex. The first 0) is ^ = 0. equations (i). + zn . and from these three equations the other equations of § 1 can be deduced.]c . we replace the fixed line by [Project the conic into a circle. ur is homogeneous of degree r polar curve of (0. 1. The result of Ex. If the range (OPQE) has a fixed crossratio. which has evi dently a (k— l)ple point at 0. . Let us now suppose that the curve has one or more multiple points with distinct tangents. § 2. (ii). Suppose the curve has a /cple point Its at (0. or Ch. Hence by Ch. if ive reckon each ordinary distinct tangents as equivalent to /= zn . the curve and first polar meet in k(k — 1) points coinciding with 0. 14. as required.+u n in x = 0. equation is . Through a fixed point any line is drawn meeting a fixed curve in P and a fixed line in Q. § 7.. [Project the fixed line to infinity plane perspective. VI. We require to show that any first polar curve and the Hessian meet the given curve k(k— 1) times and Bk(k — 1) times respectively at a /cple point of the curve.) The ..'— 1) biiangents in the evaluation of S. to possess only ordinary So far we have supposed the curve cusps and nodes. below. 0. the locus of is a curve of the same type as the fixed curve. One of the statements in this theorem is the reciprocal of the other. and invert with respect to 0. 14 holds a fixed conic through 0. To find the number of intersections of curve and Hessian at we adopt an indirect method. (See also Ex. 2.1 u rc _1 +. It will be sufficient to prove one of them. if that is so. and y. or note that the two curves arc in if Ex. (iii) of § 1 hold good when each fcple point is considered equivalent to %k(k— 1) nodes. hold.] § 3. For.k u where l . I. and D. 1). r.
§ 2. k cusps. . 31c and therefore the curve and Hessian meet at 0. .: . [If 2. (0. times at = and (iii). since k = 0. and i inflexional tangents. 1). 3 116 MULTIPLE POINTS VIII is evidently dependent solely on the nature of the and there Is therefore no curve in the neighbourhood of . We show now more generally that equations hold if (i) each multiple point of order k having I ordinary superlinear branches with distinct tangents is counted as equivalent to %k(k— 3) + 1 nodes and k — l cusps . PliicJcer's * It is supposed that the tangent touches only one branch of the curve at each of these I' points. Ex. The deficiency is taken as %(n — 1) (n — 2) — ^2fc(&— 1). Ex. VII. The deficiency of a curve with no singularity other than ordinary multiple points with distinct tangents is £ (m — 2n + 2). = 3n(n2)68a 8icSk(kl). 0. [£(m2» + 2) = £(wl)(»2)8. VI. I. 1). u = are the tangents to the curve at the multiple point the tangents to the Hessian at the point are « by Ch. (k— 1) times with The first polar of has evidently a /cple point at the same tangents as the given curve. The\i from the reciprocal curve as in § 1 n= 1) — 2r3t.] Establish directly the fact that the Hessian meets the curve 3Jc(k — 1) times at a fcple point. and (ii) each multiple tangent meeting the curve in*k' + V points at its V points of contact* is counted as equivalent to %k'(k'—3) + l' bitangents and k'—l' inflexions.] § 4. 0. 1. cla: 2 • —= dy" ' I \ ixiyj § 7. (Ch. namely 82 nodes. Multiple Points with Superlinear Branches. Ex. number m(m— 3m(m and we have just proved that m = n(nl)28 These give i 2 3ick(kl). loss of generality in supposing that the curve has besides only ordinary point or line singularities. 14 fc(fc+l) (0. . § 7 first polar meet k(k + 1) times at (vi)) so that 2/c of the tangents from a fcple point must be considered as coinciding with tangents at that point. A Hence curve and Hessian meet + 2fc(fc2) = 3fc(Jfcl) similar method will apply to § 4. Hence the curve and Ch. r bitangents. k = 2) — 6r — Si.
We deduce since c 2 = 3n{n2)682 $K result. 2 — I) . = 7»12. (ii) is the reciprocal of (i).] [(i)m (ii) <e t . t = £(»j3) (9»16).. < = nS. Find Plucker's numbers for a curve of degree superlinear branch of order (i) n — 1. As n= while m (m — 1) — 2r kz 2 —3 1 2 and m = n(n — 1)— 2S — 3k — (k 2 2 2 2 I). 8 == £(ra3)(»4). r 2 bitangents. 8) the curve and the first polar meet in k 2 l = 2{%k{k3) + l)+3(kl) points at 0. § 1). m = S(nl). D = n2. Ex. of the theorem is similar to that of § 3. 4 the Hessian of the reciprocal curve meets it k — L times at its points of contact with the multiple tangent which is the reciprocal of 0. where there is that a may be y = (Ch. b. Hence (Ch. VI. § 2. d.4 VIII ' MULTIPLE POINTS ' 117 By an ordinary superlinear branch of order q we mean one whose Cartesian equation near the origin (taken at the multiple point) is of the form y no = 12 3 x (a + bxl + cxi + dxl + between . S = r = \(nV)(n%). § 7. n with a single i = ». c. VI. I) = 0. = = »2.. VII. Ex. only note the modifications which are necessary. Since a factor repeated r times in u k is repeated r— 1 times in r^ . and no singularities t 2 inflexions.). As before. which establishes the Ex. <? 2 that the curve has besides nodes. = 3m (m — 2) — 6r — 8t — (k — by Ch.. special relation a. if the tangent to the branch is except taken as . zero. Again.. to prove that the number of intersections of curve and Hessian at is 6{±k(k3) + l} + 8(kl). . k 2 cusps. (ii) n— 2. we may suppose except before. 2 6{%k(k3) + l}8(kl). every tangent to a superlinear branch of the curve is The proof We a tangent to a superlinear branch of any first polar of order lower by unity.
6^ + 8/q + = t! /3 .f Since the curve and the first polar of meet altogether in n(n—l) points. could make the solution definite if we had a definition of the deficiency D. e. and how inflexional tangents have their points of contact at ? may consider these as ordinary nodes. (i). to be noted that and (ii) give identically 3(a + p) = . if for the reciprocal curve the line corresponding to passes through p intersections of curve and first polar and a. . t for a. bitangents. . and inflexions not coinciding with O. many We . — m(m — 1) — n — 2r — 3i = 3m(m — 2)— 6r — 8t — equations give 3(a + p) = jS + p 2 2 . P m a = n(n— 1) — m — 28 — Sk 2 2 . The general problem. o. and a times at 0. without loss of generality. The last four The equations dependent. 2r 1 + 3i 1 First of all it is = o. quantities 8lt k x t 1 tj could not be found if this relation were not satisfied. are an infinity of solutions. Similarly.118 HIGHER SINGULARITIES §5. in order that Pliicker's equations may formally hold ? Suppose the numbers required are 81} Klt tv t r Let any first polar curve and the Hessian meet the given curve in a respectively. . twice at each of the 82 nodes. 0. cusps. Hence the . &c. (ii). and (ii) are then consistent but not inOur problem is soluble but not definite there (i) . . p. . Then we must and /S points coinciding with : have 28l + 3K 1 = a. thrice at each of the k 2 cusps. j 6^ + 81! + *! = (i) a . £ = 3n(n2)682 8ic2 i 2 Likewise from the reciprocal curve a . take 82 k 2 r 2 i 2 aa the number of nodes. . and inflexions * must be considered as coinciding with 0. as required. 2 2 ac 2 . Similarly from the Hessian. which was We How many bitangents have both points of contact at 0. VIII 5 Higher Singularities. special cases of which have been discussed in §§ 3 and 4. a only depend on the nature of the curve in the immediate neighbourhood of 0. . bitangents. cusps. To show that it is satisfied.intersections of curve and Hessian. /3 + (r. . * i. namely at the points of contact of the tangents from P. is as follows Given a curve with any multiple point whatever 0. how many nodes.
(2l)(P + 2). since the curve unicursal (Ch. H=qF. /3 p l2fc 2 + 3k. bv noticing that. The reciprocal curve is a 5ic with the reciprocal singularity. ^ = ^=0.] . «ij»l. Such a definition will be given in Ch. ) = D. XVII. =iq(p + q)%p2q + %F. (a + p) = /3 + <r. 1) and (0. 0. We have not proved that Slt K lt tx i 1 thus obtained are positive integers or zero. (V) the curve has a single point singularity at inflexions. 1 Hence Ex. p = 0. T1 ^=PF. Ex. (i) and (ii) l. at (0. »i+«i + T l + « l = i(j» + sl)(p + «2). = lp(p+q)2piq+%F. These give Hence a <r = 21. since the curve is unicursal. 1) [By Ch. IV. F= 1 and «i=*Cpl)Cp + ffS). § 0t = P= '3 These satisfy *i we find (pl)(p + q). p = 3p(p + q)4p2q. r^iteljfci + aa). VIII 5 HIGHER SINGULARITIES 119 applicable to singularities other than ordinary nodes and cusps. In this case the solution of the fact the line corresponding to 0. Discuss the singularity of (yz + x*) 1 § 8 = 21. X. 3'. and three other cusps. were satisfied by S1 =^k(k3) + = kl. 2 (i) Ex. and § 8 (i) and (ii) give Ex. the reciprocal singularities at (0. 7. § 7. cr = 3q(p + q)2p4q. = «. and three other = = = = = We may «! + «! = prove k = 0.e. 1. In § 4 we proved that for the case there considered a and verified that =k ~ 2 I. 0. 4 and Ch.. for then we could add to equations (i) and (ii) i(nl)(n2)(81 + 82 + Kl + K 2 . was made definite by our knowledge that the reciprocal curve had no multiple point on Discuss the singularity of 7. (0. Ex. We may determine F by noticing that. 0) must be together equivalent to %(p + q — 1) (p + q — 2) nodes and cusps i. 81 = r1 «. 2.] = tfx at (0. 1). is where Fis any symmetric function of p and q. 12). § 4. 0. Kl (T — h— l. IX. no other inflexions. VII. 1). 0. [Using Ch. y*sfl = xv+i § 7. 1.
Let any variable transversal through G cut 2 in Q. we may take homoChoosing xy. as is easily The dotted conic in Fig. Z). P' will trace a locus c'. Y. ' ABC 2. P' lies . as triangle of reference. It and let (PPf QM) be an harmonic range on the line CQR. geneous coordinates so that 2 is z P ' . = If P' is the point (X. ^ . 1). In Fig. P is ( y . if P lies in any portion of the plane. Suppose we have a fixed conic 2 with fixed tangents GA. for these . 2 is the locus of the middle point of the chord QR. traces out a locus c. ~\ . in the other portion labelled with the verified. GB (Fig.CHAPTER IX QUADRATIC TRANSFORMATION §1. 2. same letter. c' are said to be derived from one another by 2 being the baseconic and C quadratic transformation the pole of the transformation. Definition of Quadratic Transformation. Suppose that The loci c.
z) = is transformed into f(l/x. c' is any other line. y. c' is * It is more usual for reasons of symmetry to take the transformation in the form Y : Z = 1/x 1/y 1/s so that the curve f(x. . then is the locus c' of P' * \y x z/ .: IX z1 1 QUADRATIC TRANSFORMATION C and is 121 points are collinear with = xy. P' is at B. or from the original definition. C. 1/y. a conic through A. if conjugate with respect to r Hence. y. a line through A. P' is at C. z) = the locus c of P. which is a conic If c is a line through B. = 0. c' is c' is IfPison^C. B. 1/a) = 0. P is on BG. we at once derive the following P is on AB. G. f(x . the same line. if c is the line \x + fiy + vz X/y + n/x + u/z = 0. This is equivalent to quadratic transformation followed by interchange of the vertices A and B of the triangle of X : : : reference. If c is a line through C. B. through A. From If If this result. P'isat A. c' is For instance. If c is a line through B. If c is a line through A.
p q. If c meets Pig.. the conic becomes the osculating conic to c' at G. 3). It follows that. c' touches at A the line through A corresponding to BP. and similarly if c touches GA. If c touches AB. and . — n— — n—p — n—p n—p — c * Provided P is not at A.. B. If i? approaches P. CB respectively at k. G'j4. i. two tangents to c' at G coincide . touches CP at C* at P to c corresponds the conic c' osculating c' G and passing through A and B. For if Q is close to on c. two tangents to c' at coincide. P P R Again CA at P. Then c meets J. the line joining A to the second Similarly if c meets CB at Pintersection of BP with 2. Also. Suppose c is an nia with a ftple point at G. . R' touching c' at 0. . to GA. — q. and a ^ple point at J5. a pple point at J. B. and the limiting position of CQQ' (which is CP) is the tangent to c' at G.: : 122 QUADRATIC TRANSFORMATION IX 1 Again If c meets To the tangent at AB at P. For. Q' is on c' close to C. Hence c' has k. Hence (or branch of to every intersection of c with corresponds a c' through G. to the line joining of c corresponds to any point a conic through A. P is an ordinary fcple point of c. to every intersection of c with CA c' AB CB) corresponds a branch of through A A (or B) . P' is close to A (Fig. if P is close BP meet on 2. or C and so throughout. if has k linear branches touching CP at C. 3. AP'.B. and vice versa. and conversely. n — q k other points. e.
A and B a conic through Ex. [Take three double points as A. P' are now inverses of each other with respect to the circle. inversion. IX 2 INVERSION 123 n—q—k branches respectively through C. k = is 0. B. '. wellknown theorems connected with One of the main uses of quadratic transformation is to deduce properties of a curve c from those of its transform c'. If c is an — — — For instance. then c' is a (2n—p — q — k)ic with an (n p — q)ple point at G. and an (n q k)ple point at B. Our knowledge of the theory of quintic curves (and still more of sextic) is very limited. If two curves (or branches of the same curve) have rpoint contact at a point on CA. Since c' meets at { (n —p — k) + (n — q — k) + k } points. a circle with centre From ' the results of the last section '. c' of P.B. Also it meets of AB with the tangents to c at G and similarly it meets GA and GB at p and q points respectively other than A. The properties so derived are not usually of much elegance or importance. 1. C. it is a (2n—p — q — k)ic. 1 are to be projected into the circular points. g. taking n = 2. and a qple point at B. C. B in Fig. the transformed curves (or branches) have (r+ l)point contact at A. the inverse of a circle is a circle the inverse of a curve with respect to a focus is a curve with a cusp at each circular point and so on. B. Hence AB at the k intersections AB : nic with a kple point at G.A.. p = q = 1. The loci c. be both nodes. If in § 1 we project A. we see that A the transform of a conic through and B. Hence inversion is a particular case of quadratic transformation. ' we may derive many For instance. and quadratic transformation is the generalization by projection of inversion. an (npk)ple point at A. if A.] Ex. 2. a pple point at A. P' are inverse points with respect to this circle. 2 becomes and P. . § 2. it is convenient that they should bear similar relations to the curve c (e. The main advantage of quadratic transformation over projection followed by inversion is that. Properties of any quintic with three or more double points can be deduced from those of a quartic with the same deficiency. B into the circular points. This is exactly equivalent to the process of projecting two points into the circular points and inverting with respect to some other point. C of § 1. Inversion.
P nN ABC nN—r — — N—K N 2N—K. Take the point as C. if we are to observe the convention that an nic and an iV'ic meet at points. The Number of Intersections of two Curves a given Point. VI. and therefore times at G. 1 does not lie on a side of the triangle the quadratic transformation evidently transforms two curves meeting at into two curves meeting at P'. Quadratic transformation (or the equivalent process of inversion) will also enable us to simplify the solution of problems which have been discussed in Ch. At A. INVERSION IX 2 or both cusps. The transformed curves therefore meet (n k) (N—K) times at each of and B. &c. times at C.fifple point on the two curves respectively. and G no two branches of the transformed curves touch one another. 1) so as to have no other special relation to either curve. and have multiple points of orders n and at G (§ 1). Suppose that the curves meet in r points other than G.: 124 . Therefore nN—r = s + kK. B. &c. Suppose then we have two curves of degrees n and N. so that If two curves have a kple and a Kple paint at G respecand the transformed curves meet at s points on the line AB other than A and B. the two given curves meet s + kK tively. and times at G (Ch. quadratic transformation. namely. at in Fig. A A B — A nN . s AB A = (2nk)(2NK)2{nk)(NK)nNr = nN—kK—r. VI and elsewhere. § 2). I. which is a /cple and a . such as the determination of the number of intersections of two curves at a given point.) and this restriction is unnecessary if we employ the generalized form of inversion. They meet also at the transforms of the r intersections other than G of the original curves. and take the sides of the triangle (Fig. since we took the points and in a general position. the expansion of y in terms of x near a given point. §3. The transformed curves are of degrees 2n k and have multiple points of orders n k and at each of and B. and we want to know how many intersections of these curves must be considered as lying at a certain point. § 7 Ch. . Hence they must meet at s points on other than and B. the number of tangents from a point whose points of contact coincide with the point. where If the point P ABC.
and an %ple point at. If the transformed curves intersect at a point on AB. We may find the equation of the polar An alternative method is the curve and then use § 3. though this restriction may at times be removed if due precautions are taken. It meets the first polar of B (nk)(n — kl) times at A. following. meets the Tiic a times at G. and (say) a. take the case of a curve passing through the cusp G of another. all the branches having as a common tangent. so . and at the polar of points of contact of the tangents from P. the two curves having a common tangent at G. that the lines GA and GB have no special relation to the curve.IX 4 GLASS OF A CURVE 125 We suppose. one touching AB and the other not. The transformed curves have /cple and ITple points at and therefore meet in IcK points coinciding with H. points on m m AB * a this assumption involves depends only on the shape of the nic in the neighbourhood of no loss of generality. It has ordinary (n — /c)ple points at A and B. H Hence the number of intersections is of the original curves at G 1+2. take the case of curves with k and linear branches at C respectively. (nk)(nk+l) times at B. C. Class of a Curve. K GH H . taking G at H. points of contact n(nl) times at G (Oh. The ti'ansformed curves pass through H. VIII. tangents from A to the nic corresponds To each of the tangent from B to the transformed curve whose point of a contact does not lie on AB.1 = 3. As another example. and it is not immediately evident how many times they meet at H. which meets A B at H. We shall assume the nic has no multiple poiDt other than C* Then Suppose first A The P m = n(n— 1) — a. as stated before. §4. G. § 3). we may apply the above theorem to the transformed curves. As a simple example of the above theorem. m and B have no special relation to the curve. so that they have a simple intersection at H. at the of tangents from B not lying on AB. Suppose we wish to find the number of intersections a of a curve with any first polar curve at a multiple point G of order k on an nic.' times at other than A and B. The original curves therefore meet at 2kK points coinciding with C. Now the transformed curve is of degree 2n — k.
or (ti — fc) (ti + Zc— 1) — a' = m = ii(?i— ])— a. Tangents from a Singular Point to a Curve. The other tangents from C to the transformed curve are the n tangents at G each counted twice. Now H H H Hence The number of intersections of a curve with any first polar curve at a kple point G is equal to the number of intersections k{k—\) at an ordinary kple point together with the number of those intersections of the transformed curve with any first polar which lie on but not at A or B.: 126 Therefore CLASS OF A CURVE IX 4 (2nk)(2nkl) = (nk) (nrkl) + (nk) (n4 + 1) + w(to — l) + m + <x'. Hence the class of the transformed curve is m m — \ + p+ 2n. AB the curve has k linear branches with a tangent at C meeting at H. Suppose that in the curve of § 4 X of the tangents from C have their point of contact at G. The effect of this on the class of the transformed curve is to lower it by k(k — \) and therefore the effect of the singularity at G is to lower the class of the original curve by 2k (k — 1). of a curve with the the number of intersections at first polar of is dependent on the shape of the curve in the neighbourhood of in general and not on the position of (unless is near or on the tangent at H). as in § 4. This gives a = <x' + k(k — 1). the transformed curve has a /cple point at H. common AB . and the TX tangents from C to the given nic which do not touch at C. is by consideration of the tangents from B that this class m+2(n — k) + e. where e is the number of tangents from B to the transformed curve whose points of contact are on but not at B (or A). AB Therefore X = 2k+u — e. But we see. or: The number of tangents from the kple point C to a curve whose points of contact coincide with G is 2 k. §5. if For example. and that u of the tangents from G to the transformed curve have their points of contact on AB. plus the number of tangents from G to the transformed curve whose points of .
djB. If a tangent at a /cple point C of an nic meets the curve in k + r points coinciding with C and meets AB at H. §§ 2 and 3 as i(n— l)(n — 2)— S— k. the Jple point at H' moves up to G and coalesces with the &ple point to form a higher singularity at G. AB but For example. the transformed curve has a /cple point at at which in general neither AB nor CH is a tangent. so that the original curve has an ordinary /cple point at G and a ^'ple point corresponding to at the point H' very As AB is moved hack into its origiDal position close to G.] at C or H. with the proviso that an ordinary /cple point is to count as ^k (k — 1) nodes. Suppose now that the transformed curve and the lines GA. but an attempt is here made. Suppose that the transform of a curve with a /cple point at a Jple point at on . Ex. ' We defined the deficiency D of an nic in Ch. from H Hence the points with C. § 6. The transformed curve now cuts AB in k shape slightly.) First we word the definition of deficiency when no higher We say that the singularities exist a little differently. We say that G has been analysed by quadratic transforma' tion '. [The transformed curve meets CH n times at C and nkr times not But it is of degree 2 n . of contact of C coincide 2k tangents Ex.IX the 7 DEFICIENCY on AB. the transformed curve meets CH in r points coinciding with H. distinct points not close to A or B (j of them very close to H). ' Deficiency. § 7. VIII. (See also § 12. 10. Latent Singularities. minus the. CB are kept fixed while the line AB is slightly displaced so as not Then the original curve will alter its to pass through H. The Jple G has H H ' ' point of the transformed curve is said to be latent in the &ple point G of the original curve. not coinciding with A or 5.k. 127 contact are number of tangents from B to transformed curve whose points of contact are on not at B. It is desirable to give a definition which shall be valid when the curve has higher singularities. It is difficult to do this satisfactorily without a knowledge of functiontheory. where S was the number of nodes and k the number of cusps. . if the curve has k linear branches at C with a common tangent meeting AB at H.
By § 1 the transforms of" the nic and (to— 3)ic are therefore a (2n—p — q — k)ic with (to — p — /fc)ple. That this definition of deficiency is equivalent to the earlier ' ' one is readily proved. We — ' § 8. and so for J. Now suppose k = 0. B. have assumed p. § 6). (to— p — g)ple points at A. q.C respectively. . as required for the deficiencies of the TOic We * or 1 To complete the definition add and B = 0. multiple points.has a (k of an nic. (k — l)ple points at A. The transform of the (to — 3)ic is now only of degree In— p — q— 4 (not 2to— p — q — 3.128 deficiency of coefficients DEFICIENCY an nic is one I X more than the number of arbitrary of the in the equation most general adjoined (n — 3)ic. If the A ic is • adjoined' to the nic. G and a curve of degree tic We now 2(n3)(pl)(ql)(kl) = with a multiple point of order (to 2npqk3 % — 3) — (p — 1) — (q — 1) = n—p — q — 1 at G. Deficiency unaltered by Quadratic Transformation. the summation being taken over every multiple point of the nic. and B. l)ple point shall state the fact that an iVic. But this is the number we have defined in Ch. from which the equality of the and its transform follows immediately. (to — q — /c)ple. (q — l)ple. B. show that the deficiency of a curve and its quadratransform are the same. Any adjoined gple. VIII. § 3 as Dl.* By an adjoined J\Tic we mean one which has a (k — l)ple point at every /cple point of the nic. when the curve has only ordinary Suppose as in § 1 that an nic has pTpie. II.B. by saying that the JVic is at a given /cple point r adjoined to the nic at 0'. l)ple point at a given state that a curve has a (k point is equivalent to assigning •§&(/« 1) linear relations To — — between the the coefficients of its equation (Ch. /cple points at A. G. k all greater than zero. it is adjoined at every multiple point. (n — 3)ic has (p — l)ple. Thus the transforms of the curve and an adjoined curve of degree lower by three are also a curve and an adjoined curve of degree lower by three. according as the cubic has or has not a double point ' . if » = 1 or 2 while when n = 3. Hence (n— 3)ic has J(TO3)»2ifc(ftl) arbitrary coefficients. D= '.
IX
9
DEFICIENCY FOR HIGHER SINGULARITIES
;
129
and has a multiple point of order n — p — 2 (not n — p — 1) and so for B. But if, as is lawful, we take as the transform of the (n — 3)ic the (2n — p — q — 4)ic together with
proof) at
A
the line AB, we obtain a curve of degree adjoined to the transform of the 7iic. In fact, since the transform of the nic is of degree 2n— p — q with (n—p)p\e and in — ?)ple points at A and B, an adjoined (2n— p — q — 3)ic has (n— l)ple and (w — j — l)ple points at and 5. The line therefore meets the adjoined curve in points whose number is greater than the degree of the curve, showing that the curve degenerates into and a (2n — p — q — 4)ic. Similarly, if p 0. CA is part of the transformed adjoined curve ; and, if q 0, CB is part of the curve. In practice, however, the case p, q, Jc all greater than zero is the only one we need consider. In the following we shall require to transform any curve by successive quadratic transformations into one having only ordinary multiple points with distinct tangents. To do this we can take and as ordinary points on the curve and G as any multiple point which is not ordinary '. The process of transformation is repeated till only oidinary multiple points are left.*
2n—p—q—3
A
p— AB
AB
= =
A
B
'
§ 9.
Deficiency for Higher Singularities.
The definition of deficiency can be applied to curves with higher singularities when we have defined what we mean by a curve adjoined to an nic with such singularities. Suppose that in § 6 a curve has a (j — l)ple point at H, then its transform would be considered as adjoined to the given curve at the /cple point C. If the ./pie point is not an ordinary one, it could be still further analysed by taking C at In this way we can find what is intended by a in §6. curve adjoined to the transformed curve at H, and then the transform of this adjoined curve will be adjoined at G to the given curve. __ According to this definition we shall still have deficiency unaltered by quadratic transformation. For instance, if the given curve has k linear branches touching CH at C, the transformed curve has an ordinary &ple point at H. A curve adjoined to the transformed curve has Hence a curve adjoined to therefore a (k — l)ple point at H.
H
H
* It is fairly
evident that this
is
possible.
We
do not give here the
formal proof.
:
130
DEFICIENCY FOR HIGHER SINGULARITIES IX
9
at C. the given curve has k — 1 linear branches touching Also the fcple point C lowers the deficiency by k(lc — 1), since lowers the deficiency of the transformedthe /cple point at curve by %k(k — l).
GH
H
§10.
Intersections of a Curve with Adjoined Curves.
To state that an i^ic is adjoined to an nic at an ordinary kple point G is equivalent to assigning \k (k— 1) linear relations between the coefficients of the equation of the i^ic. The number of intersections of the nic and iVic at is k(k— 1), which is twice the number of linear relations just referred to. suppose, as in § 6, that C has a latent ^'ple point. The
Now
transform of the i^ic has an equation whose coefficients are subjected to %j (j — 1) linear relations, since it has a (j — l)ple Hence the coefficients of the equation of the JVic point at H. are subjected to \j (j — 1) linear relations in addition to the £k(k— 1) relations which state that C is a (k — l)ple point of the iVic. Also, since the transforms of nic and JVic meet j (j — 1) times at H, the nic and Nic meet k (k— 1) +j (j — 1) times at C by § 8, which is again twice the number of relations between the coefficients of the i\Tic due to the fact that it is adjoined to the nic at G. is not an ordinary /pie point, it may be analysed in If its turn till the nic is finally resolved into a curve with ordinary multiple points only.* Hence
H
The number of intersections of an nic and adjoined Nic
at the multiple points of the nic is twice the number of relations to which the coefficients of the equation of the Nic are subjected owing to the fact that the Nic is adjoined to the nic.
The number of
for
relations in question is
i(nl)(n2)D;
an adjoined (n— 3)ic has
deduce that
is
D—1
arbitrary coefficients in its
equation.
We
If a pencil of Nics
adjoined to an nic
and
the base
points of the pencil other than the multiple points of the nic are also on the nic, any Nic of the pencil meets the nic in %(n—N)(N—n + 3) + D variable points (n > N).
* This statement assumes that a curve and an adjoined curve are transformed into a curve and an adjoined curve i. e. N = n — 3. But the number of intersections and of relations between the coefficients due to the adjoining at C is evidently independent of the value of N.
;
IX
11
ANOTHER TRANSFORMATION
—
131
Of the fixed intersections of the nic with a variable A~ic of the pencil (to — 1) (to 2) — 2D lie at the multiple points and * JA(iV+3)li(«l) (to2)
+D
at the remaining basepoints of the pencil.
There remain
nN{(n\) (to— 2)2jD} {fy(jy+8)l£(»l)(»2)+.D} = i(nW)(]rn + 3) + D
'
variable intersections. If to — 3, the number
N=
is
D.
If JV
=
to
—1
or
to
— 2,
the
number
is
D + l.
Ex. 1. A pencil of JVics has as its basepoints points of an nic, being adjoined to it at all its singularities with the exception of k cusps and certain of its ordinary multiple points. If any JVic of the pencil meets the nic at p points other than basepoints, while q Nics of the pencil touch the nic at a point other than the basepoints, show that
q
= 2p + 2(Dl)K'.
;
[The nic may be transformed by successive quadratic transformations into a curve with ordinary multiple points and p, q, k, D are unaltered by this process. Now use Ch. VII, § 10, Ex. 12.]
Ex. 2. A pencil of iVics has as adjoined to it. Show that
its
basepoints points of an »ic and
is
(nN)(Nn + 3)+4:D2
of the JVics touch the nic at points other than a basepoint.
§11.
Another Transformation.
form of transformation alternative to that in § 1 is obtained by taking as our baseconic a pair of lines through B harmonically conjugate to BA and BC. As before, we take and any transversal through G cutting these lines in Q and and P' on the transversal so that (PP\ QR) is take points
A
R
P
harmonic In Fig.
(Fig. 4).
lies in any pprtion of the plane, P' lies in the other portion labelled with the same letter, the hyperbola in the figure being the locus of the middle point of the
5, if
P
chord QR. ' Choosing homogeneous coordinates so that
s 2 we triangle of reference and the two fixed lines are a? find that, if P' is (X, Y, Z)„P is (X, Y, X*/Z), since CPP' is a straight line and the pencil B(PP', QR) is harmonic. is f(x, y, z) 0, the locus of P' is Hence if the locus of
,
ABC is =
the
P
=
f(x,
*
y,x 2/z)
=
0.
have
N must be such that this quantity is positive or zero. We may always N = n— 1 orm — 2, when «>2. For N to be nS, we must have D>1.
K2
132
ANOTHER TRANSFORMATION
IX
11
It is at once
by taking
§ 1,
'
shown that this transformation may be obtained in succession three transformations of the type of the baseconics being respectively
x2 — xy + z 2
= 0,
x2 — y 2 + z2
=
0,
x 2 + xy — s 2
=
and the pole of the transformations being (0, 0, 1). The properties of the transformation are somewhat similar
to those of the transformation of § 1. The reader will have no difficulty in verifying the following
statements, in which c
tively
:
and
c'
are the loci of
P
and P' respec
If e is If c is
a line through C, so is c'. a line through B, so is c'.
Pig 4.
*
If c
is
any other
line, c' is
a conic through
G
touching
AB
at B. If c meets
AB at P, c' touches CP at G. P to c corresponds the conic osculating c' AB at B.
at
To the tangent at G and touching &
To each intersection of c with BC corresponds a linear branch of c' touching AB at B. To each linear branch of c through B (not touching AB) corresponds a linear branch of c' through B, the tangents to the two branches being harmonic conjugates with respect to the two given lines. If c is an mic with a &ple point at G and a ^ple point at B, c' is a (2n — kq)ic with an (n — g)ple point at G and
IX
12
APPLICATIONS OF TRANSFORMATION
linear branches through B,
133
nk
n,kq
of which touch
AB
^The theorems in
tangents from a singular point of intersections of two curves, transformation of this.sect on the
to italics in §§ 3, 4, 5 relating
the
number
and class still hold for modifications The proofs are very similar, and the necessary apphcatmn^o w_ill require the may le left to the /eader. He which linear WcheB theorems to the case of a point at
the
touch.
each of these sections This has been given at the end of
of illustration.
by way
Fig. 5.
ThP transformation of « For !, nf a curve
the dethis section does not alter
SH
(§8).
•
the type of §
1,
equivalent to three transformadeficiency none of which alters the
it is
iturn of the
Applications of this Transformation. method of § 11 i. that the posiThe main advantage of the *or instance, disposal. „t +>,p^lin line GA is still at our
§ 12.
j»
1.
PSci Is a tangent
wS
not the
,
A
'
case In §
Much simplification in arithmetic may
other than
We may h Cthro without loss of generality which
A
or
be thus secured.
o, B, at ,' fo, .** not at A Je Ke rfTfd^LTn ot e
«.
rt „f 5
S read d
B
'
for
'
other
tlia
11
B
'.
In
134
If
APPLICATIONS OF TRANSFORMATION IX GA
is
12
a tangent at C, the transformed curve passes It is convenient to take GBA instead of ABC as This is triangle of reference for the transformed curve. equivalent to interchanging x and z in the equation of this transformed curve, which now becomes / (z, y, z 2/x) = 0, the 0. original curve being / (x, y, z) Writing it in the form f(x/z, xy/z 2 1) = 0, we see that, if the Cartesian equation of the original curve is f(x, y) = 0, the transformed curve is f(x, xy) = 0. Points oif{x, y) — near the origin correspond to points of (x, xy) = near the axis of y, and points of one curve near / the axis of x correspond to points of the other curve also near the axis of x. If the nature of any singularity of the transformed as origin curve is not immediately obvious, we may take
through A.
=
,
H
H
and repeat the transformation.
Ex. 1. If CA is the tangent at the cusp C of a curve, AB is an ordinary tangent to the transformed curve at A. More generally, if CA is a tangent at C to a superlinear branch of order r, AB is a tangent of rpoint contact to a linear branch at A.
[y
icy
r
=axr+ + bx 'y + cx''' y*+
'
,
i
...
+kyr+1 +
...
...
becomes,
....]
on
putting
for y
and dividing by xr r y = ax + bxy+
,
+ kxy'' +1 +
Ex. 2. If CA is a. tangent of rpoint contact at C to a curve, a tangent of (»•— l)point contact at A to the transformed curve.
Ex.
3.
CA
is
A
common
tangent.
quadruple point of a curve consists of two cusps with a Find the effect of this point on class, deficiency, &c.
[Taking C as multiple point and CA as tangent, the transformed (Ex. 1). have curve has two linear branches touching at shown that the effect of such a ' tacnode as is to lower the class by 4, and also the two tangents from to the transformed curve touch &^A. Therefore the effect of the quadruple point on the class is to lower it by Again, the tacnode lowers the deficiency by 2, so 18. 4 + 2 + 4.3 A that the quadruple point lowers the deficiency by 2 + 44.3 8. curve adjoined to the transformed curve at has a linear branch touching at A, so that a curve adjoined at C to the given curve has a, cusp at C with CA as tangent. The tacnode is latent in the quadruple point C, which may be considered (Ch. VIII, § 5) as equivalent to 8 nodes and k cusps where 8 + <c 2. 8, 2 8 + 3 k 18, or 8 6, k The tangents from C which touch at C are 2.4 + — 2 6 in number.]
'
B
AB A
A
We
=
=
AB
A
=
=
= =
=
Ex.
4.
Discuss similarly the case of the rhamphoid cusp.
[Taking
=
(y
C as cusp and CA as tangent at C the curve is + a#2) 2 + ya (bx + cy) + y (dx, + exi y+fxy + gya ) + hxb +
i
...
;
and the transformed curve
which has a cusp at A.
is
3
0=(y + axf + hx +
...
IX
12
APPLICATIONS OF TRANSFORMATION
135
Hence C is a double point with a latent cusp. The existence of a rhamphoid cusp lowers the deficiency by 2 and the class by 2 + 3 = 5. An adjoined curve has a linear branch touching CA at C. The rhamphoid cusp may be considered as formed by uniting an ordinary node and cusp. Since the reciprocal of a rhamphoid cusp is also a rhamphoid cusp (Ch. VI, § 5, Ex. 1), it may be considered as formed by the union of a bitangent and inflexion. The number of tangents from C whose
points of contact
lie at
C is
4.]

Ex. 5. Discuss the nature of the origin for y 1 = x 2n+1 [Put yx for y and use induction. There is a latent cusp and n  1 latent nodes. The origin lowers the class by 2.n + 1 and the deficiency by n. An adjoined curve has a linear branch with y = as a tangent of n point contact.]
Ex.
6.
Discuss the origin for the curves
(i)
y^
+ ixVy + x
1
^=
.
0.
(ii)
yv
=
x"ti l
.
(iii)
(yxv)v
= xPf1
[Put yx for y and use induction.]
Ex. 7. Show that the effect on Plflcker's numbers of a double point at which two linear branches have rpomt contact is the same as that of
r nodes and r bitangents in general. [Use induction.] Ex.
8.
Find any latent double points of the following curves
origin; where
u^x+y
u*
it?
at the
1
.
{i)
(ii)
= y*(px* + qy*). = xy
3
.
(iii)
(uOCy*)
{nPy ) (uyy2 )
1
= k(xuy
3
)
2

[(i)
Two
latent nodes.
(ii) A latent node and cusp. As another example find the double points latent in its Hessian at the origin.
(iii)
the cases k
Put in turn yx for y, xy for x, xy for x, &c. — 1, k = — Oifiy, and also
7
Consider separately
=
0,
a =
ft
/3
=y=
0,
To trace the
curve, find its intersections with
a = /3 = y u = ty*.]
=
0.
Ex. 9. JTbe order of a superlinear branch at C (§ 11) is equal to the order of the transformed branch plus the number of tangents from B to this transformed branch which coincide with BA.
If 2 a >/3 > a., [Put yx for x in the expansion (i) of Ch. VI, § 3. apply reversion of series (Ch. VI, § 1) and the theorem at the end of Ch. VI, § 5 to the transformed expansion.]
' '
Ex. 10. If
I is
the order of any superlinear branch of a curve,
2(D1) = m2n + 2(ll),
the summation being taken over
curve.
all
the superlinear branches of the
[Repetitions of the transformation of § 11 will transform the curve eventually into a curve with no superlinear branch, for which the result
136
is
APPLICATIONS OF TRANSFORMATION IX
But Ex. 9 proves
12
readily seen to be true. by such a transformation.
mln +
'S.
(I
— 1)
unaltered
We may
if
define the deficiency as
we
Ex.
choose.
11.
This
is
£{m2w + 2(/l)} + l, an alternative definition to that given in
2,
§ 7.]
Prove Ch. VI, §
2,
Ex. 7 by quadratic transformation.
[Use Ex.
Ex. 12.
§ 3,
and induction.]
Two
1
curves having at
11.]
K(K >
at C. case in
C
k)
with a
common tangent meet
superlinear branches of orders k and k (K+ 1) times at C.
[Use Ex.
Ex. 13.
and
Two curves have How many of their
which two or more
a common tacnode C with a common tangent intersections coincide with C? Discuss the of the branches at C osculate.
Ex. 14. Discuss similarly the case in which both curves.
C
is
a rhamphoid cusp of
) should *(<) <t>(tj ^(0 f(h) f(t2 ) = o . .* * We assume throughout that in general to each point on the curve corresponds a single value of "t... fcfg) (*. and that the parameter of the point considered as lying on the other branch is ty The equation (iii) must be satisfied for all values of t 2 since the points with parameters t and t x find the real nodes. (iii). Making ^ approach £C we have 2/ the equation /(<) 4>{t) <t>'{t) +(t) *'(*) = o .. . to This will give us equations in t and t x which are equivalent two equations to solve for t and £. t. which we will call the 'parameter of the point. See § i. y. . t lt t 2 (ii) /(*) for the tangent at the point with parameter The condition be collinear is that^the points with parameters f(t) /(t. Suppose the coordinates of any point (x.<)(*<!) f(t 2 ) 4>(t 2 ) that the parameter of a point considered as lying on one branch of the curve at a node is t.CHAPTEK X THE PARAMETER §1. To we suppose coincide. by the equations The equation of the t. t. 13. _ Pointcoordinates in terms of a Parameter. line joining the points with parameters £j is evidently y z f(t) t = (i) t t £.. Ex. z) of a curve given as functions of a quantity" t.
' * \r point with parameter T. If the values of a crunode . But . if 2 is real. . /(*) +" r V' = .f(t) For suppose the tangent (ii) meets the curve again in the Then *(<) «/>'(*) *(*) *'(*) 0. for f'Wf(t) = 4>W<l>® = M)/W) lc • • ( iv ) For if the values of the parameter at a double point are iandij. . i. . y.f{t)}/(Tt)\ &c.f(t)}/(Tt)> =f\t) which proves the * It is result. t the node is a real acnode. one approaches t. *'(*)/*(*) become equal to the limiting value of (k— l)/(*i — <)• The inflexions are given by the equation F(t) = \f 1/ * </. .* the node is x on giving a small increment to the value of t or 1 1 we obtain a real point on the curve close to the node. suppose the tangents at the double point approach coincidence. If the values of t and t are conjugate complex quantities. e. (v). i// are real. Now value of if T the point of contact approaches an inflexion. without altering its value.value theorem. L^.138 USE OF A PARAMETER similar (i) . fihVM = Therefore 0(*i)/*(«) = f(Wf(t) = (say). . Now make tx approach t.. 1 t])/f(t). (/>. The cusps are given by the equations determinate. z in is to equate to zero the coefficients of for the line joining two coincident points is in t and t thus obtained are real. f(t) f(T) 0(D +(T) row of the deter We may minant by replace the elements in the third 2{f{T)~f{t){Tt). supposed that/.. denote f(t). by the mean. 2 {f(T)f(t)(Tt). where /./'. Then all (kiy^t) = {/&)/(*)}/&«)/<*) =f(t+o[t where 1 > 6 > 0. f(t)/f{t) and similarly *'(*)/*(*). X 1 A method x.
• 2. If /> 0> are polynomials in t. are also given by The cusps are given by the values of t satisfying * both F(t) = and / * + F'(t)= =0. + ^(t) + „ylr(t) = n If this equation gives values of t corresponding to exactly distinct points on the curve. If a tangent to the a = by equation (ii) of is Xx + fiy + vz = curve in § 1. of § is 1. Then An f(T) 4>(T) *(<) tf{t) +(!) *(*) +'(t) {Ttf f(t) =0 . for 0. the curve is of degree n.X2 The cusps. = ffff. we may ^ § 1. and the inflexions by the single = ^ roots.'" while the parameters of the inflexions satisfy F(t) = 0. fi. but not F'(t) in general. Linecoordinates in terms of a Parameter. For instance. Just as we obtained the parameters of the nodes and cusps and the degree of the curve. (i). v=wf'<i> A. The degree of the curve noting that the line meets the curve where \f(t) (if it is algebraic) is obtained by \x + jiy + vz = 0. alternative method of finding the nodes and bitangents Suppose that the tangent (ii) of § 1 meets the as follows. USE OF A PARAMETER 139 which f'/f = 0'/0 = yjr'/yfr. f{t) Write down the condition that this equation in T has equal . so by using the linecoordinates we may obtain the parafneters of the bitangents and inflexional tangents and the class of the curve. v of any tangent to the curve are expressed in terms of the parameter t. take W0>. the inflexional tangents will be given by A '/A = (J'/fi which correspond to equation (iv) v'/v Hence the linecoordinates = . the cusps are given by the repeated roots of F(t) — 0. F(t) = f & y /'" $'" xj. curve again at a point whose parameter is T.
l.140 roots. we get an equation giving the on the curve from which two parameter T (^ t) of a point coincident tangents can be drawn not coinciding with the P tangent at F. If we write down the condition that (i) considered as an equation in t has equal roots. USE OF A PARAMETER X2 the obtain an equation in t whose solutions are contact of the bitangents and the parameters of the points of points of contact of the tangents from the cusps. . We Fig.
v If the tangents coincide in a bitangent. . As an alternative method notice that the tangent at the point with parameter t meets the curve again at a point with parameter T where 2{tl)T i + 2(tl)(2t + l)Tt[2t+l) = .1 and ^ give the parameters of the points of contact of the bitangent. bitangent. Ex. degree. inflexions. v = \ as the values of u and v required. = t+T. Hence t = .X an 2 It is clear that USE OF A PAEAMETER 141 is either an intersection of the curve with inflexional tangent. t The equation of the ! line joining the points with parameters and T is 6PT (t+T+l)x+{t+T2tT)y The t coefficients ( all + 6a{(l2tT)(fi + tT+T i) + (t+T)(tT)i } = 0. = m(2m + 1)(m 2 2m2) = 0. we see that T = 0. 1 y = 6(t + t)a. for instance. vanish if t+T=l. the denominators of these Eliminating v we get fractions vanish. = \ —1 + The 2 >/%) as the parameters of the node. i. T meet at the point y •»(2w 2 4d2 + 2u + 2uv + 2v + l where u m2») 9a 2u + 2u'vu + 5uv2v + v s 1 l tT. or is a node. is This equation is of the fourth degree in t. Find the double points. so that the curve fourth class. and obtain « = . In the latter case the coincident tangents from are the tangent to the other branch of the curve through the node. 3a. . the tangent at any point is obtained by putting 1 for z in equation (ii) of § 1. of 1. If in the original values of x and y we replace t by \/T and repeat the process. . If the curve is referred to Cartesian axes. e. The inflexions are given by 2t 2 2il = or * = l(l±v/ 3). (ii). t = a cusp. gives occurs twice on the lefthand side. t. the coordinates of P P any point being the results of §§ 1. . This gives 2tT = l. The equation (v) of § 1 giving cusps and inflexions is of the f(2P2tl) Since the factor t = 0. + l)x{tl)y9t(2t2 l)a=0. where This equation is of the fourth degree in Any tangent to the curve is s St (2t so that the curve is a quartic. and class x [The curve meets = (2tl)a/t 1 ~Kx + py + a = {2il)\6(t i + f)^ + t 2 =0. 2 evidently still hold with slight modifications . . t = <x> also gives a cusp. tangents at the points with parameters *.
17. 1) given given by fi — t+1 byi! + *+l 0. Find the double point and inflexions of x [Crunode = P+ l. + */2 the coincident values of t implying that t = gives the point of undulation (0.0. 9. or 22. 8. a(t i 2t"). 2 [Acnode is (0. 1. = t* + 255. \ are the parameters of and — 1. 0. z = *9. Inflexions are 1. 1) given by t + t+ 1 = 0. t (0. is = {t\f.] Find the double point and inflexions of s x = t* + t+l. z = cos<£.] crunode where infinitely distant tangent where t = 00 = = a(fZt).] Ex. y t* . Find the inflexions of y x = t»131.] Ex.] [Acnodes are = ( — 1. which gives the parameters of the node. Cusp is (1. 4) (3. point of undulation with . .0.1. = 3. (1. (0. xi y2 + 12 as (3 x + 2 y) +108 a* Ex. 6. Find the double points. 0) z=t*t. 5. the corresponding values of T being oo 0. co ) — 0. Find the double point and inflexions of x [Acnode is = t2 + l. inflexions. 0) given by t = 00. &c] z 0.. and bitangents of x [Cusps where t* = 1. = cos 3$. = 0.0. +i. 0). — 1. 0). Inflexions are given by t = 0. 1. oo . Hence 1. —1. given by i 3*+4 = s « 3« 3* + 7 = 0. 1). 0). 4. 7. [t= oo.8) given by 3* 2 + l =0. 3. Again. and bitangents of x = t1 z = t. 0.0) and (1.2 +10Tl = 0. given by = 0.1.142 USE OF A PARAMETER . 1. 1. (1. 1.] Ex. . 2. = <s + 3i + 4. y=\ + t\ . 1) . is (1. Find the double point and inflexions of a. .1). given by t y=t\ y=(t+\f. the equation (ii) considered as an equation in t has equal roots if 2T2 + 2T1 = 0. y = i + fi + t + l. which gives the parameters of the intersections of the curve with the two inflexional tangents. as before obtained. (8. The bitangents are x = and x + 4y = given by fi = and 2* 2 + l=0. Inflexions given by 3 t (2t1)*+1 Ex. \.1. \ . 1.5. 0) = 1. The curve is (see Fig. (8. X 2 This gives coincident values of T if t = 1. 4. = 0.] Ex. Inflexions are (0. 1) =0 . \ are the the points of contact of tangents from the cusps parameters of the points of contact of the bitangent. 8. Find the double points. [Crunode is 1) given by # = +£71. Inflexions are given by 3. 2 Inflexions given by Ex. 1. y=t + St + ±. J. 8). (2. Find the double point and inflexions of x [Acnode (1. . Ex. inflexions. 0. ^ = sin3<£.] y ii = + fi t. » and (1.
.X 2 Ex.] = f+8 = 0. with respect to t. third. 0) given by t = oo and t The third double point is also a cusp if a 4 6. a and (32 from /i<£ 2 = <t>2^t. If/. 2nd. z2 yz. and fourth derivatives of these with respect to t. 7. = + . 14.+ . quartic. t : 2t s : (* 2 +l) 2 0.(a>ai). Find the double points and inflexions of x:y [Double point at Ex. inflexions where t(t 3 i) = 0. 10. [The reader may apply the same method to find the osculating sphere any point of a twisted curve. a = a +<x + +an J»(»l). 8 /3 ) the acnodes are given by solving for + ipf. V'l/a = Wi..] inflexions at : = is y : z =f{t) <j> : ty{t) a curve and /(a + V) tf'iV'a =A (a.. . o . <f> . xy. USE OF A PARAMETER Find the double points and inflexions of 143 [Triple point where dulation where t oo .. of closest contact. The equation of the conic of closest contact at any point of x:y:z=f(t):<t>(t):^(t) is obtained by equating to zero the determinant whose first row has the elements x2 . &a. &c. . Find the double pdints and inflexions of x :y:z= (i 4 + ats ) : (t + h) : t\ [There are cusps at (1. yjff. y*. ifr are trigonometric or hyperbolic functions. £<o . 13. =/20i Apply the method Ex. . zx.+ fl . 1 (t) : (1 t + t + 2f + t i ).' we may express them in terms of exponentials.] Ex.. . +&. J a. t IXn and whose other rows are the 1st.. using the fact that if the first row of the determinant is t/ . 0) and (0. to Ex. \jr are polynomials is easily carried out by noticing that the determinant. Bitangent s = the points of contact Any tangent is t(t* + l)x(t* + l)y + tz=0.). cj>. The evaluation of the determinant when /. Find the double points and bitangent of x:y :z=(«2 l) [Cusps where given by t = + i. . o . its value is Be b where . whose second row has the elements Z2 2 ip. Find similarly the cubic.. 0. given by t= 2b] = = 0. p). whose first row is *0tl +U2 . 6 = (a. 12. + a )«+(01 +j3. is Et a where R {n— l)th derivatives of these = (a„a„i) Kanj) . <p.] . oo . #!//.. un Ex. ..(a»afi) (a»_ia«») .. at .. 1. 11. (l + * 2). 15. second. 1 % . and whose other four rows have as elements the first./3. 2 J. + a. Ex. If x : t :s = = oo . (a.. f(f>.
fourth. II. s (ii) (5t* + 5t*+l0t? + l)x* + t(t!> + 10t' + 5t + 5)i. The (n — 2)ic and 7iic meet twice at each double point of the n.] ^ p.144 Ex. so a+K = $(nl)(»2)2>. The nondegenerate conies of closest contact with a q r y p are all ellipses if (p — 2q) (q — 2p) >0. 15 are given by equating to zero the determinant whose first row is p. otherwise unreal.ic and once at each of the + other fixed points. and all hyperbolas if = a# (p2q)(q2p)<0. (p+. (Ch. in real points. USE OF A PARAMETER Find the conic of closest contact at any point of (i) X 2 x y : : s (ii) x : y : z = t" n =t f : : : 1. and fifth with yPz* = There xP + is no nondegenerate conic having sixpoint contact <>. and suppose that the curve does not degenerate into two or more curves of lower degree.* + tW (5te + 1) yzt* (f + 5) zx2? (5t e + i s + 5) xy = 0. first. [Replace q by p+q §3. meet x = intersections with z = are real if q (i). second. M f<t> and whose other rows are the Ex. [Use Ex.] [(i) i 1 l *> Ex. <j>\ *». derivatives of these. in Ex. W. 20. 16 (i).q)(p~2q)qH ix + (p + q)(q2p)pH'"'y i + (pq) 2 (p2q)(q2p)t*P +1 <tz + 4(p + q)(pq}(p2q)pt 2 + <iyz + 4(p + q)(qp)(q2p) qtp+i izx + 4t(p2q)(q2p) pqt» + " xy = 0. : 1 + I s . A curve of degree n—2 can be drawn through the (to1)(w2)Z) double points and through any l(«2)(»+l)i(»l)(m2) + 2) = »2 + i) other fixed points of the given nic. since the (n— 2)ic is determined by ^(n—2)(n + l) points. Their where p and q are positive integers. 19. 16 Deficiency not Negative.] Ex. The parameters of the sextactic points (at which a conic has sixpoint contact) of the curve of Ex. 16. §6). Suppose that now that a curve of degree n has deficiency D. n—2 D They therefore meet at n{n2)2{\(n\){n2)D}{n2 + D} = D . The nondegenerate conies of closest contact with y p zt = x? + i. ~ Ex. 17. 18. third.
) [Each intersection of two constituents of the «ic counts as a node of the nic] * If the nic this statement was degenerate. 2. Hence in this case D>0 . &<?. we have subjected the (n — 2)ic to 2^k(kl) + (n2) + D = (m2)(%+l) II. An (n — 2)ic can be drawn with a (k — l)ple point at each /cple point of the wic.] Ex. IX into a curve with ordinary multiple points without altering the deficiency.] An nic cannot have k + k 1 + k s + ki + k5 >2n. § 6 But the 7iic and (n — 2)ic meet in n(n2)2k(kl)(n2 + D) = D other points. 2216 L . 3. if l points. cannot have two multiple points of orders k l and & 2 . 1 D... in(nl)(nr)(D + Dl + +£>. and also passing through n — 2 + D other fixed points of the nw. t The summation is extended over all the multiple points of the nic. multiple points of orders k 1. 1. if &! D>0. An nic + k2 > n.. Ex. [Consider the intersections of the wic with a pic through the triple points. 1. 1 . 3 and Ch. also Ex. [Consider the intersections of the nic with the conic through the points.. § 3 D = i(nlH»8)2**(*l). It should be noticed. If the curve has higher singularities. . . If an «ic degenerates into r curves of deficiencies D1} D2 ithas double points.* It follows that n(n — 2) The deficiency of a nondegenerate curve cannot be negative. [For otherwise the line joining the points would meet the curve in more than n Ex. since the nic and (n — 2)ic meet points. VIII. 4.k !i . that there may be other restrictions on the nature of the multiple points of a nondegenerate curve in addition to (see Ex. 5 to 7 Ch.. k 3 . .. If a nondegenerate nic has \p (p + 3) triple points. 2. IV. k i. We have so far supposed that the wic has only ordinary nodes and cusps. Ex.k i . For by Ch.t and by Ch. XVII. and would not be correct. however. Now let it have ordinary multiple points. it may be transformed as in Ch.. § 6. so that the previous argument still applies. the (n— 2)ic might be part of the mic..: •X 3 DEFICIENCY NOT NEGATIVE in' 145 other points. conditions.] Ex.). § 7.
it consists of n straight If it has \n(nl)l nodes. obtaining an equation of degree Of the roots of this in x with coefficients rational in t. In a similar manner the ordinate of is expressed rationally But. curve of zero deficiency is often called a rational or The former name comes from the fact unicursal curve. DEFICIENCY NOT NEGATIVE «4 X 3 If a degenerate raic has \n (n . where u . it consists of n2 straight lines straight and a conic. e. Suppose Consider now an nic f that a variable iVic is subjected to certain conditions. If it has \n{n\)2 nodes. it consists of lines and two conies. Unicursal Curves. the curve of § 1 is of zero deficiency. The latter name comes from the fact that the curve can be drawn by a pen which never leaves the plane of the paper This implies that. the coefficients can be expressed linearly in terms of a single quantity t. giving the abscissa of P.1) nodes. Dividing the equation by the factors corresponding to the abscissae of the fixed points. If Cartesian coordinates are used (a similar process holds good for homogeneous coordinates). the jVics form a pencil. lines. and the remainequation of the wic ing root is the abscissa of the other intersection = = nN nN— P and i\Tic. . that the coordinates are expressed rationally in terms of a parameter. Thus the equation of the iVic is of the form are two such iVics i. 4.] §4. Hence P P The coordinates of any point on a curve of zero deficiency can be expressed as rational algebraic functions of a parameter.: 146 Ex. 5. Suppose also 1 times at these fixed points. Since the %N(N+ 3) ratios of the coefficients in the equation of the i\Tic are subjected to %N~(N+3) — l linear relations. [See Ex. f(t). and v u + tv 0. may be made any point of the nic. by choosing t properly. that the i^ic meets the nic We assume for the present that such a variable iVic exists. we have an equation of the first degree in x with coefficients rational in t. we may eliminate y and u + tv between/ 0. such as passing through fixed points of the nic or having assigned which conditions are singularities at fixed points of the nic in all equivalent to iV"(iV"+3) — 1 independent linear relations between the coefficients of the equation of the iVic. = nN— = = = . be taken as polynomials in t. in terms of t. 1 are abscissae of fixed points. with zero deficiency. if and \jr(t) may A . <p(t).
i. i. §3 . * See Ch. is an acnode. the point moves continuously as t varies (provided is finite). the nic. times at the fixed points as required. §i\T(i\T+3)l and the ific meets the nic 2k(k\)+n3. and subject the JTic to the conditions that it shall have a (/c — l)ple point at each /cple point of the nic. P is also given by t a — /3i. and also pass through n — 3 other fixed points of. II. Or as an alternative we may transform the nic by § 10. or to insert the acnodes in other words.JV1 . e. is a double point. First suppose the nic has no multiple points other than n — 2. We now show that there always exists an i^ic with the properties stated at the beginning of this section. where a and /3 are real. and subject ordinary nodes and cusps. This would not hold if t were complex but.e. «. Lastly suppose that the nic has higher singularities. and use the result of Ch. e. IX. 9.g. 147 except to pass from one end of an asymptote to the other. curves of deficiency other than zero may consist of a single circuit . i. nFl as required. l2 . 2JJb(Al) + »3.+ (Hi gives evidently an unreal point of the curve. We take the iVic as an adjoined (n — 2)ic passing through n — 3 fixed ordinary points of the nic. Since any value of t close to tx . XX. Of course. §§ 1. f It is often convenient to take them as points on the nic consecutive to double points. 2)=$(nl)(i»2). coefficients in the equation of the iVic are connected by N= i(»l)(m2) + (n8). y i x+ (x — 1) (x2 + 2x + 2) = 0. + pi. § 6 are connected by linear relations. the curve consists of a single circuit.2£&(*l) = 0. P P . i. if a real point on the curve is given by t — a. Then the coefficients in the equation of the JVio by Ch. the coordinates of a point P on the curve being continuous functions of t.* In fact.X 4 UNICURSAL CURVES . Next suppose the «ic has ordinary multiple points as well as ordinary nodes and cusps. i. e. We take the i^ic to the conditions that it shall pass through each of the %(n — l)(n — 2) nodes and cusps of the nic and also pass Then the through to — 3 other fixed points f of the nic. $N~(tf+8)l and the N~ic meets the nic 2 x %(nl) {n2) + (n3).e. We take ZV = n — 2.e. VIII.e. times at the fixed points For by Ch. there is no real part of the curve in the neighbourhood of i. P = P P P . linear relations .
X .
. Ill. 3y)Sx + 6xy + y* = 2 [A variable conic through the double points (0. («/ where it{2t+l)^{2fi + t + i) (iii) Meets . = (i 2 6« + 2)(< 2 3*l)r6(*l)(<2). Fig. (iv) Meets txy = 5 (a. 1 ?/ i . = (< 6< + 2)(< 3< + 5)r6(<l)(<2). = txy where (vi) Meets z(xy) x(tl) i = y{t + l) = 4:z.yf = ixyz. a.i . 3. Meets y(2 + i tx) Jrx 2x = 2 (2* + 5i + 4).y) where a:=15=(2* + 2« 2 ). (ii) v (iii) .) + {x . 4. (iv) + 2y* + l)(yx) = x*6xy + y\ x s + y s + 4xz(x + y) = 0. Ex. i (ii) 2y*(2x *5x + 4:)xy(x2)(xg) + 2x' (x2y = 0. 0.a a X4 Ex. 3. Choose B as the point on the curve consecutive to the cusp x = y = oo The conic is then . y = + 6o<) + 6a2 (2*. x = y = oo and one other fixed point E on the curve meets this quartic in one more point. (ix) r^ a 2 cos 20.3 x .0). Express rationally in terms of a parameter the coordinates of any point on the curves i 2 (i) a:V + 2x*y + xy 3x' + 2xy2y* = 0. y 2 ay s 3 2 ^ = ^"2 2 a. 1 and Ch.1) = where x~=l2tl)a + t\ y = 6(P + t)a. (i) UNICURSAL CURVES 2. x = —y = co . .] ' y _ (tf + 2t2)(? + t3) ~ 2(t* + t2) Ex.y") {hy . 1 1 2 2 2 (» . (x) (y [(i) + x^^ix^x + k). Ex. (x + y) (x . * <v) Meets (xy+l)(x + y + t) = t where 2 2 2. »y (iv) (v) (vi) (vii) (viii) xy(2xyhx10y) + 50(xyf = 0. 2. y = 1572(1 +t + t*). Meets y(l — x) = tx where (*+l)(3*)a: = 2* 2 2*+3 = (43i)^ (ii) a. (iii) + 12 a3 (Sx + 2y) + 108a4 = 0.2 xy + 9 y = i i z + z i x + x'i y i = 2xyz{x + y + z). (x + yt)(xyl) = * meeting the quartic where _ fl + 2t2){ti + tl) X~( 2{t* + t2) See § 6.x) . = 8r(2i 2 + i + 4). 2 r = o (1 + cos #). Express the coordinates of any point on the curve (x'*y 2 yi (x'i y*)(x + rationally in terms of a parameter. 149 Express rationally in terms of a parameter the coordinates of any point on the curves x3 + y 3 (2x li = Saxy.
Ill. 1).1 2 ) z\ (xi) y = tx (viii) = (t . (ii) y = t (x + 2) (iii) ty(x + z) = x(y + z). t(3t2)(tl)x = (tl) s y=t(lt 2 9t + 3)z. 0. 0) and (1. 0). y = tx. 0. —1. passes through (1. ) ) Ch. (ii)*(* + 2)*=y*(ly).150 (vii) UNICURSAL CURVES Meets x 2 ay X 4 = tx(y — a) x=(2Zt )a¥t\ 2 where 1 2 y = (23t )(2t )a + 2t\ y (viii) Meets ay x (ix) = 2 = t(x + y ) where 2(1 t 2 ) st(1 + t 2 ) 2 2 2 . s 3 (vii) (y + X ) (y + 2x ) (y + 3a. Meets y + x 2 to where x Ex. 2 2 (ix) x (xy + az 2 ) + y(xy + hz ) 2 = 0. — 1. 1). 4 (i). 0) and (0. is given in § (iv). (vii) y (x) 4xy = (1 . (\)y = tx 2 (see (i) x* . 2 2 6 3 (vi) (yx ) = x + x y. 3 2 s 2 2 3 2 3 % 2 2 3 (iv) y z + y z + z x + z x + x y + x y =2xyz(2x2 + 2y2 + 2z2 — //zzxxy).x(yz). Ex. In fact x(z+y) tz(x + y) meets the 3 curve again where (t 2 — 1) x — y (t — l)z. 1. 1) and through two more fixed points on the 5ic consecutive to It meets the 5ic again where (1. 5. 6. 0). We may illustrate the general theory for which n — 2 by taking the JVic as the cubic ' = = N= . In the case of a 5ic with a triple point and three double points the Nic of § 4 may be taken as a conic through these multiple points. and to apply the method of § 4 to the transformed curve. 2 2 (v) (x + y)(x 2y) = x 3 y. 2 = t(t 2 l)a + (t* + l). y=(t k){Jc + 2tt 2 ). An alternative method is to transform the curve by quadratic transformation. 3. 0) respectively. 0).] 3 .t z . (vi) y = tx 2 . (1. 0. 1). [This quintic has a triple point at (0. (yz + x ) = y (x + y). a node at (1. 2 2 (x) x (xy + az + y(xy + bz2 ) 2 + z(xy + az2 ) {xy + bz2 = 0. 10). 0. Meets x + y 2 = at (x — y) = t 2 where y 2 x (x) = t(t 2 + l)a + (tl + l).] Express the coordinates of any point on (x + yfz2 = x?(y + zf rationally in terms of a parameter. 2 . = 4ta + (l + t2) 2 . = h. 0. (vi). and cusps at (0. Fig. 2 3 (xi). Express rationally in terms of a parameter the coordinates of any point on the curves + y* = ax (x2 — xy + y2 ). 1.] Ex. (0. 1. replacing x y z by 1/x \/y \/z. 6.1) x. (iv) ty (x — z) =. s 3 s s 2 2 (iii) x (y + z) + y (x + z) = x y (x + z) (y + z). xyz + y 2 z + t(x 2 y + 2x 2 z — y2 z) = which has a double point at (0. which is a cuspidal cubic. [Find the intersections of the curve with (i) y = tx. The reader : : : : may Yet another method apply this process to Ex. 2 2 (ix) xy = . 3 ) = y 2 x i (viii) s« n _i + un = 0.
[The tangent at D = 0. VIII. +& = giving a 2V = iN(2nNS). the highest common factor is (t . § 5. by l/(t the curve has a cusp given — OC). while Ex.tt> + i. t — (X. 7. Ex. 2(i and Nia are « double raiVl. changed in sign by interchange of t and t t . if T is the ratio of the coefficients of any two powers of tx and is not a constant. b = N(N+3n)l. <t>. i? are polynomials of degree n with the coefficients of t n " x zero.] y»& = xv+z. Suppose that to each point on the curve corresponds r values of t..] Ex.l) Hence is (p + q) Wx =py + qtP+iz. 1.. . (f>. But the expressions whose highest common factor was taken are only See Liiroth. (t . See Clebsch. 8i Three points on a curve have parameters t lt t z ts Transform the parameter so that the points have new parameters 0.] to make zero the coefficient of I " . K = i=p + q — 2. p. Conversely. where /. § 9. UNICURSAL CURVES or 151 u—1 n — 2. The coefficients of any two powers of t in this are symmetric functions of tlt t 2 . 11. . and that the highest common factor of : = . 13. 4>. so that to each point on the curve corresponds a single is value of T.1 ... Then.1. Ex.).. p. where /3 is chosen Ex. = m=p + q. m are given by §§ l'and 2. The curve of Ex! 9 has a cusp.] . If in § 4 the »ic has only ordinary cusps and nodes. [The new parameter is (tty) (t2 : : t3 )/(t — t 3 ) : : (£ 2 — tj). 1. See also Ch. take as a new parameter T=fi + 7 1 in a. Hence > »3. Then a + 6=iJV(JV+3)l. : y : z =/(<) :<f>(t):^ (t) [n and 12.] : : Ex. X 4 Ex. For other examples find the Plucker's numbers of the examples in Ch. yjr Find the Plucker's numbers of x being polynomials in t. /(*)*(«i) /&)*(') and *W*(*i)*(*i)*(0 arranged in powers of t lt the coefficients being polynomials in t. «ic N must be points [Suppose the fixed intersections of the and 6 other points of the wic. Annalen.. lxiv (1865). f. b = r =>\{p + q2)(p + q2. Math.tj (t . 44.2 ) .). t. In the «ic x y z =f(t) <b(t) \? (*). Ill. curve with a cusp can be put in this form. ix... 9. of t. . and therefore their ratio has a single value for each point on the curve.. to each point on the curve corresponds in general a single value of t but this is not universally the case. the sum of the parameters of the intersections of the nic with any algebraic curve is zero. Oe«e. yjf are polynomials. 163. any unicursal if [The cusp is given by t = oo Conversely. In the curve x y z f(t) <j> (t) : yjf (t). 10. [If to a point on the curve correspond the values t lt t 2 .] . n (P. oo . t r . . . x:y:z can be expressed rationally in terms of the parameter T. Ex. Find the Plucker's numbers of D = 0. where /.
every trio of corresponding points is collinear. Bull. ***.. OY. /(«) Math.C. . in x 1 y 0(f) z tu . * is be expressed rationally [The degree of F(t) in 2 K + t. 1 s z : Ex. 15. n. 16. is the equation giving the where /. m : : : \j/ (t) may .. AB .] Ex. + 3< 2 +l).] Ex. [x/c = t{\t){l+tf/(\+ti f. i^ + ^lO*!. y/c 2t(lt)/(l+fl f. : (fi +lf i : t{f + : 1) : 4 : (« I s .. 18. Soc. («. f(t) *(*) I . Ex. xxiii (1916). between the equations so obtained.] < *. if /. t 3 . ijr are polynomials of degree n. *.« fo" 4 1) + *. Ex. 19. In the following curves express the coordinates in terms of a parameter so that to each point of the curve corresponds only a single value of the parameter. s If T=P + r\ x:y:z = (T2):T :l. 90.. If more than n^ + n^ + n^ trios of corresponding points on given unicursal curves of degree n lt n 2 n s are collinear. If a. p.z=T:\:T. tjjen F(a)/F(b) is the same whatever »ic is chosen. [We thus get the equation by equating to zero a determinant of order See Richmond. A line AB of length c slides with its ends on the rectangular axes OX. where t the tangent of half the angle between and OX. b are the parameters of a given node on a given »ic x:y:z=f(t):<t>{t):*(t). <t>. ty. «. of Ex. Apply § 1 (v) and the relation 2k 4 1 that. \f are polynomials.Af*<tPTf<. 17. = {<*> and F(a)/F(b) = {f(a)/f(b)}< («)/0 (ft)}' = ft. fi. if the coordinates of any point on a curve can in terms of a parameter. 14. and F(t) = parameters of the nr intersections of any ric with the nic. 13 (i) (ii) (iii) (iv) T = t\ x:y. 0(«) *(«) and eliminating 1..F. 20. Find the locus of the point of contact with of the circle inscribed in the triangle ABO. x:y:z=(T + 2):(T s + 3T):l. =S(n — 2) + 62) show D = 0. If T=t + t~\ x:y:z=T>:T:(T< + l). The line joining corresponding points (with the same parameter) of a unicursal wic and a unicursal ATic envelops a curve of class n + N. Pt*. : : t(t t + l) is (f1) (iv) x:y : z = (t^^ + tl) If 8 : (< + l) s : t\ [The H. [Ifthericis 2Axa y&z~> = F{t) 0.)/* (b)}':] to' Ex. Amer.*(l+* l )*4<.152 UNICURSAL CURVES X 4 Ex. .I f T^tt'. The homogeneous equation of x y z =/(<) <t> (t) 2 s be derived by equating to zero the coefficients of 1.] = AB is is . (i) (ii) (iii) x y z= x y: z = x y z= : : t* : 1 : i 6 . where a + j3 + y = »y then = ?. w .. Hence the locus a unicursal quartic with a node at the origin.
and sin so that the parameters lt <£„. Putting t = tan  . rational' expression in terms of cos <£ and sin </> The following theorem is of interest The coordinates of any point on a unicursal nic (n > 2) an acnode can be expressed rationally in terms of cos $ .] § 5. 0. <j> <f> . H and K being fixed points. Here the most convenient parameter is the eccentric angle of any point. HB KC Ex. it by Suppose that the triangle of reference (1.] Ex. of the inter2 sections of the nic with any ric satisfy the relation with <j). . 24. Ex. 25. B. We may is taken so that suppose the parameters of . The case of the ellipse is well known. it may be more convenient to express them rationally in terms of cos <£ and sin </> or of cosh $ and sinh (f> where <f> is a new parameter. 22. The locus of the poles of any normal to a given conic is in general a unicursal quartic. q. we can immediately change from . A Any [Its node is at the vertex of the pencil. r are linear in x.. y. <t>i + $2 + and • • • + §nr cos iri. the rational expression of the coordinates in terms of t to the or conversely. 21.: X 5 COORDINATES OF A POINT 153 Ex. 23. sin "")• (p For an nic with a crunode cosh <j) and are replaced by and sinh 0. The locus of the intersection of the tangents at corresponding points of homographic ranges on two given conies is a unicursal quartic. AP [A unicursal cubic and quartic] Ex. find the loci of the intersection of the line through A perpendicular to with the tangent and normal at P. Show that the^ locus of the intersection of and is a unicursal quartic. pt? + 2qt + i=Q. Q. —° </> (wwd. If A is an end of an axis and P any point on an ellipse. Consider the case in which the conic is a parabola. [Th* tangents can be put in the form Pl 2 + 2Qt + B=0. s. . where P. 0) is the acnode.. range of points on a conic is homographic with a pencil of line of the pencil meets the tangent at the corresponding point of the conic on a fixed unicursal cubic.. lines. Instead of expressing the coordinates of any point on a unicursal curve rationally in terms of the parameter t. p. A triangle ABC of fixed size and shape turns about C. Coordinates of a Point in terms of Trigonometric or Hyperbolic Functions.
= 1 and — 1. we have at once . + (t* + 1) ([6oA i + c If s h is the sum of the products of the roots of this equation at a time. For <p t 2 in this case also.) (Sl s 3 + s 5 'i/] li r(a a a + a4 . If a variable Nic (N<n) passes through iV(iV+3) — fixed points of a unicursal nic (as in § 4) and meets the nic 2 times at these fixed points....+<i) n ) = = (a + a a 1 3 .. nN— * coinciding with one of the fixed points.. Hence... = 1...ic. any n collinear points Similarly the sum of the parameters of the intersections of the nic with any ric is a constant (mod.. (mod. y = (t = (t* + l)(c tn *+ + 1) (b t n ~ 2 + . if (x. . or ^+ where tan a ( (^2 ... Let <f> 1 (f> 2 be their parameters and let /3 be the . <j) 2 . if any straight line meets o^.. One (or both) of the Nics here obtained may meet the wic at a cusp instead of touching it.) = 0.. ..1 154 COORDINATES OF A POINT X 5 the acnode to be i and— i (see end of §4). That this constant is zero is evident by taking the ric as r straight lines.. x=(a t« + a 1 P.1 + z .)i(a a2 + ai .. and put t — tanh <£.. equations of the form Then we have . two of the family of i^ics touch the nic in general* For the nic and iVic meet at two more points. y.)=(«„ — a 2 + a i — 6 + a/n. n).. 2 +6„_ 2 ). For the corresponding result in the case of a cusp see If the point (1. + c n _ 2 ). = — c^ + by Replacing of ftg — a + .... 0) is a crunode. <f> n the nic in the points for which <j> t&n(<l> 1 + <l> i +. so that..) = (al + a3 a + .). +an) 2 3 i»" + [6 l/t + c^] i»" +. The nic meets any straight line Xx + fiy + vz where X(a <" J = 1c + a. we take its parameters as §4.... 0. it).. x:y:z are expressed rationally in terms of cos <f> and sin <j>. z) is First take r any point of the n.i' . +a„). we have the sum of the <£ . At a point not — </> .).. 10.. parameters (mod. Ex.1 + . tt). 6 +. if we put tan </> for t.)..)r(ls2 + s4 ... §6... +<f> n = a.
. VII. This process gives a method of expressing the coordinates of a point on a unicursal curve rationally in terms of a parameter alternative to that given in § 4 which. 2{J(»l)(»2)2} =n(w3)2. IX.X 6 COORDINATES OF A POINT 155 of the parameters of the nN— 2 fixed intersections. if a and ac — b are 2 positive and similarly in other cases. N . If the JVic touches the nic. <f> 2 it). whose coefficients are rational in t. 11 and 12. is gives the two Nics which touch the nic. 1 and 2. . if homogeneous coordinates are used). If the multiple points of the nio are all ordinary nodes or cusps. Ex. may be much easier in practice. . Since of the second degree in t. the Nic may be taken as an (n— 3)ic through all the double points of the nic but two since . § 10. x where L. by the sub = 2((Xc6 2 )lrr(T 2 l). out by the factors corresponding to the nN — 2 given intersections. we have two equations for y which have a M are rational in X common solution of the form M' + L'X~z. i(n3)nl =i(nl)(n2)2. : f/3 follows from Ch. ^ = and or = f/3 + 4tt. t. and Substituting either of these values of x in the equations of nic and Nic. if less simple in theory. giving the two variable intersections of nic and JVic. in terms of a stitution new parameter T +b for instance. Ex. The coefficients in the equation of the i\Tic may be expressed linearly in terms of a parameter t On eliminating y between the equations of the nic and iVic we get an equation in x Dividing (or in x z. Solving it we have . Hence the coordinates of any point on the nic may be expressed rationally in terms of t and an expression of the X= X form (at 2 t But and + 2bt + c)k (at i + 2bt + c)^ at can always be expressed rationally . since the value of may be lower than in the method of § 4. Ch. Then sum </>i + <t>2 +P = ° (mod. we have a quadratic equation in x. where M' and L' are rational in t. *!= This result also § 10. = M+LX? is a polynomial in t. the parameter of any point on the nic being taken as in § 5.
The second Nic of § 6 is the line at infinity for this line passes through the infinite cusp. when we eliminate y between the equations of the wic and iVic. In the argument of § 6 must be n — 1. S . Then. Suppose that just as in § 4 a variable . [A variable line x + y + \ curve where = through a node at infinity meets the 2(X2)(X + l)y ' ' = X + 3X + X (3X)*. . in the equations of .y^f .] Ex. 5. Dividing the equation by the factors corresponding to the abscissae of the fixed points we have an equation of the second degree in x with coefficients rational in t whose roots and Q. N Ex. [As in § 4. the Nic may be taken as a straight line through the (n— 2)ple point and so on.%a? + 6 xy + y* = rationally in terms of a parameter. 3. we get an equation in x 2 of whose roots are the abscissae of fixed points. 2.y ) (x + Sy) . Apply the method of § 6 to the examples in § 4.] Put \=t — 1.the two variable intersections P. Express the coordinates of any point on (x + yYz2 =x 3 (y + z)* rationally in terms of a parameter. Ex. Ex. Ex. 1. 4.156 COORDINATES OF A POINT . X 6 If the %ic has a (n — 2)ple point. and' Q of nic and iVic. that the iVic meets the nic But suppose times at the fixed points (not 1 times as in § 4). while the remaining two roots are the abscissae of . or n—3 in the case in which all the multiple points of the wic are ordinary nodes or cusps./Vic u + tv is subjected to certain conditions which are in all equivalent to \ N(N~+ 3) 1 linear r relations between the coefficients of the equation of the A c.{x* . 3. Curves with Unit Deficiency. [y = \x meets the curve where \x/z = — 1 + (X+ 1)^. 7. and we have (t 1 — 1) x = y = (t s l)z as in §4. Consider now an nic with unit deficiency. Express the coordinates of any point on the curve 2 1 (a. t M and L are rational in and X is a x we substitute either of these values of polynomial in t. 4. Ex.] Ex.] §7. 2 Ex. Suppose its solution is are the abscissae of = — nN— nN—2 nN— P x where If = M±LXi. Putting X = 3 — (t+ l) a we get the coordinates given in § 4. [As another example the reader may take the general conic or the quartic with nodes at the vertices of the triangle of reference. n — 2.
XV. Ex. cnu. a & g3 1 «„ we have. 1). Ex. Ch.. or Goursat. p. § 10. the values of t given by are the parameters of the points of contact of those of the variable iVics which touch the n. If then we assume the existence of the iVic. positive. But by Ch. IV. L = w ^ «2 ^= 1. we take a as a constant defined by 3 . where M' and L' are rational in t. defined by . VII. dnu or if preferred. putting p 2 and 1. IX.ft B + 4osft + a 1 is See Halphen's Fonciions elliptiques'. well known that t and can be simultaneously expressed as rational functions of the elliptic functions snu. Hence is a polynomial of degree four in t. For instance. § 10. Ch. §§ 2 and 4. §§340 and 362. on putting ai a {a t i 3 2 . Traite A' Analyse. II. i P'up'a pu — pa a<? + 4. * If o is negative. put t r _1 + fc where' ro fc* + 4<i 1 fc 5 + 6o.aI t + 6a 2 t + 4'a st + ai }i = {pup(u + a)}* Hence: The coordinates of any point on a curve of uniUdeficiency can be expressed rationally in terms of elliptic functions. 120. we get is also of the form whose common root M' + L'Xh. Ch. Ch. and proceed as above.ic at a point other than one of the fixed points. §§ 12 and 16. XVII. a 2 . a p'a a 2 a3 — 3a a1 a 2 + 2a 1 3 *0 . we have Now X= = D= X may The coordinates of any point on a curve of unit deficiency be expressed rationally in terms of a parameter t and an expression X of the form { t* + 4a 1 £ 3 + 6a 2 i 2 + 4>a s t + a4 }i It is. Corns d'Analyse. we see that the number of such points of contact and cusps is four. II. in terms of Weierstrass's elliptic function pu and its derivative p'u.'1. = ' . where a„ 2 3 a o2 9* = a ai 4a1 a3 + 3 a 2 . and the parameters of the cusps of the nic which are not included among the fixed points. pa = — (a a2 — af). 11 and 12 (cf. X (p'u) 2 = 4(pu) s g2 pug3 if . The reader may also consult Picard.a : X 7 CURVES WITH UNIT DEFICIENCY two equations in y 157 nic and iVic.
158 CURVES WITH UNIT DEFICIENCY — X 7 To prove the existence of the Nia. x = ty.] . terms of [Consider the intersections of the cubic with a line through a fixed point of the curve. rcic [Consider its intersections with a line through a node with an (» — 2)ple point. § 7 (i). Express rationally in terms of elliptic functions the coordinates § 5 (i). we show that when the wic has only ordinary multiple points and cusps. if iV — = D=l. x = tz. § 14 (ii). the coordinates of a point on any curve with unit deficiency can be expressed rationally in terms of elliptic * also argue as follows : functions. 3 2 1 ) ) 0. IX. Ex. [Consider the intersections with x=ty. 2. y = tx. XVIII. Express rationally in terms of elliptic functions the coordinates of any point on (i) (ii) (iii) (iv) (v) (vi) = y(yx){yk' x). {yx) = (y + x) (a'x^ + Vf). of 4. any point on the quartics of Ch. Express the coordinates of any point on a quartic of unit deficiency rationally in terms of elliptic functions. however. if the %ic has only ordinary double points. z = t(x + y). i(»l)(»2)42*(fcl)= 1.] Ex. an (» 2)ic with a (fcl)ple point at each /cple point of the 7iic and passing through m 2 other fixed points of the nic satisfies n — 2 and the conditions imposed on the i\Tic. we may often take a lower value of JSf. y z = i» g xz g z x + y + z + 6mxyz = 0. Express the coordinates of any point on a cubic rationally in elliptic functions.x + cz (x* . In general we may take for the iVic an adjoined (n — 2)ic passing through n — 2 other fixed points of the nic* In practice. z = t(x + y). This follows from the second theorem of Ch. n2 + 42*(*l) =! ^iV(JV+8) — 1. But we may Since the coordinates of any point of a curve with unit deficiency having only ordinary multiple points can be expressed rationally in terms of elliptic functions. .if) = xz I i 2 (at + b i ) . we may take for the i\Tic an (n — 3)ic through all the double points but one. s s t % i i . In fact. § 10. 1. ax (y2 . 3. Ex. For instance. § 15. 1.z2 + by (z .] Ex. (x + y + zf + ^kxyz = 0. and any curve with unit deficiency can be transformed into such a curve by rational change of variables (successive quadratic transformations). and so for any Ex. tc2 + 2A(&1) = «iV2.
If gf 27grs 2 the curve has a single circuit given by real values of the parameter u.p 2u. p u. 2. 6. y. y = = where/. (ii) (iii) (u . If # 2 3 >27p 3 2 the curve has a second circuit given by values of « for which u — &>' is real. A . pu.\ Ex. where 2o>' is an unreal period of Weierstrass's function. < (iii) y = tx. [As in § in 5. . using Abel's theorem on the roots of an equation rational pu and p'u. When the coordinates of a point on a curve with unit deficiency are expressed rationally in terms of Weierstrass's function of a parameter «.. p'u. and B being Now we in differentiating repeatedly the relation (p'uY express 2 = 4(pu) g pug 3 2 3 . 3 . y\r are each of the form polynomials in pu. terms of pu. 5. = 1. « 2 m 3 are of degrees .. X X [(i) Consider the intersection of the wic with a (« — 3)ic through the . But there are exceptions e. we have <)>(u).X of 8 Ex. z) are homogeneous coordinates of any point on the curve.. where a polynomial in t.fy 2 ) (u .iff. The coordinates' of any point on an wic of deficiency 2 can be expressed rationally in terms of a parameter t and Xi. p"u. we may suppose the sum of the parameters of the intersections of the curve with any other curve to be zero.] Ex.. p'n. an nic with a (n — 2)ple point. CURVES WITH UNIT DEFICIENCY Express rationally in terms of 159 elliptic functions the coordinates any point on a a (i) s?(a? + y )=x y>.. (ii) u = ty*. p'u pu. 7. nodes. 3 y. where is a polynomial in t of degree 5 or 6. = A + Bp'ii. If (x. z * f(u) f(u). Ex. linearly &c. <j>. where u yz + x*. p'u.. where %. p u.g.. the curve is of unit : We now deficiency. The coordinates of any point of a curve are expressed rationally in terms of Weierstrass's function.] prove the converse theorem If the coordinates of any point on a curve can be expressed rationally in terms of pu and p'u. 8. . =k(xu.pu = hp'"u. For instance pho = ^(2p"ug2 ). (ii) As in § 4. . Ex. 7. [Consider the intersections with (i) yz = tx(yz). .yif) in x and z u s + zxyu 2 + x y 2 iii 2 = 0.Oty 1 ) (u 2 . It is in general impossible to express the coordinates of a point on is a curve of deficiency greater than 2 in terms of t and Xi. . .
The points of confact of any tangent through the point (x. y. The curve meets Xx + /xy + vz = where \f (u) + fi(j> (u) + vf(u) = 0. 148) will show that the deficiency But /..160 CURV. and so for ji. <j>. y. Theory of Functions.. z are linear functions of pu and p'u. The argument of § 4 (p.] Ex. . y. = 0. v. § 116. = f $/<(>'. so that A when reduced to linear form becomes of the type A + A pu + A l p'u + . and so for </>(u) and ijr(u)..ES WITH UNIT DEFICIENCY X 8 Hence we may take 2 f(u) = a + a pu + a 1 p'u+. 1) : : * Forsyth. ^ will not usually be rational functions of a single parameter. p'u. and has therefore n zeros.hi. /*=*'/*/*. In Q'yjr — <j)%lr' the terms ^n p^^u p&'Vu cancel. + A in _ i p^ n '^u. [We proved that with any choice of triangle of reference x. z) is given by \x + /iy + vz where (see § 2). Show by means as (pu. v A = *>**'. Hence three linear functions of x. Hence the class of the curve is 2n at most. is zero or unity. z can be chosen in the ratios pu p'u 1..+ an _ 2 p^. The lefthand side of this equation has n poles.* Hence the curve is of degree n. of § 8 that any point on a cubic may be taken by a suitable choice of triangle of reference. Hence in general D = 1.
for instance its polar reciprocals. M .CHAPTER XI DERIVED CURVES § 1. and 2 denoting partial differentiation with respect to the evolute of an algebraic curve is algebraic* is equation just written. In determining shall curve curve we make § 2. 2216 and then replacing obtained by eliminating x and y from the three equations f and rj by x and y. Evolutes. it is sufficient to find the number of its tangents passing through a single point.=y+m 2 +f2 2)/(W2fi2f22fnA%2) • suffixes 1 x and * Its y. the number of its intersections with every line is the same. &c. Since the centre of curvature at any point of the curve f{x. The evolute of a curve is the envelope of its normals. The case of the polar reciprocals has been already sufficiently discussed. the locus of the intersection of each normal with the consecutiye normal. 2 rj). it is sufficient to find the number of its intersections with a single line. Hence. Similarly to find its class. pedals. in order to determine its degree. if the derived is algebraic. We shall also determine the multiple points and any other peculiarities of the derived curves which may be of interest. orthoptic locus. Derived Curves. when the type of the original curve is given. . y) = is (£. 2 where 2 2 i = X+fl(fl +f2 WflfJl2f* fnfl f22)> r. Feom a given curve may be derived other curves by various geometrical processes. inverses. In this chapter we shall consider some of the other derived curves. evolute. the Pliicker's numbers of any derived use of the principle that. In particular we shall concern ourselves with determining the type of the derived curve (as denned by its Plucker's numbers). and the locus of the centre of curvature at each point of the curve.
. and the line which . Hence R is a cusp of the evolute. Then P 1 R' and P 2 R' are the normals at Px and P2 so that through R' three consecutive tangents R'Plt R'P. Hence n' = Sn + i. and meets mm' 3n times at these . In this case let Q' be a point on mm' near P. k'.162 EVOLUTES XI 2 t. Then Suppose that o>. and let Q'Pj. P'P2 can be drawn to the evolute^ all ultimately coinciding with mm'. The only other to the t infinite points of the evolute are those due inflexions of the given curve. m. QR) is harmonic. P P P P Fig. t'. The tangents from R. P PR i. Let (mm'. — m' are the circular points at infinity and that the tangent at to the given curve meets toco' in Q. e. while (o>o/. one for each of the n intersections of 0)0)' with the given curve.. D. S'. 1. touches the evolute. m. and of its evolute by n'. m'. Q'P2 be the tangents from Q' to the curve which touch at 1 and 2 adjacent to (Fig. m normals corresponding to the m tangents to the given curve QR) is from a point Q on mm' are the o>o>' harmonic. and QR coincides with o)o)'. 1). We see then that the evolute has n cusps at infinity at which mm' is the tangent. D\ We now determine n'. k. where to the evolute (mm'. If is on oxb' it coincides with Q. cusps. . and mm' meets the evolute in three points at R. denote the Pliioker's numbers of the curve by n. We S. is the normal at P. m'. i'. Q'R') be harmonic. which is therefore a tangent to the evolute. in terms of n. at which mm' is a tangent. t.
XI
2
EVOLUTES
163
counts for n tangents, since it has n points of contact, namely the cusps of the evolute on a>a>'. Hence
m'
Again,
it
= m + n.
evident that in general no two consecutive normals coincide hence the evolute has not coincident tangents at consecutive points, i. e. has no inflexional tangent. Therefore
is
;
i'
=
0.
8',
Pliicker's equations
now
give
k, t.
1 1. Find the evolute of y — 4ax. [The normal at (at?, 2at) is tx + y= a (t" + 2t), meeting the consecutive normal at x = a (St* + 2). These two equations give
Ex.
x
= a(St* + 2),
(as*
2
,
as the centre of curvature at
2 at).
2Tay*
=
<p.
= 2ats Hence the evolute 4(x2a)\]
y
'
is
Ex.
(i)
2.
Find the evolute of
(ii)
(iii)
(iv)
x x x x
=
a cos 0,
ct,
y
=
.
b sin
= = =
at
2
,
y— y =
<j6,
e/'t.
at 3
a cos 3
y
—
a sin 3 (p.
[The centre of curvature is 3 i (i) x = (a  V) cos (t>/a,
(ii)
y
=
3
(Wa*)
sin 0/6.
3
x^c(Bt + l)/2f,
i
y=e(t^V)l2t.
).
(in)x
(iv)
= %at*, x = a (cos
3
2/
= (2< + 6<
(f>
cji
+3
cos
sin
2
<p),
y
= a(S cos
2
<f>
sin
<j>
 sin3 $).]
+ p Ex. 3. Find the Pliicker's numbers of the evolute of a*y = xp i, p and q being positive integers prime to one another. [m'=p + 2q, D' — O; n' = p + 2q if p > q, »' = Sq if p < q. For other examples take the curves of Ex. 2.] Ex.4. The number of normals which can be drawn from any point to
a curve
is
m + n.
[The number is m'. Verify the result by consideration of the fact to the intersections to a curve are lines joining that the normals from of the curve with the curve obtained by rotating it through a very small angle about 0. See also Ch. XII, § 5, Ex. 12.]
Ex. 5. How contact ?
many
circles of curvature of
a given curve have fourpoint
ma>'
[Each such^circle
is
given by a cusp of the evolute not lying on
3 (ri — m')
Hence the number
K
'
is
n =
+ in
=
hn — Sm + Si.~\
Ex.
6.
How many
lines are
2
normal to a curve at two points ?
[r'\n(nl)
=%(m
+ 2mn4mic).] M 2
164
•
EVOLUTES
XI 2
Ex. 7. How many pairs of concentric circles exist, each of which osculates a given curve ? Ex. Ex.
8.
The
deficiency of a curve
and of its evolute are equal.
foci
9.
The evolute has the same
and
directrices as its involute.
[A tangent passing through a> or a' coincides with its normal. The curve and evolute touch at the point of contact of such a tangent.]
Ex.10. There are »»(»» + »— 4) normals to a curve which are also
tangents.
[The
o)
common
or a.
tangents of curve and evolute which do not pass through See Ex. 9.]
Ex. 11. Find Pliicker's numbers for the locus of the extremity of the polar subtangent of a given curve, being the pole.
[The locus respect to O.
is
the polar reciprocal of the evolute of the reciprocal with Consider the case of a conic with as focus.]
Ex. 12. Find the Pliicker's numbers of the locus of the harmonic conjugate of a variable point P on a given curve with respect to the intersections of the tangent at P with two fixed lines.
[Reciprocate and project the reciprocals of the fixed lines into a> and
a'.]
Ex.
13. If
(toco',
where
Ex.
at
<o
a curve touches aa' at P, the evolute has an inflexion at R, PR) is harmonic, a>a>' being the tangent at the inflexion.
a curve passes through a (or a), the evolute has the tangent
14. If
as inflexional tangent.
aa>'
Ex. 15. If a curve touches
in g points
and passes / times through
i'
each circular point,
«"= 3» + i3(2/+£),
What
Ex.
a>a>'
m'
= m + n{2f+g),
=2f+g.
modification must be
14.]
made
in the results of Ex. 4 to 7 in this case ?
[Use Ex. 13 and
16. If
the curve has aa>' as inflexional tangent, the evolute has as a tangent at a point of undulation, and so on.
and
Ex. 17. The number of normals common to two curves of degree n. w, class m, m 1 is mn n. 1 + J
mm
l
+m
[The normals are the finite common tangents of their evolutes. n t = 0, m t = 1 we have the result of Ex. !.]
Putting
§3.
Inverse Curve.
to the circle
The curve inverse to f(x, y) = with respect whose centre is the origin and radius k is
2 2 f(k 2 x/(x + y ), Vy/(x* + y*)) = 0. If f— u + u 1 + u 2 + ...+un where ii r is homogeneous of degree r in x and y, the inverse curve is
,
u
It is at
(x
2
+y 2 n + k u
!i
)
1
(x 2
+ y2
n~1
)
+
••
+k 2n u n =
0.
once verified that this inverse curve has a multiple
XI
3
INVERSE CURVE
165
point of order n at each of the circular points co, co' * and has also an wple point at the origin 0. (See also Ch. IX, §§ 1, 2.) To find the Pluck er's numbers n', m', 8', k', t', i', 1 of the curve inverse to a curve with Pliicker's numbers n, m, 8, k, t, i, D, we proceed as follows. First of all, it is obvious from the above that n' = 2 n. Secondly, a cusp of. the given curve inverts into a cusp of the inverse curve, and vice versa, so that k' = k. or on <oa>') Again, a node of the given curve (not at inverts into a node of the inverse curve. Also the TOple points of the inverse curve at 0, co, and co' are each equivalent to \ n(n — 1) nodes, so far as their effect on Pliicker's numbers Hence are concerned (Ch. VIII, § 3).
D
V = 8 + %n(nl). We can now deduce m', t, l, IX. We find n' = 2n, m' = m + 2n; 8' = §«(«— I) + 8,
k'
=
k,
1
/ = 2n(2n7) + 4,mn + 2r,
As a verification we The tangents from
at
i'
= 3n+i,
D'
=D
]'
shall determine
mf independently.
n tangents each counted twice (Ch. VII, § 5) and the tangents from Hence m' = + 2n. to the original curve. The same result follows from consideration of the tangents from a) (or a/), remembering that a line through co inverts into a line through co'. The result If = D is a particular case of the theorem that two curves with a 1:1 correspondence have the same deficiency (Ch. XXI, § 3). We have proved earlier (Ch. V, § 4) that the inverses of the foci of a curve are the foci of the inverse curve ; and that, if is a focus of the curve, the inverse curve has cusps at
to the inverse curve are the
m
a>
and
co'.
Ex. 1. If the original curve has a fcple point at at each of a> and ta', we have
n'
and a #ple point
= 2n2pk,
6'
= $(«2p) (n2pl) + (npk) (npk\) + h, k = k
;
and the inverse curve has a (m2p)ple point at points at each of <» and <»'.
Ex. 2. What modification touches (bo)' ?
*
and (npk)j>\e
3, if
is
required in the result of §
the curve
only
In fact any line x±iy = c meets the inverse curve whereas the curve i9 of degree 2«.
;
at
n
finite points
166
Ex.
3.
INVEESE CURVE
A
curve of degree
XI
<u
3
2N
with
.ZVple
points at
and
a' inverts
into a curve of the
same
type.
Ex. 4. Find the number of circles of curvature of a given curve which pass through a given point 0. Discuss the case in which lies on the curve.
[Invert with respect to 0.]
Ex. 5. Find the number of circles passing through a given point and having double contact with a given curve.
Ex. 6. Find the number of circles passing through two given points and touching a given curve. [Invert with respect to either point.]
7. An »ic is selfinverse with respect to a circle j. Show that n even, and that the curve has a Jwple point at each circular point. Show also that the curve has \n(n — 2) foci lying onj in general, which are the intersections of j with the locus of the centres of a family of
Ex.
is
circles
having double contact with the
nple points at
...
,
.
raic.
*
Ex. 8. A 2 wic has P meets it in Qlt Q
a
,
<» and <a'. Any transversal through Qtn Show that the product Pp =PQl .PQt :...PQ, u
;
is
independent of the direction of the transversal with respect to we have
p'
[(i)
and that on inversion
P ,=k
in
.p P /p
Use polar coordinates,
(ii)
OP" Take OP as the
1
.
.
transversal.]
§4.
Pedal Curve.
If OY is the perpendicular from a fixed point on the of a given curve, the locus of Y is the tangent at any point pedal (first pedal) of the given curve with respect to 0. The pedal of the pedal is called the second pedal, the pedal of the second pedal is called the third pedal, and so on.
P
The angles between the radius vector and the tangent corresponding points of a curve and its pedal are equal.
«orf
For let the tangents at consecutive points P, J" of the curve meet at T, and let OY, OY' be the perpendiculars from on
these tangents (Fig. 2). Then since 0, Y' Y, T are concyclic, the angles OTY', OYY' are equal. But in the limit these angles are the angle and the and the tangent to the pedal at Y. angle between
,
OPY
on
OY
The pedal
is the envelope of the circles described
the radii
vectores as diameters.
For by the
at Y.
*
last
theorem the
circle
OPY
touches the pedal
n',
We now
proceed to find the Pliicker's numbers
m', 6
V
,
k,
4
XI
PEDAL CURVE
167
t, t', D' of the pedal, the Pliicker's numbers n, m, S, k, t, i, of the curve being supposed given. The pedal is the inverse of the polar reciprocal of the curve with respect to a circle whose centre is 0. For if Q is the pole of in Fig. 2 with respect to this circle, OYQ is a straight line and is constant. . OQ have then only to interchange and n, 8 and r, k and i in .the expressions for n', m', obtained in § 3. We have
D
PY
m
OY
We
. . .
= 2m, m' — n + 2m, (Y = m(m l) + r, k' = r' = 2m(2m7) + 4mw + 2cS = 3m + D' = D
n'
(
l, i
'
i'
/c,
\
a>,
Also. the pedal has multiple points of order
m
at 0,
and
at'.
As a
verification,
is
that there
we note that it is evident geometrically a branch of the pedal through corresponding
Pig. 2.
to the given curve. Hence to each tangent from is an mple point of the pedal. meets the pedal Again, any line through times at and times where it intersects the' perpendicular tangents of the given curve, so that n' = 2m, Also the tangents from to the pedal are the tangents at each counted twice and the perpendiculars from to the n asymptotes of the given curve, as will be evident from the first theorem of this section.
m
m
m
m
Hence m' = n + 2m. The same theorem will show that
to each bitangent of the
given curve corresponds a node of the pedal and to each inflexional tangent corresponds a cusp so that k' = i. The reader will readily prove that to each of the tangents from co to the given curve corresponds a tangent at co to the
;
m
4
:
168
PEDAL CURVE
XI
pedal, this tangent at «o and <oa> being harmonic conjugates with respect to Om and the tangent from co to the given curve.* Hence a> and a/ are mple points of the pedal ; so that
S'
= m(ml) + r.
r+ccos(6OC)=f(8). The pedal of a parabola with respect
.
Ex. 1. If the polar equation of the pedal of a curve is r=f(6), the polar equation of the pedal with respect to (c, oc) as pole is Ex.
to the
2.
vertex
is
a
cuspidal cubic.
[The pedal with respect to the focus is the tangent at the vertex use Ex. 1. Otherwise: Any tangent to y 1 = 4asc is x — ty + at' = 0, and the perpendicular from the origin is tx + y = 0. Now eliminate t.]
now
Ex. Ex.
3.
The pedal of a
1.]
circle is a limacon.
[Use Ex.
4.
The pedal of rm
—
m a cos mS
is
obtained by changing
its
m
into
m/{m + Y). Ex. 5. The pedal
quartic.
of a conic with respect to
to a focus of a curve
centre
is
a unicursal
Ex.
6.
The
line joining
is
bisected
by a singular
focus of the pedal.
foci of the pedal are the feet of the perpendiculars from on the lines joining the intersections of the curve with the circular lines through 0.
The ordinary
Ex.
7.
Any tangent
is
to the curve of Ex. 1
is
x cos a + y sin
a>
=/(<»).
The normal
— xsinu> + y cos a =/' (<u), and the corresponding normal to the evolute is X cos a + y sin a = — /" (<a). The coordinates of the point of contact of the tangent and of the centre
of curvature are
(cos
o)
./ — sin
<i>
./',
sin
o>
./
+
cos
to
./')
(  sin a ./' — cos a ./", cos to ./' — sin a ./"). and The radius of curvature is /(<») +/" (a). Ex. 8. Tangents are drawn in any direction to a given curve.
Show
that the algebraic
of the radii of curvature at the points of contact Show also that the centroid of the points of eontact coincides is zero. with the centroid of the corresponding centres of curvature, and is independent of the direction of the tangents.
[If the polar
sum
equation of the pedal
is
i
where u k
radii
... +« m = 0, homogeneous of degree k in cos 6 and sin 6, the sum of the d2 u — of curvature is by Ex. 7 —uu which is zero.]
r
m + rm1 u + rm <u + x i
is
^
*
Draw
at,
close to
a diagram in which the tangent to the given curve at construct Y, and proceed to the limit.
P
passes
5
XI
ORTHOPTIC LOCUS
169
Ex. 9. The sum of the perpendiculars from the point of Ex. 8 on the tangents drawn in any direction is zero, and is the centroid of the foci.
[We have u x = 0,
Ex. 11.
to
if
we take
as pole.]
its
Ex. 10. The deficiencies of a curve and
pedal are the same.
Show
that, if a curve touches ona at
O has
in general
H, its pedal with respect an asymptote perpendicular to OH, whose distance
from
is the same for all positions of 0. Discuss the Pliicker's numbers of the curve.
Ex. 12. Discuss the Pliicker's numbers of the pedal of a curve with multiple points at O, w, and a>'. Ex. 13. Find the Pliicker's numbers of the second, third, a curve.
[See Messenger of Math., July 1904, p. 50.]
...
pedals of
Ex.
i'th
14. If the curve a is the rth pedal of the curve 6, 6 is called the negative pedal of a. .Find the Pliicker's numbers of the first negative pedal of a given curve. [It is
the polar reciprocal of the inverse curve with respect to
is
0.]
Ex. 15. Show that the inverse to the rth positive pedal negative pedal of the inverse curve.
the
»"th
Ex. 16. The locus of the intersection of any tangent to a curve with the line through making a fixed angle with the tangent is similar to the pedal.
Ex. 17. Find the Pliicker's numbers of the locus of the centre of a circle passing through a given point and touching a given curve. [Invert with respect to 0.] Ex.
18.
Find the
Pliicker's
passes through a given point
similitude.]
numbers of the envelope of a circle which and whose centre lies on a given curve.
have
as centre of
[The envelope and the pedal with respect to
§ 5.
Orthoptic Locus.
The locus of the intersection of two perpendicular tangents of a curve is called its orthoptic locus.' If the coordinates used are rectangular Cartesian, we may obtain the equation of the orthoptic locus as follows. Let be the tangential equation of the curve, and let /(A, /j.) on be the tangential equation of a point Xx + fiy + 1 to the the orthoptic locus, so that two of the tangents from homogeneous curve are perpendicular. If we make /(A, /*) in A, fi by means of Xx + y.y + 1 0, the resulting equation in to the given X//jl gives the slope of the tangents from The product of two of the roots of this equation in curve.
—
=
R
R
=
=
—
R
— X/fi must therefore be —1. If we write down (Ch. I, § 11) the condition that this should be the case, we obtain a relation
5
170
ORTHOPTIC LOCUS
a)
XI
between
locus.
and y which
is
the pointequation of the orthoptic
We see at once that the orthoptic locus of an algebraic curve is algebraic. If the equation of the curve is given in parametral form, we can express the equations of the tangents at the points with parameters t and T in the form
*+/(% = <K0
and x+f{T)y
If these are perpendicular, /(£) .f(T) + and from the three equations just written we can express the coordinates x, y of any point on the orthoptic locus in terms of a single
= ^(T). 1=0;
parameter.*
If
RP, RQ
are perpendicular tangents to a curve, the circle
PRQ
touches the orthoptic locus at R.
Fig.
3.
For suppose R, R' are consecutive points on the orthoptic 3), while Rp, Rq and R'p, R'q are the perpendicular tangents from R and Rf. Then p, q, R, R' are concyclic. If we proceed now to the limit, p, q, R' approach P, Q, R reand the theorem is proved. spectively
locus (Fig.
;
Ex.
1.
Find the orthoptic locus of the following curves
Parabola,
:
(i)
(ii)
Central conic,
Circle,
(iii)
(iv) if
=
x3
.
(v) 3 (x
(vi)
+ y) = xK
i
x 1 y1 4a(xr + f) + l8a xy27a*
=
0.
»* It is usually advisable to express t, T in terms of u *= t + T, three equations then give x, y conveniently in terms of u or v.
v
=
tT.
The
u(A. (ix) 2a. y in terms of OC.x sin a + /cos a = (— l)*/{(2ifc + l)g + a}. (a? + y 2 ) (4a:2 + y 2 ) + 9 (2«y + 1) 2a. circle. Find the orthoptic locus of the hypocycloid. (xiii) (xyX)(2xyl)+\ =0.. (a. = csin»re(^. E . 2 3a^ + y 2 a.] § 6. t. (x) (xi) (xii) + y+3 = 0. 6a:(a:y)(2a:#) + 17a: 2 18a^ + = 0. 2 (v) (vi) (vii) (viii) 81 y* (a:2 +</ 2 ). to determine the Pliicker's numbers If of the orthoptic locus of a given curve with Pliicker's numbers n. circle. = 0. F (xii) X 6 + M 5 = XV. [Two perpendicular tangents may be written a: cos (X+j/sina = /(<x). 92/ 2 Uy*16xy2 + 4:X2 + 4:y2 4:Xl =0. i.u 4 = X. D. m. (xi) X 4 + 2. + M * = A (\_ F ) (2\p). Exi 6. t'. Pliicker's Numbers of Orthoptic Locus.] = am cosm8. 5. and F divide harmonically the line joining If two points We n'. 5 a +b = 0.2 . now i'. and b the radius of the rolling The orthoptic locus of a cardioid is a circle and a limacon.] [f (6) Ex. 5 5 2 (xiii) X + XV + h = X M (X .1 Find the directrix of the parabola < x = at 2 + 2bt + c. (iv) 729^ ! = 108* 16.] a being the radius of the fixed.). [(i) (ii) (iii) Straight line. Ex. 3. S.X 4 + 2 M4 = XV (x) (X»2 M»)* + */« = 0.36 2 a*/ + 5^) + 128 = 0. [In Ex. 2. where n — (m + \)/m. k.. (ix). Show how to find the orthoptic locus when the polar equation r =f(6) of a pedal is given. 4. [/(0) Find the 'orthoptic locus of the curve rm = a cos" 0/w. k'. (xy) 2 + 2{x + y) + 4: = 0. &.. y =At* + 2Bt+C. Ex. from which we get Ex. a. .s\ where c = o26 and OT = a/(a26).XI 6 PLUCKER'S NUMBERS OF ORTHOPTIC LOCUS (vii) (viii) 171 21(x i + y*) X4 = (2x+iy. proceed m'. + y + 3 = 0. x + y + 2a = 0.•). Concentric Concentric circle.
while the interperpendicular sections of an inflexional tangent with the In general there are no tangents are cusps of the locus. Now suppose any line whatever through co meets the given and Q. Again. It is readily seen that there is no other point of the orthoptic locus on coco' in general. or vice If this is the case. 172 PLUCKER'S NUMBERS OF ORTHOPTIC LOCUS XI6 and a tangent the circular points co and co'. is equal to the class of the envelope. while the tangent at to the orthoptic locus passes through co' and PQ through co. coco') are harmonic. we obtain the number of tangents from co to the envelope. m m m(m— R P . when becomes the point of contact of either tangent. Now there are 1) such pahs of tangents so that the orthoptic locus has a ^m(m — l)ple point at each of co and co'. coPQ will touch the versa. Suppose now E. coV) is harmonic. Proceeding to the limit we see that to each pair of tangents to the given curve from co corresponds a branch of the orthoptic locus through co.&n& therefore ^approaches E F R co. E F R tangents in question.. q become the points of contact P. m(m n = m(m — 1). a tangent from on from to the given curve are perpendicular and meet at the orthoptic locus. The line TV cannot pass through co unless coPQ touches the given curve at but not at Q. while the tangent to this branch and the line coco' are harmonic conjugates with respect to the pair of Then R . so that . Q of the perpendicular tangents RP. Now in general the envelope will not touch coco' and hence the number of tangents from co' to the orthoptic locus. harmonic also approaches co. 3 RR' passes through co'. coco') and R'ipq. It suffices to find the number of tangents which can be drawn to the locus from co'. other cusps. whose points of contact are not on coco'. while R(coco'. p. P . By what has just been said. Let the tangents at P and Q meet in T curve in and consider the envelope of the line TV where T(PQ. however. EF) remains become consecuand unless the tangents from tive. To find the remaining tangents from co'. suppose that in Fig. Proceeding to the limit. pq passes through co. every tangent from co' to the envelope which does not coincide with coco' will be a tangent to the orthoptic locus. so that k = mi. the theorem of § 5 shows that the intersections of perpendicular a bitangent of the given curve with the tangents are nodes of the orthoptic locus. To find this class. Of these tangents 1) coincide with the tangents at co'. We now proceed to the more difficult task of finding m'. Then since R(pq. RQ.
D' from Ch. «' = («» — &)i. and touches it at the feet of the m(m + n — 4:) normals which are also tangents. = im{(m + l)(m2) + 2T} *' = m<. &c. neither of which coincides with the tangent at the point. If in Ex. PQ) is harmonic.] There are \m (m + 1) (m — 2) (m — 3) points from which two pairs of mutually perpendicular tangents can be drawn to a given Ex. If cu<»' is a fcple = (m — h) (m — 1). If to the curve other than ww. the general cubic has an orthoptic locus of degree 30. YZ) is and Ex. segments joining the m real foci of Ex. We now n' 5' obtain r\ i. Jare cusps. m' = m(m + n — 3). = m(3m + /c6). Hence each tangent from co to the curve is an (n — 2)ple tangent to the envelope . [The curve meets the orthoptic locus at the points of contact of the tangents to the curve from o> and <o'. 3. curve of class m. 2 § 1 and see that = m(m— 1). the locus has (m— 2) (m3)ple points. so that the class of the envelope is (n — 2). For instance. If the curve touches ma' at and as (m — 3)ple points. n' 5. We have then m m' = m(«— 2) + m(m — 1) = m (m + n— 3). m'=(m — 2)(m + n — l). VIII. »»') is harmonic. the orthoptic locus simplifies materially. &>'. and the mr nodes Ex. k' = (m2) — 1). 2. There are m(m — 3) (n — 2) points of a curve from which two perpendicular tangents can be drawn to the curve. 4. 4 the curve has oxa' as inflexional tangent at Y. t' = im{m(m + n) (6m + 6mn + n )~m + 22 + 2S}. We have n = (m— 1) (m — 2). Discuss the cases in which Y. a> and o>' as harmonic. But as will be seen in the following examples. 6. m! = (m — h) (m + n — 3 — k). It also cuts o)a>' at one other point. Z)' = (ml)(m2) + mi) ! 2 2 2 i' In general the orthoptic locus of a curve is of relatively high degree. if tbe given curve is specialized by touching the line mm. Y Z Y Z Z . The singular foci of the orthoptic locus are the \) \m(m\) middle points of the \m. Ex.XI 6 PLUCKER'S NUMBERS OF ORTHOPTIC LOCUS 173 envelope at a point U such that (all. the tangents (m — 1 Jple point.] a. There remain m(m — 3) (n — 2) intersections of curve and locus. at to the locus are the tangents from Y Y Z Z Z Ex.(m — the curve. tangent to the curve. the as an orthoptic locus has a and to' as \ (m — \) (m — 2)ple points and is a cusp at which <aa' is a tangent. inflexions. tf. (i Ex. where (o>o>'. 7. [The nodes of the orthoptic locus other than derived from bitangents of the given curve. &c. the locus has (m — 2) linear branches touching wm at Z. If a curve touches <b<o' at and (YZ. 1.
m — 3. Ex. 4 (iii) has t»a>' as quadruple tangent.6 . Isoptie Locus. of the points of contact dividing a>a' harmonically. while n = 4. The locus of the intersection of which are inclined at a fixed angle a two tangents to a curve is called an isoptie locus. = 3. But this is not usually the case. 10. If bitangents of the curve pass through and = = (m + 1) (m — 2). The orthoptic locus is a straight line when the curve (i) is a parabola. dividing toto' harmonically. . while n (ii) has toto' as ordinary and 3 inflexional tangent. while n 6. is [The locus the orthoptic locus of the evolute. = m— oxo' harmonically. 11' m = m (m + » — 5) + 4. ic' k — to mi. 4 (iii) has toto' as inflexional tangent at two points =m= n = m= . Ex. 3 still holds. are distinct and both algebraic (being in fact concentric circles). two pairs of points of contact dividing Ex. (m + (m — 2). = m(m + n — 4). The orthoptic locus is a parabola . while n = 4. while n 3 (ii) has o>to' as triple tangent. f For the circle the loci of the intersection of two tangents cutting at angles <X or ira. two 4. implied in Fig. The orthoptic locus is a circle (ii) is a central conic. 1. Ex. while n = 8. For the parabola the two loci are the two branches of the same hyperbola. (ii) touches wto' in two points dividing toto' harmonically. 12. toto' — m= . 15. m= 5. curve (i) is a circle. Ex. m' = mi — 4. The and investigation of its properties is similar to that of §§ 5 The construction for the tangent to the isoptie locus 6.j 1 Ex. Ex. to 11 on the curves of § 5. The locus has no cusp. Ex. Ex.2) + mi. If inflexional tangents of the curve pass through m' 1) and to'. 14. while = 6. If the curve passes through w and a branch through to touching the curve at branches of order 2 at to. to. 11. Verify the results of Ex. m— 5. when the curve (i) has to<o' as inflexional tangent.f * The orthoptic locus is a cubic in fourteen cases. . (iii) touches toto' at when the to and to'. 9. = m (m + n — 3) — 4. The degree and class of the locus of the intersection of two perpendicular normals are (m — l)(m + n — 2) and (m — 1) (4m + K — 6). and a quartic in thirtyeight cases. The locus is an algebraic curve if we consider tangents cutting at an angle n — a as included among those which cut at an angle a. 174 PLUCKER'S NUMBERS OF ORTHOPTIC LOCUS XI a>'. 8. m Ex. We have n' the orthoptic locus has and m — 2 superlinear k to = m{m\). The orthoptic locus is a central conic when the curve (i) has as bitangent. the points of contact dividing toto' harmonically. ni' = 2 (m to'.* § 7. 13. 16. while n 5.
3. 2 On any line through there are n1 points such as n. Tangents EP.. . which is a multiple of 2ir/k.] 0. while the Pliicker's numbers of the cissoid are and n2 . Find the Pliicker's numbers of the envelope of a chord of a given curve subtending an angle of given magnitude at a given point 0. = P P Ex.. + y2  b2) 2 4 (b 2 x 2 + a 2 y 2 a 2 b 2 ). meet on a curve which is part of an isoptic locus and which is similar to tthe pedal with respect to 0. any line is drawn cutting Suppose that from a fixed point Take fixed curves 2j and 2 2 respectively in P 1 and P2 The locus a point P on the line such that OP = 0P1 —0P i of P is called the cissoid of 2j and 2 2 for the pole 0. Cissoid. t' = m{2m(m + — (8m + 8mw + m2)2m + 12 + 25}. 5. and at the 2»t(m + )i4) making an angle OC with the points P such that a line through touches the curve elsewhere. [Reciprocate with respect to 0. x . 1. 2 ta.] Ex.. Therefore the orthoptic locus is a straght line or concentric circle. 175 of an isoptic locus For the find we n' = 2m(ml). 6.n OC (x* x'/a? + y'/b2 2 = = 1. Show that the tangents inclined at an angle OL. Ex. The isoptic locus touches the curve at the 2 m points of contact of tangents from a and a (cuts if a %ir). and the normals at P and Q meet at S. neither touching at Q. /=2mi. m.] Ex. [The isoptic locus of the two curves taken together. the instantaneous centre of rotation of the rigid body PQE.. = m(2m3)(m — ml) + 2mr. m . . Ex. m' = m(2tt + 2m4). EQ to a curve are inclined at a constant angle. m . [S is Deduce the orthoptic locus of a parabola or central conic..a XI 8 Pliicker's CISSOID numbers n\ m!.1) + 2mD An isoptic locus has m(m— l)ple points at and 5 / 2 TC) 2 2 t' 2 a> «>'. less the isoptic loci of the two curves taken separately. Show that RS is the normal at R to the locus of R. 2. Suppose the Pliicker's numbers of 2 X and 2 2 are %. = 2m(3m + /f3). D' = (m .] Ex. . one drawn to each of two given curves. A curve has fcfold symmetry about at the ends of the radii through m § 8. two . .. Find the locus of the intersection of two tangents inclined at a constant angle. . The line RS is parallel to the axis or passes through the centre. 3 [y Find the isoptic loci of 1 and y — 4 ax — 4 ax = tan2 OC (x + a)2 and 4. It cuts the curve at the tangent at 2m(»3)()i2) points Q of the curve from which two tangents can be drawn inclined at an angle OC.
. 2 to an intersection of 2j to the cissoid . and so for the to is m 2 evidently tangents 22 . Hence k =n 1 K 2 + n2 K 1 . ^ apply Ex. and OP = k OP + k OPi + k3 OPs 1 2 . to the cissoid at count aa 2TOjTO g tangents from 0. Also the Uj^n^ tangents. Hence n l 2 + n2 2n1 n2 tangents can 1 + be drawn from to the cissoid. 4. been assumed that neither curve passes through the circular points. is a line and K. Ex.] [First Ex. =k = 2 £. 2 . Extend to the case of given . and so for 22 . 3 to the locus of Q. where OQ = k1 OP1 + k2 OP„ and then apply Ex. or m i m m=n m 1 ! + ti2 m1 + 2)i1 n2 . . 2. l P2 P3 . find the Plucker's numbers of the locus of P. The ra 2 linear branches touching 2 X at each of its Find the Plucker's numbers of the locus of P if in §8 OP = k1 OP1 + k1 OP2 where k x and k 2 are constants. 3. Prove that the'cissoid of algebraic curves . Ex. 1. [Consider the cissoid of the two curves obtained by increasing the radii vectores of 2 X and 2 2 in the ratio l/&j and — 1/&. Ex. where PPj. or if the curves meet at infinity ? . each of the »ij tangents from to 2j a w2 ple tangent to the cissoid.PjP. that all the intersections of the curves are &c.. Note the case k x Ex. the line joining evidently a tangent at cissoid. meets the cissoid at Moreover. It has finite. XI 8 line P is and n2 such as P2 x n n2 points other than 1 Hence the 0. What modification is necessary in the results of § 8.] cissoid has infinite points. are constants. Again. 3 to the loci of Q and Ps . k 2 k. so that and 2 2 is a n^ple point of the meets the cissoid in Therefore any line through 2%« points. 5. if either of the given curves passes through Q.] If a radius vector meets three given curves in P. is algebraic. N curves. [This was assumed in § 8. Also it is evident from a diagram that a line joining to a cusp of Sj passes through n 2 cusps of the cissoid. respectively. Hence n= 2%n.176 CISSOID .
one of which passes through a bicircular quartic with a node at 0. the tangent at P to the cissoid passes through Q. If on the radius vector OP of a curve we measure off PQ equal to +h. s?. Ex. 11. The cissoid of two circles. and collect terms involving an odd power of k on one side of the equation and the terms involving an even power of k on the Now square both sides. P to 2^ S Glt G and OG=OG OG 2 2 . P2 and in the perpendicular to OP through . transversal P. [If O is (0. + k„OP2 Ex. 12.] . the curve is y 1 (a—x) = The area between the circle and cissoid is something Hence the name cissoid from the Greek kiooos. Ex. 0). cissoid meet Prove that l 2 . is A Ex.9 XI Ex.. [The polar subnormal ' is ^ uu m polar coordinates. How many chords of a given curve through a given ratio at O ? Ex. 0) and A is (a. Turn the given equation into polars. 11 as axes of reference.] . we may obtain the equation of any conchoid as follows. Replace r by r + k.] Pliicker's numbers of the locus of P. [Take OP and OG in Ex. 6. We have the required equation of into Cartesian coordinates. find the cissoid of the line through perpendicular to OA and the circle on OA as diameter. where k is constant. the locus of Q is called a conchoid of the curve. For instance. Expand the powers of r + k which occur by the binomial theorem. 1 OP=k OP [See Ex. 8. A are two points. If the tangents at Pj P2 to' 2 X 2 2 meet at QT. and the G. circle with centre If we are given the Cartesian equation of a curve. T and Pj bisects § 9. It is the cissoid of the curve and the and radius k.] ' like an ivy leaf. are divided in How many chords of a given curve through line ? are divided in a given ratio by a given Ex. Conchoid. If 0. 7. CONCHOID Any 2 . and P OPP P 1 1 2 through Find the 3. the conchoid. 0. 9. The normals at P. '. and turn the equation back other. 177 meets a given curve in if . consider the case of a curve of degree 2n and and at the with multiple points of order n at class circular points co and a>\ m . 10.
manner we see that the conchoid has n cusps at eo whose taDgents are the tangents at a> to the given 2Jiic. There remain m(m— 1) nodes of the conchoid which lie at the middle : . is of class 2m.k 2 (x2 + y 2)} 2 ] . and so We for a>'. conchoid is 2m. m. eo' gives rise to two nodes of the conchoid. co. and is %(= (r a homogeneous function of degree #> in cob r^%) is the same function of x and y. which with the cusps of the conchoid at co and eo' make up 2k + 2u cusps in all. see tjjat the conchoid is of degree 4n with 2nj>\e points On finding the unreal asymptotes in the usual at 0. and where vp sin 6. Turning hack into Cartesians we have + y2 f + 3 Ul (x2 + y 2) 2 + 3u2 (x2 + y 2 ) + us] + 3k2 (x 2 + y 2 )(x 2 + y 2 + u1 )}2 = k 2 (x 2 + y 2 ){3 [(x2 + y 2 ) 2 + 2u l (x2 + y 2 ) + u2 4. and so for co. <o'. VIII. and has 2k + 2n cusps. each reckoned twice (as is obvious from a diagram).9 178 CONCHOID m= 2 XI Taking curve is 3 (the method (x 2 is general) the equation of the (x + y 2f + 81*! + ff + 3 u2 (x2 + y 2 + «s = ) °. while <o counts for 2m (n — 1) nodes (and m cusps). Since the conchoid is of degree 4m. Each of the k cusps of the 2nic gives two cusps of conchoid. and the 2n tangents at to Hence the class of the the conchoid each reckoned twice. or in polar coordinates r 3 + 3v1 r2 + 3vi r + vs = 0. These are accounted for as follows by Ch. each of the 8— f m (m — 1) nodes of the 2mic other than 0. Moreover. The tangents from to the conchoid are the m — 2 n tangents from to the 2nic not touching at 0. it has {4n(4»l)2m3(2/c + 2»)} nodes. is The conchoid or {(r 3 + k) 3 + 3v1 (r + lc) 2 + 3v2 (r + k) + v3 2 = + 3^r + Sv2 r+ v3) + 3k2 (r + vj} 2 = {[(x 2 k 2 {3(r 2 + 2v1 r + v2 + k 2 } 2 ) . § 4 counts for n (2m — 1) nodes.
the locus of Q is said to be parallel to the given All parallel curves have the same normals and the curve.] § 10. [{(x2 + y2 )(ax2 + 2hxy + by2 + 2gx + 2fy + c) + k2 (ax2 + 2hxy + by 2 )} 2 = ik2 (x2 + y2 ) (ax2 + 2 hxy + by2 + gx +fy Consider the case a = b = 1. Find the nature of and at infinity. [Use Savary's theorem on the centre of curvature of a roulette. 4. Hence the at P and Q PQ — If along the pedal of the parallel curve is the conchoid of the pedal of the given curve for any pole 0.* Ex. An raic does not pass through 0. What are the asymptotes ?] its Ex. 1. We proceed to find Plucker's numbers for any curve Suppose that the given parallel to a curve of given type. How many lines can be drawn through a given point on which two given curves intercept a segment of given length ? ' . in the case f But they may each be distinct * 2 The 2reic points. or a. and ra(2n + l) The 2«(m1) remaining intersections are the ends of the chords of with length 2ft.] Ex. meets the conchoid obtained by changing ft into 2ft in 8n 2 the above reasoning 2w lie at 0. same evolute. n2 . an »ple point at each of o> and <»'.] and a circle through 0. Find the conchoid of a [A limacon r = k + b cos <?.] Ex. for instance.] Ex. 2 . and a tacnode at each infinite point of the nic. ' 3. 5. 6. a>. It has a node at tacnode at infinity with the given line as tangent. the conchoid (x — a) 2 (x2 is ) + y2 = k 2 x2 . It is called the conchoid of Nicomedes '. [If the line is x ' = a. Find the conchoid of a straight line. 2. h = 0.f are parallel and at a distance k apart. at a'. Ex. the 2»ic through algebraic curves . Parallel Curves.) XI 10 PARALLEL CURVES 179 points of the n(n — 1) chords of the 2nic which pass through and are of length 2 k. [Consider the intersections of one curve with a conchoid of the other. n(2n + l) at a. conchoid at [The conchoid is a 4«ic with a 2 «ple point at 0. The centres of curvature at corresponding points of all possible conchoids of a given curve lie on a conic. Find the conchoid of the conic ' ax 2 + 2hxy + by2 + 2gx + 2fy + c = 0. The loci of the two possible positions of Q are The tangents in general parts of the same algebraic curve. By of a circle. normal at P to a given curve we measure +h.
m' = 2m. 0. whose tangential equation therefore is 'Ki+k^+n ^' 2 i + k(\*+n*)y . k' ' = {0} + {2c}. k. 8' = {m(6m5)} + {m(ml) + 2r}. m' = 2n + 2m. k. — is a tangent = 0. Here in the expressions for S'and k' the term in the first brackets { } refers to the multiple points at 0. co and a/ being mple points of the parallel curve. i. §9 m' = 4m. &>' k = {2m} + {2i}. m' = n + 2m. numbers of any other curve. i. cY = {0} + {m(ml)2T}. * That the class of the parallel curve is twice that of the original curve is also evident from the fact that to each tangent to the given curve in a given direction correspond two parallel tangents of the parallel curve at a distance k from it.1)} + {t}. co. to each inflexion of the given curve correspond two inflexions on the parallel curve. the tangents co. »> a> being mple points of this pedal. m' = 2in4m. and that numbers are n. co. XI 10 its Pliicker's curve does not pass through 0. i = 2i. l. r = m(m— 1) — 2r. then to a curve with tangential \x + fiy + 'l+k{\* + is ^= a tangent to the parallel curve. which we are considering. t. For this inverse Oco and Oco' are mple tangents. (Y.180 PARALLEL CURVES co. Similarly. m. or co'. For the conchoid of the first pedal we have by . For at ' are 2mple points. is the the curve parallel to the given curve. t. by ti'. e. this conchoid 0. If Xx + fiy + l \l) equation /(A. on'. co' and the term in the second brackets { } refers to multiple points else where and so in what follows. *' 8' = = {0} + R { m (m . + 2m. coinciding in pairs. 8. of the inverse The polar reciprocal with respect to first negative pedal of the conchoid. For the parallel curve we have fiually n' = 2<n. We shall denote the Pliicker's for the moment By § 4 n' for the pedal with respect to = 2 m. co' For the inverse of this conchoid we have by § 3 n' = 2m.
lie Ex. Ex. 4.] with centre moves in its own plane so that Ex. parallel to the parabola y* 1! Find the tangential and parametral equations of curves = 4 ax. and draw them. 0.] Ex. b>k>Wfa. a given curve. ABCD D D . I k = b*/a. a>k>b. Find the envelope of a family of circles whose centres a given curve and which touch a given circle. a node where (k + 2 a) t* = k — la. focus being the focus of the parabola. ' y~\p~ _4a^ 2U_ l+t*' Changing the sign of k is equivalent to changing t into — 1/t. Distinguish the cases k>2a. 18 and § 10. § 8 (IV). no inflexion. Show the lines through trace traces out an isoptic locus of the curve while A. l! . properties of The parallel curve can readily be shown to be the envelope of a circle of radius k whose centre lies on the given curve. Show curves. k > a /b. A. the conies of closest contact being (y±k) = 4 ax. C. Ex. This gives in practice a convenient method of drawing curves parallel to Ex.!). and six cusps where at (oo. It has degree 6 and class. The curve is the reciprocal of the quartic of Ch. k = b. k = 2 a. It' passes through the circular points. We leave to the reader the investigation of the parallel curves by this means. . Show that their new envelope radii of a singly infinite family of circles are all increased is parallel to their old one. >k>a. x 4 at 2 kjlt 1 ) (l« 2 ) 2+ \ + fi. k < 2 a. It has also 2a(l+?) s It has = k(lt 2 s ) . ! i Ex. a?/b [{(« 4!)\1 +(b 3. 6. Find the tangential equation of curves parallel x*/a* + y*/b 2 to the ellipse = 1. (\ '+/. 2 2 [(a M X) 2 = i \ (\ + A ! ). if 2 tan. B. It has two linear branches osculating each other 2 0).] Ex. on [Combine § 4.XI 10 PARALLEL CURVES 181 This equation is rationalized by a process similar to that used to obtain the polar equation of a conchoid in § 9. that out the isoptic locus of a parallel curve . i> lf = 4i.1 * is the angle which a tangent to the parabola makes with the tangent at the vertex. XVII. 2. its singular The by a constant. h = a*/b. [Use the tangential equation. that the foci of a curve are singular foci of all parallel 1. 4. Discuss similarly ay 1 — X s and a'y = x s . B. . A rhombus parallel to the sides both touch a fixed curve. 5. Draw the parallel curves distinguishing the cases h<b}/a. Also. C. and that the normals at to the two isoptic loci are concurrent.] 2 ' h=a. 4.
§ 1 will be found in the following examples. n' m' = 2» — 2 i' = = m + 2n4k. and has an mple point at a and a>'.] We see that the Ex. a and <u'. Find the Plucker's numbers of the locus of the centre of a touching a given curve and orthogonal to a given circle with is centre O. . The tangents from to the envelope are n bitangents given by taking P at infinity and the 2 m perpendiculars from to the 2 m common tangents of j and s. are the normals at Q x and Q 2 to the envelope. 1 to 4. How many circles can be drawn orthogonal to a given circle and bitangent to or osculating a given curve ? [Such a circle is given by each node or cusp of the locus of Ex. the locus of the points Qlt Q2 on the perpendicular tangent at any point P of s. locus has the same Plucker's numbers as the reciprocal of the given curve. and those of the derived curve under consideration by n'. &.* may of even degree n. t. m. Many other cases of derived curves might be given. * If the curve has a *>ple point at each of ft. To each tangent from to the given curve corresponds an intersection of the locus with am. which will be found useful later on. Hence «' = m. i. But those already discussed must suffice. consider other > meets the given curve corresponds a tangent to the locus perpendicular to the line. The attention of the reader is especially called to Ex. Each bitangent of s gives two nodes of the envelope and each inflexion gives two cusps. k. m'. [We consider the case in which the given curve having a »ple point at w and a. The reader cases. 2(3ot + k). The foci of s are the singular foci of the envelope. and the 2 n intersections of ^ and s are ordinary foci of the envelope. The circle with centre Y and radius equal to the tangents YL and YL' from Ytoj passes through Q1 and Qt while PQX andPQ. k. The Plucker's numbers of the given curve are denoted in the examples by n. P and Bare is The envelope OY from to the . Further illustrations of the principle at the end of. r. = m(m — l) + 2i. numbers of the envelope of a circle with its and cutting orthogonally a given circle j 8' i' t' [n' = 2m. m' = 2(n + m). 8. = n> + 4:mn + 2m?l0m + 28. i Ex. If OR is perpendicular to LL'. 3. Hence mf = ». k. = = (ml) + 2X>. which are equidistant from r"and inverse forj.. XI 11 Other derived Curves. D' k' = 2i'. 2. 182 OTHER DERIVED CURVES § 11. The envelope is selfinverse with respect toj. Find the Plucker's centre on a given curve s whose centre is 0. circle 1. 1.] Ex. in To each of the n points which a given line through Hence i = k. To each cusp of the given curve corresponds an inflexion of the locus.
facts enable us to construct any number of points on the envelope. and the others by the segments in question. Prove also m' = 7 m — 3 n + 3 k. and has a node at each intersection of the given line with the curve.2) 8.] . The cusps and points of contact of tangents from a> and <u' [3 m + k. [»' = 2«i. [(i) The envelope cuts OH orthogonally at and has an asymptote perpendicular to OH twice as far from O as the corresponding asymptote of the pedal of s with respect to 0. What modification must be made in the results of Ex. (iii) So> 5.. /*) +/3 (X. the envelope is 0=02/! (*. Find the degree of the radial of a given curve .] Ex. M ) +/. e.] Ex. 4. homogeneous of degree > in X and /i. See § 4.. If j is x* + y 2 = k and s has the tangential equation where /. [n' Find the locus of the middle point of a chord of a given curve fixed direction. Of these nodes of the locus (n — 2)8 are given by the nodes of the given Mic. meets the give branches of the radial through 0. /*) is = a+A (X. Find the Pliicker's numbers of the locus of the middle point of the segment intercepted by two fixed lines on any tangent to a given curve.2) (» . 9. y)/(a? *)/(*• + 2 i/ + fc) 2 (^y)/(a 2 + 2/ 2 + ^) 3 + a. Ex. 11. 5 8' =  n (» . 6. and Sa' are inflexional tangents of the envelope. i. D' = D. k' = (n — 2) k. the equation of + f + h) + 4/s (or. = Ex. (X. Find the degree of the locus of the end of the polar subnormal of a given curve.1) (n .? are drawn. The locus meets the given line at its intersections with the tangents to the curve from a and a.] = \n(n — 1). (X. k = n.] Ex.3) + (n . 10. the locus of the end of the line through a fixed point O parallel and equal to the radius of curvature at each point of the curve. 8.] ! 8/3 Ex. 7.. [2(m + n). 3 if the curve s (i) touches aa>' at (ii) goes through a> and ra'.] • Ex. drawn in a Consider the infinite m' (» — 2) m. Consider the intersections of the locus with the line at [2n + m. points on the locus and the tangents in the fixed direction.] Ex. infinity..) + . when either s or its polar reciprocal with respect to. (iii) touches the tangents from a focus S at two points on j ? H . f. «' = t. Any line through radial in m more points. (ii) The tangents at <o and a to s are cuspidal tangents of the envelope. XI 11 OTHEK DERIVED CURVES 183 These corresponding points on s and its polar reciprocal with respect to j. m' = 2 m + n. How many lines a given nic intercepts middle point ? \n can be drawn in a given direction on which (n — 1) segments two of which have the same [In Ex. Find the degree of the locus of the centre of a circle touching a given curve and a given straight line.
] Ex.] . If P is any variable point on the first polar curve of a fixed point with respect to a given curve. in a given ratio. R. 11. 14. Find where the polar line meets the locus again. replaced by the normal at P in Ex. Ex. Discuss the locus of a point from which equal tangents can be drawn to two different curves. Find X [It is easy to find the intersections of the loci with the line at infinity. the tangent at P is • Ex.(»» Their degrees are + 2n6) and 2 («2) ii (»3). 6. [The circular points are \m(m — l)ple points of the locus.184 OTHER DERIVED CURVES it XI (ii) 11 Ex.] if Ex. 12. 13. Find the degree of the locus of X. the locus of intersection of the polar line of P and the polar of P with respect to a fixed conic is a curve of degree + 8 + «. m [The polar line of P passes through which is a (» — l)ple point of the locus. and for the other points at infinity see § 2. X X PQ again in Q. The tangent at P to a curve meets the degree of the locus of if (i) divides divides QR in a given ratio. 11. «. Find the degree of the locus of the intersection of two equal tangents to a curve.
of a curve of degree n contains y. (®r» Vr> z >) The curve will pass through them f{xv yv gj = 0.. Suppose we are given r points (*1» Vl> Z l)> ( XV . In this case an infinite number of Jiics will pass through the ^ n (n + 3) given points.. z) y. point. . Next take n = 2. namely the line through the four points and any line through the fifth . and only one curve of degree n passes given points (see Ch. . that the equations (i) are not independent. 2/2.CHAPTER XII INTERSECTIONS OF CURVES § 1. y. = 0. §6). Equations (i) have one and only one solution for one and only one straight line passes through two given points. equation f(x.+2)l = %n(n + 3) coefficients. They will determine the J(«. = one term of zero degree in x and n+l contains %(n + 1) (n + 2) coefficients. if independent ratios of the r = \ n (n + 3). the conic through the points is degenerate.. 1) (n + 2) coefficients + l)(«. two terms of the first terms of the nth degree. First take n = 1. . Hence f(x. fix. f(xr y r z r ) .. there are a singly infinite number of conies through the five points. In this case \ n (n + 3) = 2. z). being the line through these three points and the line joining the other two points. II. z) The degree. however. . (i) These are linear equations in the ^(n + of f(x. Conditions determining a Curve. If four of the given points' are collinear. Let us consider two special cases. . y. . If no four of the five assigned points are collinear. . Z 2 )> •• if . one and only one conic goes through them. Hence through It in general one ^n(n + 3) may happen.. If three of the five points are collinear. y2 z2 ) = 0. Here %n(n + 3) = 5..
In general one and only one cubic passes through nine given points. z) linearly in terms of %n(n + 3) — r arbitrary independent quantities.: 186 CONDITIONS DETERMINING A CURVE XII 1 Lastly. take the case n = 3. . Therefore the cubic passes through the nine fixed intersections of u and v eight of them being the eight given points. . w = s are s fixed nics through the given points. . [These are the only values of p and q such that \p(p + S)>pq § 2. there are a infinite number of conies through the five points. k1 w1 +. . Suppose eight points on the cubic are given. fixed point.. and k is We now a parameter not involving x. the ratios of the coefficients can be expressed linearly in terms of J to (« + 3)— s quantities. If only s of equations (i) are independent. if all five given points are collinear. .. 2 ai. . p = 2 and 2=1. + ks ws = . we must have p = q «= 1. It is necessary now to consider whether there is any exception to the statement that all cubics through eight points pass through a ninth. Here \ n (n + 3) = 9.y. 1 . so that an {%n(n + 3) — r}ply infinite number of nics pass through the given r points in general. Hence All cubics through eight fixed points pass through a ninth = = . where u = and v = are fixed cubics. z. or p = q = 2. kt are arbitrary constants. . namely the line through the points and any other line doubly whatever. The coefficients of its equation can in general be expressed linearly in terms of a single quantity. The argument which established this result * In this case any two of the nine points are said to be conjugate to one another with respect to the other seven. where kv k2 .A. Hence the equation is of the form u+kv = 0. and iq(q+ 3)2s_P2] Cubics through Eight Points. The equation of the nic is then If of the form r< %n(n + 3). . p = 1 and q = 2. an infinite number of cubics pass through the nine given points.2 + . But if the nine points are the intersections of two cubics. If p and q are positive integers such that a pic and a qio can be described through any pq arbitrary points. coefficients equations (i) express the ratios of the otf(x.* §3. and w = 0. Ex. y. w% = 0.
2 They therefore pass through a ninth point. which are such that any cubic through seven of them passes through the eighth. ABODE goes through F. the linetrio (2) (3) the linetrio C^. GH F . then A 1 A 2 A 3 and A^^. G2 pass through the lt 3 2. The case in which three or more of the given points are collinear is dealt with in a similar manner. Hence C3 lies on cubic (2). The equation of the cubics will involve not but two or more parameters in this case and the pre ceding proof breaks down in consequence. But Hence the conic does not lie on GH. 1). . It will be found that the eight points lie on one or other of two straight lines. But C3 cannot lie on the lines A 1 A 2 A 3 or B1 B2 B3 since neither of them meets the cubic in four points. But cubics (1) and (3) pass through C3 . If two straight lines meet a cubic in Bi. = not independent. Hence all cubics through eight given points always pass through a ninth fixed point.more than six points. G C2 X (Fig. . . . For the three cubics the given cubic. First suppose no three of the eight points collinear. points. 1 2 3 very close to one another. and similarly through G and H. . lines Taking the A A A BBB x 2 3 . r 8) are = . 2. B2 C2 A B3 C 2 A A 3 Blt B B eight points A A^G^ A X 2 3 . F. D. H. The eight given points lie on a conic. one (or both) of which is part of every cubic through the eight . eight points are such that all cubics through any seven of them necessarily pass through the eighth which implies that the equations (i) of § 1 (n 3. C&. . . Therefore G3 lies on . 3 3 . Let us now 3 . . one. A 2 B2 A 3 BS meet the cubic Clt G2 C3 are collinear. and may under special circumstances pass through an infinite number of other fixed points. XII 4 CUBICS THROUGH EIGHT POINTS 187 will not be valid if the. Let the eight points then be A. consider some applications of the theorem of § . while the lines again in Glt (1) C2 C3 . . The cubic consisting of the conic ABODE and the line passes through seven of the points and therefore through the eighth point F. A A A B^B^. E.. G. . B B 2 . and any cubic through them consists of this eonic and some straight line for a nondegenerate cubic cannot meet a conic in. B. G.
3. while and and FA at N. 7 .: : 188 CUBICS THROUGH EIGHT POINTS XII 4 and defining the tangential of a point 1 on a cubic as the point in which the tangent at A meets the curve again. CD. The finite intersections of a cubic with its asymptotes are collinear. If L. 3. § 5. II. its intersections with the line at infinity. BC. we have The tangentials of three collinear points of a cubic are collinear. 1. § 3. C. . Ch. 9 . then L. of course.* Fig. FA.] Ex. Ex.] [This is a particular case of the collinearity of tangentials of collinear lie. The points of contact the inflexion. CD points on a conic. Ex. This is Pascal's wellknown theorem. on the harmonic polar of Ex. (2) linetrio AB. BC and EF F are at M. M. l. meet at collinear. § 4. 2. DE. The points of contact of the tangents from an inflexion of a cubic are collinear. Ex. D. II. are AB N DE trio [Consider (1) conic and line LM. Suppose now A y and A 2 are inflexions of the cubic. E. while and B1 B2 B3 are close to one another. and we have in the limit The line joining two inflexions of a cubic passes through A A A l X another inflexion.] * EF. points. (3) line See also § 5. Cli. Then G G2 is 2 3 close to them. Ex. B. [They are the tangentials of See Ch. A. VH.
each doubly infinite in triple contact with a given nonsingular cubic. 1 is a A EF meet the cubic again at three [The second cubic of Ex. AB. 6 take E and F at the circular points.] . Q. Ex. Show number.] that the tangents at Q. a cubic at P. C and D Show Ex. D. Discuss the number of such sextactic points the cases of a nonsingular.] Show that the tangentials of P. or ousgidal cubic. having that there are three families of conies. M".] Ex. PD. C. EFN. 6. A conic has fourpoint contact with a cubic at A and meets and F. [See Ex.] Ex. If two cubics have the same asymptotes. is the line at infinity twice over. m [A particular case of Ex. C. If is meets the cubic again in N. on a conic. E. Show that PQ passes through an [See Ex. 6. B. [The conic of Ex. 13. N. PB. (3) linetrio PA. The number of points is 27. their finite intersections are collinear.] A circle meets a chcular cubic in the finite points A. The sextactic points of a cubic (points at which a conic Has sixpoint contact) are the points of contact of the tangents from the inflexions. that AB and CD meet at Q on the curve. The theorems in § 4 and many of these examples are particular cases of this result. A conic osculates a cubic at inflexion. 3. 6. Q. [Consider (1) the given cubic. C. 9. 8. CD. Ex. 4. When they are chosen. 7. 10. LM is parallel to the real asymptote of the cubic. is conic meets a cubic at A. Then there are three possible positions for R. If six of the intersections of two cubics three are collinear. Q. R.] P Ex. A conic touches B are collinear. PB. Discuss the case of a unicursal cubic.] E EF Ex. R are known. the other [Consider the two given cubics and the cubic consisting of the conic and the line joining two of the remaining intersections. CD. 0. [The tangentials of A. B are collinear. B. PC. 6. 14. Show that collinear points L. The tangents from a point P of a cubic are PA. CDM. 5. excluding the case in which P. 6. the linetrio Q. 12. 8. If AB and CD meet the cubic again in L and M. Enunciate the theorem obtained by respectively consecutive in Ex.] Ex. D consecutive in Ex.] Ex. [In Ex.XII 4 CUBICS THROUGH EIGHT POINTS lie 189 Ex. B. R are collinear. making A and B. the tangentials of P. ABL. P and Q. F. 11. and Q may be chosen arbitrarily. 13 meet on the curve. ' Show P and Q in Ex. D. the curve again at the tangential of the tangential of A. nodal. N [Take A. Ex. (2) polar conic of P and line AB. 4 Ex. 4 particular case.
B on the curve and PA. P. DSA. E.] ABE. with respect to the cubic through ' Let A. C. (3) linetrio PQT. BBC. (ii) two Ex. 17. be put into homographic correspondence with the lines through so that the lines HE. Q'. The point is called the point opposite to with respect to the cubic] A. ADF. • [Take C. Prove that QB is divided harmonically by the cubic and its point of contact with its envelope. F. C. correspond respectively to the conies ABCDE. 22. G. 17. BC. EF. 20. BS of a cubic meet at on the curve. Chords PQ. H. G. [Let ABC be any three fixed points of the cubic. B. Prove that Q'B and QR' meet on the cubic. S and any given point of the curve as vertex. they meet again on FG. Show that. Show that the tangents at A and B meet on the cubic. B. and Consider the cubics (1) given cubic. FG. if they touch at A. .4 190 CUBICS THROUGH EIGHT POINTS XII Ex. The tangents at the four points of contact all meet on the cubic.] 19. they osculate there while. P'A. P'B meet the curve again at Q. Ex. A conic passes through four fixed points A. (3) conic ABCDE'F' and line EF. Q. 16. C. and the circle of curvature at B passes through A. C. ACG. Two cubics go through the vertices A. 18. B. B'. [See Ex. The circle of curvature at A to a circular cubic passes through B. A variable point Pon a cubic is joined to two fixed points of the curve. 20. ECD. If any line through the conic ABCPQ meets the 3ic again in a (Ex. (2) linetrio SET. T DA A [Consider (1) given cubic. 17) and the line OPQ and the conic ABCDPQ have fixed point a onetoone correspondence. PB. Let the line through corresponding in the D HG H D H . 17.] D ABCDEF F Ex. that the line joining the other two intersections of the conic and cubic meets the cubic again at a fixed point. [Consider (i) two cubics and linetrio cubics and linetrio EBA. D. CD. CQD. D of a quadrangle and through the diagonal points E. B. C. meet the cubic again in E. 23. going through P. of a cubic. C. 21. D Ex. meets the D 3ic in 0.] Ex. Show that a quadrilateral can be inscribed in the cubic with the sides AB. Q. the conies through A. if they touch at E. S' through A. cubic the locus of the intersections of a pencil of conies with a homographic pencil of lines whose vertex is any given point is Any of the cubic. D Show E'. ABCDG. Through four given points of a cubic four conies can be drawn touching the cubic at some other point. [Let conies S. ABCDF. APB. Prove the accuracy of the following ruler construction for the point opposite to A. F'. B. [See Ex. D in Ex. and the joining lines meet the curve again in Q and JR. B. P' on a cubic are joined to points A. C. Two points P. I. B. (2) conic and line E'F'.] Ex. [Make P' consecutive to P in Ex. F. B. . 15.] Ex.] Ex. B. 17 at the circular points.
1 . Show that if and Q. C. 3. The theorem that all cubics through eight given points pass through a ninth fixed point can be extended to curves of higher degree in the form In general all curves of the nth degree through \ n (n + 3) — 1 fixed points pass through \ (n — 1) (n — 2) other fixed points. Deduce Ch.§3. i. the lines B3 Z)„ and D A A B B C^C D D B Cs and C D and D 1 2 . The theorem is only true 'in general' if ti>3. 7 hold good in the examples given in these sections but it must be admitted that this assumption is hardly rigorous without further inves. t lie on a conic. and t . We shall assume that the theorems of §§ 5. D. 2. where k is a parameter . 1. Ex. 4 are special cases of this result. 10 from Ex.1 + 1 (w . [Reciprocate and use Ex. 1. 4 . are the points opposite to A. H. Ex. and the curve and v 0.' Ex. Clt C2 . D lt X> 2 . the conies ABCDa and EFGHO meet OS at the ninth point of all cubics thvough A. 24. quartic. Ex. . we have tigation. Four normals are drawn to a conic from a point. . § 5.2) = TO = = = . . common Deduce a Ex. (2) the lines A X A 2 B t B 2 CY C2 DjD2 (3) the given conic and the conic A & Bs Cs S A A are three quartics through the same thirteen points. A3 and . For in general by § 1 any such curve has an equation of the form u + kv 0.] Ex.1) (« . l 2 meet the quartic again t . and from the centres of curvature at the feet of the normals two more normals are drawn. Then IT meets 2 again at 0. 3. Ex. E. D and to E. F. B. II. A t . C.: XII 5 INTERSECTIONS OF TWO mICS 191 homography E. F. H touching at H the line to the conic ABCDI meet in I' the conic 2 through corresponding to the conic ABCDH. G. construct by ruler the intersection of the cubic through them with the line through two of the points or the conic through five. G. . Show that these eight normals touch a conic. ruler construction for the ninth intersection of all cubics through eight given points. . Therefore they have three more points in common. Given nine points. conic meets a quartic at 1 2 . passes through all the n 2 intersections of u Noting that 2 •§*l (% + 3) . 2. 2 . Intersections of Two nioa. 6. G. 8. Ex. 25.e. D . s Show that A A s At B3 Bt Cs C at . the required result. . JB ? . 2 . 2. 4. 9.] [(1) The given . A A B 1 . The tangents at four collinear points of a quartic meet the curve again in eight points on a conic. F. H. only if the given points are such that not all nics through \n (n + 3) — 2 of them pass through the remaining point. B. t . .
5. C.] through (n — Ex. (2) n lines. 2. D. 6. 11. (3) given and cubic through nine of the twelve points. See Ch. (3) line of collinear points and (n — l)ic through l(nl)(n + 2) of the n(n — 1) points. More generally. 9. two quartettes of lines meet in sixteen points of which three lie on another line. § 6. A straight line meets a quartic at A. n'' + 2m — 1 finite points pass Show that the theorem of § 5 is not always true when n > 3. If a conic passes through the points of contact of a bitangent of a quartic. A n— 1 inflexions of an nic passes through an [Take the conic of Ex. line D [Consider the quartics (1) given quartic. /= twice at each node and thrice at each cusp. I. the remaining n (n —p) lie on an (n—p)ic. meet the quartic again lie on a cubic.] Ex. [Consider the mics (1) given nic. which meets 2. nth inflexion. (2) pair of conies. and [On (#£)_£ dy = (yij) _Z ox .192 INTERSECTIONS OF TWO rICS XII 5 Ex. the directions of whose asymptotes are the axial directions of / = 0. B. 8. Generalize Ex.] Ex. 12. The feet of the normals to a curve f{x. VII. Any nic through the feet of \n (n + 3)l normals from to a given nic passes through the foot of every normal from (n 2). (n 7. 5. The result also follows from Ch. 1 as a special case. This includes Ex. the line twice over and the n lines as the tangents at its intersections with the nic.] drawn. conic meets an nic in 2 m points. (3) conic and (n — 2)ic. These are divided into lines joining each pair meet the nic again in n(n — 2) points on an (« — 2)ic. Ex. The number of such normals is m + n. 9. [Consider the nics (1) given nic. 13.] pairs. Ex. All ^circular 2«ics through 2 l)' other fixed finite points. y) = of degree n class m from a point (£. if np of the riL intersections of two nics lie on a pic.] Ex. [Limiting case of Ex. a cubic will touch the quartic at its six other intersections with the quartic. 13] Ex. general. A Show that the n line through Ex. For the pic and an (n p)ic through ^(n—p)(n—p + S) of the remaining points form an nic (degenerate) through In — np + %(np)(np + 3) = %n(n + 3)l+l(pl)(p — 2) . (2) n lines. B and a conic through C. The case n = 3 is well known. § Ex. rf) lie on an nic through O and the pples of the line at infinity. > §6. 8 as. Show that the twelve points in which a conic through A. Through each of n collinear points of an nic a straight line is Show that they meet the nic again in n(n — 1) points on an — l)ic. 10. [If Ex. it may be shown to be untrue for the thirteen other intersections of any two 4ics through three collinear points. the theorem is readily seen to be not true for 4ics through the other thirteen points.
A conic through. 2. R'. § 7. C. D. 2. Show that their eight other intersections with the quartic lie on a conic. 2216 subjected. § 2. nN intersections of an nic and Nic will pass through the N= (r — iV")(r — iV+3) also arbitrary points on the nic. Take nN%(n + NrI)(n + Fr2) and JVic. P'. Ex. Take %(r— n)(r—n + 3) arbitrary points on the JVic and intersections. Ex. If a conic meets a quartic in eight points and a conic touches the quartic at four of the points. r>JV. Two conies are drawn through the points of contact of two bitangents of a quartic. 1. 4. Obtain other theorems as special cases. 5. is of the intersections of nic * For a discussion of limitations to which this result Bacharach. Q'. Ex. But none of them can lie on the pic. Note the cases n = 3 or 4.] Ex. A polygon of 2ra sides is inscribed in a conic. [See also Ch. Annalen xxvi (1886). Any conic through P. other than the vertices of the polygon. r^n + N— 3. C.] : Ex. Q. D. B. XIX. Any quartic through the intersections of a cubic and quartic meets the quartic again in four collinear points. Hence they must all lie on the (n —p)ic. Hence by § 5 all the remaining points lie on this degenerate nic. An In of the general extension of the theorem of § 5 is the following any ric through all but i{n + Nrl)(n + Nr2) remaining If n = i r^n. [Consider the two raics formed by taking every alternate side of the polygon 2ra of their intersections lie on a conic. Ex.* we have the theorem of § 5. a conic touches the quartic at the other four.] Ex. lie on an (» — 2)ic.the points of contact of two bitangents of a quartic meets the curve again in A. A conic meets a quartic in P. Q. S'. S' meet the quartic again in eight points on a conic. S. 275299. B. 6. R. Q'. 7.: XII 7 INTERSECTIONS OF ANY TWO CURVES 193 of the n 2 intersections of the two given nics. 3. Show that a conic can be drawn touching the quartic at A. S and any conic through P'. Ex. see . pp. for the pic cannot meet the nic in more than np points unless the nic degenerates. Intersections of any Two Curves. R'. provided r. R. which we suppose not to be the case. Math. Any quartic through the intersections of a conic and quartic meets the quartic again in eight points on a conic. The intersections of the sides. 3. [This and the following examples are particular cases of Ex.
4. B'. 1. Then (i) shows that the curve X passes through L. 6. intersection of the curves and n be the points together with other points which we call collectively ' the points L '. JVic is the quartic and the conic through the points. points. . 2 as the polar conic of 0. points which we call collectively ' the points 0'. The tangents at the six intersections of a conic and cubic meet the cubic again in six points on a conic. when there can be no . D. [wic is given cubic. meet the curve again See Ch. Gn Gv to be such that CiCx + Cm % + On Cr = where l + \ = m + = n + v. . C. m Let the complete Suppose that the the the the curves I and meet in the points and N~. Ex. curves A and meet in the points L and M'. The nic together with the — n)io through the J(r — n)(r — n + 3) points form a degenerate ric through the \ r (r + 3) — 1 points and so for the N'ic together with the (rN)ia through the  (r — N~) (r — iV+3) points. FF' meet the cubic again in six points on a conic. E. .] § 8. curves A and n meet in the points L and N'. CC. B. Therefore every ?'ic through the fr(r + 3) — 1 points passes through every intersection of this pair of degenerate rics which proves the result. I. ric is A other cubic and the line through two of the three. C. denotes a homogeneous expression of The curve Gr = is a curve of degree r. the lines AA'. F'. If a quartic meets a cubic in A. y. D\ E'. z. A'. F. DD'. number . n all pass through G^. Theory of Residuals. ' Suppose that degree r in Gr x. '. the curve r r. Suppose that the curves m. 2. curves I and n meet in the points and M.194 INTERSECTIONS OF ANY total TWO CURVES will be XII 7 The number of points taken (r found to be r(r + 3)l.] five of Ex. Ex. BB'. [«ic is cubic. 3. CA Gm (i). EE'.] quartic through the intersections of two cubics meets either cubic again in three collinear points. [Take the quartic in Ex. I. § 9. The tangents from a point six points on a conic. Ex. ric is the six lines. in [Take the conic of Ex. confusion with the . . JVic is quartic. m m .] to a cubic Ex. which we Suppose shall call Cj. fj. 1 as a pair of adjacent conies.
v respectively. Now by (i) the curves I and m. fi. and so on. we obtain relations between the curves and their . the complete intersection of any two curves represented by faces of the cube (not opposite) is the points whose symbols lie at the ends of the edge common to those faces. x = 1. Proceeding thus. z = 0. o2 . &c. n are so that the curve v passes through the points N. complete intersection of the curves m O'L' of u and v. y = 1. m. Suppose they meet in points on I and 0' on X. MN' of /x and n. *x Each curve passes through the points whose symbols lie at the corners of the face representing that curve. meet in points each of which lies on one of the curves n or v. Thus the curve I passes through the points OL' MN. z = 1 represent the curves I. Thus OL is the and n. the curve X passes through the points O'LM'N'. U .XII 8 THEORY OF RESIDUALS /j.intersections which will be clear from Fig. \. 195 By (i) the curves and v meet in points each of which lies on one of the curves I or \. But the only points common to the curves I. In this diagram is shown a cube with unit edge whose laces x — 0. y = 0. n. Moreover.m. 2.
e. [Tf CN = and Cn = 0. form taken all together the complete intersection of Cn with a given curve. Show that these three (nN)ica form a pencil.ATic three raics are drawn. n Cv are given polynomials such that l + \ is not and n. yjr of 0. Then use </>. = = = = = We = = M = = = M = M = = = = = + M= Similarly O. yjf be the given wics become when we substitute in them the coordinates /„. but at no other point of the given n. 1. 2. Through the remaining intersections of each pair of nics is drawn an (nN)ic.f i.given point three given nics taken in pairs form a pencil. — — XII 8 Ex. Q. what /. = with O and L o. the three «ics are Q. O+L since = 0. The »ics through any . meaning that. As in § 8 we may consider and as the complete intersection of Ci and Cn 0. O and L are residual. . Cn +CN T = and the(wJV)ics are B = r. ^ M<tA^A) +0oW'o//o<M +^o(fAi>of) = o] Ex. [Let oi and the intersections of . A = B.* &c. This will be expressed by the notation +O = where Cp less than m Gm GK . <£„. Gm For the proof of this important theorem see § 10. Suppose any curve GK through L meets Cn in the points L and N'. taken together...ic. and GK passes through the intersections of Gn and other than 0. while the curves (7j meet at 0. Cn +CN A = Cn +CN B = r 0. they are the complete intersection of Gn with some curve (Gl 0).196 THEORY OF RESIDUALS /= 0. care is necessary if some of the points are multiple points of the Mic. namely C^ 0. assume its truth and deduce some consequences. it is always possible to find polynomials C>. 0. if groups of points P. subject to certain limitations as to the positions of the points 0. t And in general. we write P+ Q+ R + . = A. B. and O and L as the complete intersection of Cm and Gn 0. Then the points and N' are the complete intersection of Gn with * some curve.] Let us now consider Gn as a given wic with any points taken on it. Gv such that = We CjCa + ®mPn 0. cm = form the complete intersection of G„ * For instance. assume that. Gm each of the points 0. =z o.. . Through the nN intersections of an nic and an . The points O and are said to be residual groups of points for the curve Cn 0. are the JVic and reic..
[P+P! = 0. + PS  = Q. L and M are an nic. — .' ... show that this point is the same whatever the curves used and whatever the value of r. in number. at some stage or other the algebraic result must be interpreted geometrically. they coincide. we see that symbolical relations such as + 0. If the points P are p n lr . In we proved M to 0. If P2I consists of a single point. The If result above obtained may then be worded any coresidual groups of points for points residual to L are residual to M. since L and were both residual residual to L were residual to M. XII 9 THEORY OF RESIDUALS 197 The points L and are said to be coresidual groups of points for the curve Cn meaning that they are both residual to the same points 0. — n a . any points N' Symbolically. of course. Through points P on a given nic any curve is drawn meeting the nic again in points P1( through t is drawn any curve meeting the «ic again in 2 through P2 is drawn any curve meeting the »ic again in P3 . if OL.] * 'The points Q and the points P taken twice from the complete intersection of the »ic with some curve. P=P Hence P^i + 2r . is points. fact. P 2 . we have m (»j + n. : . 2. we deduce from that +L = 0.. . Such a treatment of the symbolism is a great economy of labour. M L = M. then we have Ex * if ptr and ^as are single 2S =P % P=P P^O give (say). &c. Remembering the notation L = M. and the curves are of degrees m. . if it is to be of any value. We denote this by the notation . [Any line through P other than PQ meets the nic in n — 1 points which cannot form with Q the complete intersection of the 4reic with any curve.] Ex.. Two single noncoincident points P. Q cannot be coresidual (»>2).. In other words. and 2P + Q to mean + + Q 0* &c. can be added or subtracted just as if they were ordinary identities. = = — M=0 P P = . O + M = 0. OM. Multiplication by an integer will be lawful.. 1. for that is only equivalent to repeated addition but division is not allowed.. so are MN'. It is only necessary to consider such a relation as L to be equivalent to L M. P P . . We Ex.. 1= M+N'=0. LN' are all complete intersections of the n\c with some other curve.)— jp + 1 = 0. 2 P. «j. that. must be careful to confine our algebraic processes to addition and subtraction. It enables us to some extent to replace geometrical reasoning by elementary algebraical work though.. L + N'=0. Suppose other curves used. %.. and so on. P! +  P = 0.
9. Show that the conic through Q and the cubic through R meet in three points on the quartic. so do the intersections of the 4ic and the two conies are collinear. 5. ' 8.] pairs of points on an «ic (n no three of the four points are collinear. 3.] 0. P [Use Ex. R + P+Q = P+R = S=0 Ex. through them draw a conic meeting the 3ic in one more point R.198 THEORY OF RESIDUALS XII 9 Ex. Let the line joining R meet the 4ic again in two points S. remaining The four other and 4ic lie on a conic. Three cubics go through seven points. Let the 3ic and 4ic meet in six other points Q. meet cubic 3 in H. 1.] Ex. and 2P+Q = 0. through these points draw a 3ic meeting the given 3ic in seven more points. Then on cubic 3 we have 1 3. Show that the lines joining the remaining intersections of the cubics taken in pairs form a triangle whose vertices lie one on each of the three given cubics. B + K=Q. Two > 3) cannot be coresidual.] lying on an (» — 1) (m — 2)ic. see § 4. K. if [As in Ex. 0.] Ex. so that by Ex. .] Ex. Ex. through them draw a 4ic meeting the 3ic in five more points. 6. meet the curve [Denoting the n points by P=0 Hence Ex. Show that any conic through the four points initially taken meets the 3ic again in two points collinear with R. H=K. Two trios of points on a quartic cannot be coresidual (unless they coincide). 8. and the n{nl)(n2) points by Q. P+B = 0. The tangents at n collinear points of an again in n[n — 2) points on an (w2)ic. If six intersections of a 3ic six. 1 H and K coincide. For the P and the n(n — 2) points by Q. and the intersections of cubics 2 and and 3 by A. Hence Ex. by 2P+Q=0. Ex. The reader may enunciate similar theorems. [Denoting the points of contact. Through seven points P of a quartic a cubic is drawn meeting the quartic in five more points Q. Show that the remaining intersections with the wic of a pic through the inflexions lie on a (pSn + 2 S + 6)ic which passes through the intersections of the nic with its nodal tangents other than the nodes. which lie on the P. Then give Q + S=Q. An wic has h nodes and no cusp. 10. B These give P+A = 0. Let the lines joining the points A. 2. 7.] ?jic Ex. Ex. For the case » 3 see § 7.] From any point n (n — 1) tangents are drawn to a nonsingular Show that they meet the curve again in n (n — 1)(«— 2) points = = P= first polar of 0. 11. 5. remaining five intersections of the 4ic with Therefore the trios coincide. [One trio is residual to the any conic through the other trio. Through is drawn a quartic meeting the given quartic in nine more points R. 4. and Q 0. See also § 5. 3. case n = 3. [Denote the six intersections of the 4ic and the first conic by i*and the other two intersections by R. [Denote the seven points by P. A + H=Q. [See Ex. B respectively. Q=0. Through four points of a given 3ic draw a conic meeting the given 3ic in two more poinds. »ic.
then we can put/ in the form Afy + Bf. so that 1=0. XVIII. I+R Q (since the points I and R lie on the Hessian). We may suppose the axes of reference taken perfectly generally. 6.] 0. S 1. 13. Q + R 0. Then P+I=0. . as the X shall suppose that the intersections of $ 0. which is an equation giving the abscissae of the intersections = and \j/ = 0. Ex. Show that the remaining intersections with the wic of a pic through the points of contact of the tangents from and the intersections of the mic with the tangents at (other than 0) lie on a {pn + %)ic. Then from the Hessian 1 + R 0. it follows at once from the process of finding the highest common factor that each remainder found in the process. reach a remainder R involving x but not y. Ex. and in particular the last remainder R. Ex. of which R coincide with the nodes.y) 0.] = = = Ex. v/r are ordinary distinct points of 0. as a factor. and where § 9 we made use of the theorem that. is of the form A</> + fx\jf. A We C . but not (x — af. 6. The sixteen inflexions of a quartic with a biflecnode lie on another quartic. if the curve passes through the intersections of <j> (x. and taking/. and from the nodal tangents R See Ch. b) be an intersection of <p 0. <$>. [Denote by I the inflexions and by R the intersections of the 4ic with the nodal tangents. the quotient is v and the remainder 6. so that till and B are polynomials in x and y. letting I be the points of contact of tangents from and replacing the Hessian by the first polar of 0. y) y) \jf(x. = = = . To and \jr we may carry out the eliminate y between (j> process of finding the highest common factor of $ and \j/. so that P+ Q 0. = = § 9. 11. where A. The remainder R must have therefore of x — a. when A/ is divided by ^. and /* are polynomials id x and y. are here using Cartesian coordinates. and give a proof of the result with this limitation. Suppose that. Moreover. and \jr Let (a. \jr 0. 0. <p 0.XII 10 THEORY OF RESIDUALS 199 [Suppose the reic and pic meet at the inflexions / and other points P. In / (x. while the nodal tangents meet the wic at points Q and R. An »ic has a single node 0. and the eight inflexions of a quartic with two biflecnodes lie on a conic. considered as polynomials arranged in descending powers of y. [As in Ex. Equating remainder to zero we get the result of elimination required. We = = /= — = = = = = we this (j> v and being polynomials in x and y.j §10. ty CK Gm Gn of § 9. 12.
b) lie on A Then 6 A/— v\jr shows that all the n intersections of — A n— R= = = = = R — = = = n— = = = = and x = a lie on = 0. or 6 R as =A a factor. cannot meet = a in more than %— 1 points unless — a is a factor of 8. and therefore . if that is proved. But = 0. $ = 0. (\cj> + fj. The relation \<j> + /j.\jr shows that every one of the or <j> with x a lies on A n intersections of i/r 0. b).A <f>) = Now. For the case in which the intersections of / = 0. 6) lies on / = and = 0. (v + A ^)/A must be a polynomial B . being of degree n — 1.\jr). since R + A n) = A<£ + (v jxt/t R independent of i/f y. But the results there given can also be established in most cases by other methods. A being A/ = or +A and (A</> + nir) is A (/. we prove that 6 has x a similar argument will then apply to each factor of R. we have 6 a polynomial. * paper in the Mathematische Annalen from which the above proof has be'fen have applied the residuation theorem in the 9 to such cases. for (a. may refer to Noether's p. has not if we have chosen the axes of reference generally. and x a other than (a. while the other % — 1 lie on = and A = 0. But. .* We examples of § without a more careful investigation. a. 140. Hence A and \jr \jr factor. or f=A<f> + B^r as required. We should then have = AR v\jf \lt. adapted. i^ = are multiple points of any of these curves the reader \f \j[f i// a. and the result follows. any common factor of A and would be independent of y. xl (1892). Suppose \jr of degree n in x and y'smd therefore 6 of degree 1. a factor. have no common a function of x as factor.200 It only THEORY OF RESIDUALS remains to show that 6 has XII 10 For. But (the axes of reference and i/r with being general) the only intersection of <£ 1 intersections of i/r abscissa a is (a. Hence the 0. To show that has R as as a factor. which was not strictly lawful The reader may consultthe bibliography in the Mathematical Encyclopaedia. since has x a as a factor.
Since a line joining two double points of a curve meets it in at least four points. Hence there are only three types of cubic.CHAPTER XIII UNICURSAL CUBIGS § 1.* whose Pliicker's numbers are given by the table Type . Types of Cubic. a cubic cannot have more than one double point.
according as the cubic has a cusp. l.202. Since the sum of these is 2tt — <f>. according as h2 = .'] . § 4). acnode. VII.' In fact. and then inverting with respect to the acnode we get Through any point of an ellipse three leal circles of curvature pass other that the circle of curvature at 0. The real asymptote of a cuspidal circular cubic is equally inclined to the tangent at the cusp and to the line joining the cusp and the focus. or hyperbola. and the inflexional tangents of real The the cubic invert into the other osculating circles of the cubic which pass through 0. [Take the case of the acnodal cubic. and unit we obtain the conic through ax2 + 2hxy +*by 2 + 2gx + 2fy . infinity. 2. 1. and therefore the three inflexions are collinear (cf. . e. [Inverting with respect to the cusp and remembering that the focus inverts into a focus (Ch. A cubic with an acnode has three real inflexions. < . 1(4**). By a real projection it may be transformed into a circular acnodal cubic. GEOMETRICAL METHODS XIII 2 Then. Ex. iV"<t>). Ch. V. Fig. § 4) we have : ' Any tangent to a parabola is equally inclined to the axis and to the focal distances of the point of contact. ellipse. asymptote of the cubic inverts into the circle of curvature of the conic at 0. This conic is a parabola. = 0. if the origin is taken at the double point of the cubic. Similarly for the crunodal cubic] Ex. or > ab. the eccentric angles of the three points of contact of the circles of curvature are : ' *<#>. or crunode i. the three points of contact are concyclic with 0. the curve is now a circular cubic with an equation of the form ax 2 + 2 hxy + by 2 + 2 (gx +fy) (x2 + y 2 ) = Inverting with respect to a circle with centre radius. if the eccentric angle of is (f>. and a cubic with a crunode has one real and two unreal inflexions. 0.
and that the circles on such chords as diameter are coaxial. S. If O is the centre of a fixed circle touching two given circles externally. Obtain properties of a nodal circular cubic by inverting other properties of the conic. point of contact of a tangent from O to a family of confocal conies is a circular cubic through the foci with a node at O. and after reflexion passes through a fixed point B. 13.] x [a? + y ) + ax 2 + 2 hxy + by* = are perpendicular. find the locus of the inverse of with respect to any other circle touching the given circles externally. that the middle point of the chord lies on a fixed straight line. inter \x + fiy + to the origin envelope. 0). Consider the case in which O is on an axis or at infinity. if O. Ex. Generalize by projection. is reflected at P from any sphere with a given centre O. B are the origin. . s and any . 6 sin a.] Ex. 0). we get the tangential equation 1 = and 1 of the Ex. subtending 90° at Ex. Ex. Ex. line Pi. or is 0Q n zero. The cissoid becomes a parabola with If a line vertex 0. — AA = AO A'O ' . 12. 6. The locus of the. [Part of a circular cubic with a crunode at 0. 7. S' and the SO ± S'O * Ex. Find the locus of P. 10. A. A cubic has three asymptotes and a node 0. P2 .] Ex. circle is the node of a circular cubic with real foci 5. P P 2 . A ray of light proceeds from a fixed point A. 11. Show Ex. The locus of the centre of a circle whose circumference passes through two given points and meets a given line not coplanar with the points is a nodal circular cubic. 1 [(x' + y ) {2/ (6 + a) cos a + a: (6 o) sin a} =2abxy. Any . 8. z = 0. O envelops a sections of [Expressing the condition that two of the lines joining the. The lines joining the node of a nodal circular cubic to the real foci are equally inclined to the real asymptote. 0QS 0Qt . make with the tangent [Invert with at the cusp respect to 0. meets the cubic in Q and the asymptotes in P. through . show that 0P1 + 0Pi + 0P3 = ' oq. If OSS' meets the curve again at A and A'. a sin Oi.] . 4. meets the cissoid x(x i + y'i ) = ay 1 with cusp in circle meets it in Q lt Q2 Qs Qt Show that the sum of the cotangents of the angles which Ex. OPlt 0P 0PS 2 . OQi. 3. Consider the case of confocal parabolas. . 9. A chord of a circular cubic with a node subtends 90° at 0. (a cos Oi.XIII 2 GEOMETRICAL METHODS 203 Ex. (6 cos a. The chord of a circular cubic with a node conic. P 3 .
i have the Cartesian equations 2y*(xl)2y(xi 2x)+x*(x2) = 0. Hence the equation of the cubic is zy 2 axs Putting az for z. since z meets the curve three times at B. XIII 3 Cuspidal Cubics. . = . and the only term involving z is zy 2 since y 2 are the tangents at G.* Then there is no term in the equation of the cubic involving z 3 or z 2 .204 CUSPIDAL CUBICS § 3. » A real cuspidal cubic has one cusp and one inflexion. 2. Take the triangle of reference HG so that and are is the inflexion and G the cusp. we see that = = = = . 64y*(xi) + 16x'y + xi (5x12) = and x(x 2 — xy + iy*) + iy (2x — y) — respectively. 3. the equation must reduce to X 0. when we put z 0. A choice of cwpidal cubic can be put into homogeneous coordinates. while the tangents at and C (Fig. 0. s Also. 2). 2. Fig. and these must be real. B AB A AG B . the form zy 2 =x 3 by a real * The curves shown in Pigs.
the we have sum of their parameters Similarly is zero. or simply the tangential of P. The tangential of the first ' ' tangential is called the second tangential. : and conversely. the tangential of the second tangential is called the third tangential. If the points with parameters t lt t 2 . and conversely. I. 1. for the tangent at meets the curve in the points P. (2)H. If the parameter of is t. or into yx* = as . P. Q is called the first tangential. tLL ( ^2) M t = oo . and conversely. the = x 3 are sum the intersections of this cubic of their parameters is zero. = . If the tangent to any cubic (unicursal or otherwise) at P meets the curve again at Q. If the points with parameters t x t 2 t6 lie on the line — . Ex.t 5 t e lie on the conic ax2 + by2 + cz2 + 2fyz + 2 gzx + 2 hxy these quantities are the roots of at 2 = 0. t s are the roots of the equation in Xt + /i + vt 3 — 0. the so that t 1 + t2 + t s = 0. the parameter of the first tangential Q is — 2t.y + vz t = 0. Q whose parameters have a zero sum. collinear. 1. successive tangentials of a given point approach the cusp as their limiting position. t 3 ).] . t i . 2 If nine points of zy with another cubic.t 3 . + b + ct 6 + 2ft 3 + 2gt* + 2M = t s Since the coefficient of 2 If six points of zy in this equation is zero. . Any The cuspidal cubic can be projected into a2 y Ex. t being the parameter of the point. § 3 that any cuspidal cubic can be projected into the semicubical parabola ay 2 x3 Any point on the curve zy % x 3 may be taken as (t. and the parameters of the inflexion and cusp •00 are zero and infinite. and so on. 0.: : XIII 3 CUSPIDAL CUBICS 205 It follows from Ch. =x s . Hence 2 If three points of zy = x3 are sum of their parameters is zero. 2. The parameter of the nth tangential of is evidently P P P . \x + /j. t lt t 2 . lie = xs on a conic.
Through a point P of a cuspidal cubic a conic is drawn having 5point contact at P'. in three points. Show that the line PQ passes through the inflexion and is divided harmonically by the inflexion and the tangent at the cusp. Ex. 16 (i). inflexion. a tangent is drawn touching at P2 and P„ approaches the inflexion as its limiting position. 3. the inflexional tangent. 8. = [A parabola having double contact with the given curve. i ) s is seen by putting 8. conic passes through two given points of a cuspidal cubic. From a point P on a cuspidal cubic a tangent from P. 9. § 2. The tangents from P to a cuspidal cubic meet the curve again Show that the tangents at these points meet in a point Q and that. Ex. 4. If in Ex. . and so on. 11 the crossratio of (PQES) inflexion. . P P . P T Ex. similarly P2 is derived from P1. 7. 5. cuspidal tangent. The loci of the poles of the inflexional and cuspidal tangents with respect to the conic of closest contact at any point of a cuspidal cubic are cuspidal cubics with the same cusp. E. flexional tangent. is any constant other than 9/8. The conic of closest contact at P to a cuspidal cubic meets it again in P. 6. Find the limiting position of P< n >. conic osculates a cuspidal cubic at and Q. Find the limiting position of P„. Find the locus of the intersection of tangents to the semix s at points subtending 90° at the cusp. 13. X. the cuspidal tangent in Q. 16. and cuts it again at Similarly 2 is derived from t Pj. Ex. XIII 3 drawn touching so on. A P P to a Ex. 12.] Ex. closest contact at any point of a cuspidal cubic (other than the cusp or inflexion) meets the inflexional tangent in unreal points and the cuspidal tangent in real points.] Ex.206 Ex. and so on. P . The tangent at to a cuspidal cubic meets the curve. See also Ch. 11.. 7 to 13 by reciprocation. osculates it at P. Find the limiting position of n 3 from 2 A P P . 1. SHbw that the crossratio of the range {PQRT) is 9/8.] Ex. CUSPIDAL CUBICS is . Ex. z in the general equation of a conic and comparing the resulting equation for 8 with as (6tf(8t') =0. respectively. 14. Ex. (t. Show that [The parameters of P and P„ are in the ratio 1 : (£)". Obtain theorems from Ex. 10. 15. [The parameters of P and P 1 are in the ratio 1 : —5. and so on. . cubical parabola ay 1 Ex. find the locus of S. similarly P" is derived from P.] cuspidal tangent. 1. [A cubic with the same cusp.] Ex. The locus of the intersection of two perpendicular normals to a semicubical parabola is a nodal cubic. . so does Q. at P. and in Ex. and inflexional tangent. Find the locus of the intersection of the tangent at cuspidal cubic with the tangent at the second tangential of P. The conic of [The conic of closest contact at i 2 with zy 2 — x* is 4ht*x + 5t?y*z 40?yz + 15t*zx24:t?xy = 0. y. 6* for x. Ex. if P moves along a straight line.
18. the coefficients of x2z and 2 Choosing suitable homogeneous z have the same sign. The P and locus of the focus of a parabola passing through fixed and having PQ as normal at P. is a straight line. t. and CB is the harmonic conjugate of CA with respect to the tangents at C. 2at). [Any point on the curve is x Ex. Ex. and Ch. Ex. e. [If the parabola is i/ 2 = 4ax and perpendicular from the focus S on PQ. § 4. The locus of a point P.XIII 4 NODAL CUBICS 207 Ex. The equation of any wic with a tangent of rapoint contact and a superlinear branch of order re — 1 may be put in the form zyn' = xn . is the cissoid of Ex. P is (at 1 . 19. PQ .] PZ= a {! + ?)*. 21. their parameters have zero points with parameters t. CB is the harmonic polar of the inflexion A. Ex. 17. and the tangents at G form a harmonic 9 xz 2 . P and Q is + 2 < ). i. Find the circle of curvature. Show that (i) (ii) (iii) +y = i ) ay* subtends 90° at The The The locus of the middle point of PQ is a straight line. y = . The sum of the abscissae of the feet of concurrent normals of a (x — yf = x d is constant. The equation of any acnodal cubic can z (x 2 be put in the form +y 2 ) =y (3x 2 — y % ) by a suitable choice of homogeneous coordinates. fi = 4a (1 + i2 )i eliminate a and Ex. locus . the 3 2 coefficients of x and arc/ must be zero.of the intersection of the tangents at locus of the intersection of the normals at Pand Qis a circle. The chord PQ of the cissoid x(x 2 the cusp. — 3< is + lt*)x = 6t*y¥a enveloping 3 5 y 2 (ax) = Vz 3 . Ex points 20. 10. Take the triangle of reference ABC so that C is the node. remembering that the tangent at A meets the curve thrice at A. the coefficients of z equation of the cubic are zero. of which one at least must be real. A real nodal cubic has a real node and three inflexions. Nodal Cutaies. Since A is an inflexion. [If four points = a/(l + fi). which moves so that the Bum of the angles made with a fixed line by the tangents from P to a semicubical parabola is constant. a straight line. yz 2 xyz in the pencil with xy 0. 8. A is a real inflexion. The chord joining the (l on the curve are concyclic. since G is an acnode. y = alt (1 See § 2.] Q. 18.] Ex. envelope of the common chord of a cissoid and its sum. . while SZ is the SZ=at(l + t Now i )i. XI. . § 8. Since C is a node. 22. Moreover.
± /3.„ . . sin 3(/>). AB of a new triangle of reference. § 1 is equivalent to' choosing the line z as the side + y 2) + y (ax 2 + by 2 ) = 0. + a)y. XIII 4 and the equation takes the form In this replace z by £ (ba)z — \ (3b Ch. .208 coordinates. except the acnode. 0. (i) is sin (j>.. The Hessian of It (i) is 2 2 2 z(x + y )= 3y(3x*y ) (ii).. 0) of the cubic (i). NODAL CUBICS we may make z (x 2 these coefficients the same. 0). and the equation ' reduces to z{x 2 + y 2 ) „. Fig. <j> The whole curve. all.N (i). is obtained by giviDg 3). „. = 0.. /. 3. I. Any point on the cubic (cos tj>.real values between and w (Fig. = y(3x2 y 2 ) . which by + ±(3b + a)y = „ . Hence the line z = meets the curve (i) where passes through the three inflexions (1. (1.
z = The coordinates is of any point on sinh fa the line of inflexions. we obtain the parameters of the . is 77. and conversely. namely 0. fa. we may say that the whole curve is divided into two portions by the crunode.  77. 77) . iti) for (mod. = 0. {mod. The condition for three points on a line or six points on a conic is the same as that given for the acnodal cubic.. The sum of these . fa. except that the coefficients of x 2 z and y % z have opposite signs. Similarly lie An we may show conic. tan fa. owid conversely. tan fa are the roots of 2 (cos <f> + sin 2 fa (A cos <f> + u sin fa + v sin considered as an equation in tan fa tan<Jb. are collinear. (cosh$. One contains the A . The inflexions have parameters 0. on a + fa + fa: + fa + fa. The Hessian of z(aPlf)=y(aa? + y*) is (iii) z(x2 y 2 )= 3y(3x2 + y 2 ) . 2 + y 2) by a suitable choice of homogeneous coordinates. ir) . n). s as found before by means of the Hessian. (iv). fa lie 209 on the line Xx + fiy + vz = 0. \iti. on the cubic with parameters fa. fa. = inflexions. As before. fa. Hence crunodal cubic has three collinear inflexions of which only one is real. fa + (£ 2 + $ 3 Putting fa. <f> (3 cos 2 $ — sin 2 fa tan fa . %iri. If we suppose that the points of (iii) at the two ends of an asymptote are the same. substituting (mod. fa that (i) If six points fa.:: : XIII 4 If the points NODAL CUBICS with parameters fa... $ 2 <£ 3 (mod. Hence: * acnodal cubic has three real collinear inflexions . tan fa. fa. + fa be = fa.. Therefore + tan fa + tan fa = tan fa tan <f> 2 Hence i/ three points on the cubic (i) with parameters fa. fa all equal. The equation of any crunodal cubic can z (x 2 put in the form —y = l ) y (3a.. (iii) are sinh3^>). The proof is almost exactly the same as that given for the acnodal cubic. §7r.
We may mention the form x(z 2 ±y 2 ) =y 3 (v) obtained by taking A as the node. cosh <\>). is obtained by giving the value 6 + ^ni. § 1 for the nonsingular cubic. cos <£) or (sinh 3 <f>. The line of inflexions is 4x = ±Sy.210 real inflexion. The form (v) is suggested by that given in Oh. Many convenient standard forms of the equation of a nodal cubic exist. <\> cot <f> =t or coth <j> = t. by putting — oo to + oo . . The coordinates of any point on a nodal cubic can be expressed rationally in terms of a parameter t. 4). other than (i) and (iii). where 6 takes all real values between — oo and + oo (Fig. XVI. Fig. G as an inflexion. the harmonic polar of G. sinh#. NODAL CUBICS <j> XIII 4 and is obtained by giving all real values from The other portion. AB as Any sin</>. 4. and CB as the tangent at G. The condition for three points on a line or six points on a conic is the same as that given above. point on the curve is (sin 3 <£. which we shall call the loop.
Ex.1) a uu z = (n. we may use the theory of covariants. Show that. § 7. the product collinear points a conic is + 1. 3. or one (x + a ) real inflexions according as it has 3 Show from the form y (3 2 a. R. X. Ex. Q. prove that from any point on one part two real tangents can be drawn to the curve. 3 2 . Show that no real tangent can be drawn to a crunodal cubic from a point on the loop and that from any other point of the curve two real tangents can be drawn. 113.] Ex. § 2.. Show that the equation the form x s + y s = Bxyz. 6. 2^ + ^ = (mod. 4. 4. one point of contact being on the loop . — 1 and the can be projected into the folium of Ex. .' 4 the product of the parameters of three collinear points is — 1. [The tangents at the finite inflexions of y(x i + al ) = which is the harmonic polar of the inflexion (oo 0). line meets a nodal cubic in P. xxxii. sin fa) constant. Messenger Math. Show that. . Considering an acnodal cubic (supposed continuous at the two ends of an asymptote) as divided into three parts by the three real inflexions. verifying the for any special form of the equation of the cubic] = 3.2 a A . 16 (ii). 5. 1 = a that a nodal cubic has three an acnode or crunode. VII. The conic of closest contact at P with a crunodal cubic meets the line of inflexions at unreal or real points according as P does or does not lie on the loop. A t ) y = 0. a? meet on x = 0.) x + 3 (a A 3 + a % A x . 1 + f). and the lines joining these points to the node make angles 6 1 and 2 fa and </> 2 fa and fa^ with the tangents at the node. sin fa .] Ex. Or again. Ex. ttj). its ty are the parameters of a point and tangential. sin <f> 2 . Show^that the tangents at two inflexions of a nodal cubic meet on the harmonic polar of the third inflexion.(sin #2 . 14 (v) in the (a^x* l x case n p. inflexion. sin fa) 4. 7.2) a m of Ch.] . NODAL CUBICS A nodal cubic can be projected into 2 a (x 2 + »/ 2 ) = x s or into y (x" ± a ) 1 211 = 3 u. [Use the fact that in Ex. result For another proof see Bromwich.and Ch. not.4 i XIII Ex. The line of inflexions of i is + 2a xy + a2 y2) z + A^x3 + %A x2 y + ZA^xy 1 + A s ys = . ' of any crunodal cubic can be put in curve is taken as is (3 1. [See Ex. 8. 9. .o^ 2 ) z + 3 (a A 2 + « 2 A . 4 (n a 2 [It is the (n . 2.] Ex.] 3 d?V [— = 2 a _ + «')/(*" + a 2 ) 3 vanishes at an Ex. Ex. 1. A. one point of contact lying on each of the other two parts. and the other [If <f>. f2 .2 a. (sin 6X is . Ex. if any point on the product of the parameters of three of the parameters of six points on Show that any crunodal cubic Descartes x 3 +y 3 = Saxy.
= {sin 2 a + sin 2 . (iii) As P [Take the cubic of § 4 (i) the case of the cubic (iii) is similar. and 'QR at and the line of inflexions at F. 15. it [Eliminating z between the equation of the cubic and the polar conic we get the rays of the pencil. § 11. R. 14. Find the locus of P. and the tangents at Q' and R' meet on the curve at P'.212 Ex. I. Ex. QR and Q'R' * meet Ex. 11. — 10. . at is NODAL CUBICS A family line. The tangents at Q and R to a cubic with node the curve at P. . Prove that the points of contact with the cubic of the six other common tangents of conic and cubic lie on another conic. and the line of inflexions the axis of perspective. 17. If the tangents at on the line of inflexions. The — £71. (ii) EP) is harmonic. whose envelope is s2 = a. S are respectively 0. If is harmonic. the line of inflexions being the chord of contact. Now use Ch. equation of the family in the form 3 /fc Or put the 33 = (x + y + 3kz)xy. If the pencil is equianharmonic. 13. The conic through touching the cubic at Q and B touches the on OP. of P. XIII 4 two of the inflexions are a straight of nodal cubics has given inflexions and the tangents Show that the locus of the node also given. the locus is a cubic] Ex. The points of contact of the four tangents from P to a nodal cubic subtend a pencil of constant crossratio at the node.sin § 2 (a + 0)} x + {cos 2a + cos 2 + cos 2 + 0)} For the cubic of 4 (iii) replace sin and cos by sinh and cosh. Show that meet on E OP meets (i) OQ and OR and (OF. Ex. parameters of P. Show that (i) (ii) The conic OQRQ'R' meets the line of inflexions on OP and OP'. the node being the vertex of perspective. 12. The line QR meets the curve again at S. \(ir — 4>). The line joining two points of the cubic of parameters a.] § Ex. a The line QR is z + sin 2 f x + cos 2 y = 0. Q. varies QR envelops a conic touching the tangents at the node. line of inflexions at a point (iii) P and P' meet on the curve. A conic touches the tangents at the node of a cubic. 3. A nodal cubic and its Hessian are in plane perspective. 10. The other intersections with the curve of any line through a fixed point of a nodal cubic subtend an involution at the node. + y*. 16. [The harmonic polar of the third inflexion by Ex. Ex.] Ex. (3 is z 4 (a (i) with y. at also with respect to are harmonically conjugate with respect to the tangents OP and OS. the locus is the line of inflexions. The tangents at Q and R to a cubic with a node meet on the curve at P.
. . R.] Ex.4 XIII NODAL CUBICS . 213 Ex. 21. 3x Ex. the chord meets the cubic again at a fixed point. Prove a similar theorem for the six points in which the cubic of Ex. and that the centre of this circle lies on a fixed line. Show that the locus of the centre of the circle is a fixed line. 18. A = Ex. If a chord of x s + y 3 == Baxy subtends 90° at the node. Show that the points of contact of tangents from any point on = 2a to a{x2 — y 2) = xs lie on a circle through the origin. circle through the node cuts a (x2 y 2 ) x s again in four points. Q. Q. 2 If f l t it t 3 are the is (a(l + t ). line cuts a{x i + y i ) = Xs in P. parameters of three colliriear points. which shows that the lines joining the points to the origin make angles with x — [Any point on the curve whose sum is Jn. 19. A straight. 22. t^ + t^ + t^ = 1. Show that the lines joining P. 18 meets any conic. It to the node make angles with any fixed line whose sum is constant. at(l + fi)). and the sum of the angles which the lines from the node to these points makes with a fixed line is constant. 20. Ex.
Project such a line to infinity. Place the asymptote vertical (Fig. 1. remaining parallel to the asymptote. 1). we see that there are lines meeting the curve in only one real .CHAPTER XIV NONSINGULAR CUBICS 1. cubics without a double point and of unit deficiency. e. 9 (i + 4) y 1 + 2ixy + x (x2 + 2x . A Fig. XIII. and the curye will have one and only one real asymptote. § 2. Cubics with Unit Deficiency. In this chapter and the two following we shall consider the properties of nonsingular cubics. .16) = 0. As in Ch. and suppose a line I starts from coincidence with the asymptote and travels to the right. i.point.
namely the nine intersections Of these an odd of the cubic with its Hessian (Ch. The curve can have no other portion for. Hence there is a continuous portion of the cubic called the odd circuit * approaching the asymptote at its ends. namely the distant point and as a poirit travels along the curve from . The cubic has nine inflexions. For this is true of one point of the circuit. two real tangents can be drawn from to the odd circuit. the term even circuit' is applied to that part of the curve (if any) which can be projected into an ovaf. since both cubic and Hessian are real. Similarly if I travels to the left. the rest of the curve being the odd it . if two ovals existed. For any line through must evidently meet the odd circuit and must meet the even circuit in a point other than (see fig. unless the line is the tangent at P. when I travels still further to the right (or left). ' ' circuit '. 2 . and contained between two lines Zj l parallel to the asymptote. The curve will then have a portion called the even circuit * or oval contained between two lines l s Z4 parallel to the asymptote. and twocircuited cubics having an odd and an even circuit. t The names unipartite and bipartite are also used instead of onecircuited and twocircuited ' ' ' ' ' ' ' ' ' ' ' '. to P. that they are collinear. it touches the cubic again. number must be real. Of the three points thus obtained no two can coincide.f From a point on the even circuit of a cubic. a straight line cutting both would meet the curve in four points. § 7). If a cubic has three real asymptotes. Moreover. VII. for the curve in general approaches the asymptote on opposite sides at the two ends. . From a point on the odd circuit two real tangents can evidently be drawn to the even circuit (if any). a real tangent from the point remains real and an unreal tangent remains unreal. since the curve has only one asymptote. Hence cubics can be classified into two main divisions onecircuited cubics having a single odd circuit. 1). eventually these two points coincide and I is a tangent. It may happen that.: XIV 1 CUBICS WITH UNIT DEFICIENCY 215 At first I meets the curve in two real finite points. As I travels yet further must again become a tangent. As I travels. and that they lie on the odd circuit of the cubic. . P P P P P A A * The names odd or even circuit emphasize the fact that the circuit is met by any straight line respectively in an odd or even number of points. meeting the asymptote in a finite point. We shall show that exactly three are real. no real tangent can be drawn (other than the tangent at P).
. since and its interthe odd circuit does not cross itself * between the circuit has at least one section C with the asymptote. . § 4). Ie I3 I6 IS !$!$. 2). fl . &c. VII. . Secondly. But every straight line must evidently meet the odd circuit in at. suppose that ij. The circuit has then two real inflexions. Then 1^. A Fig. 1 that. Then I^I^. while 7j/7 and IS I5 meet the curve again in real inflexions Is and J which we may suppose to be their intersections with 72 /5 and Zj/7 since these latter lines also meet the curve again in inflexions. I2 ^4' ^z^i meet the curve again in real inflexions I& I6 J7 . * Contrast the case of the crunodal cubic in Ch. . . fig. The line joining them meets the circuit again in a real point which is also an inflexion (Ch. . To show that the curve has no more than three real inflexions. I2 Iz are the inflexions just found and that Ii is a fourth real inflexion (Fig. meet the curve . 4. tangent at such an inflexion would meet the cubic at three points coinciding with the point of contact. Similarly it has an inflexion inflexion between A and C. and therefore no straight line can meet the even circuit in more than two points. least one point. . 2.216 CUBICS WITH UNIT DEFICIENCY XIV 1 For the Firstly. XIII. no inflexion can lie on the even circuit. between B and G. suppose A and B very distant points on opposite ends of the odd circuit. It is evident from fig.
Circular Cubics. Then the cubic takes the form (i). and the tangent as the axis of y. Ex. 6. the cubic is the locus of the point of contact of a tangent from to any conic through A. We may project the unreal intersections of the line and cubic by a real projection into the circular points co. 1. Hence we can always project a cubic into a circular cubic. . We pointed out in § 1 that we can always find a line meeting a real cubic in only one real point. [Use polar coordinates with as pole . as the direction varies. such that the lines joining P to three fixed points meet three fixed lines in collinear points. They are parallel to the real asymptote. 5. is a cubic through the fixed points and the vertices of the triangle formed by the fixed Ex. lines. Take tbe point of contact of one of these tangents as the origin. ca. Obtain the results of § 1 by proving that at least one line can be drawn through an inflexion meeting the curve in only one real point and projecting that line to infinity. 2. x(x2 + y 2) + ax 2 +2hxy + by 2 + kx = . Ex. i. § 8. 00. We . Show that the locus of the middle point of PQ is a cubic with a node at and asymptotes parallel to those of the given cubic. § 1 Ex. Through a point of a cubic any line is drawn meeting the cubic in Pand Q.] The locus of a point P. e. a cubic passing through the circular points to. The inverse of this with respect to the circle with centre the * may apply the above proof to show that an acnodal cubic has exactly three inflexions which are real and collinear. OD are the tangents from a point of a cubic to the curve. 217 again in real inflexions contrary to the fact that the curve has no more than nine inflexions real or unreal* Ex.XIV 2 CIRCULAR CUBICS . by projecting a real inflexional Ex. Obtain the results of tangent to infinity. 4. i § 2. Ex. OB. B. C. If OA. . Show that the locus of the centroid of the three intersections with a given cubic of straight lines drawn in a fixed direction is a straight line . or see Ch. and that. 3. From the real point at infinity on a circular cubic two or four real tangents can be drawn according as the 'cubic has one or two circuits (§ 1). D. a/. XI. the locus envelops a conic. 3.
is sometimes used in the sense of ' its own . The remaining circles are all * The word ' inverse or ' anallagmatic selfinverse '. three of the circles with respect to which it is self.inverse are real. and thence that any one centre is the orthocentre of the triangle formed by the other three. found to be the cubic itself. mutually orthogonal. since the cubic is. If the circular cubic has two circuits. This system of circles is obviously selfinverse with respect any one of the circles.: 218 origin CIRCULAR CUBICS and radius It* is XIV 2 ky/(x 2 + y 2 ) for y. Hence A to circular cubic is selfinverse* with respect to four circles. and is obtained by writing Tcx/(x 2 + y 2 ) for x. It follows that the radical axis of any pair of the circles must pass through the centres of the other two. Therefore the Pig. 3 (i).
The two circles with respect to which a circular cubic is selfinverse and which have their centres on the odd circuit meet the cubic circuits. two of the circles are and the other two have unreal centres. 3. circuit. 1. [The points are the points of contact of tangents to the cubic from the centres of the circles. . has a real centre on the side of the oval near to the odd real. 3 (ii). This will be clear from Fig. Pig. The tangents to a circular cubic from a point where the tangent is parallel to the real asymptote are all equal. Ex. at two or four real points according as the cubic has one or two The other two circles do not meet the cubic in real points.XIV 2 CIRCULAR CUBICS but it 219 circle is unreal. 2.] Ex. ix(x i + yi ) + ix{x + 5y) + i0y71x = 0. If the circular cubic has one circuit.
4. then with the notation of that section. If we allow the use of unreal projection. for the cubic is of class 6. The crossratio is real. and 8 is a focus. a cubic may be (projected into co and co'. There . any two points of. or into a bicircular quartic symmetrical about two perpendicular lines. S. since (PQRS) have the same The tangents to a cubic from any point of the curve (other than the tangent at the point) form a pencil of constant crossratio. Hence m'S also touches the cubic. while the crossratio of a pencil formed by four real lines is real. Suppose j is one of the four circles with respect to which a circular cubic is selfinverse.] is selfinverse Ex. the pencils co. circles. Then the inverses of the cubic and the line coS with respect to j are the cubic and co'S respectively. Now four tangents can be drawn from co to the cubic. This is evident on considering the pencil of tangents from a real point at infinity on the cubic (see Fig. /c 2 + 6/c + 6 0. (ii) the intersection of the cubic with a circle the intersection of two such circles. the cubic being thus projected into a circular cubic. it is selfinverse. Hence we h%ve the important theorem crossratio. Foci of Circular Cubics. other than the tangent at co. § 3. Q. co (PQRS) and P. Let a tangent from the circular point a) to the cubic meet j at 8. If P. if the cubic has only one circuit.: : 220 CIRCULAR CUBICS XIV 2 is Ex. if the cubic has two circuits unreal or equal to —1. Q. For the crossratio of a pencil formed by two real and two conjugate unreal lines is unreal. Its inverse with respect to any other point is a bi circular quartic. and the sixteen foci lie by fours S are the foci on j. Any circular cubic can be inverted into a circular cubic symmetrical about a line. a/ circular cubic is selfinverse with respect to each of four circles. if the crossratio is 1. Ex. R. the cubic has the same crossratio as the pencil of tangents from co'. Hence Any mutually orthogonal on these four circles. [The pole of inversion for is (i) which 5. R. A nodal circular cubic and a cuspidal circular cubic with respect to with respect to two one circle. unless the pencil is harmonic . 1). It will be seen in § 5 that. — = .are Therefore the pencil of tangents from co to all on the circle j. The inverse of a circular cubic with respect to a point on it a circular cubic. co'. 3.
... and therefore Hence its envelope it is selfinverse with respect to (iii). (iv). real. If t is any given root (i) is selfinverse with respect to (iii). If a real focus S of a twocircuited circular cubic lies on a circle j for which the cubic is selfinverse. .. and . (iii) is one of the circles with respect to which (i) is selfinverse.. Ex. Its envelope is found by writing down the condition that It coincides with (i) pro(iv) should have equal roots in m. .. Consider the circle. the other real foci are obtained by inverting S repeatedly with respect to the other circles for which leave. is out. 4 of (v). . * is As just pointed which . if the pencil is .. Ex. 1. 3. Similarly we show that. all and therefore Take now the origin at the singular focus of the circular cubic and the axis of y parallel to the real asymptote.XIV exist 3 FOCI OF CIRCULAR CUBICS both onecircuited (k 221 (k= is —3—V 3) = —3+ </3) and twocircuited cubics. It follows at once that the equation of the cubic is (x+p)(x2 + y 2) + ax + by + c = (i). 2tm 2 (2txp)m(4ty + b) + x 2 + yT+a + 2t(xp) + 4t 2 = . and which cuts orthogonally the circle x 2 + y 2 + 2 (p2t)x + by/2t(St 2 4tp + a) Its equation is (iii). The crossratio of the pencil of tangents from any point of the curve is the crossratio of the pencil subtended by the four real foci at a> or at any point of the circle j. y 2 = 4t(x + t) . if all four real foci lie on j.* is S selfinverse. = .. Then the terms of the highest degree in the equation of the curve meet the curve in only are x (x2 + y 2). if the cubic is onecircuited. the crossratio of tangents only possible for a onecircuited cubic. two real foci lie on each of the two real circles with respect to which the cubic is selfinverse. and the lines x 2 + y 2 = one finite point apiece. The circle (iv) is orthogonal to the circle (iii). harmonic. lies the real asymptote. This case requires further investigation see § 3. (v). These inversions four real foci lie onj. vided 64t i 64pt 3 +16(p2 + a)t 2 + 8(cap)tb 2 = . . 2m£) on the (ii).' unaltered. § i. for which the crossratio of tangents —1. whose centre (tm 2 parabola with focus at the origin is where x + p = — t..
tbV). 0) and the real ordinary foci are ' ' ' The parabolas are ' (0. and the distance of the centre of j from the asymptote is equal to the semilatusrectum of the parabola. 16). 21). (i) In Fig. 3 and (ii) the foci are sh6wn by the small x . and whose centre lies on a fixed parabola with its focus at the singular focus of the cubic and with its axis perpendicular to the real asymptote of the cubic. circles o. 5). 3 (i) the singular focus of the cubic is (0. In Fig.by putting t x = They are by definition foci of (i). 2 (a: while the corresponding focal parabolas are The = 40(x + 3) and (2y + 5) = 40 2). (231. 1). . 3 (ii) the circles for which the cubic is selfinverse are 4x2 + 4:y 2 40x + 2Sy99 and 2 4x 2 + 4y 2 (2y + 5) = + 4>0x + l2y59 = 0. and the harmonic points (ff _V<r)> (ft. are the centre (2305.* The focal parabola through the real foci is 4y 2 4>x + l. at (75. circular cubic may be considered in four ways as the orthogonal to one of the circles j envelope of a circle which with respect to which the cubic is selfinverse. the cubic is . we have: from the asymptote x+p. asymptote 200a? = = selfinverse. The other focal parabolas are at once obtained from the fact that their latera recta are twice the distance from the asymptote of the centres of the circles for which the cubic is . —715) of the circle through the real foci. It will be noticed that the distance of the centre of (iii) Summing up. In Fig.1 222 FOCI OF CIRCULAR CUBICS XIV 3 The centres of those four circles of the family (iv) which degenerate into linepairs (circles of zero radius) lie both on t in (ii). (I. tact of tangents parallel to the asymptote by the dots whon * This follows from the fact that foci invert into foci inverted into itself. (416. % "5 A H bolas called focal parabolas or deferent paraof the cubic.e. (iii) and on the parabola obtained. asymptote is x = 0. The intersections of the 'parabola with the circle j are foci of the cubic. the centres of the circles with respect to which the cubic is selfinverse. i. M) of the quadrangle whose vertices are the real foci. The points which the tangents are parallel to the real 361.= is 2£. the singular focus by a cross the points of con. and the origin lies on the curve.
/30. with focus [The corresponding focal parabola. No other circle passes through four foci. 4). [The pencil (deba). dd' are also concyclic] Ex. (iv) da'. the four circles circumscribing the triangles formed by any three of them pass and their centres lie on a circle through 0. (iii). Through the intersection asymptote any chord OPQ [The singular focus of the cubic of Ex. unless the pencil of tangents from a point on the curve is harmonic. the circle (i) contains four real foci. (a'b'c'd') is If (abed) ratio. has the same crossratio as (abed). can be put x(x* + y 2 ) + ax* + 2hxy + 2gx + 2fy = 0. 3 1. XI. 4. In a onecircuited circular cubic the two real focal parabolas In a twocircuited circular have concavities in opposite directions.] at its Ex. see Ch. £V220. ad' are such that the. WW. &c] through Find the focal parabolas of 3x(x'i + f) + 2x*+6xy + 6y'> = 4x and the circles with respect to which it is selfinverse. 0) and radii Ex. The normals to a circular cubic with singular focus intersections with a circle for which it is selfinverse touch a parabola and axis perpendicular to the asymptote. (ii). § 11. Ex. cc'. so that the foci aa'. b. Ex. The centres of three circles for which a circular cubic is selfinverse are the vertices of the common selfconjugate triangle of the fourth circle and its focal parabola. Show that P. d be tangents from o> to a circular cubic. 7. radius of curvature of a circular cubic is a maximum or its intersections with the circles for which it is self Ex. (§. bb'. cc'.groups (i). (i) aa'. (a'b'c'd') and (chad) have the same crossbb'. cc'. In Ex. cd\ (iii) ca'. — h). 2. 1). (1.. J. Ex. so that ca'. bb'. Ex. 8. . FOCI OF CIRCULAR CUBICS d' 223 Let a. Ex. or else the circles (i) and (ii) contain two real foci apiece. cubic the focal parabola through the real foci has its concavity in a direction opposite to that of the other focal parabolas. cb'. and let be tangents from m. 1 the centres of the circles with respect to which the cubic is selfinverse are the centre of the circle (i) and the intersections of the pairs of lines joining aa'. ac'. [The circles have centres (—1. bd'. Many properties of these normals can be deduced . a'. 3. (bade). iV(~m J). (iv) each lie on one of the circles for which the cubic is selfinverse. 6. dd'. in the form [Take the intersection of the cubic with the real asymptote x origin. Q are equidistant from the singular focus. 4 is (— \a. bb'. db'.] = as of a circular cubic with its real of the cubic is drawn. 9. Then the foci c'. 3. 5. If aa' and bb' are real. (cdab) harmonic. ac'. dc'. aV. c. The equation of any circular cubic. V. (ii) ba'. lc'. for instance. The a minimum at inverse. XIV Ex. da". Ad' lie on the circle j of § 3. (0.
0).224 Ex. (0. y % identify (y. 0).(0 + y) (y + (X) (tX + 0)} = 4 {(a2 + 0* + y2 + 0y + ya. [See Ex. 15. Find an equation giving the other focal parabolas.] (iv) with respect to m\ Ex. [The locus of the centre of a varying parameter. Ex.] Ex. 14. The foot of the perpendicular centroid of any four concyclic foci of a circular cubic is the from the singular focus on the real asymptote. FOCI OF CIRCULAR CUBICS Two and their 11. 0). [The corresponding focal parabola meets the circle at the real foci. and that the inverse of the cubic with respect to either of these points is a circular cubic with an axis of symmetry passing through two real foci and bisecting at right angles the line joining the other two real foci. Show and singular focus. 0).] lines joining four concyclic foci of a circular cubic in pairs are equally inclined to the real asymptote. 19.] we get the abscissae or Ex. and that the inverse of the cubic with respect to any one of these points is a circular cubic with an axis of symmetry on which the four real foci lie. and the required points are the points of contact with the circle ofHhe common tangents to circle and parabola. Show that a real circle with respect to which a onecircuited circular cubic is selfinverse meets it in two real points. 12. 0). = B (j3 + y)(y + a)(OC+P) {8x(x* + y') + 4 (a* + 2 + y* + 0y + yOC + a/3) X . 0). foci. and choosing [The directrices are found by differentiating m so as to make (iv) a linepair. 10. The circle inverts into the axis of symmetry. 13. + a0) 2 + 4 a0y (a + + y)} x\ [Find the foci of x {x 1 + y*) + ax + by + c = in the usual manner. [Parallel to the axes of the two parabolas through the foci. 0). Show that its equation is (0. We have a+ + +8 = 0. that the cubic (i) x2 +y2 = of § 2 selfinverse with respect to is and that the corresponding focal parabola (y + hf = b(2x + a).] Ex. (iii) [Eliminating y or x from § 3 (ii) and ordinates of the four foci on (iii). XIV 3 circular cubics can be found with four given concyclic real asymptotes are perpendicular. The centres of the four circles for which a circular cubic is selfinverse lie on a rectangular hyperbola whose asymptotes are the real asymptote of the cubic and the perpendicular from the singular focus.] Ex. Ex. 16. and them with (a. is if t is taken as Show h. have common. A twocircuited circular cubic has x as asymptote and three real foci A (a. 15. Show that the circle through the real foci of a twocircuited circular cubic meets it in four real points. a focal parabola in that all circular cubics having the same real asymptote and selfinverse with respect to the same point. &c] . (8. The directrices corresponding to four concyclic foci of a circular cubic form a pencil of the same crossratio as the pencil subtended by the foci at any point of the focal parabola through them. 17. C (y. The Ex. Ex. 18. (iii) is 2(x+p)y + b =0.
points Ex. 2 ) . F.. &c.EF + 2PE a PF)* <r(EF2 PE2 PF1 + 2PE.PA+ where P is any point of the curved 2 1 . Prove that (HB KWHC*. 22.l i Show that. Ex.PA±(OCi. = 0. B. F(t. 24. Show that. and eliminate k. by £+ »"i/ in Ex. 23. and invert the results of Ex. Ex. PB±(HA 2 HB2 PC = 0.. OA^iaP + P' + yi + Py + yOC + OLpyiliP + tiiy + ^iCX + p).XIV 3 FOCI OF CIECULAR CUBICS C on ) 225 a line meeting . A circular cubic has three real foci A.) . i/). C curvature at . C by substituting for PA ± PB. 19 is the singular focus. 20 ABC meets the cubic in U.> + y* + 2£yy + 4y(e + ')(2Z + y)}x*. The circle of the circle ABC again at H. 0). from Ex. Therefore replace a. A E x= Show as an asymptote and is that its equation + 2(Se. 20..PF)h. By inversion obtain a property of any onecircuited cubic.J+in are connected B(iir. circle A ABC meets the cubic in K. 0. 20..)} = {(2er. 21. 27. r l [(! + *>?> 0) are foci.X+r. C(y. twocircuited circular cubic has real foci A. Show that the distances of any point P from the points i(+«. onecircuited circular cubic has three real foci {£. E[£+Y t Xn). HW = (EB + HC) (SC + HA) (HA + HB).XiY). by the relation 1 2 2 (p + i<r)PA+(p. if in Ex. KB KA . 21. Q . W. .ia) PB = p (PE + PF . 20 or 21.OAi) PB ± (OAi . Similarly we may Ex. 8HU. Obtain a relation connecting the distances of any point P on the cubic of Ex. [Write down the conditions that the pair of tangents from any point should be a circular The asymptotes of the cubics are parallel to the axes of reference. . Ex. 22. 2 + 2fy + a) = y 2 + 2 gx + b linepair. + y*2(Xx + (X')h + tfa. Ex. F(£Y. B. 19 when rationalized. 19.. p and being any real constants.n* + y* + 2iy)x£(e + n* + y* + 2t.HV. 25 in Ex. 23. the asymptote in H. K and the meets ) + . Prove that (HB*HC*) PA + tfCPHA .] 2216 to k (a. as in Ex. 26. [Invert with respect to K.] Ex. 24 from the foci E. &c.] we have Ex. The locus of the foci of a conic through four given concyclic is a pair of circular cubics with the given points as foci. 1 ) (x2 + y2 2fix + ^)i + (a ! 0 2) (x + y 2 2yx + y i )i = equation of Ex. where [(/3' 2 P is any point > 2) (x* on the curve. and that (OBl . 25.OBi) PC = .OCh) . becomes the Ex.. V.). if in Ex. i.
Hence. the coefficients of x*. Since (74. if and B are any two points on the cubic. distance of P from a fixed point multiplied by a fixed line varies as the distance of P from another fixed line. B. C. where A. A . Find the locus of (i) P if CPU have a constant ratio. Any circular cubic passing through its own singular focus may be considered as the locus of a point P such that PA PB = PC PD. The tangents of the angles APB. z3 + z 2 (x + y) + xy(ax + by + 3cz) = tz 0.226 FOCI OF CIRCULAR CUBICS XIV 3 Ex. GB are the tangents at and B. Pencil of Tangents from a Point on a Cubic. the crossratio of the pencils of tangents from and B are the same. 28. Any circular cubic passing through its own singular focus considered as the locus of the intersection of any member of a given family of coaxial circles with its diameter through 0. . Hence the tangents from (ax 2 B 2 2 ) to the cubic are + 2 cxz + z 4 bxz2 (x + z) = . Let the tangents at and meet at G and take ABC as the triangle of reference. 30. B. may be [Take as origin and the x' circles as + y^+px+a = t(py — b). The theorem is so important that a straightforward proof is given here. if + 2ct + l)yz + bty* = 0. § 4. the equation of the cubic becomes A A A B . choosing suitable real homogeneous coordinates. The product of the tangents from to the product of the tangents from (iii) The square of the the distance of P from P to two fixed circles is equal P to two other fixed circles. only involving real quantities. (ii) D being fixed points. 29.] Ex. Ex. The theorem that the four tangents to a cubic from any point of the curve (excluding the tangent at the point) form a pencil of constant crossratio was proved in § 3 by means of unreal projection. It is sufficient to show that. A. y 2 z in the equation of the cubic are zero. y 3 x 2 z. 0. C. . D are fixed points. and therefore touches {at i + 2ct + \f4>U{t + \) = 0. The o line x = meets this cubic where (t+l) z 2 +{at 2 it.
[The four tangents from a> invert into tangents from a.(2a oab + 2b*)(a + b)27a?b 2 + 36ab(a + b)c 12(2a 2 + ab + 2b 2)c 2 36abc 3 + 24. Ch.] Ex. 2"= 12 {a 2 a& + 6 2 + 3a&c2(a + 6) and 6. forming a pencil with the same crossratio. (x + y + zf + 6 kxyz = of these tangents is 227 given by . or vice versa. the crossratio of tangents is the same for the curve and its inverse. B 1. Q 2 . only if A and B lie on the same circuit of the cubic] Ex. Discuss the case tacitly excluded in A goes through B. J=4.XIV By 5 THE CUBIC I. The equation of the inverse curve is § 2 (i) with h 2fc. We proved in § 1 that a nonsingular or acnodal cubic has Choose the tangents at these three real collinear inflexions. and The real foci are readily found. inflexions as the sides of the triangle of reference. and Ex. prove that the intersection of two such circles cannot be a focus. and choose real homogeneous coordinates such that the line of inflexions 0. § 11. § A 7.(a + b)c i 8c e Since these expressions for / and J are symmetrical in a 2 . the crossratios of the pencils of tangents from are the same. Another proof of the theorem is given in Ch. Can the theorem of § 4 be deduced ? P and P of the cubic P Show P [Strictly speaking. § Ex. by using the fact that touches the cubic at and passes through the points the polar conic of of contact of the tangents from P. The Cubic (x + y + z) 3 + 6kxyz = 0. and k 0. is x + y + z The most general equation of a cubic is — 3 xyz = ax + by s + cz 3 + 3yz(fy +pz) + 3zx (gz + qx) +3xy(hx + ry). 1. 2. A and B with the pencil that the pencils of tangents from two adjacent points have the same crossratio. 3. By taking as the intersection of the curve with a real circle for which it is selfinverse. 4 in which the tangent at [Compare the pencils of tangents from drawn from any other point of the curve. the crossratio I 3 (0+1) 2 (02)2 (0i)» where = 27 J»(^0 c2 + l) 3 + c 4 }. and that the four real foci of a harmonic onecircuited cubic lie on a circle for which the cubic is not selfinverse.] onecircuited if 4& = = > > §5. If a circular cubic is inverted with respect to a point O on the curve. The curve is harmonic if ah 2 a . . XVI.
0. on excluding x Jc iy i)Z = y = z = 0. The tangential equation 9 (fi . — f and x= y = = (x + z.2fiu + p\ + \. so that z = 0. y being not concurrent. — 1. = meets. the acnode being the point (1. + 4k ( .1.: 228 THE CUBIC (x + y + zf + Gkxyz = x XI V 5 With our choice (y + zf = 0. the real inflexional tangents x 0. 1. or a cubic with three concurrent real inflexional tangents) can be Any put in the form (x + y + z) 3 + 6Icxyz = by a real choice of homogeneous coordinates.1). . found as in Ch. 0.j.A) 2 (X . 1).vf {v . it is sufficient to consider the tangents drawn from the point (1. (—1. To find the crossratio of the pencil of tangents drawn to the cubic from any point of the curve. the cubic /= (x + 3 y + z) + 6kxyz = (i) has three and only three real inflexions (0. IV. —1). Proceeding now as in § 4 we have k(k § 3. is + 4) 3 {(<l> + l)(<t>2)(<!>$}* = (]c* + 6k + 6)Z(<t>><p + l)\ of the cubic. The curve (i) has a double point when = = = ix This gives. z (1. Hence An acnodal cubic can be put in the form y + z) 3 = 27xyz by a real choice of homogeneous coordinates. —1. 0).)l2k \ fi v = 0. Hence we cannot put the equation of a cubic with three real concurrent inflexional tangents or of a cubic with a cusp or crunode into the form (i). this = —f = Similarly for y = where and cubic (other than a crunodal cubic. a cuspidal cubic. Hence : of coordinates b c p. and therefore touches 4(* + m+ 3 2 /*) it if 6 (2 + h) t* +U = 0. Now t (x + y) z meets the curve where = {t + \) 3 (x + yy + Gktxy = 0. Conversely. 0).) !i !i 2 2 ( fti'2vX+\fi)(fiv + vX2Xfi. .
However. For what value of h ihe cubic (x + y + z) s + 6kxyz [(i) = the pencil of tangents from any point of (i) harmonic. Symmetry of Cubics. but this not true of a curve of the third degree. is A curve Firstly.] Ex. § 6. 2.] Ex. 2. on the lines 0. [See Ex.] s Ex. The point and tangential equations of an acnodal cubic can be put in the form Xi+pi + v. we have : A a cubic can be projected so as to be is its own reflexion in polar.] Ex. that in general two cubics with given inflexional tangents at three given collinear inflexions touch a given line. Under what circumstances is the triangle formed by the real inflexional tangents of a cubic selfconjugate with respect to one of the polar conies of the cubics ? [The polar conic of (1. 5. H=(x + y + z)(2yz + 2zx + 2xyx y The inflexions lie zi + 2kxyz = ) 0. 4. (ii) k = i. = 0. line. Given the tangents at three given collinear inflexions of a cubic. 7. 3. the cubics are m {l^x + m^y + iiyz) (l2 x + m 2 y + ii^z) (ls x + m 3 y + n 3 z) = {(hhhfi x + [mym^m^l y+ (n^n^i z} s . xi + yi'+zi = 0. 1. the locus of the remaining inflexions is two straight lines. For suppose / any real inflexion and I its harmonic . [If the line passes through an inflexion or the intersection of two of the given tangents.! = 0. when h = —3. 1. there is only one cubic] Show Ex. 3) are the inflexional tangents and + n( z = iy the three given points are the vertices of the triangle of reference. Find the is inflexions of the cubic /= (x + y + z) [The Hessian +6kxyz t = i 0.XIV 6 SYMMETRY OF CUBICS is 229 Ex. of the second degree has always symmetry. 6. Show that nine cubics can be drawn in general to pass through three given points. 1). (ii) equianharrnonic ? & 2 + 6& + 6 2. a cubic may always be projected into a symmetrical curve. Mention any exceptions. any value of the cube roots being taken. given three inflexional tangents and the fact that the three corresponding inflexions are collinear.] f3H=4:(x + y + z)(x + ay + a>''z)(x + ca*y + B)z) = Ex. [If l t x + (i = 1.
These two theorems hold for nodal and cuspidal cubics also. Aa/ c. p.. 412) Any cubic (other than a crunodal or cuspidal cubic) can be projected so as to have the symmetry of an equilateral triangle. if we project / to infinity and the angle between I and IR into 90°. Again. In fact project the cubic so that the line of real inflexions is at infinity.: 230 SYMMETRY OF CUBICS XIV 6 Let any line through I meet the cubic again in P. A cubic can be projected so as to be symmetrical about a centre. PQ will be bisected at right angles at R.  I Pig.. * 4. Then (IR. is the result due to Clifford (Collected Works. Q and I in R. More interesting. and the triangle formed by the real inflexional . 7 will be the middle point ry. PQ) is harmonic. Hence. *» = I..\ For or if we project I to infinity. perhaps.
8. .XIV 6 SYMMETRY OF CUBICS 231 : —i 8) k K = —2*3 56' Fig.
Fig.232 tangents SYMMETRY OF CUBICS XIV 6 is equilateral. we proved in § 5 that the equation becomes + y + s) 3 + 6Jcxyz from which the symmetry is obvious. fc = '72 . m = l. taking areal coordinates. The case in which the inflexional tangents are concurrent wants separate discussion. Then. 10. (x = 0. In this case project the curve so that the line of real inflexions is at infinity and the angles between the inflexional .
e. e. and m= —\ 'S when nWlicii k — 9 2 . 4(m — l) = fc(4m — 2m + l) . .. y curve in no finite point the equation of the curve is of the form tangents are 120°. y(3x2 —y 2) or in polar coordinates = a3 . . have (4m 2 + l)J 4(ml) s = m We 2m =  §7. The cubic has one or two circuits according as y z meets the curve in one or three real points. z = X+Y2mZ.. = — . XIV 7 THE CUBIC x s + y 3 + z 3 + 6mxyz = 233 Take their intersection as origin and one of them as the axis of x. according as h is greater than or less than — f Diagrams of the cubic" for various values of k are given in Figures 4 to 10. x3 + y 3 '+ z 3 + Omxyz = by a real choice of homogeneous coordinates. y 3 2 (ii) This will be lawful except in the cases m=1 (x + when k = 0.. The and the dotted lines are explained in § 7.. . &c. i. 2(1— m)(2m + l)X = (2m + l)x + y + z. r 3 sin 3 6 = a3 from which the symmetry is obvious. m {2(2Jfe + 9)*+(2ifc)*} > {2(2/c 9)52(2/c)i}. (ii). We by a x proved in § 5 that a cubic can be put in the form (x + y + z) 3 + 6kxyz = (i) real choice of homogeneous coordinates. and since the asymptotes meet the 0. = X2mY+Z. Then the asymptotes of the curve are y + /3 a. Then straightforward reduction gives 3 y + z) + 6kxyz = Hence if we choose m as the real root of + ^^ +1 (X 3 +Y 3 + Z 3 + 6mXYZ). In this equation put = 2mX + Y+Z. i. where i. Every real nonsingular cubic can put in the form. The Cubic x 3 '+y 3 + z 3 + 6mxyz be = 0.e.
The sides of the triangle of reference for the cubic (iii) are given by the dotted lines of Figures 4 to 10. x3 + f + z3 + 6mxyz x. now consider the cases Firstly. . • = §8. Z=2$a2y. + 6mxyz = <>z : (iii) ^ <>x = ^ = ^. a> 2 z) + (8m 3 + l)a: 3 = which shows that 0..234 THE CUBIC Y. X = 2?a + &x + y.. which gives us an acnodal the above argument when k cubic (§ 5). z = XI V 7 and replace X. § 1. .. We y(Sx2 — y 2 ) and putting = a3 . * By Ch. we note that our argument does not apply to the cubic for which the tangents at the three real collinear inflexions are concurrent for such a cubic cannot be put in the form (i). (i) intersections with the sides of the triangle of reference. showed in § 6 that it can be put in the form = . satisfying this equation makes / split But up into three linear factors. are inflexional tangents whose points of contact lie on x = 0. which gives on elimination of x y z that a value of m 8m + 1 = 3 . . may be written ( — 2mx + y + z) — %mx + a> 2 y + coz) . we have X3 +Y3 + Z 3 + 6XYZ= 18cs 3 18i/ 3 + 54a. . as the sides of the new triangle of reference.f Hence (iii) cannot be a unicursal cubic unless it is three straight lines. . r ' 3 + y3 + z 3 m= —\. we note that. Y = &a&x + y. Z by * we have and the cubic in the required form. y. .= <>y : at the double point. * See the beginning of § 8. #c. I. — 2mx + m 2 y + mz = — 2mx + ay + a> 2 z = 0. if We m=1 /= x has a double point.2mx + a>y + — 2mx + y + z = 0.a 2/ = 0. This explains the failure of — •§. The are its inflexions of xs + y 3 + z 3 + 6mxyz In fact the equation ( ( = .0. this is equivalent to choosing fresh homogeneous coordinates with (2m + l)x + y + s = 0. Hence the case on 1 gives us the cubic in which the tangents at three real collinear inflexions are concurrent. Secondly.
0. cox2my + co 2 z — 2 x + y—2mz = ) cox + co y — 2mz = 2 0. (3) no two of the three inflexions lie either in the same row or in the same column.0). Another proof It will obtained by writing the cubic down the polar any intersection of and x = 0. is point of (i) the crossratio of the pencil of tangents from any to the curve. (0.1. {(</> 64m3 (m3 l) 3 + 1) (*2) (</>£)} 2 2 (t/> = (8m° + 2Qm 3 . 1. (0.1).1). 0). The nine inflexions of (i) are given by the scheme = 1.«.0). 2 0. co 2 x + coy — 2mz = Oj The harmonic polars of the inflexions are y = z. (0)2. lie y = coz. (1. . —2mx + coy + co 2 z = 0. (0. (1. 2 co. co 2 .+ 1) 3 </> . (».o> 2 .1). 2 (l. 2 ). co x — 2my + coz = 0. (0.(>.1) 2 . — 1.(1. (0.1). (I. co. co.0.«M). — 2mx + co*y + coz = x — 2my + z = 0.1)) co).0. x — y. (0.0). 1). — 2ma: + y + z = being the inflexional tangent and y z the harmonic polar of the inflexion.. (co. I. x co2 y) lines 2 co x + y + z x + co2 y + x + y + co 2 z The x y inflexions by threes on twelve = =o z = x+y+z x + coy + co 2 z x + co 2 y + coz = = — +y+z x + coy + z x + y + coz = — = 0.a>).. 0. z = cox. for instance (0. = 0] = = Oj L Three inflexions are collinear if (]») they lie in the same row of the inflexionsscheme. (a. (1. • be found to be (—2mx + y + z)(yz) = Since this degenerates.1. namely the nine inflexions and the twelve critic centres of the cubic ' ' (1. If <j> (1.1. 2 (1. co ). (a. co . (2) they lie in the same column.co\0)\ The inflexional tangents are 0) 0\ ~2mx + y + z = 0.1. 1) is an inflexion. cox y z = = co co 2 2 z\ xl .1. Another proof is given in § 9. co ).co 2 )) . ») (a. z — x.XI V 8 THE CUBIC is x 3 + y 3 + z 3 + 6 mxyz = 235 conic of 1).1. x = coy. The twelve lines intersect in twentyone points. co)) (1. .l)f.
z 0. not crunodal or cuspidal or having three real concurrent inflexional tangents. 1). &c. that the equation of a 9p + 1 cubic can be put into this form in twentyfour different ways. 0. and equianharmonic We p= Ex.] For the caso of any cubic. 4.y. The inflexions of the cubic of Ex. can be put in the form [Replace x3 + y 3 +z = h (x + y + z) s x. .curye or by putting . noting that <f> is only real (and not — 1. [Consider the tangents from (1. Ex. § 291. XVI. 0. 15. 5. 1. tion (ii) The curve has two as may circuits if < — \ and one circuit if be deduced from § 6 by means of equa. As an Ex. m V v* = 0. 1. the 0. 0) to the. 1. Find the crossratio of the tangents from any point of y*z + x'y '+ z^x + Gpxyz = 0. show in Ch.z by y + z. or 2). can be put in the form y^z + yz2 + z 2 x + zx 2 + x?y + a>y2 + 6lxyz = 0. see Elliott's Algebra ofQuanUcs. 3 lie on (x + y + z) (x + ay + a> 2 z) (x + a> 2 y + a>z) = 0. . z + x. \.] l harmonic s = 2l6p" + B6p 3 + = (0. Show that the equation of any cubic. 2. lines x 0.] !i .24m 2 A/«/ (A 3 + ^ + v 3) . y inflexional tangents being y The = = = +z = each pass through a real inflexion. x + y in § 7 (i). not crunodal or cuspidal. 0). § 6. (ii) in the expression for the crossratio found in that section. Find the crossratio of the tangents from any point of the curve. Ex.* The tangential equation of the curve is \ e + fi 6 + v a . 0).(32m3 + 2) (ji* v3 + v 3 X 3 + A 3 /* 3 ) . exercise the reader may obtain this result from the expression just obtained for (/>. &c. 3. if [Consider the tangents from or 0.] Ex. Find the crossratio of the pencil formed by the tangents from any point of the cubic of Ex. . above. if the cubic has two circuits.(48m 4 +24m) A m>— \. The pencil of tangents if is 0. 3.236 THE CUBIC x3 + y 3 + z 3 + 6mxyz = XIV 8 This may be proved as in § 5 by considering the pencil of tangents from (—1. ifl=—l. [Consider the tangents from * ( —1 . The cubic is acnodal. Ex. Show that the equation of any cubic. k = 4(ml) 3 T (4m 2 2m+l) .
any cubic through eight inflexions is f+hH= 0]. Hessian meet where xyz = i. the polar conic passes through the four intersections of £ks + ijy' + (z* = and %yz + r)zx + £xy = 0. £). z). [The roots of the equation 3 8ifcf 2m3 + 6Mm* + l = in + l< m are all real. 3. e. If the pencil of tangents from any point of a cubic the cubic is the Hessian of its Hessian.] Ex. . as proved before. line of a point on a cubic with respect to any syzygetie cubic passes through the tangential of P.: XIV 9 SYZYGETIC CUBICS ' 237 § 9. 5. The harmonic polars of the inflexions are the same for all the syzygetie cubics of the family. The polar conic of a point P with respect to any cubic of a syzygetie family goes through four fixed points lying on that cubic of the family which passes through P. zx3 — zy3 ). yz 3 — yx 3 . cubics found by varying in the equation (i) have the same inflexions they are called a family of syzygetie cubics. y. . A real cubic is the Hessian of one or three real cubics according it has one or two circuits. [If P is (£.] Ex. [If The polar (x. [8m 6 + 20m 3 — 1 = 0. 2. . The curve and its We see that A cubic has the same inflexions as its The family of all . 1. All cubics through eight given inflexions of a ctabie pass through the ninth inflexion and are syzygetie with it. (ii). if 0. Since (ii) is of the third degree when considered as an equation in m. Ex. The Hessian is of x* + y 3 + z 3 + Gmxyz = = . Syzygetie Cubics. nonsingular cubic is the Hessian of three syzygetie as Ex. Any cubics.] . The Hessian of any member of the family m belongs to the family. the inflexions of the cubic lie on the sides of the triangle of reference. 4. Hessian. (i) at once found to be x3 + y 3 + z 3 + GMxyz where M = (2m 3 + l)/6m 2 .] is Ex. harmonic. is the cubic and ff = [If / = its Hessian. P P is the tangential of P is (xy^ — xz3 . We may call the cubic 'harmonic'. i).
Q. Both conies touch 6 and c at their intersections with o.+y^y^y^ + zx z^zs I. § % Ex.] is Ex. cubic If The polar I of with respect to the polar conic of Q for a the same as the polar of Q with respect to the polar conic of P. Its real inflexional tangents are concurrent at a critic centre. Ex. If the pencil of tangents from any point of a cubic anharmonic. If it is taken as triangle of reference. and Q lie on the same cubic of the syzygetic family. and P is one of the critic centres referred to in Ex. is Ex.] ' Ex. 0). We may call the cubic equianharmonic '. ysi*j). b. The tangents at two inflexions of a cubic meet on the harmonic polar of the inflexion collinear with the two given inflexions. 7. the coefficient of xyz in the equation of the 3ic is zero. [m s (m s — 1) = 0. c. The sides of the triangle are the degenerate Hessian. 1) and (1. or (2) the cubic is equianharmonic.] ' equi Ex. (0. [If a. VII. 10. Ex. fixed and the cubic is any member of a syzygetic goes through a fixed point. One species has a Hessian consisting of three real lines. R is symmetric. show that the cubic is § 8. y = 0. 14. 1) respectively.] 12. P family. z = 0.] . 1. Under what conditions cubic a coincident linepair ? the polar conic of a point P for a [Either (1) the cubic is cuspidal. If any line meets X s + y3 + a 3 + Gmxyz (*i»yi.] ' Ex. the intersection is (1. Ex. 0.238 SYZYGETIC CUBICS XIV is 9 Ex. Obtain the equations of the sides of the triangle for the cubic of and p = — 3~$ . If R is any point on I. = (^3»y3 . 7. Each real inflexional tangent meets two unreal inflexional tangents at a critic centre which is a vertex of the triangle formed by the Hessian. with [If its The intersection of the tangent at an inflexion of a cubic harmonic polar lies on the Hessian. in 13. If all polar conies for triangle. = 0. m). 9. 6. 0). m. 11. [to = and w? — 1 in the two cases. 1. and the conies are x' i x* + 2myz=0 + 8myz = 0. The critic centres are (1. . The tangents at the inflexions on 6 and the tangents at the inflexions on c form two triangles whose vertices lie on a conic and whose sides touch a conic. Ex. 8. The inflexions of a cubic lie by threes on the lines a. (0. 0. If this point is on PQ. Ex. (0. the relation between P. Note list of Errata'. 8. in the cases p = equianharmonic of the second and first species respectively.2i)i («j.%)i then [See Oh. c are x = 0. 1) is the inflexion. 6. x^^x. 15. P and Q are I P [PQR is an ' apolar triangle of the 3ic. and Ch.] a cubic have a common selfconjugate the cubic is equianharmonic. the Hessian degenerates. 2. There are two species of equianharmonic cubic. and Pis the cusp or the intersection of the cuspidal and inflexional tangents. The other species has a Hessian consisting of one real and two unreal lines. § 3. —1.
0.] . &c] [If the sides of the triangles are x +y +z = 0. 22 pass through the same two unreal points on the line of real inflexions. 0. 18.1. o> 1). 17. z = 0.1.0). The pencil of four lines joining an inflexion of a cubic with the other eight inflexions is equianharmonic.] Ex.0). Show that these six tangents touch a conic. 20. and the remaining inflexions lie on two lines through A. and that the six tangents touch a conic. are (1. Two lines each Show that the + y 3 + z3 + 6 mxyz + hx (x + 2 myz) 1 = is a linetrio. [All the conies of Ex. (1. (1. is (0. From each intersection of two real inflexional tangents of a twocircuited cubic two real tangents can be drawn to the oval.1. with their harmonic polars are [If the inflexions are the inter sections of x+y +z = (2.0).] passing through three inflexions of a cubic points of contact of the tangents from O lie by pairs on three lines through the remaining inflexions.»l.2. [Ex. The double points of the involution (a>. (o. Ex.] The three inflexions with their inflexions. 1). 22. The harmonic polars of three collinear inflexions of a cubic pass through the same critic centre A. to any one of the collinear inflexions is divided The line joining harmonically by the cubic.] Ex. y = 0. intersections of the line through three collinear harmonic polars form an involution with those (0. and no two sides of the same triangle pass through the same inflexion. 0). — 1.] Ex. x = 0. The three lines joining two real critic centres and passing through a real inflexion meet the real inflexional tangents at those inflexions and at six points lying on a conic. [If the inflexion (0. Show that from each vertex of the triangle two real tangents can be drawn. 4 to 10. 15. (w 2 w. 2 . that the six points of contact lie on a conic. 19. and that the axes of perspective each pass through three inflexions.0. that the tangents meet the cubic again in six points on a conic.1. that their points of contact lie on a conic. meet at \a? O.0). and x + wy + aPz = 0. 1. Two triangles are such that each side of either triangle passes through three inflexions of a cubic.1.0). (1.XIV 9 SYZYGETIG CUBICS 239 Ex. The three real lines each passing through one real and two unreal inflexions of a twocircuited cubic form a triangle. 21. 1. 20 to 24 are obvious from Fig. if 8m Ex.1).. . x + a'y + az = the axes of perspective are ax + y +s = 0. A [Take the collinear inflexions as those on x = 0. 1). Ex.1). 1). the lines meet z — at (1. Then A is (1. Ex. and that the intersections of each pair of tangents with the third inflexional tangent lie on a conic. .2). 20.1). 21. Show that the two triangles are in perspective in six ways. 16. (a. s (& + 3) 3 + 27(fc + l)=0.
240 SYZYGETIC CUBICS XIV 9 Ex. 24. 28. and the chords of contact of conjugate pairs of the involution pass through I. meeting at (a. c). /= H= /= 0.] . Soc. and the nine inflexional tangents are concurrent at a point of the quintic. 15.] doubly infinite family of quintics passes through the nine Ex. § 4. ' ' Ex l{l' = ' l). touches the Hessian at three points. [(i) See Ch. (b + c)x= a(y + z). The inflexional tangents are (b+cu>)x = a(y + az). 23. 6.. [The poloconic of \x + fjy + nz = 1 is XX' (mrx — yz) + . It is possible to project a family of syzygetic cubics so that each curve has the symmetry of an equilateral triangle. If the equation of a cubic is given.] Ex. Show that I (i) The polo conic of and V is the locus of the pole of I with respect to the polar conic of any point on V.. Each such cubic has three collinear inflexions and corresponding inflexional tangents in common with the given cubic. Ex. (b + ca 2 )x= a(y + a> 2 z). and (ii) A syzygetic family of cuspidal cubics may be projected so as to be all congruent and have a common asymptote which is the cuspidal tangent of each curve. + and (fiv X'a. The six tangents drawn to a cubic from a point on the harmonic polar of an inflexion I form an involution. A [The quintic is ayz(y s + s3 2x*) + bzx(z 3 + x*2y s )+cxy(x 3 + y s 2ss)= 0.. the locus of ' The poloconic of the lines I and I' with respect to a cubic is a. it will be a linetrio Find k such that f+kH=Q on which the quartic through the twelve critic centres of a cubic has Ex. See Roberts. [Let be the Hessian of coincides with its Hessian then inflexions lie. xiii (1882). 30.] Ex. + p'y + v'z = + n'v) (mx' r. 27. show that a cubic and (i) A syzygetic family of nodal cubics may be projected so as to be all similar and similarly situated with the node as centre of similitude.m'yz) + . &c. lie The poloconic of I I touches the poloconic of I.. 26.+ . inflexions and the twelve critic centres of a cubic. p.. (ii) The polar line of any point on (iii) and vice versa. Defining a syzygetic family of cubics by the equation = its Hessian. [There are also twelve cubics such that the tangents from any point on one of them to the given cubic form an involution.. The configuration formed by the nine inflexions and the twelve lines joining them is dualistic to the configuration formed by the nine harmonic polars of the inflexions and the twelve critic centres. 26. + . XIII.] Ex. prove that the inflexions can be found without solving any equation of degree higher than 4. The poloconic of I is the locus of a point whose polar conic touches 25. Each such quiittic has an inflexion at every inflexion of the cubic. inflexion at each critic centre.] . A an [The quartic is ax 3 (y — z*) + by (z* — xs ) + cz (xs — y s = ) 0. London Math. where /= is H f+kH=Q.. 29.] Ex. = 0. point whose polar conic has I and V as conjugate lines. (ii) Project into (xk) y* = 1. which on the poloconic of I and V.. Proc.
CHAPTER XV CUBICS AS JACOBIANS § 1. means the putting x' for x. y' for y. y'. These polars are concurrent (at Q . z) with respect to the conies are W 7>U <>x  7>U 7>y IU <sz and two similar equations. V = 0. where ^—. Suppose that U = 0. = are the equations in homogeneous coordinates of three conies not passing through the same four points. z' for z in result of — <sx &c. The polars of the point P (a/. Jacobian of three Conies.
V . with the Jacobian of IT = 0. Z).V+^W. this case the equation of their U= = W= : The Jacobian of any three conies of the family XU+pV+vW = (which do not pass through the same four points) = 0. for the polars of any point with respect to all conies through four given points are concurrent. is = ^(X. *U Tz + 1 W+ iz v 3F =0 n *z  Eliminating X:pi:i> between these equations.lT+^V + h. i>'s. \U + ^V+v^W) z) d (x.e. we see that (X. Y. .W^O. V = 0. Y. if 7>U i. y. a linepair meeting at (X. Z) must lie on the Jacobian . and the centre of any degenerate conic must lie on the Jacobian. the lines and are har AP A AQ . Z). the equations give one and only one value of the ratio X: fi:v.W) S(a3. V 0. where the are constants. .: 242 JACOBIAN OF THREE CONICS of the conies XV 1 The Jacobian \ 1 U+p1 1 V+v W=0. If the conies do not pass through the same four points. z) K as is Hi *(U. their Jacobian is identical with the Jacobian of the original conies 0. Hence 0. If these conies all pass through the same four points.V + fi. Q are conjugate points on the Jacobian. W is identical is The conic X U + fiV+vW = has a node at (X. as is geometrically obvious. X. X 2 U+ft 2 V+v 2 W=0. /is. and is any other point on the Jacobian. at once verified by the rule for multiplying determinants. and if this is the case. If P. *3 H In Jacobian vanishes identically. X 2 U+fi 2 V+u 2 W. Hence Any point on the Jacobian of a family of conies is the centre of one and only one degenerate conic of tJie family. X's. X ZX + 7>V > X *X + 3TT V *X = > n ' 7>U IV iW » x . A i hi A„ fi 2 "i v2 "s = 0. y. T.
£z) = ax(y. CA. [Into the common pole and polar. incident linepair at infinity is the rectangular hyperbola through the to the conic together with the line at infinity. 2. C in common. AB. 5. Ex.] common selfconjugate Ex. C. (i = 1. their Jacobian degenerates into the lines BC. 3. their Jacobian has [Take A as (1. If three conies pass through the points A. IV.] e2 . B. For AQ is the polar of P with respect to the degenerate conic. and a coincident linepair and a conic. their Jacobian common polar and two lines through the pole. \xi + y l + = 0. If three conies have a eom'mon selfconjugate triangle. Ex. If three conies have a degenerates. Ex.] Ex. their Jacobian has a node at A and one branch of the Jacobian touches the two conies at A. 7.] conic. If three conies have points B. 0)*. 0. 4. Find the circle orthogonal to x 2 + tf + Sx + \h = 0. 6. 1. 7. z2 = by (x .] Ex. Ex. Ex.rjz).] Ex. circle orthogonal to The Jacobian of three them all. Ex. 4. If three conies pass through A and two touch at A.] feet of the normals from The Jacobian of a P [The Jacobian of ax* is + bya = z>.] sides of the common selfconjugate triangle of any two conies may be obtained by forming the Jacobian of the two conies and their harmonic locus. A in common. [Into the line 11. and a coEx. The Jacobian of two conies degenerates. a circle with centre P. their Jacobian degenerates into the line BC and a conic through B and C. their Jacobian degenerates into the sides of that triangle. 10. The triangle. Now put z = 1.] ' Ex. [The two conies and their harmonic locus have a See Ch. If three conies have a point a node at A. 2. [See Ex. 2x2y7 = 0.XV 1 JACOBIAN OF THREE CONICS 243 monic conjugates with respect to the degenerate conic with centre A. x i + y i + 2x\2y + 2l xi + if2x + 4:y + \ = 0. 3). 8. x i ^y l 2^xy2r yz + cz 2 t = 0. 9. § 4. x* circles is the line at infinity and the [Take the circles as + y* + 2g x + 2f y + c = i i Q.
[See Ex. and a coincident linepair at infinity. forming the harmonic locus of the conic and the circular points. T ne locus of the point of contact of two members of the = is their Jacobian. polar curve. 0. rj. XI V. of the point Then the P (£. § 9. The Jacobian degenerates into the axes and the line at infinity by Ex. e. the b = first the equation of a cubic. [Take A as the point (1. z) polar conic. Hence: § 3. If the Jacobian of three conies passes through the intersection of two of them.] Ex. 0). 15. y. 13.] We Ex. nonsingular cubic. i. [Into the chord of contact and a conic dividing the chord of contact harmonically. its director circle. The Jacobian of two concentric circles and any conic rectangular hyperbola and the line at infinity. = Now any cubic * is 0. 9. thus obtain the tangential equation of the circular points. The Jacobian of two conies having double contact and any third conic degenerates.] A Ex. 14. e. Then. See Ex. family X U + /x V + v W [See Ex. [We suppose that we know the equation of one circle and of the line at infinity. z) the Hessian of three different cubics i. by Ch.244 JACOBIAN OF THREE CONICS XV 1 Ex. . 16. is Suppose f(x. the two conies touch at A. 15. 12. 13. y. The axes of a conic (using any coordinates) may be obtained by writing down the Jacobian of the conic. 11. ly v= 30 0. Any Cubic as a Jacobian of three Conies. we have the director circle. () is (i) i>x <>y * <>z The Jacobian of any three polar conies Jacobian of Ix is the same as the »/_„ o. is a Discuss the case in which the conic is a circle. and is therefore 32 / ay <*xhy _ay_ (>X <>Z <>x ay f>y <>x ly* iylz = o.] § 2.] Ex. ay <>z ay lz> ix <sz<)y •which is the Hessian of f(x.
cubic is uniquely given when three pairs of conjugate points are given. A Expressing the fact that the conic ax* + by 2 + c2 2 + 2fyz + 2gzx + 2 hxy = has the three pairs of points as conjugate pairs. Z). Hence : Q should be The polar conic of a point on the Hessian of a given cubic a linepair meeting at the conjugate point of the Hessian considered as the Jacobian of the polar conies of the given is cubic. If Consider a given cubic as the Jacobian of a family of conies. so is the and R' in Fig. e. and P'.. XV # 3 RULERCONSTRUCTION FOR CUBICS 245 Any cubic may be considered as the Jacobian of a family of conies. Suppose now that the polars of P with dz respect to ox oy are concurrent at the point Q (X. Rulerconstruction for Cubics. PQ' and P'Q. so are the intersections of Q' are conjugate pairs of points on a PQ and P'Q'. P and Q are / 2 . y/ a2/ . * / . so that conjugate points of the Jacobian. 1 are conthird pair of vertices. a quadrilateral are conjugate with respect to a conic. are conjugate points on* ' P R the cubic. „ *f „ . and one vertex into the centre. > a2 / 2>'¥ _n 32 f But equations (ii) are the conditions that a double point of the conic (i). and that in three different ways. and For But when two pairs of opposite vertices of so are Q and Q'. > a2 / _ n .. we get three * Hesse's circle It may be proved by projecting the conic into a Theorem. and J" are conjugate for any conic of the family. Y. Then a2 . i.* Hence jugate for any conic of the family. . § 3. Denote conjugate pairs Q and Q'. of points on the cubic by P P and P\ Q and cubic.
Hence f \U+fiV+vW = A B C D AB D . where A. SS' meet at P'. a cubic (Fig.: : 246 EULERCONSTRUCTION FOR CUBICS XV 3 linear relations between a. g. C and are the three conjugate pairs of points. v are constants. at The tangents at conjugate points Q and Q' of a cubic Trtteet a point P on the curve. therefore. and B'. Ex. by the construction may be finite. See Ch. The Jacobian of this family is the cubic required. Q and R are consecutive points on the cubic. 9. Since the four points of contact can be grouped into two pairs in three different ways. From G and C". c. XVI. and D' we obtain similarly a fifth conjugate pair and so on. of the four tangents from any point the points of contact are two conjugate pairs Q and Q'. general construct ruler only. then the intersection of and A'B'* and the intersection D' of AB' and A'B are a fourth conjugate pair. * f We have tacitly assumed that D does not coincide with C or We do not necessarily obtain the whole cubic in this way. number C'. and QQ' meets the curve again at the point P' conjugate to P. we can in an infinite number of conjugate pairs by 1 Suppose that in Fig. we see that If. then the conjugate points Q' and Bf are also conWe obtain then the result secutive points on the curve. /i. The points of a given cubic can be divided into conjugate pairs in three different ways. h and thus reduce the conic to the form 0. 8 and tf. b. Suppose that and A'. . For special positions of the three given pairs the of conjugate pairs obtained § 6.f Given three pairs of conjugate points on a cubic. we draw P 2). and the lines QQ'.
P'Q' and with respect to P'S.XV 3 RULERCONSTRUCTION FOR CUBICS 247 This also follows from the fact that the cubic is the Hessian of three different cubics. XIV. P' (pp'. [See Ch. 1. The real points of a cubic with one circuit can be divided into The real points of a cubic with two conjugate pairs in only one way. The tangent at P' is the harmonic conjugate of PP" with respect to the linepair QQ\ SS'. P'S'.] . In one way circuits can be dividedinto conjugate pairs in three ways. This shows that the linepair is QQ'. QS) is harmonic and rp' is the tangent at P' in the limit. . By §2 the polar conic of with respect to a cubic for which the given cubic is the Hessian is a linepair meeting at P'r Also by § 1 this linepair is harmonically conjugate with P Pig. For if p is a point on the cubic close to P. respect to P'Q. SS'. and is therefore the Jacobian of three different families of conies. § 1. In the other two ways one point lies on each circuit. both points of the pair lie on the same circuit. Ex.
if we take the vertices as (1. 3. where & = 0L + \ir. 4 the tangents at the vertices of the quadrangle meet on the curve at (a. Cayleyan of Cubic. Ch. If three polar conies of a cubic are rectangular hyperbolas. Show that.y 1 ) [The quadrangle is QQ SS' in Fig. Suppose now that P. P (i) and are P on the Hessian of Ch. XIII.] meet on the curve. lie 4. Any two conjugate points 4 (ii) of the cubic of Ch. XIII.0).b. b.c) meet on the curve at (1/a. [The cubic is the Jacobian of conies having the given pairs as conjugate pairs.] See also Ex. 1/c).0). Ex. cos by sinh and cosh. 7. Ex. of contact of the tangents from O. 5. sin a. Show that the crossratios of the pencil of tangents from any point of the cubic of. 3 sin 3 a) and —3 sin 3/3). c).1). If the tangents at four points of a cubic [Consider the polar conic of (a. P and P'. and the polar conic of P degenerates into a linepair through PFor the cubic of Ch. 2. [See Ch. every polar conic is a rectangular hyperbola and the Hessian is a circular cubic whose singular focus lies on the Hessian. . XIII. 4 are (c2 — a?) r (c a — b 2 ). (a.] § Ex. 3. b. 10. VII. It may be considered called the Cayleyan of the given cubic. while § 4.] Ex. 1/6. Ex.] 1 = 0. § 4 (iii) and its Hessian we replace sin and = OL + ^ni. the cubic has the equation ax{y*z>) + hy (a» .] Ex. 9. c). 2).Ex. CC) is an involution pencil is a cubic having the given pairs as conjugate pairs of points. Ex. &c. Given three pairs of points A and A'. + 1). (0. 8. CandC'. sin 0. Any transversal through a fixed point O of a cubic meets the curve again at P and Q. VII. Ex. The tangents from P to the Hessian touch at two points collinear with P. the polar conic of any one has the triangle formed by the other three as a selfconjugate triangle. § 5. BB'. XV 3 Ex.248 RULERCONSTRUCTION FOR CUBICS . 2 (vi). 2. It also touches the cubic at the tangential of O and cuts it at the points . The double rays of the involution are the linepair of the family of conies through P.0. the locus of P such that P (AA'. (cos/3. Show that in Ex. § 4 (cos a.0. + 1.a?2 ) + cz (x* . 6. P' are conjugate points on the Hessian The envelope of the line PP' is of a given cubic (Fig. Show that the Iocub of the intersection of the and Q is a trinodal quartic.1. (0. § [The circular points are conjugate points on the 'Hessian. tangents at P [The locus has evidently nodes at the points conjugate to O in the three ways of dividing the points of the cubic into conjugate pairs. A quadrangle is inscribed in a cubic so that its diagonal points on the curve. and that the ^angents at (1.
1 suppose R. Q'. we have. But by the harmonic properties of the quadrilateral (P'E. §§ . Q' on the Hessian of a cubic are harmonic conjugates with respect to the point of contact of QQ' with the Cayleyan and the remaining intersection of QQ' with the Hessian. zx. * We see that the Steinerian 8. For in Fig. 2mgfiv — r)v2 — {/jiz = of a cubic coincide. For in Fig. of yz. is i(x 2 + 2myz) + i (y + 2mzx) + ({z i + 2mxy) = If one of the lines into which this conic degenerates \x + py + vz the conic is = 0. and Hessian See Ch. The tangential equation of the Cayleyan of x3 + y 3 + z 3 + 6 mxyz — W \ 3 + u 3 + v3 + (l4 ! m )\uv=0. Tangential Equation of Cayleyan of Cubic. QQ') is a harmonic range in the limit. ^) on the Hessian degenerates. For the polar conic of P is the linepair QQ'. 8&. and the intersection of QQ'. RR' is the point of contact of QQ' with its envelope. This polar conic T] is 0. 9. Any two conjugate points Q.XV 5 TANGENTIAL EQUATION OF CAYLEYAN P 249 as the envelope of the line joining any point on the Hessian to the centre of the degenerate polar conic of with respect to the given cubic* P the tangents from The Cayleyan of a cubic is of the third class. P E § 5. Then f is the third intersection of QQ' with the Hessian. QQ'. 2 P' to the Cayleyan are PP'. VII. rj. SS' by § 3. R' the pair of conjugate points on the Hessian consecutive to Q. 3 We proved in § 4 that the Cayleyan touches the lines into which the polar conic of a point (£. (Xx + fiy + vz) (£x/\ Comparing the coefficients + rjy/fi + (z/v) = 0. xy. Each of the lines of the degenerate polar conic of any point on the Hessian of a cubic touches the Cayleyan. and two similar equations.
] Ex. 2. is Or use the The envelope of a line divided in involution by three given of the third class. i. conies 2. which is the tangential equation of the Cayleyan.] Ex. {p Ex. P consfecutive to an inflexion in 1). 8. 5 Eliminating f we get 3 /j. and Cayleyan. tangential equation of the Cayleyan.] Ex. Ex. VII. \3 + +v 3 +. + v) (" + ty + /*) = • The Cayleyan touches the [The polar conies of the inflexions touch the Cayleyan. The [The Cayleyan of a cubic having the given conies as polar conies. inflexional tangents of a cubic. result may also be obtained by considering the tangents to the locus from any intersection of two of the conies. The Cayleyan of a nonsingular cubic is a curve of the sixth degree with nine cusps. 3. The cuspidal tangents are the harmonic polars of the inflexions of the given cubic. The poles of any line with respect to a cubic form the vertices of a quadrangle whose sides touch the Cayleyan and whose diagonal points lie on the Hessian. 6. Discuss the case m= in § 5. Or verify that the two curves touch at the points (m.t>. § 2.] Ex. 5. m. same way [The cuspidal tangents and cusps of the Cayleyan are obtained in the as the inflexions and inflexional tangents of x s + y 3 + z s +Gmscyz = 0. 1. 9. Ex. Ex.250 TANGENTIAL EQUATION OF CAYLEYAN XV ^. Fig. &c] Ex. For a particular case of the reciprocal of this theorem see § 3. . If A and A! are conjugate points on the Hessian. 4. Exactly similarly we show that The tangential equation of the Cayleyan of (x + is y + z) (* 3 + 6 kxyz = 2 (k + 4<)\(ji. The Cayleyan and Hessian touch at the point conjugate to any inflexion [Make P of the Hessian. Ex. 5. the polar of A with respect to the polar conic of A! for any syzygetic cubic touches the Cayleyan.(14to 3 )A/*i/= 0. 7. [The polar conic of any point on the line goes through the four poles by Ch. Ex. Hessian.. Reciprocate the mutual relationships of cubic. 3. The locus of a point whose polar line with respect to the cubic touches the Cayleyan is a sextic.
12.] . 11. . 2 = 9(y2 ±x2). v between the six linear equations thus obtained. Ex.x2 ) + cz (x 2 y i ) = Q. we have the tangential equation of the Cayleyan of the cubic i/. and the intersection B of the harmonic polar of C with the line of inflexions. and £y. . . //. for the coordinates of a pair of corresponding points on the Hessian or use the method of Ex. any inflexion C. 10. z 2 . yz. 9. 10. X'. Ex.XV 5 TANGENTIAL EQUATION OF CAYLEYAN 2 y z 251 Ex. Find the tangential equation of the Cayleyan of (i) = 4:a?g (iii) 2 s 2 (ii) z x = y{yx) (yk x). [Equating the coefficients of x2 y 2 . The Cayleyan of a cuspidal cubic degenerates into the cusp and the intersection of the cuspidal and inflexional tangents. [The Cayleyan of z(x 2 ±y2 ) = y (3x2 + f) is z Ex. i xs*g 3 z ax (y 2 .+ + Cy> and eliminating f f. zx. xy in >?y (\x + fiy + vz)(\'x + n'y + v'z) .z 1 ) + by (z2 . The Cayleyan of a nodal cubic is a conic having as the vertices of a selfconjugate triangle the node A. See §3.
0) }> (1. cnw clnu). the point of contact of a tangent § 2. = z 2 x^(y — ocx) (y — (3x)(y — yx) — 0. Z by x. we have the required now The point from 0. Z2 X = Y(YX) (FFZ). <x. the equation becomes = = . Suppose a > /8 > y. Cubic with two Circuits. y. a— z=(<xy)*Z. ft 0). Since the polar conic of (0. Y. . x " £=* = a—y *« 4 and the equation becomes Replacing X. yz 1 in the equation are zero. and are real since the cubic has two circuits. 1). Since the equation of the cubic must reduce to y 3 when we put x in this equation. 0. Any may = 2 y (y—x) (y—k x) be taken as (sn 3 u. Take a triangle of reference tangent at a real inflexion 0. ivith 2 The equation of any cubic the two circuits can where r>/c 2 >0.y CHAPTER XVI USE OF PARAMETER FOR NONSINGULAR CUBICS § 1. 0. 0) are the points of contact of the tangents to the cubic from (0. Coordinates of any Point on point on the cubic z2 x a. the coefficients of z 3 y 2 z. 1) is zx 0. (1. — tx y y=Y +V^L. A Standard Equation of the Cubic with two Circuits. Choosing homogeneous coordinates such that the coefficients of z 2 x and y 3 are equal and opposite. by a suitable choice of homogeneous coordinates. and ABO AB is such that BO is the the harmonic polar of 0. be put in form z2 x = y (y — x) (y —k x). (1. (1. y. The points Put now X = A. 0. 0) is z. snw. form. the coefficients of x*z and xyz are also zero.
dniij. equivalent to sn . + u2 + u3 = ) . sn(u+2mK + 2nK'i) = (l^snw.... should be collinear. by s.. s2 s2 s3 c2 c3 ct 2 a3 . c lt Then en. on (u + 2m K + 2nK'i) = (l) m+ »cn^ dn(u + 2m K+2nK'i) = (l^dnu^ u by u + 2m K + 2n K'i. we find that it vanishes. (mod. as at once Remembering that. 0. is nant But the vanishing of this determithe condition that the points with parameters u x u 2 u 3 .. if we replace u x u 2) u3 . (itj .. are collinear if u 1 + u2 + u3 This gives is = dx . 2K. Denote sn u^ .* if m and n are integers. <*. cnUj.XVI 2 CUBIC WITH TWO CIRCUITS Jc. Another sn (ttj line of argument s is to employ the formula «l + u 2 + u 3) = . 2K'i). 253 is the modulus of the elliptic functions being verified. The points with parameters we see. Ctt s3 ( ci c2 + c3 ) Substituting for d2 d 3 Si in the determinant S. d„ Si(c 2 c 3 + Ci) 'd^ : : s 2 (c z c 1 + e2 ) C. (u 1 + u2 + u3 ) = +1 + c3 = Hence enfttj + ttjj) = (s 1 c 1 d2 ~s 2 c 2 d1 )i(s l c 2 d2 s2 c 1 d1 )* d. that the point on the cubic is unaltered.
* Porsn(»+ K'i) = l/fcsni>. u + u + v5e = 6 e 0. C. [The parameters of the points of contact are \u. where Ex.da(v + K'i) = ionv/snv.* 1. v + K. Hence the points with parameters v 12 vu vB6 are collinear.ratios is & 2 [Consider the tangents from the inflexion C] .ui . 1. (c 3 . The four tangents from any point of the cubic form a pencil. u1 + u2 + v12 = 0. . = v + K'i. . U1 + tt2 +''U.3 + tt4 + tt5 (tte =0. [The method of § 7. 2. the three lines just mentioned. cn(» + K'i) see A. d denote cn( sn( — \u). Ex. . k 2 k"sc). If u1 . Ex. side of the triangle of reference. They therefore all pass through the point with parameter v5e (Ch. &2 s s k*scd). where s. Parameters of Points on a Conic or Cubic. ccP. But the given cubic. + u6 + « 6 t = 0. . is also available. v l2 v3V .u3 . §2). the conic and the line through the points with parameters v n and v3i are three cubics through the points with parameters ult u2 u3 . (1. Use § 2 to prove that the inflexions lie by threes on twelve to the curve straight lines. Ex. — \u). and so always unless the contrary is stated.u 2 . ' dn ( — 1«). XII. .254 CUBIC WITH TWO CIRCUITS XVI 2 v is real The even circuit of the curve is given by u and lies between and 2K. (d\ k*c*d. . v + K'i.] + K + K'i Ex. § 18. k'*sd). . 2 K'i . The points of contact of the four tangents from the point on the curve with parameter u are (s\s. u 1 + «2 tw3 + w. c.cd). . Hence = . one of whose cross. . \n + K. v meet a and verify that the crossratio of the pencil formed by the tangents from a point on a cubic is constant.] § 3. Dixon's Elliptic Functions. u3 + ui + v3i = 0. 4. Obtain the crossratio of the range in which the tangents at the points with parameters v. \u + K'i. and v 12 + vu + v56 0. 3. \u + K + K'i. = idnv/ksnv. . and k 2 +k' 2 =l.u 5 u 6 is are the parameters of points on a conic. ub u e . For suppose the three lines joining the points with parameters u t and u 2 u 3 and u v u 6 and u 6 meet the cubic algain Then in points with parameters v 12 v u vS6 The congruence taken modulo 2K. to4 .
xyz. Choose also the homogeneous coordinates so that the tangent at G is Then. § 4. where k2 + k' 2 = G as a real inflexion and AB as the harmonic Bnt take A. us + u 9 + r = 0. z 2 (y — x) = 1. (y + x) (k x* 2 + k' 2 y 2 ) . and that the two lines joining the points with parameters u 6 and u7 u s and u9 meet the curve again in points with parameters q. More rigorous is the proof by Abel's theorem. k. if we replace u by The point on u + 2m (K + K'i) + 2 n (K .XVI 4 EQUATION OF CUBIC WITH ONE CIRCUIT 255 Vu Uj. term in z 3 . Suppose that the conic through the points with parameters us u3 ut uB meets the cubic again in a point with parameter p. * Ch. have the same §ign. of the twelve intersections of the given cubic and the quartic consisting of the conic and the two lines just mentioned. A Standard Equation of the Cubic with one Circuit. Ex. . Ex. . Then . (2) the As in § take polar of G. u 6 + w7 + q = 0. § 7. or y2z. § 7. Now replace z by In this a and b generality in supposing (a + b)iz. of the cubic is is in the required form. u1 + u2 + u3 + ui + u6 +p = 0. pair of unreal intersections of with the cubic. There is no loss of them both positive. and takes the form AB = z 2 (y — x) = 2 (y + x) (ax + by 2 ). for we can interchange x and y if they are negative. the equation of the cubic contains no x y. and the equation Any point on this curve (snu en it. (l + cnu)dnu). i<_ u2> u 3> ui' u s> ue> u i> u$> u9 are the parameters of the 0. XII. the modulus of the elliptic functions being the cubic is unaltered. Ch> X. . Therefore the remaining three lie on a straight line. and therefore Now = u +u 2 + u3 + ui + u5 + u 6 + u1 + us + Ug= 1 0. u1 + u 2 + u3 + ui + us + u 6 + u 7 + u s + w9 = . i. as in § 1.* Hence p + q + r 0. snw. x2z. be The equation of any cubic with one circuit can the form put in 1. intersections of the given cubic with another cubic. r.57. nine lie on another cubic.K'i). B as the double points of the involution determined by (1) the intersection of AB with the tangent at G and the real intersection of AB with the cubic.
The real inflexions is part of the curve have parameters 0. when the cubic becomes = . l>fc*>0. we get k = 0. elliptic. The inflexions have parameters e (K + K'i) + e' (KK'i). 2 y (yx). §K. The real given by real values of u lying between and 4K. 0.] § 5. If (x. y = z = 0. when the cubic becomes z2 x = y(y—tt) a with a crunode at (1. k = x—y = z = .4 256 EQUATION OF CUBIC WITH ONE CIRCUIT XVI w 2 u3 . Unicursal Cubic. 0). 1. 2* II. 1. 0) 1. . z) is a double point of zi x then (Ch. e' = 0. As in § 2 the parameters u 1} are connected by the relation l of three collinear points u + u 2 + u3 the congruence = 0. [C is the inflexion. y. 2 » + (l + k 2 )i/21c 2 xy Rejecting the either = 3y* + 2(l + k )xyk*x* = case x — y = z = 0. with an acnode at or (1. 0. %K. noting that for three collinear points en (u v + u 2 + it 3 ) = +1. Show that the equation of a cubic with two circuits can be put in the form where z*{yx) = (y + as) (y^l^x"). being now taken modulo 2 (K + K'i) and 2(K—K'i). except that the congruences are taken modulo 2(K + K'i) and 2(K—K'i}. 2). Ex. The proof is as in §2. § 4) = t y{yx)(ylci x). and express the coordinates of any point rationally in terms of functions. and A and B are the double points of the involution determined by the intersections of the harmonic polar of C (1) wifh the odd circuit and the tangent at C. (2) with the even circuit. The connexions between the parameters of the points in which the curve meets a conic or another cubic are the same as in the case of the cubic with two circuits. (e. z2 x zx = 0.
and M = 2 (K + K'i).. when the cubic becomes z 2 (yx) = y 2 (y + x). (y 0. 1. since the cubic is now unicursal. 0. Hence many of the results established for nonsingular cubics hold good with slight modification for unicursal cubics. y. Applications of the Parameter. for the sum of the parameters of six points on a conic A = = This gives u = I Jf + where for the cubic ' iV. as could have been z2 foreseen. 0). namely. namely. or if k = 1. We In Ch. N=2(K. = 0. x = z = when (y—x) the cubic becomes = x 2 (y + x) with a crunode at (0. with two circuits. . i . the points of contact of tangents from the inflexions. We shall leave the reader in general to find out for himself what modification is necessary in any particular case. If the cubic has a double point. 2. 6) . 1. + x) x2 + k'2 y2 ). = 1 with an acnode at (1.K'i) for the cubic with one circuit. 0) 0. ( e . sextactic point of a curve is one at which the conic of closest contact has sixpoint contact. we must have &u 0. § 6. N = 2K' 3. But this formula includes the inflexions (the conic» being the inflexional tangent taken twice). when e and e' are 22)6 S . We now apply the parametric representation of a point on a cubic to the investigation of various properties of the curve. XVI 6 APPLICATIONS OF THE PAEAMETER (x. XIII we expressed the coordinates of any point on a unicursal cubic in terms of a parameter in such a way that the sum of the parameters of three collinear points = 0. but not that of a cuspidal cubic. e' M =2 K. see that the standard equation of a cubic found in either § 1 or §4 includes the case of an acnodal or crunodal cubic. z) is 257 Similarly either if k a double point of (k 2 z 2 (yx) = 0. 4. If u is the parameter of such a point. the elliptic functions degenerate into trigonometrical functions. A cubic has twentyseven sextactic points. = where h2 + lc' 2 y = z = 0.
. 6. namely. pairs. 4. § 3).e. inflexions. 8). we must have 9w A = and therefore u= ^M + V. If the tangential of such a point has the parameter u+u+v Hence = 0. = .258 APPLICATIONS OF THE PARAMETER There remain twentyseven values of XVI u v. It will be remembered ways are three different If. is the parameter of such appoint. e' = 0. — 8 it or 9u 0. If it. A cubic has seventytwo coincidence points. = = If the tangents at' the points with parameters u. . 7. since v u. i. the parameters of a %(M + N). giving sextactic points. the points coinciding vjith their third tangential. or v = — 2u = — e — iV. Hence when the points jugate pairs (Ch. 1. Ex. If one polygon of 2 m sides can be inscribed in a given cubic with the sides passing alternately through P and Q. If e' the tangential of a sextactic point is an inflexion. of a cubic are grouped into conpair differ that there by \M. we confine ourselves to the real points of the we have only one method of pairing the points of the cubic with one circuit. \N. where P and Q are given points of the cubic. 5. of grouping the points into conjugate however. or XV. v £ = u + %M. coincides with its third tangential if u As shown above.v meet on 2v the curve. and three methods of pairing the points of the cubic with two circuits. an infinite number of such polygons can be inscribed. 2u and therefore. number coincidence point of a curve is one at which an infinite of cubics can be drawn having ninepoint contact with the curve. the points have the same tangential. . cubic. the second tangential (tangential of the tangential) has parameter — 2 (2«) = 4u. 1. 0. 2. u + ^N. Excluding the case in which e and e' are multiples of 3. 6 (6. we have seventytwo coincidence u the tangential of the point with parameter has parameter — 2«. 3. or u + ^M+^F. and the third Therefore a point tangential has parameter —2 4u ~ — 8w. which gives the points. 6 both even.
then P. . For instance.. B t A%. Ex. The line joining two of the six points now obtained meets the cubic again in a fresh point. "^ wl) w 2nl + ^2n w 2 ^~ W S'J Ex. A n Bn are given points on a cubic such that one polygon of 2 m sides can be inscribed in the cubic with its sides passing in order through A B lt A it P2 . q are the parameters of P. R (and Q.] = = = . «!+«! «» + «„+!. Ex. then an infinite number of polygons can be thus inscribed.such a polygon. Ex.. s'2 u. .. Ex.. and so on. and p. [If u lt w 2 . Show n that for the polygons of Ex. Q' are two Steiner meet the curve again in a Steiner pair. If all but one of the 3» points. (i) 2. The reader generalize the theorem. — — x . consider the case .. . Q. (iv) Any two sextactic points form a Steiner pair for inscribed inflexions 5. 0. \£ Ay. = 2. If P.. Q is a Steiner pair for a polygon of 2w sides. «....3. J. R) is a Steiner pair for a polygon of in sides.. . sides meet the curve again 9. Show that the points thus obtained have parameters given by the expression mu + nv+pw. A n B„. If P. where P x and Q x are form a Steiner pair for inscribed hexagons. B2 . on the curve where PQ X and QP1 S meet the curve again in P2 and Q 2 and P0a and QP2 meet the curve again in P3 and @3 The conies of closest contact at P and Q meet on the cubic. If If P and Q (ii) the tangents at P and § meet on the curve are conjugate points on the cubic.q) 0. Q and P'.. . the lines PP' and QQ' Ex. «<2 + '«S=«n+l + «n+2. XVI 6 APPLICATIONS OF THE PARAMETER 259 [Suppose «. CWte. 1 . w. 6. 106.. > > Ex.. Three triangles can be inscribed in a given cubic whose in three given collinear points. . u ln are the parameters of the vertices. 7. which are the vertices of a polygon of 2n sides and the intersections of opposite sides of the polygon. 4... 2 + « 2n + «i = 0. lxiii I (1864). .. P and § are called Sterner s points. so does the remaining point. Then we must have the congruences. q + u2 + us 0.. pairs. \ dodecagons. 8. and PQ meets the curve in the tangential of R. Ex. See Clebsch. p.. Three noncollinear points are taken on a cubic with parameters The lines joining any two of these meet the cubic again in three more points. 2 + m4 + ms which are consistent if n (p . ••• consistent which is evidently the case. we have to prove . The two points in which the lines joining any point of the cubic to a pair of Steiner points meet the curve again are also a Steiner pair. . lie on a cubic. Pascal's theorem is a particular case of this result. = = so that n 3. . . 5. = may U 2nS~^ w 2n2 M 2«. v. (iii) Any two If £4 PQ and §P meet . PQ and the tangentials of P and S t §Pj meet on the curve §. u 2n are the parameters of the vertices of .
Deduce the fact that the sextactic points are the points of contact of tangents from the inflexions. B'. All cubics having eightpoint contact with a given cubic at a given point pass through the third tangential of the point of contact. XII. triangles Ex. Three of them on the odd circuit. whole . Su=Bv=3w = u + v + w. 18. D' meet on the curve. n. w are commensurable with any periods. and six on the even circuit. See also Ch. 141. Six of the coincidence points are real. PC. C. p numerically less than those considered. Now assume the are among those obtained. and that the line joining A. both inscribed and circumscribed to the cubic Ex. The tangents at A. Ex. it meet* the cubic again on the line joining the point of contact to its second tangential. Ex. or D' passes through one of four points P. and use induction. p+v+w= give Ex. C. w + 4m + k>EE0 give v = w. If a conic has fivepoint contact with a cubic. PD meet the curve again at A'. Any side is divided harmonically by the cubic and its point of contact with its envelope. D'. cvii. p.] p = 4:U.p are positive or negative integers such that a multiple of m+n+p—1 Under what conditions the total number (u> of points obtained finite ? * [The points with parameters — v—w. Ex. v. B.260 APPLICATIONS OF THE PARAMETER 3. Q. C'.. § 4. such that the tangents at P. Deduce the tangential. See Hurwitz. Ex. and the* number of conies through three given points of a cubic which osculate the curve. —2 m. If [Au + v + w=0. lie 13. S meet on the curve. The lines PA.] . or D to A'. B'.] a conic has fourpoint contact with a cubic. v. — w—u. 16. Orelle. The coincidence points are the vertices of the 24 '. 17. D on a cubic meet on the curve. C. 10. P is Ex.] Ex. 14.] and similarly Ex. B'. B. The number of points is finite if u. 15. 21. — 2» (v + iv) + + u). result true for all values of m. PB. Q. the line joining the other two intersections of the conic and cubic passes through the second tangential of the point of contact. An infinite number of triangles can be inscribed in a cubic such that the tangent at each vertex meets the opposite side on the curve. fact that a coincidence point coincides with its third * If not. and any other point of the cubic. —u—v. B. S on the curve. the construction gives the whole of the odd circuit or the of both circuits. 11. Nine or three of the sextactic points are real. They all lie on the odd circuit. [5w + » = 0. is XVI 6 is where m. Find the number of conies through four given points of a cubic which touch the curve. [If u.Hart '. Show that the tangents at A'. C. n. 12. u + v— (u + v + 2w)= — 2u>. [Easy also by quadratic transformation. if it exists. w are the parameters of the vertices. B.
pp. e a curve very close to the being a small constant.] must satisfy the relation <f> . 26. Ex. de France. [The harmonic polars of the inflexions. n'. n = 3. [As in Ex. c are constants not involving n. contact of any such conic. m. The number of coincidence points on a curve.] Ex. The locus of the sextactic points of a pencil of syzygetic cubics nine straight lines. iv (1876). m. 22. +u 6 = 0. . 359. OC+OC'). b. ffx degenerate curve ffx of n.XVI is 7 ANOTHER STANDARD EQUATION 261 Ex. 19.. If six points of a cubic lie on a conic. OL' we must have <fi (n. For a more general result see Halphen. In this equation g2 and g3 are any constants. xv (1879). m. How many Ex. XIV. 24. 20. Hence for all values = /= 0. where a are the equations of two curves. OL') = (j>(n + A'. a). OL = 18. 22. <£ = 3 = n = 3. Annalen. Bull. 3 . <j> <j> Ex. m = 3. We may get their equation by identifying P with its points . OL. Ex. so do the six conjugate points. § 7. If a second cubic passes through nine points of a given cubic. Prom the number of sextactic points of a any curve has 12m — 15« + 9k sextactic points. = m'. de la Soc. a = 10. it meets the cubic again in the points of contact of a triplytangent conic of the same family.] Ex. c are at once given by noting that = 27 n = 3. is 33m42>i + 27K. b. cubic. 25. a) + (n'. = an + bm + cOt. Let it be <t>(n. +(u i + \M) =0. 23. Now a. Math. Three families of conies can be drawn touching a given cubic If a conic is drawn through the three points of at three distinct points. OL. = an + bm+cOL.. /. p. deduce that [We assume that the number is the type.. The theory of functional equations gives readily that where a. » A cuspidal cubic has no A nodal cubic has three are real ? sextactic or coincidence points. {u 1 + \M) + {u i + \M)+ . 3. Another Standard Equation of the Cubic be The equation of any cubic can 2 put in 2 3 the form y z = 4x g2 xz g3 z by a suitable choice of homogeneous coordinates. Ex. 5985. the notation being that which is usual in the theory of elliptic functions.. sextactic and six coincidence points. m = 4. [If Mi + MjH . cj> same is e for all curves of the if same Bm + K. third tangential in Ch. 21. m = 6. another cubic passes through one of the points and the points conjugate to the other eight. m'. m. Math. The locus of the coincidence is eight equianharmonic cubics see Halphen.] Ex. a = 12. m + m'. § 9. Now is = 0.
Hence the equation of every (nondegenerate) cubic can be put in the given form. at (1. m. s = m (x — iz) the curve where x = £z and y — r\z (m' i where i 2) + (4)»2 ) xz + £2mn + 4:l g 2 z* = 0. we get the ax? + by 2 z required form. or acnode.] from any point of the curve if is crossratio of the tangents harmonic gs = 0. if (x. y. Use the equation of a cubic given in § 7 to prove that the pencil of tangents from any point of a cubic has a constant crossratio. the coefficient of x2 z zero. The cubic has one or two circuits according as the harmonic polar y of the real inflexion meets the curve in one or Z 3 three real points that is. roots. = 0). . the harmonic polar of B. 2. according as g 3 is < =. and the polar line of the intersection A of this tangent and harmonic polar. II. % 1) be a point P on the curve. The line is therefore a tangent if this equation in x/z has equal which gives m* _ 24 £m2 + 32 ijm + 1 6 g . the crossratio § 11) of the pencil of (<p+iy(<l> 2f(<i>iYP where Ex. The line through P meets 4a.262 ANOTHER STANDARD EQUATION XVI 7 Take as sides of the triangle of reference the tangent at a real inflexion B. and equianharmonio g2 = 0. /= The if 16&. yz leading to = 12x g z = y + 2g and to g = 27 g x:y:z= . = = . We see then that the cubic is of the form Replacing + cxz 2 + dz 3 = 0. Since the line meets z the tangents from P is given by (Ch. 3 2 2 3 . = .. the coefficients . cusp. Since the polar line of is (1. I. § 4) shows that. The usual condition (Ch. z) is a double point of the curve. [Let (£.48 £ 2 = 0. according as <7 2 — 27#3 is less than or greater than zero. or >0. 3 : : 2.0) is yz = 0.# * 2 2 2 2 2 xz + Sgs z 2 = 0. Ex. xy and y 3 in the equation are zero. Since the polar conic of (0. the coefficients of x y. = 2i(^cp+iyj\ ^=64^. 1. now x by ( — 4b/aftx. of xyz and yz 2 are zero. 0) is x = 0.. 0. found that in this case the double point is a crunode. 1. 6 when Since the equation of the cubic must reduce to x' 2 2 we put z in this equation. It will be readily .
e'= 0.2). except that the congruences are taken modulo 2o> and 2a>. where u is real. given by where 4 (pu) ! :i pu 2 Weierstrass's elliptic function (p'uf = g pug. If the points with parameters u. — \pu—pv) . iu u*. . are collinear.. pw This gives p'w u+v+w where 2o>.1. we may take any point on the curve as is (pu. . ^> The is real inflexions have parameters — o o the real period. Hence v w + ?'lizl'u + P'U ~P'V + P P' ' — pv — pw = — pw — pu = ~ pu — pv ' from which The (i) immediately follows.. If g 2 3 >27g 3 2 the cubic has also an even circuit given by . L' v. § 72).• .  . parameters of the form u + co. In fact (Dixon's Elliptic Functions. u+v+w = /p'w — p'u^ if 0. pu+pv + piu = 4 . Coordinates in terms of Weierstrass's Function. ) ' XVI 8 WEIEESTRASS'S FUNCTION 263 § 8.pu=\. 2a>' = (mod. 2o/). 2 y z If the cubic = ^x' — g i xz 1 — g% z z ' is nonsingular. The standard form of this and the preceding section may be used as in § 6 to establish properties of the cubic instead of the standard forms of §§ 1. The connexions between the parameters of the points in which the 'curve meets a conic or another cubic are the same as in the case of the cubic with two circuits in § 3. 1) . 4. pu pv p'u p'v 1 1 1 = 2co. p'u. (e. & ) i(*\pvpw) t /P'v — p'w^ * — =ii~ *\pwpu) 4 . where 2a> The odd circuit is given by real values of the parameter. (i).* inflexions have parameters (eeo + eV). * It is at once proved that all the plus or all the minus signs must be taken. 0. are the periods of pu.
or 3. we may draw up a table showing all possible types of quartic as follows Type A . Types of Quartie. quartic cannot have more than three double points (Ch. 2. § 1. VIII.: CHAPTER XVII UNICURSAL QUARTICS 1. so that its deficiency may be 0. § 2). 1. VIII. By the aid of Ch.
The limaijon r = a+ 6 eos for a/b = . § the conic 5). §. however.g. G each property of the curve can be duplicated by reciprocation . The method of projection and inversion will. In fact. is a cusp) . It is not readily the double points are of different kinds But for instance. (Ch. XVII 2 GEOMETRICAL METHODS 265 process which comes to the same thing is to project into the circular points and then invert with respect to A. . a cusp. both cusps or if applicable all In the case of a quartic with a node A and two cusps B. f .2fy + c = 0. be suitable only if B and G are double points of the same kind. since every circular line meets it in only two finite points Fig 1. and from each property of the conic a property of the quartic is deduced. an ordinary node.. if J. 1. 2. both ordinary nodes. An inverse of this with respect to the origin is ax 2 + 2 hxy + by 2 + 2 gx +. II. e. it is readily seen that the equation of a quartic with double points at the origin tnd circular points is of the form A B and c (x 2 + y2) 2 + 2 (g'x+fy) (x 2 + y 2 + ax 2 + 2hxy + by 2 = ) 0. when the quartic becomes a conic (a parabola. quadratic transformation is still available. and a flecnode.
if we project the cusps of a quartic with a node and two cusps into the circular points. 4""j and the origin the centre x . which A c (x 2 + 1/2 ) =x z (i). 2. is the inverse of a parabola with respect to its focus. For instance. But the parametric equations of a threecusped hypocycloid are the = %a (2 cos + cos 2<p). (ii). if we project two of the cusps into the circular points. § 4. 2)» For the polar reciprocal of the curve with respect to a cusp A and having will be a cubic touching. e. C into the circular points.e. we may obtain its properties by reciprocation from the known properties of the nodal cubic.. i. Hence. Or again. Threecusped hypocycloid. If we project B. the curve becomes a with three cusps limacon with a the cardioid finite cusp. Or again. The properties of a quartic may be investigated geometrically in various ways. j. 1. it becomes the inverse of a conic with respect to a focus. V. 266 GEOMETRICAL METHODS XVII 2 for the reciprocal of such a curve is a similar quartic. r = <z(l+cos#). if we project the points of contact of the bitangent into the circular points. 1 = a + b cos 6. the quartic becomes a conic with as focus since the inverse of a curve with respect to a focus has cusps at the circular points. and then invert* to A. and with respect A especially Ex. §tt. and conversely (see Ch. shows five limacons with the same b and a/o %. A being the origin. of . the circular lines at the line at infinity as inflexional tangent... 1 to 6 in that section). 2. as is evident from its Pliickev's numbers. y = Ja(2sin0— sin20) cusps being = 0. the curve becomes a threecusped hypocycloid (Fig. = g. a limacon with polar equation of the form r Fig. i. Its equation can therefore be put in the form Pig.
' Now invert with respect to A. 1. ABCBXJ pass through the same four points.BS) is harmonic.touch the theorem from Pascal's theorem similarly. If and BM' meet CA in and m'. B. C of a trinodal quartic curve. 2.] Ex. Ex.' It is a straight line (the polar of A). 4. C and that the envelope of the conic ABCPQ is a quartic with A as node and B. xxxiii. A quartic has nodes A. drawn BM MN m CN BA [Project B. Then we have Find the locus of the middle point S of a chord PQ of a bicircular quartic meeting the quartic again at its finite node A. Another proof is given in § 5. through the intersection of and M'N'. ABCQT. or use the equation of the bitangents given in § 3. Q.] Ex. A quartic has nodes A. B. . . and we have Through a fixed point A a line is drawn meeting a fixed conic in P.. [Projecting the two nodes into the circular points and inverting with respect to the other node. These theorems are due to Jolliffe. Derive a [Projecting and inverting we have Brianchon's theorem. T. • the conic LMPABC. B. Through A is drawn any line meeting the. C. Ex. BM' and CN. osculating the curve at A cut at L and touching the curve at P. we have ' The intersections of a conic with a pair of directrices are concyclic. and so on for the points P. B. 6. The four points of contact of tangents from two nodes of a trinodal quartic lie on a conic through those nodes. BS). quartic again in P. Show that LM. C. C into the circular points and invert with respect to A . Show that the locus of Sis a conic through B. C are conjugate chords of A quartic has three nodes A. C. we obtain <f> Sb2 {(xa)* + y 2 } +4a(x~af = as the polar reciprocal of the hypocycloid with respect to a cusp. S. Show that the conies ABCPS. C. by differentiating with respect to in the usual manner. Hence the locus of the middle point of the chord of the bicircular quartic is a circle through A and the locus of S for the original quartic is a conic through : ' : ' ..'] : Ex. CN' Show that two bitangents to the quartic pass to the quartic. V. and . and CN' jneet in n and n'. Q and S is a point on PQ such that 1/AP+ 1/AQ = 2/'AS find the locus of &'. A. C conic through A. then two bitangents pass through the intersection of mn and m'n'. Writing respect to the circle (£ca) 2 + i/ 2 =6 2 and finding its envelope. C as cusps. The first conic meets the second again in P. B. are 3. Six conies pass through the nodes A. Ma PA A The conies through A. B. Q and BC in B. But this is the curve (i). BC) are harmonic. chord PQ meets BC in B and (PQ. B. A quartic has three nodes A. Tangents BM. B. B. Find the locus of S if (PQ. Messenger Math. the second meets the third again in Q.C] Ex.2 XVII GEOMETRICAL METHODS down the polar of the point (ii) 267 with the curve. 5. [Project B and C into the circular points. A (PQ.
A quartic has nodes A. touching the quartic.C.C. B. In general there are two quartics with three given nodes. Show that the remaining intersections of the conies are collinear. Ex. 9. Show also that the tangents to two such quartics at an intersection divide BC harmonically. Q meet at T and cut the bitangent IJ at Prove that Any The tangents is (EF. . C. Ex. Ex. the locus of the remaining intersection of the conies is a conic through A. 13. C. 11. (iii) The chord of contact of a conic through and C touching the quartic at two points passes through one or other of two fixed points.] Ex. 8. A quartic has a node A and two cusps B. The locus of T is a conic through I. Ex. C and the tangent at A is Show that the locus of the point of contact of a tangent drawn from a fixed point on BC is a pair of lines through A. C touching the curve at P and Q meet again on a fixed conic through A. XIII. (i) (ii) Q. or may obtain other theorems by reciprocating those given above. If AD is the C meets and (ii) If tangents from B and C to the curve meet at the curve at P. D AD B (iv) If PAQ is a chord of the quartic. 15. Quartics have given cusps A. B. (iii) The line joining T to the intersection of PF and QE passes through the point of concurrency of the cuspidal tangents. Conies are drawn through these nodes touching the quartic at P and Q. 12. The tangents at the cusps of a tricuspidal quartic are concurrent. Find the envelope of the line through any point P of a which forms a harmonic pencil with the lines joining P Ex. fixed points cut Quartics with three given nodes and passing through four other any line through a node in involution. the conies through A. IJ) harmonic. The result is also evident on projecting the quartic into a cardioid or threecusped hypocycloid. B. B. Ex. J touching the quartic at three points. F.] [Project B and C into the The reader may Ex. § 4. tricuspidal quartic to the cusps. Through the cusps of a tricuspidal quartic three conies are drawn touching the curve elsewhere. passing through four other fixed points and touching the circle through the nodes (the point of contact not being a node). the locus of D is a conic through B and C. Prove that chord conjugate to BC of any conic through A.268 GEOMETRICAL METHODS XVII 2 Ex. B. 14.' Obtain other properties of the tricuspidal quartic by reciprocating the examples in Ch. P and E.B. t also given. circular points and invert with respect to A. ' [The reciprocal of_: The inflexions of a nodal cubic are collinear. tangent to a tricuspidal quartic meets the curve again at at P. If PQ passes through A. PA and the tangent at P divide BC harmonically. (i) 10. derive other properties of the quartic similarly. 7.
19. Ex. Hence the quartic takes the form (i). P being any point on the curve. . AS' are equally inclined to the tangents A to the quartic. or z. y Another method of obtaining properties of the trinodal quartic is to take the three double points as vertices of the triangle of reference. In the equation of the quartic there can be no terms involving the third or fourth power of x. In Ex. § 3. .S'P)/AP is constant. (AS'. H the cofactors of . If Sand S' are the foci of a bicircular quartic with a finite node A. F. A bicircular quartic with finite node can be regarded as the envelope of a circle passing through whose centre lies on a fixed conic.1 z b c 2h 2g 2f = —„+„ + A + ^L + + yi i Z yZ zg . 16. 18. y.a g. 0. 17. from Ex. SP±AS . 17 the lines AS. G. We shall denote by A G.?' at 0.XVII 3 TRINODAL QUARTIC foci 269 and common finite Ex. If this section. ay 2 z 2 + bz2 x2 + cx 2 y 2 + 2xyz (fx + gy + hz) = 2 2 2 we divide by x y z we get the equation at the head of . It can also be regarded in two ways as the envelope of a circle orthogonal to a fixed circle j through whose centre lies on a fixed conic touching. Ex. s „ „ m The Quartic . Two bicircular quartics with the same node cut orthogonally. — X . .
2 = a\ 2 2 + b/j 2 + cv 2 2ftiv2gi>\2h\ii.. 270 TRINODAL QUARTIC . For putting u= fx + gy + hz. § 3. a/x + h/y + g/z = The four bitangents fx + gy + hz either three or one of the quartic are = ± Vbc x + Vca y '+ Vab z.3 . may be written the equation of the quartic ( /a yz + Vbzx + Vc xyf = showing that 2xyz( — u+ Vbcx + Vcay + Vabz) .* where and is of the quartic. J"=2 3 + 18A u»/2A + 54AA 2 AtV. crossratio of the range in which the line divided by the { Xx + jiy + vz= quartic is given by (0_2)(^ _l )} 2 ) /3 (0 + l) = 3r</ 2(02 _^ + l 3 ) where I=32 * 2 36\fiv\. The result may also be proved by writing the equation the quartic in the form (u 2 — be x — ca y — ab z = (u— Vbcx— /cay— 2 2 2 2 ) Vabz) (u— */bcx+Vcay + Vabz) x(u+ Vbcx— Vcay+ Vabz) (u+ «/bcx+ Vcay— Vabz) '. (v). It will be found that A2 + A 2 The is = BG\ + GAfi2 + ABu 2 + 2AFfiv + 2BGv\ + 2GH\fi . of and similarly for the other bitangents. . XVII quartic at its intersections with the conic other than the nodes. if \x + py + vz = and the quartic touch one another... Or. found in the usual 3 2 2 2 (A2 + A 2) = Xfiv (18 AAS + 16 A + 27 A A = F\ + Gfi + Hu Xfiv). + signs being taken on the right.' (iv) ' ' The tangential equation way. IV. Now write down the wellknown condition that these two conies Bhould touch. so do K/x + /i/y + v/s = and ax 2 + 6j/2 + cs 2 + 2fyz + 2gex + 2 hay = 0. . ) See Oh. (iii) u is = Vbcx+ Vcay+ Vabz its a bitangent with pointe of contact on Va yz+ /bzx + Vcxy = 0. .
cusp. The fourth node is (F. each node is either an acnode. 3. Ex. 7.XVII 3 TRINODAL QUARTIC of the line 271 For the lines joining the intersections quartic to (0. 1. or is a crunode such that the tangents at this crunode and the lines joining it to the other two nodes form two non overlapping pairs of lines. In this case all four bitangents are real. &c.gpv)xy 3 + 6a /i2 y* i 2. Ex. the tangents from the nodes are all real or all unreal.] a and 6. If a quartic with three real nodes has a real bitangent. show in [The envelope is 1=0. 6. c have the same sign. Show that (2l+3J)/\/i. modification occurs in the tangential equation of the curve. x and y conjugate imaginaries. The envelope of a line divided by a trinodal quartic in a equianharmonic range is a curve of the fourth class with four nodes. Ex. 1) are and 6bX 2 x* + 12 (h\ 2 + b\pf\v)x*y + 6(a\ 2 + b/j. and the crossratio I. § 11). if one or more nodes become cusps ? What Ex. 9.'i + ct> 2 ~2fiJLi>2gi>\+4h\n)x' y 2 + 12 (h fi + aX/i. The envelope of a line divided harmonically by a trinodal quartic is of the sixth class. the two curves having the same tangents at these nodes.the inflexional tangents of the quartic. [The axis of perspective is _ x/F+ y/G + z/H = ] . 4. 8. Find the bitangents if one or more of the nodes becomes a What modification must be made in Ex. two c. h. and Ex. [The envelope is J= 0. = . Ex. 0. (p of this pencil is given by the above equation (Ch. The diagonals of the quadrilateral formed by the bitangents Gx + Fy — Cz 0. 2. /and g. H). B. The triangle whose sides are these are diagonals and the triangle whose vertices are the nodes are in plane = perspective. 1 in this case ? Ex.v = are the nodal that the twelve common tangents of this curve and and inflexional tangents of the quartic. If one of the nodes becomes a cusp. We shall touches the nodal tangents of the quartic] § 4 that 2 = Ex. The other common tangents of the curves are the inflexional tangents of the quartic. A which only 'one is real has [We may suppose z real. G.] 1=0 : is a curved of the third class and Ex. and the quartic may be put in the form y z + z x + x y'' + 2xyz (fx + gy + hz) [For a real bitangent a. b. real quartic with three nodes of real bitangents. C is zero. Three of these nodes are the nodes of the quartic and the two curves have the same tangents at these nodes. It has the nodeB of the quartic as biflecnodes. The other common tangents of the ciirves are .] i t 1 1 2 = 0. 5. one of A.
= 0. + lG{q(ps) + s(qr)} 2 = For the converse take 0. r = = 0. z. + */c).the conic touching QB at P'. [For the quartic of § 3 (i) they are ( + i/a. r. B. The four conies through A. The bitangents touch the quartic at real points positive are and 2 (a/2 three or one + hg* + ch* . &c. 10. Show that. the first degree in x. [The centre of perspective 2(jp 2 is p = q — r. 14. s. is Ex. the equation of any trinodal quartic can this form. Find the locus of the nodes of a trinodal quartic. APa? + Brh? y + Cn*f + 2 Fmnyz + 2Gnlsx + Hlmxy m.B. BP at Q'. represents a quartic with bitangents = 0. 2 = ay*. Ex. The points of contact of the four bitangents 2 2(p 2 + q2 +r + s ) all lie on the conic = {p + q + r + sy. q. in §3(iv).] if a. p p q = s. C and the points of contact of a bitangent touch in pairs at C. .^dbcf. 2 = 0. 17. CB pass through a pair of vertices of the quadrilateral formed by the bitangents. 12 are the vertices of a quadrangle whose harmonic triangle is ABC. Three bitangents of a quartic with nodes A. 11. 6. and the two tangents at C are harmonic conjugates for CA and CB. 16. = r. = = = be put in [The equation may be 2 2 put in either of the forms s 2 2 {2(p + q + i + s )(p + q + r + s) {(ps) + (qr)} i 8(pq + rs){(ps) + (qr)}* 2 } = Gipqrs. &c. r q.272 TRINODAL QUARTIC XVII 3 Ex. where p. y. Ex. c Ex.Cz 2 conies. . 13.] Ex. 12. 19. and the + 22 + r ) =(p + q + r)i] ! conic is Ex. Ex. */ ± b. where t varies .] = s s are of p= and with nodes 0. if AP meets QB in P. One of the common chords of any two of the four conies of Ex. 15. C form a triangle Show that the triangles ABC and FQB are in perspective. The quartic 1 the envelope of the conic '2 where I. = 0. given the four bitangents and two points on the curve. + signs being taken. q r s. The equation */p + >Jq + */r+ */s 0. The four centres of perspective obtained in Ex. 18. [The conies are ± */a yz + */l zx ± */c xy Ex. 12 passes through a node. The quartic the envelope of the conic fix 2 + 2 1 (ayz + hzx + gxy) By 2 + 2 Fyz . is = 0. PQ at B' passes through the points of contact of the fourth PQB bitangent. C and [6a. n vary subject to l + m + n = 0. The double rays of the involution formed by the tangents at the lines CA.] Ex. and of two similar families of . p — u — ^bcx + */ca y+ i </abz. p Conversely.abc) > ( + >/af± <Jbg±Jch.
Their properties are given in the following theorems. F. The identity of § 3 (v) is useful in the examples. Show that the locus of the other focus is a quartic with nodes at A. . U = afyz + bgzx + chxy. and one focus lies on a fixed conic S. A Ex. C. By varying the relative positions of S and ABC. Soc. C. To save space we use the following contragtions / u=fx + gy + hz. [The envelope of an asymptote or axis is a threecusped hypocycloid. iv (1903). \ 2 2 2 2 By u. The n the conies T theory. 22. p. pp. be at once deduced. [Use Ex. Math. 20. B. that their envelope and the locus of their centres are quartics with nodes at A. CO with the sides of the triangle a conic of given eccentricity is drawn meeting 'the sides of the triangle again at D.4 : XVII TRINODAL QUARTIC 273 Ex. Various forms of the equation of each conic are given. tri There are many conies of interest connected with the nodal quartic ay 2 z 2 + bz 2 x2 + cx2 y 2 + 2xyz(fx+gy + kz) = . II. iii (1902). A = F\ + Gfi. 2 = a\ 2 + bfi 2 + cy 2 . + Hi>. Conies of given eccentricity pass through fixed points A. Given the nodes and five other points of a unicursal quartic. Amer. K = ay z + bz x + cx y + 2 xyz (fx + gy + hz). 103. and vice versa.] Ex. 2 we denote the same as in § 3. construct the tangents at and from the nodes and any number of' other points on the curve. as they are needed in the examples. BE. B. Show that AD. C. BO. Conies connected with a Trinodal Quartic. 23. C. 2 2 fc — 2abfg\u. Through the remaining intersections of AO. E. If the pointequation of the conic is obtained. T = a2f 2 \ 2 + b 2 g 2 n 2 + c 2 h2 v 2 2bcghnv2cahfv\ T = ghyz + hfzx +fgxy. 22. M=f(gG + hH)x + g(hH+fF)y + h(fF+gG)z. B. 489. 21. (i). of course. A.2f/iv 2gv\2 h\/i. CF meet in a point whose locus is a quartic with nodes at A.] § 4. 154.. conic touches the sides of a given triangle ABC. ABC is a fixed triangle and a fixed point. = and U = (i. e." See Annals of Math. T = 0) are of importance They pass through the nodes of the quartic . we may find every possible shape of a quartic with three real double points. Trans. B. the tangential equation can. Show Ex.
i. 2 + c 2 Cy 2 + 2bcFij. .v + 2caGy\ + 2abH\fi = 0. = 0. Ex.af . This result is at once established by using the equations of the nodal tangents or it may be deduced from II by using the fact that the lines joining the vertices of the triangle of reference to the intersections of the conic ax 2 + hy 2 + cz2 + 2f'yz + 2%zx + 2hxy with ttie = = 0. Ax 2 + By 2 + Cz 2 2Fyz2 GzxZHxy = A2 + 2 (GHftv + HFv\ + FGX/j. bx2 + 2hxy + ay2 = 0. For we have its intersections TU+xyzM=fghK. The six tangents from the nodes touch the conic 0. e. For the intersections are z = 0. 2 . The six intersections of the tangents from a node with two nodes lie on the conic . &c. abcZ = T. aA 2 x2 + bB 2 y 2 + cO z z 2 + 2fBCyz + 2gCAzx +2hABxy = AS + A2 = 0.e. (Ax2 + By 2 + Cz 2 2Fyz2Gzx2Hxy) + 4 T = (Ax2 + By2 + Cz 2 + 2Fyz + 2 Gzx + 2Hxy) + 4 U = 0. This follows at once from § 3 See also § 3. II. (abc . i.) = 0.bg2 .ch 2 2 ) (a\ 2 + bfi + cv 2 ) + (af\ + bgp + chvf = (iv). The six intersections of the tangents at a node with the line joining the other two nodes lie on the conic bcx 2 or + cay 2 + abz2 + 2 afyz + 2 bgzx + 2chxy = 0. . or For the intersections are z = 0. 11. u = Ax2 + By 2 + Cz 2 2Fyz2Gzx2Hxy. .e. e.274 TRINODAL QUARTIC XVII 4 with the K= line I. = bcx 2 + cay 2 2 + abz2 0. The six tangents at the nodes touch the conic ! or i. 2 a2 A\ 2 + b 2 B/j. Cx2 — 2Hxy + By 2 = V. M= and meet the quartic again at 0. &c. The eight points of contact of the four bitangents lie on the conic u2 i. opposite sides all touch the conic bcX 2 + ca/i 2 + ataj^ 2 2af/ti'2tagi'X2ohX// the line joining the other IV. III.
= ax' by' cz' _ + cz' ax' + hy' + gz' hx' + by'+fz' gx' +fy' . if we eliminate z between the result of substituting 1/x for x.+.. The six inflexions of the quartic lie on the conic 2 (aA x +.. the point of contact of the conies and is the line joining this point of contact to px + qy + rz Equating to zero the cotheir fourth point of intersection. For. © = 2A. i. + 2fBCyz+ 2 2 .. } XVII 4 TR1N0DAL QUARTIC 15 v 2. or 2M(ghAx + hfBy+fgCz) U{fgh(4>abc+.. ©' = 2. and remembering 0. z') (px + qy + rz) .. 2 2 2 efficients of x y z on the lefthand side.. we obtain on eliminating z between the equations of this line and the quartic an equation in x/y with three equal roots. ....v..... + . ax2 + by 2 + cz2 + 2fyz + 2gzx + 2 hxy .. The six inflexional tangents of the quartic couch 4>(aA 2 x 2 + .{x {ax' + hy' + gz') + y (hx' + by' +fz') + z (gx' +fy' + cz') = = Xyz + /izx + vxy where (a/. . This shows that the conies X/x + ji/y + v/z = 0. + . = (ii) have threepoint contact. if. Since with the notation of the last paragraph the two conies (ii) have threepoint contact.. © = 3 A®' is the tangential equation 2 of the VII.. The conditions for this are well notation of Salmon's Conic Sections) where A' known to be (with the The relation conic required. therefore for some value of* h (ax2 + by2 + cz2 + 2fyz + 2 gzx + 2hxy) . we have that px' + qy' + rz' is = .) .e. = 2\/j. \/z for z in the equations of the line and quartic. l/y for y. if \x + jiy + vz = 2 2 an inflexional tangent. 2 + 2fBCyz + 2 .. The same will therefore be true. 2 2 3A/© = 0/0'= ©'/3A'.2fgh£)2bcfh2 2cah 2f 2 2abf 2 g 2 }..) 2A(Ax + By + Cz2 FyzGzxHxy) + AU=0.75 VI. is = A Ax + By + Cz + 2 Fyz + 2 Gzx + 2 ITxy) + 4 A U 3A2 + 4A = 0. ( + .
. e. p. . London Math. + . Of the twelve intersections of this cubic with the quartic six lie on the conic T which touches the cubic at each vertex of the triangle of reference... § 7)... now x by 1/x. . the intersections of the line with the conic 0.e.. Proc. 1) are (ay 2 + bx 2 + 2hxy) (Xx + /xy) 2 . the inflexions can verify of the quartic. The contents of the last brackets { } equated to zero give the equation of the conic through the inflexions.. = We that (afghA + bfghB + cfghC+2bcghF+2cahfG + 2abfgH) K + M{2Ax2 (gy + hz) + 2By 2 (hz +fa) + 2Gz 2 (fx + gy) + (4<abc — ifgh — A) xyz] = T{2 A (gG + hff)x2 + + . z by \/z..2xy (fx + gy) v (\ x + py) + cv 2 x?y 2 = 0. lie on a conic (Ch. + . XII. M= = . y by 1/y. . The six points where the tangents at the nodes meet again lie on the conic 2 M (beghx + cahfy + abfgz) = U(5abcfgh4>fghA2bcg 2 h 2 2cah 2f 2 2abf2 g 2 ). U= the quartic VIII. i.. point of contact of Xx + fiy + vz = any inflexion of the quartic. which is readily seen to be the same as that given above. on 2Ayz (hy + gz) + 2 Bzx (fz + hx) + 2 Gxy (gx +fy) Replacing the i. II. 130. _ (4..276 TRINODAL QUARTIC XVII 4 Hence the point of osculation of the two conies lies on ax(hx + by+fz) (gx+fy + cz) + . namely. 0... + (4*abc — Afgh — A) xyz = 0. » i (1903). the intersections lines joining The of the quartic with \x + ny + vz = to the node (0. .}. lies on 2 Ax 2 (gy + hz) + 2 By 2 (hz +fx) + 2 Cz 2 (fx + gy) + (4 ahc — 4>fgh — A) xyz = 0.. we Bee that and the quartic. e. .fF2 + 2gh A + Saf A) yz. and therefore on 2 (ghx + hfy +fgz) (ax 2 + by 2 + cz + 2fyz + 2gzx + 2 hxy) — ax(hx + by+fz)(gx+fy + cz) + . The remaining intersections of the quartic with the inflexionsconic are those intersections of the quartic with the line which do not lie on the conic T i.. Therefore the remaining six intersections of the cubic and quartic. This proof is due to Richmond and Stuart. = 0.. Soc.
the required value of k must be . We A .X y gh h I — — g 2(hH + J ' fF) . * An alternative form is obtained by omitting the A from the righthand side of the second form of the equation of the conic through the six inflexions.Gzx . the four points abz 2 A A Blt B lie on the conic + 2 (2fg — ch) xy + 2 bgzx + 2 afyz — 0.. 2bg 2) a2 cf% : . 2 They therefore lie on the conic k {abz2 + 2 (2fg . IX.+ 2hxy = Px2 + 2Qxy + By 2 = is is one pair of the&e lines and the other pair..Spgh) yz + . we have readily. + . {2h(fF+gG) + abfg}+fgh\ and straightforward verification then shows that for this value of k the conic reduces to 2bc(gG + hH)x 2 + . = a (abcf+ 2bcgh + 4af 3 . — S\f hf y+ y ab ) fz\ h S require to show that k can be chosen so that this conic also passes through G1 and C2 in othei» words that k can be chosen so that the equation of this conic is symmetrical. . on identifying the four lines with (ay 2 + bx 2 + 2hxy) (Px2 + 2Qxy + By 2) = 2 2 0.+.Hxy).. comparison of the coefficients of x2 y 2 . The six points of contact of the tangents from the nodes to the quartic lie on the conic * aA 2 x2 + bB 2 y 2 + cC 2 z 2 + 2fBCyz + 2gCAzx + 2hABxy = A (Ax2 + By 2 + Gz 2 . = gbc:fca:2(fF+gG)..i }. which is equivalent to the given form. 9 fg + cy 2 + 2fyz) lt (az 2 + ex 2 + 2gzx) z 2 {abz 2 + 2 (2fg = — cK — ch) xy + 2 bgzx + 2 afyz . P:Q:B = b cg :fg (abc + 4fgh2af \:fi:v 2 .. .ch) xy + 2 bgzx + 2 afyz} ab ) (be — = i( 2(gG + hH) x+ ca y+jz} \j..Fyz . Gfi2 is the line jx+ be « JNow. 2. ... XVII 4 TRINODAL QUARTIC 277 If ay 2 + bx'. (bz2 f since ca — y = 2(fF+qG) ' v ' z. Suppose the tangents at the nodes meet the curve again in A x and A 2 B1 and B2 G1 and C2 We have just shown that .. z 2 shows that.
we prove as in the case of A t and A 2 Bx and B2 Theorem VIII that G^C^ is . If the tangents touch the quartic in C1 and 2 . 3. In Figs.Gxz + Hxy = t . In the following examples the conic through the points of contact of the bitangents. = 0.hH) xy + af 2 By 2 . 2 + By 2 2 Hxy and the lines — bg 2 Ax 2 + 2fg (A . = (Cz 2 FyzGxz Fig. 1) to the intersections of this line and the quartic being the tangents from the nodes 4a.Fyz . is called Conic I and so for the other theorems. . ' ' .*?) (x±)(Axy 40) = 10(x 2 + 2t/ 2 12a. Then since (By 2 2 Fyz + Gz2) (Ax2 2Gzx + Gz 2) AK + H^y) 2 . 4 are shown typical unicursal quartics. 3 (</ 2 . A j x+ B = fF+gO — z —y —— the lines joining (0. lt 2 2 and also on k (Gz 2 — Fyz — Gxy + Hxy) /gG + hH B G = w A hH+fF G \ which reduces to the given form on taking k (fhF+ghG+fgC)/fgW.278 TRINODAL QUARTIC . . which is referred to in Theorem I..) 2 2 . 3. XVII 4 . the points A A B B lie on the conic Gz . 0.
is [Evident from their tangential equations. 2. I'roye also a similar result for the points of contact of the tangents from the nodes. and . ..] The conies III. the six lines AA\. Show that if in Ex. 279 intersections of A X 2 and BC.LB are collinear.B and B C. The chord of contact 0. The tangents a* the nodes A. 1 (either case) the line A X A 2 meets the quartic again in A\.B. A Fx + BGy+ CHz = and its pole is (F. and their intersections with BC. &c. CA. § 4. 0. G. VI touch at the same two points. Ex. 3. BB\.XVII 4 TRINODAL QUARTIC A . and C2 Show that the a it 1 B^an&CA.4{if lx*) (a 4) (20x + y + 160) = 10{2x 2 y i + 8x) 2 . A\. 1.] . CC\ touch a conic. AA' t BB\. Ex. respectively lie on a conic. .0. CC\. V. C of a trinodaU quartic meet the curve again in A 1 and A . [The required lines are hcghx + cahfy + abfgz = and ghAx + hfBy+fgCz = 0. AB [The equations of the lines are given in Ex. H).] Pig.
Ex. form given. The conies f7=0. Show that the common tangents AS + kh? = are the (1 common tangents of the quartic and the conic of the conic and the two cuf ves (4 of the third class . Ex. B. XVII VI. U= Ex. 6. IX pass through the same two points on the quartic. 5. and a common chord of and VIII. [k = 0. VII. meet the conic again at its intersections with the diagonals of the quadrilateral formed by the bitangents. VII pass through the same four points. The conies U= 0.] M Ex. If the conic Ax* + By 1 + Ca2 . IV pass through the same four T= 0. The conies U= 0. . 12. The diagonals of the quadrilateral formed by the bitangents are the polars of the nodes with respect to the conic IV. W. = 0.bz ) + Gy (az2 .k) A3 = {(8 . If the points of contact of the tangents from the locus of P is the cubic 1 P lie on a conic. civ. 9. A conic can be drawn touching the lines joining the nodes and the common tangents of the conies III and IV. C] . S2 = 0. IV pass through the same four points. [The intersections of U = with Obtain theorems by projecting these points into the circular points. 13. I.Gzx . Ex.ex*) + Hz (bx2 . Ex. [If first 4. V. II. 1.8.ay*) = [The polar cubic of 0. Use the tangential equation of the quartic] Ex. Ex. 4/3. VII. III. Deduce the equations of the conies III. A conic can be drawn having. 308. Ex. P touches a conic at A. 14. Conies are drawn touching the four bitangents and one side of the triangle ABC.] Ex. points. &c. l TRINODAL QUARTIC The conies 0. 11.9i) + .Fyz .x V . Fx {cy* . Gross see Stahl. the nodes as vertices of a selfconjugate triangle and touching the common tangents of the conies IV and V. The The conies conic's {7=0. Ss = we have = 3^. 10. 15. VIII. Ex. Show that the points of contact with these sides are collinear. Any conic touching the bitangents {(bg^eh^WlAff + cAS + lcGvXlbHXn} + k{aB\* + (af ch2 )iJ! + cBp + 2cFViJL2amiJ} 2 . The chord of contact passes through the intersection of a common chord of V = and VII.4 280 Ex. 16. [On AFx + BGy+CHz = 0. VI. .. AX X X F . Crelle. are the pointequations of the conies 'in the This result is due to S3 2S2 S = . III have four common tangents.Hxy meets the sides of the triangle of reference in = x Z and t 1\ and 2 the lines 1 and AX. The conic passes through the intersections of the conies V and IX.] Ex.3/fc)t} \. The conies II and IV have double contact. IX pass through the same four points.. 17. Ex. inverting with respect to a node. and generalizing by projection. I. 7. p.l is = 0. 1 and Z 2.
. P2 and B'.] Ex. [Conic I is a linepair. Blt P2 are Ax2 = By 2. if A 2 ' node with the = 4FGH.] Ex. 21. [See § 3. 9. Illustrate by tracing 2 x2 y 2 2xy(x2y)Bx 2 + 8xy + 3y Ex. the points of contact of tangents from and C are collinear. = x 2 + y 2 + z" = + 2Fx2 yz + .] threes in two The tangents from the nodes cannot meet by Ex. If the tangents at A are harmonic conjugates with respect to and AC. The intersections of the tangents at each node with the line joining the other two nodes cannot lie on two lines.. the polar. .^23. AB are harmonically conjugate with respect to the tangents is cx+gz = 0. 27. Show that AC and the harmonic conjugate of BC for BA. A = 4fgh.. the points of contact of tangents from C lie on a line through B. 22.. BC Ex. Show also that three of the points Pt P2 Qx Q2 Et P2 (say P2 Q2 B2 ) are collinear. P1....] Ex.. if A =. One diagonal AB the other two meet at and intersect AB in R1} P2 Similarly Show that AA'. 18. Ex. Similarly for conies III and IV. AC BC = Ex.. Conic II for ay 2 z 2 + is . before attempting the Ex. 28. §§ 1 and following examples.. if [Conic III is a pointpair. ¥ Ex. . = 0.] [AC and at A. —c). XVIII. = of conic 0.. Qlt @2 are concurrent.. AB A B [P=0. . is . if «A and C are not both positive. points.2fgh. + 2fxl yz + . The tangents at the nodes meet by threes in two points. . and that APlt BQ lt CE 1 are concurrent. The points of contact of the four bitangents are the intersections of the quartic with two lines.. + . 19. The tangents at A and B form a quadrilateral. . . and the point of concurrency is fx = gy = hz. 4. If the tangents from are harmonic conjugates with respect to and AC. CC we obtain points A'..f. + + . The reader should read Ch. III. V for and vice versa. . 26. and so do the remaining intersections of the curve with the tangents at C [/ o. 24. . What conditions must hold in order that conies I. The intersections of the tangents from each line joining the other two nodes lie on two lines. 25. + . and similarly Hence 6" is (g.] Ex. z 0. reciprocal with respect to Aifz 2 + . IV may be real? [The conic ax 1 * by 2 + cz i + 2fyz + 2gzx + 2hxy = is real.. Therefore is cy+fz = Q.. meet on a diagonal of the quadrilateral formed by the bitangents.. BB'.4 XVII AB TRINODAL QUARTIC 281 Ex. C . Show that conies II and V are always real.] . . 20.
282
TRINODAL QUARTIC
AC,
XVII 4
touch a conic.
Ex. 29. The lines BA', CA', CB', AB',
[aX 2 +
...
BC
+....
~(f+bc/f)ixv+
...
+
...
= 0.]
Ex. 30. The conic through ABC touching and passes through the points of contact of the tangents from C to the quartic.
AC
BC C
]
[gyz+fzx + cxy
= 0.]
to the quartic
lie
Ex. 31. The points of contact of the tangents from on a conic through C.
[(fh
+ bg) xz + (gh + of) yz
The points of through A, C and the intersections of
•Ex. 32.
= Ax + By2 +2 (ch  2fg) xy. contact of the tangents from A
2
lie
on a conic
Similarly
BC
with the quartic.
for the tangents
from B.
1
t
;•
[bgxz + ghyz
= By
+ (ch — 2fg) xy
fhxz + afyz
= Ax2 + (ch — 2fg) xy.]
Ex. 33. The conies of Ex. 31, 32 pass through the same four points.
Ex. 34. The tangents at C meet the quartic again in C1 and C2 and Cy C% meets the curve again at C\ and 2 Show that the points of contact of the tangents from C lie on a conic touching CC\ and CC\ at C\ and C'j and passing through A and B. Show also that through the intersections of the conic U = with this conic a conic can be drawn touching CA and CB at A and B.
,
C
.
[(/F+gG + hH)fH(hz2 + cxy)
=
cU.]
Ex. 35. Through two pairs of vertices of the quadrilateral formed by the bitangents can be drawn (i) a conic touching at at A and B (ii) a conic touching conic I at its intersections with (iii) two conies each touching at A and at B and each passing through the points of contact of two bitangents.
CA
CB
;
CA
CB
AB
;
[(i)
uz + cxy
=
;
(ii)
u2 — bcx 2 — cay* + abz
1
=
;
(iii)
wz + cxy+\/abz2
=
0.]
Ex. 36.
Show
v
that, if
u lEfx+gy + hz,
= cxy + uz — tz w = bcx + acy — u
2
,
2
2
2
+ 2t (cxy + uz) —
t
'z
2
,
touches the quartic at its four intersections with than A and B. Show that v = touches and CB.
iv
=
v
= 0,
other
CA
[The quartic
is
v
2
+z2 w =
is
0.]
Ex. 37. Show that w = that in the former case w that in the latter case to
= =
a linepair, if t = h or if t + 'Jab. Show is the tangents from C to the quartic, and is two bitangents. ,
=
is
Ex. 38. Show that the locus of the pole of AB with respect to v> = a conic through C, C, R lt R$; and that this and the two similar conies pass through the same four, points.
[Ax2 By 2 + Fyz Gxz
Ex. 39.
Show
that
= 0.] w — divides ^Bj harmonically.
XVII
5
TRICUSPIDAL QUARTIC
§ 5.
283
Tricuspidal Quartic.
The results of the preceding section require modification if the three double points of the quartic are not all ordinary nodes. Suppose the double points are all cusps. Since the tangents at each double point are coincident, the quartic takes the form
a —
„
ar
H
— —
b
o
c
^
i
I
y
zL
+
2V(bc) yz
2V(ca)
H —
zx
+ —
2V{ab) xy
—
u.
loss of generality.
all be taken as minus without For unless the quartic is a pair of coincident conies, three plus signs or one plus and two minus signs are impossible while if (say) the first sign is minus and the other two plus, we can make all signs minus by replacing Va by — Va. Finally, replacing x, y, z by /a. x, Vby, /cz we obtain the equation of the tricuspidal quartic in its canonical form
;
The ambiguous signs may
x2
y
2
s2
.
yz
zx
z
xy x
The cuspidal tangents y
= z,
=
x,
=y
are concurrent
at the point (1, 1, 1). Since the equation of the quartic
(yz + zx
may be written + xy) 2 = 4xyz(x + y + z),
is the bitangent, its points of contact being its x+y+z 0. intersections with yz + zx + xy The equation of the quartic may be also written
=
=
5+2/ The tangential equation is
a;

2
+z
2
=
0.
(A
+ n + vf =
27\pv.
§ 2
In the equations of the threecusped hypocycloid given in (ii) put 2Z=a + 2x; </3y, 2X=ax</3y,
2Y=ax+
=
are the sides of the triangle 0, Z 0, so that whose vertices are the three cusps.
X=
Y=
This gives
6Z =
a{l + 2cos(0 + §7r)} 2
*
67=c6{l + 2cos(0 + 7r)} 2 6£ = a{l+2cos0} 2
,
,
,
from which
is
readily deduced
any threecusped quartic is projected so the cusps and the point of concurrency of the cuspidal that
It follows that, if
5
284
TRICUSPIDAL QUARTIC
XVII
triangle, the quartic
tangents become th» vertices and centroid of an equilateral becomes a threecusped hypocycloid.
Ex. 1. The coordinates of any point on the tricuspidal quartic in canonical form are expressed rationally in terms of a parameter t by
{tiy>x
Ex.
2.
=
(thlfy
=
4z.
conic is inscribed in a given triangle and passes through a fixed point. The locus of the point of concurrency of the lines joining the vertices of the triangle to the points of contact of the opposite sides is a tricuspidal quartic.
[If the point is (X, Y,
A
Z) and the conic
is
(*/fl*
+ (yA)*+(*/fl* =
+ (r/y)* + (Z/*)*
o,
the quartic Ex.
3.
is
(X/*)l
=
0.]
variable cubic touches three fixed lines at fixed collinear points and has inflexions at its other intersections with the fixed lines. Show that .the line of these inflexions envelops a tricuspidal quartic, having the line of the fixed points as bitangent.
A
[Writing
inflexions of
down the
conditions that
(0,
—v,y.)
and
0,
(
— v,
0, X)
are
xyz+(x + y + zy(\x + /j.y + i'z)
=
we
obtain
is
which
of the third class
and fourth degree since
it
has
(1, 1, 1)
as
a bitangent.]
Ex. 4. The locus of the centre of a conic having a given equilateral triangle as selfconjugate triangle and having its asymptotes inclined at
1 an angle tan J */3
is
a threecusped hypocycloid.
Ex. 5. A fixed radius of a circle and' the tangent at its extremity intercept on a moving line a segment which is bisected by the circle. Show that the envelope of the moving line is a threecusped hypocycloid. Ex. 6. The envelope of the Simson (pedal) line of a given triangle is a threecusped hypocycloid. The tangents from a vertex of the triangle are the two sides and altitude of the triangle through that vertex.
[Writing down the condition in trilinear coordinates that the lines perpendicular to the sides of the triangle at their intersections with
are concurrent,
\a + rt3 + vy= we get the tangential equation
.
of the envelope
fiv(ii
2
(1
+
cos .4
cos
.
B
.
cos C) \/iv
.
=
2 sin 2 A
.
.
cos
C + v cos B)
which touches sin A Oi + sin B See Converse, Annals of Math.,
this result.]
+ sin C
II. iii
at the circular points. (1904), p. 105, for an extension of
y
=
may
The
Ex. 7. The equation of a quartic with one real and two unreal cusps be put in the form
(a?
+ y*2 xzf =
4z* (x* + y*).
quartic can be projected into a cardioid by a real projection.
§
[Replace x, y in
quartic see § 2.]
5 by
x+ iy.
For other examples on the tricuspidal
'
'
'
XVII
§ 6.
6
OTHER QUARTICS
285
Other Quartios with Three Distinct Double Points.
(0, 0, 1) is a cusp of ay 2 z 2 + bz 2 x% + cx 2 y 2 + 2xyz (fx + gy + hz) 0, i. e. ab = K\ a flecnode if the terms
The point
if
=
.
.
(i)
C=
It is
cx 2 + 2hxy + by 2
(ii)
multiplying s in
2
(i)
have a factor in common with the terms
®y(fa+gy)
multiplying
z.
(»i)
This
is
the case,
0,
(ii) r
if
fF+ gG =
It is a biflecnode if
i.e.
is
2fgh
=
a/ 2
+ bg\
which
is
a factor of
(iii),
only
possible if
_
_
is
I.
Quartics with three Biflecnodes.
Ex. 1. If two of the nodes of a trinodal quartic are biflecnodes, so the third.
[/=</
Ex.
2.
= =
fc
0.]
a quartic has biflecnodes A, B, C, the tangents at A are harmonic conjugates for AB, AC. The tangents at A, B, C touch a conic for which ABC is a selfconjugate triangle.
If
Ex. 3. If a quartic has three real biflecnodes, one is an acnode, and the others are crunodes. Its equation cap be put in the form
2
a?
+ jr 2
= s 2
.
If a quartic has three real biflecnodes, the intersection of two diagonals of the quadrilateral formed
Ex.
4.
acnode is the by the tangents
at the crunodes.
Ex. the
'
5.
Show
that the
'
crosscurve
x*f
carbonpointcurve
'
=
a 2 y1 + b i x*,
and the hourglasscurve
are curves of this type. Trace the curves, and show that the two former are the loci of intersection of lines parallel to the axes of an ellipse or hyperbola through the intersections of any tangent with the axes.
to a quartic with Ex. 6. The points of contact of the tangents from three biflecnodes A, B, C lie on a conic which meets the quartic again with respect to the quartic. on the polar line of _2 Then use and as (£, /?, £). [Take the quartic as ar 2 + ir 2 +.z =
the identity
{£W+e)x+
•••
, + }{f*(y +« )+
1,
•••
+
...
•••}
=
Hyz + £&x + £r,xy)
2
{( v
+ i 1)x>+
+
...
+v(ye+
...
+
...}]
t
1
286
Ex. 7. If in Ex. 6 a line whose envelope
OTHER QUA.RTICS
lies
is
XVII
6
[The conic
on the quartic, the points of contact lie on as a selfconjugate triangle. a conic having and the of Ex. 6 degenerates into x/t + y/n+z/C
ABC
=
polar line of 0.]
Ex. 8. Any point of the curve x~* + y' Four points with parameters 0,, $ a $ 3 crunodes, if <j>, + d 2 + <f 3 + 4> t = (2n — l)ir.
,
= z~
a
is
(sec $, cosec
<b, 1).
,
<p± lie
on a conic through the
t 1 t t ti i i
The four points
?,
are collinear,
</>j,
if
in
addition
=
1,
where
=
&c. Three points with parameters
tanj^,
(fj, <p 3
t)
are collinear if
3
.
k hh (V, + t s h + h
Ex.
9.
= *, + 12 +
2 The conies through the crunodes A, B of a; + y~ 2 — z~" osculating the quartic at P, Q, R, S meet the curve again at P', Q', R', S'. If A, B, P, Q, R, S lie on a conic, so do A, B, P, Q', R', 8'.
Ex. 10. Two conies through the crunodes A, meet the quartic again in P,, Q lt R 1 , Si and A, B, x Q x §„ lie on a conic, so do A, B, R u 2
P
,
,
P
,
B of 2 + y' = z~'' P Qit R S If Slt P S
a;
1
2
,
2
,
2
.
2,
2
.
Ex. 11. The sextactic points of ar' + jr' = z~ l are +2i, 1), +1, 2i). (±2*, +*, 1), ( + 1, ( + i,
[The method of Ch. X, § at a sextaetic point.]
2,
Ex.
17, gives
sin4#
=
or
cos40
=
17
Ex. 12. Through the crunodes A, B of the quartic of Ex. 8 four conies can be drawn having fourpoint contact with the curve. Their points, of
contact are the real sextactic points. The conic through A, B and three of the sextactic points touches the quartic at one of them.
Ex. 13. If in Ex. 9 the conic ABPQR meets the line joining S to the acnode on the quartic, the conic ABP'Q'R' meets the line joining S" to the acnode on the quartic.
Ex.
14.
crossratio
Obtain the envelope of a line divided in a range of given by a quartic with three biflecnodes.
in §
3.
[Put
The envelope becomes the conic touching the
biflecnodes, if the range is
f=g = h"=0
tangents
equianharmonic]
at the *
Ex. 15. The equation of a real quartic with one real and two unreal 1 (x 1 + # 2 ) 2 biflecnodes can be put in the form xyz The quartic can be projected into the lemniscate of Bemouilli
=
.
»2
=
a* sin 2
6.
II.
Quartic with, two Cusps and a Node.
[See §
2,
Ex. 10
;
and Ch. X, Fig.
1.]
Ex.
1.
A
quartic with two real cusps
2/'z
2
form
+ z x + xl y %
l l
=
and a real node can be put 2xyz{x + y + mz).
in the
The
inflexions are unreal or real according as in does or does not lie
between 1 and 7. [The quartic .meets x (yz)
(Jct^kf
=
{t
+ 1)
yz at
l,
where
+ t+l), kt' + t + 2k (iw + 1) = 1.
hfi),
6
XVII
Now
i.e.
OTHER QUARTICS
§ 1 (v),
287
using Ch. X,
we get 3kf + (8k+l)t + B = §{m + l){x2 + y ) = 2(m' + m + 16)xy
!1
1
for the inflexions.]
Ex.
lie
2.
The bitangent of the quartic of Ex.
1 is
2x + 2y + (m + l)z
=
0.
Its points of contact are unreal or real
according as
between  1 and 7. [The points of contact
illustrate
m
does or does not
lie
2
on
2
2a;
2
may
2x + 2y + 9
=
by tracing
0.]
a; ?/
+ xi + y
 (m 3) xy + 2y 2 = 0. The reader = 2xy {x + y + 8), with bitangent
1
Ex. 3. Find the value of m in Ex. 1, if polygons of 4, 5, or 6 be inscribed i.n the quartic whose sides touch the curve. [If t u tt ts tt are the parameters of collinear points
, ,
sides
can
h + h + h + h =  2 (»» + 1)> h" + h' + *s + h' Eliminating ts and tt from these and t s = tt we have 4V2 = {2 (m + 1) + (*, + *,)} {<! t2 + (t1 + ti )}. But this is the condition that the tangents at the points 2 2t2 ) on z 2 = ±xy should meet on (1, £j 4xy = {2(m + l) x + z} {y + z}.
1
1
_1
1
= ~ 12
,
,
(1, t
2^) and
,
Now use the properties of invariants of conies to obtain the condition that polygons of 4, 5, 6 sides can be inscribed in the lattericonic and circumscribed to the former. We find
— 2 the cuspidal tangents pass that in the case through the points of contact of the bitangent, and that any two lines harmonically conjugate with respect to the tangents at the node meet the curve again. at four points such that the tangents at these points form a quadrilateral inscribed in and circumscribed to the quartic.
Ex.
4.
Show
«~2, fori, Hr m=
1, 2), (1, 2, 2). two other tangents to the curve from the point with parameter T are given by 2k (T+ 1) t 2 + (kT2 + 4kT+ T+l)t + 2T (kT+ 1) = 0. 2 2 If tx *2 are the roots, we readily verify that x = kt1 y and x = kt2 y form a harmonic pencil with x 2 + y 2 — 2mxy in the case m = 2, * = $.]
[(i)
(ii)
The points of contact The points of contact
are
(2,
of the
,
Ex.
5.
Show
that the line joining the points of contact of the other
two tangents from any point of the bitangent of a quartic with a node and two cusps envelops a unicursal curve of the third class. [The tangents from the point (x, y, z) are given by 2 2 s (t + 2)x + kt (2kt+l)y2(kt + t + l) z=0. If the point lies on the bitangent z = — 4k(x + y), this becomes 2 2 {2kt 2 + t + 2} {(4/fc 2 * 2 + 6fc*+4&+l)x+fc(47c* + « + 6* + 4)y} = 0. Hence if tx t3 are the parameters of the points of contact, while 6kv + (8k + l)u + 6 = 0. and t 1 t2 = v, t1 + t2 = u But the line joining the points with parameters t lt t2 involves u and v
,
in the third degree.]
* See Eoberts, Proc. London Math. Soc, xxiii (1892), pp. 202211, for this and the following examples.
:
288
,
OTHER QUARTICS
B
and a node
C.
XVII
Show
6
Ex. 6. A quartio has cusps A, following points lie on a conic
:
that the
(i) A, B, the points of contact of tangents from points of contact of the bitangent.
A
and B, and the
(ii) A, B, the inflexions, the remaining intersections of the curve with the tangents at C.
[These and many similar results are proved by noticing that the curve can be projected so as to be symmetrical.]
and a node C. Show that the quartic has two cusps A, of the tangonts at A and B is a C to the intersection double line of the involution determined by the tangents at C and the
Ex.
7.
A
B
line joining
lines
CA, CB. Given A, B,
C, 0, find the locus of
the inflexions and of the points of
contact of the bitangent.
[A quartic with nodes at A, B and a cusp at a conic through ABO. See Ch. V, § 4, Ex. 8.]
Ex.
8.
C
touching OA,
OB
;
Reciprocate Ex.
III.
5, 6, 7.
Quartics with two Flecnodes and a Node.
real flecnodes
Ex.
1.
The equation of a quartic with two
the form
{x*
and a node
can be
putm
+ y 2) z^x^f + Zhxy
{z
+ x){z + y)
= 0.
Show
that the
C.
Ex. 2. A quartic has two flecnodes A, B and a node C. following sets of six points lie on a conic
(i)
(ii)
A,
A,
B and the points of contact of the tangents from A, B, B and the intersections (other than A, B, C) with the
curve of
the tangents at A, B, C. [These and many similar results follow at once from the fact that the quartic can be projected into a quartis with symmetry.]
Ex. 3. unicursal quartic with two unreal flecnodes can be projected into the inverse of a conic with respect to the reflection of a focus in the corresponding directrix.
IV.
A
Quartics with a Biflecnode and two Nodes.
A to a quartic with nodes A, B, C form a harmonic pencil with AB, AC, and similarly for the tangents from B. Show that C is a biflecnode and that the tangents at A form a harmonic pencil with AB, AC, and similarly for B. Show that the equation of the quartic can be put in one of the forms
Ex.
1.
The tangents from
;
z* (a?
±y i + 2mxy) ±x"yl =
0,
0.
[F = G =
gives
/=g=
if
the curve
is
not degenerate.]
Ex. 2. Show that, if a quartic has nodes A, B and a biflecnode C, it can be projected so as to have two axes of symmetry ; unless A is a crunode and B an acnode, or vice versa, when it can be projected so as to have a centre of symmetry at C. Trace the projected curves in the various cases which can occur. Prove that, if A and B are acnodes, C is a crunode.
[Putz
=
l in
Ex.
1.]
; ;
XVII
7
UNICURSAL QUARTICS
289
Ex. 3. A quartic has crunodes A, B and a biflecnode G. Show that the tangents at A and B meet at the vertices of a quadrangle having A, B, C as diagonal points; and that their remaining intersections with the curve are the vertices of a quadrangle whose diagonal points are C and
two points on AB. [This and similar theorems follow from symmetry, or by putting /=<7 = in§§ 3 and 4.]
(ii)
Ex. 4. A unicursal quartic cannot have (i) a biflecnode and a cusp, a biflecnode and a flecnode, (iii) a flecnode and two cusps, (iv) three
flecnodes.
§ 7.
Unicursal Quartics with
Two
Distinct Double Points.
Unicursal quartics with only two distinct double points include quartics with a tacnode or rhanrphoid cusp and another double point. The reader will readily obtain their properties by modification of the results of Ch. XVIII, §§ 14 and 15. He may illustrate his argument by tracing the curves of Ch. Ill, § 6 (xi) to (xv), § 8 (v) to (vii).
I.
Unicursal Quartics with a Tacnode.*
A quartic has a tacnode at C and another double point at B the tangent at C and BA the harmonic conjugate of BC with respect to the tangents at B. Show that its equation is
Ex.
1.
CA
is
(yz +
x*f
=
(1
 m) x 2 (x2 + p^y + p 2 y2 )
;
in general.
Show
Ex.
2.
that p 2
and that
B
= 0, if B is a cusp cannot be a biflecnode.
(1
— m)p^ =
4.p% if
B
is
a flecnode
The bitangents \, h 2 of the quartic of Ex. 2z = (1 + /m) (p x x +p i y).
1
are
Ex. 3. If Cj, C2 are the points of contact of the tangents from C, then Clt C2 and the points of contact of \ lie on a conic touching the
quartic at C.
[yz +
x*=
(l
+ Vm)(x2 +p xy+p 2 yi ).]
1
Ex.
4.
The
points of contact of 6 t
and
6 2 lie
on a conic touching the
quartic at C.
[2yz + 2x*
Ex. Ex.
5. 6.
AB,
= (1m) (ix^+p^y+p^).] C^, bu 6„ are all concurrent.
touches the quartic at C,,
A conic
C2 and
touches the tangents
from B.
[{PxX + p^yzf
= (lm)p
2
{xi
+ p xy+p i y 2 ).]
1
Ex. 7. The points of contact of \ and its intersections with the tangents from B form an involution with a double point on CA.
Ex. 8. The points of contact of the tangents from touching the quartic at C.
A
lie
on a conic
[Write down the polar cubic of A.]
*
22Ifi
For other examples see Messenger Math.,
xlvii (191V), p. 95,
U
7 [J? ! Ft is 2x + 9>(a — l)y + z = . meets & at L lines joining J. quartic with a triple point is converted into a unicursal cubic by projecting two points of the quartic into the circular points and then inverting with respect to the triple point. b. 1a 2 ). if B is a flecnode . a2 ). CA. Ex. [yz + x* and Cx lie on a conic osculating = 2y(x + ay). CB. 4. 1. 7. CC1 are concurrent.6.s y (x + ay). A quartic has a rhaniphoid cusp at C and another double point at B CA is the tangent at C and BA the harmonic conjugate of BO Ex. Ex. Ex. Quartics with a Triple Point. is a cusp a = 1. B. The conic osculating the quartic at C and passing through the remaining intersections the curve passes 2 of lt 2 with x through Bj. Show that a = 0. 12. 9. 2. The tangent from C is x + ay = 0. AB.a) y\] . 1 is x + ay *=z touching where x* — xy — ay* = 0. A . touch the conic of Ex. The conic osculating the quartic at C and passing through the intersections of the curve with (other than B) passes through Cv AB [yz + x* = 4:y(x + ay). 12 X harmonically by CA. F F [yz + x 1 + 2 xy = 4 (1 . and BK. CH as C to the points of contact of 6 has CA Ex. & lie Ex.290 II. The tangent from B is z = (l + a)x. if B cannot be a biflecnode. 9 at B passes through Bt [yz + x* C and touching the conic = 2xy. 0. F F is . with respect to the tangents at B. .. on a conic for If BB and by the a double ray. A lie on a conic [Write down the polar cubic of A. lt 3 Ex.] Ex. touching at B x (\a. 10. XVII 7 Ehamphoid Cusp. The points of contact of the tangents from osculating the quartic at C. 6. &c. . that its equation is = 4a. The points of contact of b the quartic at C. B 1 C1 are concurrent. UNICURSAL QUARTICS Unicursal Quartics with a 1. 13. and the conic x i —4 xy + 2yz = 0.] Ex. 3.] Ex. If the B CA is FF F : F . touching at Cj (a. The conic osculating the quartic at of Ex. 8. BK and KK X 2 is divided § 8. (yz + a?)* Show . C. [z(2x + ay) = (2 + a) a? + 2axy. 11. the involution pencil formed by CB. B lt Clt and the points of contact of which AC and BC^ are conjugate lines. K K . Ex.] Ex.1.] Ex. 5. and that B The bitangent 6 of the quartic of Ex.] line tangents at meet the quartic again at 1 and 2 the divided harmonically by 1 3 and the tangent at B to the conic touching CA at C and passing through B.
1). replacing z by z +px + qy and choosing p and q so as to make the coefficients of x3 y and xy 3 zero.. bitangents* I. we may suppose u = xy (x + y). the equation of the quartic is An zwjv = . We may simplify this equation by a suitable choice other of the two vertices of the triangle of reference. B.. 18. IX. where u = a x3 + 3a 1 x 2 y + 3a 2 xy 2 + a 3 y 3 v = A x* + 4:A x x 3 y + 6A 2 x 2 y 2 + 4>A 3 xy 3 + A^y*. the three the quartic O meet [Project A.y XVII 8 QUARTICS WITH A TRIPLE POINT 291 Of course quadratic transformation with the triple point and two other points on the curve as the points C. If we take the triple point as (0. V2 . Ex. collinear points of a cubic are collinear. ( (iii) meet the curve ( where ) Va x2 + txy + Vb y2 2 = Va x2 + Txy . = (ii). O osculating the quartic at again on a conic through A. (T0. If is A and B. a triple point of a quartic through conies through A. Hence from each property of a unicursal cubic can be deduced a property of a quartic with a triple point. The coordinates of any point can be expressed rationally in terms of a parameter by considering the intersection of the curve with a line through the triple point. r = Therefore the lines (iii) are the four bitangents of (ii). and is a diagonal of the quadrilateral formed by these Quartics with a Triple Point. B of Ch. The reader then z will easily verify that. 0.. 295. Then. comes to the same thing. O. + 2t(</ax+Vby) = (ii) z+2T(Vax. A. p.B Then : ' The tangentials of three * to the circular points and invert with respect to O. For instance.'] An exception arises if h2 = ab. B. if all three tangents at (0. Va+ Vb) 2 a + 2h + b. 0. alternative is to reduce the equation of the quartic to a simple form by a suitable choice of axes or triaDgle of reference. See Ex.Vby) = 0. (i). . 1. 1) are real. § 1.Vb 2 2 ) = 0. 19. we may reduce (i) to the form zxy(x + y) = ax i + 2hx2 y 2 + by i . if (tVaVb) 2 = a + 2h + b.
Ex. s 5= are four lines. In Ex. 3. whose reciprocal would be a quintic with 6. r = p = s 0. 11. B. and also with respect to the lines joining to the other intersections of with the quartic. A triangle is inscribed in a quartic with a triple point 0. Show that OQ and OR are harmonically conjugate with respect to OP and 0. Q. of a quartic meets the Ex. a triple tangent. For the quartic § 8 (ii) the equation of the conic is z2 + 4z (ax + by) [It is .] Ex.8 (ab + btyy*} + 64ab (a + 2h + b) {zxy (x + y). § 4. and OR trace out an involution pencil. Any diagonal of the quadrilateral formed by the bitangents of a quartic with a triple point O meets the curve and the conic through the points of contact of the bitangents in an involution whose double points lie on two tangents at O. Ex. XIII. 12. OQ curve again in A.] q= s. The points of contact of the four bitangents of a quartic with a triple point lie on a conic. Q. r q «are concurrent. R.] Ex. Show is a cubic.8 (ab + bh) y* = 0. B. Ex. CA. is a quartic with a triple point and with these lines as bitangents.292 QUAETICS WITH A TRIPLE POINT XVII 8 which osculate Ex. No and circumscribed to triple point [The quartic could be transformed quadratically into a quintic with a and three cusps. 0. R vary. If § 3. Ex. The line y = tx joining the triple point of § 8 (ii) to the remaining intersections of the curve with the line joining the sections of y = tx x and y = t 1 x with the curve is given by 2 bt + {b + av(a + 2h + b)u} t + av = 0. The conic OABQR meets the quartic again at S. B.] Ex. Ex. P. 7. + */r + a/s = Vp + x/ff such that p = s. B. XV. or we may use Ex. the quartic at P. A quartic with a triple point passes through three fixed points A.ax* 2hxi yi . 11. AB again at six fixed points. Ex. at A. § 4. {See §3. [See Ch. 3. 6. 2. 9. R. A conic through the triple point If A. The lines joining to the remaining intersections of the quartic with the sides of the triangle form an involution. 13.] triangle can be both inscribed in a quartic with a triple point. Q. XIII. C and meets BC. Ex. 10. Two P on the R AB [See Ch.) EE 1. 1 there are three conies through A.] conies through the triple point O of a quartic meet again curve. .8 (ab + ah) 2 x* . B. 4. = = 0. Show that A. 0. 9. that the locus of [Use Ex. R lie on a conic. B. Ex. inter where 91^=1. 8.by*} = * . P. 5.9. and they both touch the quartic at Q and respectively. P are fixed and Q. [Use Ex.] Ex. •» (l + sfj) (1 + fc. readily seen that these bitangents are 1 {z" + 4z (ax + by)~8 (ab + ah) x . 7 and Ch.
. The curve 16(xy)(2x + y)y with bitangents = a. t + 4(. Let Ea 2 3 3 be the three pairs of vertices of 1 the quadrilateral formed by the bitangents of a quartic with a triple point 0. 2 l! 3(a a s a 1 a )X/i3(« a i! « 2 l )n 2 = . Show that. . ts and l s form an involution. . Ex. Add and subtract these equations. . 10. The conies of closest contact at O with § 8 (ii) l are i ) zxb{x — xy + y 2 2hx = zya(x 2 xy + y 2 ) — 2hy1 = 0. (vi) and l lt lit l3 with the quartic.3 a2 Ai + 3 a^A % . . and subtract each from the equation s2 of the conic of Ex. be the conies of closest contact with the quartic the lines joining to their remaining intersections Then prove that tt and l x divide E1 1 harmonically. and the conic 52ar52jBy27 y . F and OFlt t a and \. (vii) A conic passes through E1 Ft and the four intersections of S2 and S3 It touches at the harmonic conjugate of tx for t2 and ts (viii) The conic through the intersections of St and S3 touching tx at meets t 2 and t3 in points collinear with the intersection of £./by) = 0. 2 z (x + tj) + (3a + 6)a. four (v) Let Slt S2 S3 . The conic 6x 2 + 4xy + 3y' 24:X + 8y96 i = passes through the points of contact of the bitangents.16a. contact with the conic through the points of contact of the bitangents at its intersections with tx . F E F . . . + (3a + 36 2A) xy + (a + 3b)if = 0.V b) (>/ax/by)z &(h+ ^ab) (a/ox. 12. if the tangents t lt t 2 t3 at are taken in a certain order.^/byi2 = 0.4 (os A .] 0.152^48 = touches the six inflexional tangents (only two are real). Through the four real intersections of these conies passes another conic through the 1 inflexions./a. For the curves § 8 (i) and (ii) the tangential equation of the conic is 4(A A i 4:A 1 A 3 + BA i2)v'i 4:(a . then t E F .A i Sal A s + Ba2 A 2 a A 1 ) vX .] Ex. = 0. at O. 10 to 15. The lines OE1 . 12x* Wx*y ix^y* + 8xy + Syi s Sx + 2y + l2 = 0.a A s ) v/i — 3 (a x a3 — a 22 < 3 ) X2 X 2 + Al 2 . will illustrate Ex. . A line divided equianharmonically by a quartic with a triple point O envelops a conic touching the six inflexional tangents and touching the Hessian of the tangents at O. +2 = 0. : (i) t2 t3 are the pair common to the two involutions subtended at by the points of contact of the pair of bitangents which meets at Ex and the pair which meets at F1 (ii) The two conies through and the points of contact of these two pairs of bitangents both touch t 1 at and have fourpoint contact. . . Aii + + 4^(\ + F)v + 4(^+3o6)j. 1 . (iii) (iv) The conic OE2 F2 E3 F3 touches tt at 0. Through E2 F2 E3 F3 can be drawn a conic having double .XVII 8 (ii) QUARTIGS WITH A TRIPLE POINT the diagonal is 293 [In § 8 z = 0. . x4 = 0. y3 = 0. 13. and E F1 [Two pairs of bitangents are 1 z" + 4(*/a + Vb) {/ax + */by)z8 (h.a/oB) (*/ax + .
of § 8 s (i) verify that «>) C20)a « + 2Cn .4 A i a? + 8A 1 A S Oj2 + 6 . C3l C41 .. „(*« + = ^20)2 (3C40)0 s + C aul z + 2 CM>2 ). For the quartic § 8 (ii) the inflexions lie on 2a (b + 3h) xi 4{ab + h 2 )xy + 2Ji. «+C „.* 2 [By Chi VII.. * to For this equation.2 a^ a 3 . which suggests the more general form. A x i Deduce that the six inflexions lie on the conic 3 Ci0 . ~ «aT + 2 («i «s . and write down the condition that the resulting pencil of lines is equi anharmonic. .2 a a a + a ll 3 2 1 1 i . 14 pass through th» same four points. are invariants or covariants of i u and by Cm C11 C40 . . 16...)2Ci. (3 • With the notation . v given where C20 2 &c. § 7. (i) Show which that the.2 «ias) + 8 2 «i as + ^ j4 3 (2a a3 — 6 a a2 ) . 13.. Ex.«2 2 ) «2 . 0.6 «0 a °2 a + 4 a a + 4 = {A (o a .)(C10 .2a a 2 3 ) 2 2 «2 .*) C31 != {^„ (5a a 3 MojOjaj + 9 2 ) + ^ ( — 16a a2 a 3 + 28 x Oj + v4 2 (30ff a 2 2 — 12a Oj a 3 . The inflexional tangentB are divided equianharmonically by any quartic] Ex.. C2 2. am indebted . o 294 QUARTICS WITH A TRIPLE POINT XVII 8 [Eliminates between Xx + py + vz = and the equation of the quartic. I Miss J I. Show that the bitangents of these curves z = * 4 2tx / 26f St J A where (i) (ii) 27 (18a + l)< 2 + 462 27i4 + (18a + l)r46 2 = = 0. — &A % A 3 a a x + A z A a 2} x* + .2 = {A2 «sa 6A^i«2«s + A<M 9a2 2 .ao «s) + A % (a 2 a3 2 + 3 a a 2 a 2 2 . 0.. (ii) one tangent is real.. o = AKa s A K K K . i = {a a a^)x' + •> = (a A 2a A + a A )x + = «0 "s* . (a + 3h) y 2 = {3a2h)zx+(3b2h)zy + z . 14 (v) the inflexions lie on 3C2012 * + 2C11I3 = Ex. E. ... 10. i . 15.3 ^I 2 ^ l 2 s 3 x 2 s 1 z ) + (a a i a *)(4:Ai a3 + 6A2aa 4A s a + A i a )} x + . 2 = 0.] The three conies of Ex. equation of a quartic with a triple point at three tangents are real. 14. fl Z S 3 l l 8 2 ff S .18«j2 o2 ) +A S (4a a a3 — 16a o 1 o2 + 12a! 3 ) + ^ 4 (a 2 a2 a a12 )}a: + .ij A 2 (12 a x a2 + 4 a a3 ) + . „ s2 + C 31 + 2 C22 . Ex. can be put (i) into the form (ii) z(dx2 yy s )+xi + Gax 2 y 2 + 4:bxy s z(3x 2 y + y s )x i + 6ax 2 y2 —ibxy s are ' = = 0...«i") + A i + 2 A s (Oi a2 ~ «o as) «2 . i s . Colomb. 0.«i ) 2 s ! — ^a^.a .4 2 2 a x a2 — 2A 1 A t a a..
&c. The conic through the points of contact of the bitangents the triple point O touches the linear branch at O. 3. 18.] Quartics having a Triple J*oint with two Coincident Tangents. triple point at which two Ex. + = 4 2/4 84 2/ 0. 2x + y + 2 = 0. Ex. piq. Determine whether bitangents in this case. one tangent at number of bitangents and inflexions for each type. Obtain the conies through its inflexions. its Hessian. 9x8y+36 = II. and is find the [For § 8 (ii) the condition h? = ab. whose bitangents 0. By replacing x. 19. For § 8 (ii) this becomes a + h + b = 0. and the pointpair unreal.Ex. [C41 — 0. 2. If one tangent at is real. &c] % . The equation of a quartic with a 2 tangents coincide may be put in the form zx y = y"(2 xt+ y) + ax*.. !i . xiy. Each of them is divided harmonically by two of the tangents at the triple point 0. 10. [y(y + z) = 2ax\] . 0.] Ex.+ y i ) 0. 16 should degenerate into a pointpair. b. three bitangents are concurrent. Sketch roughly the different types of quartic with a triple being the line at infinity. 22.] . x + iy. The bitangents are z + y = ± 2 </ a (x 4 y) Ex. The bitangents of a quartic with a triple point three tangents are real are all real or all unreal. 14 should degenerate into linepairs. a. Determine the condition that the conies of Ex. Their points of contact 2xy+S = lie 2xy6 = on 2 (a: 2. one is illustrated by S6y(y + l)x with bitangents 9a. and . ! + 81 10y + 45 = 0. two bitangents are real.*+ (ylf = 21. 17. Find the condition that the conic touching the inflexional tangents of the quartics of § 8 (ii) or Ex. and determine the point 0. when the conies become respectively unreal and real linepairs. h in § 8 (ii) respectively by p + iq. If the conic touching the inflexional tangents degenerates. Ex. Sx + 4y + 18 = Sx2y + 9 = 0. [Illustrate by tracing 2ixy (x + y) are the line of infinity and = (2x i. Ex. this pointpair is real ^ is or unreal. y. 20. The fourth bitangent is divided harmonically by the Hessian of the tangents at 0. r\(a + b) obtain the equation of a quartic with a triple point at which one tangent is real.] Ex. [There are four such types . Determine whether the points of contact of the bitangents are real or unreal. 1. XVII 8 QUARTICS WITH A TRIPLE POINT at 295 which the Ex. The inflexions lie on the conic Eys+2xz = 2ax [Eliminating a between the equations of the curve and wegef 2f + x 2 z = 0. 21.
a < 0.] Ex. the line joining the inflexions. Ex. 1. distinguishing the cases a 4. The tangent to the superlinear branch. = 6X». The inflexions are real or unreal according as the points of contact of the bitangent are real or unreal. 1. 2 . Any Ex.] Quartica with a Superlinear Branch. A conic passes through the inflexions and the remaining intersections of the inflexional tangents with the curve. § 8. and touching the inflexional tangents. t three can be put in the form zy 3 The bitangent is z == 0. [yz = 4aa. and the line joining the remaining intersections of the inflexional tangents with the curve. The line divided harmonically by the quartic envelops a conic touching the curve at the singular point. We find that then I — m = h = 0. 1. the bitangent. are all concurrent. A quartic has two linear branches having threepoint contact with one another.] Ex. 3. The inflexions are (±ai. the equation = (yz + x 2 f + 2 y (yz + x>) + my) + y (ax 1 + 2 hxy + by 2 ) Choose a triangle of reference ABC such that C is the double point. 2. [It is X 2 +16a/i» + 128aV = 0. 1.] TJnieursal Quartics of Class Six with one distinct Double Point. 3. Ill. of Order Three.] Ex. 1 are (p^ + iq) y = 4^2 and y" = 0. IV. ry + 12a»» III. 6. is 1 [By Ch. which touches the bitangent and touches the quartic at the singular point.] line met by the quartic in an equianharmonic range passes through the intersection of the inflexional tangents. 6. The bitangents of the quartic of Ex.. (yz + x*y =y (Ix 1 (px 1 + qy % ). 5. 36a 2 ). Ex. Show how to transform the quartic into a conic by quadratic transformation. 2. see that the curve may be projected into one with Trace this projection. B the other intersection of the osculating conies at C. 4. Show that its equation can be put in the form . we an axis of symmetry. 1. [Putting z = 1. a = 0. . 4a 2 ) and the inflexional tangents are Ex. Ex. 296 QUAETICS WITH A TKIPLE POINT XVII 8 Ex. Ex. = z+ 8aix — 12a'y at ( = 0. A the intersection of the tangent at C with the other common tangent of the conies. [Put zx /y 2 for z in Ex. * They meet the curve again + 3ai. Ex. . > 0. The equation of a quartic with a superlinear branch of order (Bay^ — x 2 ) 1. The envelope of the line divided equianharmonically by the quartic is a conic touching the inflexional tangents and touching the superlinear branch at 0.
2. ABC of conies meet the curve at seven points coinciding with (0. 6. £w 2 Conies are yz + 'dx2 = 0. If in IV. 8yz + 35x2 0. Show that the quartic can be projected into one with an axis of symmetry. 'A conic through two inflexions and the points of contact of a bitangent touches the curve at &'] Ex. Put 2=1. 1. Properties of the curve can be written down from the symmetry. Ex. Indicate roughly its shape in the cases p>0. a = 2 I .] Unicursal Quartics of Class Five with one distinct Double Point. 1). such conies. Ex. if any such conic meets the curve at P. 3. More generally. t2 t3 ti Ex. h ts tt + <x ts tt + tt t% tt + tx tt ts + 1 = The tangent at any point is (l+4?)x + P(l2ts )y + 2tz = 0.] . 2 >0.] The curve is of class 5. the equation can be put in the form (yz + x2 ) 2 {yz = xy s . 2 [Putting zx 2 /y for z we obtain z = xy. (t . p<0.XVII 8 QUAETICS WITH A TRIPLE POINT 1 297 Ex. 1. at the points with parameters and meets the curve again + (l/2£)z — t. p>0. are collinear. 0. Q. the tangent at to that conic of closest contact at C which goes through B. = R Ex. 1.g. [Replacing z by z — lx — my we reduce the equation to + x2 2 ) = y'(px + qy). [Put 2tpix V. 4. Q. — t* — t). 0. q<0. Express the coordinates of any point of the curve rationally in terms of a parameter. They lie on a is real and two unreal. but no conic meets it eight times there.] infinite An number B AB B Ex. One of the inflexions B B = . 4. through The inflexional tangents meet the curve again at points lying on such a conic. E. 5. e. . the tangents at meet the quartic again in six points lying by threes on two P. Ex. Any point on the quartic If the points with parameters t t is . Ex. 5. the point of contact of the tangent from C. The bitangents coincide with y = 0. Derive properties of the quartic from those of the conic by quadratic transformation. CA the tangent at C. h + h + h + h = 0. Choose a new triangle of reference such that C is the singularity. and C with that conic which passes conic having double contact at and has closest contact at C with the quartic. 2 . [t h i<o. q>0. = (l?)qly.
1) is a node. . 0.CHAPTEK XVIII QUARTICS OF DEFICIENCY ONE OR TWO § 1. where u2 . Nodal Quartics. Also. Fig. and take the line IJ as z Since (0. call tangents at meets the curve at another point. In this chapter we shall consider quartics with deficiency one or two. the equation of the quartic is of the form We = u4 + 2us z + u^z 2 = . 1).u3 ui are homogeneous of degree 2. and u 2 = is the equation of the tangents at the node. 1. we take this node as (0. 4 in x and y. 3. A quartic with deficiency/two has a single node or cusp. 0. If Each of the it has a node 0. these points / and J. and w 2 = since z = intersect at two points on the curve. . 0. i00{2v + i ij 1 lGxltj)(x'i + 3yi Sxl2y) + (ix + y)(2x + 3y) « = 0.
B'. OD Show that 0. 1. The six points of contact of the tangents to the quartic from the node0 lie on a conic touching OH and OK at and K. A D meet the quartic again in [If the line is z = %. Ex. the quartic at H. and OA. C. B. C". 4. 1 and H = being the equation of the lines joining to the points 2). and the required conic Ex. Putting z = 1. B. u2 v2 + zu3 z + u2 z .] Ex. OB. are collinear. Eight foci of the projected 4ic lie on j. [Project H and K into the circular points. OB. and OA. For the polar cubic of is u. lie on a conic. 2. and that the lines HABC. D'. are the points of contact withj of the common tangents of j and the polar Ex. C. are collinear. C. + u 2 z = 0. the lines OA. the points of contact of tangents from Ex. which meets the 6 quartic six times at and at the six intersections of the quartic with z 2 = v2 as is seen by writing (i) in the form . we see that the curve can be projected so as to have as a centre of symmetry. OD are « 2 vi + '2v u 3 + 1« 2 1>! 2 l = 0. is »2 = v i z ] meets the quartic again in A. Ex.XVIII 1 NODAL QUARTICS 299 of the quartic u2 v% is a factor of be written % . Show can be put in the The equation of a quartic with a biflecnode form xyz 2 + u = Q or (xl + y 2 )z ! + u = 0. Ex. 5. H K its own cubic of 0. and the 4ic is the envelope of a circle cuttings orthogonally whose centre lies on the polar reciprocal with respect to j of the . line meets the quartic in A. K. the envelope of a circle cutting any given circle orthogonally whose centre lies on a tricuspidal 4ic with an infinite bitangent is a nodal 4ic. line A through H If and are unreal. B'. 3. . 3. OC. Hence the equation may . KA'B'C meet on the conic through the points of contact of the tangents from 0. Combine the equations of the curve and its Hessian. XI. OC.polar cubic of 0. C. The foci of the 4ic on . H. § 11. OC meet the quartic again in A'. Prom this fact other properties of a biflecnodal quartic may be written down. C. H and K at which IJ meets the quartic again (Figs. B'. or as x^ + y* if the node is an acnode. A'. B'. D' A'. If lie 6. A'. where u is homogeneous of degree 4 in x and y. (i) on two lines through H. (iii) The sixteen inflexions other that is divided harmonically by the quartic and (iv) Any line through a fixed line.] [(iii) (iv) . (ii) The points of contact of the tangents from lie on another quartic. the quartic can be projected so as to be inverse with respect to a circle j whose centre is the node 0.. IJ touches that : See also Ch. Conversely. OB.. Show that K.. as 2 xy or x l —y i if the node is a crunode. u 2 (v2 ~ z2 + % z (us +u2 z) = We may without loss of generality take u ) 0. „ „ 2 = _ (i).
0. p. where u and v are homogeneous of degree 2 in a: and y.2 s2 + 2 ». Light emanating from a fixed point is reflected at the wires to another fixed point. ui + 2u3 z + u 2 z i = where in this case 0. H.0 to the other. 8. cubic * is U = zu + u3 = 2 0. J H We . meets the quartic in A. we may take in § 1 the equation of thequartic is O as (0. C. 10. it is the envelope of a circle cutting j orthogonally whose centre lies on the polar reciprocal with respect to j of the first polar curve of O. if the line is z = w. D' lie on a conic touching the noninflexional branch at 0. D'. This polar reciprocal passes through 4 {n — 2) foci of the «ic lying on j. Show that The equation of a quartio with a flecnode can be put in the form xyz 2 + 2yuz + xi v = 0. 1).. Find the locus of the point of Ex. 526.. . 2 ««i «Vi «»» »r + «» «V« See Bateman. shall leave the verification of these facts as an exercise to the reader. 48. OB. OC. 4 can be generalized as follows If an nio with an (n — 2)ple point O is selfinverse with respect to a circle j with centre O. 9. Archiv der Math. • = are the curves. Cuspidal Quartic s. we can find an »*ic with an (>2)ple point at such that any line through Ois divided harmonically by the curves r being any given number ~^\n. Most of the results of § 1 hold with slight modifications. (iii) If a line meet the quartic again [(ii) xz + u = 0. Richmond. Quarterly Journal Math. A. [If m„_ 2 z* + 2 uB_i z + un = 0. (ii) The tangents from to the curve lie on a conic touching the noninflexional branch at 0. (i) NODAL QUARTICS XVIII 1 7. : large number of polished wires in the form of concentric on a table. Show that Ex.] = Ex. xxvi (1893). xiii (1908). B'. Each curve passes through the points of contact of the tangents from . B'. (iii) x {z + w) + 2 u = 0. The points /. A circles lie reflexion. [A nodal circular quartic] § 2. and give here only the method of finding the bitangents of a cuspidal quartic* Tfle polar cubic of the cusp O has also a cusp at O and the cuspidal tangents of the cubic and quartic coincide for this" is u2 a perfect square. OD in A'. and the tangent at this intersection meets the quartic again in and K.. D and OA. C. B. Show that the theorem of § 1 can be generalized as follows Given an «ic with an (n — 2)ple point 0. coincide at the intersection of the quartic with the cuspidal tangent.\ Ex. then A'._! z + ». j'. und Physik.. C. pp.] .: : 300 Ex. As If a quartic has a single cusp 0.
. + ax) (y + hx) (y + ex) (y + dx) 0. . 8. c. f . x2 + 4 (2a p3 x pt y) y s 2 2 { (4^6 p*) ® + 3 ( 3 ^ 5 PiPz) x v + (Apt p?) y } 2 2 y) . . 2 (y d. £ can be divided into two sets of three in . y + bx. . denote the points of contact 2. Hence 2a — (bc + ca + ab + ef+fd + de)y — (abc+def)x = (i) and the nine similar equations are the ten bitangents of the . six quantities a. A p2 p3 p4 p6 p 6 are the sum a. the cusp being on Fig. 7. zx 2 + y 3 by U. U = 2 <xPy8e£ where (a/?y . /?.2 XVIII The quartic is CUSPIDAL QUARTICS 301 U = u — u2 u±. f two. £ and Now 2U = is (xPy + 8e£ equivalent to x 2 {2z — (be + ca + ab + ef+fd + de)y— (ahe + def) x} and meets the quartic — 0. u2 '= x2 ua = y s and the (zx 2 + y 3 ) 2 = y 6 — x 2 u i . 5 .. They are the points with abscissae 0. b. . four. 2 2 3 Now by a proper choice of the triangle of reference reduce the equation of the polar cubic to we may We 6 have then Suppose that zx 2 + y 3 — 0.co ). quartic* Writing the quartic in the form (2 z p x p 3 = where . e. c. tangent x The ten ways. .. quartic becomes y x Ui = where a. (y + ex) (y +fx) . I. . — * of the properties of the cuspidal quartic may be In it a. . and the cuspidal 1. y. e. e. we see that of § 1 are its intersections with and that the points H. a. six at the tangent at /(= J) is 2zp3 xp2 y = 0. of the products of a time. d. — = = 0. /3. .f are subject only to the relation a + b + c + d + e+f= We shall denote y+ ax.. e. by /?. j8 (0. three. b. K Some verified of oc = — £.8e£) 2 = 0. 2 2 2 (*Pb 2% ) * + 3 ( 3 F5 P2P3) XV + ( 4 ^4 Pi) y = 0. five. S.
302 CUSPIDAL QUARTICS XVIII 2 .
:
;
; ;
XVIII 2
The bitangent
symbol
(abc, def).
CUSPIDAL QUARTICS
(i)
303
may be conveniently denoted by the In Fig. 2 the equations of the bitangents
; ;
are as follows
(abc, def)
= 1 22 (abd, cef) x  lOy = 23 (abe, def) x5y=12; (abf, dec) 991 x  4,50y = 859 (adc, bef) 3x + 15 y =  34 (aec, bdf) Sx + 30y =  61 (afc, bde) 24 lec 225 y = 92 (ade, bcf)x45y = 68 (oc?/, bee) 65a + 50?/ = 51 (aef bed) 4\x + 25y = 38.
x + 45 y
;
;
;
The bitangents (abc, def) and (abd, cef) have unreal points of contact.
Ex.
1.
Of th^ ten bitangents of a cuspidal
b, c,
quartic, ten, four, or
If four or
two
are real,
[If a,
d,
them are
real,
/are real, ten bitangents are real. four bitangents are real. If two of
e,
them
none of are real, two
bitangents are real.]
Ex. 2. No two of a, a second double point.
[If
b, c, d, e,
f
can be equal unless the quartic has
a
=
6,
(1,
—a, a 3 )
is
a node.]
Through the points of contact of any bitangent can be drawn two conies touching the quartic at the cusp and each passing through
Ex.
3.
the points of contact of three tangents from the cusp.
\xz
= abc x* + (be + ca + ab) xy + (a + b + c) y
+ fl
l
,
&c]
Ex. 4. Show that the points of contact of e = 0, f = and the bitangents (abe, cdf) and (abf, cde) all lie on a conic through the cusp.
[0(«
= (O/3 + y8K.]
(abf, cde)
Ex. 5. The point of contact of a = 0, the intersection of /3 = with the bitangent (acd, bef), and the intersections of the bitangents
(abe, cdf)
and
are collinear.
/3
[They
Ex.
6.
lie
on 2
U+ a
2
=
The reader may
refer to
0ty8 + ape + a/3f, which is a straight line. Richmond, loc. cit., for further examples.]
What
modifications must be
made
in §
1,
Ex.
4,
in the case of
a cuspidal cubic ?
[Six foci lie oiij. The centre locus is a cuspidal cubic having threepoint contact with the line at infinity.]
Ex. 7. A quartic with deficiency two is referred to a triangle both inscribed and circumscribed to the curve, the double point being taken as Show that its equation takes the form (1, 1, 1).
ayz.y? + bzx » a
+ cxy w 1
=
0,
the tangents at the double point being an? + 6»2 + civ 2
=
;
where
(l
u
= (l + A)xyAz,
Show
also that
= Bx + (l+B)yz, iv=xCy + + C)z. and that the quartic has a cusp if au+fiv+.yiv =
v
0,
bc(X'
i
+ caP i + aby*=
is
0;
where
a,
/3,
y denote
a
(y
BC+B+1, CA + C+1, AB + A + 1.
(z
Show
that the polar cubic of the node
+z) u 2 + 6
+ x) » 2 + c (x + y) w
1
=
0.
—
304
BICIRCULAR QUARTICS
§ 3.
XVIII 3
Bicircular Quartics.
If a real quartic has a pair of unreal nodes, they may be projected into the circular points at infinity. The quartic then
becomes
c (x
2
bicircular,
)
and
its
equation
)
is
of the form
+ y2 2 + 2
(Ix
+ my) (x2 + y 2 + ax2 + 2 hxy + by 2 + 2 gx + 2fy =
0,
if
the origin is taken on the curve. If we now invert with respect to a circle with centre the origin and unit radius, the quartic becomes the circular cubic
c
+ 2 (Ix + my) + ax2 + 2 hxy + by 2 + 2 (gx +fy) (x2 + y 2 =
)
0.
Since foci invert into foci and a circle and two inverse points into a circle and two inverse points, the, properties of the foci of a circular cubic proved in Ch. XIV, §§ 2, 3, hold for
a bicircular quartic, namely
:
bicircular quartic is selfinverse with respect to each of four mutually orthogonal circles each of which passes through four foci. If the four real foci are ccmcyclic, the quartic consists of two ovals and three of the four circles are real. If the four real foci are not concyclic, the quartic consists of a single oval and two of the circles are real, while each passes through two real foci.
A
The reader
will at once verify that, if the four lines
2 2 y +(x±<x)
=
each meet a bicircular quartic at only one finite point, the coefficients of x 3 , x 2 y, xy 2 y 3 xy in the equation of the quartic are all zero. Hence the equation of a bicircular quartic becomes
,
,
(x 2
+ y 2) 2 + ax2 + by2 + 2gx + 2fy+c =
.
.
.
(i)
the line joining a pair of singular foci is taken as y 0, and the middle point of the line is taken as origin. The four singular foci of this quartic are readily shown to be
when
=
(+§/(&«),
0),
(0,
±iV{ab)).
We
shall suppose
b>a
in the following.
The quartic
(x 2
(i) may be written + y2 t) 2 + (a + 2t)x2 + (b + 2t)y2 + 2gx + 2fy + ct2 =
0.
Hence the conic
(a
+ 2t) x 2 + (b + 2t)y 2 + 2gx + 2fy + ct 2 =
.
.
(ii)
has contact with the quartic at four points lying on a circle
with centre 0, whatever
!
x2 + y 2 = t may be the value of t. If 4 i i + 2(a + b)t 3 + (ab4:c)t 2 2(ac + bcf 2 g 2 )t
(iii)
(abcaf 2 ~bg 2 )
=
.
.
(iv),
3
XVIII
BICIRCULAK QUAKTICS
305
the conic (ii) is a linepair. It is therefore a pair of bitangents in this case. may exhibit the bicircular quartic (i) as an envelope in another manner. In fact it is the envelope of a circle whose centre lies on the conic
We
4.xy(a
+ 2t)+4>y 2/(b + 2t) + l
=
=
.
.
.
(v)
and which
where
cuts orthogonally the circle
.
t
x2 + f + 2gx/(a + 2t) + 2fy/{b + 2t) + 1 is any root of equation (iv).
.
(
vi),
For, if
we
take the centre of the variable circle at
(^a2t)*eosfc J (62*)* sin $)
its
.
.
(vii),
equation
is
(a?
+ y *t) = cos(/> {(a2f)ix(a2t)*g] + sin<j>{(b2t)iy(b2t)if}
.
.
(viii),
(i),
whose envelope
is
found in the usual manner to be
on
making use of (iv). The variable circle has double contact with its envelope (i).* The foci of (v) are the singular foci of (i). The curve (i) is selfinverse with respect to the circle (vi),
since the variable circle is selfinverse with respect to (vi), being orthogonal to it. Hence the four circles with respect to which (i) is selfinverse are obtained by putting into equation (vi) any value of t derived from (iv). The ordinary foci of the quartic (i) ai e the intersections of (v) and (vi), each value of t giving four such foci. is any intersection, the circular lines through For, if

P
P
form a degenerate
circle orthogonal to (vi)
and touching the
'
envelope of the circle
The conies
or
'
'
(v),
(viii), which is (i). where iisa root of (iv), are called the
focal'
deferent conies of the bicircular quartic. The reader will notice that (ii) and (vi) have the same centre. It follows that a pair of bitangents passes through the centre of each circle for which the quartic is selfinverse, as is geometrically obvious. In Fig. 3 is shown the bicircular quartic, whose real foci (0, 1), (M, ft), (¥. I), (hi) lie on the focal conic The centres of the circles for which the quartic 9. «2 + 92/ 2 42 3 is selfinverse are (f, 3), ( T , A), (M, t o), (If, *!). being the harmonic points of the quadrangle formed by the real foci
=
* As is the case in general with any singly infinite family of the similar result for circular cubics see Ch. XIV, § 3.
951R
^*
circles.
For
306
BICIKCULAR QUARTICS
XVIII 3
X
E
:
XVIII 3
BICIECULAR QUARTICS
307
and the centre of the circle through the foci. The quartic is drawn by the method described in Ch. XI, § 11, Ex. 3. The real foci are denoted by O, the singular foci by x and
,
the centres
by
•
in the figure.
Pig. 4.
(x 2
+ y*) 2 22x*2y 2 i
=
(
0.
Singular foei
(
+
\/5,
0); ordinary foci
+
1, 0),
(0,
+2).
Ex. 1. The inverse of a bicircular quartic with respect to a point not on the curve is a bicircular quartic the inverse with respect to a point on the curve is a circular cubic.
;
Ex. 2. Through any point O of a bicircular quartic three real circles of curvature pass besides the circle of curvature at 0, and the three points of osculation lie on a circle through O.
[Invert with respect to O. may derive other theorems' by inverting properties of a circular cubic, e.g. Ch. XIV, § 3, Ex. 10, 11.]
We
Ex. 3. The circles of curvature of a bicircular quartic at its four intersections with any circle for which it is self inverse have four point contact. Ex. 4. If O is the point halfway between the real singular foci of a bicircular quartic, show that
(i) O is halfway between the unreal singular foci, and the sum of the squares of the distances of O from the singular foci is zero. (ii) The points of contact of the tangents from O lie on a conic.
(iii)
O
C,
is
A, B,
D are
The
equidistant from the middle points of and CD, where the intersections of any line with the quartic.
(i)
AB
[Use equation
of §
3,
and write down the
first
polar of O.]
bisectors of the angle between any pair of 'bitangents of a bicircular quartic are parallel and perpendicular to the line joining the real singular foci.
Ex.
5.
[Equation
Ex.
6.
(ii)
of § 3 has no term in xyi\
intersections of two pairs of bitangents lie on a circle halfway between the real singular foci. The sum of the squares of the radii of the circles obtained by taking two pairs of bitangents and also the other two pairs is the same for each way of dividing into pairs. X 2
The four
is
whose centre O
1
308
Ex.
lie
7.
BICIRCULAR QUARTICS
The eight points of contact of any two of the conies
XVIII
(ii)
3
of § 3
on a conic.
The
eight points of contact of any two pairs of bitangents lie on a conic. [The points of contact of the conies for which i = t t and t = t 2 lie on 2 2 (a + 2t1 )x + (b + 2t 1 )f + 2gx + 2fy + ct 1 " + (t 1 t 1 )lx' + ^t1 )^0,
i.
e.
ax 1 + bf + 2gx + 2fy + c +
8.
(t1
+
2)
{x 1
+ y2 )  t
x
*,
=
0.]
Ex.
The four
intersections of
0.
any two
of the conies
(ii)
of § 3
lie
on
a circle with centre The rectangular
asymptotes. The conic through
hyperbola
through
the
intersections
has
fixed
and the
intersections has a fixed tangent at 0.
6.]
[Note the case in which the conies are pairs of bitangents as in Ex.
Find the centres of the circles for which the quartic (i) of § 3 and show that they lie on a rectangular hyperbola is sellInverse through whose asymptotes are parallel and perpendicular to the line
Ex.
9.
;
joining the real singular
foci.
(iv).
f/(b42t)), where t is given by The centre of (ii) lies on b — a = g/x—f/y]
[(g/(a + 2t),
Ex. 10. The centroid of the four finite intersections of a bicircular is halfway between the real quartic with any circle whose centre singular foci of the quartic is a fixed point V. It is also the centroid of any four concyclic foci, and the centroid of the four points of contact of any pair of bitangents. It is the intersection of the axes of the two parabolas of the family (ii) and the centre of the rectangular hyperbola of Ex. 9.
;
[Eliminate x or y between (i) and](iii) or between the point (g/(ba), f/(ab)).]
of
(v)
and
(vi).
V is
Ex. 11. In Ex. 10 the ratio of the distances from x = (ii) or (vi) and of Fis the square of the eccentricity of
of the centre
(v).
circle for
[This enables us to determine the focal conic corresponding to any which the quartic is selfinverse ]
Ex. 12. The centres of three circles for which a bicircular quartic is selfinverse are the vertices of the common selfconjugate triangle of the fourth circle and its focal conic.
[They are the harmonic points of the quadrangle whose vertices are the four foci on the fourth circle.]
Ex. 13. If a bicircular quartic consists of two ovals one inside the other, the focal conic through the real foci is a hyperbola and the other three focal conies are ellipses. If the qaartic consists of two ovals external to each other, the focal conic through the real foci is an ellipse and, the other three focal conies
are hyperbolas. If the quartic consists of one oval, one focal conic a hyperbola, and two are unreal.
is
an
ellipse,
one
is
[Use Ex. 11 and'Ch. XI, § 11, Ex. 3, 4]
Ex. 14. The directrices corresponding to four concyclic foci pass through the centre of the circle and form a pencil of the same crossratio as that subtended by the foci at 0. [The line joining to the point (vii) forms a pencil homographic with that traced out by the chord of contact of (viii) with its envelope.]
XVIII 3
Ex.
15.
BICIRCULAR QUARTICS
The normals
to a bicircular quartic at the points
309
where
it
meets a
circle for which it is selfinverse singular foci of the quartic.
touch a conic whose
foci are the
[The corresponding focal conic. It follows that these normals form a quadrangle whose opposite sides subtend supplementary angles at either
singular focus,
&c, &c]
Ex.
circles
16.
the asymptotes of the corresponding focal conic. [See Fig. 3 and compare Ex. 13.]
The pair of bitangents through the centre of any one of 'the for which a bicircular quartic is selfinverse are perpendicular to
Ex. 17. The distances of any point on a twocircuited bicircular quartic from three real foci A, B, C are connected by a linear relation C.PA + + n.PC=0, where I, m, n are constants. [Invert Ch. XIV, § 3, Ex. 20, with respect to any point, and we find that the result of Ch. XIV, § 3, Ex. 23, holds for the bicircular quartic. As in Ch. XIV, § 3, Ex. 26, we obtain results for the onecircuited
P
m.PB
quartic.
As an
Ex.
alternative,
we may
invert the result of § 6
(iv)
or § 10, Ex.
3.J
If a twocircuited bicircular quartic with real foci A, B, C passes through real points P, Q, R, then a twocircuited bicircular quartic with foci P, Q, passes through A, B, C.
18.
R
[By Ex. 17 we have
interchange of P, Q,
PA PB PC QA QB QC =
;
which
is
unaltered by
RA RB RC
R and
A, B, C]
»
Ex. 19. Any bicircular quartic can be inverted into a bicircular quartic symmetrical about two perpendicular lines on which lie the real foci
and singular
foci.
[Invert with respect to the intersection of two real circles for which the quartic is selfinverse and on which lie the real foci. See Fig. 4.]
Ex. 20. Obtain the equations of the bicircular quartic symmetrical about the axes of reference whose real singular and ordinary foci are given and are on these axes. [There are three types. We may take the real singular foci as Then the curve is ( + m, 0).
{x 1
+ %/f  2
(I
+ m1 ) x  2 (I  m?)
1
where
(1)
f + fax
2
2
3
=
0,
The real foci are (±<X,0), (±/3,0), P = (m 2 a 2 )(m 2 lb 1), k = + 1. 2 2 2 2 (»i + /3 ),' h = + 1. (2) The real foci are (0, ±OC), (0, +0), P = (w + a 2 2 2 2 = l. (3) The real foci are (±a, 0), (0, ±/3), I = (m a (w + 8 ), As an exercise the reader may discuss the relative position of singular and ordinary foci, and the nature of the ovals and bitangents for each type. He may also find the relation connecting the distances of any
) )
jfc
put
point of the curve from three foci (Ex. 17). 2 = a 2 cosh 2 f— j3 a sinh 2 e in type (1).]
For
this
purpose he
may
m
Ex. 21. Find the locus of the singular foci of a twocircuited bicircular quartic, given the real ordinary foci.
[The circular cubics with the given
foci.
See Ch. XIV,
§ 3,
Ex. 27.]
310
BICIRCULAR QUARTICS
XVIII
3
Ex. 22. Show that the equation of a twocircuited bicircular quartic can be put in the form
(x 2
+y
2 2 )
2{lx + my)(x* + y' l) + ax i + 2lmxy + bif +
l
l
=
0.
as the unreal circle with respect to which the quartic is selfinverse, and the axes of reference parallel and perpendicular to the line joining the real singular foci.]
[Take x^ + y' +
l
=
Ex. 23. Obtain the equations of the conies touching the quartic of Ex. 22 in four points, the equations of the bitangents, and the equations of the circles with respect to which the quartic is selfinverse.
[The quartic
v
is
«<
2
+v
2
= 0,
u
where
1
= (aP2t)x
e
i + y — Ix — my + 1, + (bm 2 2t)yi + 2{t+l)(lx + my) + (lf).
=x
t
The
conies are
=
;
and the bitangents are v
= 0,
where
= l
or
t1
,
t3
,
t„
which are the roots of the equation
it3 2(a + b + 2)t'i + (ab + 2a + 2b—iPim?Pm?)t
The required
t
circles
,
= — I,
therefore
S1 =
+ 1 = 0, Sx = 0, St = 0, S3 = 0, where i &c] x° + y 2(t + l)lx/(2t1 + Pa)2(t 1 + l)my/(2t 1 + m?b)l,
x> + if
1
tx
,
t2
ts
,
+ (2 am + 2WabBP ni?) =*0. have their centres at the centre of v = where and are mutually orthogonal. Their equations are
Ex. 24. Show that the squares of the tangents from any point of a twecircuited bicircular quartic to the three real circles with respect to which the quartic is selfinverse are connected by a homogeneous linear
relation.
Conversely, if the squares of the tangents from P to three given circles are connected by a homogeneous linear relation, the locus of P is a bicircular quartic which is selfinverse with respect to each of the given circles, if these circles are mutually orthogonal.
;
,
[The quartic of Ex. 22
{t l
is
i 1
ts ){2t 1 + Pa)(2t + mi b)S i +
...
+
...
=
O.'J
Show how to find the foci of aS,2 + 6S, a + cS3 2 = St = 0, S2 = 0, Ss =0 are given mutually orthogonal circles. [The quartic is the envelope of K^i + pS^ + vSs = where
Ex. 25.
0,
where
»
Now make
the radius of this circle zero.]
Ex. 26. Given the circles with respect to which a bicircular quartic is selfinve\'se and one point on the curve, find the locus of the singular foci.
Ex. 27. Find the locus of P,
(i) (ii)
if
PA. PB x PC. PD, PA PB PC,
.
<r.
where A, B,
[(i)
C,
D
are fixed points.
can be inscribed in
bicircular quartic such that an infinite number of quadrilaterals it whose sides pass alternately through the two circular points.
A
(ii) bicircular quartic with and as singular foci. inverse in respect to A, C or any point ?]
A
A
B
What
is its
Under what circumstances can a given bicircular quartic be generated by one of the methods of Ex. . K K . B are not ordinary nodes. 28. . a slight modification If A and Such modification will of the above notation is necessary. St = 0. S2 = t. In our discussions of quartics with two real double points.] Ex. find the intersections of <St = S. [A bicircular quartic with singular foci at the centres of the two _ circles. 29. L X L2 ) are harmonic. double points are A and B. Trace the quartic [It'is as the intersection of circles. . where SL — 0. . S = Now easy to put the equation in the form are circles. 2729 ? For instance. [The reader may consider the question. .' XVIII 4 QUARTICS two fixed circles (i) WITH TWO REAL NODES 311 Ex. 31. if varies as the distance of varies as the product of the tangents from to circles (i) ' two fixed the product of the tangents from P to P from a fixed line. Quartics with Two Real Nodes. if we allow this circle to be unreal. The locus of the intersection of two orthogonal circles one belonging to a given coaxial family and the other belonging to another given coaxial family is a bicircular quartic through the common points and the limiting points of the two families. 5). any bicircular quartic can be generated by a point P the product of whose distances from the singular foci varies as the tangent from P to a fixed circle.] Ex. EE . 2) X (BV. Find the locus of P. MM and B2 and intersect at C (Fig. (ii) varies as the square root of the from a fixed line. for different values of the parameter § 4. The real shall adopt throughout the following notation. E F E . allow these alternatives. at A meets the curve again in 1 The harmonic conjugate of BA with respect to the and 2 The lines tangents at B meets the curve again in Zj and L 2 BL X L 2 meet at C. . .] Sl S2 = t S. is distance of P if the product of the tangents from to constant. The other diagonals of the quadrilateral formed by the tangents at A and B meet AB at R x we M H M H t . The tangents at A meet the curve t 2 again at x and 2 the tangents at B meet the curve again The line 1 2 meets the curve again at and 2 at x t The and the line F1 F2 meets it again at and x 2 2 tangents from A touch at A 1} A 2 A 3 A t and the tangents and from B touch at Bv B2 B3 B^ The ranges {ATI. AM M F . usually be quite obvious without explanation. if we refuse to ' . Find the locus of P. (ii) P two other fixed circles. or allow P to be inside the circle if real but not always. [Invert with respect to a common (i) point. (iii) varies as the taDgent from to P P another fixed circle. The harmonic conjugate of AB with respect to the tangents. 30.] Ex.
. . 5. and the other will be denoted by F. whose points of contact are A v A 2 A 3 Similarly in other cases. if B is a flecnode. Replacing x and y by suitable multiples of x and y. of contact are lt 3 2 and % coincide at is a cusp. Taking ABC as triangle of reference. the coefficients of 4 4 a. . we may simplify the equation (i). one of the points F^ and F2 The points will be at B. now the intersection of the curve The points 1 and 2 are the other intersections called F. (i). x6 y 3 x2 yz. the points Or again.312 QUARTICS WITH TWO REAL NODES XVIII 4 For instance. and 2 do not exist in this case. + qf) + 2m xyz 2 + 2 (fx + gy + hz) z 3 = and x2 +pz2 = where y 2 + qz2 = are the tangents at A and B. we may make/ and g (x2 +pz 2) 2 (y . B B B B F F K K Pig. xy 2 z are zero in the equation of the y quartic. which takes the form . For instance. . There are only three 1 tangents from B (other than the tangents at B) their points K K . . . tangents from A. if x with the cuspidal tangent. There are only three of the curve with the tangent at F. . .
&c. * We cannot. . A lt A 2 .pqt 2 ). 2 . however. d. 2 1 at A and B respectively. 2 and 3 pass through the same four which also lie on the conic (v). /org zero. b. A. g general at present. (iv). . The points of contact of the tangents from C lie on a conic. 3. B. qx2 +py 2 + 2txy + 2fxz + 2gyz+(pq + 2h — [m—t] 2 )z 2 touches the quartic at other than its — .XVIII 4 QUARTICS both unity. As . The three conies of Ex. c. 1. respect to the conic section of the pair of bitangents. is The conic (ii) of bitangents. 3. (ii) four intersections with z2 xy + (m — t) = (iii) A and B. . Write down the first polar of C] Ex.] 2 . Writing equation (i) in the form / and {xy + (m—t) z 2 } 2 + z 2 {qx2 +py 2 + 2txy + 2fxz + 2gyz + (pq + 2h[mt] 2 )z} = we see that the conic 0.fiqg. i? 15 = R (v). E F = 0. 2 . and so do The two [On and z {py' 3 1 conies touch the conic of Ex. points. A t Llt . (ii). first l Write down the polar of A and use l + (pq + 2h)z + mxy+fxz + 2gyz}+x {xy' + qxzi + myzi +fzz \ = (x* +pz2) {y 1 + qz*) + 2 mxyz* + 2(fx + gy + hz)z'i . is of The pole AB with or the inter (gtpf. qx* + (pq + 2 h) z* + mxy + 2fxz +gtjz= 0. [On qx*+py'i + 2(pq + 2h)z l + 2mxy% 3(fx + gy)z=0. We shall take the values of t given by (iv) as a.. The points E lt F. c. B lie on a conic. L^ lie on a conic.* to leave WITH TWO REAL NODES it is 313 better But in view of the following sections. Blt B B3 Bit .2pq . make / 2 : f= q : p . The locus of this point when t varies is qx2 py 2 +fxzgyz which passes through the points Ex. 2. py +(pq + 2h) z* + mxy+fxz + 2gyz = 0. 2 The points A. if t* a linepair. M Mv t .2h) t 2 + 2 (fg + mpq) t + (p2 q2 + 2hpq—pqm 2 —pf 2 — qg 2) = .] Ex. [Add and subtract the equations' of the conies in Ex. and therefore represents a pair 2mt3 + (m 2 . z [On mxy + (fx + gy + hz) Ex.] 4. .
x 3 BC. . t [Put A conic touches the — m in (ii). ST harmonically. [See Ex>7. X= . where t = a. 8. AU.hm) xy + 2 (qmg + hf) xz 2 2 2 (pf +pqm + 2hm . .] . 314 Ex.] quartic at Llt Lit M 1) M%. [In Ex. Q and the intersections of any one of conies (ii) with any one of the conies (iii). the line El Ei being Hence the required conic is = 0. 6.) {2 (xy +[mt] z>) + (2tt 1 harmonically. X X . B^Ri are double points. 9. conies (ii) at 2 = Q Ex. A conic can be drawn through the points of contact of any two of the conies with the quartic. put t = t x *= f and The conic through C and the intersections of any two of the has a fixed tangent at C and divides AU. 8. a are any two points dividing 1 1 conic can be drawn through P. l z }.] Ex. Ex.] lie on Ex.to 2 +pg. BV [See Ex. . of which The conies (ii) meet the sides of the triangle ^. 15. The points of contact of any pair of bitangents lie on a conic touching CA at A and CB at B.] t' t = t' and t = t". Through the points P. Q of Ex.] Ex.2fffm ) z = °2 If we write + qm 2 ) x + {fg. 11. The conic B1 B2 B. Ex. [See Ex.0. Ex. [On conic (iii). 6.VIII 4 lie on a conic which The points A u A 2 A s A t B.hm) y + (qmg + */) z} 2 + (f + qm?) (pf +pqm? + 2 hm? . 10. [If 2t 2 = + t" 12. The conies through C and the intersections of any two pairs of bitangents touch one another at C and divide AU. 13. The intersections of the four pairs of bitangents with AB form an involution whose double points are jB t and iJ 2 Ex. it a [On (f + qm 2 ) x 2 + 2 (fg . 14.] AH AH . Bt A K^ K2 is 1 2 1 (g +pm )y + 2(fghm)xy + 2(pmf+hg)yz . touches QUARTICS WITH TWO REAL NODES X. Q RR P [See Ex. 2 = »„ by § 1. (ii) the conies are given by in Ex. = 0. the quartic (/ becomes (y 2 + qz i )X'i + 2th X+(y i + qz2 )v2 . c. It touches one of the the and Q. 16. conies .hm) y + (qmg + hf) z. The four intersections of the two pairs of bitangents a conic touching CA at A and CB at B.2fgm) z* i\ = { (fg and «s is homogeneous of degree 3 in y and z.2fgm) z = 0. and the lines AHX AH% being v. Through the intersections of any two of the conies (ii) passes a conic touching CA and CB at A and B. or d. 9 a conic can be drawn passing through the four points of contact of any one of the conies (ii) with the quartic. 2 {qx 1 +py 2 + 2 tY xy + 2fxz + 2gyz + (pq + 2h[mt 1 f) z } [ {qx +py i + 2t 1 xy + 2fxz + 2gyz+(pq + 2h\mtiY)zi } Ex. 12. 8.] Ex. harmonically.] (h *. 7.] Ex.BCin involutions BV. 8 put *! = t. If P. 5. where . H H .m + 2hm . 8.
any root of (iv). and these lie on the polelocus. The quartic 2 the envelope of the conic where t is — p uz + 2p \xy+(m t)z } +vz = 0.——. It passes A lt A tK A . 21.— — = Vq tan 0t (x tan Vqp * an <* j= gz) + ^p cot y \ a . . a cubic. . c. The sixteen intersections of tangents to the quartic from and B lie by fours oh four conies touching the tangents at A and B. . Now use Ch. = 0. 2 (g +pm f + (2fgh . [The conic is h (f 2 + qm 2 ) x 2 + h Ex. / . VP2 tan a . d. [Dividing (xy 2 + qxz 2 + CA and CB with the myz2 +fz?) (yx 2 +pyz2 2 —xy^x + mxz2 + gs?) +pz2 ) (y 2 + qz2 ) + 2 mxyz2 + 2 (fx +gy + hz)z3 \ by z 3 we obtain the cubic xy (fx + gy) = z {qmx2 + (m? 2h)xy +pmy + z2 { (qg + mf) x + (pf+ mg) y) +fgs\] .12 ) z = 0. is a conic +py2 2 txy + (pq . b. The points A. 22. The [qx 2 locus of the pole of z = touching the tangents at A and 2 B to with respect to this conic the quartic. Another proof consists in projecting A and B into the circular points and using the fact that the foci of a bicircular quartic lie by fours on circles.mh pqm 2 2 ) 3 ) xy + (gh pqm2 ) {fx+gy) * Show that the pencils of tangents from A and = o.] A Ex. 12 and the intersections of tangents at B and A to the conic of Ex 1. 23. at C divide AB harmonically. 17.] Ex. The conic CE1 E2 F1 F2 touches the conies The tangent to this conic and the tangent to conic (v) of Ex. The where Jcpq is lefthand side of (ii) may be written in the form uv + kz 2 the lefthand side of (iv) and = .] B have the same i crossratio. [Taking = a. and p is a parameter. § 11. vpg A^pq cosec a. vpq y (y y=fz). [The intersections are the centres of the linepair conies of the family of Ex.fz).XVIII 4 QUARTICS WITH TWO REAL NODES 315 lie Ex.A lt a B. u . 18. V lie on AB with' the tangent at lt 2 3 . I. 19.] on Ex. 11. is we get the eight bitangents. . 12 at C. gz) + *lp tan ^ & * . * (y + tan a „ —==. The points of contact of any two pairs of bitangents * a conic. 21. [The tangents are i 2 2 2 s 3 i py + 2gy z + (2h + 2pqm ) y z + 2 (gqmf)yz +(2hq+pq f ) z* qxl + 2fxs z + (2k + 2pqm2 ) x2 z 2 + 2 (fpmg)xzi + (2hp+p 2 qg2 ) z* = 0.] 2 Ex. 20. [See Ex.B Bt U. C to the conic at Ex. 7^— v—Vq cot \<x (x / i t = t _ I l + —=. through the intersection of B B .] Ex. a = .
. All quartics with two given nodes and passing through seven other fixed points. case One root of (iv) = — 1. (iv).] = hm.] jp we may take = g = — 1. In § . . and a corresponding . ma 2 — 3 a). [The quartic has either two crunodes or two acnodes. 26. as explained in § 4. Ex. = 6 we get the tangent z + bx and the corresponding bitangent qbx + 2b 2 y + (2bq)z = 0. If one polygon of 2n sides can be inscribed in a binodal quartic so that the sides go alternately through the nodes. g=f. t i 2mt 3 +(m 2 2h)t 2 + 2tq = (iv). (iii). Quartics with a Node and 4 A and B were any two real double points of a quartic. Discuss the quartic for which f*/g* = q/p. and (iv) become qx2 + 2taiy + 2xz + 2yz + (2h. we obtain a tangent from the cusp z + ax = touching at the point B 1 (— 1. . [The conies of Ex. pass through an eighth fixed point. b. Ex. * Taking t = a. [Project the nodes into the circular points and invert with respect to one of the fixed points ] Ex. —1. we have p = and we may also take /= 1. d of a linepair which must evidently be If t is (ii) represents (z + tx) {qtx + 2t 2 y+{2tq)z} =0. bitangent b x with equation qax + 2a2 y + (2a — q)z touching where Similarly taking t the cusp touching at = = from b2 . The quartic can be projected so as to be symmetrical. c. an infinite number of such polygons can be inscribed. 25. a Cusp. (i) = ez meets the curve again in coincident + (2hm2)e 2 + 4:(q~m) e + 4(2qhl) = 0. if e 3 c.[m . 2 B and so for t = The line 2y points.a . . xy + (m — t)z 2 = (iii). . Discuss the quartic for which fg 1 and 5 degenerate. In the is t first the corresponding pair of bitangents meeting at (1. 2a2 y 2 +(2aq)yz + qa(a — on) z 2 = 0. and by § 4 this touches the quartic where it meets (iii). 27. Ex. . § 5. g = 1. Similarly in the second case. If A is a node and B is a cusp. [Project the Steiner points of a polygon inscribed in a cubic (Ch. 24. 01.] XVI. any one of the roots a. The equation (i) of the quartic becomes x 2 (y2 + qz 2) + 2mxyz2 + 2(x + y + hz)z3 = (i) while equations (ii). 316 QUARTICS WITH TWO REAL NODES XVIII 4 Ex. § 6. 1) into the circular points and invert. The locus of the pole of AB with respect to the conies (ii) is x = y. t = d.1] 2) z 2 = (ii).
23. 12x50y = 7.2 + 36a. cd + ab.1 XVIII This 5 is QUARTICS WITH NODE AND CUSP readily proved to be an equation in e 317 whose roots are ad + bc. a 2 d 2 b 2 c 2 ). &c. =— b i arc and the corresponding bitangents b 1% b3 . \2x\0y = 9x + l0y = 9. = 5.=l. z. + 469 (0. is shown the quartic I00x2 y 2 + 2y (100 + 90a. 4a. IOOsV + 22/(100 + 90s) . i 1 + 4> 4 Pig.) . are =5. oo ) = 0. bd+ca.Hix1 + 36rr + 469 = In Fig.144 a. &c. 6. Hence the tangents from the node 2y The points of contact (2b + A are = (ad + bc) 2 . 6 0. a. 36x + 50y = 27. A A A3 lt will be found to be 2c2a2d. z .'3a: b. 2ad2bc. The tangents from the cusp 3a. .
. Ex. 1. 12. The reader may consult Messenger Math. are represented by their suffixes only. and 5y = + 6.] lie on a conic through B. Ex. Ex. 2 (yx 1 + s3 + mz*x) = z2 {z + ax) + (z + bx) (z + ex) (z + dx) z2 (z + ax) is a bi tangent touching where Similarly 2 (ya? + z a + mz*x) = (z + bx) (z + ex) (z + dx). 6 2 touches AC. 5.] ABE^E^F touches the quartic at F. B. 8. 6.] Ex. B. The diagonals oft the quadrilateral formed by the bitangents pass respectively through A lt A % A s . The conic through B.] B2 and the points of contact of 6 1( 6 2 i lie on a conic. 81. 4.. or none of the tangents from the cusp are real. [xy + (m — a)z 1 Ex. B t and the points of contact of the bitangent b x touches AC at A and .] Ex. Ex. [2X 2 + 2 . Ex. The crossratio of the pencil formed by the tangents from B is equal to the crossratio of the range formed by the intersections of AB with the bitangents. . two.B) l have the same crossratio. A. \. B. [See § 4. The conic 1. 3. = 0. 2?!. [qx" 7. or F) lie on a line through A. 2y = 3. + (Zh — m' + am + bm — ab)z' + 2yz+(a + b) xy + 2xz = 0. two. § 1. [See § 4. Ex. = z(z + ax) (z + dx) + z (z + bx) (z + ex) + dx) is a tangent from A touching where (z + ax) (z = (z + bx) {z + ex). [In Ch. The pencils B(B B B B 1 1 s i) and A(A Ai A. 63 .b1 &c. 5 The tangents from and at the node (oo 0) are respectively 101/= 13. p. F take ABF as the triangle ABC] • . 10. The conic touching AB. Ex. Four. and ^ = 2X F . 6 Bl . The points of contact of the two real conies through A.] The conic AH^AH. ornone of the bitangents are real according as four. 64 touches AB at B. Ex. 9.] A A A s Ht H2 t 2 touches the quartic at B and touches Ex. Ex. 11.318 QUARTICS WITH NODE AND CUSP XVIII . 2.BCat B. The points of contact of the tangents from C 2. [yx^+z^ + mz'x (ya? = z* is the polar cubic of B and the quartic is Hence + + nu?x) % = (z + ax) {z + bx) {z + ex) (z + dx). Ex. xlvi (1916). . 50y = S7. osculating the curve (not at A. Ex. In Fig. Apply the method of § 2 to obtain the bitangents of the quartic (1) of § 5 and the tangents from A. 5. [See § 4. IX.
Let us take Then y ±ix are asymptotes twice over. Also y = i(x — d) which gives p touches the curve (i). The locus of a point P whose distances from two fixed points B are connected by a linear relation is a Cartesian curve. but it consists of two ovals.: XVIII 6 CARTESIAN CURVES § 6. and three real ordinary foci. which is the perpendicular bisector of the line joining the real ordinary foci E. 0. (i). the intersection of the tangents at co and co'. (iii). Hence the quartic consists of two ovals (§ 3). 0). y now takes the form '.. 0). d has the roots ex. B. (y. Or 'Cartesian oval'. which is selfinverse for a real circle j through Q. The curve is called a Cartesian curve We now show that A. 0).. When we invert with respect to Q. 319 Cartesian Curves! If a quavtic has a cusp at each circular point co. B. These lie on the real circle. Suppose the relation is .PA±l. G.PB = n. Let us consider case (1). G on the axis of (2) on the axis of and one real ordinary focus G lie The singular focus x. (8.* from which it + y 2 ^yy0itxBf + 4cx^y{2xa^y) = ' . a>\ its inverse with respect to any point Q has Q as a focus (Ch. and therefore there are three tangents from co other than the tangent at co. as origin and A. /3. The quartic with co and co' as cusps is therefore symmetrical about the line which is the inverse ofj. Considerations of symmetry show that there are only two cases to consider. F. and a = b in (i). B. ±m. 1). the equation of the quartic takes the form (x 2 + y 2) 2 +p(x 2 + y 2)x + ax2 + by 2 + 2gx + c = . Taking the line as axis of x. if = = 4gd 3 + (4ca 2 )d 2 2agdg 2 This equation in readily follows that (x 2 (i) = .AB * . Ex. G as the points (a.. and the real ordinary foci A. (ii). (1) lie The singular focus x. V^ § 4) and is a bicircular quartic (§ 3. the real foci are Q and the inverses of A. The curve is of class 6. . Hence the curve has one distinct singular focus 0./.
F. &c. 0). 5.). B. when and A. G may . is . 1. from (2).6 320 CARTESIAN CURVES on XVIII Take a point and let AB such that . 7. + y*6iy + (0.PA±0A3. 0) . 0. Case is OABC at points whose distances We have + 3PA ± 2PB = 10. G as (oc.PB = 0Cs. (8. the distances of any point on a Cartesian curve from the foci A. 0). 0). (j8. (f . (iv). Conversely. 0). (y. B is The curve meets are 13.BA where focus. 0). 0).= Singular focus ordinary foci (2. (x* 7. not so interesting geometrically. = a + /3 + y. the singular focus/and G the third real ordinary A quartic with of (ii) is two Cusps has one bitangent. The real foci E. is Then the equation of the locus P. B are connected by the linear relation P ±0Bz. (ii). which when rationalized proves to be the same as Pig. . . G be a point 0A:0B = P:m 2 on A B such that V:rn?:<n? = 0A:0B:0C. in which the three real foci of (i) are not collinear. 0) is taken as origin ± p s{(xuf+y 1 2 }i±a.i{(xl3f + y2 }i=ys(pa. 0). UH4x29). The bitangent evidently 2x In Fig. A is (2. (f . 0). 7 is (0. 0). C is (8.
] Ex. 2216 Y .c 2 ) du + M (it2 . Py+yix + otp. if in (iii) I.] Ex. and the singular focus ( ±it]. 4. 9. 4. then the ortho0. It is ? + y*2£yeT. What relation connecting the distances of a point on the curve ? from two real foci replaces (iv) in the case of the quartic [Use Ch. if PA + PB = where 2u. The curve Soc. for diagrams of this case. 8. 10.y + y* = ((XP)(0ty). and P a variable point. A B are fixed points is PBPA = 2v. gonal trajectory of the family of curves with differential equation M du + Ndv = [Use to obtain N (« 2 . or n zero ? Ex. is 1.6 XVIII be taken as CARTESIAN CURVES 321 Then (a as ovigin. Ex. m. Ex. iii. (v). 3. 25. § 3. Show and that. becomes any circle with at two points only. in § 3 Discuss the modifications which must be bicircular quartic is a Cartesian curve. Find the orthogonal trajectories of Cartesian curves (i) with given singular focus and two given ordinary foci. (v) By inversion of (iv) obtain the results of § 3. [Invert with respect to an intersection and use Ch. Two Cartesian curves with the same real ordinary foci cut orthogonally.] Ex. The distances from a given focus of the intersections of a Cartesian curve with a variable line have a constant sum. and which cuts orthogonally a fixed circle with its centre at a The other two foci are the limiting points of the fixed circles. XIV. (0. 115. /3y—ja + aft (3y+ 7 aa/3. 2. Ex. 7.c 2 ) dv = Ex. the other three make become the circular lines through a focus. 6. The roots of § 3 (iv) are = touching the curve 07 +ya + a/3.] focus. Ex. 0) are also foci. (vi) become the circles &c.) does the locus of London Math. What P become. @ in (ii) by £±ir]. x i + y i = ^y. made when the centre on y [§ 3 (ii). y). we see that any two bicircular quartics with the same four real concyclic foci cut orthogonally. § 3. 17. Inverting with respect to any point. p. § (ii) 3 (v) becomes the circular lines through or else § 3 (v). [Use polar coordinates. (iv) M and N in Ex. AB = 2c. the bitangent being %x=2£ + y. (See Proc. Hence the equation of the curve is obtained by replacing a. consists of a single oval. XIV. Ex. Ex. + 77). 5. (ii) with two given ordinary foci and passing through a fixed point. (xa. The points of contact of 2 (/3 (ii) with its bitangent are real if y + yoc + a/3) > (a 2 + /3 2 + /). Cartesian curve is in three different ways the envelope of Hence the a circle whose centre lies on a fixed circle with centre at the singular focus. The first root makes (ii) become the bitangent.*)* + 4y(? + V *)(2x2iy) = 0. (£. taking the focus as pole.] Ex.
Find the locus of a point fixed circles have a constant ratio. . if rays of light Ex. (iv). emanating from a point in one medium are brought accurately to a focus in the other. a point Ex. the points of contact ' x + y + kz=0 In fact from equation W 1 and W 2 lying on xy + mz 2 If = 0. A man B Find the surface separating two homogeneous isotropic media. take the root thebitangent If we t = of becomes z = and of the quartic (i). As explained in § 4. mz2 2 + 2(x + y + kz)z 3 = ) .b. g = =1 = /= . txy + xz+yz + (k + mt — \t 2 )z 2 = xy+(m — t)z 2 — s t{t 2mt 2 2kt + 2} = z(xy) . Ex. (i) (iii). (ii) we take the other roots z t= a. . c of gives the tangents from from — 0. . is in a pond at A.6 . 14. . and runs to on the land. . find foci. CARTESIAN CURVES XVIII (i) 10. Then and B. swims to the bank at P. 322 Ex. 11. The fourth power of the tangent from P to a circle varies as the distance of P from a line. (iv). 13. z + cy = A A 2 A 3 and the tangents z + ax — Q. and we shall put k = h — m and The equation while equations (i) of the quartic becomes (xy + (ii). We have the intersection of the tangents at p 0. 12. Quartics with. (ii) (v). Two Real Cusps. } . we may suppose G is A 2 . the locus of the other § 7. (v) become . now 1. whose normal distances from two Ex. (iii). (ii) The square of the tangent from P to one circle varies as the tangent from (iii) P to another circle. (iv). . Given one focus and three points of a Cartesian curve. . Show that the locus of P is a quartic with cusps at a and &>'. z + by = 0. q 0. If his time from A to B is the same for all positions of P. z + bx = 0. 3 . (ii). » = (iv). . find the shape of the pond. if The tangents from P to two fixed circles are connected by a linear relation. Suppose that in § 4 both A and B were cusps. (i) it is obvious that this line is the bitangent of the quartic. z + cx = B touching at B B2 B + ay A touching at x .
the Or Cassinian Oval '. with respect Cassinian Curves. But it consists of two ovals if c > a. Y2 . PF . . . 3 . 0). 2. TT2 lie on a conic. co are x — 2. + 1/) + 496 = 0. ) (0. A A A L lie t . and expressing the fact that the four lines a>. Ex. on a conic. The reader can write down a large number of theorems which follow from this fact. x [Putting z = 1 we see that the quartic can be projected so as to have an axis of symmetry. 0). E. B. The points A.. 3. A — The bitangent is 3x+3y+5 = while the tangents from the cusp x = — 3 and so for the cusp (go . Ex.A 3 .A 2 . meet the curve only at a> or a/. . . x = —§ . W .] t . 4. Sec. called a ' Cassinian curve is a lemniscate of Bernoulli. 2 . A. and so for the other points of contact. Hence. 0.B1 . a). y) from ( + c. For instance. F.* If c = a. the quartic has the equation § 3 (i)) that The lefthand side of (i) is at once proved to be the square of the product of the distances of the point (oo. 3 Ex. 0). touches the quartic at A. z') on x + y+kz = to (i) meets (i) on the conic 2z(y'x + x'y + 2mz'z) = 3s' (xy + mz 2 ). 3. A 1 . oo'. if P is any point on 1 (i). B. 1. XVIII 8 CASSINIAN CURVES 323 The coordinates of t are (a 3 ma 2 1. Consider a quartic with a biflecnode at each circular point We may choose It has two real singular foci F1 F% axes of reference such that these singular foci are ( + c. The conic [It is xy + mz1 Wlt W A lt A A = 2y (x + y + kz). Writing down the general equation of a bicircular quartic. y = + * (& ± c) we see readily (cf.] Ex.] § 8. &c] Ex.PF = a 2 2 . [See § 4.Bi B3 lie on a conic. The points of contact drawn to the curve from any point t of the other four tangents which can be of the bitangent lie on a conic through A and B. The curve * ' is Curve '. [The polar cubic of the point (x' y'. The reader may illustrate the properties of the curve by tracing 9 cc 2 2/ 2 + 96x?/ + 1 44 (a. t .
The most interesting properties of the Cassinian curve are given in the following examples. 8. (i) P is the difference 4.e. V2. any point of (i) and middle point of F1 F2 and of S^S. are l . interested will find in the examples their more important properties. 1. Fig. . P is the centre of i. rectangular hyperbolas with FF 1 2 (ii) (iii) The locus of the inflexions is a lemniscate of Bernouilli. the circle on FF l l as diameter and the perpendicular bisector of (i) FF 1 l [The polar and pedal equations of » when is pole are j* 4 . 3. A. V. the locus of a point whose distances from the vertices of a regular polygon have a constant product. (i). are is (0. ±d) where = c a c d? = a c 2 4 if c<a. 2. The angle between OP and the normal at of the angles OPF1 and OPF2 Ex. Ex. we shall confine ourselves to a very brief account of the other The reader who is varieties of quartic with unit deficiency. V3. ( + d. . A. The real ordinary foci c 2 Su S2 <2 of 4 (i) 4 . 184. 8 for the cases = •*. A family of Cassinian curves has given singular foci trajectories are Fu F 2 .. 8. See Messenger Math. the PS1 PS2 = .* In order to keep this chapter within reasonable compass. The locus of the points of contact of tangents parallel to i^Fu is . i. is r 2 +c 2 cos 2 8 = Fig. and Ex. p. PO\ Ex..324 CASSINIAN CURVES XVIII 8 a/c The shape of the curve is shown on Fig. then If c 2 . xlviii (1919). 0) where 2 c>a. 0.e. a2 . A. if 1.] * The reader may extend these results to the ' Mpoled Cassinoid '. giving their locus 2?Vcos20 = 4 + c4 a4 and 2a?pr = + a4 c The inflexions are shown in (Sr^ + c 1 — a?)p = 2aV. Show that Their orthogonal as diameter.
PD constant. Llt Lt . and so for other examples.] Ex. (aA) (0.] Ex. PS. xy = H. 9. Ex.] 6.QS=PS. Ex.B) = (Ot + kA) (P + kB) = H\ Now eliminate A. The intersections of the curve with any line through C form two pairs of an involution with one double point at Cand the other on AB. Quartics with Two Eeal Biflecnodes. 5.] symmetry . Find the locus of P. y) is a focus of A\* + 2H\p + Bfi 2 = 1 which has double contact with the confocals a\ 2 + 0/i2 = 1. § Ex. A conic touches the quartic at selfconjugate triangle. 1. 2. having . 3. Aliter. a Cassinian curve. 5. Ex.] D cut at a given angle. H. Ex. 8. Any line through A is divided harmonically by BC and the curve. A quartic has real 2 4 its equation is (a: 2 +pz ) (f biflecnodes A With the notation of + qz1 ) + 2 hz* = 0. [The curves are the inverses of the family in Fig. (OL + k)\ 2 + (B+k) M 2 = 1. C and the Cassinian curve through [Each locus is P with singular foci B. curve ? [§ What do gives £ the results of § 3 become in the case of a Cassinian 2 or ±(c i — a i )i. (iii) The Cassinian curve through P with singular foci A.PC. [If C is the centre and r the radius of the director circle. The sum of the angles made by PA and PC with a fixed line less the sum of the angles made by PB and PD with the line is constant. The tangents from A to the quartic are and tangents from B are 2 X BM BM AL t and AL^. 10. The as Ex. if the polar conic of P with respect to a given Cassinian curve has a constant eccentricity.PS' = r*CP\] Ex. If ABCD is a parallelogram. P the fixed point and C bisects PQ. Ex.PB. the envelope of a circle which cuts orthogonally the director circle of a hyperbola and whose centre lies on the curve. (i) (ii) to 0. The locus of the foci of a variable conic concentric with and having double contact with two given confocal conies is a Cassinian curve with singular foci at the foci of the given confocals. Find the orthogonal trajectories and the locus of the inflexions of Cassinian curves with given ordinary foci. project the quartic into the curve with double obtained by putting 1 for z . and B. M M 1) 2 ABC [qx* + py* + (pq + 2h)z*=0.] 3 (iv) . while S and S' are the foci of the ellipse. = ±c The Cassinian curve is § 9. 11. B. 4. The locus of the intersections of two circles having their centres at the foci of a given conic and touching any tangent of the conic is a Cassinian curve. 7. we have x*y*=*AB. 8 with respect Ex. [A Cassinian curve with the singular foci of the given curve as ordinary foci. [If (x. is find the locus of P when PA.XVIII 9 QUARTICS WITH TWO BIFLECNODES 4 325 Ex. The locus of the foci of an ellipse with a given director circle and passing through a given point is a Cassinian curve.
quartic has flecnodes at the circular points. and another on a tangent at B. 0). The { (a 2 (a. inflexions.2 a/3y (a + + y)} + a 2 directrices are y= 0. AP [Rationalizing the equation of Ex.y . Discuss the conditions under which the and points of contact of the bitangents are real. s + 25 = with bitangents x 9x + y = 4. 7 on a tangent at A. If the interof the inflexional tangents at the flecnodes is taken as'brigin and the intersection S of the noninflexional tangents as (— g.a 9 with ABC inflexions other than those at selfconjugate triangle. Ex. 2 — 5. 11. If P is any point on the quartic of Ex. The other four bitangents form a vertices lie selfconjugate triangle. § 10. Ex.BP */OA ± a/OB~ ± y'OC ~ (0y)cri (x 2 + y 2 2<Xx + <X 2 )i + . 9. — [Two roots of § 4 (iv) are t = + (pq)i.h) z* 0. giving the bitangents 2 2 qx* + 2 (pq) ixy+py + 2 hz 2 which touch the curve where qx* +py 2 + 2 (pq + h)z Illustrate = = 0. If 9h 8pq each of these bitangents meets one bitangent of Ex. Ex. t Two roots of § 4 (iv) are = ±(pq + 2h)i. 0). = 0.] quadrilateral of which two on CA. The two latter are on the diagonals of the quadrilateral formed by the tangents at A and B. [qx2 +py 2 = z = Q. C. 2.2 <X0) k. Four of the bitangents of the quartic pass through C. 8. B. 3a: 8 touch at two points on AJ$. 7. 0). 326 Ex. 10.CP _~ . Quartics with Two Unreal Flecnodes. the quartic is If these are (x 2 + y 2 ) + 0" + y2 .] = y. The conies of Ex. 1.] Ex. which touch the curve where qx 2 +py 2 + 2(pq + 2h)z* = 0. A quartic cannot have a biflecnode and a flecnode. &c] Ex. [In § 4 = we have m=f=g=0.. 6.] AB... 2(0 + ya)a. Their eight points of contact lie on a conic having ABC as selfconjugate triangle. by tracing x2 y 2 — 9 a. 6. bitangents. quartic has three real ordinary foci A. two on CB.2 yOL (x2 + y 2 ) + 8 OLpyX The .. Ex. QUARTICS WITH TWO BIFLECNODES XVIII The eight a. = y. + 2. and two on AB. (y. CA^. 0). &c.2 7 . The points of contactof the bitangents lie on a conic having ABC as Ex. (0. 7. we get . the equation of the curve takes the form section A O (x2 + y 2 ) (x 2 + y 2 + 2gx + c) = . Ex. 2 BC. = 0y+'y_a+a0a 2 . giving the bitangents 2 qx 2 ± 2 (pq + 2 h)i xy +py = 0. A and jB lie on a conic [3qx 2 + 3py* + 2 (pq + 2. 3.
The tangents at A belong to the involution. 2 g <c. BC.] = B Ex. London Math.] Ex. [Putting p= q = — 1. 2 lie on a conic touching AT at A. 2. The noninflexional tangents at A and meet the curve again in and F.g*>c. Soc. Quartics with a Node and a Biflecnode. 1 can be generated as the locus of of three ways. 5. 9. XIV. 5. Ex. A A 1 A A s B touches AE. Bj. xii. C form an involution and their points of The tangents from the intersection of AE CT and BF form an involution. Ex. [For the following examples seS § 4 or use the symmetry in x and y. and the inflexional tangents meet at 2'. from A form an involution pencil whose Ex. Ex. PO and the tangent from P to a fixed circle with (3) PO PS oc PO PS oc the tangent from the chord through P to a fixed circle with centre 0. will be found in Proc. may 1. With the notation E 2 i (x zt ){y 2 ~z*) + 2mz (xz + yz + xy) + 2hz i 0. 4. Ex. A. § Ex. s . 8. s 2 .g >c. Quartics with Two Real Mecnodes. 3. + 2(x + hz) z 3 = [The quartic by putting 2 = Ex. 2 3 . 0). Ex. 4. A it A t A Llt L .] § 12. Ch. B. f=g = m in § 4 (iv). Any line through The two tangents from 1 .on a conic touching AB at A. of § 4 1. WiEh the notation 0. Ex. t .] § 11. Discuss the case in which the real foci are 3. Ex. Plie on a conic touching AT. 1. 6. 327 (y. 3. one value of — 1.). The four tangents double rays are AB. . [Cf. 7. 2.] be projected into the symmetrical curve obtained B is divided harmonically by the curve and AC. L lt L M1 M2 lie on a conic. F. B touch at the points M M Ex. The quartic of Ex. +>. E. (1) P in one The product centre (2) S is . of constant. A its equation quartic has two real flecnodes may be written A and B. The points of contact lie on a pair of lines through B and . 4. A quartic has a node of § 4 its equation may be written 2 (x 2 +pz ! ) (v* + qz ) A and a biflecnode B.XVIII 12 Ex. . p 22. NODALBIFLECNODAL QUAETICS (£. (3)&>0. P 2 B lie on a conic. . Ex. . [The three cases are given by (l)k>0. The tangents from contact lie on a conic. AC. BT. Three pairs of bitangents intersect on and the other pair t is on AB. A. Diagrams {2)k<0. A lt A it A E. P perpendicular to PS of a fixed 2 circle with centre S. 25. The conic Ex.
B. Transform the curve of Ex. NODALBIFLECNODAL QUARTICS XVIII B. The tangents from tangents meet AB form an involution with BA. 1 into a cubic by quadratic transformation. The conic tangent at B. We take the taenode G (0. B B . With the notation 0. 7. of § 4 The quartic has a node equation (x* and a flecnode B. 24 . B B B A 2 s touches AC at A and the noninflexional Ex. 5. [Putting g on AC = 0. i Es. lie Ex. The intersections of the bitangents lines through B. The quartic and a flecnode. 1) and the tangent at C as y = O. Hence. and investigate properties 8f the curve by this means. BC. Ex. We now consider at as quartics with a taenode. z 3 y. The points B. » takes the form 2 2 2/ 2 + 202/w 2 + m4 O Hence the equation (i). + 2(fx + y + %z) 2s = 1 Ex. 0. 6. B § 14. readily seen that the coefficients of . the The points of contact of a taenode C. /=1 in §4 (iv). four tangents to the tangents at G. 1. its may be written . other than touching the quartic at C. t Ex. z 3 x.e. 5. [Replace Ex. Quartics with a Taenode. 6. if (ad — be) (bd — ca) (cd — ah) = 0. or otherwise. y.] A Quartics with a Flecnode and another Double Point.B 328 Ex. 3. or by pairs on = 0.Mlt M lt t a 2 lie on a conic touching the inflexional tangent at B. Then by Newton's it is diagram. u2 does not contain y as a factor. lie a quartic from on a conic conies of closest contact to the two branches of the quartic at G meet in two points at G and in two other points The . t 2 — pq + h± (h?+p)§. The conic ABE E F touches the inflexional tangent at B.z2) (f + qz*) + 2xyz. The biin two pairs of points forming an involution with A. 12 Elt E m= 2 are collinear.] quartic cannot have a cusp and a biflecnode. a double point which two linear branches touch. i. the bitangents are z* qx 1 +py* + 2txy + 2xz + [pq + 2hf) where § 13. of § 5 has a cusp A Ex. 4. = where u2 and ui are homogeneous of degree 2 and 4 in The polar cubic of G with respect to the quartic is y(zy + u2) = 0. if x.B . zx 3 » in the equation of the quartic are zero. x by z (z—x)/x. 2. z 2 x 2 z 2 xy.
we have u2 x 2 and equation (i) v ' = = becomes (yz or. 16(3t/ + a. C3 (c. G2 (b. x + by — 0. squaring this equation and subtracting from (ii). (ii). z ) 2 = Fig. . a 2). the lines x + ay = 0. Then with a proper choice 2 oi homogeneous coordinates. + x 2) 2 = (yz (1 _m (1 ) (as* + Pl a*y + Pi x 2 y 2 +p3 xy* + Pi y*) (ii). . for. c% Ct (d.m) { x 2 + \ Pl xy + (%p a . . factorizing the righthand side of + x2f = m ) ( X + ay) (x + by) (x + cy) (x + dy) . (iii). 6 2). 1. 9. 0).l. and a line through A as a pair of these conies have y ai e the conies of closest contact  = common chords. we must have p1 = a + b + c + d = . 1. The conies yz + x 2 = ± </(l .±p 2 ) y 2 } with the quartic at 0. Let this line meet at A (1. 0. x + cy = 0. to the conic zy + u be B (0. . (iv). we see that Since either conic meets the quartic in seven points at G. x + dy = at being the tangents from G to the curve which touch G^a. 1.cP).XVIII 14 QUARTICS WITH A TACNODE 329 joined by a real line. y = and let the point of contact of the second tangent from A 0). 1. 7 (x + 3)(x + 2)(xl)(x4).
excluding the tangent at G. In Fig. [On yz + mx i + \{ml){ab + cd)yi * = 0. Ex. Ex. . 3). .s/17 3a. 88. For a more detailed discussion of the tacnodal quartie see Messenger Math.1) (<b 4). touching where a + ay) (x + by) = @(x + cy) (x + dy). (2.122/13 = 0r 12a.] conic touching the The points of contact of b n) b Si lie on a 4ic at C. . 3. *). . The bitangent 6^ meets the 4ic. namely bu and 623 b2i and b3l b3i and 612 The bitangents are shown in Fig. 9 the suffixes only of Ca 6 I2 &c. ' lie on a conic touching [On yz + a? .] xlvii (1917). 6 ]2 and b u involution with a double point on CA. are given. . four lie on a conic touching the 4ic at C and dividing Ex. The other harmonically. This may be seen by writing (v) in the' form yz + x 2 = a. = 2<X{x + ay) (x + by). Cx C4 and C2 Cs . .330 Writing QUAETICS WITH A TACNODE a XVIII 14 = i(lWm). 1 = 2 (# + cj/) (a. b si »nd b u . 6 2S two are the intersections of Cj C2 and Cs C4 . = 4zx (Pi = 0). (4. (x + ay) (x + by) + /3 (as + cy) (x + dy) and then squaring this last equation and subtracting from Interchanging a and /8 we have another bitangent b 3i ' (iii). Of the AB [On i(yz + x 2 ) = (1m) {2x + (a + c)y} {2x+(b + d)y}. 4. We have thus three pairs of bitangents. Their equations are Sx +12y + 43 = 0i 12a... 5. . p. j8 = i(l/m) we is find that the line b 12 (or b 21 ) {<x(a + b) + l3(c + d)} x+(aab + Pcd)y2! (x = . 2.242/29 = 0) 15x 122/ + 23 5 = 0) = 0)" The points Clt G2 C3 Gi are . The intersections of each pair of bitangents lie on the same conic through touching at C.] Any pair of bitangents divides CA harmonically. ). 1. Cs Ct and the points of contact of 6 12 the 4ic at C. C17 C2 and the points of contact of 6 W the 4ic at C. vertices of the quadrilateral formed by b n . 7. if). 16 (32/ + x 2) 2 = 7 (x + 3) (x + 2) (<c . bu and 6 25 in an Ex. 9 for the quartie * . + 24j/ + 59 = 0) 15oj+12. A BC [On p 3 y i + 2p 2 xy Ex. + dy). . (v) a bitangent.] . on a conic touching [On yz + x Ex. 6. (1. (3.] lie Ex.
See Ch v III. the points of contact Cu C 2 C3 of the tangents from the cusp on a conic osculating the § 8.&!.] In Fig.axf/y. properties its 7. 10. 1 at B.XVIII 15 QUARTICS WITH RHAMPHOID CUSP 331 Ex. are represented by their . &c is [CA * (3. oo ) and GA the line at infinity. Clt Ca C. XVI. Ex. 1 to Ex. and AB touches the conic of Ex. curve at C. b b. . The equation of a quartic with a rhamphoid cusp can be put in the form 4 y (x + ay) (x + by) (x + cy). 2. are 2x + 5y = 5._8. [Write down the polar cubic of C. In this case the points of contact of the four tangents from C are collinear. (0.* rhamphoid cusp lie Ex.] of the tacnodal quartic by quadratic [We can transform § 14 (iii) into the quartic of Ex. = 2 . 3x = 5j/ + The equation of the curve is (6» + ^' = 4*(x8) (a + 6). suffixes only. is bitangents b\. [The quartic can be projected into a symmetrical curve. 5) &!. 0). and properties deduced from this symmetry or by modification of Ex. by replacing a. &c. i). 9. (?. &c. Or we can transform it into the cubic of Ch. Quarties with a Rhamphoid Cusp.ax)/y. 15. If in § 14 « 2 has y as a factor. . . Ex. zby z(y. 10 G is (0.] Fig. If a quartic has a C.] The (a. . (yZ + x *y C. Obtain transformation. The equation of the quartic may be put in the form My^z* + (x + ay) (x + by) (x + cy) (x + dy) = 0. and the equation of the bitangent 6 12 in the form 2a/Mz = (a + bcd)x + (abcd)y = 0. (5. y by x*/y. 6 a & 8 are 6x = 5y + 3. In the figure . (5y + x*y = ix {x 3) (as + 5). 2. are (6 + c + l) x + (a + bc)yz = 0. the curvatures of the branches touching at C are equal and opposite. the tangent at C.. § 15. § 8.a2). 1. by x (y . l. 8 by replacing z by z — x^/y.
Obtain properties of the curve by transforming it quadratically into a cubic in this case.y. (ii) Replace y by x'jy and .] x on the conic of Ex.332 QUARTICS WITH RHAMPHOID CUSP XVIII . [(i) Take B as C z by z .. The and and 6. = 2y(x + ay). Ex. There is no loss of generality in taking a = in Ex. The points of contact of quartic at C. 2. C and CL lie on a conic osculating the [yz+x* Ex. AC passes through C 63 to the intersection Ex. of 6j 5. Ex. 4. 7. . The points of contact of 6 2 and 6 3 lie on a conic through C. 1. 6 2 6 3 form together with a range of the same crossratio as the pencil C(Ci C1 C3 A). 3. The intersections of AC with 6. C2 Cs . 6.] line joining the intersection of 6 2 x . 15 Ex.
[The notation is that of Ch. 1) is [The polar cubic of canonical. The properties of the twentyeight bitangents have been worked out in very great detail. &c. xix and the references in Grelle. and Dickson's Finite Groups. ax* 3. ix (1910). p. § 8. cxxii. 1. 145. XVII. 1 . Math. . p. 205 &c.JlB)\z = with two of the — \/ changed into + \A § 3. The reader who wishes to pursue the matter further may consult Weber's Algebra. as twenty = ( — l)iz. . Find the inflexional tangents and bitangents of xi +yi +zi [y = 0. pp. We cannot spare space for more than the most interesting of these results. counting are four inflexional and twelve bitangents. The equation of any nonsingular quartic can be put in the form 2 2 + by* + czl +fyz (2 y* . 209 Proc.3 zx + 2 a?) i hxy (^x^Sxy + Zy^+px1 (3yzxyxz) + gy (3zxyzyx) + + rz2 (3xy—zx—zy) = 0. . 1.CHAPTER XIX NONSINGULAK QUARTICS § 1. It is of class 12. xxii (1891). 275 cvii. London Ex. except such properties as are special cases of theorems relating to the inflexions of a curve of any degree. ch. 173. x+1? y + X*z = 0. 0. . are Bz* — 2Fy z' + Cy* = and the same (F. London Math. The remaining bitangents Ex. (1. II. has twentyeight bitangents. 2.i &c] The bitangents of 1 ax t + bif + cz* + 2fy'i z i + 2gzi x i + 2hx i y = and eight similar bitangents. Soc. We shall return to the subject in Ch. In this chapter we consider the nonsingular quartic without double or triple point.] See Mathews. . &X. are twelve undulatory tangents. p. and twentyfour inflexions. &c. II. Of the geometry of the inflexions practically nothing is known.J~CA)\ y + (H. xcix. . Blickfeldt.3 yz + 2 z ) + gzx (2 z . §§ 112122 Miller. Ex. Proc. p.] .^SC)i x±(G. Nonsingular Quartics. Soc.
NONSINGULAR QUARTICS Bitangents of Nonsingular Quartics. s4 + xyu + 2az 3 x + (a 2 + 2c)z 2 x2 + 2aczx 3 + c 2 xi + 2bz 3 y + (b 2 + 2d)z 2 y 2 + 2bdzy 3 + d2 yl is = 0. y. XIX 2 Suppose that any two of the twentyeight bitangents of a nonsingular quartic are taken as x = 0. . the lefthand side reduces to a perfect square.. . q W= = = = = = = Hence Through the four points of contact of two bitangents of a nonsingular quartic pass jive conies each of which passes through the points of contact of two more bitangents. 0. Let its equation arranged in descending powers of z be z* + 2z 3 (ax + by) + . of factors p. When we put 0. y = 0. y 0.. 2 V= z so that + (ax + by)z + (cx2 + dy 2). where u= This may a conic. may be chosen in general in five ways. of bitangents. so that k. z. it reduces 2 to a perfect square. Now these six pairs have the property that the eight points of contact of any two of these pairs lie on 0. say (z 2 + azx + ex 2 ) 2 and when we put x 0. = 0. q 2 linear in x. p 0. where U= 2abz2 + 2 (bcx + ady) z + 2cdxy — u. Now may be written xy(U+2kV+k 2 xy) = Choose k so that (V+Jcxy) 2 . say (z 2 + ~bzy + dy 2 ) The equation must therefore evidently be of the form y = = . namely x = y = a conic. we obtain thus six pairs and five pairs such as p = 0. and that 0. The equation of the quartic H+2kV+k 2 xy has a pair U+2kV+k then takes the form where W = V+ xypq Jcxy. q the conic passes through the eight points of contact of the four bitangents x 0. Since k can be chosen in five ways. = W 2 (ii)» It is evident that p are also bitangents. be written xyU =V 2 (i).: 334 § 2. q = 0. U= (i) and V= are conies. The condition that xy should factorize is readily seen to be of degree 5 in k.
or W = i(W+Z) + %(WZ) will vanish when x . p= q = 0. s= lie on a conic. . = and / = xypq —W'2 = xyrs — Z 2 . is a constant. 3/ = and pqrs only meet the curve in their points of contact. since x = and. q Then the quartLe is / = 0.XIX r 2 NONSINGULAR QUAETICS 335 = For suppose that any two such pairs are ^ = 0. Consider now the pair of bitangents x There 0. Let another 0. deduce xy(pqrs) = (W£) (W+Z). according as their six points of contact do or do not lie on a conic. 2 W is of the form xy/fi + fi(pqrs) similarly it is of the form xp/v + v(yq — tu). 1) = y = 0. For otherwise Now we W—Z and y of W+iJ. Then we have Now — 4/= — 4xypq + 4tW 2 = —4xypq + (xy+pq — rs) = x2 y +p2 q 2 + r 2 s 2 — 2 xy pq — 2 xy rs — 2pq rs. Three bitangents of a quartic are called syzygetic or asyzygetic. We have shown that. One such pair is y and p = = = = = = pair be t = and u = 0. W= y is any conic through the 0. 2 2 The equation of the quartic (xy)i is therefore + (pq)? + (rs)i = (iii). and the quartic will go through which is impossible. the point (0. . which 0. W= We and Z— being conies. cannot have x a factor of vice versa. Hence there is no loss of generality in supposing fi = 1 and the upper signs taken in the ambiguity. where 0. p are five pairs of bitangents such that the points of contact of any pair lie on a conic through the points of contact of x and q 0. if points of contact of x = 0. = 0. Hence we must have xy where ft = n (W +Z) = 1 n~ (W±Z). q = is the required result. ' a bitangent is not altered by multiplying its equation through by a constant. = shows that the eight points of and r = 0. The symmetry of this result contact of p 0. s = 0. 0. Similarly four bitangents will be called syzygetic or asyzygetic according as their eight points of contact do or do not lie on a conic.
B.AR2 = .] Two bitangents. Ex. . D.] is the triangle formed Ex. quartic are syzygetic. D. If points of contact are P. C. QiQi).] Ex. If four bitangents. 19] Ex. . u = passes through the intersection of either x = 0.] .) are overlapping in the two cases respectively. [By a proper choice of coordinates the bitangents become and y = 0. t = and s = 0. (AB. .: 336 NONSINGULAR QUARTICS for 2 XIX 2 Comparing these expressions W we get Therefore the line through the intersections of r = 0. CQ1 CQ2 AR1 . . limiting case of this. then all four bitangents are syzygetic. See also Ch. 3. the required [Taking the conies as V+k. AQ t AQ2 CPt CP2 Qs . 2. (CA. Ex. taking x = The four common tangents to two ovals of a Ex. Ex. Ex. . . . . [If the quartic is conic is xyU = V* U+2kV+ k^xy = 0. . £> = 0. § 6. § 6. (PC. Four concurrent bitangents are syzygetic. P . Three concurrent bitangents are syzygetic. Show that a conic can be drawn touching the quartic at A. 5. C. and the conic is V+kxy = 0. Compare Ch. and V+k t xy = 0. we find that will 17+ 2 k V+ k 2 {x 2 + 2 hxy + y 2 ) have y as a factor for a suitable value of k. II. lie 2. 2 . x 2 + 2hxy + y 2 = The equation of the quartic is (a. [Use Ex. 1 A conic through the points of contact of two bitangents meets the curve again in A. [Project the points of contact of one of the bitangents into the ftircular and remember that a pair of common chords of a circle and conic are equally inclined to the axes of the conic] points.xy = The preceding example is a conic is D + {k 1 + k 2 ) V+kJc^xy = 0. the upper or lower sign being taken according as the bitangents are An even or odd number of the involutions syzygetic or asyzygetic. P1 P2 ). [Use Ch. 4. the required For the quartic may be written xy (U+ 2kV+k 2 xy) Ex. if y = is a bitangent. BR. B. not all concurrent.or j/ = 0. 6. Ex. conies are drawn through the points of contact of two Show that their eight other intersections with the quartic on a conic. 5. [As in Ex. £.£. BP PP 1 . § 3. by three real bitangents whose Rx and P2 then + BR. 7. = {V+kxy)*. and 2 Qx and ABC .] 5. Hence The 378 intersections of the bitangents lie by threes on straight lines. as the fourth bitangent. are such that any three are syzygetic. I. q = 0. 3. XII.2 + 2 hxy + y 2 ) U = V 2 and.
c may be changed without loss of respectively. the bitangents are the diagonals of a quadrilateral whose vertices are their points of contact.f The twelve bitangents of a complex touch a curve of the third tion. . y = 0. If the sides of the triangle of reference are three bitangents of a quartic. See Humbert. f=g = h=0. 423... But it is at once proved that these the conies U+2kV+ conies cut x = in an involution. r. exact differential. = as a e b e . syzygetic where p. a x (3 x a 2 /32) is an involution. If the lower signs are taken. second case If in the q = —hx + by+fz. f Another proof is the following § Take x = 0. ' x . . P and Q. A Either bitangent of five pairs ' any pair of a complex is met by the other in an involution. The signs of a. 8. an involution. y. We . The lines ax and b x the x conic A^'B^A^A/. A lt B B in . p. z 1 2 i 1 (by* + 2jyz + cz' ) 1 . and similarly (PQ. . b. to + ax i + 2hxy + by i f. the equation of the quartic can be put in one of the two forms 2 i i 1 p x + q^y + r z' ±2qryz±2>pzx±2pqxy 1 = xyzw. ( + cz + 2gzx + ax' ) . q= hx + by +fz. Let a 6 be the bitangent. and similarly for the intersections of a 6 with a3 and b 3 &c. vi (1890). . If pdx + qdy + rdz is an exact differential in this latter case. and meets a x b x a t b 2 in oc x /3lt a 2 /?2 Then A[. the bitangents are asyzygetic. pdx + qdy + rdz is an r = gx—fy + cz. and the conic A X A[B X B{A 2 A£B 2 B 2 Suppose a e meets this last conic in are three such conies. an involution. o 5 6 5 are included h?xy = 0. signs are taken. (PQ. the linepairs a^. The points of contact of the given It bitangent also belong to the involution. the equation reduces or p = ax + hy — gz. is a^. IV. the bitangents are and pdx + qdy + rdz an exact differential. Six pairs of bitangents such that any two pairs are syzygetic are said to form a Steiner's Complex* shall denote the pairs by a x b v a 2 b 2 as b s a 4 6 4 a s b 5 a 6 b 6 The points of contact of a. * Or (i) Steiner's Group or Steiner's Set : or Doublesix. A A 6 ') 6 .. . meets all conies through . If the q. . &c. Liouville's Journal. will be denoted by 1 and A(. . Then taking 2 among as the equation of the quartic. p = ax + hy + gz. z. Steiner's Complex. class. .] § 3. We shall say y ' a complex '. w are is upper linear functions of x. . . . . . — ( 0.XIX 3 STEINER'S COMPLEX 337 Ex. . [Suppose that when we put x = 0. a? /3 2 A 6 A/) is an involuHence (A 6 A 6'. generality and we readily obtain in the two possible cases r = gx +fy + cz.
.. as the condition tliat the intersections should lie on a conic.. This is equivalent to proving that the six centres of the six linepairs included in the family of conies S+21cS' + k 2 S"=0 (i). The conditions that the centres of the linepairs lie on a conic . on a conic. Taking the bitangents as the sides of the triangle of reference.. z = a 3 xi + b s y'i + 2yxy = lie on a. B.. S' = a'x + . . when we replace S by S + 2^8" + h^ S" (i = 1. by Three bitangents of a complex. &A i + bBi + cCi + 2fF + 2gG + 2hH = i i i 0. . . 8'. 2 S"=a"x2 +. a linepair. the points x = b 1 y 2 + c 1 z i + 2atyz = 0. are asyzygetic. conic. V. . or otherwise. For the sake of symmetry we have replaced the U. which is excluded. y = c 2 s 2 + a 2 a. A — bc—f 2 . &c. F= gh — af Suppose now (i) is a linepair when h = k lt k2 . become A t Bit . gx+fy + cz = 0. 2 6). where 8 lie = ax + by 2 2 + cz2 + 2fyz + 2gzx + 2hxy. .. In this case we get .. ax2 + by2 + cz 2 + 2fyz + 2gzx + 2hxy are = = 1. = 0. its centre (x. and can readily be shown not to lie by § 2 The (iii). . z) is given by the equations ax + hy + gz = 0.. or 7c6 and that the quantities A.. (i 6).. if a^t^ = a 2 6 s c. the equation of the quartic takes the form (l t x2 + m x xy + n^zfi + (l 2 yx + my 2 + n2 yz)~* + (l3 zx + m3 zy + n3 z 2)? = 2 intersections of this with the sides of the triangle of reference are at once found. 2. hx + by+fz — which give x2/A =y/B = z 2/G = yz/F where 0. t By Carnot's theorem.. . * See Ch.338 For STEINER'S COMPLEX # XIX 3 it is readily proved that the envelope of a line divided three linepairs in an involution is a eurve of the third class* touching the lines in question.. on a conic. y. 1 m 2 » 3 = 0... Ex. 8". = zx/0 = xy/H. XV.. § 5.f The six intersections of each pair of a complex lie on a conic.z + 2/3sa. and If xy of S= § 2 by is S. no two of which belong to the same pair.
b^. We call three such complexes a complextriple.. it C.* There are sixtythree complexes of bitangents. [The quartic is the envelope of the conic U+ 2kV+ l?xy = for varying h. London Math. .6 = 63. P is The the corresponding Cayleyan. § 4. of the sixth order whose rows are the determinant A At B . . . [See Ch. Between them they contain all the twenty. V= 0.. 3. to deduce from Ex. Use Ch. For by the result that three bitangents of a complex are asyzygetic.. . § 6. pairs of bitangents of a complex are the polar conies of at an intersection of the fixed conic with the Hessian. say 04 and a2 with the pairs ct^a^. t . . The remaining bitangents of this new complex are not in the original complex. .6) vanishes. no one of a s b 3 . Ex. a 2 b 2 a 3 b3 a4 &4 a 6 b6 a 6 b 6 take any two bitangents not These determine a complex forming a pair. Soc. . XIX 4 RELATIONS BETWEEN COMPLEXES if 339 Therefore the six centres lie on a conic. There are 28 (7 = 378 such pairs.. all three This proof is due to Baker. F Q t. Hence the number of complexes is 378 J. . Ex. . But is this is the case. We' have seen in § 2 that from any two bitangents a single complex may be obtained. 2 that the intersections of pairs of bitangents of a complex lie on a conic. if no two belong to the same pair. The if P in Ex. z2 . p. four of the bitangents lying in Proc. for A considered as of the fourth degree. 1. . H. (i = l. curve of the third class touched by the bitangents of the complex is 2. XV.] Ex. ix (1910). Relations between Complexes.. . .2. Hence A vanishes identically. Any one of the six pairs of a complex 2 determines the same complex. xy = U= is the Hessian. a 6 b6 can be syzygetic to both ax a 2 and'&j&g. In the complex whose pairs are %&!. . .] Ex. 148. 25. . . II. § 4. 1. Similarly a x and b 2 determine a complex containing the pair a2 bx and no two of the three complexes thus determined by aj) x a x a 2 a 1 b 2 respectively have a bitangent in common other than a lt b x a 2 b 2 which are common to all three. Any nonsingular quartic is the envelope of the polar conic of any point P on a certain fixed conic with respect to a fixed cubic.. The cubic is a curve of which the Jacobian of 0. . XVI. For any two bitangents determine a complex. an equation in /Cj = k 2 k3 kt k6 and k 6 . and has five roots /Cj = . eight bitangents * ..
. their pairs are of the form We b^ a i C l! a2 C 2> a3 C 3> ai Ci> a6 C 5' OjCj. Two pairs of the given complex such as a. Suppose then that ax a2 .. . we may without" loss of generality suppose a3 to be Let in the second (and b 3 in the third) complex of the triple. whereas a v a 3 b 3 are syzygetic. For the bitangent Cj must lie in the complex determined There is no loss of by a x a2 or in that determined by a x b 2 . which with the tbiityone complexes just mentioned make up the total of sixtythree. . a 4 6 4 a b b.a2 . i.. tt 6 C 6' b3 c3 . For they are not a pair of this complex. &. then c^Cj. Hence the complex with the pairs b^. b 5 c 5 . . b i ci . to be in the former . Neither of them is in the first complex of the triple. . Returning now to the given complex with pairs = a^. Hence each complex is a member of fifteen complextriples including in all thirtyone complexes. we get thus thirtytwo more complexes.. . . Oy 2 J C1 C2 . take any bitangent c 1 not in the complex. o6 c6 .& 2 can be selected in e 2 15 ways.. . a 2 c 2 a 3 c 3 be three pairs of the second complex. b 2 c 2 .. and consider the three complexes with pairs generality in supposing it <XyOj 2. These form a complextriple and therefore contain all the bitangents. They cannot both be in the second (or both in the third) complex. Now the complex with the pairs a 2 a3 c 2 c3 contains the pair b 2 b3 since a 2 a 3 b 2 b 3 are syzygetic. and three bitangents of the complex no*two of which are a pair are asyzygetic. Since a 3 and b 3 are interchangeable in the given complex. 340 RELATIONS BETWEEN COMPLEXES XIX 4 complexes. . and the result proved. e. either four syzygetic bitangents in another complex containing the same four bitangents a complextriple including all the twentyeight Summing up we have Two complexes have common and form with .a 6 6 6 . b 2 c2 contains the pair b3 c 3 This argument is at once extended. chosen in sixteen ways.. and the other twentyfour lying in one of the complexes only. have six bitangents in common. . . a2 b 2 a3 b3> . shall'show that the complexes determined by a^. a2 c 2 . . c x c 2 are pairs of a complex. and in particular a 3 and b 3 . We get two new Since c x can be complexes determined by c^Cj and b^. . 5 ttjCj.. for the given complex determined by 04 & x is not altered by interchange of a2 and 62 . .
. b3 . . .. 2. 6. oe ce are such that any two have six asyzygetic bitangents in common. therefore ai a r is .XIX 5 THE HESSIAN NOTATION 341 bitangents. . . is cij. . c. The bitangent paired to it in this complex denoted by ij or ji. is a{ bj. . . \. 72. and form a complexpair of eighteen bitangents. . * Seven is b1 . 4. . and the bib v b 5 b 6 by 17. . b2 . b2 c 2 . c6 . b s . . 75. . if they were. 84. § 5. '> j j I of the digits 1. OjCj. 71. i andj being any two occur in the third. The seven bitangents ax a2 a3 at as c 6 be are such that no three of them are syzygetic. .. 74. a complex . 18. o B c 5! . For. 37. b 6 . 73. Denote the bitangents a lt a2 a3 ait ab a6 by the symbols 68 or 81. c is ae c. and so is b x . 76. 48. . «A. The Hessian Notation. There are 8 C2 = 28 symbols such as ij = ji. 83. o 4 c . The complexes formed by a iK a n°2> asK «A> a5&5 a 6 & 8 a i C l> *2 C 2' a3^3> ai®l> ab C5> a i C 6' . 28. b 2 c 2 . . «2 & 2> a 3°3' a J>i> a b°s. 27. 57. 6 c 3 . b t . ajbi. b 2 . tangents b 1 b2 b 3 . b i ci Then ax a 2 a 3 a4 a5 ae . 5. Such a set of bitangents is called an Aronhold Seven '.. 38. bjbs aja B . . contrary to the supposition that ai} a it ar are asyzygetic. 3. b 6> <x 6 . three of or else they have six bitangents in common no which are asyzygetic. by interchanging a6 and become In the three complexes at the end of § 4 alter the notation b 6 and by writing c for c 6 They . o 6 c 5 3 4 .> ^2^2' a6 & e! ^6 C ' a3 CA> ^4 C4! a6 C63 b3 c 3 . 86. . and cannot occur Hence it must in the second as it is asyzygetic to a^. It does not occur in the first by definition. 67 or Denote c by 78 or 87. ai Cl! OjCj. b 3 . 82. a complex. a r bs a s b r . Another Aronhold . Hence we have a notation which will include all the twentyeight bitangents. 47. 85. 58. an Aronhold Seven . . Now c must occur in one complex of the triple ai hj> a b i> «A> a bj> •'•> a i aj> hi hy . We note that the bitangents ij and rs cannot be the same. .. . .
g.342 THE HESSIAN NOTATION XIX 5 The notation is not symmetrical. 7. . 4. 6. see Fig. their from § 2. Ex. 78 * or 12. e. 34. (ii) their symbols involve two digits twice and two digits once. that four bitangents are syzygetic. p. e. 4. It follows if The criterion of § 2. 1. but it is perhaps as convenient as any we can find.. The bitangents are now denoted by the symbols 12 or 21. 56. . 8 in the figure. Ex. applied to show that three bitangents are syzygetic. Buil. g = fg + e equal ellipses of small eccentricity and nearly concentric. &c. . These portions are narrow ovals each touched by one bitangent. 12. the major axis of one being parallel to the minor axis of the other. symbols are of one of the two types 12. 3. and / 0. 2. 2). de France. Fig. proceed to the limit in which the ovals are indefinitely we may consider the bitangents as the twentyeight lines joining in pairs the eight points 1. 34.. xxvii. 1 will readily if (i) their symbols in Fig. 23. this section. 34. 12. If l. 3. 5. = = Kg. We shall show that this is the same notation for the bitangents as that before derived at the beginning of we narrow. 229. 34. where e is a small constant. We may visualize * the notation by considering the quartic are 0. 23. 13 or 31. de la Soc. 8 (Fig. 56. FontenS.g. 2 involve six distinct digits. 2. 41. Math. The quartic consists of four portions each lying inside one ellipse and outside the other. while any two ovals have four tangents in common. Hence the quartic has twentyeight real bitangents each touching the curve in points which lie very close to two of the points labelled 1.
have the possible types f We We We j7. since we have just shown that i8j7. rs. 1 * begin with the type pq. 78 56. a complex in Fig. retracing our steps. 58 67. Fig. tu are syzygetic. 3. r. qr. 6. k.ji. It follows at once that we have complexes of the two types or 18 17.. 68 75. 68 67 14 23. 2.. Hence we have thus all possible h ^4 sixtythree complexes and then. 28 27. 5. 58 57. we see that we have given above every possible set of three or four s • The = = . 34 12. .. 8i and 7j. * The figure only gives a upon for a proof. s are 7 and 8. 48 47. 1. 38 37. qr. rs or three of the type pq. 24 31. q. 2. XIX 5 THE HESSIAN NOTATION 343 In Fig. denote any of the integers . First take the case in which two of p. 8i and 87.i7j8. rs. 2 the four bitangenta either join four distinct pairs of points or else are the sides of a quadrilateral.ji and 7j. chapter.ij78. we see that the notation for the bitangents obtained at the beginning of the section agrees with that obtained from Fig. 4. though it . 7j. fost type contains S C2 28 complexes and the second 35 complexes. 78. .. Moreover. f i. 2. j. and therefore must not be relied assists us to visualize the results obtained in this 1.. special quartic. is can deduce the fact that three bitangents of' the type pq. syzygetic bi tangents. from the original definition without the use of Fig. j8. I. 18.
Hence . qr. 37 38. 5.. 4. 6.. 28 2k. jk. 8p 8q.. rs.. It is not in the first. Repetition of this argument shows that is lZlj. for j7. interchange 7 and 8. a complex by definition. . . 78 7k is a complex. It follows that the other type 8i. p and q are any of the integers 1. i8j7. i7 j7. . j8jk. . 81 8j..jk 7 or 8." are complexes... 3. ik. Therefore it is in the third which makes i8. 58 57. 2.. It is not in the second . and 17 18. 7. _ j 8 jk.. i7j8. 8j.. . We have now shown that there are complexes of the type . a complex... 38 37. 8j. 68 67. 21 2j. ij . .. % Now We take the type ij. 48 47. .. is ij 78.j8jk. A similar process holds if we kl. Therefore for any value of k 18 Ik. . where . kl are syzygetic. It follows that i8 is in the complex j8 jk But...344 These are THE HESSIAN NOTATION all syzygetic.. Iplq. since this complex is %8 ik. ik. for otherwise we should have three bitangents in different pairs of a complex syzygetic. kj. changing the notation.. 8. 78. q... 2p2q. 18 78 7k. 81 8j a complex and ij. ik are syzygetic. 78. i7 j8. . . 8i. 71 7j.jk syzygetic. for XIX 5 i8j7. jk are syzygetic. i8j8. whence klkj. This exhausts all trios of the type pq.. is a complex. r... is For the first type we note that jk complexes in one of the triple of 18 17. We and 8i. i8 is ik. . s is Now take the cases in which one of p. have shown that Ik. 27 28. jk. j8.. have the two possible types i8. 28 27. kj is also syzygetic...
72 74. The points of contact of this bitangent e. where u r is homogeneous of degree r in x. u2 — 0..] Ex. 61 63. tu. . c5 of § 5 are 16. Conversely. 62 64. 52 54. . 56. it is in the third. 56. rs. 41 43. and is not in the are syzygetic.ws + 2u 2 w + u3 = 0. for the equation of any quartic * The proof is similar to that which Shows that the points of contact of the tangents from to a plane curve lie on the' first polar curve of 0. take any point on a cubic surface as the point the equation of the surface in homogeneous 1). . 12. Hence 12. coordinates is If (0. tu are syzygetic. 46. i. / = «. Obtain all the 288 Aronhold Sevens of bitangents. 12 34. 14 32. lie on w.= 0.j = 0. 82 84. 2 joining one point to the other seven or lines forming a triangle together with the lines joining one of the remaining points to the other four. we have the equation u u3 = w2 for the intersection of the tangentlines with the plane w = 0. cs . 3. 1. 32 34. 26. 34.XIX 6 LINES ON A CUBIC SURFACE 345 consider three bitangents of the type pq. y. 0. c4 . on the inflexional tangents of the cubic at 0. Since 56 is in one complex of the triple Now 12 14. Ex.. 71 73. 36. 2. qr. How many sets of four syzygetic bitangents exist ? [105 + 210. 51 53.* w between / = and ^. This is a plane quartic with the interof w = and the tangentplane at to the section Mj = surface as a bitangent. we 0.] The bitangents c u § 6. Ex. 21 23. c2 . 56 This completes the proof that three bitangents of the type Then the proof that pq. complexes are of two types and that there are only two types of three or four syzygetic bitangents follows as before. first two. The tangentlines from O to the surface have of contact on their points ^=2(^u.+w dW Eliminating 1 2) =0. rs. z. . for instance. Lines on a Cubic Surface.. 81 83. rs or pq. 34. any nonsingular quartic can be derived from a cubic surface in this manner . [Lines in Kg.
Hence either 3. Hence the curve of intersection of the plane with the surface has and Q for nodes. or 27 of the lines on a nonsingular cubic surface are real. z(U+2kV+k2 u Hence the conies in 1 z) = 0.w = u 1 = Q make up a Steiner complex (§ 3). p. we obtain five pairs of bitangents of the quartic which together with the pair w z = 0. 129. we shall show that either 4. which planes through the line w = z — meet the surface again project from into the family of conies U+ 2kV + k u . . z. homogeneous of degree 2 in The surface meets the plane w = Icz where U and V are x. * The method is due to Geiser. Q one on each of the lines joining to the points of contact of the bitangent. Through any line m drawn m i. evidently no loss of generality in supposing that through a line of the surface. This result enables us to deduce properties of the plane quartic from those of the cubic surface and vice versa. of the lines on on to Hence the projections from the surface are the bitangents of the quartic other than I. Hence = w of a cubic surface five planes can be each touching the surface in three points and meeting it again in a linepair. by I. = kz for such a value of k meets the surface in The plane three lines which form a degenerate cubic with three nodes. 15. where ut = any bitangent. Math. w= In Ch. u x For the present let us denote the bitangent to any bitangent of the quartic. y. » By choosing k to make such a conic a linepair. The plane joining other than I. i 1 z = 0. The projections of these five linepairs form with I and the projection of a complex of bitangents of the quartic curve. and must therefore degenerate into the line PQ and a conic. Annalm. touches the surface in two points P. 16.* 0. or 28 of the bitangents of a nonsingular quartic are real. But the curve is of w= = P degree 3. 7. It follows that there are twentyseven lines on a cubic surface. at each of which the plane touches the surface. 8.: 346 LINES ON A CUBIC SURFACE XIX is 6 can be put in the form u^u^ =u 2 2 . XX There is the plane the line as w = passes w — z = 0. Take The equation of the surface is u 1 w 2 + 2Vw+Uz where = 0.
.planes of the ci Ne two of c lt c 2 c 3 can meet. The plane of a 6 cB bB meets the surface in the three lines a e> c5> "6. § 7. Z> 6 Cj. . tangent. hyperboloid with e lt c 2 c 3 as generators in two points. meet c 6 Let the planes . . Sehlafli's Doublesix. 3 (diagrammatic only). <x Suppose that the linepairs (§ 6) line e of a cubic surface is met by the five Vl. b B c3 b 6 cit 6 6 c s meet the surface . . behig . . . . It cannot meet a 6 .Hence b 6 meets one of these lines. & 2 C 2> &3 C 3. . the planes a 6 6 1 c 1 surface. . Since b 6 meets c x . triple &c.c 2 c 3 c 4 in four points lying on the surface. We now proceed to find the other thirtysix complexes. . Suppose that meets the Oa cs Fig. be c 2 . there is no loss of generality in supposing it to .XIX 7 SCHLAFLI'S DOUBLESIX of such triple tangentplanes is 347 evidently The number 27 X 5 j 3 = 45. b i c t> 6 5 c 5. 6 6 is a line of the surface. Then it is at once seen again in the lines a lt a% a3 at> as that the configuration of twelve lines on the surface ( ax a2 a 3 a4 a 6 a 6 b2 b3 l J t h 64 b5 b6 . The above result gives us all the twentyseven complexes which contain I. Through each of them passes a generator of the opposite family to c x c 2 c3 One of these is ae let the other be b a ' .
and 6 4 b 5 b 6 which meet them. . . . . the number of double. We showed that each line on a cubic surface meets ten other and therefore does not meet sixteen lines.planes through a x are ' «A> »A> «A> a A> a A> follows readily that we might have constructed the doublesix starting from the pair afi^ instead of a 6 b e In other words. These lines can be chosen out of a 1> a 2 . hut does meet the other five lines...sixes is 216 r 6 = 36. Hence all the complexes of bitangents are projections of lines. as each doublesix may be determined by any one of its pairs. . . . . a configuration is called a Schlafli's doublesix'. namely the conic in which the tangentlines from to the hyperboloid meet the plane. The projections of the generators of a hyperboloid from any point on to a plane all touch a conic. . Such . sets of bitangents of a quartic there are twenty of six bitangents touching a conic. a 1. the doublesix is symmetrical as regards its six pairs it . Since the triple tangent. «A. . are 27 — doublesixes or else contain I. For the plane through and a generator touches the hyperboloid. A doublesix projects into a complex of bitangents. ... Now the lines a x a 2 a z 6 4 b & b6 of a doublesix lie on a hyperboloid. a2 a3 being generators of one family. Hence the corresponding bitangents of the quartic touch a conic. No two of these six bitangents belong to the same pair in the complex. being generators of the other family. Hence there x 16 2 = 216 nonintersecting pairs of lines on the surface. . In any complex From a complex the theorem that the intersections of each pair of lie on a conic we obtain li/nes The six drawn through any point of each intersecting a pair of a given doublesix the second degree. . . . a x meets b2 b3 bt bs b6 only. Hence: . . «A> as&fi> a ehDenoting the bitangents into which the lines a x b x c15 . or ax bx is a pair of the complex determined by ae b e Similarly for a2 b 2 a 3 b it &c. project by the same symbols. «A> «A.: : '348 is SCHLAFLI'S DOUBLESIX XIX 7 such that any line does not meet each of the lines in the same row or column in the array { }. For instance. . . a cubic surface lie on a cone of . . a e Hence in twenty ways. Therefore. we showed in § 6 that a x be a B b x are pairs of the complex determined by I and c l Hence ai> ae> &i> ^6 are sy z ygetic.
These cusps are the intersections of a conic with a cubic toughing the sextic at the cusps and the twentyseven bitangents of the sextic are the projections from of the twentyseven lines on the surface. 2. By taking at the intersection of two lines of the surface obtain properties of a binodal quartic. 1. Ex. . By taking of § 6 on a line of the cubic surface obtain properties of the bitangents of a quartic with a node. 3. Ex.XIX 7 SCHLAFLI'S DOUBLESIX 349 Ex. The intersection of any plane with the tangentlines from a point to a cubic surface not passing through is a sextic with six cusps. .
CHAPTER XX CIRCUITS § 1. &c. unless the contrary is stated. Circuit and Branch. so that the direction of the tangent 'point'. ' oo Pig. Such a statement as the number of inflexions curve. 'inflexion' P . and travels along the curve. Suppose a point starts from any point on a curve. Heretofore. when we have discussed singularities on a we have not always distinguished real from unreal singularities. and by — mean 'real point'. 2)' is only true if we include of a nonsingular nic is 3n(n unreal inflexions. 1. 'real inflexion'. In the present chapter we shall make the distinction between real and unreal points on a curve.
.. i. 'tripartite'. e. The part of the curve thus traced out by P is called a circuit of the curve. if an oval shrinks into a point. number of its intersections with the Circuit alters. 'bipartite'. § 2. The reader must distinguish between the use of the word branch as curve at a multiple defined above and its use in such a phrase as branch of a ' = = . all Suppose. that it does not Such cross itself.. it next travels inwards from infinity to a very distant point B such that A and B are neighbouring points on any projection of the curve. Suppose that. though we shall be mainly concerned with algebraic curves. Zug circuit. one three branches. * In German. If line is moving in the plane of a circuit. § 3. has no crunode..* These definitions will apply to any continuous curve with continuously varying tangent. Odd and Even Circuits.XX 3 ODD AND' EVEN CIRCUITS 351 at P varies continuously. e. Ovals. A curve loses one of its circuits. if in its journey P travels outwards to infinity from a very distant point A. two. Thus a hyperbola has one circuit composed of two branches. . The curve need not be algebraic. its Suppose that a circuit is closed. the the number is increased or decreased by two (or some even Suppose that a number) as is obvious from Fig. i. unipartite '. Ast circuits as 'part' for 'circuit'. as shown in Fig. 1 shows a curve with' two circuits. It is to be noticed that an acnode is not counted as a circuit. Fig. The portion of a circuit traced out by P between two infinitely distant positions of P is sometimes called a branch of the curve. An a circuit will be called an oval. ' point ' . and speak of curves with one. oval can be reduced to a point by continuous deformation. . 3. three. and the other closed and crossing composed of itself. This means that in general P jumps from one end of an asymptote to the other. Some English writers use branch. moreover. points are finite. 2 which gives successive positions of the oval as it shrinks to a point. which shows three conIt follows at once that secutive positions of the moving line. Eventually P returns to 0.
In general an nia has an odd or even number of odd circuits. two an odd and an even circuit. as n is odd or even. in particular any oval. In fact tRe argument of Fig. and odd circuits which are met by every straight line in an odd number of points. The least number of points in which a circuit is met by any The greatest number of points is The index and order are both odd or both called its order. The theorem will therefore be established if we show how to deform the first circuit so that all its odd circuit in lines points are distant. 3. Two odd circuits meet even circuits. line is called its index. Fig. or in an odd number of points.352 ODD AND EVEN CIRCUITS XX 3 circuits can be divided into even circuits which are met by every straight line in an even number of points. and the number of its intersections with the other circuit alters. even according as the circuit is odd or even. Its intersections with the second (fixed) circuit will then be practically identical with its intersections with the asymptotes of the second circuit. First of all deform the circuit so that its crunodes are cut '. The oddness or evenness of a circuit is evidently unaltered by projection. meet in an even number of points. is even. This is done by replacing the part near the node by the dotted ' . if one of the circuits is continuously deformed. Any. and any number of lines meets an even circuit in an even number of points. for it meets the line at infinity in an even (zero) number of points. closed circuit. To prove this we first notice that. the number is increased or diminished by two (or other even number). We now show that the first circuit can be deformed so that all its points are very distant. an even number of meets an odd or even circuit in an even number of points. But an odd number of lines meets an an odd number of points. 3 still applies.
4 (i). but for our present purpose we shall think of these distinct circuits as a single circuit whose Kg. 4 (ii) is would necessitate the alteration of the direction which a moving point traces out the circuit. 5. a 2216 . 4 (i). we can evidently withdraw such a branch continuously to an indefinitely great distance as shown in The ovals. as shown by the arrowheads. Keeping the two distant ends of. A nodeless curve cannot have more than one odd circuit. 4 (ii). for it in shown in Fig. The cutting of the nodes may really resolve the circuit into two or more distinct circuits. The method of Fig. is evidently unaltered by cutting the circuit now consists of distinct nodeless branches and no two of which cut.XX 3 ODD AND EVEN CIRCUITS 353 line as inferior. A . oddness or evenness nodes. each branch fixed. Pig. Fig. Also each oval can be deformed into a point and we have achieved our purpose.
On the sphere an even circuit is deformable into a pointpair and an odd circuit into a great circle in such a way that each pair of diametrically opposite points remains diametrically opposite. Use Ex. 2 to prove the theorem of § 3. Ex. 4. In particular. it cannot meet an even [If it met in two. 2. The difference between the order and index of a and greater than zero. If an even circuit is projected similarly. would meet in one or more nodes of the A 71 is nonsingular nio has therefore one or no odd circuit as odd or even. 3. Pig.354 ODD AND EVEN CIRCUITS circuits XX 3 For two odd curve. Ex. 1. 5. the line joining them would meet the quartic in five points. Ex. 1.] Ex. more than one circuit of a (singular) quartic cannot meet any other point. 5. circuit is even An odd circuit in circuit. whose points can be divided into diametrically opposite pairs. that the projection is a closed circuit on the sphere met by any great circle in ir+2 points. Show Ex. the projection Is a pair of diametrically opposite circuits each met by any great circle in an even number of points.] . An odd circuit is projected on to any sphere from its centre. [Project after cutting the crunodes.
the cutting of a node either divides the circuit into two even circuits or into two odd In the latter case it is possible. 4. circuit ' We call now ' an odd pointeven number of points. if a node is cut as the circuit is divided into two closed circuits. tangenteven. 6. on a circuit id As a point P travels from a fixed point (§ 1). tangenteven. the tangent at a fixed direction. What happens if a node is cut the wrong way as in Fig. by cutting a second node. tangentodd. quintic with A two intersecting even circuits cannot have more than one odd Ex. 4 (i). A ' pointodd To its and an even circuit meets any line in an odd reciprocal with respect to any base ' pointodd ' circuit Pointeven. 4 (ii) ? [See Landsberg. reconvert the figure into a single even circuit with two nodes less than the original. conic any given draw an odd number of tangents from '.XX 4 Ex. 355 circuits of a quintic cannot meet in more than one point. 6 shows the four possible varieties we may therefore point. Pointeven. A by an even number. Show that. reciprocal will be called tangentodd Similarly we obtain tangenteven circuits as the reciprocals of pointeven circuits. Fig. but not closed. 6. if the circuit is even. RECIPROCATION' OF CIRCUITS Two odd circuit. Annalen. if all the nodes are cut. the numher of circuits obtained differs from n + 1 in Fig. p.] § 4. Also that. 7. Beciproeation of Circuits. [See Pig. Pointodd. to circuits.] Ex. 8. lxx (1911). Show that. The ' ' ' of circuit. 563. '. if the circuit is tangenthas passed an odd number of times odd. For the tangent at P P Aa2 . Fig. Math. Pointodd. finally returning to turns through an odd multiple of tt. closed circuit has n crunodes. tangentodd.
which proves the result. arriving at A A left A. pointodd or pointeven circuit contains respectively an A tangentodd or tangenteven circuit contains respectively an odd or even number of A odd or even number of inflexions. As starts from 0. cusps. Now a is the number of tangents from A to the circuit. through an even multiple of n. according as the circuit is tangentodd or tangenteven. and b is the number of tangents parallel to I. A § 5.356 RECIPROCATION OF CIRCUITS XX 4 through any fixed point A. since the number of tangents from Similarly the tangent at P turns to the circuit is odd. Q starts from and in the same direction as it finally returns to A. travelling along the Suppose that the tangent at a point Let the tangent at a fixed circuit meets a fixed line I in Q. of these statements is the reciprocal of the other. a being the number of times Q passes through A* b the number of times Q jumps from one end of I to the other. We confine our attention to the first. if and only if is an inflexion or an intersection of the circuit with I. 7. point of the circuit meet I in One may P A P Fig. Therefore a + b is even so that c is even. Hence a + b + c must be even. . 7). . But Q reaches a limiting position on I and then moves back again. P A * The initial starting from A and the final arriving at A taken together count as passing through once. (Fig. Hence a and 6 are both odd or both even. and c the number of times Q travels up to a certain point of I and then begins to move back again. Inflexions and Cusps of Circuits. and returns to 0. traverses the circuit. if the circuit is tangenteven. .
9. since the circuit is of order 3.. The inverse of a circuit of r branches with respect to a point not on the circuit is an even closed circuit with an rple point at 0. [Project one inflexion to infinity. 2. Let the tangent at P meet the circuit again at B. and [State whether point. 4.. and notice that during Ex. [(i) Invert with respect to 0. . An odd circuit has 2r+3 inflexional tangents no one of which meets the circuit at more than three points (coinciding with the point of contact). circuit of Ex. If the tangent at does the product PQ 1 PQ. cannot lie in this order on the circuit. 3. 4. [Project one inflexional tangent to infinity. 7. [The curve and inflexional tangents divide up the plane into these various portions. An odd circuit of order three without node or cusp has exactly three inflexions (Mobius's theorem). Describe the circuits of the curves in the diagrams contained in Ch.or twocircuited cubic and show the portions of its plane from which 0. and therefore coincides with P twice. Ex. 6. 8. C Ex. The reader may enunciate a theorem concerning the positions of the points of contact similar to that of Ex. or 6 tangents can be drawn. [Deform the circuit into some standard position. curve ? Deduce the theorem of § . PQa meets the circuit at Q„ Q2 Qa . Treat similarly some nonsingular quartic curves.. the inverse circuit is odd or even as r is odd or even. Draw a one. Show also that 2r+l tangents can be drawn from any inflexion. (ii) cut orthogonally a given circle.] Ex. 5. 10. Ex. If is an rple point of the circuit. An odd circuit without node or cusp has at least three inflexions. give their order index. (ii) Deform the circle continuously into its centre. . 2. 1. change sign as P travels along the . B'. Ill to XIX. 11. 5. but has 2p (2p + 3) and q bitangents. An the deformation p—q is never increased or diminished by an odd number.or tangenteven or odd. . when . A'B'C A. [Project one inflexion to infinity. 4 . Show that 4) — q is even. C. 5. R travels along it twice in the opposite direction. Hence there are two finite inflexions.+ 3) bitangents. 357 Ex. . Ex. .. &c] Ex. B. An even number of circles of curvature of a circuit (i) pass through a given point 0.] Ex. A nonsingular curve of odd degree has at least three real inflexions.] Ex.] .] inflexions even (odd) circuit has no node or cusp. and 2r + 2 tangents t'rom any other point of the circuit.' Show that it has r(2».] Denoting the three parts into which the inflexions divide the by llt l2 l 3j show that there are two tangents from any point of ?i whose points of contact lie on l 2 and l 3 and that the tangent from the inflexion at which l 3 and ls meet has its point of contact on lt Show that if two lines meet the circuit in ABC.XX 5 INFLEXIONS AND CUSPS OF CIRCUITS P . A'. Ex.. There is only one other tangent through the infinite inflexion. As P travels along the circuit.
2. 14 has no inflexion. 15. Similarly for a cusp.] true for each circuit of this quartic. where the curve C is shown with a crunode. Method of Variation of Coefficients. This is illustrated in Fig. § 6. 8. represents a curve of the same degree C and lying very close to it. Another example is The theorem as a whole. 5 it Every tangent to a circuit without node or bitangent meets in the same number of points. An oval without double point has at least four vertices '. and for the quartic Ex. [E. 8. 14.g. INFLEXIONS AND CUSPS OF CIRCUITS XX 12. a cusp. In general also an adjacent curve on one side of C has as . 13.] Ex. 12 cannot be true for a circuit with a bitangent. vP — yL is = x l yi . an acnode. Show that the result of Ex. points at which the circle of curvature has fourpoint contact. and' two adjacent curves also. remembering that the circles of curvature at two points of the oval lying between two consecutive vertices cannot both pass through 0. if has a crunode while an adjacent curve has none. Ex. If the oval of Ex. We method enabling us to obtain the equaan assigned degree. number of circuits. the sum of the maximum radii of curvature less the sum of the minimum radii of curvature is half the perimeter of its evolute. equation thus obtained.358 Ex. • ' [Invert with respect to a point of the oval and use Ex. Such a method consists in giving small increments to the coefficients in the equation of some given curve 0. It will be noticed that in general. The new Fig. the adjacent curve has an inflexion close to each inflexion of C and two more inflexions close to the crunode of G. but may be true for a circuit with a node. i. &c. shall require a tion of a curve with inflexions. and the radius of curvature is a maximum or minimum. e.
the adjacent curve has an equation of = a + bx + cy i px1 + 2 qxy + ry1 + sx3 + . & The from results for the case of a cusp are at once written down Fig. The number of ideal bitangents of an adjacent curve is the same as for G and these bitangents are adjacent to those of G. 424. Annalen. bitangents whose points of contact are unreal or lie on the same circuit. F are conjugate * the form .. c are small. Two . of these must be real bitangents in their final position just before becoming unreal. Clifford. F as nodes consists of one circuit and has two real bitangents t lt t as 2 may be readily proved by projecting E. where q2 < pr and a. shall call a real bitangent with unreal points of contact The name is due to W. approximates to the curve G near EF. For take the case in which E. F. Let Q be the quartic with a real cusp which approximates most closely to C at E. The quartic in the position in which it has E. F. vii.. F of these bitangents are consecutive to t x t 2 The other Now since the quartic Q to EF. In the neighbouring position Q will have four bitangents of the first sort. . e. pro We an ideal bitangent. which is in general a convex oval without inflexions. i. or no real bitangent. reasons as follows The line joining f two nodes counts as four bitangents. Two . as C varies. F and as C passes into a consecutive position let Q pass into a consecutive quartic still approximating to G near E.XX 6 VARIATION OP COEFFICIENTS 359 a small oval enclosing the acnode of G. the line EF counts as the limiting position of four bitangents.* The adjacent curve on the other side of G has no circuit near the acnode. G has a pair of conjugate imaginary nodes or cusps E. and the other two must be unreal becoming real.f similar argument holds if E. vided If. G has no imaginary nodes or cusps. an adjacent curve has respectively two or three ideal bitangents adjacent to EF. This will be proved in § 8 by reasoning which does not involve the use of the result we are here establishing. . To each line joining two nodes of corresponds four real bitangents of an adjacent curve having real points of contact. are cusps. p. . . 6. F are nodes. : imaginary. however. Math. into the circular points and inverting with resp*ecfc to the cusp. or no real bitangent. K. 8. two must be consecutive A F Taking the acnode of C as origin. If the nodes E. a bitangent of C must approximate to each bitangent of Q near EF. The approximation near the origin is the ellipse = a + bx + cy + px 1 + 2 qxy + ry2 Zeuthen. To each tangent from a node of G to G corresponds either two bitangents of an adjacent curve which are both real with real points of contact.
2. Then if / = 0. If h is even. e is any small constant. is an (n+2)io meeting the ellipse in the 2 (n + 2) real intersections of It has also (n + 2)n inflexions and no ideal e = 0. g = 0.ic and n 2 ic. i + n^ic with n (ni — 2) + n2 (% — 2) + 2%! %2 = (%! + to 2 (nt + 2 — 2) a nonsingular (n^ inflexions and no ideal bitangent. If a nonsingular n x ic and n^ic exist having respectively ^(TOj — 2) and n 2 (n 2 — 2) inflexions and no ideal bitangent. r may have any of the values 0. For yj/2 if = is /x = 0.. 71. [(i) x =f(t). prove that one Ex. which. g = and lines. The reader will readily establish the truth of this statement in the cases n — 2 and n = 3. Jc. %K2) +%(%. Take n + 2 lines each meeting the ellipse in real points. 2 Obtain the equations of (i) a unicursal curve. there exists a nonsingular (n l + n i )ic with (n 1 + n 2 )(n l + n 2 — 2) inflexions and no ideal bitangent. where *n. where /. is inflexions. where e is a small constant. But the result is only required in the second part of § 7.2) and no ideal bitangent. (ii) a nonsingular curve. /= Ex.] /= is any unreal non For we shall see in § 7 that the number of real bitangents of any nonsingular «ie adjacent to an nio with nodes at E and F is the same. where singular («— 2)ic or (» — l)ic with real equation. i (ii) (x + y*—l)f=f or (x\)f=e. are the equations of nic. ef = eg. and does not affect our discussion of the circuits of a nonsingular curve in §§ 7. bitangents as in the proof of the preceding theorem. 1./2 = are the equations of the an (n t + « 2 )ic with n x n 2 crunodes. 0. For any given value of n with it \here exists a nonsingular nic n(n — £) inflexions and no ideal bitangent. of degree n consisting of one circuit only which is of index or 1 as n is even or odd.. Now use induction. . if h is odd. y degree n and \jr = = <f) (t). ellipse. If (where q> = of the adjacent curves /+ e$ = does not go through and e is a small constant) has 2r inflexions near and that the other has 2 (Ic — r). let us take above remarks which will be useful one or two theorems. has ~k real branches through a point 0. Let us assume that such a curve exists and that moreover meets an ellipse in the 2n (real) points. Perhaps neither line of argument will appear quite convincing. e = 0. .. 1. Then /j/g ) = e.360 VARIATION OF COEFFICIENTS illustrations of the XX 6 As later on.. z = yjr{t). r = \lc but. meet in n^n^ real points. 8. \jr are polynomials of has not more than one real root. .
(The straight line is a bitangent in each case.. Now in the process no two ideal bitangents can coincide.* Hence the only way in which i + 2 1 can alter is (i) by the coincidence of two inflexions at a point of undulation and their subsequent disappearance (ii) by the points of contact of a bitangent becoming real instead of unreal. Then the curve is continuously changing its shape. [/ = ci where /= is the curve in but not if p > 3. or vice versa. and we may thus continuously deform the nic till it coincides with some standard nonsiugular nic. Show that an «ic exists cutting the pio in np real points. First consider the case of a nonsingular nic. * This would not apply to two bitangents with real points of contact. for .] p real points. and d is the number of real acnodes. XX 7 KLEIN'S THEOREM 361 Ex. n being any given integer. A pie is cut by a straight line I in p real points. For instance. but not a quadruple tangent. for this would imply the existence of a quadruple tangent. Suppose the coefficients in the equation of the curve to vary continuously. or one node. I. no two of which are consecutive. § 7. k is the number of real cusps. where i is the number of real inflexions. (iii) by the appearance of a node. In any curve of degree n and class m n + i + 2t = m + k + 2d. &c. 9. t Fig. or vice versa. Klein's Theorem. Now Fig. or pair of unreal nodes. &c. During the deformation the curve may have one triple tangent. 9 shows that cases (i) and (ii) are really the same and that such an event decreases i by two and increases t by one. they might coincide to form a triple tangent. is the number of ideal bitangents (real. or flecnode. Also we have shown in § 6 that in case (iii) the numbers i and t are not permanently altered.) We may suppose that during the deformation there is never more than one relation between the coefficients. the equation of n straight lines each meeting The line I can always be found if p = 2 or 3. with unreal points of contact). 3.
i is diminished by two. But we have shown in § 6 that an nic exists for which i = n {n. as required.2t by Klein's theorem. \d' pairs of conjugate imaginary nodes. Ex. besides its k real cusps. For n = 2r + S.: : 362 if KLEIN'S THEOREM XX 7 a crunode appears. Klein (Math. Verify this. The theorem of this section is due to F. the following: For n = 2r. and d 1 crunodes. n + i + 2t = n(nl)2(d 1 + d + d')3(k + k') + 2d + k = n(nl)2$3 K + 2d + k = m + k+2d. d acnodes. XI. 5.' As a corollary from Klein's theorem we deduce No nic has more i than n(n — 2) real inflexion. n+(i + 2d1 + 2k) + 2(t + d' + ilc') Hence = n(nl). t = 0.] Prove iSi + 6d = *3k + 6t. which proves the result. &c. and is increased by two again as the node disappears. Ex. k =d= 0. 6. Let it pairs of conjugate imaginary cusps. § 3. 199). i real inflexions. a curve adjacent to the curve consisting of r equal concentric ellipses with major axes parallel to the r sides of a regular polygon. is established for a nonsingular curve. For an nic with double points it is sufficient to notice that we can find an adjacent nonsingular nic for which vmay be greater. m = n (n—l). but cannot be less. We have already shown that for any assigned value of n an nic with n(n—2) real inflexions exists. What property of a curve do we obtain by applying Klein's 2. theorem Ex. 2. Annalen. Hence the theorem II. to its evolute? [Use Ch. Klein takes as standard nio with n(n — 2) inflexions. Ex. Now have •§/«' An flexions and of the proof adjacent nonsingular ?iic has by §6 i + 2d x + 2k inTherefore by Part I t + d' + %k' ideal bitangents. suppose the nic has nodes and cusps. x (1876). add to the ellipses a cubic whose odd circuit cuts all the ellipses in six points. It may be put in the form 'The quantity n + i + 2t is the same for a curve and its polar reciprocal. 7.2). 1. . p. and t ideal bitangents. For a nonsingular nic = m — n— 2t = n(n — 2)~. Hence n + i+2t is the same for all nonsingular nics.
For if it had five. There is some line meeting the curve in no (real) point. which consists of one. For if a quartie had three ovals A. Each oval is of order two or four. 6 for the case of quartics with double points. the curve inflexions. if Ex. for an inflexional tangent to the inner oval would meet the inner oval in four points and the outer oval in at least two. a line cutting B and C would meet A. The reader may illustrate on r = a cos 35. 3. Klein's relation (§ 7) becomes i + It 8 for the nonsingular quartie. Circuits of a Quartie. * f There . 6. If there is an inflexion. and G in at least two points each. a cubic. there are four ideal bitangents each meeting the curve in no real point. XIV.] Ex. infinity as ideal bitangent. result. [Apply Klein's theorem to an adjacent nonsingular curve. Ex * A curve has no acnode and not more than a third Show that not more than a third of its inflexions [Use Ex. It cannot have more than four even circuits.XX 8  CIRCUITS OF A QUARTIC 363 of its cusps are real. the conic through a point on each circuit would meet the curve in ten points. which has the line at 1. is then a circuit approximating at infinity to a semicubical parabola (ay 2 = x s ) and with one asymptote also.f If there is no inflexion. it will be readily seen that they have four and only four common tangents circuits of We have already discussed (Ch. Cf. B. no oval can lie inside another. § 6. B. If there are three or four ovals. they may lie one inside the other or be external to each other. We = See Ex. This will be seen to have two branches with a common tangent. are real. or four ovals. 5. and discuss in detail the nonsingular quartie* Such a quartie cannot have an odd circuit (§ 3). In the former case the inner oval has no inflexion. The quartie can therefore be projected into a closed curve. and a line adjacent to this tangent will meet the curve in no real point. G with B inside A. the reader will easily convince himself of the truth of this statement by projecting the inflexional tangent to infinity. If there are two ovals. three. An wic with ^ crunodes cannot have more than »(»2)2d. What modification must be made in Klein's has a multiple point with distinct tangents ? . Ex. § 1) the possible consider here the case of the quartie. If two ovals are external to each other.] § 8. which is impossible for a curve of degree four. two.
AB = Fig. of course.bitangents of the second sort. where r is 1. 3. it has also )'(rl)x4 bitangents of the second sort. Imagine a string wrapped round an oval of order four. 10. 2 + 4'i/ 2 36)(4a. If it has r ovals external to each other. * If the curve such common is of degree higher than tangents. 8. The equation (9a. We give an example to show how an equation may be found which shall represent a quartic belonging to an assigned class. is a bitangent subtending a 'bay' AGB of the Hence the quartic has quartic with two inflexions (Fig. 2. twice as many inflexions as it has bitangents of tho first sort with real points of contact. be also eight real inflexions and twentyfour . Hence it has 0. 16. Bitangents of the first sort are those whose points of contact are either unreal or lie on the same oval. or 4. which proves the theorem. Nonsingular quartics may be classified by means of their ovals and the nature of their bitangents. or 28 real bitangents. 4. It has four bitangents of the first sort. = e. Suppose we require a quartic with four ovals each with a bitangent. 10). A nonsingular quartic has 4. There will. or 24 bitangents of the second sort and four of the first sort. 2 l92/ 2 36) five. The relation i + 2t 8 now proves the result. 12. A nonsingular quartic has four bitangents of the first sort. there might be more than four .364 CIRCUITS OF A QUARTIC XX 8 which are bitangents of the quartic* Such bitangents are called bitangents of the second sort. If AB is A and a straight portion of the string touching the oval at B.
will represent a quartic adjacent to the pair of ellipses 9a 2 + 42/ 2 and lying inside one = 36. Ex. 3. (yl)(y2)(y3)(*4). Ex. Ex. sd>. x 2 + §y" + 2x i x*y2 = 9.] Ex. 1 x2 . . + 3 = 0. Ex. 0. 2 + 9f = 36. (y1). 4^^ = 4.y2 = = 2 x. (xl)(yl)(x + y). XIX. Fig. Each of the involutions is nonoverlapping. § 2. Describe the nature of the ovals and bitangents of the quartics are the fg = ±e. (i) 2 2 2 Describe the nature of the quartic (x + y —5y) = + is (ii) (yl)(y2)(yS). first sort can [The argument of Ex. and / = 0. XIX. 11. we take we bitangents. ^+ x* + 4a. 4a. 11. outside the other. Fig. as shown by the narrow line in Fig. g = curves given below (i)a»« (ii) + y' 2 «/ = 1. 3. § 2. where 2 2 as* + 4^5. for the case of four concurrent bitangents. 6. where t is a small positive constant. 2. 1. 11 by the dotted line. x'y\ (ii) (x + y2){xy + 2). It will evidently be a quartic of the and as shown in ellipse kind If stated. a small positive obtain a quartic with two ovals. 4. The eight points of contact of the four bitangents of the first sort lie on a conic. The outer oval has eight inflexions and four e constant. where rf> Ex. (yl)(y2). (iii) (iv) (v) (vi) (y2x){y4){y + x). See Ch. Ex. 1 would show that the conic through the points of contact of the three bitangents passes through the points of contact of one of the common tangents to the two ovals which is impossible.: XX 8 e is CIRCUITS OF A QUARTIC 365 * where a small negative constant. 5. (iii) (iv) (xy)(x + y2)(x + y + 2). Describe the nature of the quartic fg = +e#. g = 4x + y 5 /= and <j> is (i) x+y2. 2x2 + y* = (iv) (v) (vi) (vii) x2 + y*±4:X5 2 2 =4. [Use Ch. j/ + 4y = 0. No triangle formed by three bitangents of the enclose two ovals external to one another. x + 4y' = 4. one inside the other. 8. (iii) + y 2 =4. . 2*s + y» + l = l.] . .
0. ideal. due to Harnack. They lie on 1/x + l/y+l/z = 0. x (1876). Ex. — \(1 — t 2 ). and similarly for §§ il. common No oval of a quintic can have more than twelve inflexions or six bitangents. For §§ 5. 7.e. 12. and no quintic has more § 9. A quartic with one real and two unreal cusps has one bitangent with real points of contact. d are unreal. The real inflexions lie on a conic. 7. Show that x'y + y'z + z^x = has exactly six real inflexions and four real bitangents. suppose the curve to be of § 4). we see that the quartic consists of a single oval touched in real points by three bitangents. relative to the tangents at A and B. or four of a. A curve of deficiency B has at most D + l This result. (1 + t.366 * CIRCUITS OF nature A QUARTIC circuits XX the quartics 8 in Ex. degree n with r odd circuits. A A quartic with three real cusps has one bitangent which is conic meeting the curve in eight real points intersects the bitangent. a bitangent such that the ovals lie on the same side of it) and two internal bitangents (the ovals lying on opposite sides).] Ex. In the latter case all the six common tangents of curve and conic are real. .* is familiar if (Ch. circuits. unreal. Maximum Number of Circuits. 1— *). since two odd circuits meet in one or more crunodes. 15 we may classify the various types of curve by considering the cases in which none. §§ 315. after putting z = 1. or lying on odd or even circuits. To ovals of a quartic external to one another have two external bitangents (i. X. c. Let the curve have s other double points. For § 9 project the curve into x^y 1 + ax 1 + by 1 + c = 0. 189. &c. 14. For § 4 take A and B at infinity and consider the possible positions of the conic of § 4. p. A conic meeting the curve in eight real points does or does not intersect the bitangent. § 3. Then D D= = i(wl)(TO2)fr(rl)8. Ex. The fourth bitangent is x + y + z = 0. To two ovals of a quintic external to one another four tangents can be drawn. These odd circuits meet in at least fr(r — 1) crunodes. 1. 10. where Ex. [The quartic can be projected into a threecusped hypocycloid or a cardioid.] circuit. 8.0). b. 9.] fi + 1 = t' + 9 1. Armalen. To prove it in general. 6. * Math. of the of [The nature of the bicircular quartics may be obtained by inversion The nature of their bitangents is given by from the circular cubic. 11.] Ex. Ex. 13 and 16. [Drawing the curve. Describe the Ch. The real inflexions are (1. two. [Three real inflexions lie on the odd than fifteen real inflexions. Ex. XVIII. acnodes.
four being on one side and side of the line of inflexions. 3. ovals. = n(n2) + l which is impossible. First consider the case in which n is even. = Ex.] the cases in which [To find is nic = = ' : ' . Then r is even.XX 9 MAXIMUM NUMBER OF CIRCUITS It will 367 that. least Hence the nio and (n — 2)ic meet in at 2{%r(rl) + s}+r + 2(Dr+l) + (n + r2) =n(n2) + 2 which is impossible. with twentyfour inflexions and eleven ovals. For the case of n and r odd. e. 6. another point on each of the r odd circuits and n — 4 more points on an even real points. Show that not every nonsingular nic can be projected into a closed curve. If a sextic has eleven ovals. protectable into such a sextic). Suppose — D— — (Dr+2)+%r(rl)+8 + {n + r4) = \(n + l) {n2). Let (j> + 8^ + 32^ + 28) (a. and all meeting the odd Then / 2 + ?0 are (1) a sextic circuit of the cubic in three points. the (n — 2)ic meets each odd circuit in three points at least. = 2. take one point on each of the even circuits. and n + r 4 more points on one of the even circuits. the Jr(r— l)+s double points. Take one point on each of the r + 2 even circuits. (tftfV. . (2) a sextic with eighteen inflexions and nine ovals (i. = two on the other = all possible arrangements of the circuits of a nonsingular a difficult problem. The total number of these points is circuits. the curve had Df 2 or more have at least JD r + 2 even circuits. which ha9 been solved only for re 3. is a sextic with eleven ! ! O. circuit. for 6 take (a. for meets them each once at a crunode. 1. if n is a given number greater than 4. 2 it has an odd circuit. odd. if possible. the r(rl)+s double points. [If n is For n Ex. they consist For n 6 we have the result of ten ovals all external to each other and an oval enclosing either nine as in the above example. The reader may discuss of them or only one represent six or less lines in various positions. the curve can never be closed.] = be the equation of six lines parallel to the line of inflexions of a twocircuited cubic / 0. This each odd circuit of the nic at least twice. 2 + 8y2 . Since n — 2 is odd. so that it meets one of them in at least w+r — 2 points and the it (n— 2)ic meets others at least twice. Prove that where f {2^sin3(0 + S) + 3^l}{8r sin30 + llr }+<r = and 8 are small and positive. Hence a (n — 2)ic can be drawn through them. Also it meets each even circuit in au even number of points. and n — 2 is even.32^ + 28) = e. 4. and the nic and (n — 2)ic meet in at least 2{%r(rl)+s} + 2r + 2(Dr+l)+(n3) real points. Ex.
ovals. (ii) n odd. a line cutting this other circuit Fig. In this diagram we have the three nests abed. Then. ?*<£ (n — 2). a> 2 2 inside co^. the nest and one odd circuit being the ivhole nic. and the innermost oval would cut the nic in mofe than n points. . as in Fig. we have : (i) n even. Then. abf. 12. Suppose n odd. if the curve had %n nested ovals and another circuit as well. a line cutting this even circuit (or two odd circuits) and the innermost oval would cut the nio in more than n points. In the following section we prove the converse of this theorem. the nest being the whole nw. o> 3 . 12. ae.368 NESTED OVALS § 10. XX 10 Nested Ovals. or r = \(n — 1). Suppose n even. A o> set of ovals o^. showing that for an assigned value of n it is always possible to find a nonsingular nic with its maximum number of nested ovals and its maximum number of circuits and real inflexions. if the curve had %(n— 1) nested ovals and also an even or two odd circuits in addition. or r = \n. It will be noticed that a nest may contain ovals not belonging to the nest.j inside w r such that ooj lies inside <o 2 w r is called a family of r nested . r^^(n — 3). If an nic has r nested ovals.
number ^(n — 2) %(n — 3) q/ nested ovals. . (iii) It has the maximum. Suppose we have also a thick line) with equation e curve (shown by the thin line) with equation f=0.. The theorem may be considered to hold even in the cases 2 or 3. 13...: XX 11 HILBERT'S THEOREM is 369 even. if we adopt the convention that a single oval shall count as a nest of one oval when n > 3 and shall not count as a nest when n = 2 or 3. Fig. (a?s + y»{»l})(3!») = «=. Ex. y"i„) = «.. Suppose that in Fig.] § 11.» + (a? + tfl)(a* + y*2) .as the maximum number n(n — 2) of real inflexions. such as is shown at the bottom of B b 2216 = . which possesses a nest of p ovals. 0. Find the equation of an «ic with \n nested ovals when n and  (n — 1) nested ovals and an odd circuit when n is odd. has the maximum number \n{n% — 3% + 4) or of circuits. (iv) ii A.For every assigned value of n greater than 3 there exists nic tulth the following properties (i) an It is (ii) It nonsingular.. IX^ + ^l) (a!» + y»2) . Hilbert's Theorem. 13 we have a circle (shown by the n= . (a.
. circle. if e is a small constant of suitable sign. If we take r = n + 2. 13 is purely diagrammatic. now only to establish the existence of the n\e = eg. meeting the circle in 2r points passed through in the same order whether we traverse the circle or the oval.. the derived curve an (% + 2)ic with a nest of ^n or ^(n — 1) ovals and an oval such as 12 meeting the circle in 2 (n + 2) points. e = is a and g = a straight line not meeting the three straight lines all meeting the circle. 2 the lines B2 Bs Bi . since each intersection of the circle and original curve gives rise to is two extra It inflexions by § 6. . with §(« 2 — 3n + 4>) circuits of which (ti — 2) or § (n — 3) form a nest as in Fig. B2r _ 1 B2r be g 0. . and an oval 12 meeting the circle in 2q points (only) passed through in the same order whether we traverse the circle or the oval 12. 13 and one is an in 2n points (q = n). On this arc of the B2r and let the equation of circle take 2r points B t . 13. and inflexions are shown in the figure exist. (n + 2)n its circuits is S! < and the number of !(7i 2 37i with the nest. There is a portion of the plane not containing the nest. . which is bounded by an arc of the circle. = = = = a curve such as is showrj by the broken line in Fig. It has a nest of p + 1 ovals and has in all 2^—1 more ovals than the original curve. q Then. For n = 2 it is sufficient to take any ellipse meeting the For n — 3 we may take the curve circle in four real points. and the n(n — 2) inflexions for the cases n = 2 and n = 3. . h = * Fig.* has also an oval of the same nature as the oval 12. eh We have + 4) + 2%l = §{(% + 2) 3(7i + 2) + 4}.. the oval 12.. . It has too 4g more inflexions. r 2. B . and an arc of 12 lying inside the circle. where circle.370 HILBEET'S THEOREM XX 11 the diagram. oval such as 12 meeting a fixed circle e = Suppose also the two has n(n— 2) real inflexions. The result will then follow by induction. The derived curve has also is i n(n— 2) + 4% = real inflexions. 3. Supposing now we have an nio / = satisfying the conditions of the theorem. B . which do not really . 1 In the diagram p 2.
319 Rendiconti del Reale Istituto Lombardo. xlix (1916). p. III. lxxvii. where a. (ylx). p. 2. having its of circuits. cxiv. . lxiii. p. The r lines g meet e in »• points. p. + y'i)(yax)(ycx) 1. Berichte der K. and use induction. xxii.] .. x. 349. and has a circuit meeting e in r points only which are passed through in the same order whether we traverse the line or the circuit. pp.: XX _ 11 HILBERT'S may consult THEOREM 371 The reader who wishes to pursue further the subject of • circuits American Journal Math. xxxviii. II. xlviii (1915). 177. Annali di Mat. 245 xxix. Soc. i. New York Math. p. k.. 189. Crelle. . Pura ed Applicata. . Annalen. p. p. 489 and 797. such that the two segments containing the q points and the r points on e do not overlap. p. A /= = = = = = = /= nonsingular raic exists for any given value of n. xli.. p. Math. zu Leipzig. 495 and 577. 540. Bb2 . Ex. vii. der Wiss. p. 182 Trans. American Math. r = n + l in Ex. . I is magnitude. b.. p. lxvii. xliii (1910). has q — 1 more circuits than 0. xlvii (1914). Soc. iii. 218. p. e. pp. 416. 48 and 143. Sachsischen Gesell. 197. with a circuit meeting a given line in n points passed through in the same order whether we traverse the line or the circuit. p. 305. p. A maximum number [Take Ex. 126. 117. Bull.. 3. q= (x i n. pp. tg.. d. p. p. xiv. are 2n — 4 constants in ascending order of an nic of zero deficiency with a single circuit of index n — 2.. 170. Ex. 1. 388.(tjkx) = y {y~bx)(ydx) ... where e is a small constant of Prove that ef suitable sign. lxxiv. 410. lxxiii. lxix. p. 115. curve has a circuit meeting a line e in q points (only) which are passed through in the same order whether we traverse the line or the circuit.
.CHAPTER XXI CORRESPONDING RANGES AND PENCILS § 1. and that the curve lias no multiple point other than A and B. if certain relations hold between the coefficients of equation (i). — A ^ m P AB * Assuming that the tangents at A and B are all distinct. VIII. Correspondence of Two Pencils. D = (p1)(?1) The tangents from A to the curve + q 2 pq). Suppose we take two lines PA. To every PB PA PA B PB A Eliminating t and T between (i) and y tz.l + . remembering that a &ple point counts as %h(Jc— 1) nodes (Ch. x Tz we It is a (p + q)ic with multiple points obtain the locus of P. Pliicker's numbers are at once written down. The to every position of are said to trace out 2^>encils with vertices and lines and having a p:q correspondence. 0. This assumption may not he valid.+u p + n' (b fP + b tP. + (Jc tP + k tP. These facts are obvious either from the equation of the curve or from simple geometrical considerations. If a = 0. .. In (i) we have supposed a 0. We have * = = A A .. . We verify easily from this that = 2pq. The locus of is now = 2 (#V.. S = i (p c 2 are the p tangents at each counted twice and the 2(q l)p lines given by those values of t which make (i). position of correspond p positions of PA. have equal roots. tz. at A and B of orders p and q respectively. § 3). the line corresponds to itself in the two pencils.+kp ) = 1 ) 1 1 .. considered as an equation in T.+bp ) + . such that t and Tare connected by a relation of the form PB = — B Ti(a tP + a htP.. (i). n=p + r q..l + . k — =3(2pqpq). x Tz respecy of the triangle of reference tively through the points A and ABC. and correspond q positions of PB.1 + . The p tangents are the p lines of the pencil with vertex at which correspond to the line BA with vertex B and similarly for the tangents at B. m=2pq..5z>q + 2p + 2q).
o . I) is = (p1) (q1) \ very well known. any %ic with an rple point and an sple point B * may be considered as the locus of P._ s _ 2 . The case p = q = 1 ). We have then that the intersection of corresponding rays of two homographic pencils is a conic. .. the lines and joining any point on a (p + q)ic to multiple points and B of orders p and "q respectively have a p q correspondence. . . the locus of P is similarly a (p + q — 2)ic with (p — 2)ple and (2 — 2)ple points at A and B respectively p = q 1 and so on. o n _. The polar reciprocals of two pencils with a p:q correspondence are ranges of points on two lines with a p q correspondence.] T=2(p 2 q*7pq + 2p + 2q + 4 [. ! n=p + ql. o1 . We readily show that. .. a1 .] Correspondence of Two Ranges. conversely. This is evident on putting t for y/z. To any point of the first range correspond 2 points of the second. r = 1 if A is an ordinary point of the curve and so for B. when we get a relation of the form (i) with PA P PB A : = A PA PB BA AB a all zero.. = S(2pqpql). We have a relation such : * r = . . if A is not on the curve.. T for x/z in the equation of the curve.. If a — a = b = 0.. Bb3 . More generally. In fact. when the pencils traced out by and have a (n — s):(n — ?•) correspondence such that n — r — s of the lines through A corresponding to coincide with AB. and n — r — s of the lines through B corresponding to coincide with BA. or is a straight line if the line joining the vertices of the pencils corresponds to itself. and to any point of the second range correspond p points of the first. putting y = tz and x Tz in the equation of the curve we have a relation of the form (i).. Ex. XXI 2 CORRESPONDENCE OE TWO RANGES 1 373 the line point at AB together with a (p + q — l)ic having a (p — l)ple A and a (q — )ple point at B. ci n _ r_ s _ 1 . For this cuvve m=2(pql). K i = Q.. The quadratic transform of the (p + q)ic of § 1 with respect to a conic touching CA and CB at A and B is the intersection of pencils through A and B having & q:p correspondence. The reader may also verify the results obtained by taking p = 2.. q 1 or = = 2. A is a line through B. 5 = i (p 2 + g 2 3p3g + 4). . [The quadratic transform of a line through § 2. .
a node at B ] : The locus is Ex. B. The line joining corresponding points of two ranges with a p:q correspondence in general envelops a curve of class p + q having the lines on which the ranges lie as pple and gple tangents respectively. 5.] : AB Ex. § 2. BC being self . to a given family of confocal conies. Two pencils with &p:q correspondence meet any two lines p q correspondence. if t and T are the distances of corresponding points measured from fixed origins on the two lines. [The pencils Ex. Two ranges on the same line with p q correspondence have p + q self corresponding points. conic through intersection of C are fixed points. (ii) is. and B have a 2 2 correspondence in which The locus is a cubic through A and B. one of the given lines.] Through A and C are drawn conEx. A. A A [The pencils Ex. The tangent at locus of the intersection of BQ and CP.374 CORRESPONDENCE OF TWO RANGES XXI 2 as (i) of § 1. (ii) A (2m — 2)ic with (m — l)ple points at the circular points. B. 1. AP and BQ have a 1 : 2 correspondence. B. Find the locus of the foci of ftrves of the mth class touching given lines when the line at infinity (i) is not. Find the locus of the intersection of BP and CQ. Find the locus of the intersection of tangents from two fixed points A. 3. BQ and is CP have a 1 : 2 correspondence. C are fixed points. * Ex. A. jugate chords of a fixed conic through A and B meeting the conic in P and Q respectively. Two pencils with a common vertex and p q correspondence have : : p+q selfcorresponding rays. The [(i) A (2w — l)ic with (m — l)ple points at the circular points. Consider the correspondence of the tangents to the curve from the circular points. B [The tangents from is A selfcorresponding. C to touch meets I at Q. and the pencils formed by joining any two vertices respectively to the points of ranges with a p:q correspondence have themselves a p q in ranges with a : : correspondence. C. 7.] conic is drawn through three fixed points A. 5. Find the a fixed line I at any point P. B. V.] drawn through four fixed points A. D and meets a fixed line through A in P and another fixed line in Q. 2. through A meets a fixed Find the locus of the have a therefore a cubic through C with 1 2 correspondence. Any line [The B in P and BP and CQ.] BP and CQ have a 1 : 2 correspondence. singular foci of the locus lie on the locus. Ex. A conic [The pencils corresponding. pencils CQ and BP a fixed line in Q. See also Ch. Find the locus of the intersection of AP and BQ. Ex. 6. m— 1 4.
H [The tangent and 12. B. Ex. 9. C. <p (x) are polynomials of degree n. = k<t>(x) and are said to be projective. D. How many circles can be drawn through a fixed point A touching a given line at P such that P and the intersection Q of another ? fixed line with the tangent at A are collinear with a fixed point [OQ and OP have a 1 : 2 correspondence. F are fixed points. Find the locus of the intersection of CP and DQ. C. Any conic through A. J are fixed points. B. variable conic touches fixed lines at A and B. 13. that the involutionrange has 2(» — 1) double points. 11. H. D.] Ex. 8. 10. B. Discuss Ex. IP have a 1 : 3 correspondence.' A similar definition holds for pencils.] : Ex.ndEQ. F. 14.] : is 6. G. 0.] : ' Draw a circle Ex. Therefore by Ex. F. [The pencils EP and FQ have a 2 2 correspondence. C. A.] Ex. D meets two fixed lines in P and Q respectively. How many solutions are there to the problem through two fixed points A. Show that two projective involutionranges selfcorresponding points and that the line on the same line have n + joining corresponding points of two projective involutions along different having the lines containing the lines envelops a curve of class n + Show also that involutions as «ple and IVple tangents respectively. A. Q collinear with a given point ' ? [OP and OQ have a answer is 2 + 2=4. E. The two involutions f{x) of degree n and F(y) N given by the equations = &'* {y). E A [The pencils DP and EQ have a 2 : 4 correspondence. fixed conic passes through C. Therefore the answer is 1+2 = 3. How many solutions are there to the problem Draw a circle through two given points to meet two given lines in points P. N . Any cubic through A. D. C. G. Find the locus of the intersection of the tangent at A and the line IP. A common tangent of the two conies meets this tangent at C in Q. Find the locus of the intersection of EP and FQ. D A and B P and Q [The pencils CPand DQ have a 2 :2 correspondence.] ' : Ex.XXI 2 CORRESPONDENCE OF TWO RANGES 375 Ex. If x is the distance of a point on a given line from a fixed origin and f(x). 1 the Ex. and meets the tangent at C to the fixed conic in P. A. E. A. B meeting a given conic at P and a given ? line at Q so that PQ meets the conic again at a given point ' [OP and OQ have a 4 the case in which The answer 2 correspondence. Find the locus of the intersection of DPa. are fixed points Tangents are drawn from to any conic of a given confocal family to meet a given line in respectively. D. if k and k' are connected by a relation of the form akk' + bk + ck' + d = 0. meets a fixed line at P. 15. are fixed points. a corresponding result holds for pencils. E. B. C. B.] 2 : 2 correspondence. 16. the group of n points given by f(x) — k<p(x) is said to trace out'an involutionrange of Show degree n as k varies. the given line passes through A. N . B. A.
Suppose Q. (p 2 '»)• Show that. If and classes m.) m : . Now by a series of quadratic transformations we may transform a given curve successively into curves with less and less complex singularities. AQ. The are found by eliminating k from intersections of the locus with y = and /2 (a:. IV. k) = 0. till we obtain a curve with no singularities other than ordinary multiple points with distinct tangents. § 8.376 COKRESPONDENCE OF TWO RANGES XXI . may use our knowledge of the locus of the intersection of corresponding rays of two pencils to establish the theorem : We If the points of two curves have «1:1 correspondence. R and in PThe locus of is an algebraic eliminate the coordinates of Q and from the rational equations connecting them. 2 Ex. 0. while the coordinates of P' can be expressed rationally in terms of P" and vice versa. k) = § 3. p x and p 2 in k. y. then the curves have the same deficiency. characteristic n x and n 2 . guch that the coordinates of P can be expressed rationally in terms of P' and vice versa. y' may be expressed rationally in terms of x. the curves have the same deficiency. IX. Zj) families have a 1 1 correspondence. fc) = 0. the locus of the intersection of corresponding curves is in general of degree n l p 1 + n l p 1 and of class mjZj + t l 1 . Now we have proved (Ch. thus getting an algebraic eliminant which is the equation let . if to each point P with Cartesian coordinates (x. y and vice versa. Hence it suffices to prove the theorem for two curves with ordinary multiple * points with distinct tangents only. then the coordinates of P can be expressed rationally in terms of P" and vice versa. Two families of curves have respectively degree. k) = where/! and/2 are of degrees » x and n 2 in x and y. (x. It will suffice therefore to prove the result for any two curves into which the given curves may be transformed by a series : of quadratic transformations. This means that. §§ 1. It is at once seen that. P'. m [Corresponding curves have equations and /z (x. /x (x. class. B. m^. 0. Take two fixed vertices A. M. from the equations of the given curves.] /. Bit meet P curve for we may R BR . and from the equations of the lines and AQ. if the lt (p 1 . (See Ch. P" are taken on three curves. Curves with a OnetoOne Correspondence. if points P. y. 8) that two curves derived from each other by quadratic transformation have a 1 1 correspondence and have also the same deficiency. 17. corresponding points on two such curves of degrees n. y) on one curve corresponds a point P' with coordinates* (x\ y') on the other so that x'.
For two of the lines corresponding to such a position of coincide. xvi (1888). 132.y) = 0. Ex. Similarly for y. Hence and have an correspondence. we express x rationally in terms of f. the two pencils of planes in two pencils of lines with a p q correspondence. y) of a given curve f(x. we pointed out that the coordinates (f.) of a point on the evolute corresponding to a point (x. making use oif(x. A curve and its reciprocal have the same deficiency. [See Ch. so we may have two pencils of planes with p q correspondence. 3. 1). and q are t t a* + Steinerian. M— (Ch. § 2. : : . and have therefore the same deficiency. de la Soc.y from these equations. Hence the class of the locus of Similarly it is 2n + M. §3. 77 by eliminating. Ex. de France.VIII. the establishes a 1 1 correspondence [The transformation x = £ 9 y = between the given curve and xPiyp +l = Q which is nonsingular of degree p. ?j. Hence a curve and its evolute have a 1 1 correspondence. For example. Conversely. VIII. tf+\ = Q is \{pl){p2). Bull. § 1 2. same deficiency. . For otherwise (£.] § 4. : ^ p. is Therefore in—2n + 2 and 2iV+2 are equal. Jamet. Ex. and Cayleyan of a curve have the relatively prime integers. : Ex. The tangents from to the locus of are the lY tangents at each counted twice. ?. : Just as in § 1 we had two pencils of lines with p q correspondence. or it may be proved independently as in §1. together with the lines through touching the locus of Q. with and B as ifple and %ple points. XI. y) = are expressible rationally in terms of x and y by means of equations (i) of Ch. Math. To eacli position of correspond n positions of BR. when we solve for x in terms of £. Correspondence in Three Dimensions. the planes of each pencil passing through a given line called the axis of the pencil. and the locus of is an (n + iV)ic. (vi).y) = 0. See V. If p deficiency of p being positive. 1.] The Hessian. But these are twice the deficiencies of the curves since they have no cusps AQ BR AQ BR P N:n A A P A m A BR AQ P 2N+m. Corresponding planes meet on a ruled surface of degree p + q having the axes of the pencil as pple and gple lines respecThis is evident from the fact that any plane meets tively. which is not in general the case. and to each position of correspond impositions of AQ. 77) would be the centre of curvature at more than one point of f(x.XXI 4 CORRESPONDENCE IN THREE DIMENSIONS 377 of the locus of P.
and let any plane II meet the generators through in A and B. But these tangent planes are the planes joining I to the generators through the intersections of I with 2.'. one on each of two given twisted curves. Reciprocating. The (p + q)ie of § 1 may be employed to study algebraic curves on a conicoid. these two generators not being counted as part of the curve of inter. If the coordinates of two points P and P. and a generator of the same family as OB as an (n— £>)ple line. 1 is well known. Suppose that the curve on j we are investigating is the intersection oi' j with a surface of degree n having a generator of the same family as OA as an (n — <7)ple line. P Curves on a Coniooid. the line joiniDg corresponding points of two ranges on nonintersecting lines with a p:q correspondence generates a ruled surface 2 such that p + q tangent planes can be drawn to 2 through any line I. and the p + q seltcorresponding planes give the generators of the surface which meet I. P and spondence.* Let be a point of the conicoid. the whole or partial intersection of the conicoid with any algebraic surface. which does not pass through 0.4 378 CORRESPONDENCE IN THREE DIMENSIONS XXI If the axes of the pencil intersect at 0. for . that is. then the line traces out an algebraic ruled surface of degree p + q. Then OA meets the curve in p points all projecting into the point A. * Every twisted curve of the third or fourth degree is of this nature a conicoid through nine points of the curve must contain the curve.] § 5. Moreover. . and OB meets the curve in q points all projecting into B.. Similarly the other family of generators project into straight lines through B. Consider the projection from on to IT of any curve on j. Any generator of the same family as OA . section. are connected by rational relations so that p points P correspond to each position of and q points to each position of P. p:q corre[The planes joining any given line I to. PP P P have a. the reciprocation shows that the lines on which the ranges lie are gple and pple lines of 2. The case p q = = Ex. All generators of one family on j intersect OA they therefore project into straight lines through A. so that these intersections are p + q in number. We see then that the line joining corresponding points of two ranges on nonintersecting lines with a p q correspondence generates a ruled surface of degree p + q with the given lines : as gple and pip\e lines respectively. the ruled surface becomes a cone with vertex 0.
we obtain a (p + q — l)ic with J. through 0. = = V OA OB P VP P VP A B AB V A B AB A B envelope. and the tangent at P tangent at to the projection lies in the osculating plane of the curve at 0.e. The harmonic envelope is the envelope of a line divided harmonically by the conies.XXI 5 CURVES ON A CONICOID 379 meets the curve in p points. the into the tangents at pole of the plane of a conic onj projects into the pole of with respect to the projected conic. Hence the projection of the curve is a (p + q)ic with . if the planes of two conies onj are conjugate. If OA and OB meet the curve respectively in p — 1 and q — 1 points other than 0. Hence the lines and VQ project planes at and to the projected conic.* We have so far supposed that the point does not lie on the does lie on this curve. This is clear on putting^ 1.A as £>ple point and B as (/pie point. q Let be the pole of the plane of a conic on j. in other words. and the harmonic locus is the locus of a point from which a harmonic pencil of tangents can be drawn to the conies. the projections are two conies with degenerate harmonic locus and and through A A B . vice versa. If we project on to the plane II. the generators given twisted curve. as (p — l)ple point and B as (q — l)ple again at a point P on the The projection meets point. If the plane of any other conic on j passes through V. which meets and in and Q. Each inflexion of the plane curve is the projection of a point of the twisted curve at which the osculating plane passes . or otherwise. Projecting we see that. Then and VQ are the tangent and Q to j. The tangents at A to the projection lie in the planes through A touching the twisted curve where it meets OA and so for B. conic on j projects into a conic through and and conversely. the polar of with respect to it is the line of intersection of its plane with the polar plane of V. when the conies become orthogonal circles. to the given twisted curve. . A twisted curve on j meeting all generators of one family in points and all generators of the other family in q points AB p * This is obvious on projecting A and B to the circular points. and projects into a line meeting the projection of the curve q times at B and p times elsewhere. Similarly for a generator of the other family. Every property of the plane curve gives a property of the twisted curve on projecting back on to the conicoid j and . i. their projections for and such that the pole of are two conies through one conic is the pole of their other common chord for the second conic .
Q. For since a twisted cubic meets any conicoid not containing it in six points. 9. 3. 5. [Projecting from we huve Ch. Projecting from A.]^ Ex. XIII. otherwise of type 3 + 1. With any point as vertex three cones of the secon°ree can be drawn having sixpoint contact with a given twisted cubic. The points of contact of the three osculating planes of a twisted cubic which pass through any point lie on a plane through 0. The tangent planes to the cone at the points of contact pass through the points of contact of the osculating planes through 0.second degree. may Projecting from [The cubic lies on a conicoid through 0. 5 the pencil is harmonic. Ex. Ex. we have the fact that the pencils of tangent planes through and have the same Similarly project from C for the chords CA and CD. and the crossratio of the pencil of planes is constant. it is of type 2 + 2 . Similarly a twisted quartic lies on the conicoid through any nine points of the curve. 8. the pencils AB{PQRS) and AC(PQRS) have the same crossratio. Any twisted cubic curve is the partial intersection of two conicoids with a common generator and is of the type 2 + 1. [Project from 0. [Project from a point on the line. The planes joining four given points P. If in Ex. Nine osculating planes of a curve of type 2 + 2 pass through a given point on the curve. we see that [Let AB and CD be two chords. 4. Four tangent lines of a given twisted cubic meet any given line in space.380 CURVES ON A CONICOID XXI 5 be called a curve of type p + q on the conicoid. § 4. it must lie wholly on all conicoids through any eight points of the curve. have The three inflexions of a plane nodal cubic are collinear. § 4.] Ex. XV11I. S of a twisted Projecting from A. The tangent planes through any point to a curve of type 2 + 2 at the points where it meets the generators through of a conicoid containing the curve meet the curve again in four coplanar points. [See Ch. 7. the tangent at osculating plane at in a point on the plane RQS. P meets the R [Project from R. R. 3 + 1 are well known. Ex. Through the line joining a fixed point to any point of a twisted cubic two tangent planes are drawn to the curve touching at Q and R.] crossratio. [Let and CD be two chords. The plane through and any two of the points of contact passes through a third point of contact. Similarly project from C for the pencils joining CA and CD to P. 2. AB AB AC Ex. S.'] : ' we Ex. R. 6. 2 + 2. 13 (iii). Ex. l.] Ex.] Ex. Through any variable chord of a curve of type 2 + 2 four tangent planes can be drawn. Q.] Ex.J . cubic to any chord form a pencil of constant crossratio. If it lies on a second conicoid. Show that the plane OQR envelops a cone of the. The curves of type 2+1. 1.
The properties of conies on j proved in § 5 now become the wellknown theorems that a cirqje on j and the pole of its plane project into a circle and its centre. If we take n as the diametral plane parallel to the tangent plane . m m . The a curve on j. it has S nodes and k cusps curve is of degree n and class besides a ^Tiple point at each circular point. The line of striction of one family of generators of a hyperboloid is a curve of type 3 + 1. r great circles are bitangent to the Then the projected curve. If the conicoid j of § 5 is a sphere. 10. and that two circles onj whose planes are conjugate project into orthogonal circles.] Ex. all great circles project into circles orthogonal to a fixed (unreal) circle which OB A B the projection of the circle at infinity. The generators OA. while it has t bitangent circles and i osculating circles which are orthogonal . § 6. the projection becomes the wellknown stereographic projection. become the circular lines through in the tangent plane at 0. Suppose a spherical curve df even degree n has 8 nodes is We F F F great circles can be drawn through any and k cusps while point to touch the curve. generators through Hence the stereographic projections of the foci of a spherical curve are the foci of the projected curve. 11. These results also follow from the fact that a spherical curve are inverses of each and its stereographic projection from other with respect to 0. Show that the plane OQR envelops a cone of the second degree. Ex. and Pis any other point on the curve. project into lines through <o and to'. while and become the circular points a> and a/ in IT./ touch If the generators through a point may be called a focus of the curve. on the sphere. is a fixed point on a curve of type 3 + 1. 12. . we must have p = q when the curve onj is real and the degree of the curve is even. at 0. § 8. Ex. XVII. to any fixed circle. The tangent planes through OP to the curve touch at Q and R. Curves on a Sphere. and l great circles osculate it.XXI 6 CURVES ON A SPHERE 1 381 by projecting Ex. In particular. The three points of contact of the osculating planes of a curve of type 3 + 1 which pass through a point of the curve lie on a plane through 0. Obtain properties of a twisted curve of type 3 + from any point of the conicoid containing the curve. may show similarly that arjgles are unaltered by stereographic projection. 13. [See Ch.
B. Ex. 5. C. § 11. A chord PQ of a circular cubic subtends a right angle at a fixed point of the curve. See also Ch. C. 3. Ex. if the line traces out a locus on the sphere. (iv) A selfinverse spherical 2reic is the intersection of the sphere with a cone of degree n. whence we get (ii) and (iii). 2 2<53k. A AB [The tangentarcs to a spheroconic from a point P make equal angles Project from the point diametrically with the focal distances of P. If are inverse with respect to the circle. BPD. Show Ex. so pole of the plane of. Show that foci are concyclic : may be bicircular quartic has real concyclic foci A. Ex. AQC. passes through the respect to a circle on the sphere. 2. general a spherical 2raic of the same type (with the same Pliicker's numbers). See also Ch. Obtain properties of the spherical 4ic from those of tb^e plane bicircular 4ic.] Ex.the circle. .382 CURVES ON A SPHERE results of Ch. Two bicircular quartics with the same four real concyclic foci cut orthogonally. D. P and Q are two points on a bicircular quartic with real concyclic foci A. does P.] Ex. 5. Ex. It follows that the stereographic projection of a circle and two inverse curves on the sphere is a circle and two inverse curves . (ii) The angle between two spherical curves is equal to the angle between their inverses with respect (iii) The inverse of a spherical 2mic is in to any circle of the sphere. opposite to P. § 6. § 3. Ex. Show that the circle through PQ bisecting the circumference of a fixed circle with centre passes through two fixed points. ' P ' PP P P ' ' [For (i) use the fact that the circle is cut orthogonally by any circle through the points. n = m (m — 1) — 2t — 3i. as Pliicker's equations for a spherical may be regarded on a sphere are said to be inverse with Ex. BQD are all touched by the same two circles. properties of a plane bicircular quartic whose real obtained from the properties of a conic as follows Project the conic on to a sphere from the centre. 1. second degree. Ex. 7. Two points P. [Two confocal spheroconics cut orthogonally. that the bitangents to the curve from the intersection of and CD make equal angles with OAB. i = §n(n2)68$K. 6. D. 6.] The foci of a spherical 4ic lie by fours on four circles with respect to which the 4ic is selfinverse. 8. Show that the four circles APC. XVIII. XI. Then project the spherical curve thus obtained (a spheroconic) stereographically into a bicircular quartic. XXI we get 6 Looking at the m = f™ These curve. B. 4. and the loci of jP and Show that (i) the stereographic projection of a circle and two inverse points is a circle and two inverse points. The 4ic is the intersection of the sphere with four cones of the Ex. XVIII. OCD. 1.
and that their points of contact lie on a circle through O. section is a circle. The locus of a point on the Earth's surface at which a given place has a given azimuth is a spherical quartic. (sin and foci S. 16.. The spherical quartic has a node point on the curve. Q in . cosec \ OS') cosec \ OP is is constant. 4 applicable to any curve which is selfinverse with respect to a circle. sin 1ST' . VII. cosec  OS± sin ±S'P . find the locus of P.. Through a given point If drawn meeting a given spherical 2»ic in Q lt Q2 .XXI 6 CURVES ON A SPHERE 383 Ex. [Project stereographically from the point diametrically opposite to and use Ch. pp. . sinO02 sin \OQr . Show that. Show that the method of Ex. 1873).. 15. cosec \ PQi . S'. if P is any Ex. [Projecting stereographically from collinear inflexions. the plus or minus sign being taken according as acnode or crunode. %SP .] 2n C. A circle spherical quartic has a node touching the curve is met again by the orthogonal circles Show that Y and Y' lie on a fixed and 8. S'. in Darboux's [The reader will find a very interesting discussion of spherical quartics Sur une classe remarquable de courbes et de surfaces alg4briques (Paris. cosec \OY cosecOr . [Projecting stereographically from is a hyperbola.'] we have is : 'A cubic has three real Ex. 13. cosec P§2 cosec \PQ r summation extending over § 1.. Show that the two properties are obtainable from each other by ordinary inversion. Two properties of a plane curve are derived by projecting a given spherical curve stereographically from two different points of the sphere. is constant. 17. drawn through the node The locus of their second of a inter Ex. Y' A and that sinJSr. Ex. The locus of their second intersection is a spherical quartic with a node at 0. 18. Two circles are drilwn through the cusp of a spherical quartic touching the curve and cutting each other at a constant angle. and loci S. A spherical sextic has a triple point 0. of a sphere any great circle OP is Ex. and S' at Y.] Ex. terms). O = 2sinJ0& (the . Two orthogonal circles are spherical quartic touching the curve. a parabola Ex. 9. 160. Two spherical quartics with a common node and foci cut orthogonally. 12. . 11. Ex.'] we have ' : The isoptic locus of through through circle. 14. 10. Show that three real circles of curvature of the sextic pass through 0. an [Projecting stereographically from we project the curve into a conic] Ex.
.
296. Canonical (contd. Anallagmatic. nonsingular. 127. 135. 288. 229. 175. 297. Ceratoid cusp. 217. 55. 341. 1. 64. 333. 326. 299. 111. 92. 330. Coordinates. 71. 218.: : : ) INDEX (The numbers refer to the pages. 304. 21. 327. 295. tacnodal. 107. 171. Carnot's theorem. Asymptote. Axis of pencil of planes. 337.83. Coincidence point. 8. 323. 75. Class. 7. for double point. 351. 206. 120. 1. 312. Cardioid. 238. 363. 190. 30. 233. 377. 207. 332. 53. 285. quartic. 325. 248. 339. 177. lines and points. 228. AByzygetic bitangents.) equation of quartic binodal. 9. Biflecnodal quartic. Cissoid. curve or oval. 283. 328. 328. 351. 236. 12. Axial direction. 313. 62. 66. with triple point. . 285. 305. 185. 207. 25. 351. 266. 320. 288. 291. 21. 261. 236. Areal coordinates. 270. 296. 329. Cassinian curve or oval. &c. 25. 335. Bay of quartic. 228. 63. Carbonpoint curve. 252. of reciprocation. 75. 376. 69. 334. tricuspidal. Anautotomic curve. Basepoint of pencil. 322. 58. 256. Conchoid of Nicomedes. 8. 31. Canonical equation of cubic. Conic of closest contact. 363. 91. . 284. Characteristic. 87. 63. 290. 299. Conditions determining a curve. 261. 5. 366. 34. 9. Algebraic curve. 316. 128. 2. 304. Centre of curve. Bipartite. unicursal with one distinct double point. Baseconic of quadratic transformation. 203. Complextriple. 364. 21. 269. 331. 51. 323. Cayleyan. Confocal curves. 255. Apolar triangle of cubic. 333. 337. 36. flecnodal. 326. Analysis by quadratic transformation. Circle of curvature. with rhamphoid cusp. 125. Ast. of perspective. 61. Cartesian coordinates. 331. 262. 31. 294. Centre of curvature. 192. Cn . 364. Bicircular curve. Circular cubic. 161. Biflecnode. Complex of bitangents. 382. Bitangents of quartic. 33. Conchoid. Acnodal cubic. 283. 177. Acnode. curve. 351. Branch. 143. Adjoined curve. 37. Bitangent. 23. 319. 215. Aronhold Seven. 26. trinodal. 358. 22. 257. Circuit. &c. Canonical equation of quartic bicircular. biflecnodal. 179. 215. 289.
Klein's theorem. 256. 3. Director Ideal bitangent. Harmonic 237. 244. Intersections of curves. 74. 235. of tangents to a cubic. 138. 369. 116. Inversion. 260. Lame' curve. 372. Cubic curve. 98. 85. 285. Hessian. 286. 299. 164. equation. Deficiency. 352. Hart triangles of a cubic. tangent. Fourcusped hypocycloid. 127. 226. 220. envelope and locus. 127. 68. point. 181. 37. Crunodal cubic. 366. 241. 108. Even circuit. Hourglasscurve. 61. 22. 45. 305. i 54. 101. 363. 63. Critic centres of cubic. 96. Double cusp. 324. 61. 63. Inflexion. 188. 3. 209. 86. 95. 129. 173. of inverse curve. Conjugate point. 109. 63. Doublesix. 304. 95. quartic. 130. 302. of pedal. 74. 16. 374. 115. Evolute. Hilbert's theorem. 257. 379. 144. 262. 70. 208. pencil and range. 10. 169. 222. Invariants of cubic and quartic equations. 236. parabola of circular cubic. Cusp. pencil and range. 375. Curvatures of touching curves. Homogeneous coordinates. Latent singularity. . 22. Limacon. 227. 8. Index of circuit. 235. 306. 348. 168. 245. Flecnodes. of quartic. 19. 52. Degeneracy. 85.: : . 8. 373. of circular cubic. 305. 228. 7. 78. of cubic. Focus of bicircular quartic. 323. 164. Directrix. . 179. 9 Deferent conic. 17. 378. 3. 229. 89. 113. Focus. 74. Inflexional tangent. 83. 182. 164. 64. 223. 229. Coresidual groups of points. 245. 113.24. Bquianharmonic cubic. 386 INDEX Focus (contd. 22. Isoptic locus. Curvetracing. 54.. 248. 262. Focal conic of bicircular quartic. 220. Higher plane curve. &c. 63. 204. points of involution. 361. 118. Crunode. Genus. surface. Isolated point. 76. 215. cubic. Jacobian. 63. 241. singularity. &c. of spherical curve. Involution. Expansion near origin. 220. 352. 139. 237. 62. 124. 2. 64. 108. 22. Hesse's notation forbitangents. 14. 229. 300. 325. 9. 300. 224. of evolute. 376. Crosscurve. 169. 77. 285. 262. 165. 164. 81. 129. Degree of curve. 197.) Conicoid. Crossratio. 136. 55. 16. 308. Correspondence of pencjjs and ranges. 207. 215. 345. Cutting of a crunode. 192 356 of cubic. 161. 359. 253. points on cubic. 209. 235. 222. 9. 341 theorem. of curve and adjoined curte. 356. 21. parabola. 88. 254. 171. Harnack's theorem. 235. 381. 63. 258. 86. 28. 138. 69. 352. 3. Cuspidal cubic. 123. see Contents ' '. 238. First polar curve. 227. 19. 165. 24. polar of inflexion of cubic. perspective. 164. 361. 21. 204. Lemniscate. circle. 174.
88. nic. curves. Quartic curve. 364. 363. 252. reciprocation. 296. 368. Menelaus's theorem. Order of Ruled surface. 245. 137. numbers for spherical curve. 355. 352. Pencil of curves. Septimic curve. see ' Rhamphoid ' Nodal quartic. pair on a cubic. Singular focus. Rhamphoid cusp. 65. 102. 337. Sort of bitangent. 116. 382. '. 3. raic equation. Multiple point. Spherical curve. 33. 104.3)ple point. 2. 135. 33. Ruler construction. 240. 193. 127. Quintic curve. Pointeven and pointodd circuits. 126. Loop of nodal cubic. 5. branches touching (see 'Taenode '). of orthoptic locus. 107. 9. 79. &c. 129. 348. 259. 230. 65. 74. Residual groups. 259. 51. 352. Spheroconic. Orthoptic locus. 23. 264. 33. 183. 12. Linesingularity. 36. 23. 179. 298. with (n — l)ple point. Spinode. Rational curve. 181. 168. 31. 25. 112. Symmetrical bicircular quartic. 88. 189. 173. 21. 377. Schlafli's doublesix. Species of cusp. coordinates. 224. with (« . 57. ' Negative pedal. 76. circular cubic. see Unicursal Reciprocal of circuit. Pippian. 357. . Standard equation. 261. Stereographic projection. 260. 134. Node. 9. Oval. Reciprocation. Partial superlinear branch. Parallel asymptotes. Symmetry of cubics. 137. Poloconic. 216. 381. 79. Steinerian. see Contents '. 190. Pliicker's numbers. 351. 209. 107. Pascal's theorem. 76. of parallel curve. 35. 257. 196. 51. Perspective. Mobius's theorem. 229. Linear branch. Pointequation. 215. Point of contact. 65. 58. 88. 62. 22. Sextic curve. Quadruple tangent. of superlinear branch. 87. 16. 144. 13.' INDEX Line at infinity. curve. Points at infinity. 115. Pedal curve. 52. 120. circuit. 71. Osculation. 191. 382. Newton's diagram. Superlinear branch. Quadratic transformation. 73. 29. of inflexions of cubic. 22. 88. 38. 381. 367. Ramphoid. 366. 367. 21. 123. of multiple point. 59. 355. Oscnodal quartic. 336. Oscnode. 86. Sextactic point. 25. on cubic surface. 201. 86. 63. 169. 98. Opposite point. Reversion of series. Pointsingularity. 366. of quadratic transformation. 4. 76. 346. 135. 120. 25. 16. 104. 53. Pole. 169. 387 Polar conic. Nonsingular curve. Radial. see 'Canonical equation '. 225. of pedal curve. 309. 249. Nest of ovals. 188. 80. Steiner complex. 76. 125. Radius of curvature. 210. 211. 375. 62. 53. Projection. Odd circuit. line. 166. Parameter. equation. Projective involution. 215.
Unipartite. 3. 358. 335. Triangle of reference. 264. 23. Theory of equations. 263. 86. Tangent. 13. 201. 283. Triangularsymmetric 102. 62. 289. 113. 64. Type of curve. 146. 24. PRINTED IN ENGLAND AT THE OXFORD UNIVERSITY PRESS . 266. 380. of point on cubic. Tacnodal quartio. 351. Tacnode. 328. Tangential equation. 137. of quartic. 91. 355. 77. curve. 63. 16. 163. 351. of projection. &c. cir Undulation. 215. 108. 290. Tangenteven and tangentodd cuits. 264. 333. 258. 20. of perspective. 188. pencil. 5. 59. 201. 161. Vanishing line. 57. 157. 240. 20. Threecusped hypocycloid. 211. 2. 4.388 Syzygetic bitangents. 64. 45. tangent. Unicursal curve. »*point contact. 62. 378. 52. Tangent of 92. Zug. cubics. 237. of twisted curve. &c. 205. Vertex of curve. 59. Twisted curves. INDEX Triple point. Weierstrass's function.
This action might not be possible to undo. Are you sure you want to continue?
We've moved you to where you read on your other device.
Get the full title to continue reading from where you left off, or restart the preview.