This action might not be possible to undo. Are you sure you want to continue?
THE BOOK WAS
DRENCHED
CO
>
164418
00
OSMANIA UNIVERSITY LIBRARY
Call No.
4~/3
.
g/
??f
.
Accession No,
2
67.2
Author
Title
This book should be returned on or before the date last mrked below,
D. HARVARD UNIVERSITY OXFORD AT THE CLARENDON PRESS 1909 .THE ELEMENTS OF NONEUCLIDEAN GEOMETRY BY JULIAN LOWELL COOLIDGE ASSISTANT PROFESSOR OF MATHEMATICS IN PH.
EDINBURGH.A. LONDON. M. PUBLISHER TO THE UNIVERSITY OF OXFORD NEW YORK TORONTO AND MELBOURNE .HENRY FROWDE.
such the analysis situs. quite as well established as any strange. or. * THE and the new subject appeared as a dangerous lapse from the orthodox doctrine of Euclid. as in all other changes. there is subject both for rejoicing and regret. Liebmann.* who * Detailed references given later. at least. any more need than any other adds to our supply of books. it solid basis than . without feeling any need to justify himself. It is a satisfaction to a writer on noneuclidean geometry that he may proceed at once to his subject. In this.PREFACE heroic age of noneuclidean geometry is passed. he will miss the stimulus that comes to one who feels that he is bringing out something entirely new and who subject of noneuclidean geometry is. or and. Recent books dealing with noneuclidean geometry In the one we find fall naturally into two classes. to the mathematician. The other branch of mathematical science may some branches. and those who do undertake it. A2 . the works of Killing. The attempt to prove the parallel axiom by means of the other usual assumptions is now seldom undertaken. are considered in the class with sad circlesquarers and searchers for perpetual motion byproducts of the creative activity of modern science. and Manning. On the other hand. lay claim to a decidedly more as the theory of assemblages. It is long since the days when Lobatchewsky timidly referred to his system as an imaginary geometry'. in fact.
and nothing that all of we make ' A The author would have preferred. and the principal noiieuclideaii systems in the way that he wished. and the authors of a goodly number of articles on the foundations of geometry. not because it is one whit less important than the first. These writers deal at length with the consistency. had he been familiar with any that seemed to him suitable to establish simultaneously the euclidean else. The present work is. . The system of axioms here used is decidedly more cumbersome than some others. In the other category are Hilbert.4 PREFACE wish to build up certain clearly conceived geometrical systems. and the theory of transformation groups. lies mainly in the fruits. This method is developed in the openand is the most obvious. subject like ours must be built up from explicitly stated assumptions. rather than in the roots but the day is past when the matter of axioms may be dismissed with the remark it . in the first chapters. in a measure. Vahlen. be stated at the outset. but do not go very far towards raising a superstructure on any one of the foundations suggested. to start from some system of axioms already published. an attempt to unite the two tendencies. Euclid's assumptions except the one about parallels'. Veronese. (2) Projcctive ing chapters geometry. The author's own interest. but and distance. There are three natural approaches to noneuclidean geometry. but leads to the desired goal. significance. (1) The elementary geometry of point. and are careless of the details of the foundations on which all is to rest. line. and logical independence of their assumptions. This method is not taken up until Chapter XVIII.
has a more lasting geometrical importance than a nondesarguian plane that cannot form part of a threedimensional space. to what may be called the himself. at any rate in a first study of the subject. the gain in generality obtained by studying the geometry of ^dimensions is more than offset by the loss of Secondly. XIX. more any closely allied to the euclidean system than are others. old. and space constant. reader. may find naturally. may and XVI. he has confined almost exclusively. evident that a system which gives a simple and clear interpretation of ternary and qua ternary orthogonal substitutions. noneuclidean systems. one whose points are determined by numerical values in a nonarchimedian number system. it wiser at the first reading to omit Chapters X. cepts This method is explained in the last chapter. On the other hand. with the conbeginning. The author has imposed upon himself one or two very definite limitations. of distanceelement. let us say. new to the subject. and have by far the most historical imporIt is also tance. a reader already somewhat familiar with noneuclidean geometry. This is because of his feeling that. find his greatest interest in Chapters which contain the substance of a number of X . he has not gone beyond three dimensions. in the present work is. XVIII.PREFACE because it 5 seemed better not course of the narrative by to interrupt the natural interpolating an alternative (3) Differential geometry. extremal. XV. a noneuclidean plane which may be interpreted as a surface of constant total curvature. has a totally different sort of mathematical significance from. To begin with. Or again. and XIX. These are much classical ' ' clearness and naturalness. The majority of material A XVI.
we say abetted?) him in the . Professor Eduard Study of Bonn and Professor Corrado Segre of Turin. and more comprehensible than that of Bianchi. He would also offer affectionate thanks to his former teachers. which are not. who has XV and XIX. of Chapter XIV which contains a number of neat formulae relative to areas and volumes published many years ago by Professor d'Ovidio. and all others who have aided and colleague. AssistantProfessor read in manuscript Chapters encouraged (or shall present work. hoped that the introduction to noneuclidean geometry in Chapter XV may prove to be more comprehensive than that of Darboux.6 PREFACE recent papers on the extraordinary line geometry of Mention may also be made rioneuclidean space. and Chapter XIII. where Staude's string construction of the ellipsoid is extended to noneuclidean space. very familiar to Englishspeaking readers. perhaps. The author takes this opportunity to thank his It is differential Whittemore.
. their .. Nature of the congruent group Definition of dihedral angles.. . 21) 30 32 Comparison of triangles Side of a triangle not greater than sum of other two Comparison and measurement of angles . . 35 37 38 3d CHAPTER IH THE THREE HYPOTHESES A variable angle is a continuous function of a variable distance Saecheri's theorem for isosceles birectangular quadrilaterals ... 51 54 55 57 .. their properties .17 CHAPTER Axiom of continuity Division of distances II CONGRUENT TRANSFORMATIONS 23 23 26 Measure of distance Axiom of congruent transformations Definition of angles. .... . . . . .. their properties .TABLE OF CONTENTS CHAPTER I FOUNDATION FOR METRICAL GEOMETRY IN A LIMITED REGION PAGE Fundamental assumptions and definitions Sums and differences of distances Serial arrangement of points on a line Simple descriptive properties of plane and space IS 14 15 . ... hyperbolic.. . categorical nature Definition of the euclidean.. 46 47 CHAPTER IV TRIGONOMETRIC FORMULAE Limit of ratio of opposite sides of diminishing isosceles quadrilateral Continuity of the resulting function Its functional equation and solution 4H 50 Functional equation for the cosine of an angle Noneuclidean form for the pythagorean theorem Trigonometric formulae for right and oblique triangles . 40 43 44 45 4fr The existence of one rectangle implies the existence of an infinite number Three assumptions as to the sum of the angles of a right triangle Three assumptions as to the sum of the angles of any triangle.. and elliptic hypotheses Geometry in the infinitesimal domain obeys the euclidean hypothesis . .
Imaginary elements geometrically defined. .8 CONTENTS CHAPTER V ANALYTIC FORMULAE PAGE Directed distances 62 Group of translations of a line 62 64 and negative directed distances Coordinates of a point on a line Positive .. 81 84 85 88 .94 . new form for the definition of distance Extension of the distance concept to the complex domain Case where a straight line gives a maximum distance . 65 67 68 .. .. Congruent transformations and orthogonal substitutions Fundamental formulae for distance and angle 69 70 CHAPTER VI CONSISTENCE AND SIGNIFICANCE OF THE AXIOMS Examples of geometries satisfying the assumptions Relative independence of the axioms .. ... made . . free mobility of the whole system One to one correspondence of point and coordinate set in euclidean and hyperbolic cases Ambiguity in the elliptic case giving rise to elliptic and spherical Euclidean and hyperbolic space Contradiction arising under the ... 77 78 78 New 80 81 geometry Ideal elements. 94 .... 89 . extension of all spaces to be real continua . but permitting the extension of every segment by a definite amount Last axiom...64 65 Coordinates of a point in a plane Finite and infinitesimal distance formulae..... extension of all spaces to be perfect continua in the complex domain .... .. line hyperbolic plane . . .91 CHAPTER VIII GROUPS OF CONGRUENT TRANSFORMATIONS Congruent transformations of the straight . .. . Cayleyan Absolute. . the noneuclidean plane as a surface of constant Gaussian curvature Equation connecting direction cosines of a line Coordinates of a point in space . 72 74 CHAPTER Possibility of extending a VII GEOMETRIC AND ANALYTIC EXTENSION OF SPACE segment by a definite amount in the 77 euclidean and hyperbolic cases elliptic hypothesis assumptions identical with the old for limited region.
AND PLANE.. ..135 138 139 Spheres Poincare's sphere to sphere transformation from euclidean to noneuclidean space .. . and centres of similitude Circles through two points. .. TREATED ANALYTICALLY . 125 Clifford angles 126 128 Chain and strip Chain congruence 129 CHAPTER XI THE CIRCLE AND THE SPHERE Simplest form for the equation of a circle Dual nature of the curve 131 131 Curvature of a circle Radical axes.. .. and resulting desmic configurations Invariant formulae for distance and angle of skew lines in line coordinates' 108 110 and parataxy in Relative moment of two directed lines Criteria for parallelism line coordinates.104 106 107 Equations of parallels to a given line Notable points of a tetrahedron. 113 114 CHAPTER X HIGHER LINEGEOMETRY Linear complex in hyperbolic space 11 ti to interpret the geometry of the 118 complex plane 119 Chain. 101 . . . . The cross. Clifford parallels. synectic and nonsynectic congruences 124 Dual coordinates of a cross in elliptic case .CONTENTS Congruent transformations of the .96 .. its coordinates The use of the cross manifold 117 Condition for parataxy . or tangent to two lines 133 134 . .121 Chain congruence. . . Notable points of a triangle in the noneuclidean plane Analoga of the theorems of Menelaus and Ceva Formulae of the parallel angle .. LINE. .. . elliptic plane euclidean plane . . . 9 PAGE 96 97 and spherical space . and chain surface 120 Hamilton's theorem . hyperbolic space elliptic . or paratactic lines 98 99 99 The groups of right and left translations Congruent transformations of euclidean space 100 CHAPTER IX POINT.
148 150 152 Normals Confocal and homothetic conies Elliptic coordinates 152 152 CHAPTER XIII QUADRIC SURFACES Classification of quadrics Central quadrics Planes of circular section and parabolic section . . focal lines. 186 186 . Conjugate and mutually perpendicular lines through a centre Confocal and homothetic quadrics Elliptic coordinates. . . 14B 144 Relations connecting distances of a point from and their duals foci. 159 160 ..174 175 Area found by integration Area of circle Area of whole elliptic or spherical plane Amplitude of a tetrahedron . . directrices. . . foci.10 CONTENTS CHAPTER XII PAGE CONIC SECTIONS Classification of conies 142 148 Equations of central conic and Absolute Centres. directrices.161 166 CHAPTER XIV AREAS AND VOLUMES Amplitude of a triangle Relation to other parts Limiting form when the triangle Deficiency and area .  . . . 170 infinitesimal is . .171 . and director points . . . .158 . . various forms of the distance element String construction for the ellipsoid .154 157 . axes.181 184 185 Reduction to a single quadrature of the problem of finding the volume of a tetrahedron Volume of a cone of revolution Volume of a sphere Volume of the whole of elliptic or of spherical space . . . &c. Conjugate and mutually perpendicular lines through a centre Auxiliary circles . .178 178 179 Relation to other parts Simple form for the differential of volume of a tetrahedron . .176 178 . . .
.CONTENTS CHAPTER XV INTRODUCTION TO DIFFERENTIAL GEOMETRY Curvature of a space or plane curve Analoga of direction cosines of tangent..195 196 197 Dupin's theorem for triply orthogonal systems Curvature of a curve on a surface .... . principal normal. . and 11 . 202 203 204 205 206 208 210 Ruled surfaces of zero Gaussian curvature in space Geodesic curvature and geodesic lines Necessary conditions for a minimal surface elliptic or spherical Integration of the resulting differential equations .199 201 Dupin's indicatrix Torsion of asymptotic lines Total relative curvature.. PAGE 187 bmormal 189 Frenet's formulae for the noneuclidean case 190 191 Sign of the torsion Evolutes of a space curve Two fundamental quadratic differential forms for a surface Conditions for mutually conjugate or perpendicular tangents Lines of curvature 192 . . its relation to Gaussian curvature Surfaces of zero relative curvature Surfaces of zero Gaussian curvature . and congruences of normals to surfaces of .194 . . 225 227 Representation of normal congruence Isotropic congruence represented complex variable 228 by an arbitrary function of the 229 232 Special examples of this representation Study's ray to ray transformation which interchanges parallelism and parataxy Resulting interchange among the three special types of congruence 23& 235 .212 CHAPTER XVI DIFFERENTIAL LINEGEOMETRY Analoga of Kurnmer's coefficients Their fundamental relations Limiting points and focal points Necessary and sufficient conditions for a normal congruence MalusDupin theorem Isotropic congruences.. .. 215 216 217 222 225 zero curvature Spherical representation of rays in elliptic space .. . .
287 287 239 240 248 245 245 . integration of equations of geodesic Reappearance of familiar distance formulae . their differential equations Determination of a geodesic by a point and direction cosines of tangent thereat Determination of a geodesic by two near points 275 275 277 278 278 279 279 280 28 1 284 284 285 287 . .. .. CHAPTER XVIII PROJECTIVE BASIS OF NONEUCLIDEAN GEOMETRY Fundamental notions Axioms of connexion and separation Protective geometry of the plane Projective geometry of space .. .. Axiom Definition of angle of congruent transformations Simplified expression for distance element Constant curvature of geodesic surfaces Introduction of new coordinates Recapitulation . 247 247 249 Projective scale and cross ratios Projective coordinates of points in a line 250 255 261 Linear transformations of the line Projective coordinates of points in a plane Equation of a line.. various types of line in them Hyperbolic case little known relation to automorphic functions Nonexistence of multiply connected elliptic planes Multiply connected elliptic spaces ...... INDEX ..... .12 CONTENTS CHAPTER XVII MULTIPLY CONNECTED SPACES Repudiation of the axiom of free mobility of space as a whole Resulting possibility of one to many correspondence of points and coordinate sets Multiply connected euclidean planes Multiply connected euclidean spaces. ....... .. its coordinates Projective coordinates of points in space Equation of a plane Collineations Imaginary elements Axioms of the congment collineation group Reappearance of the Absolute and previous metrical formulae 262 262 268 264 264 265 265 268 272 CHAPTER XIX DIFFERENTIAL BASIS FOR EUCLIDEAN AND NONEUCLIDEAN GEOMETRY Fundamental assumptions Coordinate system and distance elements Geodesic curves..... ..... PAGE .
the invariable custom in works on the foundations of geometry. i. called respectively points and distances.* explicit assumptions shall be as follows : * There is no logical or mathematical reason why the point should be taken as undefined rather than the line or plane. 1891. 1904. Serie 3. 1902. Iv. Accademia di Torino. Vahlen. Moore. vol. i. as does * Padoa.' Periodico ' * A * uses line and separation. use distance. vol. the raw material with which we are to work. however.' Memorie della E. v.' Atti della R. What names we choose to attach to these objects is obviously a question quite apart from the nature of the logical connexions which arise from the various relations assumed to exist among them. Serie 2. and by a desire to make our mathematical edifice as well adapted as possible to the needs of practical life. one is obliged to introduce the sixparameter group of motions (as in Ch. xlix. as it seemed to me to give the best approach to the I cannot but feel that the choice of segment or order problem in hand. 1905. Schur. 1899. vol. With geometria. uses practically the same indefinables as Veronese. and congruence of segments. vol. the lastnamed authors. considering the weight of historical and psychological tradition in its favour. and I fondamenti della goometria metricaproiettiva. I have made the same choice as Torino. there is no such unanimity. use segment and order. Abstrakte Geometric. 19028. and. This in. vol. Accadcmia dcllc Scienze di Torino. will probably continue to stand among the fundamental indefinables. the point. Uii nuovo sistema di definizioni per la geometria euclidea. vol. on the other hand. 1902. Fondamenti di Padua.* Transactions of the American Mathematical and Veblen. uses segment and motion. liv. ' 1 ' . Fieri. segment. It is. and in choosing these names we are guided principally by tradition. in spite of the very condensed For to reach consystem of axioms which Veblen has set up therefor. Leipzig. Levy. an inelegance to base projective geometry on a nonprojective conception such as betweendi niatcmatica. of Axioms for Geometry/ sanio System Society. vol. would be a mistake for our present purpose. takes the line. Ueber die Grundlagen dcr Geometrie.CHAPTER I FOUNDATION FOR METRICAL GEOMETRY IN A LIMITED REGION IN any system of geometry we must begin by assuming the existence of certain fundamental objects. introduces motion alone. elementare come sistema ipotetico deduttivo. In the present work we shall assume the existence of two sorts Our of objects. o. Veronese.' Mathematische Annalen. Die Grundlagen der Geometric. The projective Axioms of Geometry. t regard to the others. gruence and measurement by this means. 1899. Memorie Accad. XVIII of this work). xxxviii. Serie 2. base metrical geometry on projective. La geometria basata sulle idee di pun to e di distanza. Hilbert. iii. 1904. Leipzig. Bella geometria Journal. Peano. ' 1903.
as the distance betiveen the two points. If AXIOM V. AB = AXIOM VI. two identical points shall be Definition. or from one to the other. called points. the fundamental part of the subject. If the points be A and B it will be convenient to indicate shall speak of this also their distance by or BA. whereas writers like Vahlen require both projective and 'affine' geometry. It will CH. a very roundabout way to reach what is after all. AB = AB. There exists a class of two members. : AXIOM III. AB = AB + CO. AB = CD and CD = EF. In place of the words we shall write the symbol == The following assumptions shall be made with regard to the congruent relation . If AB = CD + PQ. then AS = CD + MS. containing at least be convenient to indicate points by large Roman letters as A.14 FOUNDATION FOR METRICAL GEOMETRY AXIOM I. These might have been put into purely logical form by saying that we assumed that every distance was congruent to itself. AXIOM II. The distance of called a null distance. If AB = CD + PQ and PQ = M. AA = KB. VII. before reaching metrical geometry. If AB = CD + PQ and A'B* = AB. B. AXIOM VIII. AB We We next assume that between two distances there * may ' exist a relation expressed by saying that the one is congruent is congruent to to the other. and that the congruent relation is transitive. then AXIOM IX. This shall be written CD + PQ. objects. then AB = EF. then AB = PQ + CD. ness'. AXIOM IV. that the distances of any two pairs of identical points are congruent. The existence of any two points implies the existence of a unique object called their distance. Let us next assume that there may exist a triadic relation connecting three distances which is expressed by a saying that the first AB is congruent to the sum of the second CD and the third PQ. G. 7 .
and written (AB) or (BA). In the first place it clearly involves the existence of an infinite number of points. we have merely premised the existence of such extension. Between any two exists one. Theorem 2. In the second it removes the possibility of a maximum disIn other words. AXIOM XI. We shall make the concept of extension more explicit by the following definitions. Definition.i IN A LIMITED REGION and Definition. If AB = CD. AB < CD. AB = AB + PQ contrary to X. then AB shall be said This } is written AB > CD. possessing the segment of B and A. however. there is no distance which may not tance. For we could not have not null. of the three relations Theorem 1. be extended in either direction. VIII. If AB = CD + PQ and <77/ = CD. If not null distance a 15 AB exists is congruent to the to be greater than CD. It is. then The proof is ABCD + PQ where _ PQ was by immediate. distances AXIOM X. then CD = AB. and points all other points thereof shall be said to be within it. or of AB == AC+CB shall be called the A B Definition. fundamentally important to notice that we have made no assumption as to the magnitude of the amount by which a distance may be so extended. CD = AB f PQ for then. If A and C be any two points there exists such a point B distinct from either that This axiom is highly significant. Nor could we have = CD\ AB>CD. all The assemblage of points C property that A and B. The shall be called the extremities of the segment. A and C such that AB = AC + CB shall The assemblage of all points different from be called the extension 5 of (AC) beyond C. This is CD < AB. less AB > CD written then CD shall be said to be than AB. AB and CD there and only one. . Definition. If CD be two such distances that there PQ fulfilling the condition that AB sum of CD and PQ.
a point of (AC). CH=EF. The effect t)f this axiom is to establish a serial order among the points of a segment and its extensions. If OIL AXIOM then AB = AC+ CB where AC = AD + DC. and this will be identical with D. Theorem 7. If 9 9 which is see as a corollary. If AB = PQ + RS and A~B = PQ + W. OR = AB AXIOM XIII. then every point D of (AB) is^either_a point of (AC) or of (CB). while Theorem If AB = PQ + Rti and then The proof is left to the reader. by XII. CB < DB. D AB We D that segment. If C be a point of (AB). If AB > CD and CD > EF then AB>JEF. If AC = Ip we have C and J) identical. AB = AD + DB where DB = DC + CB. where EG = AB. points It will follow from XIII that two distinct points of a segment cannot determine congruent distances from either end also see from XII that if C be a point of (AB). itself. We shall also be able to show that our distances are scalar magnitudes. it is likewise a point of (AB). thereof. Let and the reader show further that every point of a segment. a point of 9 We against our hypothesis. Theorem If AC>AD 9 then 6. let us put E~F=EG + GF. AB =J&+CB and AB = AD + DB A^W = PQ + RS. Theorem 4. and hence. then . that if C and be_any two of (AB) one at least being within it. Then Then CDCK + KD. If AC < AD we find C as a point of (AD). 5. is. CD = CH + Hb'.16 FOUNDATION FOR METRICAL GEOMETRY XII. to this. If AC > AD wo may find a point of (AC) [and so of (A JS)] whose distance from A is congruent to AD._If then EF>A #. To begin with AB = EF is impossible. AB PQ + RS there is a single point = PQ CB = Bflf. and that addition of distances is associative. _ D is a point of (CB). as will be seen from the following theorems. > CD. whose extremities belong to a given segment. (7 of (AB) such that AC Theorem 3.
necessary in order to proceed beyond the geometry of a single straight line. F . If 4JB = P'Q' + R'S' then Definition. RS. ^B will be convenient to write and say that PQ is the difference of the distances A B and When we we are uncertain as to whether AB > RS or RS > AB. = PQ + LM and AB = AS PQ = FQ'. The line determined by two points or BA. RS < WS7 . ~ then C5 = M = LM. Two lines having in common a single point are said to cut or intersect in that point. its PQ > FQ'. however. Theorem while If AB RS\. AXIOM XVI. = PQ + RS and 4JB then Theorem while 9. The second specifies certain conditions under which two lines. Two lines having two common distinct points shall be written are identical. shall write their difference 8. AXIOM XIV. If B be a point of (CD) and E a point of (AB) where A is not a point of the line BO. Definition. make this assumption We explicitly. The assemblage of all points of a segment and extensions shall be called a line. intersect. If C be a point of the extension of (AB) beyond B and D another point of this same extension. AC = PQ. a point of (BD). A and B AB Theorem 10. Notice that we have not as yet assumed the existence of two such lines. shall soon. then D is a point of (BO) if C = TH) or EC > /) otherwise G is . not given by means of common points. neverIt is clear that some such assumption is theless. All points do not lie in one line.IN For If it if A LIMITED REGION and 17 AB = AC + CB. The first of these axioms is clearly nothing but an existence theorem. then the line DE contains a point of (AC). must. AXIOM XV.
The proof is left to the reader. both of which are inconsistent E E We D D with our data. But then. and G be identical. there an infinite number of distinct points which belong to their segment. by 13. Theorem dense. D. If A. The points originally chosen shall be called the . Lines which contain the same point shall be said to pass through it. The mainfold of all points of a segment is form the configuration of Theorem 13. If two distinct points A and B be given. are collinear. C is a point of (BL). But if // be a point of (BG). Points which belong to the same line shall be said to be on it or to be collinear. Definition. (7. Hence of (EG). and D. If A. Definition. should either that this were not the case. no three points. the point point of (DF). andD a point within (AB) while is a point of the extension of will contain a point F (BC) beyond C. If L will contain a point L of (BE). B. one within each of their three segments. or to be concurrent. C be three noncollinear points. B. belongs to (AC). while G belongs to (AL). G would be a point of (DE) and in the second would be a point of (BC). If L be within (BC). by 4. It may be interpreted otherwise by saying that there is no minimum distance. then the line E DE of (AC). Then Take (?. in the first case. If three noncollinear points be given. G be three noncollinear points. 12. If L be a point of (CE) then within (AG) as goes through required. Suppose have Fas a point of (DE) or as a point of (EF).18 FOUNDATION FOR METRICAL GEOMETRY OH. E AG EG F K H EG H F H We A F Theorem 15. and all points of the segment of the other two. G and (AC) and of (BK). Theorem 14. If A. then goes through 11 of are points of (AE) y so that. This theorem is an immediate consequence of the last two axioms. shall be called a Triangle. contains see also that it is impossible that C should belong to (AF) or Hence to (FC). either be a point of (BG) or of (GK). G will be the point required. other than the null distance. B. IB a point of (GK\ and (AH) which is untrue. a point of different from (ED). the locus of all points of all segments determined by each of these. is a points described in Axiom XVI. is Theorem 11. must then.
It is our next task to show that this specialization of function is only apparent. while the other belongs to (BO). at once from 16. The assemblage mined by the vertices of a of all points of all lines detertriangle and all points of the opposite sides shall be called a plane. Theorem 18. Any point of the triangle. and that any other three noncollinear points of the plane might equally well have been chosen to define it. B2 . or of one of its The plane determined by three points as A. and a point of the line of the other two vertices. the following points lie therein (a) All points of every line determined by a vertex. 1882. I have followed Veblen. Leipzig. it : of a will contain a point of the third side. G extensions.. their not on one of sides. and from the conpoints of a line. He in turn confesses his indebtedness to Peano. It should be noticed that in defining a plane in this manner. every other point thereof is either a point of their segment. may be summed as follows * up Theorem 16. are thus led to the Jiideration that if The proof come we know two We following theorem. and 13 and 14. loc. B.T IN A LIMITED REGION segments the its sides. give Axiom in weakening the axiom to the form given. It is interesting to notice that XVI. f The treatment of the plane and space which constitute the rest of this chapter are taken largely from Schur. p. (b) All points of every line which contains a point of each of sides of the triangle. shall be written the plane ABC. J3. loc. The plane determined by three vertices of a triangle is identical with that determined by two of their number and any other point of the line of either of the remaining sides. cit. 19 vertices. as Pasch. XVI in this form. If the three given points be A. shall be said to be within it. C their triangle shall be written A ABC. Neuere Geometric. (d) All points of every line which contains a point of the (c) two line of each of two will sides.f Theorem 17. cit. the vertices of the triangle play a special rdle. * Some writers. 21. p. All points of every line containing a point of one side of the triangle and a point of the line of another side. If a plane be determined by the vertices of a : triangle. 351. Let the reader show that this triangle is completely determined by all points of all segments having A as one extremity. Definition. If a line contain a point of one side triangle and one of either extension of a second side.
All points of each of the following figures The tetrahedron itself. Definition. and the four triangles its faces. is made up of fifteen regions. described as follows : (a) of the extensions beyond each (h) vertex of segments having one extremity there. as a matter of fact. It will be seen that a space. Four regions composed . its Theorem 20. in the which lie same plane shall be called coplanar. Planes which include the same line shall be called coaxal. like lines. the tetrahedron will be determined completely by means of segments. All points do not Definition. a point of the triangle of the other three. or two points of two opposite edges. thereof 22. lie AXIOM XVII. Definition. Four regions composed of the other extensions of the (c) segments mentioned in (6). if it do not lie in that face. The four given points shall be called its vertices. not collinear may with the two remaining determining points. their six segments its edges. If four noncoplanar points be given. Let the reader show that. Edges having no common vertex shall be called opposite.20 FOUNDATION FOR METRICAL GEOMETRY Theorem OH. Two planes having three noncollinear points common Theorem lie are identical. as so defined. to be opposite to a face. Theorem in 21. and the other extremity in the opposite face. Any one of the three points determining a plane be replaced by any other point of the plane. Points or lines in one plane. Planes. all points in that plane. shall be called a tetrahedron. A plane points which are not may be determined by any three of collinear. shall be called concurrent. The assemblage of all points of all lines which contain either a vertex of a tetrahedron. shall be called a space. Theorem 23. and for the other. while the other extremity is in the A vertex may also be said face of the other three vertices. which include the same point. (d) Six regions composed of the extensions of segments whose extremities are points of opposite edges. 19. the assemblage of all points of all segments having for one extremity one of these points. If two points of a line lie in a plane. and a point of the opposite face. all having a common extremity at one vertex.
line containing a point of one edge.i IN A LIMITED REGION defined 21 of a given will lie in the space tetrahedron. four noncoplanar Theorem 28. however. strictly speaking. (a) by the vertices A plane containing an edge. and the definitions of A line. on the line of an edge through the given vertex. It would not. Two spaces which have common are identical. In determining a space. may any vertex of a tetrabe replaced by any point of that space. and a point of the opposite edge. and space. must lie in one of the fifteen regions individualized by the A A E A E F t 9 F 9 * This means. that we shall not consider geometry of more than three dimensions. Theorem hedron 25.* of the Suppose that we have a plane containing the point but no point of the segment (BC ). Theorem 24. Theorem points in 27. Take and ( B) segment G two other points of the plane. space contains wholly every plane whereof it contains three noncollinear points. The proof will come directly if we take the steps in the order indicated. not its coplanar with the other three vertices. line containing a point of the line of each of two (d) A A A opposite edges. and a point of the plane of a face not containing that edge. A space may be determined by any four of points which are not coplanar. and hold fast to 16. line containing a vertex. and a point of the (c) line of the opposite edge. and a point of the plane (6) of the opposite face. of course. In determining a space. for we shall make various mutually contradictory hypotheses about space. be accurate to say that we consider the geometry of a single space only. any vertex of a tetrahedron may be replaced by any other point. not a vertex. and a point of the (e) plane of a face not containing that edge. plane. . not collinear with and construct the including space by means of the tetrahedron As C lies in this space. space contains wholly every line whereof it contains two distinct points. Points belonging to different spaces shall not be considered simultaneously in the present work. Theorem 26. Theorem 29. (/) A line containing a point of the lino of one edge. it whose vertices are A J3. line containing a point of one edge. PRACTICAL LIMITATION. 0.
If a line in the plane of a triangle contain a point of one side of the triangle and no point of a second side. two planes have a common point they have common Let P be the common point. Definition. and thus determine another line of the first plane which intersects the second one. it must lie in a plane containing one edge. the theorem is proved. we may take two points of this line on opposite sides of the second plane. Theorem 31. If two points be on the same side of a plane. If a point within the segment of two given points be in a given plane. Theorem of the third. If line. Theorem a 33. not collinear with the three already point chosen. If not. In the first plane take a line through P. it must contain a point of the third side. Now any other of the first plane. opposite to both. it must contain a point triangle.22 FOUNDATION FOB METRICAL GEOMETRY . If this be also a line of the second plane. It is immediately evident that all points common to the two planes lie in this line. We hereby reach a second point common to the two planes. The proof comes at once from 30. i tetrahedron or. a point opposite to one is . . more specifically. they shall be said to be on the same side of the plane. and 16 will show that in every case this plane must contain either a point of (AC) or one of (BC). Similarly. If a plane contain a point of one side of a but no point of a second side. will be opposite to one of the last two points. and the line connecting the two is common to both. . OH. and a point of the opposite edge. a point opposite to one is on the same side as the other and if two points be on the same side. we may define opposite sides of a line. those points shall be said to be on opposite sides of the plane otherwise. Theorem 32. 30. Every such plane will contain a line of the plane EFG> as may be immediately proved.
we have laid the basis for a metrical geometry. If the second is not null. however. while C may be ascribed to either. Let us divide all points of the segment into two . An We have essential point in our system of axioms is this. there will exist in the segment (AB) a finite or null number n of points Jc possessing the following properties AB P : AB < PQ then. space wherewith we shall. taken as a fundamental indefinable. rather than in a positive manner. Yet. AXIOM XVIII. presence of this point common to both. that makes it advisable to describe the two classes in a negative. the other to establish the possibility of congruent transformations. It is manifest that A will belong to the first class. we constructed that threedimensional type of We laid the foundation for the present work. There remains the third case where AB > PQ. If all points of a segment (AB) be divided into two such classes that no point of the first shall be at a greater distance from A than is any point of the second. If. Imagine the theorem to be untrue. clearly. is a magnitude. that == x follows. the one to give us the concept of continuity. We shall arrive at a contradiction as Suppose. be occupied. distance.CHAPTER IN Chapter I we II CONGRUENT TRANSFORMATIONS made a number of explicit assumptions. the principal use that we have made of these metrical assumptions. and PQ be any two distances whereof Theorem 1. In other words. that no point of the first class is within (OS) and none of the second within (AC). In order to complete our metrical system properly we shall need two more assumptions. AB PQ then 11 = 1 and P is identical with B. and. and this. from now on. n = 0. being subject to the categories greater and less. has been to prove a number of descriptive theorems. then there exists such a point C of the segment. and B to It is the the second. building thereon. firstly.
xxxix. Ch. In our present work we shall leave nonarchimedian or discontinuous geometries entirely aside. we may find such a positive integer All other the succession of points & being taken as above. we shall either Then. IV. cit. The point so found shall be called the middle point of the Theorem * A good deal of attention has been given in recent years to this axiom. has shown that a good deal of the subject of elementary geometry can be built up without the Archimedian assumption. it is convenient to do so.. and whenever. in nontechnical language. The proof is left to the reader. and Vahlen has shown.4#). Hilbert. for to belong to it both classes at would require Pn+ i or Pn+2 The theorem is thus once. or an adjunction of transfinite elements thereto. If.24 classes. Let D l)d< PQ. . or in the second ^^ be a P within (OB). that if a sufficient number of equal lengths be laid off on a arc not line. cit. that volumes may be similarly handled. we cannot have of the first class.. in the first case we construct M+1 . CONGRUENT TRANSFORMATIONS A point // shall belong to the n that first class if OH. for then we should find AK = APn }PnK'. loc. insert a point TO+1 n+l within (AC) making then. Halsted. We P 2. see Stolz. make use of our axiom of continuity We XVIII wherever. 2978. for we cannot be sure that our space shall include points of the type n in the extension of (AB) beyond B. loc. We PnK < PQ have contrary to the rule of dichotomy. which accounts for the otherwise somewhat obscure title of his book. Rational Geometry (New York. 1891. pp. therefore a cut of the type demanded by Axiom XVIII. as we may find n have PnC < PQ or so large that < PQ. vol. was the first writer to set up the theory of area independent of continuity. 1904). in fact. If be a point is clear that neither class will be empty. as > PQ. one of the second. however. 'Ueber das Axiom des Archimedes.* which says. proved. and be such a point of (AC) that a point of division 0. In any segment there is a single point whose distances from the extremities are congruent. For an account of the connexion of Archimedes' axiom with the continuity of the scale. n else we shall be able to P D 6 < PQ.' Mathematiscfie Annalen. and this n w+2 f P P P 3 ^w ^ P PB involves a contradiction. and that for the reason that their analytic treatment involves either a mutilation of the number scale. and P K H Ajmthin_(. It points of the segment shall be assigned to the second class. any point of that line may be surpassed. shall. able to state the principle in exactly this form. These questions are of primary importance in any work that deals principally with the significance and independence of the axioms. It will be seen that this theorem is merely a variation of the axiom of Archimedes.
AB be given and a distinct points of i Theorem If a not null distance positive integer m.n segment. it is possible to find m the segment (AB) possessing the properties AP\ It is = Jc merely necessary to take so that 2 7c > m f 1 and find Theorem 4. y so that last D k B < ADj_. we reach ing AD >7Q > Dk B. P B> AP .. B will clearly be a point of the second class. . any one of these congruent distances to AB by writing 3. n When any exist are given. will itself be a point of the first class...+1 = ^ AB. and a point of division D l and this point is different from A. but every point of (AB) at a lesser distance from A than a point of the first We have thus class. to be untrue. integer.__^AiP^ __ 9 ZP > ~PQ kP^D^t < P^S^B. ^ be such a point of Then. OtherLet us divide the points of suppose (AB) into two classes according to the following scheme. PP. the wise. CONGRUENT TRANSFORMATIONS we may 25 It will follow at once that if k be any positive find a set of points of the 2 . and let D k be the last extremity of the resulting segments. . If the distance it ^L^ be null. a distance that ~l Let D^i be the other extremity of this Let PQ bo such first.P2 1 1 PP ^ segment (AB) possessing the following properties We may express the relation of .. by mark(ADJ k successive distances by our previous device. may be marked in (AB) is &. by 1. reaching such a point Pn of (AB) that P n l all other points of (AB) shall be assigned to the second class. there segment (^^) and a positive integer n 1 points D l D 2 .Dn _ l of the segment (AB) such that theorem is trivial. A point 1 shall belong to the first class if we may construct a congruent distances according to the method already illustrated. that segment. once more a cut as demanded by Axiom XVIII. Let us next assume that the number of successive distances congruent to AD l which. that k < n. Suppose.
26
CONGRUENT TRANSFORMATIONS
such a point of
CH.
Ph
(AD k )
that
But this is a contradiction, for 7c is at most equal to n 1 and as P3 is a point of the first class, there should be at least one more point of division P k+r Hence k^.n. But k > n For we might then find Q leads to a similar contradiction.
3
l
Then mark of the second class so that (A 2) l Qi< ^AD\. & 2 successive congruent distances, reaching Q&_ 2 such a point of (ADj^) that Q&_ 2 ^i > ^A Hence,
and we may
D
we shall find that D k and B are identical. For otherwise we might find Q l of the second class so that nD^ < D n B and marking n successive congruent distances reach Q n within (Dn B), impossible when Q belongs to class two. Our theorem
l
But i 1^ n and find a (&l)th point Q k _i. this leads us to a contradiction with the assumption that Q l should be a point of the second class ; i. e. k n. Lastly,
=
is
thus entirely proved, and
It will
D
v
is
the point sought.
be convenient to write
5.
AD, 1
=
n
I
AB.
is
Theorem
not null,
If
AB and PQ
find
be given, whereof the latter
we may
n
so great that
_ _ AB <
PQ.
7i/
The proof
is left
to the reader.
are at last in a position to introduce the concept of and number into our scale of distance magnitudes. Let be two distances, whereof the latter is not null. It may be
We
AB
PQ
possible to find such a distance fiti that qtiti
= PQ, pR& = A K.
;
shall be called the numerical number __ ?__ measure of A B in terms of P(^, or, more simply the wieasure. It is clear that this measure may be equally well written
In
T)
this case the

f) L
or
WO nq
.
r
^
q
that
There may, however, be no such distance as HN. J
Then, whatever positive integer q
qLM = PQ,
we have
may
be,
we may
find
LM so
this
and
p
so that
1
TM>(ABpLM). By
number system
process
defined a cut in our
JD ft a nature that  and q q
"4"
of such
appear in the lower and upper
ii
CONGRUENT TRANSFORMATIONS
If
7?  be a number
27
divisions respectively.
of the lower,
and
*,
one of the upper division, we shall see at once by ? #/ f 1 reducing to a lowest common denominator that  < ? ? Every rational number will fall into the one or the other division. Lastly there is no largest number in the lower
> .
division nor smallest in the upper.
largest
For suppose that 
is
the
number
of the lower division.
Then
if
LM > (ABpLM)>
we may
put
L
__ =
l
find
n
so large that

LM
< (ABpLM).
as
M
l
LM.
At the same time
l
_ PQ =
__ nqL M
1
Let us
.
l
we
).
may, by
1,
find k so large that
L l M > (AB(np{lc)L l Ml
is
division, yet larger than
Under these circumstances p
a number of the lower
n(J
In the same
way we may
prove
that there is no smallest number in the upper. therefore defined a unique irrational number, and this in terms of PQ. taken as the measure of
We
may
have be
AB
7?
Suppose, conversely, that
there exists such a distance
is
any
rational
fraction,
and
AB'
that
(AB')
we may
find such a point
B
qABf > pPQ. Then in that AB =  PQ, i.e. there
M
I
will exist a distance having the measure  in terms of PQ.
let
ry\
...
_
Next
r be any irrational number, and
in the
let there
be such a number
rational
J 1
corresponding upper division of the
number system that a distance qAB' > ((p+l)PQ) may be found. Then the cut in the number system will give us a cut in the segment (AB'), as demanded by XVIII, and a point of division B. The numerical measure of AB in terms of PQ
will clearly be
r.
Theorem 6. If two distances, whereof the second is not null, be given, there exists a unique numerical measure for the first in terms of the second, and if a distance be given, and there exist a distance having a given numerical measure in terms
28
CONGRUENT TRANSFORMATIONS
OH.
thereof, there will exist a distance
having any chosen smaller
numerical measure.
Theorem 7. If two distances be congruent, their measures in terms of any third distance are equal. It will occasionally be convenient to write the measure of PQ
in the
form
MPQ.
8. If r > n and if distances rPQ and nPQ exist, > nPQ. rPQ When m and n are both rational, this comes immediately by reducing to a common denominator. When one or both of
Theorem
then
numbers is irrational, we may find a number in the lower class of the larger which is larger than one in the upper class of the smaller, and then apply I, 3.
these
Theorem
9.
If
AB >
CD, the measure of
AB
in terms of
any chosen not null distance is greater than that of CD in terms of the same distance. This comes at once by reduction ad absurdum. It will hereafter be convenient to apply the categories,
less, to segments, when these apply may respectively to the distances of their extremities. similarly speak of the measure of a segment in terms of another one. Let us notice that in combining segments or distances, the associative, commutative, and distributive laws of multiplication hold good e. g.
congruent greater and
We
;
r

nPQ = n rPQ = rnPQ, n(AB + CD) = nAJt + nOD.

Notice, in particular, that the measure of a the measures.
Definition.
sum
is
the
sum
of
all possible a halfline.
The assemblage of all points of a segment, or of extensions beyond one extremity, shall be called The other extremity of the segment shall be called the bound of the halfline. halfline bounded by and including a point B shall be written AB. Notice that every point of a line is the bound of two halflines thereof.
A
A

relation between two sets of points (P) and Definition. (Q) such that there is a one to one correspondence of distinct points, and the distances of corresponding pairs of points are in every case congruent, while the sum of two distances is carried into a congruent sum, is called a congruent transformation. Notice that, by V, the assemblage of all congruent
A
transformations form a group. If, further, a congruent transformation be possible (P) to (Q), and there be two sets / of points (P ) and (Q') such that a congruent transformation
ii
CONGRUENT TRANSFORMATIONS
9
29
from the set (P) (P') to the set (Q) (Q') then we shall say that the congruent transformation from (P) to (Q) has been enlarged to include the sets (P') and (Q').
is possible
a congruent transformation will carry of a segment, line, or halfline, into points of a segment, points It will also carry coplanar line, or halfline respectively. points into coplanar points, and be, in fact, a collineation, or linear transformation as defined geometrically. In the eighteenth chapter of the present work we shall see how the properties of congruent figures may be reached by defining congruent transformations as a certain sixparameter collineation group.
It is evident that
AXIOM XIX. If a congruent transformation exist between two sets of points, to each halfline bounded by a point of one set may be made to correspond a halfline bounded by the corresponding point of the other set, in such wise that
the transformation may be enlarged to include all points of these two halflines at congruent distances from their
respective bounds.*
Theorem
points into
10. If
two other chosen
a congruent transformation carry two chosen points, it may be enlarged to
include all points of their segments. Theorem 11. If a congruent transformation carry three noncollinear points into three other such points, it may be enlarged to include all points of their respective triangles.
Theorem 12. If a congruent transformation carry four noncoplanar points into four other such points, it may be enlarged
to include all points of their respective tetrahedra.
Definition. Two figures which correspond in a congruent transformation shall be said to be congruent. We shall assume hereafter that every congruent transformation with which we deal has been enlarged to the greatest Under these circumstances possible extent.
:
Theorem
their line.
13.
If
two
congruent transformation, the same
distinct points be invariant under a is true of all points of
Theorem
14.
If three
noncollinear points
be invariant
* The idea of enlarging a congruent transformation to include additional points is due to Pasch, loc. cit. He merely assumes that if any point be adjoined to the one set, a corresponding point may be adjoined to the other. have to make a much clumsier assumption, and proceed more circumspectly, for fear of passing out of our limited region.
We
30
CONGKUENT TRANSFORMATIONS
CH.
is true of all of their plane. points Theorem 15. If four noncoplanar points be invariant under a congruent transformation the same is true of all points of space.
under a congruent transformation, the same
Definition. The assemblage of all points of a plane on one side of a given line, or on that given line, shall be called a halfplane. The given line shall be called the bound of
the halfplane. Each line in a plane is thus the bound of two halfplanes thereof. Suppose that we have two non col linear halflines with
a
onehalfline,
and C be two other points of two points of the other. Then by Ch. I, 16, a halfline bounded by A which contains a point of (BB') will also contain a point of (0(7'), and vice
common bound A.
Let
B
and B' and
(7'
versa.
We may
this
thus divide
point, into
all
halflines of this
classes.
plane,
bounded by
two
The assemblage
of all halflines which contain points of segments whose extremities lie severally on the two given halflines shall be called the interior angle of, or between, the given halfThe halflines themselves shall be called the sides lines. of the angle. If the halflines be AB, \AC, their interior angle may be indicated by %.]BAC or $_ CAB. The point shall be called the vertex of the angle.

A
Definition. The assemblage of all halflines coplanar with two given noncoil inear half lines, and bounded by the common bound of the latter, but not belonging to their interior angle, shall be called the exterior angle of the two The definitions for sides and vertex shall be as halflines. If no mention be made of the words interior or before. exterior we shall understand by the word angle, interior Notice that, by our definitions, the sides are a part of angle.
the interior, but not of the exterior angle. Let the reader also show that if a halfline of an interior angle be taken, the other halfline, collinear therewith, and having the same bound belongs to the exterior angle.
Definition. The assemblage of all halflines identical with two identical halflines, shall be called their interior angle. The given bound shall be the vertex, and the given halflines
the sides of the angle. This angle shall also be called a null The assemblage of all halflines with this bound, and angle. lying in any chosen plane through the identical halflines, shall be called their exterior angle in this plane. The definition of sides and vertex shall be as before.
ir
CONGRUENT TRANSFORMATIONS
opposite.
31
Definition. Two collinear, but common bound shall be said to be
not identical, halflines of
The assemblage of all halflines having as bound the common bound of two opposite halflines, and lying in any halfplane bounded by the line of the latter, shall be The called an angle of the two halflines in that plane. We notice definitions of sides and vertex shall be as usual. that two opposite halflines determine two angles in every
Definition.
plane through their line. We have thus defined the angles of any two halflines of common bound. The exterior angle of any two such halflines, when there is one, shall be called a reentrant angle. Any angle determined by two opposite halflines shall be called a straight angle. As, by definition, two halflines form an angle when, and only when, they have a common bound,
Two angles shall in future cease to mention this fact. will be congruent, by our definition of congruent figures, if there exist a congruent transformation of the sides of one
we
into the sides of the other, in so far as corresponding distances Every halfactually exist on the corresponding halflines. line of the interior or exterior angle will similarly be carried into a corresponding halfline, or as much thereof as actually
exists
and contains corresponding
distances.
Definition. The angles of a triangle shall be those nonreentrant angles whose vertices are the vertices of the triangle, and whose sides include the sides of the triangle.
Definition. The angle between a halfline including one side of a triangle, and bounded at a chosen vertex, and the opposite of the other halfline which goes to make the angle
of the triangle at that vertex, shall be called an exterior angle of the triangle. Notice that there are six of these, and that they are not to be confused with the exterior angles of their respective sides.
Theorem 16. If two triangles be so related that the sides of one are congruent to those of the other, the same holds for the
angles.
is an immediate result of 11. The meanings of the words opposite and adjacent as applied to sides and angles of a triangle are immediately evident, and need not be defined. There can also be no ambiguity in speaking of sides including an angle. Theorem 17. Two triangles are congruent if two sides and the included angle of one be respectively congruent to two sides and the included angle of the other.
This
their logical sum. and 12. is Two Definition. Let the halflines be AB. which includes the common side. If two points be at congruent distances from two points coplanar with them. An angle shall be said to be congruent to the sum of two nonreentrant angles. Two with a sides. opposite to the halflines which are the sides of a given interior angle. the theorem is at once evident. Connect B with and K.BAD. If does not belong to the interior 4. angles congruent to two adjacent angles a straight angle shall be said to be suppleEach shall be called the supplement of the other. in fact. If two sides of a triangle be congruent. whose sum mentary. coplanar therewith. all points of the line of the first two are at congruent distances from the latter two. shall be The angle bounded by their remaining said to be adjacent. Definition. An angle which is congruent to its supplement shall be called a right angle. AC. It is clear that this is. Theorem 18. If three halflines lie in the same halfplane and have their common bound on the bound of this halfplane. nonreentrant angles of the same plane but no other common halflines. shall be called the vertical of that angle. The interior angle formed by two halflines. contain points of the same two sides plane. shall be called their sum. AD.  \ \ H  AD BHK Theorem then  AC be a halfline of the interior 20. one of shall be called a right triangle. recall the The truth of this is at once evident definition of congruent angles. if of the triangle one contain a point of (BH) and the other a point of (BK)< then B belongs to _ GAD. Definition. then one belongs to the interior angle of the other two.  AB 4^0AD. respectively congruent to them.32 CONGRUENT TRANSFORMATIONS when we on. all common Definition. the opposite angles are congruent. common side. triangle. containing points. points of the opposite halflines bounding this halfIf 4(7. naturally. Definition. . and having the same bound. The vertical of a straight angle will be the other halfplane. when it is congruent to the sum of two adjacent angles. Theorem 21. be called isosceles. A whose angles is a right angle. Theorem 19. Such a triangle shall. Definition.
fear of mishap. and J. if in any segment the points be paired in such a way that the extremities correspond.  AB AC 1 AA and meet AB  or AA{ be taken sufficiently small. keeping AB invariant and We may suppose   AA  H AA HA  We } AA . while the latter are interchanged. are opposite halflines.BAB^ to them will correspond AB^ and AC^ the latter being in the interior angle of 4BAB^ while by definition. is to be a halfline of the interior BAD. ACl being in the interior angle of 4. 1898. 16 does not hold. If AA{ be a halfline of the interior $ BAA l9 then we cannot have a congruent transformation into AA^. Let be the point of the segment (A^^A^) equidistant from A. Theorem 22. we choose any halfline AL of the interior 4BAA}. while A l goes into A[. Theorem it 23. There is one case whore this reasoning has to be modified. l carrying that A l and A are at congruent distances from A. have 2_JMC' and 2(LlLiZ) mutually ^ Enriques. and the greater of two distances from an extremity correspond to the greater of the two corresponding distances from the other extremity. These corresponding halflines selfcorresponding point.     corresponding halflines AL V AL so situated that AL l belongs to the interior ^b^BAL^ and 4~L^ AL is congruent to LAL V The proof is tedious. This will involve a if contradiction. If  AC impossible congruent. for here namely. when AC and I. and depends on showing that as a result of our Axiom XVIII. Let us notice. corresponding halflines always determine congruent angles with AB. however. 16. that we may and l enlarge our transformation to include the 4and ACl be two halflines l 4BAA^ respectively.IT CONGRUENT TRANSFORMATIONS 33 For we may find a congruent transformation keeping the former points invariant. * Cf. COOLIDQK C . If. AA{ will AC as we see by I. for to see that we shall have D be the intersection. it may be shown that we may find two DA == D   AB  AB BAA j . 80. may find a congruent trans/ into 'HA l Let this take formation carrying 1 l the halfline into Then if (in the same plane). then.* we may apply the first part of our proof without being found. If of the first angle. then there is one '  \ {  . it is easy ZM/ and simultaneously l /) A > or J^^i < O^i'j for is unaltered by the conl DA^ gruent transformation./. p. Geometria proieUiva. however. Bologna.
If two angles of a triangle be congruent. Theorem triangle 25. of greater than. and equal as applied to infinite assemblages. the verticals angles will be right angles obtained by extending (AB) beyond thus have four mutually congruent to the other two. and a not null angle. Given two nonreentrant angles. and < for less than. is that. Two angles between which there exists one of these three shall later see relations shall be said to be comparable. The reason why we cannot at once proceed to prove this fact. then we should have an angle that would be both greater than and less than itself. while D lies on the extension of (BC) beyond C. Let E be the middle point of ( AG) and let Df^meet^AB) that FE = EG. and congruent when applied to angles are mutually exclusive. we are not very clear as to just what can be done with our congruent transformations. is an immediate result of 18. we have perfectly clear definitions of greater than.34 CONGRUENT TRANSFORMATIONS We CH. . we see that the relations greater than. 1028. be said to be less than This As the assemblage of all congruent transformations first. a group. the is isosceles. pp. The second shall under these circumstances. An exterior angle of a triangle is comparable with either of the opposite interior angles. Paris. less than. For if we had two angles whereof the first was both greater than and less than the second.* Theorem 26. Let us take the triangle ABG. 1898. Borel. If ~DE > RF find of (DE) so BAG congruent to 4ECG Then we have * Cf. when it is congruent to the sum of the second. The first shall be said to be greater than the second. as we see from 23. As for the a priori question of = We comparableness. make right see as a result of 24 that if a halfline with the opposite halflines 4(7. Under these circumstances we shall say that they are mutually perpendicular   We AB A  We there. but are entirely in the dark as to whether when two such assemblages are given. one of these relations must necessarily hold. have merely to look at the congruent transformation interchanging a side of one with a side of the other. an absurd result. We shall write > in place the is means congruence. An angle is congruent to its vertical. AC'. that any two angles are comparable. in F. congruent right angles at the point A. and these alone. Theorem 24. Lemons sur la theorie de$ functions. less than. so far.
c2 .BD^O^ the supplement of C 1 which is congruent to 4^AG1 D 1 whose supplement is greater We have therefore returned to our original than 4. the side opposite the first is greater than opposite the second. remains to show that AB < AG + G1 B. If two points of (AB) be taken indefinitely close to D the angle which they determine at any point of (BG) other than B will become On the other hand as Ct approaches C indefinitely small. greater than that opposite Theorem the sum 30. Two angles of a triangle are comparable. where other than J5. P . One side of a triangle cannot be greater than of the other two. D. The proofs of these theorems are left to the reader. find such a point G1 of Theorem than that then ABG . we might find D l of = ^AD 1 A . BAG We may. and by the same reasoning D. sides of a triangle be not congruent. Were such problem. a 28. If in any triangle one angle be greater second. and if our original assumption be false. where have the triangle by the definition of congruence. (BG) that %Cj^AG It thus is congruent to %C^GA and hence (\A EE C^Cr. Let us > 2L EGA. If = is at once from this contradiction. (BA ~~BG) Our theorem comes Theorem 29. Evidently these sides cannot be congruent.B@i not the case. by our axiom of continuity. is any point of (AB) 2f_ APC$ will tend to increase. this time. two the angle opposite the greater side the lesser. and similarly the points D$ of (AB) tend to approach a limiting point.BAC greater than an angle congruent to 4. cannot be other than B so that the difference between BGt and BD i can be made as small as we please. Now in A BD 1 C1 we have ^.ii CONGRUENT TRANSFORMATIONS less 35 and than 4ECD. is less Tfieorem 31. the inequalities must always lie the same way. Next notice that. on the other hand f  AGV (AB) so that AD 1 and the problem reduces to comparing BCl and BJ\. with a smaller triangle. The difference between two sides of a triangle than the third side. in which case the angle is constant This shows that (7. Theorem 27. the points G^ of (BG) must tend to approach a point G of that segment as a limit. For they are comparable to the same exterior angle. But. however. Now this reduction process may be continued indefinitely.EGD. If RE < EF we have &.
so that BF==~JE. and we have a point of division D. appeal explicitly to intuition.. Many modern textbooks prove that all straight angles are equal . This is not usually sound. Two triangles are congruent if a side and two adjacent angles of one be respectively congruent to a side and two adjacent angles of the other.36 CONGRUENT TRANSFORMATIONS Theorem 32. F which Theorem 33. have worried over this theorem. all other points of (OB) shall belong to the second class. Let us take two such points B. Euclid puts it as his eleventh axiom that all right angles are equal. is PB < PB\ In the same and this way we would be contrary to the law of the > Dff. . may either of (CB) or of (CB'). We may make a cut in the points of (CB) according to the following principle. or to a vague continuity axiom. It is truly astonishing how much geometers. If a line be perpendicular to two others at * This is substantially Hilbert's proof. Through any point of a given line will pass one line perpendicular to that line. us suppose that CB > CB'. not. and C a point in the plane. Theorem 34. cit. in fact. as halves of equal things. hence right angles are equal. and perpendicular to Suppose. for it is not clear by definition why a right angle is half a straight Others observe the angle of a fixed and a rotating line. If If then CB =777?. 31 assume at A. and perpendicular to it at the same point. and either angle. B' on the given line. Theorem. let is E belong to (Off). It is clear that the requirements of Axiom XVIII are We could not fulfilled. p. Two distinct lines cannot be coplanar with a third. not on the chosen line. that A is the middle point of (Bff) and 11 B' < OB. ancient and modern.* definiteness. The locus of points in a plane at congruent distances from two points thereof is the line through the middle point of their segment perpendicular to their line. For will contain a single point AC AD BR We E AB = AD Then take on (BO). that we have AB* so that by I. CH. 16. A point shall belong to the first class if no point of the segment (PB) is at a distance from B greater than its distance from B'. Hence congruent 4~BB?F is congruent to %B'BE and therefore congruent to %BffE\ which contradicts 23. take that for all points E a point P of DE cut. it lying in any given plane through Let A be the chosen point. Theorem 35. 36. Hence could not have DB AD the perpendicular required. AC is the line required. by 31. lot P have J)B of < DB' so (DC) then near to very t for we D might. to (#'(7). loc. Bff< Off.
We find a point of division D. AB into any halfplane bounded We have merely to find and 0' the middle points of (AB) . and congruent > PJD there exists a dis+ AB congruent to PD + DC. Then as to + AD. be a point within the triangle and Theorem 39. AC+CB A and use our cut proceeding in shall point (0(7). There exists a congruent transformation carrying any segment (AB) into any congruent segment (A'B') and any halfplane bounded by by A'B. hence congruent to %A'O'C'. All lines perpendicular to a given line at a given point are coplanar. then D To prove this let BP pass through AC > AD tance a distance exists AB + A~D>BP + PD. these right angles be 4AOC and to be the middle point of (AB) and 0' the assume We middle point of (A'B') y where OA=0'A'. Theorem 38. AC > A'C'. Definition. The plane of all perpendiculars to a line at a point._WQ may also to suppose that distances exist congruent to Then^lC > AO and AC/ > A*0'.ii CONGRUENT TRANSFORMATIONS 37 their point of intersection. may Y AW + CFW. Theorem 41._we assume that AC = AC/. It is also clear that the locus of all points at congruent distances from two points is a plane. 1 > PC. Lastly . may 40.4'0'G". it is perpendicular to every line in their plane through that point. For if we had say. A ABC congruent to bA'KC'. Theorem 37. shall be said to be perpendicular to that line at that point.AO(! AC^AW. A congruent transformation which keeps all points of a line invariant. if no point of (OP) determines with A a distance greater then A'C'. Any two right angles are congruent. we have . The proof given in the usual textbooks will hold. Replacing the letter D by C f . first class. and hence PD + DC 1)0 > PCPD. . and see at once we might P belong to the that AD = A'C'. will transform into itself every plane perpendicular to that line. Theorem Let As AS AD of (AC). If P ABC there exist a distance congruent to AB + AC. AB + AL > RP + PC. otherwise it shall belong to the second class.
38 CONGRUENT TRANSFORMATIONS CH. The essential thing We shall demonstrate at length in Ch. and carry any chosen lino through this point.AOB congruent to any chosen angle. and 0' so that 00 = O'C". : We may be found to leave point. apply to angles system of measurement entirely analogous An angle may be represented to that applied to distances. in terms of any chosen A have also found out a good deal about the congruent The principal facts are as follows group. no further infinitesimal congruent transformations are possible. and any line through it. Several results follo^ from the last four theorems. and G' on the perpendiculars and (A'E') respectively. the sum of their measures. into any other such plane. (d) If a point. and so on. XVIII that is this. A congruent transformation may be found to leave any (ft) chosen point invariant. a line through it. any two angles are comparable. whose correreentrant angle sponding interior angle is the greater. may be at once translated into the geometry of the angle if straight and reentrant angles be excluded. will be the logical sum of two nonreentrant angles. . at in J. We may then not null angle. satisfying these four requirements. We see also that our Axioms IIIXIII and XVIII. there is only one congruent transformation of the line possible. We may extend our system of comparison to include straight and reentrant angles as follows. and a plane through the (c) A congruent transformation invariant any line be invariant. (a) A congruent transformation may be found to carry any point into any other point. there will exist Theorem 42. A straight angle shall be looked upon as greater than every nonreentrant angle. besides the identical one. If any chosen halfplane bounded by OA a unique halfline OS making the ^. To begin with. The proof of this theorem depends immediately upon the preceding one. Of two reentrant angles. as we see at once from 42. into any other such line. but to carry any plane through this line. and less than every reentrant one. the congruent group is completely determined by the requirement that it shall be an analytic collineation group. and shall have as a measure. unequivocally by a single number. that one shall be considered the less. let the reader show that if a point and a line through it be invariant. be a given halfline. and to AB and A'ff. The last assertion has not been proved in full.
45. their dihedral angle. Definition. The proof is immediate. and. If two dihedral angles be congruent. a transformation may properly be called a reflection in that plane. the dihedral angle itself shall be called right. in fact. Every segment whose extremities are one in each of these half. and conversely. The definition of the dihedral angles of a tetrahedron will also be immediately straight. A plane perpendicular to the edge of a dihedral angle will cut the faces in two halflines perpendicular to the edge. containing points of segments whose extremities lie severally in the two given halfplanes. bound. more simply. or. and their bound the edge of the Definition. Two plane angles of a dihedral angle are con We have merely to take the congruent transformation which keeps invariant all points of the plane whose points Such are equidistant from the vertices of the plane angles. and reentrant dihedral angles. Given two noncoplanar halfplanes of common The assemblage of all halfplanes with this bound. and the planes shall be said to be mutually perpendicular. .n CONGRUENT TRANSFORMATIONS 39 that we have two halfplanes on opposite aides _ of a plane a which contains their common bound I. Theorem gruent. The interior (exterior) angle of these two shall be called a plane angle of the interior (exterior) dihedral angle. dihedral angle. We evident. by following the analogy of the plane. all such points will lie in one halfplane of a bounded by I. shall be called their interior dihedral angle. any two of their plane angles will be congruent. as may easily be shown from the special case where two segments have a common ^6 extremity. Theorem 44. then the first plane is also perpendicular to the line of intersection of the other two. The assemblage of all other halfplanes with this bound shall be called their exterior dihedral angle. Theorem. and the three be concurrent. Let us next notice that we may measure any dihedral angle in terms of any other not null one.planes will have a point in a. define null. If a plane be perpendicular to each of two other planes. The two given halfplanes shall be called the faces. and that its measure is the measure of its plane angle in terms of the plane angle of the latter. If the plane angle of a dihedral angle be a right angle. 43. may.
If a point of a segment as desired. as we shall amply see by elaborating consistent systems of geometry where this sum is greater than. be congruent to the sum of two right angles. or shall not. however. belong to the internal l(^. by III. and lying in that halfplane. while B3 is a point of the extension of (AB) beyond B. at as small a distance from P A point. a 4 AGP < 4_ACD.AGD. tfLACD will necessarily differ from 4ABD by less than any chosen angle. a fortiori. or less than two We must first. did not take up the question of the sum of the angles of a triangle. Theorem 2. while the other side including this angle may be diminished at will. and if we take P within (AE). The axioms so far set up are insufficient to determine whether this sum shall. Let Bl be the middle point of (AB). 4. one side and an adjacent angle remain fixed. while G is the variable vertex within a fixed segment (BD). once more \ 4_ACD 9 and. CD   AB 4. then the external angle opposite to the fixed side will take and retain a value differing from that of the fixed angle by less than any assigned value.CHAPTER III THE THREE HYPOTHESES IN the last chapter we discussed at some length the problem of comparing distances and angles. the it^ACP may be made less If C be a point of the theorem is trivial. and that shall be our next task. give one or two right angles. in any triangle. so that iAQD is congruent to the given If then angle. theorems concerning the continuous change of distances and We angles. we in the halfplane bounded by CA may. find which contains B. Through each of the points JS1 52 J53 construct a halfline bounded thereby. Theorem 1. . equal to. If not. we have AB (AB) may be taken and G be any other than any given angle.ACB If less than AD belong to the internal %ACB. We wish to show that if BG be taken sufficiently small. AD must contain point E of CAB. If. 4_AGP<4AGD. and of giving them numerical measures in terms of known units. Let the fixed side be (AB). . . and B2 the middle point of (Bj^B).
be at a small distance from (7. 41 bounded by AB which contains D. is B J^C^ possible to extend CjjD r AD^ will surely is (B^) meet 2 2 in a point 7)2 when B^L\ differs infinitesimally from very small. and hence exceeds AB by FIG. with regard to the 1 1 I) l (i. and hence on the other side of J? 2 from A and jD Let D^C' meet B 2 D<2 at We now 2 2 the external angle at see that. may more. say (715 (72 C.6 respectively. and let the angles so formed at J3 15 J?2 J53 all be congruent to i^ABD. and 1 Dt congruent to = D H . a amount.e. AC We over take BO beyond C^ to D so tiny that^it so that l . We may contains a point of each certainly take EG so small that of these halflines. it is greater than 2AGl which We may thus finitesimally from 2AB19 or AB.CH. and as AG^ AB%. 1. one of the mutually vertical external angles) is l) 1 C' + 4LC'i> 1 l)s ). Ill THE THREE HYPOTHESES . finite differs in find C" on G' will the extension of (ACJ beyond C^ so that AC^ G^T'. AAB D : > (S .
and see similarly that the difference between and ^LJ. 9 this let us first notice that and AB will form ~ BG 77L . 2. .JSZ) will become. the sum of the angles of this triangle will differ infinitesimally from a straight angle. Theorem 6. and ^AGI) is Lastly. is congruent to 4The l l and. Now such points will lie as close as II. the 2 2 approach 2 and //2 at every point in angles determined by J5 2 2. If in any triangle one side and an adjacent angle remain fixed. H ^S^D^ B D %B^H^ . hence congruent to and ^LC"jf)1 J?2 is the difference between and as and as a limiting position. D space decrease together towards a null angle as a limit. Theorems. in fact.AG be perpendicular to HG then which contain . and of the variable angles will be continuous functions of tho measure of the variable side first mentioned. If a set of lines perpendicular to a line I.. and remain infinitesimal. meet a line m. where m and n are integers. If two lines AB. then the measures of the third side.ABD becomes and remains in DD we finitesimal. by III. Several corollaries follow immediately from this theorem. B&. by our previous reasoning. the difference between which will. If in any triangle one side and an adjacent angle remain fixed. the distances of these points from a fixed point of and the angles so formed with m. Hence 1^C'D^^ becomes infinitesimal. Theorem 4. be congruent to every other perpendicular distance from A to BC. as and 4B^BD ABl B^B.4 and points of BO are perpendicular BC\ and all points of BC are at congruent distances from A. Such a distance will be the distance from A to the middle point of (BC) and. 2 . and the difference between . AB D D ^ABD. to every point of BC whose distance from B may be expressed in the To prove Theorem 5. will vary continuously with m . Of course a constant is here included as a special case of a continuous function. differ from a right angle. 31. hence y no distance from A can differ from AB and no by angle so formed can. while the other side including this angle becomes infinitesimal. all lines to our A A BC is isosceles.B1 1 2 and ^. 4AB^ ^B^D 4B^D 4B^A with therefore The difference between ^.42 THE THREE HYPOTHESES OH.ABD and 4ACD will become and remain less than any assigned angle. while the other side including this angle varies. we please to every point of BC. become infinitesimal But = are congruent.
and the given segments The foulits sides. than he knew. internal angles 4. B. and proceeds according He builded better to the modern method of assuming the postulate untrue. The theorem given above covers Sacckeri's theorems 1 and 2 on p.* In an isosceles birectangular quada line through the middle point of the side adjacent to both right angles. adjacent three right angles it shall be called trirectangular. however. The given points shall be called its vertices. Saccheri's is the first systematic attempt of which we have a record to prove Euclid's parallel postulate. the quadrilateral will be transformed into itself. (DA). the right angles having their vertices at and B. 1732. 4. Definition. Given four coplanar points A. as he attempted to show.CD A shall be The definitions of opposite sides and called its angles. the opposite sides being interchanged.DAB. 4_ ABC. opposite vertices are obvious. A convince himself that. . This theorem may be more briefly stated by saying that A E F Saccheri. which is perpendicular to the line of that side. 4. birectangular and trirectangular quadrilaterals necessarily e^dst. shall be said to be isosceles. (BO). (7. 1895. and four Let the reader right angles it shall be called a rectangle. will be perpendicular to the line of the opposite side and pass through its middle point. Let the quadrilateral be ABCD. quadrilateral with right angles at two If it have vertices shall be called birectangular. and not inconsistent. * Engel und Staeckel. A rilateral Theorem 7. It will be easy to pass a plane through this line perpendicular to the plane of the quadrilateral. and by taking a reflection in this latter plane. as are the definitions for adjacent sides and vertices. Then the perpendicular to Att at the middle point of (AB) will surely contain point of (CD). Accessible in Thcorie der Parallellinien von Euldid bis auf Gams. Definition. 1) so situated that no segment may contain points within three of the segments (AB). under our hypotheses.in THE THREE HYPOTHESES 43 the distances from a fixed point of I to the intersections with these perpendiculars. The proof comes easily from 2 and 5. (CD). 50 of the lastnamed work. Eudides ab omni naevo rindicatus. for the system so constructed is selfconsistent. Milan. birectangular quadrilateral whose opposite sides adjacent to the right angles are congruent. Saccheri's.BCD. Definition. The other two angles of the quadrilateral are mutually congruent. Leipzig. The assemblage of all points of all segments whose extremities lie on these segments shall be called a quadrilateral.
TJteorewi 9. 4AEB. that the distances so called for exist simultaneously on the sides of a birectangular isosceles Distances so indicated will be everywhere quadrilateral. under the present hypothesis. be the rectangle.^I'C'JB'.EBC. In a rectangle the opposite sides are mutually congruent. hence there will exist such a point . If and be the middle points of (BC) and (AC) respecand A EBC are isosceles. EGA are congruent. If there exist a single rectangle. the right angle being ^. If there exist a single right triangle the sum of whose angles is congruent to a straight angle. the same is true of every right triangle.44 THE THREE HYPOTHESES OH. Then EBC ECU since 4. Be it noticed that. dense on any line. provided. _ __  ~ t <  * AC. Passing now to the we see that the perpendicular to A'C' at F' the E . all trirectangular quadrilaterals are under the present circumstances rectangles. by 6 we may construct a rectangle having as one of its sides one of the congruent sides of any isosceles birectangular quadrilateral. this line divides the quadrilateral into trirectangular ones. the right angle being 2jL ACB so that the sum of the other two angles is congruent to a right angle. hence. and 4. We see that both it^ABC and ll^BAG are less than right of (AB) that angles.same process. and quadrilateral. in the tively. a rectangle which is identical with this All isosceles birectangular quadrilaterals. EAC D = . and any isosceles birectangular quadrilateral whose opposite sides are mutually congruent is necessarily a rectangle. Theorem 5 is superfluous. 0. as EDGF right angles at D. Theorem 10. and hence. by a repetition of the . We have to prove that the sum of its remaining angles also is congruent to a right angle. every isosceles AB birectangular quadrilateral is a rectangle. hence 4EDGF is a rectangle. Let A A BC be the given triangle. and F. The angle quadrilateral at is also a right angle.ACB is congruent to the sum of i^EAG and 4. F &EAG E . of course. we have. Continuing this process we see that we can construct a rectangle whose adjacent sides may have any ABCD measures that can be indicated in the form i AB. two mutually congruent Theorem 8. for it is one half the straight angle. Let A A'B'G' be any other right triangle. The line perpendicular to Let at the middle point of (AB) will divide it into two smaller rectangles.
or change continuously. all will. and then the other of the sides which include the right angle. and the same will hold for every right triangle. by 2. If there exist any right triangle where the of the angles is less than a straight angle. some will meet (AC) and some (BG). Then kC'E'R is isosceles From this conies of i^E'R'C' and i^E' we wished to show. f is clear that 4D'E B = 4D'E'C' since 4F'E'D' is a right f angle and like &A'EC 4F'R'A' = jFE'C'. If no two of the perpendiculars intersect. f . pass through this point. Let us notice. from any right triangle to any other by means of a continuous change of first the one. If two of be a perpendicular to (in the plane BO). Theorem 13. If there exist a right triangle where the sum of the angles is greater than two right angles. divide our triangle into two right triangles. by an easy modification of 9. every case we may. but may never become congruent to the sum of two right angles. the same is true of every right triangle. the sum of the angles will either remain constant. Thewem sum We Theorem 12. we shall find a point of division D. by 5. cut will clearly. therefore. say 4At each point of (AB) there will are less than right angles. In one of these the sum of the angles must surely be less than (congruent to) a straight angle. by XVIII. and. then. hence it must always remain less than that sum. will meet (A'B') in E\ and the perpendicular to ETF' at E' will meet (_S'C") in D'. AC f r is immediately that the sum congruent to a right angle. then in every triangle the sum of the angles is less than (congruent to) a straight angle. This comes immediately by reductlo ad absurditm. or greater than right angles. and a line hence to C will surely be perpendicular to AB. A AB D 19 . thus be determined among the points of (AB). under our present circumstances. must have at least two ABC AB these perpendiculars intersect. Next observe that there can. But. For we may pass see the truth of this by continuity. exist no triangle with two angles congruent to. to begin with.in THE THREE HYPOTHESES 45 middle point of (A'C'). the It trirectangular quadrilateral E'F'D'C' is also a rectangle. the same is true of all right triangles.BAO which angles. In this change. If there exist any triangle where the sum of the angles is less than (congruent to) a straight angle. that our given AABd and ^. as 11. as there exists one rectangle. It is at once evident In that the perpendicular to at will pass through 0.
viii. vol. * we ' * *( . various possible assumptions in evidence. the sum of whose angles is less than a straight angle is called the Lobatcheiuskian or hyperbolic. There will exist. xvii. the angles is less than (congruent to) a straight angle. though the credit for discovering the hyperbolic system is generally given to Gauss. ). with explanations in the appendix by Weber. Hence every triangle can be divided into two right triangles In each of these triangles. author $ Riemann. This was followed by an ' article Geome"trie imaginaire '. 1887. Ueber die Hypothesen.* The assumption that there exists a triangle. The three hypotheses were certainly familiar to Saccheri (loc. where the sum of the angles of a triangle is still congruent to a straight angle. This comes at once by reductio ad absurdum. Die Legondro'schen Satze iiber die Winkelsumme im Dreiecke. hence in the triangle chosen. liii. Dohii. Theorem 15. g. numerous geometries. hypothesis. Theorem 14. 272 of the second edition of his Gesammelte Werke. Crelle's Journal. L. of course. All spellings of Ix>batchewsky*s name in Latin or Germanic languages are phonetic. Geometry in which the sum of the angles of a triangle is two right angles/ Transactions of the American Mathematical Society. e. other than those which give in the following pages. the same will be true of every triangle. the sum of the angles is less than (congruent to) a straight angle. and R. the sum of whose angles is greater than a straight angle.' Mathematische Annalen. The has seen eight or ten different ones.^ Only under the elliptic hypothesis can two lines intersecting be perpendicular to a third line coplanar with them. Lobatchewsky'g first work was published in Russian in Kaaan. vol. If there exist any triangle where the sum of the angles is greater than a straight angle. f The assumption that there exists a triangle. see p. Definition. is called the Riemannian or elliptic hypothesis. first read in 1854 . The difference between the sum of the angles of a triangle. who speaks of it in a letter to Bolyai written in 1799. The assumption that there exists a single triangle. the sum of whose angles is to a straight angle is called the congruent Euclidean or Parabolic hypothesis. wdche der Geometrie zu Grunde liegen. We have now reached the fundamental fact that the sum of the angles of a single triangle will determine the nature Let us set the of the sum of the angles of every triangle. 1900. those lacking our strong axiom of continuity. the sum of the discrepancies of the resulting triangles is congruent to the discrepancy of the given triangle. and a straight angle shall be called the discrepancy of the triangle. 1907. vol. If in any triangle a line be drawn from one vertex to a point of the opposite side. Moore. in 1829.46 THE THREE HYPOTHESES CH. the sum of as we have just done. cit. Cf.
f is which infinitesimal. triangle. will diminish towards zero as a limit. is taken as the basis of a number of works on noneuclidean geometry. Etudes analytiques surla theorie ties paralleks Paris. which start in the infinitesimal domain. the discrepancy of the triangle. Then. way of stating this theorem is to say that in the infinitesimal domain. then the discrepancy may be made less Theorem by saying If. the geometry of these triangles may be made to differ from the geometry of the euclidean hypothesis by as little as may be desired. in any than any assigned angle. if loose. loosely proved. one vertex remain fixed. Theorem 17. Euclid. the other vertices lying on fixed lines through it. If it do tend to approach a straight . that if in any one angle remain constant. and if a second vertex may be made to come as near to the fixed vertex as may be desired. becomes we may apply 1 to show that If C do aroach A we may pproach take I) the middle point of (AC) and extend (BD) to E beyond A == EB.#4(7 tends angle. If C do ^AGB not approach A. Cf. Proposition 16. g. while the third vertex does not tend to recede indefinitely. and work to the finite by integration. Theorem 18. e. ( * y . as neither BO nor must be very close becomes indefinitely great. infinitesimal. it will remain less than some nonreentrant angle. we have euclidean geometry. is yet greater Book I. Then we may apply Euclid's own so that proof* that the exterior angle of a triangle is greater than either opposite interior one. Notice.in The proof THE THKEE HYPOTHESES 47 is immediate. If neither angle adjacent to the diminishing side tend to approach a straight angle as a limit. and 16 will apply to all such angles simultaneously. let the diminishing side be (AB). hence. or to AC A A itself. while neither of the other sides becomes indefinitely large. so that the exterior angle at D DE than 4ACB. This theorem. while one or both of the triangle. A specious. the discrepancy may be made less than any assigned angle. we see that to some point of the extension of (AB) beyond A. other vertices tend to approach the vertex of the fixed angle. 1871. along fixed lines. while to approach a straight angle. all thus diminishing together. If.Marie. If in any triangle one side may be made less than any assigned segment. Flye Ste. when not We shall make zero. while no side becomes indefinitely great. one side of each may be made less than any assigned segment. in any system of triangles. this more clear 16.
Draw P B on either half of A X bounded drawn perpendicular to AB at any AX be two mutually perpendicular AB E P a line there perpendicular to AD. terms of some convenient unit. fundamental question with which we shall have Suppose that we have an BCD. If AD and we may find such a point by A. and worse still. and most important. Paris. Suppose. meet the perpendicular at at a point Q. There is no proof that the required limit does actually exist. PQ differs AX E but little from AE> and. A right angles are at A AB the AD . which lies in the same halfplane bounded by a point whose distance from is greater than AD. If B be a point of very close to 4. further. thereby rendering valueless his solution of its functional equation. several important points are omitted by him. that becomes infinitesimally small. that. Let be a point of the extension of ( A D) beyond D. is greater than AD. It has been possible to shorten some of his work by the consideration that we have euclidean geometry in the infinitesimal domain. a line being of (AB) we may find on the half thereof bounded by point as does D. and several of the actual proofs in this chapter are taken directly from G<3rard ? La geometric noncuclidienne. he gives no proof that the resulting function of M AD is necessarily continuous. P. first convince ourselves. and if a line perpendicular to AB at P of (AB). as null distance. we must show that a definite limit does measure of of XY in all.CHAPTER [V THE INTRODUCTION OF TRIGONOMETRIC FORMULAE THE first to deal in this chapter is the following. that. 1892. AB decreases towards the Theorem lines 1.* But. when is given construct a suitable quadrilateral secondly. On the other hand. * The general treatment. hence. remaining constant what isosceles. whose birectangular quadrilateral and B. . will be the limit of the fraction MAB J^ where M XT means AD . we must we may always necessarily exist for this ratio.
I may find such and R. on the other hand. Theorem If in a triangle may sides of the triangle. that if in any triangle. Q n __ will be within (Q$). and PQ < is and PQ < ^ __ QR. the ratio of the measures of this and the opposite side approaches a definite limit. having right angle. A similar proof holds when $~PQR greater than a right angle. tf^QRS will differ infinitesimally from one half 4PQR. If in an isosceles birectangular quadrilateral. and neither of the other angles approaches a straight angle as a limit. Now if RQ remain constantly greater than a given not null distance. we speak of the ratio of two . 4PRQ becomes infinitesimal. so that QR = QS. during the rest of this chapter. if the ratio of the measure of the first to that of the second may be made less than any assigned value. and (AD) as one of the congruent sides. matter how large a positive integer n may be. Then. by hypothesis.OH. while Suppose that we have. as a corollary. we require 4QRQ n ^ ^ e ^ css ^ nan named amount. in fact. the congruent sides remain constant in value. the side opposite this angle will be infinitesimal compared to the other 2. that n points Q i may be found on positions for so that $PRQ~4QRQi 4QkRQk+i> y et tQRQ n is less P = j PQ than any chosen angle. It is clear that one of the angles &. the theorem is perfectly evident. we may find S on PQ but not in (PQ). It will follow. while the side adjacent to the two right angles decreases indefinitely. A PQR with ^LPQjR fixed. If. RQ decrease indefinitely. If. infinitesimal We v shall say that a distance may be made compared with a second distance. one angle become infinitesimal. a second angle whereof one angle is constant. Theorem 3. then. be made as small as desired. then the side opposite the infinitesimal angle becomes infinitesimal as compared with either of the other sides. AX y AX AD A A Definition. It will save circumlocution and involve no serious confusion if. and the right 2jLD may be very near to any point of taken as the vertex of a second right angle of an isosceles as the vertex of one birectangular quadrilateral. no Suppose it be ^QPR. angle. iv TRIGONOMETRIC FORMULAE 49 The net result of theorem 1 is this.PQR or 2(L QPR must be greater than a right Then.  as geometry in the infinitesimal domain obeys the euclidean hypothesis. If be given.
while A A B perpendicular to Saccheri's theorem. Let us next show that the function A! ABE' as before. as _ AB such points gets nearer D decreases. Take is continuous. every ratio jFyT1 '.AB and D D that A A' = DD 19 _ /./> TT  Hence ^=^ AC AB . (x). as AB approaches _ the null distance. Let us imagine that and A' are is on a fixed line at a very small distance from A. CH. becomes infinitesimal. 6^ 4C'AA differs infinitesimally from a right angle. and let the their vertices at fixed points. Let us then take the isosceles birectangular quadrilateral A'ABB'. and l nearer to r=r AB Lastly.approaches a definite limit. for. at G meet (A'B') at C". In the other cases it is a variable depending on the measure If this measure be a?. by the middle point of (AB). _ . we may call our limit of A A'. so that if 6\ be the point as is AB AC < of (?'). rt But i A = ==r AB AC LJ jA. if P be any point of (AB)> and P lie on the perpendicular at P so that A A' == PP 19 we may find one of our points recently called D of such a nature that DP1 and JDjPj are infinitesimal as compared with AB. and write such a ratio simply XY ~ . Now. By a repeated use of this process we see that if D be such a point of ( A B) that at AD = . is however small more. Let G be the middle point of (AB). Hence A'Pl A _ .!? AP AB 7 zrrrr _ ft < where e is infinitesimal with AB. what we may take all _ AB be.50 distances. THE INTRODUCTION OF instead of the ratio of their measures. while A l and B1 are respectively on the < < .0 C". for which Ji. l such a point of the perpendicular then. so small that this inequality shall hold for at once. which. e may A^I) l A'B' AD AB =r < and. 7 C^ = ZZ A. Jj . . or (CG') extended beyond Csv'/ . by III.< 5 where 8 may be made less than any assigned number. _ This shows that . This limit is constantly equal to 1 in the euclidean case. the right angles having and B. n AC L \77ir . G.
(x y) that of AO^ equation. 4CC2 and ^LC^jR will differ exist. We shall find the actual form of from its functional Let x be the measure of AC. that this construction is possible. in fact. as compared with V V ^__. opposite C^PC^R (PP') is <F. We shall presently suppose AB to be infinitesimal. < (7P 2 __ we have 8 0^=0^. on (GP) or" (GP) extended beyond P. is <j> and (AC2 ).   + y) at will meet respectively. however great that m may be. ^LC^CJt EE ^L0 2 OP C^ = <7<72 Also CC2 P and ^. of the second order. < A d> (#) < h S. that of AG^ where (7 and (7T are points within Take a corresponding set of distances upon a line near = AC. v ' m a continuous function. the four lastnamed points wilJ surely be very tiny.C and J5D by. M(72 P meaning thereby the measure of 2 P.. But 2 ~CoP =  _ C' J? + 2e where c is infinitesimal. let us D 2 1 _ as the angle infinitesimal. let us compare AC (7 P and AOC^jR. beyond the measure of A A' be x. thereto. = AL\ while J. A B f . For. if from right angles.TV TRIGONOMETRIC FORMULAE 51 and of (BB') beyond extensions of (AA'). BD are in the same halfplane bounded by AB and perpendicular We know. Hence. we take f t> . . while that of A'A l is A#. by 1.(7(7^ differ infinitesimally. The perpendicular to CD at C will meet CZ D 2 and G^D^ in P and R (x . r t B '. Lastly. while the perpendicular to these lines at Q and & . BD. SS1 = Z^. _ . if. so that by 2 infmitesimally CD D AB P This infinitesimal 6 is. Let ~A  B A  1 Now and. hence. we may take A^A' so small A*A' < K AB> 1 2m then and.
is infinitesimal. OH. i. know that $(&) cosr is a continuous function. Let us assume that the unit of measure of distance is well fixed Let x 1 be a value for x in the interval to which the equation applies. have immediately ^ </> = We If. the measure of an actual distance. We may find k so that <f> (x^ = /yi cos rk/  We . distance. be any value of the argument.52 THE INTRODUCTION OF 8. or . OP and 0~P is infinitesimal. e. e. ^)(aj) x cos will be j less than any assigned quantity. then. use letters of the type to indicate infinitesimals. 7?. and remember that AB is an infinitesimal But GP > (TO. cos ~ = /j\ cos also a. Substitute in the first equation connecting G2 P and C\R _ __ Hence <(#f 7/) + </)(iC y) 2(t)(x)(t>(y) < made less than any assigned value 77 where ?/ may be This wellknown equation may be easily solved. f "7? JT we may find u and Hence m such large integers that x ^~ .
at most. however. in the elliptic is. signs of imaginary values from (2) by writing </> ik. we may. of course.iv TRIGONOMETRIC FORMULAE The function cosine has. and if & be the measure of the latter. loc. equal 1 /L" hence . In the hyperbolic case. it more fully in subsequent chapters.BAC is constant. We shall that it gives the radius of a sphere (in our usual euclidean geometry) upon which the noneuclidean plane shall. then. suppose that we have a right triangle. A ABC whose right angle is 4AC\ We shall imagine that It is now of two simultaneously diminishing AB becomes infinitesimal while ^. define the constant may be developed. notably XIX. cit. and let A# be the measure of .)  As a matter doing ratio this. of course. Of fundamental importance find We p as the Medsure of Curvature of Space* k. the possibility that this We . i.CAC'. we see immediately that in the < ^ 1 . we may write our function /(#) including therein. beyond (7. We shall This fundamental concept is due to Riemann. loc. We seek the limit of AB AC . If A0 be * consider ease. that of the Hides of a right Let us.e. f It is strange that Gerard. there is little or no gain in necessary to calculate another limit. if we choose. To find the nature of the value of parabolic case to 1. p. 53 meaning. is positive. or k a pure imaginary. of fact. remove k' all = Under these circumstances. function should be a constant. the reader.f That such a limit will actually exist may be proved by considerations similar to those which leave the details to established the existence of </>(#). (x) = ^ v cosh (^. First of all it is incumbent upon us to show that this Take C' on the extension of (BC) function is continuous. assumes this ratio from the euclidean . we a purely analytical write is the constant k. therefore. < con stitutes a minimum value for and 1 fa r% is negative.. . eit. The limit is a function of the angle ^jLjEMC*.
  within the interior angle certainly exist if OF be very small. AB r=r will become and remain AB than any assigned number. infinitesimal. and through F.A respectively. F these points will OB 01) OB UK . infinitesimal. Suppose. then. D.54 THE INTRODUCTION OF __ OH. that we have two halflines OF. and f(B) is continuous. if OF be small enough. Take JF on OZ. and meeting it in E. .C. <f><0. If. * =8. + $  . _ Hence by 2 CC =rr f is infinitesimal as compared less with AC. be separated from the middle point of (EA) by a distance infinitesimal compared with EA. is  OB . so that OF=~OB 4YOF = and B by a line Connect OF in /). Let XOF and . so that it = cos y and B must be real. then. for the perpendicular to OX at such a point would meet (BF) at a point whose distance from D was infinitesimal as compared with OF. This f is the functional equation that I we had before. which C will points also are sure to exist. and have the measures 0. B draw three lines permeeting pendicular to OX. \OZ lying in a halfplane bounded by \OX. now. and find  J3.XOF. we so choose that the _ measure of a right angle shall be ^ f(9) * = cos<9.XOZ be each less than a right angle.
(3) The extension of these functions to angles whose measures are greater than ~ will afford no difficulty. c treated from the cases where they do. by 1. lim. and J^ on the extension of (AB) beyond A. now make rather an elaborate construction. the measures of BC.iv TRIGONOMETRIC FORMULAE Let us not fail to notice that since ^L4jB(7 have. and we may draw G1 C3 perpendicular to CC%. a construction 1 which. the geometric extension may be effected as in the elementary books. is certainly Draw A^B2 We saw that decreases) We from i^EGA by an infinitesimal (as Bt B and &CCl B l will approach a right angle as a limit. Let will differ but little from Bl Cl meet (AC) at <72 AB We = AA = . and construct B1 GI A ABC. G1 lying not far from (7.* Take B1 in (AB) as near to B as desired. 2 which . so that A^A B^B. on the other. on the one hand. b.i l \J\J n for CC1 cos y BB is * infinitesimal in comparison to BB 1 or See figure on next page. . such will still be the case if the sides of the triangle be not large. Let the measure of ^. 17. the defining series remains convergent. Let us next find 2 on the extension of (A C) beyond A so that A%A = GZ C and B 2 on the extension 4C&C A of (Cj^Bt) beyond B% so that possible as 6^6^ is very small. We shall assume that both ^ and are less than ~.BAG be \(r while that of 4BGA is 0. GA. respectively. and the case where the inequalities do not hold may be easily Let us also call a.BCA. Let this be the A ABC with ABC as its right angle. while under the elliptic. and. $. for. where Ca is a point of (CG^). thus get two approximate expressions for sin# whose comparison yields /T7T OOo 6^3 ^^^Z 7v77 CG ^ZZ. is surely possible if BB1 be small enough. ' cos T j. ^_(71 (72 (7 will differ B^ = CjC/g. an obvious necessity under the euclidean or hyperbolic hypothesis.I. to find the relations which connect the measures and sides and angles of a right triangle. is 55 a right angle we == cos (~ 6 ) = sin 9. Our next task is a most serious and fundamental one. l Zm a k ~ ^l l "^(f ""  OO. by III.
CH. Next we see that AA^ = BB^ and COSy /L' hence = COS k Y  = JBjJSg. OG 2 B ^^4 2 4 per" Moreover. ^~00. COS f. 2. CC Again. cos k . we see that a line through the middle point.56 THE INTRODUCTION OF .6 u = 6U FIG. by construction C\U2 to AA l from the middle point of (AA 2 ) will be pendicular perpendicular to A^B 2 and the distance of the intersections will differ infinitesimally from each of these expressions . l of (AAi) perpendicular to AA% will also be perpendicular Cl and the distance of the intersections will differ into l A . COST k sm\//r JTT AA. finitesimal ly from Qin\lsAA v We see that C { G^ differs by a higher infinitesimal from sin^cos 7 fc rAA^ T k  so that b . .
multiply through by fe . as is usual in elementary work. getting the usual Pythagorean formula. Begin by assuming that ^AJBC is a right angle so that is We have B= . the whereof merely requires a little analytic skill. fc T =  COS 7 k a. and then put p= now 0. we should 2 cosines in power series. and let us use these same letters. A. (t . 3 cos ~r COS /i I K + bn\ * k ) ' ) ^ = / COS y k = a COS y COS y k k / y OS a C /l r COS A? v 1 y A* ( COS 2 9 2 a i\ T K' 5 / COS \ ' ) 9^1 "7 \J h ~ COS 0<^ A T 9 2 / 2 ^1 COS ~~ .iv TRIGONOMETRIC FORMULAE 6 57 Hence a I) c cos T K cos Y cos 7 1C 1C < c A/' . 6.a. B. sin k . 3 COS 60 2 = k COSy k cos ~ a.w . Let A. O a C f ( ^ 2 'l COS" ~ a k 2 rt = a k COS y i 1 C 3 7 k COS T k . sin a. b c a = sm T sm T k . C be the vertices of a triangle.> CO 9 2 k\j COS 2 o f</ k COS * 7 COS y k k 9. 1 yfc sm y Ic . a sufficient basis for trigonometry. 3 fc sin T . AD G and CD being . COS T 1C = a COS y COS A/ (4) To get develop all the special formula for the euclidean case. development It may not perhaps be entirely a waste of time to work out some of the fundamental formulae. c respectively. to indicate the measures of the corresponding angles. Let D be such a point of (AC) that BD bl and perpendicular to AG\ the measures of 6 2 while the measure of BD is </ a . while the measures of the sides shall bo a. COS 6.a k . . .
Let the reader show that the limit for each series cannot be other than the point itself. ABC a : sin j1C b sin y tv . c'. (5) Let the reader deduce from tan r A/ (4) and (5) that (6) = tan r cos A. AADB as whose we did with the A ABC we shall reach two more sines ratio is % A* and so forth. sin A . . we may connect any vertex with a point of the opposite side by a line perpendicular to the line of that side. C' . A* (7) If none is any triangle. Let us assume that B>^ than We may for the is legitimately assume that A and C are less extreme case under the elliptic hypothesis where such not the fact may easily be treated after the simpler case has been taken up. : sin C. We shall still have sm Let to is . sin j A/ sin A. c T A/ = . A/ cos B= cos f sin A. and we see at once that . : c sin r It = sin A A : sin n n B sm 0. while the measure of $~ABE is A' and that of 4CBE E BE AC. we have just seen in (3) that the limit of this ratio is sin . even when this construction is not possible. and CE be a' b'.58 THE INTRODUCTION OF Now proceeding with the OH. : Let us show that this formula holds universally. b . a r : K sm . . '. be that point of (AC) which makes perpendicular Let the measures of AE. BE.. M . Let us next suppose that A of the angles be greater than a right angle. hence A Now a sin r A/ = .4. Continuing thus we have in (AB) and (AC) two infinite series of points.
a COS r l\j COS y.6 sm y / / ' Q\ /^ "~ _____ (O) sin jl sin sin G let us suppose that no angle of our triangle than a right angle. 8m sinC" . sm y/c c a /c sin y tan . 6' j ^ . sin x > r j sin y sin y c COS T /C = of V COS . sin ^ B . sin y k k' a' + c' = 6 . sin . = . and let D be such a point of greater (BO) that AD is perpendicular to BG Once more is : COS = COST =cos b k COS y k C 008 Jfc r a M&D . sin c y sin ^4. . c' 7: . ~. sin sm y // . c rC > cos C' = tan . x (. tan .. ) = /o r k K .IT TRIGONOMETRIC FORMULAE 59 cos A' =. / (^ c . y = a r In sin (7 = . sin (A + A . a . k /c . a . . c' V COS y/C COS IV y A/ sm . c r cos y sin + cos T a .9 = . c D . a' yfo . = a c cos T cos 7 k k + sm y sin T cos i5. /C . . sin a c y sin A' y A. a . a =1C T sm .
Here.4 cos 2 7 A: sinjB sinC' ' 4 fA sin ^! sin 2 Ocos 2 B 2A 2 sin4 sin C cos 5 cos T /c = C0g 2 (. a' y .r= X db sin. k a c sin . following our previous notation . 2 2 2 2 2 r 2 2 /c /c * I owe this ingenious trigonometric analysis to my former Dunkel. .V sin 2 r cos j. sm . cos Y cos T . be deduced as follows * : Let sm T . a sm r =r . c' a c sin T sin T jfc yfc 6 a k c cos 7 fc . (' r K> . c = 2 . c' tan" _. = k a +c .A sinM . ^ = = 1 . however. k (9) v 7 A correlative T formula may 6 sin T .sin 2 5 2 1 = sin A sin 6 sin T + 2 sin ^1 sin G cos 5 cos k k A sin G' + sin A sin (7 = sin A sin 6 cos 7 2 sin . sin c sin y sin k a r sin T k 7 a cos r /c = V cos c' T /c cos T A . Otto . .A 2 2 sin 2 6 + A sinMsin 2 0. B> <l this proof is invalid. c .r*/\ 4 C") ' = c y cos r cos j A? . ^ _ COS o 2 a A' A. pupil Dr. > COS T /o = cos V cos a' 7 7? A. . f 4 sin 2 4 + sin sin 2 2 (7.4 sin C cos 7 cos 1? + cos 5.60 THE INTRODUCTION OF If OH. a' ri\> sm r \\j .cos rC rC . 2 f 2 7 . . cos J5 = / < / cos x (A r . a' . Jf 5 a .sin A! A* b cos T = k c a a c cos 7 cos 7 + sm T sin T cos k k k . B. .
= cos A cos C f sin 4 sin cos . lc Ic k The proof of this is left to the reader. our formulae pass over into those for the euclidean plane. h sin ivtZTJ = sin . B= ir C. * that case Iu finding this formula wo have extracted a square root. as well as the task of showing that the formulae which we have here established are identical with those for a euclidean sphere of radius k. Let him also show that when Ic =0. we have but to consider the limiting A = 0. the right angles being at A and J?.iv TRIGONOMETRIC FORMULAE cos 61 A cos G = cosJS cos T sin rC A sin (7 cos B. cos = cos k MAC K j wRD cos 7 cos wAB 7 . . To he sure wo have taken the right sign.* 7 A* (10) If A BCD be an isosceles birectangular quadrilateral.
however. where this distinction is made between the two points shall be called a directed distance. let us imagine that a distinction is made between the two points. The directed distance from A to 5 shall be written greater than. other congruent transformations of the line into itself besides reflections. into A' and B'. Conversely. AB. also transforms f case the middle point of (AA ) will remain invariant. tances. there will be no invariant point on the line. Leaving aside the trivial case of the null distance.CHAPTER V ANALYTIC FORMULAE AT the beginning of Chapter I we posited the existence of two undefined objects. Such a transformation shall be called a reflection in this middle point. we easily see that a congruent transformation whereby goes into and one other point of (A A') also goes into a point of that segment. These concepts may be further sharpened as follows. the directed distance from the initial to the terminal point. the extremities of every segment having this middle point will be interchanged. more specifically. For if go into A\ and f will any point of (AA ) go into a point not of (AA')> then be the only point of (AA') which goes into a point thereof. The concept distance. determine two directed distances. There are. In this relation the two points entered symmetrically. is a reflection in the middle point of the segment. A A . It is at once evident that every congruent transformation A B A A A '. points and distances. and less than. The relations congruent to when applied to directed dis mean that the corresponding distances have these Suppose that we have two congruent segments (AB) and It may be that a congruent trans(A'B') of the same line. Between the two existed the relation that the existence of two points implied the existence of a single object. the one being called the initial and the other the terminal point. and transforms f and In this into A. shall relations. and we have a different form of congruent transformation called a translation. Any not null distance will. or. formation which carries the line into itself. their distance. thus.
the we or same sense. If then. Suppose. and a second whereby A' goes into A".CH. Otherwise. The inverse of a translation is another translation . P P P P P y i Definition. is a translation. Suppose. a sense like (opposite) to that of one is like (opposite) to that of the other. We wish to show that the product of these two is not a reflection. next. also say that AB and A B' have AB and A'C' have the same sense. We see. and not a translation. the inverse of a reflection is the reflection itself. If A were a point of (A'A"). they shall be said to have opposite senses. and if two directed distances have like senses. (A A') P2 would belong to the extension of (A' A") beyond A'. The two directed distances determined by a given distance have opposite senses. like senses. have like senses to one another. that every congruent transformation has an inverse. to begin with. Let the reader show that the product of a reflection and a translation is a reflection. The product of two translations The assemblage of all translations is a group. in fact. and hence If A" were a point of of (A'A")> leading to the same fallacy.V ANALYTIC FORMULAE 63 of the line into itself is either a reflection or a translation. A'C' upon the same line. and that the product of two reflections is a translation. There will then (XIII) be a single such point ff of (A'C') that AB = A'B shall f . so that A'C' > AB. Two directed distances which have like or opposite senses to a third. if the congruent transformation which carries the initial and terminal points of the one into the initial and terminal points of the other be a translation. . Theorem 1. the of (AA") close to A". suppose that we have a translation whereby A goes into A'. They shall be said to have opposite senses if this transformation be a reflection. A point l of (AA'f ) close to A must then go into another If A' be a point of (A A"). Two congruent directed distances of the same line shall be said to have the same sense. that it were. that we have two noncongruent directed distances AB. The group theorem for translations gives at once Theorem 3. 3 point into a point of (A'A") and first translation will carry l 3 is also a point of (A' A") the second transformation as 3 would be a reflection. This premised. and P3 would belong to (A'A") and not to (AA"). The following theorem is obvious Theorem 2. P2 would be a point of (AA').
64 while ANALYTIC FORMULAE OH. If. OF making a right Their lines shall naturally be called the coordinate angle. every We D of the line will be completely determined point coordinate > P by a single x = sin MOP . with the further equation . If. while   the measure of being are a and ft respectively. Let P be any point of the plane. . YOP = k sin y r\ = ksm j cos/3. and nearly all that is necessary to set up coordinates in space. if they have opposite senses. Let us begin with the plane. further. (1) CO P = COS T . If in %~ABO. Opposite directed distances of the same line shall have measures with opposite algebraic signs. Wo may /c while those of _XOP and then put cos a. to directed angle. group of congruent transformation which keep the vertex or edge invariant. Let us now make suitable conventions for the measurement shall take for the absolute value of directed distances. We have at last elaborated all of the machinery necessary to set up a coordinate system in the plane. is the origin. the measure of the corresponding distance. the resulting directed angle shall be written 4ABC. of the measure of a directed distance. OP p. We thus arrive at the concept for sense of an angle. In an entirely similar spirit we may enlarge our concepts of the system for halflines or halfplanes of common bound. We choose an initial and a terminal side or face. and define as rotations a certain one parameter. we choose a fixed origin upon a line and a fixed unit for directed distances. axes. we assign the measure for a single directed distance of a line. then. and choose two halflines OJT. j 1C and dihedral angle. and set up a coordinate angle. AH be taken as initial side. that of every other directed distance thereof is uniquely determined. a sense like (opposite) to that of one is opposite (like) to that of the other.
for their distance would be /CTT. same point. . CD = 1..e. 77. for the distance of two points . so that the radical must have the In the elliptic case it is not possible to have same sign as XQ the other with two points. case. P and P' with j tc p = cos y cos fc p' jic 4 sin p jfc . the radical must have a welldefined sign in order that equations should give a point of our space. The formula coordinates cos (#). co should refer to the f. co and 77. The sign of the radical in the numerator of (4). should be so taken as to give a positive The . which is contrary to Axiom II. ?. would be straight. well determined. _ (4) signs of the radicals in the denominators are. : 65 ( homogeneous In practice it is better to use in place of coordinates defined as follows  . (of) is 77 = p cos /3. In the limiting parabolic case = p cos a. a) C03 . they might be connected by many On the other hand.v ANALYTIC FORMULAE f. for 7. as we have seen. then that point would determine with itself two distinct Hence.. _ o _ . it is not possible that straight lines. n%Cj ~ = What shall we say as to the signs to be attached to the In the hyperbolic radicals appearing in these denominators ? case co is essentially positive. i. . and the opposite angle of every triangle containing them both . one with the coordinates ?. sin y cos (a /c p' . in every distances. co and the coordinates co.
In the euclidean case MPP = 7 Returning to (4) and putting infinitesimal element of arc or' x^ + dx^ we get for the dx l n A rut dx<2 x i  .0) (xc?cc + ydy) dz + (x 2 + y 2 v2 = ds2 . / y = ds* == In the limiting euclidean case ^ ds (5) as follows let 2 0> 2 . that of every other is completely determined. after the adjunction of the value o'f the sign of a single directed distance. y = 3 x ' 7 x + dx> . then.^ f T/^ If  3 y A. = dx + dy 2 Returning to the general : case.  A: z u' +v 1 2z + 2 (fc . distances Should we seek the measures of directed PP'. we may improve our formula dz z = VFT^T^/^ d .66 ANALYTIC FORMULAE on the line CH. = V k* H. value to the whole.
as we shall introduce them merely through the expressions cos a. These shall be called the direction cosines of the halfshall be the origin. in halflines. 1868. cos/3. and Saggio d'interpretazione della geometria noneuclidca '. ' vol. We shall imagine that OP and OP' are * The idea of interpreting the noneuclidean piano as a surface of constant curvature in euclidean spare must certainly have been present to Riemann's mind. vi.    whose measures shall be a.ZOP fact. due to Beltrami. cos y'. cos y. XV We and XIX * shall return to questions of this sort in Chapters of this work. The credit for first setting the matter in a clear light is. while the planes determined by them are the coordinate Take a second halfline OP'. Serie 2. and OX. and a 4XOP. vol. These angles are not directed. Suppose. 4YOP. Giornale di Matematichc. but this will cause no inconvenience. The angles that we have three OX. with direction cosines planes. Let us now take up coordinates in three dimensions.   cos a'. OZ the coordinate line. ii. Annali di Matematica. OF. mutually perpendicular fourth halfline  OP. E2 . cit. 1868. sHall be called the direction angles of the halfline OP. 1 rt 2/fc =[' 4F~J Theorem 4. axes. loc. OF. 4. See his 'Teoria fondamentalo degli spazii di curvatura costante'. y respectively. Wo must make some preliminary remarks about the direction cosines of a halfline. The noneuclidean plane may be developed upon a surface of constant curvature p in euclidean space. p. cos /3'. however.ANALYTIC FORMULAE [ 67 Comparing this with the usual distance formula Now if K be the this distance 1 measure of curvature of the surface having formula S^lVVE V*. OZ.
(10) We then write (ll) . cosetcosa'f cos/3 cos /3'f. (9) = K am . perpendiculars to the axes through f Let Q' be them. C' bearing the same relation to that point of OP' which makes %PQ'O a right angle. &c. in the infinitesimal domain obeys the euclidean hypothesis. T/. and A'. B'. circumstances. as before. But (xy) pass. In the same s P int But clearly MO A = M OP cos a f . ANALYTIC FORMULAE Under these CH.cosy cosy'. G where  P P . . . cos a.B. infinitesimal as compared with MOP. 7 k sin MOP .  j A/ cos y. Hence MOPcosfl = M OP [cos a cos a' f cos p cos/3' f cosy cosy'] + M OP. From x : : these : we xl X 2 <# 3 . we may find meet A. to homogeneous coordinates first we shall introduce a new symbol : =x y + x yl + x2 y2 + x. hence we have = the c is MOPcos0f *.^ l . rj = 7 Ic  sin fc ^ cos ^3. and let Now we know that geometry 4_ POP' have a measure 0. or dividing out cos0 = (7) In particular we shall have 2 2 2 1 = cos a + cos /3 + cos y (8) : We now set up our coordinate system as follows 0> = COS = .68 infinitesimal.
This distinction between coordinates (x) and coordinates (x) shall be consistently maintained in the hyperbolic case. one We see. so enlarged as to carry into itself every halfplane through that axis. We . new coordinates (x'). The cosine of the measure of distance of two points (x) and We see at once that we shall have (y) is easily found. f 2 2 2 has a constant value. . i. and every halfplane with this axis as bound into itself. We see at once that Let us * now see what effect the new direction cosines.v ANALYTIC FORMULAE 69 Here. d > f sin r . Let us next suppose that we have a congruent transformawhich carries the planes = and r. and x is a pure imaginary. as in the case of the plane. To real. The are thus linearly transformed in such a way variables 17.. that by this transformation every point receives just the coordinates that it would obtain by a translation of the axis OZ into itself through a distance d. 2 a> are in fact.77 f f Hence x Q9 x l oc2 . will be linear functions of the old ones for a plane through the origin will be characterized by a linear relation connecting the direction cosines of the halflines with that bound. and so the . they are subjected to an orthogonal . in the coordinates (x). d . . = into themselves.e. V(xx) V(yy) a congruent transformation will have upon our coordinates. we must have (xx) < 0. modification which we shall occasionally make. First take a congruent transformation keeping the origin invariant. substitution. that in the hyperbolic case. There is. remedy this let us write KXQ = XQ. and this group may be represented by tion . that 4. . d smr +CCOST* see. # 3 are linearly transformed so that (xx) is an invariant (relative). while o> is unaltered. since & < ?/. The assemblage of all such transformations will form a oneparameter group. Once more wo find that. in fact. a) = o> cos  d f . however. there is no ambiguity arising from the double sign of the radical. X^ = & 15 X% =# 2J ^3 == ^V A point will now have real coordinates.
70 this will ANALYTIC FOKMULAE be an orthogonal substitution. . Its equation is y 2 x i 2/1^2 = 0. ' sides of this equation are absolute invariants for all congruent transformations. and we shall use them to find expressions for the distance from a point to a plane and the The existence of the former of angle between two planes. : : : Theorem . in general : Once more we have an invariant form. meets it. Now. and as we may pass from one of these to any other plane by linear transformations. V(xx) V(xx) V(uu) so that. We see that (xy). the distance from the point to the find the distance We plane. these quantities is contingent upon the existence of a point in the plane determining with the given point a line perpendicular to Let the with the cosines of are the x% plane (u) be that which connects the axis xl = x2 = The point (y). in general : ~~ But both from a point to a plane in the same Let the point be (x) and d the distance thence to the way. : Every congruent transformation of space is represented by an orthogonal substitution in the homogeneous variables # x l x 2 x 3 In Chapter VIII we shall make a detailed study of these congruent transformations. For the present. (ux). (uv) are concomitants of every congruent transformation. the angles which this makes with the plane v 1 x 1 = direction cosines of the 0. Hence. we may write. CH. lot us begin by noticing that the coordinate planes have linear equations. by definition. the measure of the angle be we have COS = fJiL== = ^ V(uu) V(yv) = JL^. every congruent transformation of space may be compounded out Hence of transformations of these two types. d sin* =  * = ~~ + d ' V V n u ii <r . If then. the plane. two halflines of OP. sin 7 J = (ux) L ' T (14) v(uu) v(xx) . so the equation of any plane may be written 5. point where a perpendicular to the plane u l x 1 = this being. lastly.
but have dealt with the geometry of such a region as the inside of a sphere. and what meaning shall we attach to coordinates to which no point corresponds? We must also adjoin the com plex domain for coordinates. extremity. must show that the system of axioms which has carried us safely so far.e. and give a new interpretation to our fundamental formulae (12). Yes. for in elliptic space no distance may have a measure kir under our axioms. a system of trigonometry. Then only shall we be able to continue our subject in the broadest and most scientific spirit. i. will not break down later. At present our task is different.v ANALYTIC FORMULAE 71 of </(xx) is determined. The matter may be otherwise stated. (14) covering the most general case. As for that of v/(uu). we should have run into a difficulty. Every point will have a set of coordinates in our system. . rendering trebly difficult many a proof and definition. that these axioms are essen We We must also grapple with a disadvantage which has weighed heavily upon us from the start. All of these things will be of The sign reversing We use later. In Axiom XI we assumed that any segment might be extended beyond either tially compatible. What is the extreme limit of possibility for making points correspond to coordinate sets. we get opposite directed distances of the same line. not including the surface. had we assumed that every segment might be extended a given amount. but how far may it be so extended ? This question we have not attempted to answer. Our system of axioms has given us a large body of elementary doctrine. and a system of analytic geometry wherein the fundamental metrical invariants are easily expressed. In fact. by it. have now reached the end of the first stage of our journey. (13).
any system of geometry corresponding to these formulae will be of the euclidean type. and employ its terminology. How shall we show that those which we made at the outset are. conversely. . namely. By the distance of two the last chapter fundamental question suggested at the close of points we shall mean the positive value of the expression The sum sense.CHAPTER VI CONSISTENCY A SIGNIFICANCE OF THE AXIOMS THE first was this. is this. We shall take as our may . assumptions mutually consistent ? We need not here go into that elusive question which bothers the modern student of pure logic. Let us now exhibit the existence of a system of geometry obeying the hyperbolic hypothesis. If we take our fundamental dis assumptions and the euclidean hypothesis. y> z. of two distances shall be defined in the arithmetical It is a perfectly straightforward piece of algebra to . The elementary geometry of Euclid fulfils In what immediately follows we shall these conditions. Let us begin with the geometry of the euclidean hypothesis. bhow that such a system of objects will obey all of our axioms and the euclidean hypothesis hence the consistency of our axioms rests upon the consistency of the number system. and take as points any class of objects which may be put into one to one correspondence with all triads of values of three real independent variables x. to familiar sets of objects which do actually fulfil our point fundamental laws. points and be put into one to one correspondence with expressions of the above types and. whether any set of assumptions can ever be shown All that we shall undertake to do is to to be consistent. and that we may take as indubitable. Be it noticed that we have another system of objects which obey all of our axioms if we make the further assumption that The net tances result. assume this geometry as known. in truth. so far.
but not upon. If we interpret the equatorial plane as the Gauss plane. Let us take for outclass of points the assemblage of all points of a euclidean sphere which are south of the equatorial circle. passes through the north pole. the x5 'I/ natural logarithm of a cross ratio which they determine with the intersection of their line with the imaginary surface By a proper choice of the cross ratio and logarithm. . we see that we must be under the hyperbolic hypothesis. or rather so much of this group as will carry at least one of the southern hemisphere into another such point. we see that the congruent group will be A ^=as + /3. The We times. vi class of points the lie AXIOMS 73 assemblage of all points in euclidean space within. We may obtain a simultaneous bird'seye view of our three systems in two dimensions as follows. yet there is an infinite number of lines through a given point. and as the distance of two points . Lastly. aa=l. that we are under the euclidean hypothesis. hypothesis will prevail. shall mean by the distance of two points one half the real logarithm of the numerically larger of the two cross ratios which they make with the intersections of their line with The reader familiar with projective geometry the sphere. take as points elliptic case is treated similarly. this The congruent expression may be made positive. group will bo so much of the orthogonal group as carries at least one point within our sphere into another such point. which yet do not which We meet it. b is Eointevident from the conforinal nature of the transformation from sphere to equatorial plane. a sphere of radius unity.CH. will see that the segment of two points in the noneuclidean sense will be coextensive with their segment in the euclidean sense. the assemblage of all points within a euclidean sphere of small radius. We shall define the distance of two points in three successive different elliptic The ways (a) : The distance of two points shall be defined as the distance which the lines connecting them with the north pole line will be a circle which cut on the equatorial plane. as before. for a line is infinitely long. for two coplanar lines perpendicular to a third will tend to approach one another. and the congruent group will be the group of collineations which carry this sphere into itself. coplanar with a given line.
74,
CONSISTENCY A SIGNIFICANCE
CH.
(6) The distance of two points shall be defined as one half the logarithm of the cross ratio on the circle through them in a vertical plane which they determine with the two intersections of this circle and the equator. line here will be the arc of such a circle. The congruent group will be that group of (euclidean) collineations which carries into line will be infinitely itself the southern hemisphere. there will be an infinite number of others through long, yet i. e. we are under any chosen point failing to meet it the hyperbolic hypothesis. (c) The distance of two points shall be defined as the length of the arc of their great circle. Noneuclidean lines will be arcs of great circles. Congruent transformations will be rotations of the* sphere, and it is easy to see that the sum of the angles of a triangle is greater than a straight angle we are under the elliptic hypothesis. We have now shown that our system of axioms is sufficient, for we have been able to introduce coordinates for our points, and analytic expressions for distances and angles. The axioms are also compatible, for we have found actual systems of objects obeying them. Compared with these virtues, all other It will, qualities of a system of axioms are of small import. however, throw considerable light upon the significance of these our axioms, if we examine in part, their mutual independence, by examining the nature of those geometrical systems where first one, and then another of our assumptions is supposed not to hold. Axiom XIX is popularly known as the axiom of free mobility, or rather, it is the residue of that axiom when we It puts into precise shape are confined to a limited space. the statement that figures may be moved about freely without We have defined suffering an alteration either in size or form. congruent transformations by means of the relation congruent which is itself defined in the logical sense, but not deWe might, of course, have proceeded in the scriptively. reverse order.* The ordinary conception in the elementary textbooks seems to be that two figures are congruent if they may be superposed ; superposed means that they may be carried from place to place without losing size or shape, and this in turn implies that throughout the transference, each
A
A
;
;
remains congruent to itself, f With regard to the independence of this axiom, we have but
* Cf. Fieri, loc. cit. Cf. Veronese, loc. cit., p. 259, note 1, matics, vol. i, Cambridge, 1903, p. 405.
t
and
Russell, The Principles of Mathe
VT
to look at
is
OF THE AXIOMS
75
any system where the measure of distance in one double that of all the rest of space. A triangle having piano two vertices in this plane, and one elsewhere, could not be congruently transformed into a triangle of a different sort.
Axiom XVIII is the axiom of continuity. have laid special stress on it in the course of our work, although the subject of elementary geometry may be pushed very far
without
We
its aid.* We are not here concerned with the question of the wisdom of such attempts, considered from the didactic point of view. Systems of geometry where this axiom does not hold will occur to every reader; e.g. the Cartesian euclidean system where all points whose coordinates It is interesting to note that are nonalgebraic are omitted. whereas the omission of XIX runs directly counter to our sense experience, no amount of observation could tell us whether or no our geometry were continuous, f Axiom XVII is an existence theorem, not holding where It is a very the geometry of the plane is alone considered. curious fact that the projective geometry of the plane is not entirely independent of that of space, for Desargues' theorem that copolar triangles are also coaxal cannot be proved without the aid either of a third dimension, or of the con
gruent group. J
gives a criterion for circumstances under which It is evident that must necessarily intersect. without some such criterion we should have difficulty in proceeding any distance at all among the descriptive proIt is difficult to show the independence perties of a plane. The only dense system of geometry known of this axiom. to the writer where it is untrue is the following. Let us denote by R the class of all rational numbers whose
Axiom XVI
lines
two
denominators are of the form
where a. and b+ are integers or one may be zero. Let us take as points the assemblage of all points of the euclidean plane whose Cartesian coordinates are rational numbers of the class E. The whole field will be transported into itself by a parallel translation from any one point to any f other. Moreover, let #, y and x\ y be the coordinates of two
* Cf. Halsted, loc. cit. Cf. K. L. Moore, loc. cit. t Cf. Hilbert, loc. cit, p. 70
Moulton, A simple nondesarguesian plane geometry,' Transactions of the American Mathematical Society, vol. iii, 1902 ; Yahlen, loc. cit., p. 67. Cf. Levy, loc. cit., p. 32.
'
;
f
76
CONSISTENCY AND SIGNIFICANCE
f
CH. vi
2 points of the class, where x in fact that
y
2
= #'
2 4 2/'
2
.
We may imagine
Then the
cosine
and
sine of the angle
which the two points
subtend at the origin will be respectively
pp' 4 qq'
pq'
p'q
p
2
f
q*
and these are numbers of the class R. The whole field will go into itself by a rotation about the origin. Our system It is of course twodimensional will, therefore, obey XIX. and not continuous. Moreover XVI will not hold, as the reader will see by easily devised numerical experiments. There are, also, plenty of geometries of a finite number of points where this axiom does not hold.* Axiom XV is, of course, an existence theorem, untrue in the geometry of a single lino.
lines.
Axiom XIV gives the fundamental property of straight As an example of a geometry where it does not hold,
us consider the assemblage of all points within a sphere of radius one, and define as the distance of two points the length of an arc of a circle of radius two which connects them. The segment of two points is thus a cigarshaped region see that the extensions of such a segconnecting them. ment and the segment itself do not comprise the segment of two points within the original, and the extensions of the latter. Axioms XII and XIII are also in abeyance, and ifc seems possible that these three axioms are not mutually independent. The present writer is unable to answer this
let
We
question.
Axiom XI implies that space has no boundary, and will be untrue of the geometry within and on a sphere. The first ten axioms amount to saying that distances are magnitudes among which subtraction is always possible, but addition only under restriction.
*
Veblon,
loc. cit., pp.
85051.
CHAPTER
VII
THE GEOMETRIC AND ANALYTIC EXTENSION
OF SPACE
WE are now in a position to take up the second of those fundamental questions which we proposed at the close of Chapter V, namely, to determine what degree of precision
be given to Axiom XI. This axiom tells us that, popularly speaking, any segment may be extended beyond How far may it be so extended? Are we able either end. to state that there exists a system of geometry, consistent with our axioms, where any segment may be extended by
may
any chosen amount? Or, in more precise language, and PQ be given, can we always find C so that
if
AB
case,
are already able to answer this question in the euclidean it affirmatively. have seen that there is no inconsistency in that system of geometry, where points are in one to one correspondence with all triads of (real and finite) values of three coordinates #, y, 0, and where distances are given by the positive values of expressions of the form
We
and answer
We
Here,
if,
as
we have
said,
we
restrict the values of x, y, z
merely to be real and finite, we have a space under the euclidean hypothesis, where any segment may be extended beyond either extremity by any desired amount. Such a space shall be called euclidean space. The same result will hold in the hyperbolic case. We shall have a consistent geometrical system if we assume that our points are in one to one correspondence with values
<0, + xf + xf + x* <0. Wxf
XQ'.^.X^IX^
&2
Here, also, there will exist on every line distances whose measures will be as large as we please. The space under the
78
THE GEOMETRIC AND ANALYTIC
OH.
by any chosen amount
bolic
hyperbolic hypothesis, where any segment may be extended shall be called hyperbolic space. To put the matter otherwise, we shall have euclidean or hyper:
geometry if we replace Axiom XII by AXIOM Xir. If the parabolic or hyperbolic hypothesis be true, and if AB and PQ be any two distances, then there will exist a single point (7, such that
turn to the elliptic case, we find a decidedly Suppose, in fact, that there is a one to one correspondence between the assemblage of all points, and all sets of real values XQ x l & 2 x. r The distance of two points will depend upon the periodic function
different state of affairs.
:
When we
:
:
cos
_
.
minimum positive to avoid ambiguity, we value should be taken for this expression, we should easily find two not null distances, whose sum was a null distance, which would be in disagreement with Axiom X. The desideratum is this. To find a system of geometry where each point belongs to a subclass subject to Axioms IXIX, and the elliptic hypothesis, and where each segment may still be extended by any chosen amount, beyond either end.
If,
V(xx) V(yy) assume that the
AXIOM
least
I.
There exists a class of objects, containing at
points.
two members, called
AXIOM II'. Every point belongs Axioms IXIX.
Definition.
region.
to a subclass obeying
Any
such subclass shall be called a consistent
AXIOM
III'.
common
point, have a
Any two consistent regions which have a common consistent region including
this point and all others determining therewith a sufficiently small, not null, distance.
AXIOM IV. If P and Pn+l be any two points there may be found a finite number n of points Plf P 2 P3 .Pn possessing the property that each set of three successive ones belong to a consistent region, and P^ is within the segment (Pj ~i JPfc+i)
,
. .
c
Definition.
and a line.
of all points of such segments, all possible successive extensions thereof shall be called
The assemblage
vii
EXTENSION OF SPACE
79
points
important implication of the last axiom is that any two may be connected (conceivably in many ways) by a chain of consistent regions, where each successive pair have a consistent subregion in common. This shows that if we set up a coordinate system like that of Chapter V in any consistent region, we may, by a process of analytic extension, reach a set of coordinates for every point in space. We may also compare any two distances. We have merely to take as unit of measure for one, a distance so small, that a distance congruent therewith shall exist in the first three a distance congruent with overlapping consistent regions this in the second three and so on to the last region, and then compare the measures of the two distances in terms of the first unit of measure, and the unit obtained from this by the Let the reader show that series of congruent transformations.
;
An
>
if
once
we
find
AB == PQ
<
the
same
relation will hold if
wo
proceed by any other string of overlapping regions. Having thus defined the congruence of any two distances, we may state our axiom for the extension of a segment, as follows
:
PQ be any two distances, there a single point G such that EC = PQ, while B is exists within a segment whose extremities are C and a point of(AB). An important corollary from this axiom is that there must exist in the elliptic case a point having any chosen set of homogeneous coordinates (x) not all zero. For, let (y) be Consider the line the coordinates of any known point. through it whose points have coordinates of the form
If
AXIOM V.
AB
and
\(y)+(ji(x).
As we proceed along this line, the ratio  will always change in the same sense, for such will be the case in any particular consistent region. Moreover we may, by our last axiom, find a number of successive points such that
the
sum
of the measures
first
of their
distances
shall
be
kir.
Between the
and
last of these points the value of  will
,
M have run continuously through all values from co to oo and hence have passed through the value 0, giving a point with the required coordinates. The preceding paragraph suggests two interesting questions. Is it possible that, by varying the method of analytic extension, we might give to any point two different sets of
and the coordinates of found directly as in Chap ter V. Proceeding thus. as a result of this.80 THE GEOMETRIC AND ANALYTIC en. as we shall see at more length in Chapter XVII. a sufficiently small congruent transformation of the one may be enlarged to be a congruent transformation of the other. if the connexion be made by means of a different succession of overlapping consistent regions ? It is make impossible to answer this question a priori the following explicit assumption : . if we take any two consistent regions of space. Will the original transformation give rise to the same transformation in the second space. A congruent transformation of any consistent region may be enlarged in a single way to be a congruent transformation of every point. it is not possible that every infinitesimal congruent transformation which keeps invariant shall also keep Q invariant. and carrying Q to an infinitesiinally near point Q'. while those of Q are found by an analytic extension through a chain of overlapping consistent regions. We reach the desired limitation sufficiently small congruent transformation of any consistent region will effect a congruent transformation of any A by means of the following considerations. But in . There will be no limitation involved in assuming that the coordinate axes were P this consistent region. and so of any consistent region including It thus appears that if two consistent regions this latter. For the present it is. Suppose. have a common subregion. however. so that a transformation of this sort may be found set up in P Now P transforming each overlapping consistent region infinitesiinally. we therefore AXIOM VI'. a congruent transformation of one consistent region can be enlarged in only one way to be a congruent transformation of any other. chosen subregion. it may have several different sets of coordinates. and connect them by a series of overlapping consistent regions. homogeneous coordinates in the same system ? Is it possible that two different points should have the same homogeneous to the first of these questions. that sistent region have the coordinates (x). Let us next observe that it is impossible that two points of the same consistent region should have the same coordinates in any and Q of a consystem. then a small congruent transformation of the one may be analytically extended to operate a congruent transformation in the other. wiser to limit coordinates'? is a fact that With regard under our hypotheses a point ourselves to the classical noneuclidean systems. on the contrary. where a point has a unique set of coordinates. Evidently.
. Let us consider the assemblage of all points whose coordinates are linearly dependent on those of three noncollinear This assemblage of points may properly be called points. This proof is independent of Axiom VI'. the coordinates of points on each being represented in the form \y i + fjLXi The two lines are identical. For let a point Q on a line through one of the points have coordinates (y). or transform them infinitesimally. the ratios of the coordinates (XQ) must be unaltered by every infinitesimal orthogonal substitution which leaves (x) invariant. But there is a point distinct from and close to it which has the coordinates (x"). and will contain every line clearly whereof it contains two nonequivalent points.vii EXTENSION OF SPACE 81 the analytic expression of this transformation. conversely. and by the same chain of extensions as gave these coordinates to Q. Every line through one of these points will pass through the other. But this is impossible for the coordinate system explained in Chapter V. will either keep (x') invariant. (t) be the coordinates of three points. let us say. Let us call two such points equivalent. so that Q" will also have the coordinates (x). Let (?/). ir '= px$. a connex assemblage. when we set up a coordinate system in the last consistent region. Hence. i. in the form of an orthogonal substitution (in the noneuclidean cases) the values (x) will be invariant. and the two lines through (Q) lie in part in a consistent region.e. For. which we have just seen to be impossible. these two points will have the same coordinates. They may not lie in the same consistent region. for a consistent region gives distinct coordinates to distinct points. to a set of values (#"). that if each point have but one set of coordinates. that which includes Q and (/. Hence. by supposing that two distinct points shall have the same homogeneous coordinates. no two of which are Our P P We . for those points thereof which lie in any consistent It is region will lie in a plane as defined in Chapter II. It is evident. and every congruent transformation which leaves one invariant. and this gives two points of a consistent region with these coordinates. (s). It is time to attack the other question proposed above. desired uniqueness of coordinate sets will follow at once from the foregoing. a plane. Every infinitesimal transformation which keeps the values (x) invariant. will leave the other unmoved also. may connect it with the other by a line. not proportional to one another. suppose that a point have two sets of coordinate values (x) and (a/). reversing the order of extensions. Axiom VI' must surely hold.
the sum of the measures of the angles of all the triangles and quadrilaterals will be less than. and our new Axioms I. in fact. in case the determinant of the original orthogonal substitution is negative. We may therefore pass from to any equivalent point by a transformation which leaves in place every point of a plane. P and Connect them by a line whereon are n+l were equivalent. and (xX) is an points satisfying the condition (xX) invariant under every orthogonal substitution. a space obeying Axioms IVI'. then. have but one equivalent at most.82 THE GEOMETRIC AND ANALYTIC Let us consider the point (ux) (x) CH. and If n successive segments whose .TT. as we have exists. But clearly the sum of the measures of TT t) (71 the angles at points P^ and P/ is 2nir. of Hence every point in space can course. there might not be any such point. for it constitutes the assemblage of all 0. are whose coordinates = \uyzt\. It is clear. at the end. however. Let us next make a congruent transformation whereby '. P/ w '. But there is only one congruent transformation of space which leaves every point of a plane invariant. or equal to 1 2 TT f. quadrilaterals. Under the elliptic hypothesis there are two possibilities . . . each too small to change P' to an equivalent point. as follows. In the elliptic case. yet keeping the values (x) invariant.. Suppose. there is but one point for each set of coordinates. with a reflection in a plane perpendicular to the given one. equivalent. so that the sum of the two angles which the two lines make at and w+1 is null an absurd result. the goes into an equivalent point plane of (y) (z) (t) goes into itself congruently. P P P P P : Elliptic space. This is the elliptic hypothesis. P seen. that in these cases. After has been carried to P'. the identical one.P W Pj. such a point surely In the hyperbolic or parabolic cases. and this too will leave P' unchanged.. and as we are under the hyperbolic or euclidean hypothesis. besides. P P We P P . Move this line slightly so that the connecting 2 . Our results are.VI' will yield us nothing more than euclidean or hyperbolic space. and (n 1) points. very near to the former string have constructed two triangles. Equivalent points can then or negative occur only under the elliptic hypothesis. and there will surely be a point with the coordinates (x) above. there can be no equivalent points. of points are P/. each point of the plane may be returned to its original position by means of a series of congruent transformations. coupled. Under the euclidean and hyperbolic hypotheses. .
let us take as points concurrent lines of a four dimensional space (euclidean.vii EXTENSION OF SPACE  83 measures are last n be taken upon a lino as indicated in V' the r will the extremity of the last segment extremity of the first. = r^=~ =^ For instance. Spherical space. be called halflines. Any two noiv . n successive also a space obeying Axioms IVF Each point will have one equivacongruent distances bo taken upon is  a line whose measures are the last extremity of the last will be equivalent to the first extremity of the first. Two . the details to the reader. be modified plane. Two will have one and only one common has an equivalent. If This elliptic hypothesis. The measure of the distance of two equivalent points shall be defined as the number kir. so that no point a consistent region the assemblage of all points whose distances from a given point . with propriety. The definition of an interior angle given in Chapter II may be retained. We leave may also. but the concept of halfplane is illusory. An example of elliptic geometry will be furnished by any set of points in one to one correspondence with all sets of homogeneous values x xl a" 2 x : : : {) :i where also cos . much as we have modified the definition of a halfline. however. and from it a definition built up for a dihedral angle. for example) and mean by distance the measure of the angle ^ ^ formed and the lent. unless the measure of their distance is in & which case they determine two segments with the same segments These last two extremities. We may take as first be identical with lines of the same plane point. may take as a consistent region the assemblage of all points the measures of whose distances from a given point are less We than . points will always have a determinate distance and a single segment. are of measure less than If 4 two points be of such a nature that the expression for the cosine of the measure of the fcth part of their distance vanishes. by two lines. for a line will not divide the It may. we shall say that the measure of their distance is ~.
two planes which meet in an ideal These definitions of parallel are for euclidean space line. Now a set of values y l y 2 ?/3 will determine at each real point (x) a line. while the ideal plane is called the plane at infinity. not all zero. as also. A two dimensional spherical geometry is clearly offered by the euclidean sphere. to be parallel. Ideal points and lines shall also be called infinitely distant. and assign to Two ideal points will determine a it the coordinates (y). an ideal Two lines whose intersection is ideal shall be said line. in that there is in the first two cases always a point to correspond with every set of real values. we get a second line coplanar with the first. The elliptic and spherical spaces which we have thus built up are. of the limiting great circle are considered as identical. This we shall call the equation of the ideal plane which is supposed to consist of the assemblage of all ideal points. that may be attached to our four homogeneous coorWe bring dinates #. A simple example of a two dimensional elliptic geometry is offered by the euclidean hemisphere. may therefore call the bundle an ideal point. the length of the shorter arc of a great circle connecting them. and so for halfplane.84 THE GEOMETRIC AND ANALYTIC We OH. where opposite points : : . may equivalent points will have a welldefined segment. while in the latter cases this is not so. find a definition for a halfline analogous to that given in the elliptic case. or have in common. and dihedral angle. the coordinates of whose points are of the form Xy^ + ^x^ and if (x) be varied off of this line. internal angle. : : : () : : : & t u. more complete than euclidean or hyperbolic space. and this will have exactly the same We descriptive properties as a bundle of concurrent lines. We shall therefore say that they determine. meaning by the distance of two points. Let us begin with the euclidean case where there is a point corresponding to every real set of homogeneous values a? x l # 2 # 3 provided that x 9^ 0. We shall . Our coordinates y^ y% ?/3 will thus serve to determine a bundle of lines. our euclidean and hyperbolic geometries up to an equality with the others by extending our concept paint. as pencil of planes having the same descriptive properties a pencil of planes through a common line. in one respect. the equation ___ n by : An example of spherical geometry will be furnished by the geometry of a hypersphere in four dimensional eucHdean space. The assemblage of all ideal points will be characterized only.
and plane to cover both ideal elements and those previously denned. any two of which are coplanar. for which this inequality does not hold. There will thus be a plane corresponding to each set of real homogeneous coordinates (u) not all zero.vii EXTENSION OF SPACE 85 in future use the words point. By the introduction of ideal elements we "have made each of our spaces a real analytic continuum. In all but the spherical case there is a one to one correspondence between points and sets of real homogeneous values not all zero. and the distinction between actual and ideal shall be the same as there given. Through each actual point will pass two lines An equation of the type to a given line. We shall later return to the meaning of such words as distance where ideal elements enter. In the hyperbolic case we may apply the same principles with slight modification. There will be a real point corresponding to each set of real homogeneous coordinates (x) for Which J2^2 + ^2 + ^ + ^2 < We 0> A set of real homogeneous values for (#). which latter may be called. If the inequality be not fulfilled. Actual and ideal elements stand on exactly the same footing with regard to purely No congruent transformation can descriptive properties. will give a plane. this bundle determines an ideal point having the coordinates (4 If x* k 2 x* + + ^ x? + = 0. can interchange actual and ideal elements. as defined so far. the assemblage of all ideal points whose coordinates fulfil the equation (and there can be no actual points which meet the requirement) shall be called an ideal plane. An ideal line shall be defined as in the euclidean case. parallel (ux) 0. will determine a bundle of lines. Two lines having an infinitely distant point in common shall be called parallel. in spherical space there is a one . a bundle with the same descriptive properties as shall therefore say that a bundle of concurrent lines. the ideal point shall be said to be infinitely distant. No congruent transformation. If the ideal point shall be said to be ultrainfinite. Let us take account of stock. the coefficients (in) being its coordinates. line. interchange actual and ideal elements. one through every actual point. in distinction. 7^ u 2 + u? + u^ + ?V > 0. actual.
86
THE GEOMETRIC AND ANALYTIC
OH.
to one correspondence of coordinate set and pair of equivalent Each of our spaces will fulfil the fundamental points. postulates of project! ve geometry, as we shall develop them in Chapter XVIII, or as they have already been developed Let us show hurriedly, how to find figures to elsewhere.* Four distinct correspond to imaginary coordinate values. points will determine six numbers called their cross ratios,
which have a geometrical significance quite apart from all concepts of distance or measurement. f An involution will
arise when the points of a line are paired in such a reciprocal manner that the cross ratios of any four are equal to the
corresponding cross ratios of their four mates.
no selfcorresponding points, the involution
elliptic.
If the
points
of a line
:
be
homogeneous coordinates A
involution
ju,
it
may
If there be said to be located by means of be shown that every
is
may
if
be expressed in the form
(A//
AM + B
=
+ XV) + Vw'=
0.
be the coordinates of two points, (?/) there will exist an involution on their line determined by the
In particular
equations
arid (2)
(a .)
X (j,) + M (*),
()'= M (y)A(*),
and by a proper choice of running coordinates any elliptic involution may be put into this form. Did we seek the coordinates of selfcorresponding points in this involution,
we should
get
set of
(x)
=
(y)i(z).
(y)
Conversely, every
homogeneous complex values
to a
definite
+ i(z)
way elliptic involution. The involution may be taken to represent the two sets of conjugate imaginary homogeneous values. may separate the conjugate values by the following device. It is not difficult to show that if a directed distance be determined by two points, it will have the same sense as the corresponding directed distance determined by their mates in an elliptic involution. To an elliptic involution may thus be assigned either one of two senses of description, and we shall define as an imaginary point an elliptic involution to which such a sense has been attached. Had we taken the other sense, we should have said that we had the conjugate imaginary
will lead us in this
We
* Cf. Momorie della Fieri, 'I principi della geometria di posizione.' R. Accademia della Scienze di Torino, vol. xlviii, 1899. f Cf. Pasch, loc. cit., p. 164, and Chapter XVIII of the present work. The idea of assigning to four collinear points a protectively invariant number originated with Von Staudt, Beitmge zur Geometric der Lage, Part 2,
1922, Erlangen, 185866.
vii
point.
EXTENSION OF SPACE
An
87
imaginary plane may similarly be defined as an involution among the planes of a pencil, with a particular sense of description; an imaginary line as the intersection of two imaginary planes. It may be shown that by introducing imaginary elements under geometrically
elliptic
these definitions we have a system of points, lines, and planes, obeying the same descriptive laws of combination as do the real points of lines and planes of protective geometry, or the assemblage of all real homogeneous coordinate sets, which do not vanish simultaneously.* Introducing these imaginary
expressions, and the corresponding complex values for their homogeneous coordinates, we extend our space to be a perfect
We must now see what extension must be given to the concept distance, in order to fit the extended space with which we are, henceforth, to deal. To begin with, we shall from this time forth identify the two concepts distance and measure of distance. In other words, as the concept distance comes into our work effectively only in terms of its measure, i. e. as a number, so we shall save circumlocution by replacing
the words measure of distance by distance throughout. The distance of two points is thus dependent upon the two points, and on the unit. In any particular investigation, however, we assume that the unit is well known from the start, and therefore give as the definition disregard its existence. of the distance of two points under the euclidean hypothesis
analytic continuum.
We
d
This
=
>
When it takes is, at worst, a two valued function. a real value, we give the positive root as the distance, when it is imaginary we may make any one of several simple conventions as to which root to take. If one or both of the points considered be ideal, the expression for distance becomes infinite, unless also the radical vanishes when no distance is
determined.
Under
these circumstances
we
shall leave the
concept of distance undefined, thus getting pairs of points disobeying Axiom II'. Notice also that whenever the radical vanishes for nonideal points we have points which are distinct, yet have a null distance, and when such points
are included, Axiom XIII may fail. shall in like manner identify the concepts angle and
We
* Cf.
Von Stfmdt, loc.
cit.
,
7,
and
und das Reclmen mit Wurfen,'
l Liiroth, Das Imaginare in der Geometrie Mathematische Annalen, vol. ix.
88
THE GEOMETRIC AND ANALYTIC
in terms of the unit
CH.
measure of angle
which gives to a right
angle the measure &
We may
cases.
proceed in a similar manner in the noneuclidean
If (x) and (y) be the coordinates of two points, shall define as their distance d, the solution of
we
/oi
COB ^
*
=
<W)
/7
v
/
//
\"
^
*
V(j'x)
V(yy)
This equation in d has, of course, an infinite number of Before taking up the question of which shall be called the distance of the two points, let us approach the matter in a different, and highly interesting fashion due to Cay ley.* This theory is of absolutely fundamental importance in all that follows. The assemblage of points whose coordinates satisfy the
solutions.
equation
/ \ (CC>^)
__
A vJ.
/Q\
V
/
shall be called the Absolute. This is a quadric surface, real in the hyperbolic case, surrounding, so to speak, the actual domain imaginary in the elliptic and spherical cases in the
; ;
the locus of points which coincide with their equivalents. Every congruent transformation is an orthogonal substitution, i.e. a linear transformation carrying the Absolute Let us, by definition, enlarge our congruent group into itself. so that every such transformation shall be called congruent; certainly it carries a point into a point, and leaves distances In the euclidean case we take as Absolute the unaltered. conic
lastnamed,
it is
a;
= 0,
x* + x* + x.*=Q,
(4)
and define as congruent transformations a certain sixparameter subgroup of the seven parameter collineation group which
We shall return to the study of the congruent group in the next chapter. Returning to the noneuclidean cases, let us take two points Pj, P 2 with coordinates (x) and (y), and let the line We connecting them meet the Absolute in two points Q 19 Q2
carries it into itself.
.
obtain the coordinates of these by putting \(x) + ^(y) into the equation of the Absolute. The ratio of the roots of this equation will give one of the two cross ratios formed by the pair of points and the pair Q^o', interchanging X 2
PP
l * Cayloy, A sixth memoir on Qualities/ Philosophical Transactions of the Royal Society of London, 1859.
vii
EXTENSION OF SPACE
we
*.
89
the roots
points
The value of such a
(xy)
get the other cross ratio of the two pairs of cross ratio will thus be
+
*~(xx)
(yy)
By
interchanging the signs of the radicals
cross ratio into its
changing
we change this reciprocal, and this amounts to interthe members of one of the two point pairs. Let us
2)<t
denote this expression by
eT

k
^.
</(xx) V(yy)
(5)
If we write the cross ratios of the pair of points 3 2 and the pair Q T Q^ as (PjP2 QiQ^> we may redefine our noneuclidean distance by the following theorem
>
:
P P
Theorem.
whose
line
If d be the distance of two points I\ and meets the Absolute in Q 1 and Q 2
,
P.,
The great beauty of this definition is that it brings into clear relief the connexion between distance and the congruent group, for the cross ratio in question is, of course, invariant under all linear transformation which carry the Absolute into itself, i.e. under ail congruent transformations. Let the
reader show that a corresponding protective definition may be given for an angle. Our distances, as so far defined, are infinitely multiple
valued functions. There is no great practical utility in rendering them single valued by definition. It is, however, perhaps worth while to carry it through in one case. If we have two real points of the actual domain, the expression (1\P*, QiQ 2 ) will have two values, real in the hyperbolic, pure imaginary in the elliptic and spherical case, and these two are reciprocals, so that the resulting expressions for d will differ only in sign, for each determination of the logarithm. We may therefore take the distance as positive.
* For the geometrical interpretation of a cross ratio when some of the elements are imaginary, see Vori Staudt, loc. cit., 28, and Luroth, loc. cit.
90
THE GEOMETRIC AND ANALYTIC
OH.
Did we seek, not for a distance, but a directed distance, then it would be necessary to distinguish once for all between Ql and Q% and in each particular case between the pair PX P2 and the pair P2 Pi ^ e directed distance will have a definite
,
5
value sometimes positive, sometimes negative. Let us specialize by confining ourselves to the hyperbolic We have defined the distance of two actual points. case. Still restricting ourselves to the real domain, suppose that we have an actual and an ultrainfinite point. Let us choose 1. such a unit of measure that A: 2 = Our cross ratio is here with an absolute value r let us say, so that the distance negative, expression takes the form ^[log?* (2m + !)TT ]. Let us choose
in particular
Next consider two ultrainfinite points. If the line connecting them meet the Absolute in real points, we shall have
If,
a real cross ratio as before, and hence a real positive distance. however, this real line meet the Absolute in conjugate imaginary points, the expression for the cross ratio becomes imaginary, and the simplest expression for their distance is pure imaginary. The absolute value of this expression will
run between
and
,
for the roots of
^logA
=X
differ
by
7i
i.
We
may, hence, represent
the Gauss plane
by points
all of these cross ratios in of the axis of pure imaginaries
between
and
9
If the line connecting two ultrainfinite points be tangent to the Absolute, the cross ratio is unity, and we may take the distance as zero. The distance from a point of the
Absolute to
a point not on its tangent will be infinite the distance to a point on the tangent is absolutely indeterminate, for the cross ratio is indeterminate. may, in fact, consider the cross ratios of three coincident points and a fourth, as the limiting case of any cross ratio which
;
We
we
please.
case, we are thus able to represent the distance of any two real points of hyperbolic space in the Gauss plane by a point on the positive half of the axis of reals, by a point of the segment of the origin
Leaving aside the indeterminate
and 
iy
or
by a point
of the horizontal halfline
i
oo
;
vii
EXTENSION OF SPACE
91
and as two points move continuously in the real domain of the hyperbolic plane, the points which represent their distance will move continuously on the lines described. Let us now take two points of the hyperbolic plane, real or imaginary. We see that the roots of ^ log A = X differ by multiples of iri, so that we may assign to (/ an imaginary
part whose Absolute value

Moreover, by choosing
properly between the two reciprocal values of the cross ratio, we may ensure that the real part of d shall not be negative. If two points be conjugate imaginaries, while their line cuts the Absolute in real points, the cross ratio is imaginary, and the expression for distance is pure imaginary, which we may represent by a point of the segment of the origin and
i.
2>
If
both pairs of points be conjugate imaginaries, the ^
is
cross ratio
real
and negative, so that the distance may
be represented in the form
distance of
ratio
X
i.
We
shall define as the
two points that value of the logarithm of a cross which they form with the intersection of their line and the Absolute, which in the Gauss plane is represented by
a point of the infinite triangle whose vertices are cc + ~ I>  i. The possible ambiguities for points on the sides of
,
have already been removed by definition. have already seen that when euclidean space has been enlarged to be a perfect analytic continuum, imaginary points and distances come in which do not obey all of our axioms. In the hyperbolic case we shall find real, though ultrainfinite, points which do not at all obey the principles laid down Let us take three points of the for a consistent region.*
this triangle
We
ultrainfinite region of the actual hyperbolic
plane x%
=
0,
As these points are supposed to be real we may assume that x l9 x% are real, while x is a pure imaginary, and that a like state of affairs exists for (y) and (z). We shall further assume that the lines connecting them shall
say
(x), (2;), (%)
intersect the Absolute in real, distinct points.
(y*Y dm) (**) >  (zz) (xx) > (zxf
>
0,
(yy)
(zz)
M>
>
We
have then
>
>
0,
(7)
(xy)*(sM)(yy)>0,
0.
'
* The developments which follow are taken from Study, Beitr&ge zur nichteuklidischen Qeometrie/ American Journal of Mathematics, vol. xxix, 1907.
)(^)>().triangles whose vertices are (y) and (z) and within the verticals of these see (9). that (yz) (zx) (xy) we are in the qu asitriangles < 0.2x + xy. Under what circumstances cosh shall we have ? yz^>.nr) (M) (zz) +2 \{yz) (. (10) .. that under these circumstances. zx) ^ cosh j.//. those on the right are positive. The conic and tangents determine four quasi. (2/5 ~~ V (20K^) V ^ " / (yrf I V RT^T) V farftyy) llyW^iyWs) ''(yy)(n)~ J^f I T^) 2 V/MM ^^) ~(sz)(ra) The terms on the left are essentially positive as they represent hyperbolic cosines.r) (xy) \ ~^x) (yz)* (8) (9) We we (yy)(~xY~(zz)(xy)*<(). on the contrary. lateral of these tangents or the vertical angle at (y) or (z). are at liberty to drop the absolute value signs in the second term. let To by what region X two < angles. see that if (? p)(::. COS j A* = u COsh / O/.  \ \yz\ ^ \zx\ + \j'y\. indicate the distance from (x) to (y) 7p ^ zx ^ xy. and draw tangents must lie within the quadrithereunto from (y) and (z). Our original inequality (8) will *  still hold if ocyz 4 (yz) (zx) (xy) < 0.M by ~xy. Let us )ther now assume. beinohyperbolic sines we may therefore square the inequality . therefore. Since (yy) > our inequality (9) will hold within the quasi. of the ultrainfinite domain is determined us sketch the Absolute as a conic.triangles with two rectilinear and one curvilinear side each. Let us. for the moment. and the whole expression is the square of the determinant xyz which is zero or negative. THE GEOMETRIC AND ANALYTIC and assume shall also take <f / // CH. (. We rr: > . We see.
If we look on (y) and (z) as fixed. . 1038. loc. a rectilinear path is the longest from * (y) to (z). this inequality certainly holds if (7) does. ^. pp. Fig. will play In a region where (8) holds. the curve  xyz 2_ 4 ( yz ) ( zx ) ( xy ) the _ 0) we are in so far as it lies in two quasitriangles now FIG. 3 is taken direct. cit. and (x) as variable. see Study. the part of the segment of (y) and (z)* considering.VII EXTENSION OF SPACE 93 and.. conversely. For a complete discussion.
and say that every analytic transformation which carries the Absolute into itself alone is a congruent . i '= coshd + #! sin d. pointing out besides certain interesting sub groups. for were it a function of (x) the Absolute would be carried into itself.) 0. as they are merely the quaternary orthogonal group. and our present task shall be the actual formation of those groups. V= P Suppose that we have x i = fi (Vi^A)> /o ( X o x i x 2) <. ( x{ x^ sinh d+x^ cosh d. In the real domain. transformation. which (y) is contrary to hypothesis. = P (xx). (xfx') <= /a (^1^3). of course. the structure of the two is quite different. however. the congruent groups of elliptic and hyperbolic space are identical. evident.CHAPTER VIII THE GROUPS OF CONGRUENT TRANSFORMATIONS THE most significant idea introduced in the last chapter was that of the Absolute. that in the complex domain. $ #3 r= JT3 (Xg ij X% 3J.?. easily show that the transformation a It is. and into some other surface (x) P= (. and may be written. ^ ' We get a reflection by reversing the signs in the second . and its connexion with the concept of distance. Replacing must be a constant. We shall incidentally treat the euclidean group as a limiting case where j^ A/ = 0. The group of translations of the hyperbolic line will depend on one parameter. is whence we may collineation. if A2 = 1. by A 4 & we see that (x'y') we shall also have = P(xy). Every collineation of noneuclidean space which the Absolute in place was defined as a congruent keeps transformation we had already seen in Chapter V that every congruent transformation was such a collineation. We may go one step further.
the transformation. too. itself. 2 O/ f 2 <*"i/^c/ As the Absolute must be projectively transformed into itself. we see that there are three distinct possibilities. The real line connecting them is ultrainfinite.4A = ( a il. # sin c + # x cos c?. (3) The ternary domain is more interesting. VIIT CONGRUENT TRANSFORMATIONS In the elliptic 95 shall equation. similarly or spherical case we have .2 p . This will give a rotation about this point. . In the third case the two fixed points of the Absolute conic fall together. . 2 a 22 * + 2 ) (a u a 21 ~a ]2 a 22 ) + 2(a u a 22 + a 21 a 12 )i 2 ^ .. * ' To and cZ pass to the euclideaii case.a 22) 2 + 4c( l2 a 21 < 0* If the fixed points of the Absolute conic be real.. if A > 0. and we shall have ( a il + a 22)2 . 2 )a. (4) /= 2 (a u a 21 + a. ~~"~ / 2 / 2 1 "^ 2 ' _ / 2 1 1 / 2 2 ' A. If we view the matter geometrically. or a sliding combined with a reflection a perpendicular plane through this line if A < 0. Let us express the Absolute in the hyperbolic plane in the following parametric form j. replace # # ' by kxQ kx^ . and the third fixed point of the plane falls there The transformation carries a pencil of parallel lines into . in the actual domain. we may put and this will lead to the general ternary transformation + 2(a n a 12 a 2l a. x Xi ' = = x cos d + ajj sin d.OH. will appear as a sliding along a real in line. ^0 ^ = ^=xd. by r Ic > divide out fc. First the two fixed points of the Absolute conic are conjugate imaginaries. and has an actual pole with regard to the Absolute. and then put 7^ l\' = 0.
2 2 f + B* = ^ 2 + 52 = 1. in fact. that we have the ^ y 2 + Z'2 1 The geometry thereof will be exactly our spherical geometry.96 THE GROUPS OF The elliptic case is treated similarly. considering this as the Gauss plane. take the linear transformation this ' ~ (y. three dimensions are of the same form as those in two. and. euclidean sphere jf2 These forms remind us at once of like forms occurring in Suppose. .dij~3 + (a~j3i)' " ~ (y ' + bi)&~+ (a + pi) These equations are seen at once to be transformable into the others by a simple change of variables. and we wish for the group of congruent transformations of sphere into itself. To pass over to the euclidean case. Let us project the sphere stereographically from the north pole upon the equatorial plane. put l . duction of imaginaries. en./'2 l j . . Notice that here the group y= is an invariant subgroup. the theory of functions. We may by a judicious introwrite the Absolute Let us now take the binary substitution We come thus to the general group of congruent trans formations + ^1 + (a xl + 2 yaXv + ^~y 2 6^). albeit the structure is a trifle general The congruent groups in .
p&/ = (aaf f$p (7) p^/ = (ay + ay + ^36 + pb) x + (ay + ay . long more complicated. The sixparameter group is made up of these and the assemblage of transformations of negative total * The literature of this subject is large.d6 . xxxii. #a x.)Sy) i dy 3 . an leaving imaginary quadric of real equation./^S .* The Absolute of hyperbolic space may be interpreted as a euclidean sphere of radius one. Tho treatment here given follows broadly Chapters VI and vol. ' 1893. VII of Klein's Niehteuklidischo Geometric'. i (z 0).^6) ^ + (aS f ab + /3y + /3y) i2 + i(ag . lithographed notes. 1846./3y 4. 'Sur quolques propriety's des determinants gauchcs. is the same as that of finding all collineations carrying such a sphere into itself. and an imaginary conic with two real equations.68)^ + (a^8 + a/3 + y f 78) x2 + i(ap a/3 + y6y6) + yy 6g)i' + (au^ yy f 82) ^ + (a^ + a/J~ySy8)u.2 hi(a)3a/3yS + y8)^ 3 . The solution has of course. The first writer to express tho general orthogonal substitution in terms of independent parameters was Cayley. and the problem of finding all congruent transformations of hyperbolic space.\ 1 i . COOUDQK 0 . Let us represent this sphere parametrically in terms of its rectilinear generators '/ *t ~>^y ** t . } 5 + 2. 3 . nonruled quadric. been known.vin CONGRUENT TRANSFORMATIONS 97 We wish for the sixparameter groups invariant respectively a real. This subgroup might properly be called the group of motions. = Let us now take the linear transformation ys + 5 yz + positive The sixparameter group of congruent transformations of modulus will be pxt = (au + )3jS 4.yy + 85) + (ad /3/3 + yy .' Crelles Journal.
k. An extremely elegant way of expressing these is offered by the calculus of quaternions. under the most general motion. the generators On the conof each set are permuted among one another. these latter discriminant called symmetry transformations._ + p' /5 + jf" a'z _. The subgroup of motions includes the identical transformation. so. the most general symmetry transformation will arise from the combination of the most general motion with a reflection. and it is easy to see that a reflection will inter change the two sets of generators. and we shall have . under causing the modulus to vanish. j. each generator of the Absolute stays in place. We assume that they obey the associative and commutative laws of addition. following the Hamiltonian notation. the product of two quaternions is the product of their tensors. Let us next write is where P and Q are quaternions.98 THE GROUPS OF OH. assume three new symbols i. the identical transformation. and identifying the real parts and the coefficients of i. is called a quaternion. by writing ^ We reach . * _ ~ u' +_*' y'cfg" The distinction between motions and symmetry transformations stands out in clear relief when we consider the effect upon the Absolute.j. without ever This shows that as. In the elliptic case we shall have the group of all real quaternary orthogonal substitutions. the associative and distributive laws of An expression of the type multiplication. whereof It is easy to show that the tensor of called the Tensor. Multiplying out the righthand side. k i : 2 =/ = /*= ijlc= 1. Let us. we have x^x^x^x^ expressed as linear homogeneous functions of x^x^x^ The modulus of the transformation will be different from zero. and any motion may be reached by a continuous change in the six essential parameters from the values which give the identical transformation. trary.
sots of generators of the Absolute by calling the one left. vol.Yin CONGRUENT TRANSFORMATIONS 99 These equations will give the sixparameter group of motions. have already donned parallels as lines intersecting on the Absolute. Clifford's discussion is in his * Preliminary Sketch of Biquaternions '. Two conjugate imaginary left generators will also be invariant. 1902. Now were it possible to divide 6r into invariant threeparameter subgroups in two different ways. yet it is better to be consistent in our terminology. Proceedings of the London Mathematical Society. <r/ ?t (f/ .j + X3 'l: = P' (. Hence the groups // 3 <// are identical with G G. and although in the present case such lines cannot both be real.^ by resolving it into the two groups which keep invariant all generators of the one or the other set on the Absolute. 1873.X.<V . Of course there are possible complications in the imaginary domain. * The more common name for such lines is 'Clifford parallels'.f . + X. We obtain their geometrical significance as follows The group of motions G G can be divided into two invariant threeparameter subgroups r/ 3 <j. the group of symmetry transformations will arise fr m X '+ < . iv.J. This may be done analytically by adjoining a number I to our domain of rationals.. It is well worth our while to look more deeply into the Let us distinguish the two properties of these subgroups. We . especially since we shall find in Chapter XVI a transformation carrying parallelism into parataxy. Let us now look at a real congruent transformation which keeps all right generators invariant. being made up of two ' invariant subgroups G3 G3 obtained severally by assuming that Q or reduces to a real number. Our group of motions is halfsimple. and the other rlyld.k) Q'. xi. Zur Nichteuklidischen und Linien* geometrie.xj . Two lines which cut the same left (right) generators of the Absolute As the conjugate shall be called Left (right) paratactic. the distinction between motions and symmetry transformations being as in the hyperbolic case. :] . The word paratactic is taken from Study. the highest common factor of 3 or #. we see that through each real point will pass a real left and real right paratactic to each real line and the same will hold for each real plane. arid the latter keep the one or the other set of generators all in place. for f/ a is nothing but the binary projective group which has no invariant subgroups. not only of G 6 but of 3 This may not be.' Ja/iresbericht der deutschen Mathematikcrvereinigung. and every line meeting these P : (. but these need not concern us here./ with G would be an invariant subgroup.* to each generator of the Absolute belongs to the imaginary same set as itself.
shall be called a left translation.A. is \\ the matrix of a ternary orthogonal sub. any two points will be transported through congruent distances. In three dimensions. we may split up the transformation into a reflection in a plane. (9) where \\A 1 B^G. since the path curves of all In fact points will be a congruence of left paratactic lines. and a rotation through an angle TT about a line perpendicular to both planes. CH. or a rotation through an angle ir. of all translations There will be a threeparameter invariant subgroup x'~ A + x that ?/=/* + y. let us notice that in the plane a reflection in a line is identical with a reflection in a point. there is never any identity between a rotation and a reflection. this congruence will give the path curves for a whole oneparameter family of left translations. in a spherical plane they are different. In like manner we of may find the sixparameter assemblage symmetry transformations. for as each plane is hereby transformed into self.100 THE GROUPS OF CONGRUENT TRANS.z. Before leaving the elliptic case. on the other hand nothing new is brought in by interchanging each point with its equivalent. vnr two will be carried into itself.3 stitution. . ' / (/ _i~ w^ p '. Let the reader show that under a translation. a reflection in a second plane perpendicular to the first. and a reflection in a line is the same as a rotation through an angle TT coupled with an interchange of each point with its equivalent. To pass to the limiting euclidean case elliptic ^x 4 A 2 y 4. . every other line will be carried Such a transformation into a line right paratactic to itself.
for the most part. if one of two such points be actual. as promised in Chapter VI. Two points which are conjugate with regard to the Absolute shall be said to be mutually orthogonal. Two intersecting lines or planes which are conjugate with regard to the Absolute are mutually perpendicular. Let us notice. x logarithm of the cross ratio that they form with the points where their line meets the Absolute. shall leave aside the euclidean case. the other must be ideal the converse is not necessarily true. The distance of two points is We We . . us take two points A. and. from the higher point of view gained by extending shall find in space to be a perfect analytic continuum. Otherwise stated. to the problems of elementary noneuclidean geometry. LINE. B with coordinates (x) and (y) respectively. say x3 0. at the outset. Let Let us begin in the noneuclidean plane. the distance from a point to a plane is minus its distance to the pole of that plane with regard to the Absolute. that the principle of duality plays a fundamental role. and plane. tangent to tho Absolute . our present task is to express the fundamental metrical theorems of point. the Absolute a Dens ex Mdchina to relieve us from many an embarrassment. the angle of two planes is  . leaving to the reader the simple task of making the distinction between the elliptic and the spherical cases.CHAPTER IX POINT. & x logarithm of the cross ratio which they form 'i/ with two pianos through their intersection. In the real domain of hyperbolic space. and find the two points of their line which are at . AND PLANE TREATED ANALYTICALLY THE object of the present chapter is to return. in terms of the invariants of the congruent group. line. handle all of our noneuclidean cases together.
in fact. The bisectors of angles formed by three coplanar but not concurrent the lines are concurrent by threes in four points. BO we see that the coordinates of its . (z).r) + n (y) (?/) that w/ The coordinates of the centres of gravity ( v ' congruent distances from them. LINE. Theorem 1 . (z) are easily (*. The centres of gravity of the pairs formed from three given points are lines. a polar reciprocation in the Absolute: . Lastly. may arise in the Let us next take three noncollinear points A. Theorem Y. these The centres of gravity of the points seen to be ~ (x).102 POINT. where the centres of gravity will be equivalent points. point and line. B. The necessary and sufficient condition that the point A. may be called the centres of gravity of the three On the other hand the centres of gravity of points. notice that a dual theorem might be reached by interchanging the objects. A line connecting (x) with a centre of gravity of (y) and (z) will be v/^7) It is clear that \ Xxz + V^T) XMJ i 0. and the will thus be Let the reader discover what complications ideal domain. These shall be called the centres of gravity of the two points. points bisectors meet the given lines are collinear by threes on four lines. G with the coordinates (a?). The collinear by threes on four The.+ ?'  M^)Returning to the line + V s$j). and are. We purposely intersections of their line with the Absolute. should be at congruent distances from (x) and (. distance and angle by taking. in fact. lines from the given points to the centres where of gravity of their pairs are concurrent by threes in four points. (?/).v&)> OZ) v. (//). \  such lines are concurrent by threes. exclude the spherical case. the two points which divide harmonically the given points. AND PLANE CH. in four points which given our pairs of points are collinear in threes.
Let us the given lines. Returning to a centre of gravity of the two points see that a Jine through it perpendicular to the line BO BU. The lines through each of three given noncollinear points. now suppose that besides our three original points. sec immediately from the form of this equation. 33. are concurrent. .e. and the sum of the three is identically zero. The equation line through A of the line connecting this point with A. II. If three co collinear points be given.<*) ! dp] z/) ()  o. we get two other equations of the same type. the perpendiculars to the lines of their pairs at the centres of gravity of these pairs are concurrent by threes in four points. we will have the equation (xy) (.ix TREATED ANALYTICALLY 103 (s). The points on each of three coplanar but not concurrent lines. perpendicular to J5C. each at congruent dis but not concurrent be given. Theorem !2'. The&rem If three non Theorem planar lines 3'. the points orthogonal to their interseetions on the bisectors of the corresponding angles are collinear by threes on four lines. will be the we permute the letters #. are collinear. . so that If Theoreiib '2. y. pole with regard to the Absolute will have the coordinates where for every value of (r) (rs) = \ryz\. r L A) Jij ^^ ii f^L ** J L j) 'j~3 J H 1=0 when The (y) first factor (:) and will vanish (in the real domain) only are identical. orthogonal to the intersection of the other two.cyclically twice. the equation will then be We all (z). that points of this line are at congruent distances from (?/) and thus confirming 3. perpendicular to the line of the other two. i. tances from all three of the making congruent angles with all three of given points.
suppose that we are restricted to a consistent region of the plane. we easily see that if A'. ~~ . we see that BA' __L whence rv . sin AC' >. we have first set three others lying one on each of the lines of the as follows G' = (rx f $y). AND PLANE CH. Then we shall easily see f from Axiom XVI that if A A'. AB'  . Ipr mqs = 0.??i Xzx =  0. BC' A* sin if A'. specifically. CB' y sin sin ism On the other hand. G' be collinear. mqs _ R. CA' A/ .2/0 1 = 0. more 2 GA' . If A\ B\ G f be three points lying respectively .  And and the condition for concurrence for the three lines (Ipr this will give f mqs) 2 a. LINE. C' be collinear On the other hand. EA' sin sin Off . Theorem 4. B'. CC be concurrent Let us. BB'.sm AB' ~ /' sin z^ BC' 7 > A.* . BA' . moment. BC' to Q\Y\ QITI si ri A* The equation of the line I \ A A' will be Xxy 4. AC' sin sm Now. CA' A' sm .104 POINT. AB' A. for the .
. k Theorem 5. then the expression sm . is equal to the corresponding ratio for these two with the third point. meets BC. Let us next suppose that A' is a point where a bisector of an angle formed by the lines BA. thus getting m We A AB (y GA' sin : sin sin BA : sm CA . A'. k. BB'. These are. BA' . CA. If three cobut nonconcurrent planar lines be given. . each centre of gravity of a pair of points where two of the lines meet a third determines with the intersection of this pair of lines such a line. B '. is Theorem 6. second order. all six points being in a consistent region. is an envelope of the second class. AC* k 7 sin 7. of course. equal to the corresponding ratio for the two lines with the third. AB' . and only G' are collinear. in such a way that the ratio of the sines of the &th parts of its distances from two points is constant. when. and only when. when. CA. that the ratio of the sines of the angles which it makes with these two lines. Theorem 5'. r find I and must be easily in this case. AB. GC' are concurrent. If three noncollinear points be given. merely the analoga of the theorems of Menelaus and Ceva. The envelope of a line which moves in such a way in a plane.IX TREATED ANALYTICALLY 105 on the lines BC. Cl r r~ sin sm sm . CA' . that they will afford a sufficient ground for a metrical theory of cross ratios. It is worth noticing also. each bisector of an angle formed by the lines connecting two of the points with the third will meet the line of the two points in such a point that the ratio of the sines of the kth parts of its distances from the two points. while it will be equal to 1. is a curve of the Theorem 6'. AA'. that the ratio of the sines of its angles with two fixed lines is constant. The locus of a point which moves in a plane. by noticing that at congruent distances from and AC.sin k k will be equal to 1 when.
LINE. . Our next investigation shall be connected with parallel lines. Let D be the point where the perpendicular to (u) through (y) meets (n). Let the reader show that if an angle inscribed in a semicircle be a right angle. This shall be called the parallel angle of the distance from the point to the line. V(vv) V(uu) (yy) squaring. and remembering that (vy) 2  0. (4) cos * Ann 4ABC =  v _i^l y ADn srn2LABC= . the euclidean hypothesis holds. Let (v) be the coordinates of a parallel to (u) through (y). (3) From these we if easily see sin n(rf) Furthermore. (lf!f)(uy) cos n (d) (ny) (uu) (uv) (ur) (vv) (vv) [(uu) (yy) . . and the notation n einU(AB) (d) are due to Lobatchowsky.__. We seek Since (u) and (v) intersect on the Absolute 2 (uu) (vv) The equation of the The line PD  (uv) will be = 0. while the given line has the coordinates (u). cosine of the angle formed by v and PD ( will be .ACB be a right angle . sech (d) tan U(d) ^. a circle.(uy) 2 ] ' . We shall hunt for the expression for the angle which a parallel to a given line / makes with the perpendicular passing through a point to I through P. AND PLANE OH. and that k = i. that we are in the hyperbolic plane. /K (5) . xyu =  0. in general. = = csch (d). Let us give to the point P the coordinates (.106 POINT. and if the latter be d the parallel angle shall be written * P n (d). for the moment. We suppose. It would be quite erroneous to suppose that either of these curves would be. cosll(47y) The concept parallel angle.//).
and this will pass through / : (uy) (vy). limit cos FT (d) d Let us line ('&) ' d = now find the equations of the two parallels to the which pass through the point (y). (7) Let the reader prove the correctness of the following construction for the parallels to Drop a perpendicular from P through P on meeting I : it in Q.Btan^L ABC. cos 4I nn A BG = . naturally. sin   (6) v ' = sinn(J?6>inn(6^) = tan^G'A.2 (uy) (uy) (. a pair of lines meeting in (y) the polar of (//) with regard to it will be illusory.* ff$ in T. w will pass through the intersection of (u) with the polar of y with regard to the Absolute. is structure. . Let the coordinates of the line which connects the other intersections of the parallels and the Absolute be (w). sn ^ T". 107 _. Noneuclidean Geometry. by the harmonic theory of a quadrangle inscribed in a curve of the second order. The general form for an equation of a curve of the second order through the intersections of (u) and (w) with the Absolute will be / we (ux) (ivx) (y) if m (xx) : 0. These two cannot. and a radius equal to (QR). and let the perpendicular to PS at S meet I at R. .*%) = may / ( o. be rationally separated one from the other. Then with P as a centre. Be it noticed that. Take S a convenient point on the perpendicular to PQ at P. u: Substituting b rrt ^ [2A(uy) * + (ux)(uy)\ The formulae given may be used Of. This last equation \U'f/) I be written ' ' I I V 1 '') ( W/ ~f* (''''/) ' ' ~ ^ f ('/ i ^') iW) Now. construct an arc meeting PT will be the parallel required. so that wo may write . as we should expect. the coefficients of (x) will vanish in m = (yy) Since this curve is (yy) (uy) ( w) + (yy) (wy) (^) . reasoning as the basis for the whole trigonometric Manning. 1901.ix TREATED ANALYTICALLY . Manning'* open to very grave question on the score of rigour.e. so that shall find the equations of both at once. i. Boston.
replace # by /kc # and divide by k. (2). LINE. in fact. vol. be called the centres of gravity of the four points. will pass through a point of the third. The centres of gravity will form with the given The meaning of this phrase points a desmic configuration* is as follows. homogeneous point Let us have four points.' Bulletin den Sciences mathematiques. (10) will be points of concurrence. may then divide our twelve points into three lots as follows: We H We (y) (z) (0 (+ + + +)(+ + __)(+_ + _)(+__+) (ii) see that a line connecting a point of one lot. 1878. Let us indicate the centres of gravity by the signs prefixed to their radicals. in. not in one plane. s6rie 2. of lines from each of the given points to the centres of gravity of the other three. giving always to the first radical a positive sign. These eight may. four * figuration The desmic configuration was first studied by Stephanos. point twelve points will thus lie by threes on sixteen lines. with the coorWe easily see that the dinates (x). We eight points 4~ = ~ +  ). 'Sur la condosmique de trois tetraodres. We get the square of the usual expression [(uj/KM^J^O. four by four. .(2/y) (UB) f 2 Which leads to the required equation (uyY (M) + M 3 (?/?/)  2M (wj) (xy)  0. (?/). The x^\x^ will vanish Under these circumstances (wx) = . (8) To get the euclidean formula. The (9) principles which we have followed in studying the metrical invariants of the plane may be extended with ease have merely to adjoin the fourth to three dimensions. with any The of a second.108 POINT. or line coordinate. (/) respectively. coefficients of AND PLANE if CH.
the eight points which are severcongruent distances form. Let the reader who is unfamiliar with the desmic configuration. If four non concurrent planes be given. (y). Let us seek for a point which is at congruent distances from our four given points. f the planes which severally are coaxal with each of the three through eight points which form. form a desmic configuration. with the original ones. and dually for the planes bisecting the dihedral angles. As there are eight points at congruent distances from the four given points. Let the reader show that the centres of gravity of the six pairs formed from the given points will determine a second desmic configuration. Their coordinates are found to be (. (t) but their four . If four non coplanar points be given. we take the twelve planes obtained by omitting in turn one point of each lot. It is easy to see that there cannot be more than eight such points. meet them in congruent dihedral angles. eight points with regard to the Absolute. its centre.s) where. with the original four. and the ideal points of concurrence of its parallel edges. the eight planes which sever at from them figuration. the lines from each to the four centres of gravity of the other three will pass by fours Theorem 7 If four nonconcurrent planes be given. if we consider not the points (#). (Vs) ~ i rztx i + (12) Theorem 8. (z). Theorem 7 . the lines where each meets i.IX TREATED ANALYTICALLY 109 In like manner we shall find that if passing through each. lie by fours in eight planes which. form. with the four. a desmic conoriginal ally Theorem 8'. If four noncoplanar points be given. study the particular case (in euclidean space) of the vertices of a cube. we have but to take the polars of the In like manner. a desmic conally figuration. so there will be eight planes at congruent distances from them. a desmie configuration. with the original ones. for all values of r. remaining planes and a plane bisecting a dihedral angle of the two still left. . two planes of different lots are always coaxal with one of the third.
If four noncoplanar points be given.*) = k tan ' (13) line shall be said to be parallel to a plane. A if (wy)* (xj:) + (uxf (yy) . form. The coordinates of these latter eight will be Theorem 9. we shall have for the parallel cos FI (. If the distance from the point to the plane be x. arc at congruent distances from the planes of the first four. AND PLANE distances from planes. severally. will be <lij = 1>W two The condition for the intersection of (jt> . a desmic con figuration. LINE. the eight planes which. severally. a desmic configuration. and (_//) = (15) . with the first four planes.110 POINT. naturally /} = 2 ^ lines (p) . for the line joining (x) and (y) the usual Plueckerian form We now We The coordinates of the polar of this line with regard to the Absolute. are at congruent distances from the points of concurrence of the first four.2(ux) (wj) (xy) = 0. with the first four points. there will be eight points at congruent them. Theorem 9'. cone of parallels to a plane (u) through a point (y) will have the equation Definition. the Absolute polar let us say. form. The the point common to the two be on the Absolute. If four nonconcurrent planes be given. (14) pass to certain metrical invariants of noneuclidean take as coordinates space expressed in line coordinates. the eight points which. The parallel angle of a point with regard to a plane can be defined as its parallel angle with regard to any line of the plane through the foot of the perpendicular. will be.
(. there will be two lines meeting both at right angles. (18) sin  . while 2 p^ p'}./) (y'y') 2 p'J. which we now explicitly exclude both for (p] (yy) (aw:) = 2 Pif.IX TREATED ANALYTICALLY will 111 if Each meet the Absolute of polar of the other 2p<jl>'ij = 0 (16) Notice that (p \p') is an invariant under the general group of collineations. that is. both. sin 2 _ will be a root of IV an irreducible quadratic whose coefficients are rational invariants under the congruent group. while For the sake of the other is (//) given by (j/) and (?/). (p f (xx) (xx (yy) ) (yy (x'x') ) (xx') (2/2/0 (s/Y) 8 V)(a^) ]. thus.is invariant under the congruent group only. d be taken to indicate the distance of two lines.( W )(^(j^)l. in the general case. If. Let us seek for this equation. (y).'. legitimate assumptions The distances which we wish to find are We have and a possibility (xx) _ (ajy)2 = (yy} v ^ = f (yy) this will vanish only when (p) is tangent to the Absolute. simplifying our calculations we shall make the obviously equation. and these are indistinguishable in the rational domain. We shall mean by the distance of two lines the distance of their intersections with a third line perpendicular to them It is easy to see that if two lines be not paratactic. Let one of our lines be p given by the points (a?).
(16). ^ With regard to the signs of the roots in (19) we see that in the hyperbolic case.112 POINT. = *> Then. . * See D'Ovidio. thus bringing us back to equations (15). one of Studio suIla geometria proiettiva. since lira d T A k } k. and if each intersect the absolute polar of the other. where the two lines are actual. AND PLANE ' . and have been studied under the names of moment and commoment of the two lines. 1873. divide out and put ^ . i._! fc The square roots of the products of the roots of these two equations are wellknown metrical invariants. If two lines intersect the moment must be zero. CH. the commornent must vanish. and ( Le funzioni metriche fondamentali . (FO ^ .* We shall return to the moment presently. To reach the x by kx .. LINE. oo We have * the usual formula. ( vi. Memorie <ki Lincei.. as usual. attaching a particular value to the signs of the radicals in the denominator. T ' = 1 k sm.' Annah di Matematica.  (2/2/0 k 2 + cos. limiting euclidean case k*. we replace. +8m "I . negli spazii di quantesivogliono dimensioni '. 1877.coa . ' sin = d.
so that We shall mean by the angle of two nonintersecting lines the angles of the plane. our equations (23) become (p\p ) = H sin = 0. In To tions get the euclidean formula we make the usual substitu and divisions. (24) Let us now look at paratactic lines. and that there should be but a tangent to the Absolute. which contain the same common perpendicular.e. by the equations (P P') I single line of their pencil These conditions will be expressed = [2ft/ ZpfppijPi/r] = when we f 0. (23) Let the reader notice that usual way pass to the limit in the for the euclidean case. thus getting the well known formula ' (X* + ffj + X*) (X* + X* + X. i. and put jlc = 0. The necessary and sufficient condition that two lines should bo is . one through each.^) line q cutting ~ The coordinates of the will bo given by (/>!?) p and PtfVij jf at right angles = (/ 1 ?) = spy iij = 2 = (s I ?) = o have defined as a parallel. This will be k times the corresponding distance of the absolute polars of the lines. the elliptic case the two distances will be real. k k The square of the moment of the two lines is negative. two lines whose intersection on the Absolute let us now give the name pseudoparalld to two coplanar lines whose plane touches the Absolute. Of course COOLIDOE . We either parallel or pseudoparallel is that they should intersect. We thus get for the angles of the two lines (p). so that . one distance will be real and the other pure imaginary.ix TREATED ANALYTICALLY 113 the points chosen to determine each line will be actual and the other ideal. lines which meet the same two generators of one set of the Absolute.
This puts in evidence that intersecting lines cannot be paratactic unless they be parallel. now proceed to replace our concept of a line by the sharper concept of a ray as follows. let us return for an instant to the moment of two real lines.. suppose that the distances of two lines be congruent. \ We We . '* Ssp. sin rf. ( the lines are paratactic. 2 ~ sin ~ .114 it is POINT. LINE. that every point of the line (x V(yy) + iyV(xx)} r (x V(y'y ) +i and of the line (xV~(yy).iyV(xx)) . AND PLANE OH. The necessary and sufficient condition that two lines should be paratactic is that their distances or angles should be congruent. Besides our previous equations connecting (x) (y) (x') (y'). paratactic. however. (xx)(tfaf)~(yy)(y'y'y p' meet the Absolute respectively in the points (x </(yy) It is iy V(xx)) (x' V(y'y') iy'V(xx')). This condition may be expressed analytically by equating 2 to zero the discriminant of either of our equations (19). the absolute polars of paratactic lines are. _ / /vo _jfW)L_. d k 7 . we have. merely to look at the oneparameter group of translations of space which carry these two lines into themselves. so that the sign of the moment is that of (p //). In conclusion. belongs to the Absolute Lastly. Let us. It is in the elliptic case only that two such lines can be real.* </2 PiJ (P\P) ~ ==~=== shall assume that the radicals in the denominator are taken positively. we have / MO __^]!__ The lines p. immediately evident that two paratactic lines have an number of common perpendiculars whereon they always determine congruent distances. . clear. Hence Theorem 10."}=0. Coninfinite versely. in fact. (20). . or pseudoparallel. themselves. i/spij' } {[(P M(25) S^/2^.
Intergive a second ray. We shall later see various applications of this concept. and this shall be considered equivalent to any other ray whose coordinates changing to this.IX TREATED ANALYTICALLY > V 0. both in therefrom by a positive factor. said to be opposite The relative moment of two rays is thus determined. Vi Pij *j and in the = Vi shall be called the coordinates of the ray from (y) to (2). 115 in the hyperbolic case assume always The coordinates elliptic case X Q > 0. H2 . differ (y) and (z) will magnitude and sign.
see the dissertation of Beck. The polar of the given complex will have the coordinates ajk = . For the hyperbolic case. continue the subject in the special direction where the fundamental element is not. ' . which are the directrices We get of the congruence. by 'Zur nichteuklidichen etc. : m = pXi> = pX 2J * (1) this chapter is sketched. d oi = rl> jk and the congruence. shall be called axes of the pencil. k = 1. 1905. and the two mutually absolute polar lines. 3. (Zdjja f m& 01 ). The elliptic case Tho dual projective is developed at length in the author's dissertation. cit. . . geometry of elliptic and spherical space. but a pair of lines invariantly connected.rfitf. dots indicate . Practically the whole of Study in his article.2 djk j>Jk = 0. j. *Die Strahlenkotten im hyperbolischen Raume.* Let us start in the real domain of hyperbolic space and consider a linear complex whose equation is <&!$ = The . that the coordinates of a point aro k* = xlt x2 1 . <>. (a p) will be composed of all lines of our complex and its absolute . i. 2. whose equations are = 2 atfprt .CHAPTER X THE HIGHER LINEGEOMETRY IN Chapter IX we took some first steps in noneuclidean The object of the present chapter is to linegeometry. 1904.  polar. and choosing such a unit of measure that we have for the Absolute #a . a line. in general.' Greifswald.' Hannover. without proofs. such values their pllickerian coordinates by giving to I Let us now write that the complex shall be special. or common to all complexes of the pencil (ld ol w 23 ) ./ loc. These complexes shall be said to form a coaxal pencil.
and only one namely where case. i. These coordinates (X) are homogeneous in the complex (i. i.e. All complexes of the pencil will here be special. the directrices of the congruence are tangent to the Absolute. A pair of real lines which are mutually absolute polar. If either of the lines have the pllickerian coordinates (a). where our equations (2) (XX) This will arise = 0. and therefore to leave the axes of the pencil unaltered. will. shall be called a proper cross. X. conversely. serve to determine the coaxal system.e. We shall therefore define such a pencil of tangents as an improper cross. neither of which is tangent to the Absolute. we merely have to reverse the sign of one of our radicals. i. and.CH. They will determine a pencil of coaxal complexes. then the three numbers (X) given by equations (1) may be taken to represent the cross.e. conversely. suppose that we have a triad of coordinates (X) which are homogeneous in the imaginary domain. Any mutually polar lines of the pencil of tangents. the Absolute polar of the line (d). when = . x THE HIGHER LINEGEOMETRY 117 complex coaxal with the given line will be obtained by multiplying the numbers (X) by (i + mi).e. X. and will be determined severally by lines intersecting the various tangents to the Absolute at this point. every such set of homogeneous values will determine a pencil of tangents to the Absolute. The coordinates of the lines of the corresponding cross will be found from (1) by assigning to p such a value that the coordinates For (a) shall satisfy the fundamental pluckerian identity. become illusory. We may then represent such a pencil of tangent lines by a set of homogeneous values (X ) where (XX) 0. imaginary) domain. when (aW^ZaJZatf^Q. Conversely. for the result of multiplying them through by (l + mi) is to replace the complex (a) by a coaxal complex. this it is necessary and sufficient that the imaginary part of 2 A P (XX) should vanish. There is one. To get the other line of the cross.
We sect. Improper crosses will correspond to points of the elliptic Absolute. Theorem . Theorem 1. (3) When will this indicate a transformation of point space? It is certainly necessary that improper crosses should go into improper crosses. This assemblage. and the assemblage of all points of the complex plane of elliptic space. condition for this in the case of two proper crosses will be (XY) (XY) * Geometrically a line may intersect either member of a cross. S. or pure imaginary points. shall say that two crosses intersect if their lines inter The N. is the general depending on eight complex. If a cross be improper. cross space let us say.z/= 2 i </**. and all lines in the plane of contact. hence the substitution must be of the ortho. reducible in point space. There exists a perfect one to one correspondence between the assemblage of all crosses in hyperbolic space. if the variables and coefficients be treated Geometrically we shall imagine that our contragrediently. The necessary and sufficient condition that two crosses of different layers should intersect orthogonally is that the corresponding line and point of the complex plane should be in united position. Kl^o. This ambiguity disappears in the case of perpendicular intersection. The assemblage of crosses which intersect a given cross orthogonally will be given by means of a linear equation. the crosses of one layer being represented by points and those of the other by lines in the complex plane. it collineation group of cross space. is irreducible in cross space. A linear equation will be transformed linearly into another linear equation. of crosses. crosses intersecting orthogonally will correspond to orthogonal points of the elliptic plane. we have then Theorem 3. . Two intersecting crosses will correspond to the cosine of whose distance is real. is doubly overassemblage laid. the assemblage of all crosses cutting orthogonally will be made up of all lines through the point of contact. 2. or sixteen real parameters The group .118 THE HIGHER LINEGEOMETRY OH.
3. the Absolute of hyperbolic space be transformed into itself. the real lines. Every collineation or correlation of hyperbolic space which leaves the Absolute invariant will be equivalent to an orthogonal substitution in cross space. 2. Also. A transformation which carries intersecting crosses into intersecting crosses may thus be interpreted either as a collineation. Suppose that we have a * ' chain of imaginary points is duo to Von Staudfc. p X = aY^ bZ { t . There chains.THE HIGHER LINEGEOMETRY 119 gonal type. namely. Accademia delle Sciense di Torino. . on those of two given crosses. Conversely. If this line be repre&ented in the Gauss If the line plane. or a congruent transformation coupled with a polar reciprocation in the Absolute. * is a theorem of very great generality connected with which we shall now give. it is immediately evident that a congruent transformation will transform cross space linearly into itself. pp. See his For an extension. loc. Let us now inquire as to what are the simplest figures of cross space. cit. The simplest one dimensional figure is the chain composed of all crosses whose coordinates are linearly dependent. xxv. will generate a linear pencil or a regulus. for a polar reciprocation in the Absolute of point space appears as the identical transformation of cross space. (of the other layer) called the axis of the chain. the pencils of tangents to the Absolute where it meets the actual line of the chain. Theorem 4. campo di ricerche geometriche/ Atti della E.* The crosses of the chain will cut orthogonally another cross The axis the chain will contain two improper crosses. one through each point of the chain. the chain will be represented by a circle. 1890. The concept 1 Beitrage'. Moreover. an orthogonal substitution in cross coordinates will carry an improper cross into an im proper cross. or a correlation of point space. be imaginary. see Segre. by means of real coefficients. so that our transformation of point space must be a congruent one. and every such orthogonal substitution may be interpreted either as a congruent transformation of hyperbolic space. ' vol. (4) Interpreting these equations in the complex plane we see that we have oo 1 points of a line so related that the cross ratio of any four is real. point space is not completely determined. being proper. and will carry intersecting crosses into other The corresponding transformation in intersecting crosses. i = 1.. Su un nuovo 13742.
get X = b(r + 2 si)^ JT3 = 0. Such a congruence shall be called synectlc. in point coordinates. perpendicular to (U) and (6) but one common all adjacent crosses. .120 THE HIGHER LINEGEOMETRY CH. v. We easily see that there will be two crosses of the chain which intersect orthogonally taking these and the axes to determine the coordinate system. xv. the equation of the surface obtained by splitting off from a chain its improper crosses. se Transactions of the Royal Irish Academy. and each Let us find. congruence of lines of such a nature that the corresponding cross coordinates (U) are analytic functions of two real parameters u. The common perpendiculars to a line. have then the theorem * : We Theorem 5. on Systems of Kays'. (5) There are two sharply distinct subcases. congruence. JSfj = a (p + qi). +~ We shall mean by the general position of a line in such a congruence. Let us exclude this case for the moment and pass to the other. of a nonsynectic adjacent line will generate a chain. vol. The cross of common perpendiculars to the cross (U) and the adjacent cross (U+dU) will be given by UJ dv. we may express our chain in the simple form . one where this determinant does not vanish. (a) U Here there is ^ ~ = 0. for euclidean space is due to Hamilton. in the general position. Eliminating a/ b we This gives the equation of the chain surface in point coordinates (ps * qr)( xQ 2 f x%) *i ^2 + (P r + ?s ) (^i 2 + *22 ) Va =  (8) The analogous theorem * his paper 1829.
The tions reciprocal to the chain congruence (9) will have equa.e. considered as an assem . If by a one Again. 0. This is made up of all crosses which have coordinates linearly dependent with real coefficients on those of three given crosses which do not cut a fourth orthogonally i \XYZ\^0. Leaving aside the special cases the following theorems may be proved for the general case. zi *J T .. c are real homogeneous variables. but not under the general group.THE HIGHER LINEGEOMETRY If 121 (psqr) = or (pr + qs) = 0. the congruence will be transformed into itself parameter group of rotations. we have two real and two imaginary linear pencils the conditions for this in cross coordinates will be invariant under the orthogonal. 0. = l. is real for any two the con gruence consists in all crosses through the point (1. There are various subcases under the congruent group. . i. The common perpendiculars to pairs of crosses of a chain congruence will generate a second chain congruence Each congruence is the locus of the axes in the other layer. having a strong similarity to the euclidean cylindroid. ^' TT' \ (10) Let the reader show that the chain congruence reduced to the canonical form may be where a. Theorem 6. The simplest two dimensional system of crosses is the chain congruence. b. +r *i TTTT T. (9) crosses which correspond to the assemblage of all points of the real domain of a plane will generate The a chain congruence. Here we see that (XX') crosses of the congruence. Theorem 7. of the GO 2 chains of the other . The chain congruence. 3. the two are said to be reciprocal to one another. The general form of our surface is a ruled quartic. let Qw _ qr) _ . 0). ( pr _ qt) = 0.j. 2. Theorem 8.
c.122 THE HIGHER LINEGEOMETRY OH. of the reciprocal congruence.) there will be but one X = so that our congruence is represented at by a curve. There will be found to be one singular case where the same cross has determinations. This corresponds to the fact that there will exist an equation 0. Another simple twoparameter system of following crosses is the pY { + qZ{ + sTi ^ 0. and. the tangent any point representing the common perpendicular just mentioned (in the other layer). we may suppose Y =Z =T = aY + bZ + c 2\ = aF2 + bZ2 + cT2 = 1 l 0. (abcpqr) real. The points and will be represented by two synectic congruences so tangents . If common perpendicular to a cross and its adjacent crosses. YZT =  0. orthogonally generate a quartic surface. r. / and these may always be solved. conversely. every curve will be represented by a synectic congruence. All these crosses cut orthogonally the cross YJ Conversely. let us show that every cross orthogonally interAs such a form secting (U) may be expressed in this form. is of the third order and class. blage of lines in point space. The two congruences have the same focal surface of order and It is generated class eight. + ir') A 2 which amount to four linear homogeneous equations in five unknowns a. those which meet such a line obliquely generate a regulus whose conjugate belongs to the reciprocal congruence. r . of crosses cutting a cross orthogonally is but a special case of what we have already defined as a oo ' The assemblage synectic congruence. by common perpendiculars to the pairs of lines Those lines of the congruence which meet a line of a regulus. b. We have then the equations (r (r f tr') A\. as this is invariant for all linear transformations. j> ( Xl z X.
the real point of interest of otc. ~ r k Our congruence curvature zero. . We shall not enter points with hypercomplex coordinates. 1 r./ cit. after * Cf. congruence is one of normals. Their orthogonal trajectories are a second set of geodesies whose tangents will generate a like congruence. if we look upon the congruence of lines generated by our crosses. ~ A. k tan 1 . It will be made up of crosses whose lines are normals to a series of surfaces of Gaussian curvature zero. then the Gaussian expression for the curvature of the surface at that point will be chapters. let us emphasize the distinction between these congruences and the nonsyneetic onevS. where the common perpendiculars to a cross and its adjacent ones A generate a chain. 328. real Did we wish to represent the imaginary as well as the members of a synectic or nonsynectic congruence. be written is made up i. we see that the two focal points on each are orthogonal and the two focal planes mutually From this we shall conclude that our lineperpendicular..x THE HIGHER LINEGEOMETRY We 123 related that each cross of one is a cross of striction of a cross of the other. may reach a still clearer idea of these congruences by anticipating some of the results of differential geometry to be proved in later" For. 1 7 k tan . We shall. &a In the present instance as the two focal points are orthogonal 1 k tan . we should be obliged to introduce into our representing plane. all. for. show in a later chapter that if r x and r 2 be the radii of curvature of normal sections of a surface in planes of curvature. moreover. r J^ k r2 k tan ~ k . Study. to which the lines of the other congruence are binormals. and the characteristics of the developable surfaces of the congruence will be geodesies of the focal surface. and all its adjacent crosses. of normals to surfaces of Gaussian to surfaces whose distance element may Theorem 9.* synectic congruence will represent the points of a curve of the complex plane.e. into this extension. The characteristics of the developable surfaces are geodesies of the focal surfaces. < Zur nichteuklidischeji p. . In conclusion.
start. Our doubly homogeneous coordinates have a second pretation which is inter Let us write the of the highest interest.124 THE HIGHER LINEGEOMETRY OH. so we may relate a cross of elliptic (or spherical) space to a pair of real The modus operandi is as follows points of two plane. a01 a23 = <r r A l . The assemblage of all real crosses of elliptic or spherical space may be put into one to one correspondence with the assemblage of all pairs of points one in each of two real planes. As we identify the geometry of the cross in hyperbolic space with that of a point of the complex plane. as before. and. namely.e. to give a real interpretation for the geometry of the complex plane. cross defined thereby. we merely multiply dX) ( r X) by two different constants. = pi A ! . may Tfoorem 10. The two separately homogeneous coordinate triads faX) (r X) be taken to represent this proper cross. the subject lies merely in this. every real pair of triads will correspond to a single cross. when we wish to move back from the independently homogeneous sets of coordinates (}X) ( r X) to the we replace our complex degenerate complexes of the pencil. with a pencil of coaxal linear complexes defined by v v : We a ol + a23 .e. Conversely. as all quantities involved so far are supposed to be real. ra oi = v()+ 7 . of the parameters determining the one and the other set of linear generators vn v/t *o The pliickerian coordinates of a generator of the left or right system will thus be Pol = P23 = 2 V 2 > Sfol = ?23 = . i. to the lines of the we have to take for p and a such values that the fundamental pliickerian identity is satisfied. i v. coordinates of a point of the Absolute in terms of two independent parameters. conversely. If shall by another coaxal therewith. i.
should be represented by identical points in the one or the other plane.x THE HIGHER LINEGEOMETRY (X) of 125 The parameter a left generator which meets a given line (a) will satisfy Similarly. In order that two crosses of different layers should intersect orthogonally. and assign a cross to the upper layer if it be determined by two points in the representing planes. however. 0. it is necessary and sufficient * The whole question < Study's Beitrage '.12) = 02Mlf*2 Kl We thus get as a necessary and sufficient condition that 2 (f*l tt lines should be right (left) paratactic. of left and right is considered pp. so here. 0. 0) to the two lines of the cross. while the second is 2 [(pMSpijPijTSpif Zpij' = If these equations be multiplied together. we shall look upon cross space as doubly overlaid. where the Chapter IX. If a line pass through the point = is met respectively by the left and the right paratactic through the point (1. plane # = called. we get (25) of (1. while it shall be assigned to the lower layer if it be determined by two lines. 0. It thus proportional to those of its intersections with x appears that in (11) the coordinates dX) and (r X) are nothing more nor less than the coordinates of the points. cit. shall speak of two crosses as being paratactic. while the first three are 0. 126. is that they fore. 0. respectively. It will. for a right generator ) we have tt . that the differences (sums) of complementary pairs of pllickerian coordinates in the one shall be proportional to the corresponding differences (sums) in the other.. If the lines be (p) and (p'). for a line paratactic to the one is also paratactic to the other. most carefully in . when their We and the necessary and sufficient condition thereinvariant under the group of cross space. 0) its last three pliickerian coordinates will vanish.* As in the hyperbolic case. the left and rig fit representing planes. 0. be more convenient to consider (lX) and ( r X) as standing for points in two different planes.<%) + i (Ml2 + M22 (02 ~ 3l) ~ /V) K>3 . the first of these conditions will be two [(P P') + I IPijPijTS Pij* Spy* = 0. Under these circumstances we may say lines are so. 156. : Theorem 11.
* We may. thus. from equations (19) and (20) of Chapter IX. the cosines of the Clifford We may We mean by angles will be fcXjF) (T X r Y) V( r Xr X] iYiY) Now. 0. however. or the mentary. the angles of right and loft paratactics to them through any chosen Let the reader show that the magnitude of these angles point.= sin  d' . The necessary and sufficient condition that two lines should intersect is that their Clifford angles should be equal or suppledistances. we easily find V(r X r X) V( r Y T Y) /d d' (13) or else / /d COS ( T \k + d =. = . . state the following theorem : Theorem 12.' cit. especially p. into which we shall not enter.) kJ = The ambiguity can be removed by establishing certain conventions with regard to the signs of the radicals. If. we see that d sin T sm 7. 0. cos cos ~= cos0 cos0' = _ " hence. The Clifford angles of two lines have the samemeasures as the sums and differences of the kth parts of their sums and differences of their angles. sin & . 130. shall go still further in this same direction. 0). the right and left Clifford angles of two crosses.. that they should be represented by line elements in the two planes. 'Beitrage. see Study. * For an elaborate discussion.126 THE HIGHER LINEGEOMETRY CH. is independent of the choice of the lastnamed point. we choose the point (1.
* Cf. and denote thereby a oo right improper crosses. KM%*0. while under 1Q the two sorts of parataxy are interchanged. cit.x THE HIGHER LINEGEOMETRY 127 When we adjoin the imaginary domain to the real one. 3 <^/ = 2%r*/. It will be made up of the sixteenparameter subgroup (? 16 of all transformations of the type crosses. . loc. corresponding to sets of values for which (jX }X) = ( r r X) The definitions of parataxy and orthogonal intersection may be extended to all cases. serious complications will arise which can only bo removed by careful definition. Improper crosses of the second sort shall be defined.. The general group of linear transformations of cross space will depend upon sixteen essential parameters. X = PlXi^aijlXj.* If dXfi) 0. ( r r X) = X ^ left generator of the Absolute. . wo merely give the facts. while u includes the assemblage of all symmetry transformations. . Without going into a complete discussion. conjoined to a nonparabolic involution 3 among the right generators. J (14) and the sixteenparameter assemblage J/16 of tions of the type all transforma Notice that under G 1Q left and right parataxy of crosses of the same layer are invariant. being composed entirely of two invariant subgroups j(? 8 r (?8 of which the former is made up of the general linear transformation for dX) with H H . the author's ' Dual projective Geometry '. 3. There will be co such improper left a improper 3 cross. Let the reader show that there can be no collineations of point space under G 16 except congruent transformations. as a subgroup. 0. and obvious. we shall say that we have 0. as in hyperbolic space. as pencils of tangents to the Absolute. and that the necessary and sufficient condition that (14) should represent a motion of point space is that the transformations of the two representing planes should be of the orthogonal type. The group G? 16 is halfsimple. the group of all motions. their analytic expression being as in the real domain. whose definition is Left and right improper crosses together will constitute what shall be called ivnproper crosses of the first sort. The group Cr 16 will contain.
take a congruence of crosses whose coordinates are analytic functions of two essential Let us further assume that (jF) (T T) parameters (u). For instance. when the relation is a congruent one. congruence. next. is a new oneparameter family of crosses called a strip. Let The assemblage of crosses so defined shall be called a chain. and its immediate neighbours. composed of two pencils. more A . this restriction is that neither (jF) nor ( r F) can be expressed as functions of a single parameter. arises. The properties of these chains are entirely analogous to those in the hyperbolic case. although possessing more variety in the real domain. and each strip shall be said left strip of the upper layer to be reciprocal to the other. factors of the group of motions with j(?8 and r Gs respectively. the common perpendiculars to a line in the general position. to pairs of crosses of the left strip will generate a right strip (whereof the definition is obvious).128 THE HIGHER LINEGEOMETRY CH. (v). the The highest common roles of (iX) and ( r X) are interchanged. The common perpendiculars exactly. will generate a chain. being crosses of the system r Y. Suppose. Chapter IX). that we have This or. the identical transformation for (r X). will be the groups of left and right translations (cf. while in the latter. The simplest assemblages of crosses in elliptic space bear a close analogy to those of hyperbolic space. represented by means of a homographic relation between the The meaning of points of two linear ranges in the representing planes. The chains of elliptic cross space will have the same subclassifications under the congruent group.Y r * ~ 0. so that the crosses of the congruence cannot be assembled into the 1 generators of oo surfaces. a left strip. as in the hyperbolic Let the reader show that the general chain may be plane. and that the special chain. those of each surface being paraLet the reader then show that for every such tactic.
The general congruence will have all of the properties mentioned in (8). (17) This congruence will contain oo 1 strips. In point space. no two of our quantities a^ in (16) be equal. and are minimal surfaces. The reciprocal strip in the lower layer will be represented by the pencil through the point in the left plane. stitute in the last b. whose other generators belong to the reciprocal strip. xxxvii. since they are generated by paratactic lines. and the line of the range in the right. vol. Zur nichteuklidischen Geometric. The common perpendiculars to all nonparatactic crosses of the congruence will generate a bundle. the lines of a strip are generators of a quadric. Owing to the parataxy of the generators of such a quadric. r We may solve the first three equations for a. c. it will intersect the Absolute in two generators of each set. we have the general congruence. The reciprocal congruence will be given by lU{ If the squares of =a ir U { . whose reciprocals generate the reciprocal congruence. ' COOLIDQE I . shall show in We We that these surfaces have Gaussian curvature zero. \ 1 YZT 1 1 x \ r Yr Z T =0. Kloin. may easily be reduced to the canonical r form (16) X = Utl X t t . the reciprocal * Cf. we have a bundle of crosses through a point.* The simplest two dimensional system of crosses will be. and sub This. if we have one such equality. There are various subclasses under the congruent group. when we wish to refer to it as a figure of point space. and a linear range of the right plane. A different sort of congruence will arise in the case where \ 1 T Z T\=^ 1 1 r YrZ r T\^0. since their asymptotic lines form an orthogonal system. the congruence will be transformed into itself by a oneparameter group of rotations. the chain congruence Chapter XV. 1890.x THE HIGHER LINEGEOMETRY 129 will be represented by a point of the left plane. as before. again.' Mathematische Annalen. those to paratactic crosses. shall call our quadric a Clifford surface. If all three squares be equal.
X = Pl Pl l <r cr r X l == 6.2 (18) = Pl X 3 . having oo 1 deter minations. we have ll F( X X X = 2l 3) Q. all to vanish. In general. require the first minors of have a bundle of paratactic on the other hand. and oo We shall show surfaces will have Gaussian curvature zero. r X. 3) the linecongruence can be assembled into oo 1 surfaces with l Such surfaces with right paratactic generators.WI = without the vanishing of the . Apparently nothing has ever been published concerning this type of congruence. <}>( r X lr X 2r X = 0. 0. . in addition to (17). The canonical form will be * 1 X = a lr X 1 19 If. we have fonn minors of either determinant. The crosses cutting a given cross orthogonally. ' 2 6. but a common line and paratactic axes. also in Chapter XVI that the lines of such a congruence are normals to a series of surfaces of Gaussian curvature zero. c. x congruence. 0) if l (0. equations of the congruence may be reduced to the canonical first oo 2 X = a. we have we we shall crosses. left. * .130 THE HIGHER LINEGEOMETRY CH.Fr z/r =o. cit. The theorems here given are taken from an unpublished section of the author's dissertation. In point space the line congruence will be of order and class two. V^3 = The cross (1. 0) will be singular. If. Such a congruence will be generated by the common perpendiculars to the paratactic lines of two pencils which have different centres and planes. 1.
shall be called the axis of the circle. 12 . j Note that a circle of radius radius is two lines. 1880. this curve has double K contact with the Absolute. 1874.xii. A To get the equation of the circle whose centre is (a) and whose radius is r. Let the reader show that all points of a circle are at constant distances from the axis. Sul rapporto anarmonico sezionale e tangenziale delle coniche. so that the circle is selfdual It is evident that : when Theorem \.(ax) 2 = 0. conversely. I cercoli nella geometria noncuclidea. (1 ) cos T =0. The locus of all points of a plane at a constant distance from a given point which is not on the Absolute is called a circle. The locus of all points of a plane at a constant distance from a given point thereof is a circle whose centre is the given point.' ibid. a distance whose measure becomes infinite in the limiting euclidean case.e.CHAPTER XI THE CIRCLE AND THE SPHERE THE simplest curvilinear figures in noneuclidean geometry are circles." Giornale di Matematica. hence.* Definition. Definition. * Theorem 1'. we have cos 2 * (ax)  V. see Riccordi. and it is now time to study their properties. be another circle. The envelope of all lines of a plane which make a constant angle with a given line is a circle having the given line as axis. which will also turn out to be its polar with regard to the circle. ' Battaglini. The given point shall be called the centre of the circle. Riccordi's results had previously been reached analytically by xviii. and. this shall be the measure of the distance of all points from the centre.. the secant of contact being the axis. ' ~. line through the centre of the circle shall be called a diameter. its absolute polar. every such curve of the second order will be a circle. and that circle of * For a very simple treatment of this subject by means of pure Geometry.(aa) V(xx) (aa) (xx) _= r 2 r COS 5 . The absolute polar of a circle will.is a line. i.
Let the reader show that these theorems hold also in the case of the horocycle. the circle will have two centres. 2'. The diameter through If (y) will   = \ have the equation 0. If the point (a) tend to approach the Absolute (analytically speaking) the equation (1) will tend to approach an indeterminate form. we indicate these two  lines yay (uv) \aya\. Theorem is A These simple theorems may be proved in a variety of ways. for the moment. + C (uxf circle and the equation of the horocycle takes the form (V + u 2 2 ) (xx) = 0. the axis will be actual. disappear in the elliptic case . For instance every circle will be transformed into itself by a reflection in any diameter. hence the tangent where the diameter meets the curve must be perpendicular to the diameter. the coordinates of the axis. naturally. if AC. while an equidistant curve would These distinctions will. Such a curve shall be called a horocycle^ the point of contact being If (u) be called the centre. to the real domain of the hyperbolic plane. The limiting form for the curve will bo a conic having fourpoint contact with the Absolute. Notice that to a dweller in a small region of the hyperbolic space. have what may be more properly called a proper circle.132 THE CIRCLE AND THE SPHERE CH. we see that if the centre be ideal. A tangent to a circle is perpendicular to the diameter through the point of contact. then . = (aa) by . a proper circle would appear much as does a euclidean circle to a euclidean dweller. G will be perpendicular to BG'. which are equivalent points. and the curve will appear in the actual domain as the locus of points at a constant distance from the In this case the circle is sometimes axis. the equation of the tangent to the circle (1) at a point (y) will be AB B (xy) (aa) N(ax) xya (ay) = 0. Restricting ourselves. then let up on this centre of gravity. Theorem 2. an actual line. appear like two parallel lines. If the centre be actual we shall called an equidistant curve. and the common tangent the axis. == Or. in the spherical. a line from A to one centre of gravity and G close of B. lastly.N (ay) (u) and (v). we have (uu) = 0. again. Or. point on a orthogonal to the point where the tangent thereat meets the axis.
PO = r tan T A0.7 . limit ^i= = _. If two tangents a Theorem 4'. 133 Theorem 3. r /O sin y tan A ^ (W . and the diameter perpendicular to these lines. the locus of their point of intersection is a concentric circle (horocycle). pole.. in the infinitesimal. If two points of are at a to a circle (horocycle) make a constant angle. T dO. . . The APAP' and AP'PP" are isosceles. 7 = d / v . by dx = r ksin. . The locus of the centres of gravity of pairs of points of a circle whose lines are concurrent on the axis. f limit pp' =< pptA PP'dn jr But tan r5 /O PQ = ' .. is the point of concurrence.XI THE CIRCLE AND THE SPHERE Theorem^'. The element Chapter IV (5). and let P" be the point on the tangent at P whose distance from P equals PP'.theenvelope of their line is a coaxal circle (horocycle). . equals ds. hence tan Fp T)/. and its absolute Theorem 4. A0 = . Let A</> be the angle between the tangents. fv by IV limit tan (6). PP' k sin T~ tan A 4* tan ^ ~ 4 ' = P P" 2/o . circle (horocycle) constant distance. of arc of a circle of radius (r) will be. 2P'P i . . Ic The circumference of the circle is thus r k sin 7 dO KjQ circle r C"2ir = 277/0 sin 7 Iv Let the tangents at P and P' meet at Q. or. The envelope of the bisectors of the angles of tangents to a circle from points of a diameter. is this diameter. the centre of the being A. limit .
The difficulty of visualization disappears in the hyperbolic case where we take one at least of the circles as an equidistant curve. which are also the absolute polars of the equivalent circle. by Chapter VIII.. the equation is reducible. sixth edition. known. expect. in pairs. In the general case two circles. point will have one absolute polar. When. we see how two circles there also can intersect in four real In the spherical case. . shall subsequently define this expression as the curvasee that. Two real circles cannot intersect in A more than two which intersect in four points will have three The problem of finding the common pairs of common secants. the two conies have double contact with a third.. secants of two conies will. however.134 THE CIRCLE AND THE SPHERE v . and one pair of secants appears which intersect on the chords of contact. the locus of all points which are identical with their equivalents. The absolute polar of a circle is two equivalent circles. p. If we identify the euclidean hemisphere. the absolute polars of such circles will intersect in four real points. They will Theorem 5. of course. it is constant. lead to an irreducible equation of the third degree. or imaginary. 1879. well 242. as we should We shall next take up simple systems of circles. London. will intersect in four points. a line two equivalent absolute poles. neither of which is a line. L r IV j We We ture of the circle at the point (P). with the elliptic plane. leave to the reader the task of making the slight modifications in what follows necessary to adapt it to the case of spherical geometry. Salmon. are circles * Two real points. in general. real.* In the case of two circles these secants shall be called the radical axes. This theorem is. common secants called radical two intersections of these. lie axes of the circles and haron the line of centres. We will If two circles lie completely without one another they have four real common tangents. If two circles Theorem 5'. Conic Sections. Cf. two have four common tangents. where opposite points of the equator are considered identical. the Absolute is points. pp'z k tan . axes are concurrent with the called centres of similitude. If two circles intersect in four points. and are harmonically separated by them. limit ~ OH* Hence ds =r limit .
They are perpendicular to one another and to the line of cencentres of gravity of the intersections of the circles with a radical axis are the intersections with the other radical axis and with the line of centres. angles of the tangents at a centre of similitude are the line of centres and the line to the other centre of similitude.(bx)* = 0. point. If the equations of the two circles be cos 2 ^ (aa) (xx) V (66) . the points of contact of tangents to all of them from a point of the line lie on a circle Theorem 6'. be the cosines. Consider the assemblage of all circles through two given If the line connecting the two points be a radical points. the envelope of tangents to whose centre is this them at the points where they meet a line through this a centre of similitude will be circle with this line as axis. Theorem 6. of the kth parts of the distances from (x) to the points of contact of tangents. by Ch. axis for two of these circles it will be perpendicular to their line of centres at one centre of gravity of the two points. thence to the two circles.XI THE CIRCLE AND THE SPHERE centres 135 monically separated by them. the equations of the radical axes will be ( cos ~~ (ax) f cos ~ V(aa) (bx)) r~i _____ ______ cos(ty* 1C V(bb) (ax) cos ~r V(aa) (I = 0. tres.(ax) 2 = 0. cos 2 ^ (66) (xx) . IV (4). (3) The last factor equated to zero will give (ax) (bx) _ ?\ 1 V(bb) COSy AC cos~ k ? and the two sides of this equation will. If a set of circles through two points have the line of these points as a radical axis. and in every case a perpendicular from the centre of a circle on . harmonically separated by the and are mutually The The bisectors of orthogonal. If a set of circles tangent to two given lines have the intersection of the lines as a centre of similitude.
they are concentric. circles touch one another. We thus see Theorem of all 7. Two circles of the same family. Four circles will three given Each line connecting points. a secant will meet at a centre of gravity of the two points of the circle on that line. same family.(cuf = 0. circles The assemblage through two Theorem 7'. two of the given points will be a radical axis for two pairs of circles. Theorem pass 8. The radical axes of three circles pass by threes through four points. of all circles tangent to two lines will fall into two families according as the centres lie on the one or the other bisector of the angles of the lines. The assemblage common two points will fall into families according as the perpendicular from the centre on the line of these points passes through the one or the other of their centres of Two circles of the gravity. . ^ (66) (*)  ( = 0. The centres of similitude of three circles lie by threes on four lines. Two circles will have double contact when. and consider the circles that pass through all three. It is here centric. their common replaces one radical axis. will have the line as a radical axis. and the point of contact one tangent centre of similitude. may be constructed so that the points of contact of tangents common to them and to each of three given circles form two pairs of orthogonal points. Four circles will lines. intersect assumed that no two of the given circles are conThere is no reason to expect that because two circles at right angles in two points they will in the other circles . Four circles Theorem 10'. Theorem 9'.136 THE CIRCLE AND THE SPHERE it CH. and they only. Let us now take a third point. Theorem 9. two. through Theorem 8'. will have the intersection of the lines as a centre of similitude. and only when. Let the r ] be cos 2 cos 2 (aa) (xx) . get at once Of course when two We from (6) and (9) Theorem be 10. and they only. Each intersection of two lines will touch three given be a centre of similitude for two pairs of circles. Four circles constructed to cut may each of three circles at right angles twice.
/c  = } 9 *J (n<i\ =J?^===.(by)* (4) T T / / cos~ cos ~ V(aa) V(bb) sin A/ / sin ~ ic V(aa) V(bb) This gives two values and only when J for the angle . The cosine of the angle formed by them (yy) ( (t will be y) COS0 = (a6) (by) (ab) V(yy) (aa)(ayf V(yy) (bb) . geometrically evident.xi THE CIRCLE AND THE SPHERE Let (y) 137 be a point of intersection . 7 . which will be equal when. \/ (hl>\ . The necessary and sufficient condition that two circles should determine by their points of contact. (6) these last two facts being.= . The necessary and sufficient condition that two circles should cut at the same angle at all points is that their centres should be Theorem 11'. (ab) = 0. see that two circles cannot have four rectangular intersections. common tangents. for if (a&) We = 0. is that their axes should be mutually perpendicular. The condition cos of contact will be cos f (^ ~ V / = ) \ 2 A. the lines thence to the centres are j xya \ = 0. K> (7) the circle is a line. congruent distances on all four mutually orthogonal. Theorem 11. Notice that these two conditions are really identical. also. cos 5 = 0.  xyb =  0. (5) and of orthogonal intersection 9 l cos T} cos  r r = (ab) ^L . .
The common tangent planes to two spheres envelop two cones of revoluare vertices whose tion and orthogonal mutually harmonically separated by the centres. The necessary and sufficient condition that two spheres should. a quadric with conical contact with the Absolute. by their contact. point. the plane of conical contact the axial plane of the sphere. Theorem 13. with the centres of the spheres. Theorem not 14. Theorem 15. containing Three spheres a common circle will meet in three pairs of circles whose planes are collinear by threes in four lines. form a desmic configuration. Theorem. Three spheres not tangent to a cone of revolution have three such pairs of common tangent cones whose vertices are collinear in threes on four lines. when not a plane. determine congruent . and are harmonically separated by the axial planes. 13'. sphere is the locus of all points at a constant distance from a given point not on the Absolute. a quadric A Note that a plane and point are special cases of the sphere. It is. Two spheres will intersect in two circles whose planes are perpendicular to the line of centres and to one another. It is. We Theorem 12.138 THE CIRCLE AND THE SPHERE CH. Theorem 14'. Four spheres whose axial planes are not concurrent are enveloped in pairs by twelve cones of revolution whose vertices lie by sixes in eight planes which. when not a with conical contact with the Absolute. A Theorem 12'. sphere is the envelope of planes meeting at a constant angle a plane which is not tangent to the Absolute. with the axial planes. shall define as a sphere that surface which is the locus of all points of space at congruent distances from a point not on the Absolute. Theorem 16. a tangent plane is perpendicular to the line from the point of contact to the centre. intersect in twelve whose planes pass by sixes through eight points which. The fixed point shall be called the centre. The necessary and sufficient condition that two spheres should cut at the same angle along their two Theorem 16'. determine a desmic configuration. A line connecting any point with the centre of a sphere is perpendicular to the polar plane of the point. to the diameter through the point of contact let us say. Four spheres whose centres are not coplanar circles Theorem 15'.
To each point Vx^x^ We = Conversely. Let us then write shall terminate this chapter We = px x^ py = &2 . will correspond a point of euclidean space above the plane z 0. = 2yt. is that their axial planes should be mutually perpendicular. not on. by adjoining to our domain of rationality a square root of minus one. THE CIRCLE AND THE SPHERE is 139 should gonal. c. t and a hyperbolic 1. not on the Absolute. a&i = 2xt. 0. we may. or on the z plane. b. ' Nichteuklidische Cykliden/ Munich. 1906. y. Let us choose that for which the real part of x^ x^ is greater than zero. have adapted the notation to conform to our own usage. thus say that to every point of hyperbolic space. or on the Absolute. distinguish between the may imaginary roots.xi circles . p. Leipzig. by Poincare. so that real and actual points will correspond to real ones. We . (8) of hyperbolic space will correspond two points of euclidean space. The transformation is real. and so choose one in particular. p =# &J. 258. d) ((6 + & 2 + c 2 cZ 2 r2 )=p 2 . pz Vx*~x*x*'x*. This is fruitfully used in the dissertation of Munich. by giving an unusually elegant transformation from euclidean to noneuclidean space. <T#O we 2 get from (8) 42 r = x* + y 2 . If we write for short 2 (a. 1906. will correspond a point of hyperbolic space. and to each points of the Absolute will correspond points of this plane. and L. (10) * This transformation seems to have been first given in the second edition of Wissenschaft und ffypothese. be that their centres mutually ortho distances on the generators of the two circumscribed cones. the equation of this sphere may be written = 0. translated by F. When the real part vanishes. Suppose that we have a euclidean sphere of centre and radius r. = a* + y* + z*t* *xz crx2 9 (9) and to each point of euclidean space.* Let us assume that we have a euclidean space where a point has the homogeneous coordinates #. Lindemann. above. a point being given by our usual (x) space for which k* coordinates.
that is. a hyperbolic sphere for which that is.il'.. d. (a J' (6j ii'j t\ =1 ^ . Transforming we a factor #3 which corresponds to the euclidean plane at infinity. will which is perpendicular to the plane (i 3 correspond to a euclidean line.v2 ^j^a) 3 ) (* *i * a a *A 2 we get from (9) . one of th6 euclidean spheres would lie wholly above the o plane. two spheres which are reflections If the hyperbolic sphere were of one another in the plane. Suppose that we have two euclidean spheres given by an equation of the A We . We may say that (leaving aside special cases) a hyperbolic sphere will correspond to so much of a euclidean sphere as is above or in the plane of so much plane. and to the reflection in the of the sphere as is below it. 26d. (XQ + #3) 2 rf (ax { + Ix2 f c This is a sphere of hyperbolic space whose centre 2 is ((P+y> . ^/a 2 2 2 6 if we have 2 the hyperbolic sphere 2 2 a.> = coshv^rcw'j . i.a 3 ) (^'2 + [(fl // I s a) . j^d  a ). real and actual. one whose centre is in the plane will correspond to a plane in hyperbolic = space. may go a step further in this direction. and the other wholly below it.140 THE CIRCLE AND THE SPHERE get.2^^ 2(i 2 = 2 cosh rj v/V ~ ^+ + (fi^ V VA  aa)^]_ 3 c ^ ( 13 ) We have here two spheres which differ merely in the coordinate of their centre. A euclidean sphere for which c 0. 2 ad. will correspond to a plane euclidean circle perpendicular to the in euclidean space.e. after splitting off CH. s plane will correspond to a hyperbolic line. and whose radius r x is given by c i cosh Conversely. one whose centre is in the plane which corresponds to the euclidean piano at infinity. a hyperbolic circle <i = 0.
and the condition that they shall be mutually orthogonal is that %%' + Cti&i + A 2 A. surface will go into any lines of curvature of the corresponding surface. But this gives immediately that the corresponding hyperbolic spheres are also mutually orthogonal.xi THE CIRCLE AND THE SPHERE 141 type (13). and conversely. thus have a correspondence of orthogonal spheres to orthogonal see next that the lines of curvature of any spheres.2 cosh r l cosh r/ cosh rj cosh r/ ' A/ei 2 A^ A/ * A. namely./ */A '* 2 2 <i/ cr/ f <i /2 3 (W = 0. we might in We We a similar manner establish a correspondence between spheres of euclidean and of elliptic space. the intersections are lines of curvature. and hence the DarbouxDupin theorem must hold in hyperbolic space. in any triply orthogonal system of surfaces. Were we willing to sacrifice the real domain. .
Four imaginary absolute points and tangents. vol. real and distinct absolute points Two coincident.' American Journal of Mathematics. Convex hyperbolic parabolas. Four real absolute points. This may be done analytically by means of Weierstrass's elementary divisors.' and 'Teoremi sulle coniche*. Two coincident. (6) * The treatment of conies in the present chapter is in close accord with three articles by D'Ovidio. and two and distinct absolute points.CHAPTER XII CONIC SECTIONS of the metrical properties of conies in the nonis. (4) Ellipses. and two conjugate (7) Elliptic parabolas. two real (5) Concave hyperbolic parabolas. See also Story . real distinct absolute point. Nichteuklidische Geometrie/ in the Sammlunft ' * Schubert^ xlix. 1891. however. Four real absolute points.* First of all. * On the noneuclidean Properties of Conies. THE study euclidean plane. not uncommon in Italian mathematical publications. four absolute tangents. that the theorems are not given in distinctive type. We We Jl) absolute tangents. . Leipzig. Die nichteuklidische Geometrie in analytischer Behandlung. in relation to their intersections with the Absolute. vol. Concave hyperbolas. imaginary Three real coincident. no real (1) Convex hyperbolas. Two real and two imaginary absolute (3) Semihyperbolas. 'Le proprieta focali delle coniche. all in the Atti deUa R. xxvi. 1904. Two coincident. Accademia delle These articles suffer from the curious Scienze di Torino. v." Killing. in the last analysis. but the geometric question shall begin is so easy that we give the results merely. and Liebmann. but rather try to pick out those metrical properties of noneuclidean conies which bear the closest analogy to the corresponding euclidean properties. shall not. and the same for absolute tangents. that is.' Sulle coniche confocali. let us classify our conies under the real congruent group . 1882. 1885. Two coincident. blemish. and one (8) Osculating parabolas. nothing more nor less than a study of the invariants and covariants of two conies. points and tangents. and two conjugate imaginary absolute tangents. ' Leipzig. go into general questions of invariant theory here. and and tangents. absolute points and tangents. with the real conies in the actual domain of hyperbolic space.
Each axis of a central conic is a bisector of an angle of each pair of tangents to the current thereon. conic con The three pairs of lines which connect the pairs of intersections of a central conic with the Absolute shall be called The three pairs of intersections its pairs of focal lines. Theorem 2. and not the exact definition of Chapter I. which is meaningless except in a reThere will. Theorem I'. Taking this bola. or spherical. except in the case of the semihyperIn the elliptic case it will surely be real. as the coordinate triangle we may write the equation of the Absolute in typical form.e. In the real merely (1) Ellipses. we shall use the letters h. (2) Circles. 1. arise no confusion stricted domain. 143 (10) Proper circles. Theorem 2'. Conj ugate lines through a focus of a conic are mutually perpendicular. 1. but not concurrent lines. vertices of the from this. assume that no two of our c's are equal. that in speaking of triangle in this sense we are using the terminology of projective geometry where a triangle is a figure of three coplanar. we shall have i. Theorem Each centre of a central conic is a centre of gravity for every pair of points of the conic collinear therewith. Our plane being x. (11) Horocycles.CH. of its absolute tangents shall be called its pairs of foci. xii (9) CONIC SECTIONS Equidistant curves. while its sides are called the axes. k I as a circular permutation of the numbers 0.. 2. Conj ugate points of a focal line of a conic are mutually orthogonal. to those which cut the Absolute in four distinct points. however. and that none them are equal to zero. . plane. and define the of y We common selfconjugate triangles as centres of Be it noticed the conic.3 Q. while that of the conic is A In what follows we shall limit ourselves to central conies. elliptic. real central conic in the actual domain of the hyperbolic plane will have a common selfconjugate triangle with the Absolute which is real.
ck _____ __  ) : Vc k (ty <?. and are perpendicular to the opposite axis.. PFl  J!l"~^) + c l ( c l ( 'k .144 CONIC SECTIONS 3. Theorem Two focal lines Theorem central Two foci of a of a central conic pass through each vertex. (3) The polars of the foci with regard to the conic shall be called directrices. directrix d h perpendicular to the axis x h will have the A equation y ^_^ ^3^ + C = Q (4} Let (x) be a point of the conic. through the centre 0. and are orthogonal to the opposite centre. will be uh uk : : Ui = Vc : Vc h . quadric lie on each axis. the poles of the focal lines its director points.c k : Vl^7 h __ . Eliminating x h by means f(1)weget We then h ( h have If dfr be the corresponding directrix ( rfa^Hfa^ h <h the signs of the radicals in the numerators of the two expressions being the same . The coordinates u = jt of the focal lines /////. CH. 3'. (2) The coordinates will be of the foci t F^ Fk on the opposite axis xh xk x : : = ___ : { (c h .).
. Theorem 4'. (11) COS ck c. lv T{ ^ K (10) csc r.+ csc r2 csc r^ + csc r1 esc PF . sin PfY rC . and so It is also geometrically evident that for the lower sign. = : 1. : A/ 7^ sin Pf] sm 1 y. The ratio of the sines of the angles which a tangent to a central conic makes with a the absolute focal line and polar of the corresponding director point is constant. T7T ** ** 7JT f TX tan^ (9) Vf > 2 ^ 2 . and in the case of a hyperbola the lower sign (when in the real domain). : sin PF~ k j** rC . . .The ratio of the sines of the &th parts of the distances from a point of a central conic to a focus and to the corresponding directrix is constant. . in the case of an ellipse we must take the upper.XTT CONIC SECTIONS 145 Theorem 4. sin y^ sin 1C ft . (13) With regard to the ambiguity of signs the upper sign in (12) will go with the upper sign throughout in (13).esc ~.
and the pairs of lines thence to its pairs of corre sponding foci. semihyperbola make a with two real focal lines is through a centre Reverting to our point (x) constant. The pole of a with regard to a central conic lies on one bisector of the angle determined at each Theorem 8. The sum of the distances from real points of an ellipse and the difference of the distances from real points of a hyperbola or semihyperbola to two real the same axis is Theorem 5'. Let us now recall Desargues' theorem. 8'. A dual theorem will of course hold for a central conic. we see V/^ sin y^ sin ch Pf k Pf ^ K ' = c f ~~ CT. and with its pairs of correspooding focal lines. or the difference of the angles which the real tangents foci on to a concave hyperbola or constant. Theorem 7. The product of the sines of the ith parts of the distances from a point of a central conic to two focal lines through the same centre is constant. The polar of a point with regard to a central one conic passes through centre of gravity of the intersections of each focal line with the tangents from the point to the conic. and the pairs of foci. all have the same centres of gravity. whereby a transversal meets the conies of a pencil in pairs of points of an involution. This will apply to a central conic. Theorem 7'. form angles with the same two bisectors.146 CONIC SECTIONS CH. . The intersections of a line with a central conic. Theorem line focus by the lines thence to the intersections of the given line with the conic. Theorem 6'. ( Theorem 6. the Absolute. The tangents from a point to a central conic. and the pairs of focal lines. The sum of the angles which the real tangents to an ellipse or convex hyperbola. The product of the sines of the &th parts of the distances to a tangent from two foci of a central conic on the same axis is constant. the Absolute. Theorem 5.
the points of the focal lines orthogonal to their intersections with the line.xii CONIC SECTIONS 147 variable point of a conic will determine projective pencils at any two fixed points thereof. the distance which these lines cut on any focal line is constant. whence tangents to two coiiics form a harmonic set. as in the euclidean case. 2 "2 = 2 ( W ) + 2 M . If a variable a central conic be Recalling the properties of the given conies and a line : el even point conic of two Theorem 10. the two lines through the point which are mutually conjugate and perpendicular. Theorem 10'. and these will meet any line in projective ranges. The envelope of lines which meet a central conic in pairs of mutually orthogonal points is a conic touching the tangents to the given circle from its director points. and the three centres lie on a conic. A Theorem Theorem to tangent brought to intersect two fixed tangents thereof. ?<.22 (uv) = 0. the angle of the lines from a chosen focus to the two constant. Theorem II'. diculars on the line from the and the three axes all touch a conic. dicular tangents from the point (y) be (ux) "2 = 0. If a point and a central conic be given. is a conic passing through the points of contact with the common tangents. be If the mutually perpencircle. intersections is 9'. If a line and a central conic be given. "2 ^^ = ^^= ^ u. It is a wellknown theorem that the locus of points. a It is clear that neither of these conies will. The locus of points whence tangents to a central conic are mutually perpendicular is a conic meeting the given conic where it meets its directrices. If a variable of a central conic be point connected with two fixed points thereof. (vx) 2 = 0. the two mutually conjugate and orthogonal points of the line. . Theorem 11. in general. the perpenfoci. hence 9.
centre is constant. of a real focal line. the corresponding focal Let (y) be the coordinates of a point P of our conic.. .P. is is whose a circle whose corresponding axis line. The sum of the squares of the tangents of the angles which an axis of a central conic makes with a pair of tangents to the curve from two conjugate points of this axis is constant.2 CH. The sum the squares of the tangents of the &th parts of the distances from a centre of a central conic to any pair of intersections with two conjugate lines through this Theorem 13'.. yTT. is a centre is the focus. sn Sin <y/ ( . We shall call two such diameters as 0. J. (14) h Let the reader show that the equation of the other conic will be 0. Oj P' conjugate t diameters. The envelope of the reflection in a variable point of an ellipse. circle Theorem 12'.^ z .2 We may extend the usual euclidean proof to the following theorems  first of the Theorem 12.i  rx II This will meet the conic in two points P' having the coordinates _ .J. The equation of a line through the centre O h conjugate to the line n P will W \J 1 . f tan" y /V of 9 = (15) Theorem 13.1J hp> ILJC si ft I /y I st ni fj..148 CONIC SECTIONS 0. The locus of the reflection of a real focus of an ellipse in a variable tangent. tan 2 OP ~T/I* .
Theorem The product Theorem The product of the tangents of the /cbh parts of the distances from a centre of a central conic to two intersections with a pair of conjugate diameters through that centre. The equation of the tangent t' at the point P' is From this we get : sin 2 tan Oh P tan (17) . Theorem 15'. The sum the squares of the cotangents of the fcth parts of the distances from a centre of a central conic to two intersections of the curve with mutually perpendicular diameters through this centre is constant. multiplied by the sine of the &th part of the distance of these points is constant. h CM 2 of A. ( cu ( CH  2 cur . 149 14'. multiplied by the sine of the angle of these diameters is constant.XII CONIC SECTIONS 14. The sum of the squares of the cotangents of the angles which an axis of a central conic makes with two tangents from a pair of orthogonal points of this axis is constant.. k r (16) Theorem 15. . . of the tangents of the angles which an axis of a central conic makes with two tangents to it from a pair of conjugate points of this axis. The equation of a line through the centre Oh perpendicular This will meet the conic in points P" having coordinates * .
already studied in the last chapter. If a circle have double contact with a conic. Theorem The product Theorem The product of the tangents of the &th parts of the distances from a centre of a central conic to a point of the curve and to the tangent where the curve meets a diameter conjugate to that from the centre to the point of the curve. The product of Theorem The product of the squares of the tangents of the 2 /cth parts of the distances determined by a central conic on the axes is 1. its Theorem 19'. Pfr be the intersections of the x h axis with the conic cos ' ^VY = * fl < tan f 2 p jjT* . CH. 16'. the figure of two conies having double contact with a third. is constant. stant.150 CONIC SECTIONS 16. of the tangents of the &th parts of the distances of two conjugate points of an axis from either centre on this axis is constant. axis is y G y ~ c Theorem 17. If a circle have double contact with a conic. i^jA tan. is con of the tangents of the angles which an axis of a central conic makes with a tangent and with the absolute polar of a point of contact with a tangent from a point of this axis conjugate to the intersection with the given tangent. equal to the squares of the tangents of the halfangles of the pairs of tangents to a central conic from its centres is constant. The equations of two conjugate diameters through 0^ have already been written The product of the tangents of the angles which they with the Xj.1 WP' ** . make The product Theorem 17'. tan 2 \ ~ = L 18'. Theorem 19. The product of the tangents of the angles which two conjugate diameters through a centre make with either axis through this centre is constant. If a circle have double contact with a conic. its axis and the lines connect centre and the intersections . we have. Let Ph . (18) Theorem 18. with the Absolute.
their centres being the centres of the conic. that the tangent of the kih part of the distance from a point to a line. is the cotangent of the /cth part of its distance to the pole of the line. having . ( h .xii CONIC SECTIONS of the 151 ing the points of contact are harmonically separated by a pair of focal lines. P 0. Q'. P'. If the tangents of the angles which the tangents to a circle make with a diameter be altered in a con . There will clearly be six such circles. <> J: Let us remark.2 Of course we mean by foci and focal lines of any we mean in the special case of the central conic. having The same coordinates line will meet the conic in points P. pair of A circle which has double contact with a central conic where the latter meets an axis is called an auxiliary circle. conic what This will meet the line (u) through coordinates O h in points Q. and that if the tangents of two distances bear a constant ratio. If the tangents of the /cth parts of the distances from the points of a to any diameter be circle Theorem 20'. Consider the circle having its centre at O k while it has double contact with our central conic at the intersections with x = 0.~~~ 1 U T.. common tangents are harmonically separated by a foci. finally. so do their cotangents : Theorem 20.
II. homothetic to a given conic will pass through every point of space.2 with the absolute pole of __ *y J*1 ^^. One conic conline. The product of the tangents of the angles which an axis of a central conic makes with the lines from a point of the curve to the intersections of the curve with this axis is constant. One conic Theorem focal 22'. (20) ' V The tangents to a central conic from a centre shall be called asymptotes. altered in a constant ratio.152 CONIC SECTIONS CH. (21) (y) will The tangent at the point P with coordinates them in two points jR. The equation of the pair of asymptotes through the centre (0. the locus of the resulting points will be a conic having the given circle as an auxiliary. If they have the same four absolute tangents they shall be called confocal. 0. =0. This line is also perpendicular to the absolute polar of the given point. and two will touch #very line. The product of the tangents of the &bh parts of the distances from a centre of a central conic to the intersection with the asymptotes through that centre of a tangent is constant. so that the conic and its absolute polar conic are geodesically parallel The equation of the normal to our conic (1) will be curves. not through a point with a given conic will every touch and two will pass through every point not on the common tangents . the envelope of the resulting lines will be a conic having the given circle as an auxiliary circle.t) will evidently be 0. R\ whose coordinates are meet tan k tan = k Theorem 21. stant ratio. Theorem 21'. is The normal it at any point of a conic its the line connecting tangent. sot of conies which meet the Absolute in the same four points shall be said to be homothetic. get at once from Desargues' involution theorem : A We Theorem 22...
2 " . 153 common to all the conies. however. and the calculations are too fatiguing to make it advisable to carry them through twice. The tangents to these will bisect the angles of the pairs of tangents from that point to all of the confocal two conies. (2 Q\ ' * ^n^hci) h 0.XII CONIC SECTIONS to all. i. the general equation = If Aj 0. for in the next chapter we shall work at length the more interesting corresponding problem in three dimensions. that point will trace a confocal ellipse. namely. and pulled taut at a point. in the centres of gravity of all pairs of intersections of the homothetic conies with this line. Let us write ci in place of a/.. then..2 (27) With the aid of these coordinates. It is sometimes useful to modify the second of these equations. X} _ y V(xx) Our confocal conies have.c l)( c h*i)(Ch*2) 0. . (25) and A 2 be the parameter values of the conic through (X) ( we have c k.e. Concentric circles are a special case both of homothetic and of confocal conies. The general form for the equations of conies homothetic and confocal respectively to our conic (1) will be (23). if a loop of thread be cast about an extremely thin elliptic disk. we may easily prove for the noneuclidean case Graves' theorem. We shall not give the details here. the two parameters giving the conies of the confocal system which pass through it. in order to introduce the elliptic coordinates of a point.
imaginary developable. The names adopted are intended to We shall give a certain idea of the shape of the surface. Imaginary quartic curve and developable. A. and developable.CHAPTER XIII QUADRIC SURFACES THE discussion of noneuclidean quadric surfaces may be carried on in the same spirit as that of conic sections in the preceding chapter. Let us begin by classifying noneuclidean quadrics under the group of real congruent transformations. however. Elementary Di visors. makes use of Weierstrassian ibid. Real quartic curve. vol. fact that they had been preceded by rather a crude article by Barbarin.*" begin in the actual domain of hyperbolic space.' Memozrcs couronnes par The ' article . iv. the curve of intersection of the surface and Absolute. The Classification of Quadratic Loci/ The latter. Both Professor Bromwich and the author wrote in ignorance of the. Central Quadrics. (5) Onesheeted ruled hyperboloid. nonruled hyperboloid. ' ' 1'Academie de Belgique.. Concave. last Real quartic curve and de velopable. Real quartic curve (4) Twosheeted ruled hyperboloid. and it seemed wiser to avoid the introduction of these into the present work. J903. The * two surfaces differ from the preceding ones in that ' classification here given is that which appears in the author's Transactions of the American Quadric Surfaces in Hyperbolic Space This classification was simplified and Mathematical Societyj vol. 1905. vi. however. put into better shape by Bromwich. giving only those surfaces which have a real part in that domain and a nonvanishing discriminant. 1900. developable. There is not. Real quartic curve. the same wealth of easy and interesting theorems. Real quartic curve and (6) Nonruled semihyperboloid. (1) (2) Ellipsoid. vol. Real quartic curve and developable. owing to the greater complication in the formation of the simultaneous covariants of two quadrics. imaginary developable. (7) Ruled semihyperboloid. . while developable is the developable of common We tangent planes. vi. mean by curve. Etude de ge*ome"trie noneuclidienno. (3) Convex nonruled hyperboloid.
real mutually tangent conies. pidal quartic curve. is two real conies whose planes meet in an ideal line. real developable. B. real developable. . (18) Nonruled osculating semihyperbolic paraboloid. Imaginary quartic curve with real acnode. (16) Horocyclic elliptic paraboloid. Curve is two imaginary conies in imaginary developable. real (19) Prolate spheroid. and two imaginary lines. conies whose planes meet in an ideal line. imaginary developable. real ultrainfinite planes. Curve is two mutually tangent imaginary conies. Curve two real (23) Ruled hyperboloid of revolution. quartic. The curve is a real conic and two conjugate imaginary generators meeting on it. xia QUADRIC SURFACES 155 here two vertices of the common selfconjugate tetrahedron (in the sense of projective geometry) of the surface and Absolute arc conjugate imagiiiaries. Absolute curve two real conies whose planes meet in an ideal line. Real crunodal Real cus(13) Cuspidal nonruled hyperbolic paraboloid. Curve (21) Concave nonruled hyperboloid of revolution. quartic (15) Horocyclic nonruled hyperbolic paraboloid. real developable. imaginary developable. Surfaces of Revolution. Real (10) Cupshaped nonruled hyperbolic paraboloid. real developable. (8) Elliptic paraboloid. developable imaginary. (22) Convex nonruled hyperboloid of revolution.OH. imaginary developable. C. Real quartic (9) Tubular nonruled hyperbolic paraboloid. Curve is two (17) Horocyclic ruled hyperbolic paraboloid. developable real. imaginary developable. The curve is two mutually tangent conies. imaginary developable. quartic. with acnode. (12) Gathered ruled hyperbolic paraboloid. conjugate imaginary planes meeting in an ultrainfinite line. developable imaginary. while in the first five cases all four are real. Curve is two imaginary conies in (20) Oblate spheroid. developable. quartic Real acnodal (11) Open ruled hyperbolic paraboloid. The developable is a real cone. with acnode. Real cuspidal (14) Cuspidal ruled hyperbolic paraboloid. curve. imaginary developable.
(33) Equidistant surface. Canal Surfaces. conies. and an imaginary one in a real plane. E. Curve two real conies whose planes meet in an actual line. (4) (5) Oblate spheroid. Spheres. Developable same two lines and imaginary cone. Ruled ellipsoid of revolution. . In elliptic or spherical space the smaller. (1) (2) (3) be much We have Nonruled ellipsoid. Curve and developable two con(34) Horocyclic surface. (30) Non ruled whose planes meet cones. (7) * Sphere. hyperbolic canal surface. The curve is a leal (24) Semihyperboloid of revolution.156 QUADRIC SURFACES CH. (27) Cupshaped semihyperbolic paraboloid of revolution. . Curve two real coincident conies. Curve is two coincident imaginary (32) Proper sphere. Curve and developable two generators of each set. loc. developable two real (31) Ruled hyperbolic canal surface. cit. each counted twice. developable two real coincident cones. Prolate spheroid. of course. Real conic and two imaginary lines not meeting on it. and the same two generators. Ruled ellipsoid. (28) Clifford surface. conic. number of real varieties will. (6) Clifford surface. imaginary developable. is an conic in an ultrainfinite plane. The curve is a real conic and two imaginary generators not intersecting on it the developable is the same two lines and a real cone. The developable is an imaginary cone.* Curve is two imaginary conies (29) Elliptic canal surface. (26) Tubular semihyperbolic paraboloid of revolution. whose planes meet in an actual line. ' Called Surfaces of Translation in the author's article Quadric Surfaces '. The absolute curve (25) Elliptic paraboloid of revolution. Two real conies in an actual line. the developable is a real cone and an imaginary one. and two imaginary imaginary generators not intersecting on the conic. developable imaginary. jugate imaginary intersecting generators. D. developable imaginary.
eight planes of parabolic section will pass through an arbitrary line. In an arbitrary plane there will be twelve points. sixteen points in every plane are the intersection of two tangents. meeting the curve twice. eight double tangent planes pass through every The developable will. two secants. the Absolute may be written in the typical form while the equation of the surface involves none but squared terms. Theorem 2. and which are distinguished by the existence of a non. this tetrahedron is chosen as the basis of the coordinate system. An arbitrary point will be the centre of one section. An arbitrary plane will . eight points on an arbitrary line are vertices of circumscribed cones which touch the Absolute. An axial plane of a central quadric bisects a dihedral angle of every two tangent planes to the surface which meet in a line of this axial plane. and its planes the axial planes. two lines characteristics. When Theorem 1. A Theorem I'. Let us next turn our attention to that class of quadrics which we have termed central. possess the dual point. of course. Through an arbitrary point in space will pass twelve planes cutting a central quadric in osculating parabolas. lines in i. pass every point not on the curve. centre of a central quadric is equidistant from the intersections with the surface of every line through this centre. The vertices of this tetrahedron shall be called the centres of the surface. It has sixteen stationary tangent planes.degenerate tetrahedron (in the protective vsense) selfconjugate with regard both to the surface and the Absolute. vertices of cones cirto a central cumscribed quadric which have stationary contact with the cone of tangents to the Absolute.e. twelve. We obtain a good deal of information about our central quadrics by enumerating the Cayleyan characteristics of their curves of intersection with the Absolute. Its osculating developable is of order eight and class one. and the corresponding The curve is a twisted quartic of deficiency developables. sixteen points in an Theorem 2'.xfir QUADRIC SURFACES 157 It is worth mentioning that the Clifford surface of elliptic space has real linear generators. thirtyeight every plane lie in two osculating planes. Sixteen planes cut the surface in horocycles. while that in hyperbolic space has not.
p. It may be shown that not No two zero. 164.158 QUADRIC SURFACES CH.* / )*< = 0. and that only two of those will cut the surface in proper circles. eight planes of circular section pass be a plane of symmetry for one circumscribed cone. In like manner there will be four focal conies in the axial The equation of the focal cone whose vertex is O h planes. Sixteen through an arbitrary point. will be a fc 2(c. and let the equation of the surface and 9 We * See the author's ' Quadric Surfaces '. more than six real planes of circular section will pass through an actual point.. cit. Sixteen planes through an arbitrary point are perpendicular to the axes of revolution of circumscribed cones of revolution. and which pass through the Absolute curve. of the <. (2) The focal conic in the corresponding axial plane will be two Let the reader show that each of these conies passes through foci of each other one.* Let us write as the equation of a typical quadric The planes of circular section are those which touch the cones whose vertices are the centres of the quadric. points are vertices of circumscribed cones which have fourplane contact with the Absolute. (o>). arbitrary plane are the centres of circular sections. . and none shall equal The cones whose vertices are the centres and which pass through the Absolute curves shall be called the focal cone*. next seek the locus of points whence three mutually tangent planes may be drawn to the surface. loc. Let these be the planes (v) (w).'s shall be equal.
we see that we may pass from any line through set of three concurrent conjugate diameters to any other such set through that same centre by changing two diameters at a time.. where (x) is the point of concurrence of the planes (v). Returning to actual coefficients. and keeping the third one fixed. Vy = 0. hence interchanging y and a. We may thus continually apply Theorem 14. and its focal conies through a line form five sets of dihedral angles with the same bisectors. u>a = ^ va j w u = wa Wy (Dy o> a = a> nva = ' 0. The tangent a central quadric and focal planes to a central quadric cones in five pairs of points with the same centres of gravity. (4) This quadric is also the locus of points whence triads of tangents to the Absolute are conjugate with regard to the given quadric. vanish.3 We 2 // ch (Wl + Wm + (t m<'k) V= 0. the coefficients of x$Xj will . Theorem 3. is The proof If of these two theorems immediate.. a a centre.or a . In the same way we may pass from any set of three mutually perpendicular diameters to . of Chapter XII. (itf). the former of these loci will intersect along a curve where generators of different sets intersect at right angles. (co). a^ which are zero. shall find eventually 0.. A line will meet its Theorem 3'. for they involve y^y. ua . we get the locus of points whence triads of mutually perpendicular tangents may be drawn to the quadric (1) (5) it If the quadric be ruled.xiii QUADRIC SURFACES uy 159 the Absolute in plane coordinates be. we mean by a diameter of a quadric. in the ClebschAronhold notation * ^ = 0.
perpendicular lines in that axial plane is constant. The sum of the of the cotangents of squares the angles which an axial Theorem 4.160 QUADRIC SURFACES set. say the intersections of the axial planes through Ojr We thus get ci ) K K 1C set of quadrics having the same absolute focal curve. 5. Three confocal thetic quadrics will touch an in three plane arbitrary mutually orthogonal points. 6'.of a central quadric makes with three tangent planes through three conjugate lines in that axial plane is constant. the same focal cones. 5'. focal planes quadrics through an arbitrary line. set hence. quadrics will pass through an arbitrary point. any other such chapter. line will Theorem 6. The sum of the squares of the tangents of the /jth parts of the distances from a centre of a central quadric to three intersections of the surface with three conjugate diameters through that centre is constant. The sum of the of the cotangents of squares the /jth parts of the distances from a centre of a central quadric to three intersections with the surface of three Theorem Theorem a central quadric with three tangent planes through three mutually plane of makes mutually perpendicular lines through that centre is constant. The sum of the squares of the tangents of the angles which an axial plane . we have but to choose a particular set of diameters. Three homo Theorem 7'. An arbitrary meet a set of con Theorem to quadrics in pairs of points with the same centres of gravity. and. CH. To find the values of the constants referred to in Theorems 4 and 5. and intersect orthogonally. The tangent a set of confocal Theorem 7. shall be called homothetic. I A A inscribed in the same absolute developable. and possessing. in consequence the same focal conies shall be called confocal. form dihedral angles with the same bisectors. . and apply Theorem 15 of the same Theorem 4'.
we got we The analogy is striking. . divide the terms into partial fractions and reconibine. confocal with that given by 0. are the same for euclidean as for noneuclidean space.* if xr 2 we replace c^ by  > ci 2 If the roots . We thus get 1 frfr cfo* ~ If V give to c h each of its four values.. m\ (11) wish to express this in terms of our elliptic coordinates. as will be shown in Chapter XV.A 2)fa^) (C*Cfc) (C A C) For the differential of distance * We It will be 2 = (xx)(dxdx)(xdx)* ^ )2  (c A ) we have = (flYrlY\ (dXdZ). =  W C m ' be A x A 2 A 3 we have . this is a priori evident if we have in mind that our coordinate system is a triply orthogonal one. indeed. to the corresponding formula in euclidean space * The residue of the present chapter is closely analogous to the treatment of the corresponding euclidean problem given by Klein in his * Einleitung in COOLIDQE L . and the general formulae for orthogonal curves. Yh " V/^( A i)(^.3 (1) will be. we (8) These coordinates (X) are inapplicable to points of the Absolute we imagine that all such points are excluded from consideration. and.xin QUADRIC SURFACES 161 Let us now set up our system of elliptic coordinates as did in the plane l. The general equation for the system of quadrics . found that the coefficients of cZA d\ q will vanish.
1893. = confocal cones form a oneparameter family all 2 0. Any two of thes< cones will intersect orthogonally. they will have four cornmoi tangent planes which touch the Absolute. and is somewhat prolix in its attempts die hdhere Geometric ' . . 3878. will be that the coefficients of the expressions d\ p dK^ for the axial planes of the cones will be given by tan We gents to A p = 0.. usual textbooks on analytic geometry. wil themselves be confocal.3 y 2 0. These facts. and Staude. The the same tangent planes to the cone ds touching lithographed notes. 1882. xx. This shows that the con gruence of lines tangent to two confocal quadrics will b< a normal one. pp.. A g = 0. Ar = 0. Fadenconstruktion des Ellipsoids/ Mathematiscke Anncden. as umbilical points. The difficulties which aris< tangents for special positions. GOttingen. will be proved for the non euclidean one in Chapter XVI. anc which are circumscribed to a set of confocal quadrics. i.e. t The QO l 1 at simplicity. Staude returns to the subject in his Die Fokaleigenschaften der Fldchefi This book is intended as a supplement to tlu zweiter Qrdnung Leipzig. the edges of regression of their developabl< surfaces being geodesies of the quadrics.3 *S? ^i CA jrg p "3 *S? Putting X = Y+dY we get the differential form _ notice Let us change this also to the elliptic form. wel known in the euclidean case. vol. 1896.162 QUADRIC SURFACES CH The cones whose vertices are all at an arbitrary point. Notice that we get th< 3 system of geodesies of a quadric by means of its oo commor with confocal quadrics. need not conceri us here. The equation of the cone whose vertex is (Y) and whicl circumscribes the quadric (1) will be 2* 'A i 0.
It is clearly a poly nomial in powers of X. m We finally get for our cone  For progress along an arc of a geodesic of have rfA.xni QUADRIC SURFACES may 16S be thrown into equation of one cone of the family the form where L is a function of A. we must quadrics = 1.*)II(c V / 22Z2L =0. = L2 . Again. Hence the general form will be 2 It remains to find the value of Lp ~p. as two of these confocal contain every line through the vertex. depends merely upon elliptic integrals. have hence A A == A ^p Aq ^r* where Ap is a constant. so that the problem of finding the geodesies of a quadric If we take A f A. We thus get for then only shall we have d\ p = A^. Lastly.V^r_ (\. q. ^fi^VVX)".. r. Ar = const.__. we / . A. which vanishes only when A * 0. our expression is symmetrical in p.
whereof much more later. then..3 7% (\7~ A J( x <7. the surface.164 QTJADRIC SURFACES double tangents to CH. then any of these lines of curvature will cut congruent distances linear of these on all two Theorem 8'. where also will be found a bibliography of the subject. Theorem 8 may also be easily proved by showing that the generators of a set of confocal quadrics form an isotropic congruence.x r) 0.. rectilinear (iX^ (tX n 0. If from a set of confocal central quadrics a oneparameter set of linear generators be so chosen that all intersect the same oo 1 lines of curvature of oo 1 confocal quadrics of the system. If from a set of homothotic central quadrics a oneparameter set of linear generators be so chosen that all touch oo 1 developables of circumscribed to pairs quadrics of the homothetic then the tangent system. is \>\iHA p.e. planes to any two of these developables will determine congruent dihedral angles whose edges are the given linear generators. we have generators.3 The general differential of arc on a surface X r . = const. i. generators. for a distance along a generator k This expression is independent of X r> whence Theorem 8.X r) 7V o we have. .* * The general theorem concerning isotropic congruences upon wjiich this depends will be proved in Chapter XVI.
\'. X 1 1 1 VA A r .3 0...XIII QUADRIC SURFACES common 165 We now upon a seek for the expression for the element of distance tangent to two confocal quadrics \..X A ^. 0.3 (14) A/ rri V *i (c< L. ? A.
two possible cases can arise in the actual domain. In hyperbolic space. (x'A ). assuming. whose tangent touches A' we have = 7r 2 For a line of curvature common to X = A 1? A'= A2 (17) = 77 2 It is now necessary to look more closely into the signs of the radicals in (15). and summing for ds . (15) For a geodesic on \ = Xj. of course. that we are dealing with the case of central quadrics. three confocal quadrics will pass through each point. If the developable be . We know that. and the other two not ruled .166 ds QUADRIC SURFACES CH. 3 1 ). In elliptic space one of these will be ruled. 2. at least in a restricted domain. 2 p= Multiplying through by (AA 1.
If we take a point in this axial plane. while Aj The elliptic case will follow immediately gives the ellipsoid. Once more. or when passing along ?: a tangent to one of these surfaces. the radical associated with it in (15) changes sign also. will replace a nonruled hyperboloid for each point without the hyperbola. 1C5. Similar considerations will hold in the elliptic case. but envelopes. and a notruled hyperboloid. be surrounded by the ruled ones. The plane counted doubly. as we pass through a point of the axial plane with an imaginary focal conic (which we shall call 7r3 ). ruled. If a point be taken in the remaining axial plane. In (15) let us assume that A refers to an ellipsoid. which is without this hyperbola. will replace a nonruled hyperboloid for each of its points. focal conic will be a hyperbola. constitutes the transition between the ellipsoid and the ruled hyperboloid. or points of ?r 2 within the focal hyperbola. a ruled. and one nonruled hyperboloid will pass each point. taking A 3 as the parameter of the nonruled hyperboloid. ruled. d\ i will change sign as a point passes through an axial plane that counts doubly in the A family. let us look at the signs of the terms in (15). and A' In two of the three actual axial to a ruled hyperboloid. The radical associated with d\ will change sign. The other real will shrink into the plane counted doubly. The radical with d\% will change sign for points of TT the plane of the focal ellipse without this The radical curve. looked upon as a point locus.xiii QUADRIC SURFACES two 167 real. without the focal ellipse. counted doubly. looked upon as an envelope. . the ruled hyperboloid. and surround all ruled hyperboloids. A 2 as that of the ruled one. if we suppress the word hyperboloid substituting ellipsoid. and a confocal. It will be surrounded by all ellipsoids. The loop is supposed * See the Author's l Quadric Surfaces'. d\i changes sign. the point of contact is On the other hand we see from (14) that when traversed. looked upon as It will surround the nonruled hyperboloids. onesheeted hyperboloid A'. We next suppose that a loop of inextensible thread is slung about an ellipsoid A. planes we shall have real focal conies. and vice versa. the ellipsoid and nonruled hyperboloid will subsist. with d!A. this plane. and for a point of the axial plane 7r 2 of the focal hyperbola. There will be a real focal ellipse which.* confine ourselves to this case. will change sign for points of TT I within the focal ?t I ellipse. p. and will serve as a transition between the two sorts of hyperboloids. If it be imaginary we shall have an through Let us ellipsoid. and pulled taut at a point P.
and touching a confocal one. while A 3 refers to a nonruled hyperboloid.168 QUADRIC SURFACES CH. moreover. cZA 3 changes sign with the radical. and this will be true throughout the continuation of that geodesic. so that it winds partly on each of the portions of the hyperboloid. The same form of distance element will hold for the rectilinear parts of the loop. : hyperboloidal. . portions and these two are separated by an intersection with TT V We have then pX 2 F d\ 2 p\' 2 JA 2 =\ JA F d\ 2 pr. and. hyperboloidal. to surround the ellipsoid. ellipsoidal. twice when the loop passes 7r 2 the plane of the focal hyperbola. which. of a confocal system will touch any line hence A and A' are the only two which will touch the rectilinear parts of the loop. straight. 2 \ F d\ 2 pA' 2 PC! + \ 2 JA' Jc. can never vanish. as we have We thus have seen. For when we pass from the ellipsoid to the hyperboloid we pass along a geodesic whose tangent touches both surfaces. Jet us limit ourselves to those regions of the plane where the various portions of the loop may be named in order straight. length throughout the whole string will be that given by (15). The constant length of the thread may be written = We p Ai F^^ p*2 + JA 2 7 JA. integrate right up to these limits. that two. for a geodesic is traced by a line rolling on a quadric. It will become infinite four times. The form for the element of are separated by the ellipsoid. and only two surfaces. F% will become infinite twice when the loop passes the plane It will vanish throughout those two of the focal ellipse ir lt of the loop that lie on the ruled hyperboloid A 2 = A'. and twice when it passes 7r3 We may. in a restricted domain. F pA 3 JA 3 ^3= pr 3 JA 3 =4 We may *\d\* = const. approach the second integral in the same spirit. however. f\d\s\ F2 d\ 2 + JV pA \ Jc. Lastly. v = 4 r F d\ = 2 2 Jc. We . . const. ellipsoidal. see. for A and X are the parasee that 3 meters of an ellipsoid and ruled hyperboloid respectively.
we have in the limiting case : Theorem 9. If an ellipse and hyperbola in mutually perpendicular planes pass each through two foci of the other. in conclusion. P P and hyperbola. Lastly. > cZx = 1 1 f JA. J^dXt f JA V^ = first af A J^A! = and the We sum have therefore. but will vanish along those two portions We have of the loop which lie on the ellipsoid A = A r therefore A : X f JA.xni QUADRIC SURFACES 169 We must. since the two integrals are constant. and if a loop of inextensible thread be slung around the ellipse in such a way that it meets and pulled taut at a point will be an then the locus of the two curves alternately. let the ellipsoid and hyperboloid shrink down to the focal ellipse and focal hyperbola respectively. and the locus of the moving point is an ellipsoid. consider the first integral. It will never become infinite.. = const. ellipsoid confocal with the given ellipse .
B. might carry We We We through the same sort of work for any three points. vol. in the hyperbolic case. We shall reserve the name area for the second of these.* A first notable difference arises from the fact that. We begin by rewriting IV. in the hyperbolic case at least. i. The first is present in the analoga of those formulae which give the area in terms of the sides and angles the . but. as a definite integral. and. shall define triangle as in Chapter II. as we saw in the closing pages of Chapter VII.^ Let us. C to indicate.an incorrect formula for the volume of a tetrahedron. hold for all the others. 1893. 20 ? . 1904. two different functions of a triangle appear to play the role of the euclidean THE subjects urea and volume of area. we should thereby be compelled. as in elementary geometry. delle Science di Torino. author gives. either the vertices of a triangle. p. l> .itnd Volumeriberechnu'ny fur Lobatschefskijsche Rdumo. xxviii. Gbttingen. assume that these points are real. in the noneuclidean cases. established for one region. not only in our analytic expressions. Die Oberjldchen. giving to the first the name amplitude. but even in the trigonometric formulae first introduced in Chapter IV. second appears when the area is defined as the limit of a sum. * seen at once to For a bibliographical account of fcho subjectmatter of the present chapter see the dissertation of Darinmoyer. 9 . to introduce certain very delicate considerations as to algebraic sign. e. use the letters A. or the measures of its angles.CHAPTER XIV AREAS AND VOLUMES offer some of the most between euclidean and nonstriking points disparity euclidean geometry. <' sin j sin 1C r cos 1C A= A cos T cos y K 1C be a cos 7 iC is This formula. situated in the actual domain. Unfortunately the .
the if left side is essentially (. sin 7 in k a AA' sm ~ . k sm . . side is symmetrical in the three letters a. r k .. . written sin (ABC). fi/ = . B h G sin (ABO) k sin If sm 7 a sin T . 171 k = L sin 2 7 sin 2 7 k k cos 2 T cos 2 7 k k cos 2 7 + + 2 cos T cos T cos r k k k cos 2 7 r k\ 1 cos 2 r cos 2 7 2 cos r cos . C be Iii the real domain. we have . r. . k (4) . B'. and shall bo called the Sine Amplitude of the triangle.. (7 rC /i/ A/ 1 COS 7 COS 7 a COS 7 1 COS COS 7 COS 7 1 & K the measures of sides and angles be negative in the hyperbolic case. AREAS AND VOLUMES k A sm A . XIV b c sin T sin 7 . . taken positively. .'). c 6 sin 7 sin 7 sin ^1 . MJfo) (1) M sin  xy (2) We may rewrite in the form sin A ~ sin . sin 7 sin sin B = sm a sin 7 sm 7 . c A/ . . and positive in the elliptic. a A/ 7 . k n A ~ sm o sm CC' = sm (ABC). cos j The righthand so that . C' be the points where the sides of the triangle meet the perpendiculars from the vertices. (y). B.   (3) k sm T sm k 4'. ft 7 k Bff sm . .CH. Let the reader show that if the coordinates of A. . (z) respectively (xx)(xy)(xz) * (ABC) = . It will vanish only when the three points are collinear (under the restrictions made at the outset of this chapter). so that the radical on the right must be chosen accordingly. 6. write . we may .
A = This bolic . sin (ale). c . r cos 7 A/ ' = /{/ A/ 6 r sin T sin 7 /c fc s sin  J. pure imaginary in the hyper a L 6 . sin r T k sin 7 k sin A sin B sin G (abc) sin sin J. . 6 sm sin sin A _ k Y  sin (A ftd\ sin (^^g) sn sin6 sin (a6c) (*) sin ^ sin sin (7 ' sin^sin/C A /u If cos v = a COS^. > in the elliptic case. sin B sin r.= A/ sin C G GO' sin rA/ sin (ale). B (7 (t>j sin A sin r . b . I e cos . k 2 sin (ABC) = 2 Area A ABC. Let the reader show that Lim. sin sm . B sin C sin 7 = if IV C sin A = sin A sin J5 sin T 1 ft/ /f cos 1 (7 cos cos 1 cos cos B A (5) 5 cos .172 AREAS AND VOLUMES CH.= A/ . p = 0. We see at once the close analogy of the sine amplitude of a noneuclidean triangle to double area of a euclidean triangle. / i \ /f. 2s. (7) a k sin T sm r yfc . = sm r sin r . may be A function correlative to the sine amplitude obtained from the correlative formula sin sin B i^in G cos r = k sin cos sin T if A/ B cos (7+ cos A.
s .XIV AREAS AND VOLUMES sm sm cos . s s  6 v . f I COS (vB) COS ((rc)"! 7 I L a cos <r cos((t a) J sin (abc) = 2 >/ cos cos (o A) cos (a B) cos : (<r (7). ^ n (. 8' smj k cos = sin (a&c) 4 sin ^ J. as we have already seen. .4. . sin J? sin sin (12) o = 4 cos A t  ^ . 9 ^ \J let / am y sm k I . 173 s . sm a k 7 . 8 A= sm k . (11) asinl sin ^ 4 sin  J? sin ^ (7 = _^ . sin k sin si SC j (10) us put . a = = _  r L cos : o cos . c cos ^ r j cos f T (13) It should be noticed that the denominator on the right of equation (13) is essentially positive. . b . ^ T V) i sin e T sin a r sm r sin . 7 sin 7 k .00 sin i A pr\ ) In like manner. The numerator is negative in the hyperbolic caso.6 . (av  ^ )l sm ^ sin (7 J COS k = i jy a y A. . A. sin . = . s~a ~ir tc 6 c sin T sin T k s sm ctn f .
the absolute value of the difference between the sum of the measures of the angles and TT. .(ABC) = 4 lim. (cos f cos ^ T cos T . (ABC) Theorem 1. In the or elliptic case the numerator is positive but > ^. Let us next examine the infinitesimal quadrilateral. In an infinitesimal triangle the limit of the ratio of the excess to the product of the euclidean area and the measure of curvature of space is unity. In Chapter III we defined as the discrepancy of a triangle. we have a / . points remaining vertices are K .C. sinjr& . Bin.B. D.15 .CD /t __ sin A. cos ^ T: cos  t c (14) ^ ' n . AB and CD shall intersect in // (actual the latter two or ideal) while AC and BD intersect in at a finite distance from A. J5.  r c\ ) Bin lim.174 AREAS AND VOLUMES <r CH. Let us now define as the excess of our triangle the expression > the measure of curvature This will have the same sign as p of space. BUI . . 2 T Passing to the limiting case where the triangle becomes infinitesimal. sin A ' sinl7 5fc sin .. sin . whose A. (7. =4 e = 4 lim.and & cos a o > 0.  We have e sin2 = cos o = sin  (ABC) x i^ 4 cos ^ a r: b . cos A < 0.6 ^ Lim..4 sin . . but here also < . . D.
which is limited by one or more closed curves. 83102. This follows immediately from the definition given above. and need not detain us here.xiv ABEAS AND VOLUMES r\ . e. the limit of the ratio of the areas of the triangles into which it is divided by a diagonal is unity. ^. sin(CM. When the surface integral of the function 1 exists over a region of the plane. and the euclidean expression for its area.g.^. common halfamplitude. /C . Traite cC Analyse. IS . AC  k   K AB. Theorem 3. = We the 1. As an application of these principles let us determine the * Conf. The proof of the existence of such a limit. Paris. .swA DC. included angle. The area of a region of a plane is the sum of the areas of any two regions into which it may be divided provided that these two have no common area. vol. . .sm y sm Z> . i. Theorem 2. pp. tn v = lira. shall define as the area of an infinitesimal triangle its value of k 2 times its excess. and let this be covered by a network of inLet the finitesimal quadrilaterals of the sort just described.* point. A sm ~T~ sm T~ sm A .. 7 . Suppose now that we have a region of the plane. DBDC k ^~. The sum of these two infinitesimal areas shall be called it will be the area of the infinitesimal quadrilateral equal (always neglecting infinitesimals of higher order) to the product of two adjacent sides multiplied into the sine of the . Picard. first ed. sin D . connex right up to the boundary.AC..JB) ' lim. 1891. (DAB) r n sm r. Definition. that integral shall be defined as the area of the region. If the opposite sides of an infinitesimal quadrilateral do not intersect in points infinitesimally near the vertices. area of each of these be multiplied by the value for a point therein of a continuous function of the coordinates of the The limit of this sum as the individual areas tend uniformly toward zero shall be called the surface integral of the given function for the given area. A 175 . and its independence of network employed will be identical with that used in the corresponding euclidean case.
Now it is perfectly clear that if a triangle be divided in two by a segment whose extremities are a vertex and a point of the opposite side. j<j> rf> = M k tan j sin </>. ii. The element of area will be (15) k CR / r sin . In Friachauf.) COS r R= ~ k cos r r Remembering that the * It is surprising to see this. (6). Leipzig. will be found a geometrical proof applicable to the hyperbolic case but not. Ekmente der absoluten Geometric. 1876. Jo k tan dr = /c 2 / ( 1 cos R\ k/ ) . the excess of the original triangle is the sum of the excesses of the parts. 1891. 49. and the same remark will apply to the book of Liebmann. It will be sufficient to do so in the case of a right triangle. IV. Let us give a second demonstration of this fundamental theorem with the aid of integration. hence * Theorem 4. In ClebschLendemann. p. We may introduce polar coordinates M /* k tan cos </>. so far as I can see. and we shall take a right triangle with one angle at C the intersection of x l = 0. the right angle being at B a point of the axis x2 = 0. to the elliptic.176 AREAS AND VOLUMES CH. limits for y cos c + </> tan 2 ^ k are and G how unsatisfactory are the proofs usually given for the bestknown theorem of noneuclidean geometry. apparently. cit. but the analysis is unnecessarily complicated owing to the fact that.. x% = 0. is . (Ch. excess divided The area of a triangle is the quotient of the by the measure of curvature of space. and we may establish our network by a repetition of this process or division. or by (1) the limit 2 of the sum of k times their excesses. vol. and r = const. the author overlooked the consideration that it is sufficient to prove the theorem for a right triangle. Vorlesungen iiber Geometrie. area of a triangle* It is the limit of the sum of the areas of a network of infinitesimal triangles. the elements of arc along = const. makes an attempt at a general proof. Manning. loc. \ j = tan ~ 2 </> sec rf> </>. Leipzig. but the use of intuition is scarcely disguised. cit. a proof by integration. will be dr and k sin T rC 7 d <f> respectively.
Bologna. p.xiv A Area AREAS AND VOLUMES 177 = 72 f k2 I /o c '^ d$ i /cos 2 \r / </> The first integral is k 2 C. due For an admirable discussion of the question of to Hilbert.* Definition. (16) Two have very regions with the same area may. I believe. cit. the other hand by Chapter IV. . when they are composed of the same number of nonoverlapping subregions have in common a region (i. 1900. our second integral becomes Area = i 2 (J. are.r 55i cos r T \ const J Hence our second integral will be . p. loc. in the fifth article in Enriques. e. where by the . cit. we put sin <j>=x.A8  sin. different There however. The COOLIDGE M . 109. ffi + Cw).1 [sin cos ^1 On __ HC k 1 c . <fa i $cT~^ = sm . by completion. This vanishes at the lower limit. shapes. three simple cases ... First. line other than one (4&4fc +1 ). where the equivalence of area is immediately evident. If. further. where the two figures are congruent second. Questioni riguardanti la geometria elementare. adjunction of pairs of mutually congruent nonoverlapping subregions to them. naturally. subregions no two of which which has an area) congruent in pairs third. Given n successive coplanar segments (^A. (4i^i) so situated that no through a point i can contain points of more than two of the the assemblage of all points of all segments whose . loc. they may be transformed into congruent In this latter case they may be said to be equivalent regions. 40. (7) cos A sin C cos . area see Amaldi.). A segments * The term distinction between equivalent and equivalent by completion is. equivalent by completion is borrowed from Halstd.
178 AREAS AND VOLUMES CH. is certainly a polygon. Let the vertices of a tetrahedron. The area of the polygon will thus be the sum of the areas of these triangles. extremities are points of the given segments shall be called a The definition or. (t) respectively. no two of which have in common any area. g. triangle 1 sides exist. tudes. On the other hand. as defined in Chapter II r be A B. (7. following. The opposite faces shall be a. The area of a convex polygon is the quotient of the excess of the sum of its angles over (71 2)71 divided by the measure of curvature of Space. with ampli an analytical rather than a trigonometric method. more simply. Theorem 5. s (w). (y). r(coZ) = (Xxyz). e. if a polygon of ?i~l sides may be divided up in the manner suggested. as before. a polygon. D with the coordinates (&). Let the reader show that the area of a proper circle is (17) The total areas of the elliptic and the spherical planes . (z). y. may convince ourselves of the compatibility of these statements as follows. will be respectively 2 77 & 2 In the hyperbolic plane regions desired area. (v). vertices. 6 with coordinates (u). /3. say be connected with all the others. and if a polygon of n we may easily enlarge it to have n sides by taking an additional vertex near one side. of sides. it is immediately evident that one of n sides may be so convex polygon A We A divided also. shall be to see how far the methods which we have established We for studying areas are applicable shall begin. of the tetrahedron We shall define as sine amplitude sin (A BCD) = COS j AA K COS SB 7 GO COS 5. the polygon will be 19 divided into 71 2 triangles. however. and angles is immediate. If one vertex.  (axe) (yy) (zz) (tt) .COS DD  i K K \ K . may be found haviug any Our next undertaking in three dimensions. (co). so that.
we get sin j~ sin j~ (Moment AB. C'. CD) (ABC) f = sin (ABCD). (21) The geometry of lines through a point is an example of the M 2 . D be the points where perpendiculars from the vertices of a tetrahedron meet the opposite faces.O = sin (ABCD). . B'. Then __ sin A A~r ry Ti/ (BCD) sin r~ = sin (CDA) sin y = sin (DBA) sin (20) = If sin (ABC) sin ^. two we mean by %a/3 the dihedral angle of these ^ a cos 4_ a/3 = "" ^ faces xz) (xt) (zy) (zz) (zt) (ty) (tz) (tt) V I *\(xx)(yy)(zz)(tt)\ *(x) sin v I ^\(xx}(yy)(zz)(tt}\ (ABCD) sin ~ A ft sin sin (BCD) sin (4C7D) O.xiv AREAS AND VOLUMES 179 _ V(zz) We shall give to the radicals involved such signs that k sine amplitude shall have the sign of k 2 Recalling the concept of the moment of two lines introduced in Chapter IX. (19) sin = V(xx) V(yy) v(zz) Let A'..= sin (ABCD).
geometry of the elliptic plane. (25) sin^_a6sin^_/38sin sin y6) = sin (a sin (26) mn(BCD) sin(/3y5) _ "~ s _ "" sin(y8a) sin (8/3a j sin (a/3y) sin (a/3y5)" ^ ^ Our two tetrahedral functions are connected by the relations sin (5CD) sin (CDA) sin (554) sin (45) sin 3 (a/3y5) sin (/3y8) sin (y6a) sin (6j8a) sin (a/3y) . (20). Sill sin (a/3y8). relative to sin There will be a function corshall call sin (ABCD) which we sn sn (Moment 45. 4C AD) = Y . (23) sin (a/3y) sin sin (a/3y5). 2 = 1. and analogy between the sine amplitude and six times the euclidean volume. CD) = sin (aj3y5). = ^ 56. 4D) = sin sin sin sin (45. (22) The reader (22) the striking will not fail to notice in formulae (19). sin (45CD).180 AREAS AND VOLUMES CH. smsin (ay6) sin . where k of the sine amplitude of a trihedral angle speak We may thus sin (45. 4C?.
when the figure becomes infinitesimal. BAG \\j = ~Area A ABC. suppose that we have six no three coaxal. besides half the sine amplitude and the euclidean area. sin (ABCD) = Km. sad to relate. * Die eulersche Formel in Zusammenhang mit dem Inhalt in der nichteuklidischen Oeometrie. It is the lack of this function that renders the problem of noneuclidean volumes difficult.xiv AREAS AND VOLUMES 181 The analogy between the sine amplitude and the sextuple of the euclidean expression for the volume appears even more distinctly in the infinitesimal domain. Each time we called sine amplitude corresponding in have had a function many particulars to a simple multiple of the euclidean area or volume. In the plane there appeared. planes. a third expression. on the analogy of a euclidean region parallelepiped. In three dimensions this function is. fc 2 Lim. sin . that we have a three dimensional . of any two is unity. the function of the sum is the sum of the functions. namely. So far the analogy between two and three dimensions has been sufficiently good. Mathematische Annalen f vol. passing by fours through four actual Let the remaining interor ideal. which may be divided into six tctrahedra of such sort that the limit 01 the ratio of the sine amplitudes. or Six times the of the euclidean volumes. . will thus be formed. sin (ABC) = ^ IS AC. in general. ABCD. and ap proaching a multiple of the area or volume as a limit. sections be at a finite distance from the three chosen points. Lim. and that we divide it * It is highly interesting that in four dimensions a function playing the role of the discrepancy appears once more. (28) Following our previous analogy. 1 . 1906. the discrepancy or excess. A/ (ABCD) tetrahedron = ^ Vol. there is no simple function of the lacking measures of a tetrahedron which possesses the property that when one tetrahedron is the logical sum of two others. euclidean volume of any one of these tetrahedra may be defined as the euclidean volume of the region. Ixi. but not collinear points. See Dehn. entirely that is to say.* Suppose. region connex up to the boundary. An infinitesimal but infinitesimally near one another.
one of these angles is obtuse. we may subdivide into three smaller tetrahedra. next. Two regions unity will have the same volume if they be congruent. or if by the adjunction of such pairs they may be completed to be congruent. and if no line. the volume integral of the function integral. then. as all the volumes approach zero uniformly. The limit of the sum of the euclidean volume of each. The perpendicular on the plane of this face. each of which possesses . by a familiar theorem in elementary geometry.182 into a AREAS AND VOLUMES CH. and its independence of the method of subdivision. mutually congruent in pairs. in the plane of this face neither of the face angles whose vertex is not at the foot of the perpendicular is obtuse secondly. The proofs for the existence of that volume integral. for each of which the line of one edge is perpendicular to the plane of one face. a tetrahedron where the line of one edge is indeed perpendicular to the plane of a face. from the opposite vertex. First. is To begin with. shall be called the volume integral for that region of that function. to that of finding the volume of a tetrahedron of particularly simple structure. and thus. (The case where both were obtuse could not occur in a small region. pass through a point within the face. then it is easy to show that the region may be divided up into a number of tetrahedra. in turn. number of extremely tiny tetrahedra. There are two possibilities. can contain more than two points of the surface. Consider. If the limiting surface of a region be made up of a series of plane surfaces. and.) In the first case we might draw a line from the foot of the perpendicular to a point of the opposite edge in this particular face. with the help of this perpendicular. shall be called the volume of the region. made up of the same number of parts. divide the tetrahedron into two others. . which holds equally in the noneuclidean case. and the problem of finding the volume of any such region reduces to the problem of finding the volume This problem may. be reduced of a tetrahedron. are analogous to those already referred to for the surface In particular. perpendicular to the line of that edge. multiplied by the value for a point therein of a continuous function of the coordinates of that point. not lying in a plane of the surface. will. we may assume that there one face which makes with the three others dihedral angles less whose measures are than > for the bisectors of the dihedral angles of the original tetrahedron will always divide it into smaller tetrahedra possessing this property.
where is perpendicular to BCD and DC perpendicular to ABC. (7. and one of the sort considered in case 1. Let a contain a point Bl of (AB) whose plane perpendicular to distance from A shall have the measure x while this plane AB AB . 64t>. when the science of noneuciidean geometry had not reached * its hampered by a ' It is very general. our tetrahedron of the simplest type is a special case of an Artiothoscheme. Ch. vol. and and wonderful terminology. We wish triangle D P l to find r r T sin T cos k Let are r dr d<j>. however. merely to consider the volume of a tetrahedron of the simplest type.. limits of integration for T to find v B^ meet (C D 1 1) in EY The and BE l l .. g.) This integral takes a striking form. mulof the cosine tiplied by the surface integral over the A B l Cl l of the fc th part of the distance of a point from B lt (Cf. while $ is the measure of 2jL Cl B P. It is. it was originally written in 1855. will be dx. Tfieoric der vietfachen Kontinuitat. IV. The tetrahedron will be divided up into a tetrahedron of the simplest type. 1855. These we shall for the moment all simplest type. draw a line perpendicular to the line of the opposite edge in this particular face (and passing through a point within this edge).* Let the distance from Bl to a variable point of the be r. In the second case. We have. Lionville's Journal. xxii. B^Cly hence Now C D 1 1 is perpendicular to * The integration which follows is a very special case of a much more general one for n dimensions given by Schlafli. e. 1901. hence 2  wo have merely k r Jo . meets (AC) and (AD) in Cl and D l respectively. a striking piece of geometrical work. . p. Zurich. and connect the intersection with the vertex opposite this face. and an adjacent one of the same type and the three faces through J. 7). The volume of the region bounded by this plane. (2). extremely difficult reading. then.xiv AREAS AND VOLUMES 183 the property that the lines of two opposite edges are perpendicular to two of the faces. fearful 1'aire d'un triangle spherique comme cas particuliers '. This paper of Schlaiii's is posthumous . J5. Schlafli gives a shorter account of his work in his Reduction d'une integrate multiple qui comprend Tare d'un cercle et present recognition. from the vertex of the obtuse angle mentioned. Let the vertices of the tetrahedron be J.
r = cn = Z. . JJjOj Let the reader note the astonishing feature of this result.) (Ch. 381. xxxiv.E.62 ctn 2 ____ d ] 0.3 (  cn = Vol.1 [a v'l . this will also be the measure of %AC^^ which is^the plane angle of the dihedral one. and another trigonometrically. * apparently independently.184 . namely. AKEAS AND VOLUMES k . ^ sin ic ~ sm B~C\ . 0. where it is stated that this integral cannot be evaluated by integration by parts. vol.j_i. (5).C\ 1 cos 7 k k . (29) We can easily express this integral in terms of tan i = A/ sin i tan ^ DAG = a sin b ctn ^. thus get for our volume the strange formula * VoL = * fc^da. k kjo. . B& = tan B G 1 ^ l \ l k sec (6). ^ A/ ___ A/ ^i/ fJ/i VX A/ Our required /I/ integral P I is then 7 TTT xy / . A^ T> rt . by Richmond. p. . tan. j tan. sm * sm ' A.G. B. sm = We . IV. Let the measure of the dihedral angle whose edge is (C^Z^) be 0. 175. Reduction. that it involves one side of a triangle directly. This same integral was discovered. p. (30) * See Schlilfli. p rt JJi\Ji /. ' . = cos E. 1902. cos = cos AB  l si sin OdQ 1 = r sin /c . The Volume of a Tetrahedron in Elliptic Space/ Quarterly Journal of Mathematics. CH.
Schlafli. P^K rh pan = k. A/ tan 2 Vol. = 7T& 2 sin ^ cos 6 1 4. while Let Q be the intersection of (AO) with a perpendicular cone. while the base circle meets the plane AOP in B. If a right triangle be rotated completely about one of the sides adjacent to the right angle. tan 2 Put * tan = x. p. from P. in the general case. . Vielfache Kontinuitcit. op (T A. (Oh. Let the vertex of the cone be A is a point within the and the centre of the base 0. (AB) shall meet PQ in R. Let us also write P AB = . and not suitable to reproduce here. = /v Jo Jo Jo R ) am^coa^f& & nv~ mn^eo&^'f* OP OP tan AO = ~ 1C r tan T cos K cos sec 0. (6).* it does not seem possible to effect the quadrature in terms of elementary functions. the figure so generated shall be called a cone of revolution. .XIV AREAS AND VOLUMES 185 This formula apparently represents about as close an approach as can be made towards finding the volume of this tetrahedron.) 2 dAQ = 1 k 7 dr + cos 2 tan 2 \ sin  = sin r sin 0. 40 AB = 0. The proof is highly intricate.cos 2 2 dr. s. for. 95 gives a formula for tho special case whore the sum of the squares of the cosines of the dihedral angles is equal to unity. The volume within the surface may be found as follows. V. ZK = r.
tedious demonstration was subsequently worked out by Ktfrperinhalt des senkrechteii Cylinders und. where the distance from the centre to every point of the surface has the constant value R.. = 7r& cos 3 sin 2  2\7T^~2 ^m = 7T/0 3 COS tan" 1 te cos r"** Jo * COS0 [1 0) tan" 1 # ~l Jo (31) ]* To find the volume within a proper sphere. p. . absoluten Geometrie. = 2 Jc \ sin 2 Jo Jo Jo /lf f sin dr d f sin 2 Jo (82) Let the reader show that the total volumes of elliptic of spherical space. * This formula * loc. 99. where k 1 will be. respectively. Der given without sufficiently detailed proof by Frischauf.* Grunerts Archioen. vol.Kegels in rler is lix.186 AKEAS AND VOLUMES tan* CH. xiv Vol. cit. A Von Frank. PR pTT P27T O Vol. 1876. and = 2 7T . 27T 2 .
as to be scarcely suitable to serve as an introduction to the subject. 1899. d COS i ~=T /I/ _ f " (xx . ... IV.CHAPTER XV INTRODUCTION TO DIFFERENTIAL GEOMETRY task which we shall undertake in the present chapter develop the differential geometry of curves and surfaces in noneuclidean space. x + dx) ds 2 = k^ (xdx) = = (dxdx). and assuming that W. Vorlesungen uber Diffirentialgeometrie. that we are limiting ourselves to a real region. . this proceeding is not points we shall therefore assume. tth'" _ ^_ \XX) \CtXCtX) / / > \ . in exceptional cases.* We shall introduce a notable simplification in our work by abandoning homogeneous coordinates. and VI. It is. l* . where no absolute or ultra infinite points are included in the hyperbolic case. III.^ f dx^ we have n y/ow . where we wish to include of the Absolute or beyond. (3) analytic curve. further. I W/ /* A/ /'*^ /o /* /l/ //** When a /= of distance x. (#'). have for the distance of two points (#). y \ (dxdx). for the square of the differential \XCvX) \ r . unless we specifically legitimate state the contrary. The formulae developed in this chapter will hold * The developments of this chapter follow the general scheme worked out for the euclidoan case in BiarichiLukat. however. such a curve that the coordinates of its points are analytic functions of a single variable. In Chapters XXI and XXII of the same work will be found a different development of the noneuclidean case. yet so concise. so general. Chapters I. . ~ ) 9 UJLJUL T. in the hyperbolic. XQ = . Of course . (1) (xx) is to THE = In the elliptic case we shall take XQ ^ . We shall. We shall mean by an Leipzig. XQ j> for all real points.
(2). lim. p. If we compare with Chapter XI. we shall have Theorem 1. 603. and we shall assume from here on that when we speak of curve we mean analytic curve.188 INTRODUCTION TO CH. say P. Let us imagine that at a chosen point of a curve. Then P we shall define * . equally well under the supposition that the functions and their first three partial derivatives exist and are finite in our region. is taken from Bianchi. and rejecting powers above the second (t ) * This definition ascribed to Voss. and define as the osculating circle to a curve at a point. It is there . but the gain in generality is of little interest to the geometer. We shall take two near points P' and P" and on the curve and tangent respectively... so situated near on the same side of the normal plane that PP' = PP".  5 PP" as the curvature of the given curve at that point. The curvature of a curve at any point is equal to that of its osculating circle. of Developing by the binomial theorem. cit. A = r . loc. and is equal to the absolute value of the product of the square root of the curvature of th space and the cotangent of the /c part of the distance of each point of the circle from its centre. the limit of the circle through that and two adjacent points. Let us now suppose that the equations of our curve are written in the form Then for a point on the tangent we shall have coordinates To get the value of X (XX) = xx = k\ (xti) = 0. a tangent is drawn.
so that * ~~ ' ~~ da its As (t) lies on the tangent. which figure so prominently in the euclidean theory. = (xx)=tf. (f\ \tAjlif /^. In hyperbolic space these points lie without the actual domain to which we suppose (x) confined. take as our parameter on the given the length of arc. and () orthogonal to (x) on the binormal. (0) orthogonal to (x) on the principal normal. These three will replace the direction cosines of tangent. principal normal. / If a point trace an infinitesimal arc is the angle of the * ds ^A/ corresponding absolute polar planes / ^\/ We curve shall.xv DIFFERENTIAL GEOMETRY t 189 Subtracting from the series development of x^ we get for our curvature . minus the measure of curvature of the noneuclidean space. (x'x')=l. be orthogonal to (x) three other points allied to it.* 2 ~ r2 \vU \ A/ P (rr"<Y"\ "i T tt \ >*J\fj \('r''r'\ "T" (v<r j \ *AS \fj j 4\h ) 4> / / i& \ n 1 ~j~i yu* A/ If' vU \vU \ w\* \vU (v / t N P 1 (x'xj __ (a'VQ _^ ^ Theorem 2. It will be convenient to consider. (5) ti .\ \JU^J (r\ V^C/ "~~* (t*\ \ ^/ (t\ \ / (r\ Y**/ tfo. coordinates will be of the form (tt) = (xx) = (z) /c 2 . (^/) = 0. and binormal. hereafter. The square of the curvature of a curve is the square of its curvature treated as a curve in a fourdimensional euclidean space. (to) = 0. . s. (t) shall and on the tangent./< have For the point we shall (zx) = (zx'} (zz) = (xx') = (x'x') + (xx") = 0. besides our point (x).
. (10). and it is important now to find the geometric difference between the case where the torsion is negative. = 7 /x . so that (aiVO = 0. . (^0 ) = .. . . analoga have. (11) ai^e the of Frenet's formulae for euclidean curves. but that is well determined from the above formulae. We two successive osculating planes thus get 1 to the . the elliptic case only. will see at once that (9). As before we shall choose s as the independent variable. rff more specifically We have also . so far. and that shall carry through the work for where it is positive. Hence or. 0. /9^ (y) = (off) = (a"0 = = ('O = (&') = 0. overlooked the question of the sign of the torsion. (7) We shall define the torsion of our curve as the limit of the ratio of the angle of differential of arc. (xz) (xz) = . ) (6) To determine we shall have the conditions ^ = P ~\yxx'x"\. the hyperbolic may be treated in the same way. . (0V) = .190 INTRODUCTION TO CH. . but it is wiser there to replace the coordinates Y The reader We We (x) by (x). k Reverting to our formulae da (atf (5) 3T" and k ) ' (8) (6) <**<_** ~ P ~ S.
that of z$ from (6). so that. in general. Again. while that part of the principal normal shall be called positive which lies on the same side of the plane of tangent and binormal as does the curve.XV DIFFERENTIAL GEOMETRY 191 The sign of t { (which may be ideal) will be found from (5). while the sign of T will be given by (10). Putting in the coordinates of a nearby point of the curve. we see by (6) that we may give to a point on the principal normal close to (x) the coordinates Substituting in the equation of the plane we get on the same side as the curve if e > 0. Let us call positive that part of the curve near our point for which As > 0. so that this will lie We Pit = Xi e Xi 4 As x 6 :i AS X* Xj' In like manner for a ray from part of the curve (x) to a point on the positive we get ?/rf = moment two 8 <**"*" of these relative close of Chapter IX. the curve will not cross this plane here. 1C ~" 2 2 As2__ & (As) and this is essentially positive. The positive part of the tangent shall be that which lies on the same side of the normal plane as the positive part of the curve. The equation of the plane of the tangent and binomial wil1 be \Xxt\=(Xx) + k 2 (Xx") = 0. will be The rays. and that of $ from (7). Let us find the Plueckerian coordinates of a ray from x i + x/^s on the positive part of the tangent to # + #/' on the positive get part of the principal normal. as defined at the .
192 INTRODUCTION TO all. we may write ^+ Now y^ is dw tf fc +U Z: . . and a ray from a point on the positive part of the tangent to one on the positive part of the principal normal is negative positive. and this by (7) is by hypothesis. sm T It . We shall next take up the evolutes of a curve. du dv linearly dependent on (x) and (x). The torsion at a general point of a curve is positive when the relative moment of a ray thence to a point on the positive part of the curve. positive otherwise Let (x) negative.. be a point of an evolute. the torsion is negative. A tangent to an analytic curve at a general point will be in the osculating plane at the corresponding point of any evolute. The factors outside of the determinant are positive. . so that the sign depends merely determinant. . Ic Now Hence the relative moment will have the sign of =Theorem 3. upon that of the ' equal to . OH. when the latter product is is when Intuitively stated this means that the torsion the curve resembles a lefthand screw. =  p ^~ Remembering that tts normal of the evolute. Since (x) lies in the normal plane at (x). fl rp = kt^ while ^ is on the principal Theorem 4. Then Xi jJ as = .
x _ C_^L.rrrr=irrrr==.?) ~~ JL JL k k in question will therefore have the coordinates (i sin (o + C) + t cos (o f (7). and so proportional to wo} Xj = u^ du 3?du ~cte ~~ p kT 1 T + tan. cos 1) ( + C) +  ^ sin (a + The coordinates of the point of the line (x) (x) orthogonal to (x) will be _^ . = The point . . < + 0.xv and DIFFERENTIAL GEOMETRY 193 and. &2 .1 i u dp ~ p k t / / To get (w) we have K.r=i  = U. _ . A AC f" ~ ". the assemblage of all terms in () (z) must be a linear combination of (x) and (#). for the same reason.
and in the actual domain of hyperbolic space. If the parameters be (u) and (v). then. e. for it will vanish only when the tangent plane to the surface is also tangent to the restrict ourselves. surface the locus of a point whose coordinates are analytic shall exclude functions of two independent parameters.194 INTRODUCTION TO CH. will Absolute. If the generators of a developable surface be turned through a constant angle about the tangents to one of their orthogonal trajectories. (a. F=(r rr '<)X <^ /# <>X (13) This is a positive definite form in the elliptic case.4. e. and we shall find a no less striking shall mean by an analytic analogy there. This gives us the significance of <r. meet on the involute. The tangents to an evolute of a plane curve make a constant angle with the plane of the curve. namely (<r + C) is the kih part of the distance from this point to (z). is i. we take two evolutes of our curve the angle between their corresponding tangents. Theorem 6. from consideration all singular points of such surfaces. to which we shall The discriminant. It is our next task to take up the theory of surfaces. Corresponding tangents to two evolutes of a curve meet at a constant angle. The equation of the tangent plane at (x) will be = CV 0. .C) represents the angle which this normal makes with the principal normal. we shall have for the squared distance element ds 2 Edu 2 + 2 Fdudv + Gdv 2 We We = . those which Theorem 5. Theorem 7. tion. The foregoing theorems and formulae exhibit sufficiently the close analogy between tho differential theory of curves in euclidean and in noneuclidean space. under this same restricalways be greater than zero. i. If. the resulting surface is developable.
or the cosine of the angle of the two arcs from (x) to (x + dx) and (x + b x). V du "ev 7 ^x [i <y <) 2 x I ~l . Neglecting differentials of higher order than the first. 195 The Absolute pole of this plane (14) consistently use the letter (y) throughout the present chapter to indicate this point. and the point (x + dx). The equation of the plane through the normal.. will be We shall (Xx) (xx) (x~ E F 0. we have TT XX . The condition for perpendicularity between the parameter curves will be The equation of the tangent plane at (x f dx) is = 0.XV DIFFERENTIAL GEOMETRY will be ^ rx.e < < U . will be Edit. The cosine of the angle which this plane makes with that through the normal and the point (x + bx). t6 dv 2 1 J N2 . bu + F(du bv + udv) + Odv bv (15) The two will be mutually perpendicular if = 0..
the tangent plane may be said to rotate about a line .e. J/. the pole of the tangent plane. D". the equations determining differThe parameter curves will entials in conjugate directions.196 INTRODUCTION TO CH. These are the equations for tangents to conjugate systems of curves. (18) The differential equation for selfconjugate. This line will contain the point (x f. D= . The line of intersection with the tangent plane at (x) will be found by equating to zero separately the first and the last four terms. i.dv f D"dv* = 2 0. Under what circumstances will the normals at two adjacent points intersect. when we do progress along such an infinitesimal arc. be mutually conjugate if V= 0.bx) if Ddu bu + D'(du b v + dvbu) f D"dvbv = 0. Z)'. we have (an/) = (ycfa) = (xdy) = 0. or. D" = The (17) signs of D. when will their minimum distance be an infinitesimal of higher order than the element of arc? Geometrically we see that the characteristic of the two adjacent tangent planes must be perpendicular to its conjugate. D' = "\ "\ "\ >. D" to be determined presently. Conversely.^ . will be Ddu + Zlfdu. = DcZu 2 f 2D'dwdv f J9"c?v 8 (20) These equations will determine the signs of Z>.^L^r. briefly put. fix \*v d . Returning to the point (y). or asymptotic (19) lines.
This fairly plausible geometrical reasoning may easily be put on a sound analytical basis. In any triply orthogonal system of surfaces. The normals to a surface may be assembled into two families of developable surfaces. mentioned above occurs in the case where Theorem 8. This is the theorem of Joachimsthal. well known in the euclidean case. and these give the elements desired. lies in one surface of each family. . No less celebrated is the beautiful theorem Theorem 9. with the exception of those at umbilical points. The necessary and sufficient condition that the four points (03). by (14) Edtt + Fdv Ddu+D'dv J)'d Fdu+Gdv This is = 0. (y). they intersect at a constant angle. (y + dy) should be coplanar is  yxdxdy (xdx) \ = 0. (x + dx).XV DIFFERENTIAL GEOMETRY 197 perpendicular to the element of progression. = 0. and if two surfaces intersect at a constant angle along a curve which is a line of curvature for one it is a line of curvature for the other. (xdy) (xx) x x . except at an umbilical point where the involution of conjugate tangents is made up of mutually perpendicular tangents. If two surfaces of Dupin. there will be just two tangents which are mutually conjugate and mutually perpendicular. The integral curves of the differential equation (20) are see at once that called lilies of curvature. We intersect along a line which is a line of curvature for each. Theorem 10. . At any general point of the surface. and adjacent normals are coplanar. (21) and the Jacobian of the binary homogeneous forms (13) and gives the two tangents which are both mutually the indeterminatiou perpendicular and mutually conjugate (20). Each normal. the curves of intersection are lines of curvature.
therefore. since. The normal at any point will determine. generate an evolute. carry any surface all of whose curves are lines of curvature into another such surface. into another such system. n X The vanishing of D' and ^ proves our theorem. Every carries a sphere into conformal transformation a sphere. hence Theorem 11. by hypothesis. with any other point Q of the surface. orthogonal surface to a bundle of concurrent lines is a sphere. or normals must pass through one point. one normal Hence the normals at P and Q intersect. hence a line of curvature into a line of curvature. Our statement in Chapter XIII that confocal quadrics intersect in lines of curvature is hereby justified. and hence. by (7) make a fixed angle with the plane . It will. is here regarded as a special case of a . The normals to the surface along this curve. a plane. will. of space Of course a plane sphere. mutually per o __ X \ ) f /oX ( v X \ I (oX = /uX ( o ~'X \ I + /dx I u* 1 X *\ _ ( /_^ ^^'\ ) ^ ") + ( =O ") t / x ^^\ J = r /^fl/ (^ <^^\ /^^ ( )= = / ( ^x ^x\ ) == 0.198 INTRODUCTION TO CH> Let the three families of surfaces be given by the equations As the parameter pendicular lines are. in every case. A P and all lies in must be null. Evidently the this angle the plane. by hypothesis. surface all of whose curves are lines of curvature must be a sphere. Let us suppose that we have a conformal transformation of It will carry a triply orthogonal system of surfaces space.
A/ tan du* + ~^~~~ dv*. dy$ sin = \(x i f fxy^). if we eliminate . r 5 ds dy. lli sm r . On the other hand.Gdv = . hypothesis. > = 0. T . ._ = dx: COS r p sin r _^ k k ds ds r \_ r XiSmr k r~\ dr *l ^COSy y k] d$ Now. (dxdy) =   2 rfit 4 tan . (ajy) But (xdx) = (aJcZy) = A (y^fe) = fx = (yc%) = = 0. = (3v y r* tan/ /c *. a point whose coordinates shall be called (x). Let r be the distance from the point (x) to the intersection of the normal there with the adjacent normal.tan 7* j r Eliminating tan A/ du. . <>yi ^p 0?. curvature ^ xi let us take as parameter lines the lines of ^y<> r* Oit = ri tan f A. dv we get we get our previous differential equation for the lines of curvature. xi daT* =x r cos T . Edu+Fdv= Fdu 4. by and (y). ^ au ^x > ~~^ <.) is linearly dependent on (x) T dxi cos r .xv DIFFERENTIAL GEOMETRY 199 Let us now examine the normals along a line of curvature. cfa^ = cZ^ tan y In particular. (22) In the general case. (T.
Notice that they are absolute simultaneous invariants of the two binary forms (13). tures of normal sections through the tangents to the lines of curvature. r. + ~ ED"+GD2FD' : rg i tan / K . 1 & tan . mean relative respectively. the curvature at any point of a curve of the surface is identical with that of the curve of intersection of the osculating plane with the surface. / L . this does not involve derivatives of higher order than the second. .200 INTRODUCTION TO 2 CH. j I MI ~6 ^* \ I /du^ ~7 I 9 <w du dv + 2Fdudv + The indetermination of sign may be used to make the curvature essentially positive.F) = 0. As. by (4). Along our curve u and v will be functions of s the parameter of length of arc. 1C (23) 1 .D' ) tan 2 A/ + [ED" + GD. These last two expressions shall be called the curvature and the total relative curvature. XL Let us now look at the more general question of the curvature of a curve on our surface.2FD'] tan j + (EG. so that. k . I / ' ^v dv ds_\ The cosine of the angle which the principal normal to this curve makes with the normal to the surface may be written COS (r = " f 'T<>~ (yz) > Of = dtj "7 p ds "t" r X% COS ' or 7 K 2 c) p i/^ ^ Vi t _ . by (2). the sum and the product of the curvaThey are. . (20). using our previous notation. 'l 7 / IV f^* du * N L<w 668 I c)' I I dv~] 7 I I . (DD" .
1f) cos 2 sin 2 k tan / A* k tan  k The normal sections of a surface at any point the greatest and the least curvature are those deterhaving mined by the tangents to the lines of curvature. they will come very easily from considering the equation of a central conic as given in Chapter XII. Of course the theorem is untrue at a point where the tangents to the two lines of curvature coincide. If on each tangent to a surface at a point Theorem 13. Reverting to our previous expressions r15 r 2 and taking the parameter lines.. Theorem 14. if be the angle which the chosen tangent makes with that to v = cons.xv DIFFERENTIAL GEOMETRY 201 Theorem 12. Meunier's. /. The curvature of a curve on a surface at any point is equal to the curvature of the normal section with the same tangent divided by the cosine of the angle which the principal normal makes with the normal to the surface. the locus of the points so formed will be a central conic. We leave to the reader the task of filling in the details of the proof of the last theorem. 7 k. This central conic is called Dupin's Indicatrix in the euclidean case. fctan^ . * r\ds) rMs) V 3 7x k tan  . a distance be laid off' equal to the square root of the reciprocal of the measure of curvature of the normal section with that tangent.tanj dj. tan ~ I or. . and we may well use the same name in the noneuclidean case also. the curvature of the normal sections through the tangents to the lines of curvature lines of curvature as 1 i _ . . { = tan dy{ bx{ = tan by h r_^_A + _j_An.
2 (edu + 2fdudv + gdv = ~~ l tan ^ ri 1 y /u 2 (c^ + 2 JW. i. ofw + Grfy ) + 2 i ' tan ^ > . the tangent plane to the surface is the osculating plane to the curve. the tangent plane to the surface tends to rotate about the tangent to this curve. in the case of an asymptotic curve. such a curve the point (y) on the normal will replace the point we previously called (). and the normal In dealing with to the surface is the binormal to the curve. the second part of the righthand side of (25) will be zero. 7* (25) tan^ A > An asymptotic curve has the property that as a point moves along it. by (8). (dxdy) (dydy) = _ ~ ' EGF'1 EGF* \<)V<>V. The curvature of a surface bears a close relation to the element of arc of the point (y). kds But.302 INTRODUCTION TO CH. e. while the paren . The torsion of any asymptotic line will be.
The two asymptotic lines at a point. point will therefore be distinct. curvature in terms of E. F. hence : real. and the necessary and sufficient condition that a surface should have constant total relative curvature is that the asymptotic Under these lines of one set should have constant torsion. have torsion with opposite signs. and the square of either torsion will be the total relative curvature. Theorem 15. those whose assume that our surmce is not a developable circumscribed to the Absolute. (xx) = P. v the shall take as parameter lines tangents isotropic curves of the surface. . for an asymptotic It is not difficult to see that the two asymptotic lines at a point. The We isotropic curves at every shall have E=. G and their derivatives. when real. In any surface of constant total relative curvature. e. circumstances the asymptotic lines of the other set will have a constant torsion equal to the negative of that already given. y k f k For the sake of simplicity we u. Theorem 16. when have torsions equal to the two square roots of the negative of the total relative curvature of the surface.G = 0. the torsion of every asymptotic line is constant and equal to a square root of the total relative curvature.xv DIFFERENTIAL GEOMETRY (its 203 thesis in the first part is equal to 2 . we have but to look at the special case of a ruled quadric. It is now time to explain why that Let us try to express our total relative adjective is chosen. also touch the Absolute. In speaking of the total curvature of a surface we have used the word relative. hence. We have ZP B * 7c V ' rs 72* 2 1 tan tan ^u. and that in the region considered no tangent plane to the surface touches the Absolute. We i.
as and there will be the 18. Notice that this theorem remains true in euclidean space where the measure of curvature is 0. by an equation just preceding (22). cit.. 2 F ^~ F ~ t> n \o n" f c t}~. A along u = const. loc. The Gaussian curvature may also be called the total absolute curvature. Bianchi. 1C Then. surface of total relative curvature zero is same tangent plane all a developable. p. A: 2 F 1 F _jF A. Let us assume that r = GO tan t . 609. p. cit.f The total relative curvature of a surface is equal to the difference between its total Gaussian curvature and the measure of curvature of space. Theorem * Cf.. C8. loc. U* ~ F* i r i ( F* I F " ' The fii\st expression on the right is the Gaussian curvature of a twodimensional manifold whose squared distance element is 2Fdudv* Theorem 17. f Cf.204 INTRODUCTION TO *F OH. The problem of finding all surfaces of total relative curvature zero is quickly solved. . Bianchi.
In the case of a horocyclic surface (yy) we may not co assume k 2 . We Manning. but must treat ordinates where (yy) = 0. and whose centres lie on these axes. those which are developable upon the euclidean plane. In elliptic space there is a peculiarly notable class of We have surfaces of Gaussian curvature zero. The centres are thus mutually orthogonal points. and the question arises. i. cuts the Absolute in two generators of each set. In the hyperbolic case let (y) be the centre of a sphere. hence the total relative curvature is > /c* 7 2 4 and the Gaussian curvature is zero. be it remembered. We get then 1 __ _ _ 1 ~~ ' (y) as homogeneous o r 7 9 k2 tan2 T .. notice also that the generators of either set are paratactic. . ruled surfaces. k* Theorem 19. This statement was given without proof in Chapter X. 1898. Much more interest attaches to the surfaces of total Gaussian curvature zero. the constant distance thence to points of the surface being r k2 If the surface is to be actual (xx) If the sphere be a proper one (yy) = k 2 the total relative curvature will be . Now Dupin's indicatrix shows that the normal sections of greatest and of least curvature will be determined by tangents bisecting the angles of the two generators. will not this fact alone constitute a sufficient condition that a surface should have Gaussian curvature zero ? * Cf.e.n/b .xv DIFFERENTIAL GEOMETRY 205 Clearly every developable has total relative curvature zero. already seen one example. 52 . The total relative curvature will be A/ ^ There is an advantage in considering the hyperbolic and elliptic cases separately. > . Killing. . Lie Grundlagen der Geometric. p. 33. and the planes of these normal sections will cut the surface in two circles whose axes are the axes of revolution of the surface. cit. p.* The horocyclic surface of hyperbolic space is developable on the euclidean plane. This quadric. and its own generators form an orthogonal system. loc. Pader born. the Clifford Surface of Chapter X.
For an interesting treatment of these surfaces see Bianchi. and we may choose 11 so that it We shall be equal to unity ds 2 =du + dv*. The square of the element of arc may be written Since the Gaussian curvature is zero On the other in the form hand we may write our surface parametrically 'i = /<() cos v* + <t>i(u) sin * with the additional conditions kF = (</>/') cos 2 . Serie 2. Le superficie curvatura nulla nella geometria ellitica/ Annali di Matematica. suppose that we have a ruled surface of Gaussian curvature zero. These are identical with previous E = [* only when We may. . moreover.(/*') sin^ = 0. take * a. We have for our distance element Let us imagine that know. Hence E is a function of u alone. then. by Chapter IX that if two lines be paratactic they have an infinite number of common perpendiculars on which they determine congruent distances.206 INTRODUCTION TO we have CH. 1896. a surface generated by co 1 paratactic lines. (ft') = (tf>) = 0.* The parameter v shall give the actual distance measured on each line from an orthogonal trajectory v = const. 2 (27) and the Gaussian curvature is zero. Conversely. ' Tomo 24.
the If we compare . and so are paratactic. The necessary and sufficient condition that a ruled surface in elliptic space should have Gaussian curvature zero is that its generators should be paratactic.xv and this DIFFERENTIAL GEOMETRY 207 distances shows that two adjacent generators determine equal on all their orthogonal trajectories. to a curve v const. e. Our given surface may be written in the form v v cos This shows = xu 2 = dv This reduces to 4 cos 2 T 4 jf^ ^ L ^ u2 * difi + dv 2 . and only when Theorem 21. i. are = a rx that the generators are binormals to their orthogonal trajectories. Theorem 20. however. Another highly interesting criterion for a surface of constant Gaussian curvature zero is obtained as follows : x \ ^x *u dudt. The proof given holds equally surface is. we get in hyperbolic space . in that case imaginary. The necessary and sufficient condition ruled surface should have Gaussian curvature zero is should be generated by the binormals to a curve squared torsion is equal to the measure of curvature of that a that it whose space. when. theorems 16 and 21./ are __ ~ 2 V5tT ^v) /x <)x\ __ fix ~ Vdu aW ~ <*x\ __ / ' \ X x\ __ ^u 2) The coordinates of the absolute pole of the tangent plane The coordinates of the absolute pole of the osculating plane to the orthogonal trajectory of the generators.
V = . then. from to the cos  o = 1 f) 1 i fix / 1. the curvature of its orthogonal projection on the tangent plane at that point. . i.1 t 1 r\fll (29) f \ Pg For the other parameter line Let us now. Then. it We shall. . = (z) e. Theorem 22. find the geodesic curvature of the curve V (U ). It will. Let us denote this by >. let us find the geodesic curvature of one of our parameter lines ds v ' F= = _ = </~G dv. assuming 0. The intersections of the given surface with the various families of paratactics will be the asymptotic lines of the former. ~ P c d* k </GU>v\Jd To find cos <r we must determine the distance of the point orthogonal to (x) on the curve v const. define as the geodesic curvature at any point of a curve on our surface.. applying Meunier's theorem to the projecting cone q ! = C09<T Pg . as in euclidean space. is that the given surface should have Gaussian curvature zero. P (28) As a first exercise. while o is the angle which the p osculating plane makes with the tangent plane to the surface. more generally. be possible to assemble the normals into families of right (left) paratactics also. The necessary and sufficient condition that should be possible to assemble the normals to a surface into one parameter families of left (right) paratactics.208 INTKODUCTION TO CH.
e. VIII. Ch. shall get * For a proof of of Variations. du " " 1 9 *u*v v ' + ^"^ + "^ '2 I. and. ds = ' . If we assume that two sufficiently near points can be connected by a curve of minimum length * we always This merely the first i. tlie existence of this curve. see Bolzn. 1904.xv DIFFERENTIAL GEOMETRY shall 209 Once more we so that make use of the isotropic parameters. i. we dv shall have * VJ tells us that our given curve is an extremal. variation of the length between two fixed points is zero. Lectures on Chicago. What will be the nature of those curves whose geodesic curvature vanishes. the Calculus . evidently. For an orthogonal trajectory to this curve cv ~ ~7 == V . e. those curves whose osculating planes pass through the normal? These shall be called geodesic lines.
p. and the formula for the distance element is written in the same shape. In any congruence of normals. have given this formula in terms of the Beltrami We have. is CH. geometry. purposely avoided the introinvariants. we have further trajectories of a family of of the surface generated by If we consider the two planes through the normal to a surface and the two tangents to the lines of curvature. Differentialgeometrie. The orthogonal paratactic lines are geodesies these lines. G and their derivatives are the same in cuclidean and in noneuclidean space. loc. pp f 307. we see that they are mutually perpendicular. the edges of regression of the developable surfaces are geodesies of the focal surfaces of the congruence. f As a last problem in the differential geometry of surfaces let us take up that of minimal surfaces. It is also clear that as the expressions for the geodesic curvature of a parameter line in terms of E. therefore. however. g. . To begin with. by (15) _ and the area of the elementary quadrilateral * For e. 308. a geodesic for all space a result also evident from Chapter II. for instance.210 INTRODUCTION TO 23. what It is perfectly clear that the will be the element of area ? expression for this will be the same as that in the euclidean The sine of the angle formed by the parameter lines case. so will the formula for the geodesic curvature of any curve be the same. Theorem 24.* Theorem 25. Blanch i. The osculating plane to any straight line is indeterminate the line is. points of a surface Remembering 21. and that each touches the focal surface of the congruence of normals at the point of intersection of the two adjacent normals in the other plane. F. cit. i. will be. vol. duction of these into the present work. a simple proof of this general theorem see Picard. 253. We might.. . 30. cit. Theorem The curve of shortest length between two a geodesic line. and will therefore merely refer the reader to the current textbooks in differential .
and neglecting higher powers of w we have VEO w Ik IVt M ^tan / + tan tV\ ' k M 2 tan r tan 7 k k dudv.~ k Ic . . w dx 7 __ =  w w j i r r w a^cos. "fc + tan F . w . ~ 7 tan 7 . The formula for the area enclosed by a given curve will be N^EGdudv. Let us compare this with the area enclosed by this curve upon a surface reached by laying off on each normal an extremely small distance _ . w (uv). by (22) w sin COS W 71C f sin w j1C . in particular. take the lines of curvature as parameter lines. when we neglect powers of w above the first. T2 I This becomes. . ' 1 * 9 7 tan  Developing by the binomial theorem. 7/^smr mo.* dudv.tan . w~\ j The squared element of arc for this surface will be .XV DIFFERENTIAL GEOMETRY 211 Let us. For the surface element we have X VEO .
It is very interesting that in noneuclidean space we should have an algebraic minimal surface (other than the plane) whose order is as low as two. exhibit the differential equations on which they depend. The necessary and sufficient condition that a surface should be minimal is that the asymptotic lines should form an orthogonal system. Darboux. may go one long step further towards the solution of the problem of minimal surfaces. The necessary and sufficient condition that a surface should be minimal is that the mean relative curvature should be zero. The reader is . hence We Theorem 27. and only when. iii. vol. This theorem justifies our statement in Chapter X that a Clifford surface is a minimal surface. Let us put * Cf. 1894.212 INTRODUCTION TO . If wo define as a minimal surface one where the first variation of the area is zero. harmonically separated by the asymptotic lines. strongly urged to read this interesting chapter in connection with the present work. see from (23) that the numerator of the expression for the relative mean curvature is the simultaneous invariant of (13) and (20). we have Theorem 26.CH. hence We We JO n and take for (x) four It is merely necessary to find k2 solutions of (3) subject to the restriction (ocx) F = .* shall once more take as parameter lines the isotropic These will form a conjugate system. namely. and vanishes when. since they are ones. the tangents to the asymptotic lines are harmonically separated by those to the isotropic ones. Lemons sur la theoric generate des surfaces. Paris. ch. certainly a necessary condition. xiv.
and cannot be real in the actual domain. since xii 213 ~~ ou c We easily find 3Px$ 1 <>F ^Xi ~ Now <$v \<)u 2 <H Hence If developable.xv which is DIFFERENTIAL GEOMETEY certainly possible. by (24). Conversely it is clear that every developable circumscribed to the Absolute is a minimal surface in that its asymptotic lines are mutually perpendicular. In the second case let us suppose (u) ^ 0. curvature is zero. The total relative (/> Let us replace replace the letter u by Tb H(u) so that letter (   ~ j = . even though it lie in a region of our space where the concept area has not been defined. hence a minimal developable must be circumscribed to the Absolute. Then by the u once more. In like manner * = F ^v Zv k* Multiplying through by ^~ and adding 2 1 <x> k 2 * F^ 1 *F_DF Yv . and the surface is In a developable surface the asymptotic lines fall together.
14 DIFFERENTIAL GEOMETRY the other hand CH. F Lastly. tf^~ + sin2w = %u<*v When from (33) ' (x) F has been found we may. . as already noted. let us put J (32) ^ p #iw 0. xv On yFL ~~ (^ *J?\ fa 2 ^ 2 X\ 1 /^X ^ (~<Px "~ ^?^Tv" k*(^u^ _ 1 2 _ /^ ^. find (31).x 2 2 ' F *H Zv' *F.
shall write the following fundamental quadratic expression = &(dydy)(xdy)* Edit? + 2 Fdndv + Qdv*. . in particular. Scuola Normale Superiore.CHAPTER XVI DIFFERENTIAL LINEGEOMETRY IN Chapter .dv + gdv\ * The first part of the present chapter follows. and. whose coordinates aro analytic functions of the two independent parameters in question. (xx) = (yy) = lc\ we : (xy) = (\ (I) Following Rummer's classical method. a rather inaccessible memoir by Fibbi. (2) k*(dxdy) = edu* + (f+f)du. a theory of twoparameter line systems or congruences. say u and v. This is equivalent to supposing that our lines are determined by two points. ' I sistemi doppiamente infiniti di raggi negli spazii di curvatura costante/ Annali della R.* We shall define as an analytic linecongruence a system whose Pluckerian coordinates are analytic functions of two independent parameters. IX we gave the foundations of the Pluckerian linegeometry. and the fundamental invariants of a metrical character in Chapter X we saw what advantages arose from taking the cross instead of the line as element. = + 2 F'dudv + G'dv*. Chapter XV was given to the It is the object differential geometry of curves and surfaces. and introducing suitable coordinates. of the present chapter to draw all of these threads together into a theory of differential linegeometry. XJ^K^UV). Pisa. yi = yi(u>v). 1891. with slight modifications. which we may assume mutually orthogonal.
(8) ~ 1 (9) .216 DIFFERENTIAL LINEGEOMETRY CH. (3) (4) (5) (6) The following expressions : relations will subsist between these various since A2 y^ E= ~ 2 [G'e* 2F'ef + E'f*].
) the point of the first line orthogonal to the first point.G')dv 2 ] sin ~ cos = 0. (11) . as a first in pairs of mutually orthogonal points. The coordinates of an arbitrary point of our line may be written (#cos.F')dudv + (G. problem.+ y sin j while an arbitrary point of an adjacent line will be X(x + dx) + ^(y + dy).(x + dx) \(y + dy). (ydy) = .E')du2 + 2(F. ryi we shall 2 have %(dxdx)"\ sin fC [k (ydx) cos = fC M  [P \(dy dy)] sin rC (xdy) sin (xdy) cos K> ryt . There will result two linear homogeneous equations in A. find where the common perpendicular to a line of our congruence and an adjacent line meets the given line. = 2 [!c . We remember from Chapter IX.%(dydy)] cos (ydx) sin + 2 [k t  (dxdx)] cos j =0  (10) Casting aside infinitesimals above the second order . while. Let us begin by writing that the second of these points is x (7* sin r 7*\ 2/ cos T. (xdy) + (ydx) /y> = . and whose determinant must be equated to zero. the first point lies in the absolute polar plane of p. on the other hand. (edu 2 + (f+f)dudv + gdv2 ) 2 ^cos ^ sin 2 } + [(E.(ydx) 2 k*(dydy) + (xdy) 2] sin r cos ~ = fc fa 0. When this is simplified in view of the identities /ut (xdx) =  (dxdx). that two lines which are not paratactic have two common perpendiculars meeting them Let us.% (dydy).(dxdy).xvi DIFFERENTIAL LINEGEOMETRY 217 Notice that A and A' being square roots of positive definite forms cannot vanish in the real domain.[k*(dxdx) .
we saw sponding We inequality (12) holds a ' general congruence. Equating to zero the partial derivatives to du and dv we get = A/ 0. It gives. these common perpendiculars will generate a surface of the fourth order.218 DIFFERENTIAL LINEGEOMETRY oo 1 CH. we have \e(GG')fj(EE')]dudv V* = 0. those points of the given line where the two perpendiculars coalesce. On the other hand. (14) The lefthand side of this equation is the discriminant of (11) looked upon as an equation in du dv. Theorem 5. with the correelliptic case. (13) Each root of this will give two values to to tan . therefore.corresponding K two mutually orthogonal points. if we eliminate duidv we get . This will give determinations for r in the general case where e f : ( ~f) : <J $ (EE') : (FF') : (G.F')du + (G. Such points shall be called 'limiting : .(F~ F') (/+/) + g (EE')] (tan" l) t&n* tan + [(EE') (GG')(FF')*] = fc 0.G'). analogous to the shall call a congruence where euclidean cylindroid. (12) and. P du + gdv\ (tan 2  1 ) [(F. as in Chapter X.G')dv] tan A/ Eliminating r = 0. Let us now ask what are the maximum and minimum values for r in (11).
when real. 26. determine two real regions of the line where it meets the real com mon perpendiculars with adjacent lines of the congruence. determine two real regions of the axial pencil through the line which contain all planes wherein are real common perpendiculars to the line and adjacent lines of the congruence. (16') we shall have (17) 0. line of a general analytic congruence will pass four limiting and these. which contain the intersections of the line with the real common perpendiculars. They are also the planes in which the two perpendiculars coincide. vV = In like manner for the limiting planes 2(^2 + ^2)2 0. + 4a*x*X* = 0. Reverting to equation (8) of Chapter the equation of the ruled quartic surface will be. determine all planes wherein lie all real common perpendiculars to the given line and its immediate neighbours. a2 ) 2 = (x* + x*) 4 fa atfx*x* Dual (17') * See the author's Projector e Geometry. They will determine two regions (when real) point by point mutually orthogonal. equate an equation in #! : X2 or xl : x. planes. now look more closely into the question of the of limiting points and places. and these.xvi DIFFERENTIAL LINEGEOMETRY 219* points '. . mutually perpendicular in pairs.. may so choose our reality coordinate system that the equations of the line in question shall be x 1 x<z 0. have (15') foaa) 2 (X + x^)x l x^(a + a2 )(x^ + x/)x l = find the limiting points on the line x 1 to zero the discriminant of this looked upon as To xt 0. mutually orthogonal in pairs.A . Through a general analytic congruence contains four limiting points. A line of a Theorem 1'. when real. cit. and which when real. They are also the points where the two perpendiculars coincide. 3 shall 0. in pairs.A = () 0. Theorem 1. in the hyberbolic case shall We We = = X 2 a(^xQ + x^)x I x2 + b(x1z + x^)a t x. p. with their verticals. mutually perpendicular. (15) Let the reader show * that in the elliptic case we x. In the same way we might find limiting planes through the line determining two dihedral angles whose faces are.
Theorem 3. Giving to # : #3 one of the values from (16') we see that Substituting in (15') we have two The four limiting points hence will yield but these planes. 0. In elliptic space this may occur. angles of the limiting planes If we look more closely into the roots of the last four equations we see that the roots of (16) are all real. and the limiting planes imaginary. 1. whence TJieorem 2. or the planes may be all real and the points all imaginary. The perpen' diculars at the limiting points line in two planes called 'principal planes' whose dihedral angles have the same bisectors as pairs of limiting planes. 1. 0) (0. are Notice that the centres of gravity of the limiting points while the bisectors of the dihedral (1.^2)^0 + . Let us pick out a pair of limiting points which are not mutually orthogonal. 0. 0. The perpen Theorem 3'. In hyperbolic space the limiting points of an actual line are real. 0. 0) (0. 1) . A/C^ A/a 2 ) (v^f Va2 . the other imaginary ones. say 2. Reverting to : (16') we ( see that v'ttj we may ) : also write x o X3 = Va 2 ( A/% + Va2 ).220 DIFFERENTIAL LINEGEOMETRY OH. Calling d^ d2 the distances thence to the limiting points just chosen we have = = ttj f tan k = ( ^. 0. x%). 0). diculars in the limiting planes meet the line in two points ' called principal points whose centres of gravity are those of two pairs of limiting points. are (0. 0. 0. 0. The perpendicular from the point (x) to the line x1 #2 meets it in the point (x^ 0. 0. 0. 0. the one will have real roots. those of As for the two equations (16') and (17') (17) all imaginary.
be the angle which the plane through the principal plane == 0./V shall have f cos yi sin = ^ . Fibbi's work one cannot help admiring his skill is burdened with many long formulae in handling such cumbersome expressions at all. = (xdy). Differentialgeometrie.* to the notations wherewith we opened the Returning present chapter. the points where it intersects adjacent lines of the congruence. (18) of course. cos 2 co = y1^ tan ^ 2 co . 57. (xi + dx { ) cos j .xvi DIFFEEENTIAL LINEGEOMETRY x^ 221 Further. * For the Hamiltonian equation see Bianchi.sin y cos . p. cit.. cit. let us find the focal points of our line. 26] For the noneuclidean form here given.j. + gdv\ cos r 1C f [J^'cJu sin T 1C = 0. or rather. cf. f (yi + dy4 ) sm J/^ 7 x . the points where the distance becomes infinitesimal to a higher order. sin 2 . Fibbi. let ojj = = (co) and (x) makes with xl f x. This known formula / r (x COST we . then multiplying through by . .and adding again /v /v cos r + [ff'cJttr + F'dv] + Gr'dv] sin T = 0. & d?' dxi cos T f dyi si n j)tc/. Here. co = co ^ cos + tan ^ sin 2 = 0. . e.r (x. if the focal point be is. loc..) = 0. the direct analog of Hamilton's wellfor the cylindroid. r~ f /^ 2 L c)'?/ Multiplying through by  : and adding. i. . p.
and these will vanish when. The centres of of the focal points are gravity identical with those of two pairs of limiting points.FeZiH. and 4. Theorem Theorem 4'.//') tan* = Q l + [JS<jF(f+f) + (21) Ge] + (EG F2) Subtracting one of these equations from the other F*)(< gfffl ! = 0. Eliminating r (E'f .222 DIFFERENTIAL LINEGEOMETRY (xdy) we sin ~ A* OH. We have then . focal properties of a congruence of Here we may suppose that (y) is the cially interesting. similarly cos j /y [edu + fdv] r/cfo.F'(f+f) + G'e] tan /^ /c ff) (/.FJ tan. Absolute pole of the tangent plane to the surface described by (x). x (19) ' [/CM.F'(ff) .G'e] du dv Eliminating (E'G' du dv : + (F<jGf)dv 2 =0. Replacing (ydx) by have.Gdv] COST = 0. ?' . + sin j + [.* + [E'g .Fe)du. (22) We see at once that the middle coefficients are identical in (22).J + [E'g . The bisectors of the dihedral angles of two focal planes are ideutical with those of two pairs of limiting normals are espe The planes.] + \_Edw + Fdv] = 0. we (14) are measuring from a centre of gravity of the roots. and only when.
and (ycfe) must be an exact differential. _ FgGf . Writing their equations in the form  Xxydx =  0. find the necessary and sufficient condition that the focal planes should be mutually perpendicular.  Xxybx = \ 0. by dvbv (20). and sufficient that the point of the be orthogonal to every displacex r (7* sin fc line orthogonal to (x) should y cosy).fc5 ~~ Tlf^e Tllf^Pe Hence rdw ' Idv + 5ul bv bv] _G ~~ ~ EfFe Let us give the name pseudonormal to the absolute polar thus get of a normal congruence. We . in fact. /=/ . ' /oox (23) This condition can be put into a more geometrical form. = kdr. (ydx) ^(dxdx) For perpendicularity.xvi DIFFERENTIAL LINEGEOMETRY 223 Suppose. Lot us. conversely. the numerator of the expression for the cosine of their angle will be k2 2 (ybx) *(bxbx) (dxbx) = 2 /c [k?(dxbx) (ydx) (ybx)]. Edubn f F(du bv + bu dv) f Gdv bv = Now. 0. e. T ^ f y$ sin ^ and show that we may find T so that our lino is For this it is necessary normal to the surface traced by (x). that /=/' Let us put X} = T x$ cos . we must A// 7*\ have 7* _ 7* sin j (xdx) /V (?/<rte) cos T A/ (ydx) = 0.. . i.
lines in this surface the isotropic curves. Let us suppose that we have a normal congruence deter mined by mutually orthogonal points (x) and (?/). where x$~Xi(uv) traces a surface. and only when /=/. The necessary and sufficient condition that a general congruence should be composed of normals is that Theorem 6. so that __ "~" We x parameter __ \& The normal sine of the angle which our given line to this surface is makes with the . If a constant distance be laid off on each normal to a surface from the foot. the focal points should coincide with a pair of limiting points. constant. not one of the orthogonal trashall choose as jectories of the congruence. tually orthogonal. Theorem 5. (xdy) = 0. In a normal congruence let us suppose that (x) traces a surface to which the given lines are normal so that (ydx) Let us then put =. The necessary and sufficient condition that a congruence should be normal is Theorem 5'* The necessary and sufficient condition that a congruence should be pseudonormal is that the focal points on each line should be mu that the focal planes through each line should be mutually perpendicular. If we subtract one of the equivalent equations (20) from the other. in such a way that the points on adjacent normals are on the same side of the tangent plane corresponding to either. ~ where y is sin 2ft = a*sin0. 2ft cos . we get an equation which reduces to (13) when.224 DIFFERENTIAL LINEGEOMETRY CH. We (ydx) see at once that = .(xdy) = Theorem 7. the locus of the points so found is a surface with the same normals as the original one.
=/ = /= = </ <>. that the first is a normal congruence. for focal points expressed in (23) Our condition was inde pendent of (12). also finite Theorem 8. contrary to There are two sharply distinct sub. We shall now abandon the general congruence and assume that. and this shows that our two subcases just mentioned differ in this. the focal points will fall into (x) and (y) likewise. We see also by theorem (22) of that chapter that it is possible to assemble the lines of our congruence into families of left or right paratactics according as we assemble them by means of the one or the other set of asymptotic lines of the given surface. that the total relative curvature of the surface will be jk We see that or the Gaussian curvature zero. these perpendiculars will either all meet the given line at one of two mutually orthogonal points. the resulting conis gruence normal. as we readily see. we shall have point. If a normal congruence be subjected to any number of reflections or refractions.xvi DIFFERENTIAL LINEGEOMETRY 225 Let all the lines of our congruence be reflected or refracted in this surface in such a way that sin 6 =n ^ 4 sin 0. while the second is not.cases which must not be confused : In either case. These are mutually orthogonal. We must replace y by y where  _ 1 ^X ^X I ^^{ ^ ^' ^ ^ ' It is easily seen that for the new congruence /=/'. (11) is illusory. (y) being thus the other such Then under our first hypothesis. or two adjacent lines will be paratactic. . and there no ruled quartic determined by the common perpendiculars to a line and its neighbours . and have oo is l common perpendiculars. Let (x) be a point where our line meets a set of perpendiculars. Conversely. and so by equation (26) of the last Chapter.
looked upon as an equation in r. xxxix. A discussion of these definitions will be found in a note at the beginning of the second of the author's articles. distinguishing the latter by the name of normal '. 1903. we shall define as * pseudoisotropic '. later. we have given a congruence of normals to a surface of Gaussian curvature zero. and the focal surface must be a developable circumscribed thereunto. two values of are paratactic thereunto.226 if DIFFEEENTIAL LINEGEOMETRY CH. 1904. Atti delict li. then. Hence the focal planes all touch the Absolute. and as we have entirely illusory. becomes Hence (24) must hold. Xj = f T y sin r > iv h (dxdx) = cos Y (dxdx) sin 2 T fy* * The earliest discussion of these interesting congruences in noneuclidoaii space will be found in the author's article Les congruences isotropes qui servont a representer les fonctions d'une variable complexe*. two normals adjacent to a given one There must be. ' ' ' * . In the same number of the same journal as the first of these will be found an article by Biancht. on the other hand. included in his definition of isotropic congruences those which.' Professor Bianchi uses the word isotropic to cover both what we have here denned as isotropic congruences. Accademia ddle Scienze di Torino. Sulla rappresentazione di Clifford delle congruen/o *rettilinee nolle spazio ellitico. It is clear that the lines of such families. so that we may put We may take as coordinates of a focal plane (uu) = But by (20) this expression vanishes. 0. and xl. The author. a congruence cannot bo assembled into paratactic c This type of congruence shall be called isotropic '* Let us take an isotropic congruence. We now make the second assumption : We various shall still take (x) as a point where the line meets the common perpendiculars. du do for which (11). and also congruences of normals to surfaces of Gaussian curvature zero. normal congruence (23) is also true. or congruence of normals to a surface of Gaussian curvature zero. and choose (x) and (?/) so that e = *(/+/) T x cos 7 k 2 ij* =g= .
does not. suggested by the developments of Chapter X.^o) .xvi DIFFERENTIAL LINEGEOMETRY 227 This expression will be unaltered if we change r into /. and the congruence will be either isotropic. there is advantage in studying our last two types of congruence from a different point of view.rX ( r X) = 1. we must have (dxdy) = 0. called. II parallelismo di Clifford negli spazii The latter writer ellitici. (26) were formerly looked upon ( r X) as giving the lines through the origin (1. which pairs of points determine always the same distance. respectively. They may now be looked upon as coordinates of two points of two unit spheres of euclidean space.' and Fubini. Let us rewrite the equations (11) there given. 0) respectively left and right paratactic to the given line. distinguish with sufficient clearness between rays and 4 ' * lines. The centres of gravity of these pairs of points will be the points where the various lines meet the common perpendiculars to themselves and the adjacent lines.xk Vj) = rX i ( 25 ) These equations were originally written under the supposiAt present if we so tion that (x) and (y) were homogeneous. or composed of normals to a surface of Gaussian curvature zero. . The necessary and sufficient condition that a congruence should be either isotropic. Scuola Normale di Pisa. We shall assume that by reversing the signs in one triad of coordinates we replace our ray by a ray on the absolute line is 4 '. ix.' Annali della K. doubly overlaid with two opposite rays meaning thereby a line with a sense or sequence attached to its points. Zur nichteuklidischen etc. however. 0. Vol. is that it should consist of lines connecting corresponding points of two mutually applicable surfaces. as indicated in the beginning of Chapter V or end of Chapter IX. or composed of normals to a surface of Gaussian curvature zero. on the other. 0.* The representation is not. we get the same line if we replace either representing point by its diametrical We shall avoid ambiguity by assuming that each opposite. In elliptic (or spherical) space. (WJi . On the one hand the two lines of a cross will be represented by the same points. the left and right representing spheres. unique. Conversely. Theorem 9. . This representation was first published independently by Study.( xj VU . p 2 . 1 we have choose the unit of measure that k = GA>Y) These coordinates (jX). when such is the case. however. 1900.
The condition that a congruence should be pseudonormal is either normal or = (d r Xd r X). or intersect one another's polars if = (d r Xd rX). while by reversing both sets of signs. l^i l X i(r X ) l fX2 A)' Two adjacent rays will intersect. of the pairs of representing points shoiild be equal each will intersect the absolute polar of the other if these spherical distances be supplementary. Theorem 13. polar of its line. r x = r^*M> i or else. we Theorem 10. rays on mutually absolute polar lines by identical points on one sphere and opposite points of the other. will have The common perpendicular coordinates to two adjacent rays r Xi JL i frXdtX . lines of Two Theorem 12. Reverting to Theorem 12 of Chapter X. sphere.sufficient condition that the two rays should intersect is that the spherical distances . There is a perfect one to one correspondence between the assemblage of all real rays of elliptic or spherical space. Tlieorem 11. = ~  Z XdX rrr r . The necessary and . Rays on left (right) paratactic lines will left (right) be represented by identical or opposite points of the rays shall be said to be paratactic when their lines are. in general. replace the ray by its opposite. Each ray of a common perpendicular to the two rays will be represented by a pair of poles of two circles which connect the pairs of representing points. . great lines of It is clear that an analytic congruence may be represented i in the form i x = zi'tM. The perpendicular distances of the lines of two rays or the angles of these rays are half the difference and half the sum of the pairs of spherical distances of their representing points. Opposite rays of the same line will be represented by diametrically opposite pairs of points. and that of pairs of real points of two euclidean spheres.228 DIFFERENTIAL LINEGEOMETRY CH.
Here two isotropic congruence. wherein the sign in equation (27) will not be changed. On a transformation which changes the signs of all areas. in absolute In the parvalue. Such a congruence will not have a focal surface at all. in we take the congruence of normals to a sphere is (1. 0. we may pass from one normal congruence to another by a continuous change. 0. Study. 0. the congruence of rays in the absolute polar of this plane will be this transformation sign. (27) Let us determine the significance of the double sign.xvt DIFFERENTIAL LINEGEOMETRY Xd r X)(d r Xb r X) 229 (r from these (dtfbtX) particular. perpendiculars to two adjacent lines necessarily intersect. 0. A normal con gruence will be represented by a relation between the two spheres which keeps areas invariant in actual value and sign. ticular case of the isotropic congruence of all lines through the point (1. Lastly. * Cf. in the case of either of these congruences. 0) the relation between the two representing spheres is a directly conform al one. 321 . 0) whose centre we shall get the equations and keeps areas invariant in value and the other hand. We may now repeat p. 0) common ' ' we have an inversely conformal relation. let us say. hence * Theorem 14. Fubini. while in the case of the pseudoisotropic congruence of all lines in the plane (1. 0. cit. The same will hold for the absolute polar of an isotropic congruence. 40. secting arcs of one will make the same angle. congruence will be represented by a relation bcTheorem normal tween the two spheres where the sum of correspond]" Dg areas on the two is zero. A pseudo14'. but a focal On the representing curve. as the corresponding arcs on the other. or each intersects the absolute polar of the other. and every such relation will give a normal congruence. two interspheres. and every such relation will give a Let us next take an pseudonormal congruence. which lies on the Absolute. p. . = (d r Xb r X). loc. a pseudoisotropic congruence.. 0. If.
our transformation must be such as to carry a rectilinear generator into another generator. Let us take up the isotropic case more fully. and get Theorem 15'. e. l t U. the reasoning by continuity used in the case of the normal * congruence. . a/ 1 =  . The necesTheorem 15.>==1^2+ ' U * A = I>   Y A o We shall get a real ray when In order to have a real directly conformal relation between the two spheres. i. Let us give the coordinates of points of our representing spheres in the following parametric form : L 7 A .^ 3? _ u2 = * 1 wt . Any directly conformal relation between the real domains of two euclidean spheres of radius unity may be represented by an analytic function of the complex variable. & \ First given in the Author's first article on isotropic congruences. ' i i ry \ ft l 7i7 i n> \ /OO\ For an inversely conformal transformation ~"~ 1 1\ 2/' 2 IV I/* \ / All will thus depend on the single analytic function u^z). The necessary and sufficient condition that a sary and sufficient condition congruence should be isotropic that a congruence should be is that the corresponding repseudoisotropic is that the belation between the representcorresponding relation t ween the representing spheres ing spheres should be directly should be inversely conformal.230 DIFFERENTIAL LINEGEOMETRY CH.. conformal. recently cited. The opposite of the ray (u) (z) will be >_ t&j l ' 9 ^i ~>.
_  "ifa) If this hold identically. ray of the second sort. There are four solutions to these equations. interpreted in cross coordinates. Let us see under what circumstances such a ray (uz) will intersect a proper ray (uV) orthogonally. If not. Geometrically. we see that either the proper ray must pass through the vertex of the pencil.1 = %iZ 2 + 1=0. the opposite of every ray of the congruence will belong thereto. and analytically we shall We X have Uji6 2 4. which is made up of a pencil of tangents Such a pencil we may also call an improper to the Absolute. those are the equations which characterize an improper cross of the second sort. By considering a special case we are able to pick out those two where the ray lies in the plane of the pencil = u' 1 J 1 or else The proper ray (u ) (z') was supposed to belong to our congruence. This amounts to putting saw in Chapter that. there will still be To begin certain rays of the congruence for which it is true. The condition that the improper one (u) (z) shall also belong thereto will be f . with it will be satisfied by all rays of the congruence for which U i u2 f 1 = 0.xvi Let us DIFFERENTIAL LINEGEOMETRY now what circumstances the ~ } *l : 231 fol inquire under lowing equation will hold 1V . z^ +1=0. or lie in the plane thereof.
(29) is identically satisfied. The focal surface of an isotropic congruence is a developable circumscribed to the Absolute. If. very easy to observe the distinction between the two cases in the case of the linear function If it (3 y. d). and the focal surface which is its own conjugate imaginary.232 DIFFERENTIAL LINEGEOMETRY CH. namely. In these planes only shall we have improper rays of the second sort belonging to the congruence. c. Theorem 17 may now be given in a better form. 13. The two . There are two distinct possibilities first. The necessary and sufficient condition that an isotropic congruence should contain the opposite of each of its rays It is is that the focal surface should be irreducible. our congruence is nought else than the assemblage of all rays through the point (a. p. . those which touch the Absolute at the points of intersection of the two curves of contact with the two portions of the focal surface. Theorem 17. it is not. the focal surface be irreducible. Theorem 16. 8 = a a. every point of the curve of contact may bo looked upon as being in the intersection of two adjacent planes tangent to the Absolute. o. . two conjugate imaginary portions in the second there is one portion which is its own conjugate imaginary. In the first case there will be a finite number of planes which touch the Absolute and also each of the two portions of the focal surface at the same point. and will have a real equation when the congruence is real. . y c + dl. But here will be seen that if we write = a + bi. on the other hand.* The necessary and sufficient condition that the opposite of a real ray of an isotropic congruence should also belong thereunto is that the ray should be coplanar with an improper ray of the second sort belonging to the congruence. The focal surface is the cone of * Sec the Author's second note on isotropic congruences. the congruence contains the opposite of each of its rays. When the latter are present in infinite number in an irreducible congruence. cases here given may be still more sharply distinguished by geometrical considerations. the equation of this surface is reducible in the rational domain In the first case the surface is made up of second. The tangents at each of these points will be improper rays of the second sort of the congruence.
infinite number of lines of the congruence left paratactic to each line thereof. For if it 2 remain constant while t^ varies. we shall have a line congruence of It is the fourth order. but a focal curve composed of two conies. /3. und zweiter Ordnung der Liniengcometrie. have still to consider the congruence of normals to a Here surface of Gaussian curvature zero in ray coordinates. Leipzig. 18926. as is easily verified. we must subtract from this the order of the curve of contact of the focal surface and Absolute. there will be an lines in the congruence. work (1. clearly its own conjugate imaginary. 0. 0) will trace a pencil. by a group m . so our present congruence has as focal surface two conjugate imaginary quadric cones which are circumscribed to the Absolute. we revert to the previous case. y. If ttj be a function of ^ that possesses an essential singularity corresponding to a certain value of z l} we see that as u^ takes all possible values (except at most two) in the immediate neighbourhood. i. or merely write We We tdX^XJ = t(.XVT DIFFERENTIAL LINEGEOMETRY 233 tangents thence to the Absolute. When (u) and (z) are connected by the vanishing of a in i^ and order n in z v in the general polynomial of order case where (31) does not hold identically. there will be a whole bundle of right paratactic If be periodic. When their conjugate imaginary centres fall together in a real point. however. (32) here developed for the elliptic case may be brought into immediate relation with the hyperbolic case. l % of left translations. 0. If it 1 be one of the functions of the regular bodies. when a. however. ii. ct'ster and this pencil will lie in Sturm. On the other hand. may therefore express faX) and ( r X) each as functions of one independent variable. 320. and in so doing we shall get to the inmost kernel of the whole matter. (31) does hold. the left paratactics to the ray in question passing through the point * Cf.J = 0.X All our lr X tr X. b are not connected by these relations. we shall have a linecongruence of order (mf w) 2 When. . and then divide by 2 to allow for the fact that there are two opposite rays on each line. They have. The parameters u l v 2 will determine generators of the left representing sphere. well known * that a congruence of the second order and fourth class has no focal surface. Gebilde p. e. there will be oo 1 paratactics of each sort to each line. a more direct significance. we have a congruence which is transformed into itself r by a group of orthogonal substitutions in ( r A ). and second class. Vol.
Z) = OZ') or (r X] i = ( r X'). synparallelism and antiparallelism. and this shows that u l u 2 are the parameters determining the left generators which the ray intersects. only one value  . It is developed. for there for Uj. 1906. and that. Ueber nichteuklidische und proof. and the conditions for parallelism will be v^= To pass (r uv 2 '= z^ or u/= 'U 2 to the hyperbolic case. a plane tangent to the Absolute. Such a congruence will contain cc rays pseudoparallel to a given ray. e. gruence. We may tabulate our results as follows. a congruence of normals to a surface of Gaussian curvature zero (30) will give an isotropic conwhile (32) will give a pseudoisotropic congruence. while (34) give those for pseudoparallelism. let are two points of the hyperbolic Absolute. . namely. Two rays will be parallel if (lX) (34) z{ = z v us now assume that . but we shall not enter into such questions here. or isotropic congruence. they give a ray from dX) to ( r X). will be that two rays shall have the same value for one (u) and for one (z). Equations (33) will give the conditions for parataxy. X) (. they are either parallel or pseudoThe conditions for parallelism or pseudoparallelism parallel. in fact. Let us. If two rays meet the same two generators of one set they are paratactic. but only a finite number The conditions for pseudoparallelism will parallel to it. taken in order. * The Author's attention was first called to this remarkable correspondence by Professor Study in a letter in the summer of 1905. If they meet the same two generators of different sets. Equations (29) will give a congruence whose rays can be assembled into surfaces with paratactic generators. in his second memoir. while z l z 2 in like manner determine the right generators. * Liniengeometrie. is 1 given by equations (29). i.* .234 DIFFERENTIAL LINEGEOMETRY is CH. without but in detail. xv. which will make the moving ray tangent '1^2 When. We might push the matter still further by distinguishing between syntaxy and antitaxy. to the Absolute. e.' Jahresbericht der deutschen Mathematikervereinigung. u 2 is fixed. one of the left generators of the Absolute met by the ray in question is fixed. therefore. assume that the subscripts are assigned to the letters u^u^ z^2 in such a way that a direct conformal transformation. thus be On given by the other hand a pseudoisotropic congruence will be (30). their lines are. i.
or pencil of tangents to Absolute. Imaginary pseudoparallelism. Real congruence of normals to a surface of Gaussian curvature zero. Real pseudoisotropic con conpseudoisotropic gruence.parallelism. Imaginary parallelism. Real ray. 235 Ray. Real congruence of normals to a surface of Gaussian cur Real vature zero. Elliptic Space. Real isotropic congruence. Real parataxy. Ray. . Real ray in actual domain.XVI DIFFERENTIAL LINEGEOMETRY Hyperbolic Space. Imaginary parataxy. Real parallelism. Imaginary pseudo. Real isotropic congruence.
Our spaces so defined were not.CHAPTEK XVII MULTIPLY CONNECTED SPACES we laid down a system of axioms fundamental objects points and distances. we might build up the geometry of a restricted We also saw that with the addition of an assumption region. To reach such continua it was necessary to assume that any chosen segment might be extended beyond either extremity by a chosen amount. Our next task was to show that under Axioms I'V each point will surely have one set of homogeneous coordinates (#). . . concerning the sum of the angles of a single triangle. For this new type of space we set up our Axioms I'VI'. One further point was established in connexion with these developments to each point there will correspond but a single set of homogeneous coordinates (x). IN Chapters I and II for our or euclidean geometry of the restricted region in question. we found it necessary to distinguish between elliptic space where but one point goes with each coordinate set. thereby. saw in the beginning of Chapter VII that this assumption. at most under the elliptic hypothesis. and conversely. on the contrary. The difficulty was overcome by assuming the existence of a space which contained as subregions (called c. we were in a position to develop fully the elliptic.onsisteut reyious) spaces where our previous axioms held good. and in euclidean space x ^ 0. > 0./ Hin elliptic space (.cx) 2 3 < 0. and showed how. though allowable in the euclidean and hyperbolic cases. hyperbolic. even in the real domain. perfect analytic continua. Under the there will surely correspond one real point. euclidean or hyperbolic hypotheses each set of real coordinates can correspond to one real point. to each set of real coordinates subject to the restriction that in hyperbolic space IPx* + x^ f x. will involve a contradiction when added to the assumptions We already made for elliptic space. and the spherical case where two equivalent points necessarily have the same coordinates. The proof of this depended . however.
assumption. let this region be connected with the given point by two different sets of overthen (x) and (x') shall be two lapping consistent regions different sets of coordinate values for this point. Van Vlock. is in close accord with Killing. if these equations hold for any point. and give two sets of coordinate values for every point of the region. published in Lectures on Mathematics. obtained by two different sets of analytic extension of the original coor upon Axiom tion . which required that a congruent transformaof one consistent region should produce one definite Of course such an transformation of space as a whole. 1905. a statement which will always hold true when there is a single point so reached. Die Orundlagen der Paderborn. and come back to the region in which we started. In the present chapter we shall set ourselves the task of examining whether. and then make successive analytic extensions for the change of axes from one to another of the overlapping consistent regions. one point we shall have 0. or all but a finite number of points of the region. Woods. Let us first assume that there is a consistent region which is reached by each chain of overlapping consistent regions. one point of the region have different values for its coordinates from what it had at the start. and White. xvii MULTIPLY CONNECTED SPACES 237 VI'.OH. in some consistent region. then. New York. the same will be true of all.3 Conversely. (x ) belong Let a coordinate system be set up. What will be the meaning of the statement that under our f net of axioms two sets of coordinate values (#). dinate system. * The present chapter Geometric.. and the new coordinate values will be obtained from the old ones (in the noneuclidean cases) by means of an orthogonal If (x) and (x} be two sets of coordinates for substitution. under Axioms I'V of Chapter VII. until we have run through the whole circuit. If. when applied to our space of experience. as to the same point? in Chapter V. 1898. it be possible to have a space where each point shall correspond to several sets of coordinate values.* For simplicity we shall assume that no two different points can have the same coordinates. Part iv. Another account will be found in Woods' 'Forms of NonEuclideanSpace'. can neither be proved nor disproved empirically. We may set up a coordinate system in this region. they will represent an identical transformation of the region. .
called singular points. In speaking of spaces which obey Axioms I'V. We may connect P with a chosen point A of the original region by two continuous curves. If (x) and (. thus making. These will be connected with the second set by a relation 0. We shall. for then we should have two different points of the same . Suppose that we have a third set of coordinate values for a point of our consistent region. exclude this possibility by sticking closely to our axioms.. There is one possible variation in our axioms which should be mentioned at this point. Let us suppose that we have two overlapping systems of consistent regions going from the one wherein our coordinate axes were set up to a chosen point P. We by analytic extension that these equations will every point in space. and there are special points. It is entirely possible to build up a geometrical system where IV holds in general only. in all. This shows that.') be two sets of coordinates for the same point cos 1 . PJ be a point which will have two different sets of coordinate values. according as we arrive at it by the one or the other set of extensions.238 MULTIPLY CONNECTED SPACES see also sets of coordinate values for CH. our space is multiply connected. we shall use the term 'multiply connected spaces. we see that our loop is of a sort which cannot be reduced in size beyond a definite amount without losing its characteristic property.3 give two t see that (x") and (x) are also connected of this type. in the case of the geometry of the euclidean cone with a singular line. however. the expression (xx')  cannot sink below a definite minimum value greater than zero. a continuous loop. The assemblage of all coordinate transformations which represent the identical transformation of a multiply connected space form a group. in the sense of analysis situs. but where each point can have several sets of coordinate values. which can lie in two consistent regions which have no subregion In two dimensions we have a simple example in common. hence We by a relation Theorem 1. If now.
We shall begin by asking what groups of congruent transformations of the euclidean plane fulfil the requirements of Theorem 2. The points of 8 may be put into one to one correspondence with those of a fundamental region of this sort or of a portion thereof. for in 8 we are interested in actual points only. while n is an integer. and. The group of identical transformations of 8 will appear in 2 as a group of congruent transformations. may be taken as the group of identical transformations of a multiply connected space whose points may be put into one to one correspondence with the points of a portion of any fundamental domain of the given space for that group. Let us further define as fundamental such a region of 2. We Theorem 2. yet no two points of a fundamental region are equivalent to one another. regions will be strips bounded by lines . that every point of 2 has an equivalent in this region under the congruent subgroup which we are now considering. this group may be expressed in the form x'~x + nl The fundamental t 2/=2/. nor produce an infinitesimal transformation of that domain. conversely. and giving to each point one set of coordinate values only. For the sake of clearness in our subsequent work let us introduce. positive or negative. which we have seen is impossible (Chapter VII). Every real group of congruent transformations of euclidean. Our interest will. Every congruent transformation of the euclidean plane is either a translation or a rotation. a space 2. hyperbolic. nor transport such a point an infinitesimal amount. a group which has the property that none of its transformations can leave a real point of the actual domain invariant. What then are the groups of translations of the euclidean plane ? The simplest is evidently composed of the repetitions of a single translation.xvn MULTIPLY CONNECTED SPACES 239 consistent region with the same coordinate values. besides our multiply connected space $. which carries the actual domain into itself. such a fundamental region will furnish an example of a multiply connected space obeying Axioms I'V. lay stress upon the actual domain of 2. and none of whose members leave an actual point invariant. centre in the space 2. If the amplitude of the translation be I. but the latter type is inadmissible for our present purpose. shall also find it advisable to treat the euclidean and the We two noneuclidean cases separately. or elliptic space. from now on. having the same value for the constant k as our space S.
any point in the finite domain. unless we are under our previous first case). except the vertices which are not common. their amplitudes were commensurable. for every point within such a parallelogram is nearer to one vortex than any two vertices are to one another. or a finite number of such. CH. including two adjacent sides. 2/=7/fmA. If the amplitudes of the two be I and A. nearer to it than any other. excepting two extremities. and we are back on the second previous case. A corresponding space 8 will be furnished by a euclidean each strip including one of the bounding cylinder of circumference I. This parallelogram. The points equivalent to it under the congruent group in question may not cluster anywhere. If these nearest equivalents do not all lie on a line with P. the group compounded from repetitions of two nonparallel translations will suit our purpose very well. but if the amplitudes were incommensurable. regions are parallelograms. noncollinear with P. we should fall back upon the preceding system. The Clifford and surface discussed in Chapters offers an excellent example of a multiply connected surface of this type. that is any point of this plane. and n are integers. Suppose. in fact. hence there is one equivalent. For if. that is to say. hence. we may write our group in the while . in that case. What translation groups can be compounded from two given It is clear that the lines of motion of the two translations ? should not be parallel. m form tf=x + nl. two only in number).240 MULTIPLY CONNECTED SPACES y axis. the group would contain infinitesimal transformations and these we must exclude. It is interesting to notice that with these two examples we exhaust the possibilities of the euclidean plane. or three independent transThe fundamental regions will be respectively layers lations. parallel to the lines. two. will constitute a fundamental region. In a threedimensional euclidean space we shall find suitable groups compounded of one. On the other hand. we may pick out two of them. and thus construct a parallelogram within which there is no equivalent to P. each including two adjacent sides. thus determining onehalf of a fundamental parallelogram. If the nearest equivalents are collinear with (and. . we may pick out one of them and one of the next nearest (which will be oft* that line. Let the reader notice an exactly similar line of reasoning will show that there cannot exist any single valued continuous function of the complex variable which possesses more than The fundamental X XV P P two independent periods.
the line in 8 will still be a loop. in the form x' = x cos nO y sin nd. lines. but of greater length. bounded by In S we shall have various types of straight axis will be a simple closed loop of length d. Let us glance for a moment at the various forms of straight line which will exist in a multiply connected euclidean space which corresponds to a euclidean parallelepiped in 2. the line in 2 do not contain any other point equivalent to the vertex. will be layers z'= z + nd. but. we see that every parallel to the Z axis will go into a closed lino of the typo required. nd = The last of these equations COOLIDGK T cos y. /S'. lastly. there being an infinite number of points of each at the same distance from that axis. The fundamental regions in 2 parallel planes. There are other groups of motions of euclidean space. It is 241 between parallel planes. if followed sufficiently far. n being an integer. vertex with any other equivalent point. the Z axis is the only closed line. = z f T cos y. which fulfil the requirements. y' = x&mnO + yeosnO. fourfaced prismatic and easy to determine how much of the parallelepipeds. bounding surface should be included in each case. x + r cos a? x sin nO + y cos n6 = ?/ + rcos/3.xvn MULTIPLY CONNECTED SPACES spaces. It is also evident that there can be no other groups composed of translations only. will pass again as close as desired to the chosen point. [ The necessary and sufficient condition that they should be equivalent is x cos n Q __ y s n n Q _. Will there be any other closed lines in $ 1 The corresponding lines in 2 must be parallel to the axis. The corresponding lines in 2 shall all pass through one vertex If the line in 2 be one of the fundamental parallelepiped. This may be expressed. besides translations which give rise to multiply connected An obvious example is furnished by the repetitions spaces. If. of a single screw motion. When 6 and 2 IT are commensurable. the line in 8 will be a simple loop edge If the line in 2 connect the of length equal to one period. when and 2?r are incommensurable. of the parallelepiped. Let us now take two points of 2 separated by a distance r The Z = f # frees a. the line in S will be open. shows that a line in 2 per Q .
will pass an infinite y . /3. coupled with the fact that the sum of the squares of the direction cosines is unity through each point in $. We number of straight lines. cos a = cos ft = .242 MULTIPLY CONNECTED SPACES OH. and possessing double points. parallel to a line meeting perpendicularly) cannot return to itself. r may be determined from the given equations. we get a bundle of loop lines in /S with direction m y cosines cos a  _ * _^=__ =_. Such lines will. since the determinant of the coefficients will not. n be given. having this The planes in & will be of three as a double point. x. On the other hand. y. /3. Those which are perpendicular to the Z axis will contain open lines only.e. /3.=^=_ . those whose equations lack the Z term will contain till sorts of lines. will form an infinite discontinuous assemblage. 2x + 'ftia _____( _ . will be triangular right axis will appear in $ To find lines which cross Z themselves.1) + lib. y.____ . nO = 2 7/177. however. in general. and each integral value and 7i. If. ma 22 2 When number I is odd. We see. vanish. we shall have through each point an infinite of lines which have a double point there. /. let us write x f r cos a y Z + r cos j3 f T COS y = l) x + ma = . y. c. thus see that in S the lines with direction angles a. ti be given. = + l ( . Other planes will contain no lines which are simple loops. If a. The fundamental regions Lines in 2 parallel prisms. in general. of For each even integral value of I. &c. in 2 to the as simple closed loops of length 20. y from the two preceding. not on the Z axis. that . be open. n being integers. r may be determined by the last equation. . Another type of multiply connected space will be deter mined by x'=~l m. the direction cosines being cos a =  __ _:^_^ . y 0. if and we have a closed loop of the type just discussed. it pendicular to the Z axis (i. a. 7 ( // '^ If. sorts. on the other hand.2 </( z . and x.2x + ma)* + .
been found.supposed to define a point of the absolute conic. The path curves are horocycles touching the absolute conic at the fixed point having in fact. the &bh part of their distance will be (m 2 f>)). for the equation of a horocycle tangent will be This will intersect the two lines # &#! = (). while the absolute conic has an equation of the form : xf + x^ The general type at (0. but such is not the case. ideal. worked out. (P + p)} (m. this & happen ~ point reached again after a distance C. and its pole. form of a lernniscate. The cosine of * loc. The last is not. Giundlagen.* When we turn from the euclidean to the hyperbolic The hypothesis. 1. 0. therefore. It is merely necessary to show that a horocycle of the family may be found which cuts two lines through the fixed point in two points Let this fixed point be (0. These and the preceding example are taken from Killing. if the point ^& is . The two fixed points must be real. we find a less satisfactory state of affairs. 1. that in such a substitution there will be points of the plane which are transformed by as small a distance as wo please. as yet. It might seem. fourpoint contact with it. we wish for groups of binary linear substitutions which contain members of the hyperbolic type exclusively. ij 243 plane whereas the length of a loop perpendicular to the is 2c. cit.xvii MULTIPLY CONNECTED SPACES x. in order that the line joining them shall be actual. in fact. the fixed point. This loop has. to be on such a loop. would also answer. In other words. 1) = 0. . that parabolic transformations where the two fixed points of the conic fall together. We may show. Apparently such groups have not. 0. 1) as near together as we please. in the points (Z. the general real congruent group of the hyperbolic plane was shown in Chapter VIII to depend upon the real binary group the homogeneous coordinates (t) being . at first. x m&'j == 0. however.
vol. ideal lines. seo Fricke.244 MULTIPLY CONNECTED SPACES CH. depend upon the linear function of the P complex variable a? f # The group which we require must not contain rotations about a line tangent to the Absolute. i. = x sinh &3 cosh 0. Leipzig. limit. as  approaches The group of hyperbolic motions in three dimensions will. we may not have rotations about actual lines. in like manner. There does not seem motions. 1807.* The group of repetitions of a single rotation about an ideal line may be put into the form (IcP = 1). * For the general theory of discontinuous groups of linear substitutions. trajectories of planes through this line will line in E be equidistant curves whose centres lie thereon. to be any general theory of groups of linear transformations of the complex variable. '= XQ sinh nO 4. as we saw in Chapter VIII. = x. and screw which give loxodromic ones. allowable motions of hyperbolic space are rotations about which give hyperbolic substitutions. . the substitutions may not be elliptic. an expression which will approach unity as a zero. hence the complex substitution must not be parabolic. write the group of repetitions of a single screw motion The orthogonal A &/= B 2 tf3 'j\ L cos n (f> x 2 sin n $. hence the path curves on the Absolute may not be conies in planes through an ideal line (the absolute polar of the axis The only of rotation) . which include merely hyperbolic and loxodromic members only. connecting two points which are equivalent under the group will appear in $ as a lino crossing itself once. We may. for the reason which we have just seen. '= X sin n<f) + x. t = 0.^ The fundamental regions in 2 will be bounded by f pairs of planes through the line a. x '= XQ cosh n XA sin n 0.2 cos n<f>.Klein. x.#3 cosh n0. Again. Vorksungen uber die Thcoric dcr automorphen Funktionen.
(x'] i. of length  Notice the close analogy of this case to the simplest case in cuclidean space. for otherwise the group would contain rotations. (x). p. (. for our If A \L we have = transformation may be written in the quaternion form * : (ATT of rotation. and consider the group of repetitions The angle of rotation about one of a single screw motion.   > v f U7T 4 x cos n U7T ? V where It will A. us assume that k = 1. Hence. sm n cos 7i . cit. .o? 3 A"). there are no twodimensional multiply connected In three dimensions the case is different.or^. these two fractions must have the same denominator. . there being no ideal points. 68. p. erroneously states that these translations are the only motions along one fixed line yielding a group of the desired The mistake is corrected by Woods. . cos n h sm n . and the two distances or angles of rotation must be of the form v > v T in order that there shall be no infinitesimal transformations in the group. IJLTT v V x.rft sm'H ATT > v . 342. the identical transformation.r 4. loc. transformation of the real elliptic plane is Every congruent a rotation about an actual point. n ATT %i ATT fa?.xvn In MULTIPLY CONNECTED SPACES elliptic space 245 we obtain rather more satisfactory results. XQ . axis is equal to the distance of translation along the other.T 1 = . . Chapter VIII).r cos . Grundlagen. so that there are no we have real fixed pointy nor points moved an infinitesimal distance.. * Killing.e. ATT A i\ . n a variable integer. a translation (cf. type. ATT 9 . . 4 x$j 4. v are constant integers. /m. Let elliptic spaces. . The pathcurves in 2 will be lines paratactic to either axis and they will appear in 8 as simple closed loops . cit. v x9 x x 9 cos n # sin n . sm n HIT . Moreover.. We may therefore write the general equations . and be found that the cosine of the distance of the points will be equal to unity only when n is divisible by v.
the inner reason for thi. loc.* If a multiply connected elliptic space be transformed identically by a group of translations. In particular. x. whence Theorem 3. rise to : if we put Jr +/.(c + di) A . that group is isomorphic with one of the groups of the regular solids.^\^ (XQ then the translation (jcj f a?/ i + a'2'j + s^k) A t '= A2 ' == (a + bi + cj + dk) + xi 4. and a real point of a euclidean sphere is given by the value of its coordinate as a point of the Gauss sphere. The line Of course problem of finding elliptic translations. . depend therefore. are therefore identical with those of euclidean rotations about a fixed point which contain no infinitesimal members.246 MULTIPLY CONNECTED SPACES OH. merely on the problem of finding linear transformations of the complex variable which transport diametral values into diametral values.ix.xj + #3 &). while diametrically opposite points will be given by diametral values of the complex variable. The groups of elliptic translations which contain no iniinitesimal transformations. Each linear transformation of these will determine a translation. = (c ( 2 . 08.e. i. or euclidean rotations.rfi) A! + (a . Now this is precisely the formula for the rotation of the euclidean sphere.r 1 = \ 1 . p. . The cosine of the distance traversed by the point (x) will be which becomes equal to unity only when 6 = c = rf = 0. Conversely each group of the regular solids gives rise to a group of right or left elliptic translations. the generators of one set on the Absolute. * Cf. xvn There is another translation group of elliptic space giving a multiply connected space of a simple and interesting Let A a A 2 be homogeneous parameters.s identity is that a real meets the elliptic Absolute in conjugate imaginary points. locating description. suitable to define a multiply connected space of elliptic type. corresponding to diametral imaginary values of the parameter for either set of generators. Woods.. cit. when we have the identical transformation. may also be written + hi) A! .bi) A.
sub class called a The points two shall also be said to be on the line.' American Journal of Mafhcttiatics. Vol. Abstrakte Geometric. rests in the last analysis. If the given points be * A and . ' system of axioms for projective geometry. made in the cuclidcan case where the congruent group was a sixparameter subgroup of the sevenparameter group which left a conic in place. Each pair of distinct points belongs to a single line. namely. successors. Parts II and III. or Vahlen. The A 1908. starting once more at the very beginning. as developed in Chapter VII and subsequently. containing at two distinct members. which left invariant a An exception must be certain quadric called the Absolute. and that noneuclidean metrical geometry may be built up by positing the Absolute. xxx. points have thus ruled out the possibility of building up spherical geometry upon the present basis.#. Lezioni digeometria proiettiva. AXIOM II. A point common to to pass through the points. shall be called their intersection. in his Ptincipii detta geometria di posisione. We thus come naturally to the idea that a basis for our whole edifice may be found in project! ve geometry. Two distinct points determine among the remaining points of their line two mutually exclusive subclasses..CHAPTER XVIII THE PROJECTIVE BASIS OF NONEUCLIDEAN GEOMETRY OUR noneuelidean system of metrics. upon a projective concept. the line lines II are identical. cit. The group of congruent transformations appeared in Chapter VII as a sixparameter collineation group. and defining distance as in Chapter VII. . Bologna. It is the object of the present chapter to show precisely how this may be done. 1898. neither of which is empty. It is evident from lines two Axiom that with two common We AXIOM III. as Enriques. Voblen and Young. the cross ratio. two points belonging He to was first writer to set up a suitable set of axioms for projective geometry has had many Fieri. called points. cit. There exists a class of objects.* least AXIOM I.
i. If AC\BD j and AE^CD. we shall write PQ^AB. then then A and B. though on a line. two belonging III shall be said to be to the same class not separated* We shall call such classes separation classes. a chosen pair them will either separate two of the pairs formed by the other three or none of them. and hence AC^DE. Again. If and are separated P Q be separated by by A and 2?. B. U u J For C and D determine but two separation classes on line. if EC^AD] hence BE^AD. AE^JBC. If and Q be separated and Q.248 THE PROJECTIVE BASIS OF Axiom CH. Let the five points be A. if EC f 7)7?. If five collinear points be given. . But we have * AE \CD J hence AE \BD J a contradiction with caratteristiche delle varieta a The axioms of separation were first given by Vailati. BC^DE.e. of Theorem 2. Let AC \DE. j To begin with BC&AD. 'Sulle proprictk una dimensione. J we had same J BC^DE j AB^DE. AB f CD and AE CD. J If PQ be not separated by A and B. and both B and E belong to that class which does Theorem 1. 1895. But if Then. \j then AE\BD. are separated by P P by or A and J?. E. AXIOM Q and P IV. v. [ the not include A. is AXIOM VI. we y should have AB^EC. AXIOM A and B V.' Rivista di Matematica. different classes according to separated by them. We shall write this relation PQ Jf AB 4B fpQ. j ABC would belong to the separation class with regard to DE. <j Theorem 3. then EB^CD. we had AB J ED. ^ AB^DE. If four distinct collinear points be given there a single way in which they may be divided into two mutually separating pairs. J and and if AB \DE. D. or collinear. with them. C.
there will exist a single point C of this separation class of such a nature that no point of the first subclass is separated from A by B and C. If all points of either separation class determined by two points A. AXIOM VIII. This axiom is one of continuity. and say that a point is between two others when it be separated thereby from the chosen point. then PQ f C7). It is clear that C may be reckoned as belonging to either subclass. have BE^AD and AB^ED. . the theorem is immediate. Points or lines in the same plane shall be called coplaixcr. Let the plane be determined by the point A and the line If the two given points of the given line belong to BC or be A and a point of BC. be clear that this theorem includes as a special case We have but to take A at a great I. let the reader make a careful comparison with XVIII of Chapter II. we must AE\BD.classes that no point of the first is separated from A by B and a point of the second. intersects the third line. If not. PQ ^AB. All points do not belong to one line. Theorem 4 If PA fcD. and none of the second is separated from B by A and C. The proof is It will We AXIOM VII. The assemblage of all points of all lines determined by a given point and all points of a line not containing the first shall be called a plane. AXIOM IX. have but to choose one point of the line. A BG. since 249 AB\ED. plane will contain completely every line contains two points. but that no other point enjoys this property. follow from the fact that neither of our separation classes is empty that the assemblage of all points of a line is infinite and dense. Definition. left to the reader. A line intersecting in distinct points two of the three lines determined by three noncollinear points. XVI Theorem whereof it 5. It will Theorem 3 of Chapter distance. PB fcD.xvni NONEUCLIDEAN GEOMETRY As a result. be so divided into two sub. Let the reader compare this with the weaker Axiom of Chapter I. B.
then the planes determined by A'E'C' and ABC are identical.6'. C. AXIOM X. The assemblage of all points of all lines which arc determined by a chosen point. B C be three noncollinear points. respectively. B and C be two points of the one line. B\ 0'. Then BP will intersect AC. by C and are identical. Let us assume that $'. AB AC Let P AP Theorem 6. C' be three noncollinear points of the plane determined by ABC. respectively.250 let the THE PROJECTIVE BASIS OF line contain cm. of the other. If we may A be also a point of BG the theorem is proved. C'. Definition. D' be four noncoplanar points of the space determined by A. leave to the reader the proofs of the following very simple theorems. A space contains completely every line whereof contains two points. D lie in that determined by D. A. B. and by the planes determined by A and J?(7. Theorem Let If not. All points do not lie in one plane. and A a point 8. then and CA. B'. D' are points of AB. hence will intersect BC or will lie in the given plane. 11. then the two spaces determined by the two sets of four points are identical. '. With regard to the last theorem it is clear that all points of f the space determined by A'. (7. B. C'. A plane Theorem plane plane BCD is CD A. The planes BCD and B'C/D' have a common AD . If A' . the points B' and C' of and be any other point of the given line. If A'. y B AB We lie have but to notice that the wholly in each of the others. Theorem whereof it 10. If A. space contains completely every contains three noncollinear points. identical with that determined The space determined by a point by A B and the and the Theorem 12. and all points of a plane first point shall be called a space. D. Two lines in the same plane always intersect. use the point A and the line BC to determine must be identical with a line the plane. lines generating each plane Theorem 7. This will come immediately from the two preceding. AC. not containing the We it Theorem 9. and our second line through A meeting EC.
of SO and B'G'. This is The following the usual proof. then the intersections of and A'B'. let us suppose that the The lines AA'. To begin with. so that B will lie in both. BB\ planes ABC and A'B'C' are distinct. Theorem. of CA AB and is C' A' are collinear. the intersections must be on L Considering the lines AA'. . which naturally belongs to both spaces. so that R"C' meets and B'G' in a point of I. and VC will meet V'V in C". if the lastnamed three pairs of lines meet and conversely. BB\ 0(7 be concurrent. and similarly for C"A". All points. Theorem line. Furthermore A B will meet A"B' in a point of both spaces. Two planes in the same space have a common Theorem line or a 14. common Three planes in the same space have a point. V V . on the other hand. they intersect in a point which must lie 011 the line I of intersection of the two planes ABO and A'B'C' and a similar remark applies to the intersections of BC and B'G\ of CA and G' A' Conversely. by similar reasoning. when these lastnamed three pairs of lines intersect. 13. we see that each two are coplanar. Then A"R meets both A E f r and Z. BC . as do the line A' A" and the point A the two spaces are identical. and will meet A'J3' on L Conversely. lines. V AB in points of a line I in their plane. and must intersect. BB' and CO'. we may find A"Bf'C" r noncollinear points in another plane through I. In the same way CA will meet G' A' on I. C and D . Then the plane BCD lies in both spaces. hence. The planes ABC and A"R"G" will meet in a line I.15. Then as and A'B' arc coplanar. and. and J3"<7" will meet both BO and B f C/ in a point of /. theorem of two triangles. Then VA will meet A' in VB will meet ff in JT. If. If three lines A A'. and planes hereinafter considered are supposed to belong to one space. Let V and two points without this plane collinear with the point of f concurrence of AA BB. or lies in each. assume that AA' does not intersect this line. common Practical limitation. but all three are not coplanar. Let A" be the intersection of AA' with BCD. has two points in each space.xviii line NONEUCLIDEAN GEOMETRY 251 Let us first I. Desargues' AB . f lie in both. A". and CC' will be concurrent in outside of both planes. The second case occurs where A f B'C f are three noncol linear be points of the plane determined by ABC. Hence the three are concurrent. '. . CO'. A A' meet I in A" then A lies in both spaces.
no three of which are Two sides which do not contain a common vertex be said to be opposite. A given point has a unique harmonic conjugate with regard to any two points collinear with it. Their six intersections shall be called the vertices two vertices being said to be opposite when they are not on the same side. be given.252 THE PROJECTIVE BASIS OF f CH. the figure formed by the three pairs of lines determined by them is called a complete quadrangle. each Theorem 18. after drawing two or three lines. C'C" concurrent in V. have already remarked in Chapter VI on the dependence of this theorem for the plane either on the assumption of the existence of a third dimension. B Definition. If two complete quadrangles be so situated that five sides of one meet five sides of the other in points of a line. CO'. we leave the details to the reader. . Lastly. and A'A" B'B". If plete quadrilateral. the sixth side of the first meets the sixth side of the second in a point of that line. the sides. If A and C be harmonically separated by and D. formed by four coplanar lines. and C be two opposite vertices of a comDefinition. BB'. is . This is an immediate result of 16. If A B and and C. then B and and C be harmonically separated by D are harmonically separated by A The proof will come immediately from 15. AB. shall collinear. figure The are concurrent. the three coaxal planes VV'A". Then by the converse of our theorem AA"\ BB". then A B A E to be harmonically separated by C and /). or of a congruent group. . Theorem 17. The original points are called the vertices. said to be the harmonic conjugate of the other with regard to these two points the four points may also be said to form a harmonic set. C of the f A and for first 9 ABC A We Definition. The three lines which connect opposite pairs of vertices shall be called diagonals. the pairs of lines Theorem 16. VV'C" will in three concurrent lines meet the plane A'. D. no three of which shall be called a complete quadrilateral. The intersections of pairs of opposite sides shall be called diagonal points. CA. VV'B". A'B'. A" B" . while the diagonal which connects them meets the other two in and shall be said and 7). If four coplanar points. CC" will be part concurrent in F.
If a point be connected collincar with it by lines OA. & y. If a. 6 pass respectively where A and C are harmonically through four points A. and these lines meet another line in A\ C". by means of concurrent lines. then we may speak of a and y as separated by harmonically separated by ft and 8. through to A B y 0. . harmonic conjugates with regard to ft and <5. B> C. or say that the four lines form a harmonic set. if to B and J5. tively in A. lastly. We may also speak of a and c as harmonic conjugates with regard to 6 and d.xvin Theorem A. C'. f f Definition. B f A assume that the quadrilateral conwhich yielded A. may properly be said be harmonically separated by 6 and d. AG{BD then A'c f \Biy. and D. if C'. understand by projection the transformation (recently used) whereby coplanar points and lines are carried. Definition. B. C'. D was in a plane which did not contain 0. into other coplanar points and lines. and (7. /3. shall We Theorem 21. y. with four points OB. B D respectively. B. c. d be four concurrent lines which pass and C be harmonically Theorem 20. G. 0(7. It is sufficient to draw the line AD and apply 19. then A' and C' are harmonic conjugates with f and D'. Then radiating lines through struction will B We may legitimately transfer this quadrilateral construction into another giving A'. or /3 and 5 as harmonically shall also say that a and y are separated by a and y. Any finite number of projections and intersections will carry a harmonic set into a harmonic set. we have the theorem. AXIOM XI. and 6 and d harmonically separated by a and c. (7. D'. OD. B'. B. If four coaxal planes a. D y '. regard to . and if A and then a and c 6. D meet another line in four points A'. 6 determined by a line I and four points A. and if A and C be harmonically separated by B and D. B\ . then A and C' are harmonically separated by J3' and /)'. if If four coaxal planes meet two lines respecand A' B\ C' D' distinct points. for this construction may be effected in any plane which contains AD. With this in mind. separated by /). and C be harmonic conjugates with regard and. NONEUCLIDEAN GEOMETRY 253 D not 19. If four planes a. D' respectively. 7)' respectively. B. or that the four D B We planes form a harmonic set.
IV. VI may be shown II is lacking in the geometry to be serially independent. Besides being independent. A.* of four points. B. cit. we shall say that the planes If the planes a A and 1C separate the /3 planes IB and ID. and y separate the planes plane meet the four in a.. They will be satisfied by any class of objects in one to one correspon1 Hence our theorem. V. AC f BD j and / be any line not intersecting I AD. VIII fails in the geometry of the single line. Ill. If J. and if a fifth Theorem in 22. so We D D that if we had AB CD j f we should also have CB f AD. then we shall say that a and c separate b and d. b. to C\ B.. must be assumed as an axiom. then AC BD. The laws of separation laid down for points Axioms IIIVII hold equally for coplaiiar concurrent lines. let us glance for a moment at the question of the independence of our axioms. by any Theorem 24. (7. our 15. loc. number The relation of separation is unaltered of projections and intersections. C. D be four collinear points. d respectively. and coaxal planes. and 8. j and geometry where XI is lacking. * Vailati. Theorem finite 23. while VII is untrue in the system of all points with rational Cartesian coordinates. The author is not familiar with any system of projective vice versa. A complete justification for this terminology will be found in Axiom XI and in the two theorems which now follow. B.254 THE PKOJECTIVE BASIS OF If OH. have merely to observe that our quadrilateral construction for harmonic separation permits us to pass by two projections from A. X naturally fails in plane geometry. We have merely to bring the four lines or planes to intersect another line in distinct points. and Desargues' theorem. c. Here IX must be suitably modified. IX is lacking in the projective euclidean geometry where the ideal plane is excluded. note quoting Padoa. and apply XI. Definition. respectively. our axioms possess the far more important characteristic of being consistent. and A and C be harmonically separated by B and D. Before proceeding further. Definition. .
A line may be defined as the assemblage of all objects whose coordinates are linearly dependent on those of two./' : JLI not the case. in general for the case We P also . and call it JF>. and 7^. Theorem "6. in general.. n+ Ap j n /^ Pn+l6 ^P. collinear with 7^. ~ the generic Theorem P^P \P n+l u if 11 > //. . and call it P 15 and so. while # and D have the coordinates A (x) + (y) and \'(x) + pf (?/). separated As a next step in our development of the science of project! ve geometry. into two subclasses as follows. they shall be said to be not and D. . Suppose that we have three distinct collinear points 7^ 7fJ5 I\. Suppose that n ^o^J^i^Q n n n+l $P P clearly P F and. The conthis is When by B . let us take up the concept of cross ratio. then such a positive integer /* may be found Let us divide all by 7^7^ which include P and P the positive separation 1 let us say.xvni NONEUCLIDEAN GEOMETRY : : 255 deuce with all sets of real homogeneous coordinate values x x. A similar proof may be found where negative subscripts are involved. . If P be any point which satisfies the condition 7<J P LfJ JP. If A and have the coordinates (x) any (y) respectively. 0. and 74. will always intersect on l ly name for such a point being 25. Q n+l :Y /J. Construct the harmonic conjugate of jP with regard to P. The theorem certainly holds when that = 1.2 # 3 not all simultaneously zero. P P P struction is very rapidly performed as follows. construct n+1 and n^ harmonic conjugates with regard to n and 7^. A point A be assigned to the * points of the separation class determined class shall find such a positive first class if we may See Fig. then A and C shall be said to be separated by B and D if . know P ^Pn n We r notice also ^P P^ that f P Q Hence. 4 on page following. while our given points lie Let l l be the line from the intersection of l OP Then OP +1 n and VPn Take and V on the line 1 and VP to 7^ Q . that of 7^ with regard to 7^ and 7^. that of I\ with regard to P. and call it P.
for a point separated from ]^ by have to belong to both classes. Let us OD meets ^ in T) and VD meets l in (7. and a point of division D. . so that or G is a point of the first subclass. y {} We know D are that Hence lines from . Qt Our net either result is P P P^P^ in exactly the that if P be any point of the it is a point of the system we have constructed. then. as y long as A and B are distinct we shall have P B\AP^ that i giving a dichotomy of the sort demanded by further assume Axiom VII. find n so great that P PU Q \C2^ . however. otherwise it shall be assigned to the second class. . by Q 3..e. hence Q^^ADP^ w JDP . Our theorem this contradiction. D from We line l might treat the case where same way. and every positive integral value of n+ ^7? \P iP^^ Then.. are not separated by those to lines from 7) to 7^ and J^. P {} to V and not and P^ Hence separated by those to and V.. We may.256 THE PKOJECTIVE BASIS OF n that CH. is But P Q P^DP^ hence PQ P n+l This. lj n+1 would results and absurd. for every point of the second class 7?. and P^+JW. integer P$P +i \AP^> n i.
Let l k be the line from P^ to the intersection of OPk with VP Then I say that VP and OP This is certainly true when meet on l k m m+k k Let us assume it to be true in the case of l k ^ so 1.m n+ n+k n+h A particular result of thia will be that Pk Pk+n Pk+2 n^> f rm a harmonic set. l . Negative rational subscripts might be treated in the same way.P are projected from k the line VP m These points lie on the lines I^ m9 ^m> hm* Find the intersections of the latter with VP and project back n from on we get the points P _ m P . that VP^ and OP meet on l k _ v Then l k is constructed with k regard to l k _ l as was l with regard to Z for we take a point and find where that line meets of l k _u connect it with Our next may . V!Q. m P i. = 9 . . find the on points into which the points P^. .Pj. we get points which we may call JP. +2 so on. n+ 1 that PP n n+l PP^ . t .P^. where Pn n P^ l Connect P^ with the intersection of VP Q and OP l by a line r '71 used ^ to We may use find P We k . Jf^. . As an application of this we observe that l n meets VP n on the line OP% n hence we easily see that P and P^ are n harmonically separated by T^J and P% n Secondly. or P^ n is identical IT with ^. VP meets OP m m+k on l k which was to be . this line to find Pk ti n as formerly we are shall thus find that P^ and P2n harmonically separated by Pn w. Subdividing IT further we shall find that Pr rn is identical with P l or 7J m ~r7t n identical with 7^. care shall be to find points of the line to which we properly assign fractional subscripts.P . . Let us now draw a line from JJ to the intersection of VP Q and l n and let this meet P^ V in V Then if 7J Pk P Pa be t . on l k and In like manner VP meets OP 2 k+l on l k ^ and OJF^. . . and eventually we shall find a single point whose sub . and 7^. proved.xvin or else NONEUCLIDEAN GEOMETRY we may find 257 (calling two such successive integers an integer) n. and similarly still Jfjl w is identical with 7^. n projected from upon TjjV and then projected back from V\ upon Z . . . n We have thus found a single definite point to correspond to each positive rational subscript. . .
is effected in the system of positive rationals defining a precise irrational number. There will be no largest number in Q the lower class. if by reducing to a common denominator. We can thus find so large that . l n. in fact. Let P be any point of the positive separation class determined by P^ and P^ Then either it is a point with a rational subscript. to our scheme. similar proceeding will assign a definite subscript to A each point of the other negative separation class of I^P^. as separation is invariant under projection. This will be given by a dichotomy in the system of positive rationals. approaches . Conversely. We r Q may express this by saying that Hence. We that oo> shall also find. P.P ^{PP^. <a and corresponding thereto we may establish the points of the positive separation according to the requirement of Axiom VII. P as a limit and P w L approaches P as a limit. Finally each number of the upper is Hence a perfect dichotomy greater than each of the lower.\RP^. that wherever R may of such {] be in the positive separation class of P^P^ so great that we may find /*/ P In. It remains to take up the irrational case. however great soever n may **' I be.. we may find m so that J^P f^Z.H 7i ll> P n n W In the l is also number of the first class. and this may be assigned as a subscript to P. in fact. We know. according . ^P with a similar rule for negative numbers. or else. suppose that we have given a positive irrational number. m u _ r I We thus J n it J have a dichotomy of the positive rational number system a nature that a number of the lower class will correspond to a point separated from P^ by P and P while one of the upper class will correspond to a point separated from P by P and P^. CH. or ?&' Pm + n QO j approaches tn w/ P a nti as a limit. n approaches P^ as a limit as n' increases. and surely 7 > 'tb same way we show that there can be no smallest number in the upper class. We shall.258 script is THE PROJECTIVE BASIS OF any chosen rational number. assign a point A of this separation class to the lower subclass if we may find such a number in the lower number class that the point with the corresponding classification class of P P^ among .
0. remembering the definition of cross ratio. as a matter of fact they have but six. the subscript which should be attached to D. For we may use these three elements in each case as GO. respectively the roles of P^ l shall be called a cross ratio of the four given points. and then. Theorem 28. then every element of the first lies in the corresponding element of the second. and Four points which are indicated by the symbol (AB. G are made to play in the preceding discussion. CD). The one to one correspondence between points of a line and the real number system including oo is thus complete. distinct would thus seem to have twentyfour different cross ratios. finite that the harmonic relation is unaltered by any of projections and intersections. we may assign as subscript to the resulting point In the same way we of division the irrational in question. or four coaxal planes. This shows that all of be four collinear points. D P P . We know number We Theorem number of of all 27.xvin NONEUCLIDEAN GEOMETRY 259 otherwise a point subscript is separated from Q by P^ and If thus and B shall be assigned to the upper Hubclass. If A. If in two project! ve one. and. Cross ratios are unaltered by any finite projections and intersections. . B. and the pencil of coaxal planes shall be called fundamental onedimensional forms. by the corresponding cross ratios of the points where they meet any other line. Definition. the pencil concurrent coplanar lines. when A. P A A tively. make use of the fact that the construction of the harmonic conjugate R2 . hence. the first three are necessarily distinct. be said to Definition. may assign a definite point to any negative irrational. be any two points of the lower and upper subclasses respec. The range of all collinear points. 1. Two fundamental onedimensional forms shall \>e protective if they may be put into such a one to one correspondence that corresponding cross ratios are equal. that P P \AP^ m riJ whereas P B f Pw P^ J . 5. PB f J n the requirements of Axiom VII are fulfilled. we can find in the lower number by class so 3. whereof Definition. G. may therefore define the cross ratios of four concurrent coplanar lines.dimensional forms three elements of one lie in the corresponding elements of the other.
they may two fundamental onedimensional forms be be connected by a finite number of and intersections. This theorem known as the fundamental one of projective geometry. p. n. m. then. to least now by reducing all rational common /c. In like manner we see that set. I. Theorem projections 30.260 THE PROJECTIVE BASIS OF OH. the intersections of corresponding lines are collinear.1 are also Q . as do we .* Theorem 29. 161. see Vahlen. have already seen that in the case of integers.. next remark that the cross ratio of four points is that of their harmonic conjugates with regard to two fixed points. . But * For an interesting historical note concerning this theorem. hence. . m. This comes immediately from 27. in fact. __ V/oo *to /p p  L p n = Putting V. and. If two fundamental onedimensional forms be connected by a finite number of projections and intersections they are projective. Reverting to our previous construction for I{ we see that it is u We collinear with Vl n and Q r VQ^P. It is. therefore. n become applying a limiting process. We turn back for a moment to our cross ratio scale. If. loo. on a 1 line. in the case of numbers I. If projective. is of a point with regard to two others is unique. /. . and see that this equation is Pu P + k ^ q + q . By letting k. denominator. easy to connect them with two other projective forms whereof one contains three. irrational. . cit. we compare the triads of points VP Q VPQ 1 1) since n lines connecting corresponding points are concurrent in P^. P^ form a harmonic In general. P^ . always true. one at a time. and hence all corresponding members Let us of the other. P P P2q q .
cross ratio of four points We A (x). 0). 1). Then replace y + a by y. If three points be taken as fundamental upon a straight line. by J^ and the intersection with . (z). the line n and identical with FQ meets V1 I^ on Furthermore n n n <?! are harmonically separated by the intersections of their line with FPj and TJ^. i. P P\Theorem 31. P P number. (y). (0. n Project these four upon 1 from the intersection of shall find n and l are harmonic conjugates F/ij. (t) will then be 2/o*o (2) Mi . &c. This will add a to each subscript.XVIII NONEUCLIDEAN GEOMETRY to the intersection of 261 the line from VP l l with VP Q (or FQ X) n Hence VP_ Jt which is is. The preceding considerations will enable us to find the cross ratio of four points which do not include P^ in their . To begin with 78 __y x /3a ~0 y' Let us project our four points from V upon Zu then back upon Z from 0. (1. by construction. We P P OP with regard to l and Let the reader show that this last x relation holds equally when n is a rational fraction. and is left to the reader. when it takes any real value. shall assign to the fundamental points the coordinates (1. and. . x Four elements of a fundamental onedimendetermine six cross ratios which bear to one another the relations of the six numbers sional form 1 * 5 1 ' X~l \ ' X X 1x' X^l' The proof is perfectly straightforward. OP OP . n VQQ and . 1). any other point thereon may be located by a pair of homogeneous coordinates whose ratio is a definite cross ratio of the four points.e. . hence.
262
THE PROJECTIVE BASIS OF
OH.
project! ve transformation of the line into itself, i.e. any to point transformation which leaves cross ratios unpoint altered, will thus take the form
Any
Px './
*t/2
a 0' 4tf sr ^11^1 ^10^0
'
To
demonstrate this
we have merely to
point out that surely
points into any other three, the maximum amount of freedom any protective transformation of a fundamental onedimensional form. Let the reader show that the necessary and sufficient condition that there should be two real selfcorresponding points which separate each pair of corresponding
for
this transformation is a protective one, and that we may so dispose of our arbitrary constants as to carry any three distinct
points
is
,
!
a;
_
<
Two projective sets on the same fundamental onedimensional form whose elements correspond interchangeably, are said to form an involution. By this is meant that each element of the form has the same corresponding element
be assigned to the first or to the second set. found that the necessary and sufficient condition for an involution in the case of equation (3) will be
it
whether
It will be
a^ > 0, there will be no selfand the involution is said to be elliptic. corresponding points, Let the reader show that under these circumstances each pair
the determinant
 \
When
of the involution separates each other pair. Our next task shall be to set up a suitable coordinate system for the plane and for space. Let us take in the plane four points A, B, C\ D, no three being collinear. shall to these respectively the coordinates (1, 0, 0), (0, 1, 0), assign Let meet BG in A^ meet GA (0, 0, 1), (1, 1, 1). in Cr The intersections of AB, A l l in .Sp and CD meet of BC, 5j6\, and of CA, Ol A ly are, by 15, on a line d. Now let be any other point in the plane
We
AD
BD
AB
,
P
(ABAC, ADAP) = (P(\PC, PDPA) (BCBA, BDBP) = (PCPCv PDPB)
(CACB, CDCP)
=
(PC PC,,
PAPD)
=
(POl P0
9
PAPB) =
product of the three
is
From
to unity,
this it is clear that the
equal
and we may represent them by three numbers of the
xvn
type
r
NONEUCLIDEAN GEOMETRY
1 5
263
2
5
3?2
We may
therefore take
x :x 1 :x 2 as three
()
XQ
X}
for the point P. One coordinate will vanish for a point lying on one of the lines AB, SO, CA. Let the reader convince himself that the usual cartesian system is but a special cane of this homogeneous coordinate system where two of the four given points are ideal, and
homogeneous coordinates
x^
XQ
lines
___
The equations of the
A, B,
C
connecting two of the points
'J.
are of the form
OUn
Those which connect each of these with the point similarly x . x =
.
D
y
are
.
(z) be two points, not collinear with A, B or G, a variable point with coordinates \(y)f /x(0), the lines connecting it with A and B will meet BG and (CA)
If (y)
and
is
while
P
respectively in the points
(0,
X^ + fxCp Ay + f*s
2
2)
(A?/
+ ^0
,
0,
Av/,
+ f/.3 2 ).
It is easy to HOC that the expressions for corresponding cross ratios in these two ranges are identical, hence the ranges are The pencils which they determine at and projective. are therefore projective, and have the line selfcorrespond
A
B
AB
ing, for this will correspond to the
parameter value
\:/*
= V2fe
But it will follow immediately from 28, that if two pencils be coplanar and projective, with a selfcorresponding line, the locus of the intersection of their corresponding members Hence the locus of the point is also a line. with the coordinates A(y) + ^(s) is the line connecting (y) and (z). Conversely, it is evident that every point of the line from (y) to (z) will have coordinates linearly dependent on those
P
of (y)
and
(z).
If,
then,
we put
as equation of the line
and eliminate X
:
p,
we have

xyz
=

(ux)
= 0.
evident that such an equation will always Conversely, a line, except, of course, in the trivial case where represent the u's are all zero. Let the reader show that the coefficients
it is
264
THE PROJECTIVE BASIS OF
CH.
U} have a geometrical interpretation dual to that of the for this purpose the line which we have above coordinates ;
^
called
d
will be
found
useful.
of homogeneous coordinates may be extended with great ease to space. Suppose that we have given five be any no four being coplanar. Let points A, J3, (7, D, other point in space. may write
Our system
P
We
(ABCABD, ABOABP) =
^
,
(ACDACB, ACOACP) = ^,
2 .
(ADBADC, ADOADP) =
xi
We
shall
then be able to write also
(CDA CDS, CDO CDP)
=X
*
,
(DBA DEC, DBO DBF) =
XQ
^ XQ
>
(BCD BOA, SCO BCP)
=
X*
In other words, we may give to a point four homogeneous Two points collinear with A, B, coordinates x :x l :x z :x z will differ (or may be made to differ) in one coordinate C, or An equation of the first degree in three coordinates only. will represent a plane through one of these four points. Every line will be the intersection of two such planes, and will be represented by the combination of two linear equations one of which lacks x.i while the other lacks x>. The coor.
D
dinates of all points of a line may therefore be expressed as a linear combination of the coordinates of any two points A plane may be represented as the assemblage of thereof. all points whose coordinates are linearly dependent on those of three noncollinear points. Eliminating the variable parameters from the four equations for the coordinates of a point in a plane, we see that a plane may also be given by an \ equation of the type / n /*\ Jr H
(use)
=
0.
(5)
Conversely, the assemblage of all points whose coordinates satisfy an equation such as (5) will be of such a nature that it will contain all points of a line whereof it contains two distinct points, yet will meet a chosen line, not in it, but Let the reader show that such an assemblage must once.
be a plane. The homogeneous parameters (u) which, naturally, may not all vanish together, may be called the coordinates
xviii
NONEUCLIDEAN GEOMETRY
265
of the plane. They will have a significance dual to that of the coordinates of a point.* If we have four collinear points
(y),
(*),
%)+?(*),
A//
*'(</) +/*'(*),
one cross ratio will be
The proof will consist in finding the points where four coaxal planes through these four points meet the line
and then applying
(2).
Suppose that we have a transformation of the type
0..3
3
This shall be called a collineation. ourselves to those collineations for which
We
shall
restrict
The transformation is, clearly, one to one, with no exIt will carry a plane into a plane, a line ceptional points. into a line, a complete quadrilateral into a complete quadriIt will lateral, and a harmonic set into a harmonic set. therefore leave cross ratios invariant. Moreover, every point to point and plane to plane transformation will be a For every such transformation will enjoy all collineation. of the properties which we have mentioned with regard to
a collineation, and
when once we know
will, therefore, be completely determined the fate of five points, no four of which
are coplanar. But we easily see that we may dispose of the arbitrary constants in (6), to carry any such five points into any other five. It is worth while to pause for a moment at this point in order to see what geometrical meaning may be attached to coordinate sets which have imaginary values. This question
The treatment of cross ratios in the present chapter is based on that of loc. cit. The development of the coordinate system is also taken from the same source, though it has been possible to introduce notable simplificaThis method of procedure seemed to tion, especially in three dimensions. the. author more direct and natural than the more modern method of
Pasch,
*
*
'
Streckenrechnung
of Hilbert or Vahlen, loc. cit.
266
THE PROJECTIVE BASIS OF
Every
set
/
\
CH.
of
has already been discussed in Chapter VII.
complex coordinates
^/ z
\
may
To
be taken to define the
elliptic
involution
0.
(7)
verify this
statement
involution will,
by
to notice that an definition, be carried into an involution
we have merely
by any number
equations such as
of projections and intersections, and that But (7) will go into other such equations.
M/O
in the case of the line
__
""
,
'*'
>
__ ~~"~
n *')
these
equations
(x)
between and (4).
will give an involution, for the relation and (#') may readily be reduced to the type of (3) Did we seek the analytic expression for the coor(7)
dinates of a selfcorresponding point in the values
(y)
it is eas}^
we
should get
+ i(s)
to elliptic involution Conversely, be reduced to the type of (7). There is, therefore, a one may to one correspondence between the assemblage of all elliptic point involutions, and all sets of pairs of conjugate imaginary coordinate values. The correspondence between coordinate sets and elliptic involutions may be made more precise in the following fashion. Two triads of collinear points ABC, A'R'G' shall be said to have the same sense when the projective transformation which carries the one set, taken in order, into the other, has a positive determinant when the determinant is negative they shall be In this latter case alone, as we said to have opposite senses. have already seen, will there be two real selfcorresponding points which separate each distinct pair of corresponding Two triads which have like or opposite senses to points. a third, have like senses to one another, for the determinant of the product of two projective transformations of the line We shall also into itself is the product of the determinants. find that the triads ABC, BGA^ GAB have like senses, while each has the sense opposite to that of either of the triads ACB) GBA) BAG. We may thus say that three points given in order will determine a sense of description for the whole range of points on the line, in that the cyclic order of any other three points which are to have the same sense as the It is immediately first three is completely determined. evident that any triad of points and their mates in an We may therefore elliptic involution have the same sense.
;
show that any
xviii
NONEUCLIDEAN GEOMETRY
267
attach to such an elliptic involution either the one or the other sense of description for the whole range of points.
Definition. An elliptic involution of points to which is attached a particular sense of description of the line on which they are situated shall be defined as an imaginary point. The same involution considered in connexion with the other sense shall be called the conjugate imaginary point.
Starting with this, we may define an imaginary plane as elliptic involution in an axial pencil, in connexion with a sense of description for the pencil; when the other sense is taken in connexion with this involution we shall say that we have the conjugate imaginary plane. An imaginary point shall be said to be in an imaginary plane if the pairs of the involution which determine the point lie in pairs of planes of the involution determining the plane, and if the sense of description of the line associated with the point engenders among the planes the same sense as is associated with the imaginary plane. Analytically let us assume that besides the involution of points given by (7) we have the following involution of planes.
an
(u)
=
I
(v)
+ m (w),
(u)
(vy)
=
V (v)
(wz)
+ m' (w),
IV
4
mm' =
0,
=
0.
I
(8)
while
The plane (u) will contain the point its mate in the involution contains
(vz) (y)
m (ivy) (2)
if
the point
These points will be mates in the point involution,
tell us that the imaginary plane (v) f i (w) will contain either the point (y) + i (z), or the point (y) i(z). An imaginary line may be defined as the assemblage of all
and these equations
points
lines,
real
common to two imaginary planes. Imaginary points, and planes obey the same laws of connexion as do A geometric proof may be found based upon the ones.
but
it is
immediately evident analytically.* onedimensional form be transformed into itself there will be two distinct projectively
definitions given,
Theorem
32.
If a fundamental
or coincident selfcorresponding elements.
We
*
have merely to put (px) for (x) in
(3),
and solve the
See von Staudt, loc. cit., and Luroth, loc. cit. It is to be noted that in these works the idea of sense of description is taken intuitively, and not given
by
precise definitions.
* * It is highly remarkable that this axiom is superfluous. and depends upon six parameters. II. (d) If a real point. group of congruent transformations which we considered in Chapter II. lor that reason. should find no difficulty in all of the wellknown theorems of a descriptive sort proving connected with quadrics in terms of our present coordinates. called congruent transformations. is a group of collineations which leaves invariant either a quadric or a We also saw in conic. at length. however. 38) : We (a) Any real point of a certain domain may be carried into any other such point. and a plane through congruent the line be invariant. vol. is included here. does. quadratic equation in p obtained by equating to zero the determinant of the two linear homogeneous equations in # x r The assemblage of all points whose coordinates satisfy an equation of the type shall be called a quadrie. (b) Any chosen real point may be left invariant. obeying the following laws: AXIOM XII. Theorie der Transformationsgrupptn. more fully. including the inverse of each member. infinitesimal It shall be our present task to show that these assumptions. and. reached the point where we may Let us recall that the profitably introduce metrical concepts. .268 THE PROJECTIVE BASIS OF . Leipzig. will serve to define hyperbolic elliptic and euclidean geometry. and. Ch. 188893. The assumption that our congruent transformations are collineations. save an incredible amount of labour. iii. in Chapter VIII. p. Cf. a line through it. no further transformations are possible. 98. XXII. and any real plane through this line may be carried into any other such plane. The assemblage of all congruent transformations is a group of collineations. CH. joined to the ones already made in the present chapter. that the congruent group may be characterized Chapter as follows (cf. It is assumed that there exists an assemblage of transformations. (c) Any real point and line through it may be left invariant. LieEngcl. and any chosen real line through it carried into any other such line. We have now. or rather the last three.
generate a surface or set of surfaces. Tho transference from a point to a point imposes three restrictions. The congruent group depends on is six essential The number of parameters arises certainly finite since the congruent group fifteen essential parameters of the general collineation group. from analytic relations among the A independent parafixed point being chosen. or the lines in question. yet carrying any real plane through that line into any other such plane. if 33 were untrue. then. and a real plane through that line. For the tangents to all possible paths which a chosen point might follow would. i = I.xvni NONEUCLIDEAN GEOMETRY 269 AXIOM XIII. 3. shall be called a restricted region. A congruent transformation may be found of a restricted region. There exists no continuous assemblage of congruent transformations which leave invariant a point of a restricted region. a real line through that point. Theorem parameters. The group of congruent transformations may be expressed by means of analytic relations among the parameters of the general collineation group. and any real line through that point. AXIOM XV. The tangent planes to the surfaces. necessarily distinct. be freely interchanged with other planes or lines through the point. and this assemblage of surfaces or lines would be carried into itself by every congruent transformation which left this point invariant. The congruent group ciently small restricted region. or a line or set of lines. Definition. and transforming any real line through that point into any other leaving invariant any point such line. is transitive for a suffi This comes at once by reductio ad absurdum. AXIOM XVI. two more independent restrictions are imposed by . as three meters are needed to determine a point. 34. A congruent transformation may be found leaving invariant any point of a restricted region. The assemblage of all real points whose coordinates satisfy three inequalities of the type &< AXIOM XIV. "' ^0 < X { . 2. could not. Theorem 33.
contrary to our last axiom. and for all transformations obtained by this integration. hence also. need not detain us here. When. LieScheffers. a'. a' into two planes /3. for otherwise there would be infinitesimal congruent transformations which left a point. These planes are surely conjugate imaginary. Vorlesungen uber continuierliche * Cf. as no further infinitesimal transformations are possible. an absurd result. as none but analytic functions are involved.* Let us deter mine any plane through this parameters A 1 :A 2 and take an . What is essential is that this pair of planes will be invariant for the whole group. p. /3' invariant. For suppose that S^ indicate a generic transformation of the subgroup which leaves invariant the two planes a. the transformation obtained by integrating this equation Now this integral will involve one belongs to the group. Let us now look more closely at the one. a real plane. however. constant. a real line therein. which may be used to make the transforarbitrary mation transitive. line by two homogeneous infinitesimal transformation of the group The product of two such infinitesimal transformations will belong to our group. The question of whether our whole oneparameter group is generated by this integration or not. and leave the planes /3. and a real point in the line are fixed.270 THE PROJECTIVE BASIS OF CH. and combining these with the transformations $^ we have a twoparameter subgroup of our oneparameter group . determining the fate of any chosen real line through it. Our oneparameter group has thus a transitive oneparameter subgroup with a single pair of planes invariant. 1893.parameter family of project! ve transformations of the axial pencil through a fixed line of the chosen restricted region. . 125. and real plane invariant. that pair of planes will be invariant which . the limit of the product of an infinite number of such transformations as dt approaches zero that is to say. there can be no independent parameter remaining. Gruppen. When a point and line through it are chosen. line. one more restriction is imposed by determining what shall become of any assigned plane through the line. Then all transformations of the type TST~^ will belong to our group. Leipzig. was invariant for the infinitesimal transformation. and the transformation T carries the two planes a. j3'.
giving rise to invariant real planes through this point. loc. Here r must be a constant. .* The first case is surely excluded for such lines would have to appear in conjugate imaginary pairs.xvni NONEUCLIDEAN GEOMETRY 271 Let us next consider the threeparameter congruent group composed of all transformations which have a fixed point. The general invariant yf is linear transformation r sin Ov. Ai? 2 = udO. let us take an infinitesimal transformation Au = Integrating vdO. are proportional respectively to (0. The cone we seek therefore a quadric cone. (1. as otherwise we should have congruent transformations of the type u'= ru. f which kept a point. to see what sort of cones are carried into themselves by this group. p. * Cf. are invariant for the total The envelope of all these invariant oneparameter group. LieScheffers. 289. If a real line I bo carried into a real line l\ then the two planes which were invariant with I will go into those which are invariant with V. Let us fix our attention upon one such oneparameter group and choose our coordinate system in such a way that the nonhomogeneous coordinates u. for if it depended on two it would include It is well known that real planes. this system of planes must envelope lines or a quadric cone. The envelope is therefore a cone with no real tangent planes. and there are no such in the threeparameter group. the cone we are seeking for being necessarily of the number. i. Each pair of conjugate imaginary tangent planes must touch it along two conjugate the plane connecting these is real. To prove this we have but to repeat the reasoning which lately showed that the two planes which were invariant for a subgroup. v. i. and imaginary lines invariant for the oneparameter congruent group associated with the line of intersection of the two imaginary planes. 0). a line. 0). 1 of our three fixed planes . planes which pass through a point will thus depend upon one parameter. and this is not the case. u* + v is ^C. and all planes through that line In order invariant. 0. cit. 1). v '= TV. yet depended on an arbitrary parameter. keeping these three = r cos Ou v' = r sin Ou + r cos Ov. (1.
The congruent group is a sixparameter collineation group which leaves invariant a quadric or a conic. Let the reader show that this definition is legitimate as all by A and B points separated from A by B and (7.272 THE PROJECTIVE BASIS OF repetition of the sort of reasoning given CH. but the restricted region in question within (6) The quadric is imaginary. (c) Theorem 35. We AXIOM XVII. It make remains for us to find the expression for distance. AXIOM XVIII. the following assumptions. if If B be separated by ABC be three collinear real points. The envelope of these cones must be a quadric or conic. invariant under the envelope whole congruent group. i. thus. a surface which meets every plane in a conic is a quadric or quadric cone. In our present case our quadric must have a real equation. There are. from equation e=f(d). This theorem is simpler when put into the dual form. If we call a distance d. or from will belong to the restricted region. We see by a that if above that carries into a point P'. then the distance from A to O is the sum of the distance from A to B and the distance from B to G. is real. For it has just the same points in every plane as the quadric or cone through two of its conies and one other of its points. The distance of two points of a restricted region is a real value of an analytic function of their coordinates. and A and G from a point of their line not belonging to this restricted region . (1) and Axiom XIII. : (a) is The quadric it. . and the corresponding cross ratio of this type c. we must have Moreover. The distance must be a continuous function of each cross ratio determined by the two points and the intersections of their line with the quadric. it will carry the invariant quadric a point cone whose vertex is The into that whose vertex is of these quadric cones is. The quadric is an imaginary conic in a real plane. Let us first take cases (a) and (6) together.e. hence. three possibilities P we take a congruent transformation P P . since it touches the conjugate to each imaginary plane tangent thereto.
or^ + x^ + x^ = 0. e.  :i3 1 substitution. 275. and a plane through the line are fixed. rf.xvin NONEUCLIDEAN GEOMETRY this functional equation is well is 273 Now continuous solution * known. From this we may easily work back to the familiar expressions for the cosine of the fcth part of the distance. 3 gonal substitution has the value a^ for then only will there be no further infinitesimal transformations possible when a point. 2 equation (5) of Chapter VII Q^ . in particular. # = 0. (>]) f (])\ real. Pari^. These are unaltered by a sevenparameter group a xl a 2 2 a is the matrix of a ternary orthogonal For our congruent group we must have the sixparameter subgroup where the determinant of this orthowhere . Theorie des fonctions d'unc variable.g. The case of an invariant conic is handled somewhat Let the equations of the invariant conic be differently. and the only d c = e^. d If. Tannery. second edition. satisfies This function is functional equation continuous and the f(D)+f(D')=f(D + V). We shall find that. p. a line through it. * Cf. we shall have If the distance of ^ _ two points and (x). the two points be J^P while their line 2 meets the quadric in we shall have for our distance. 1904. under the present circumstances . the expression is be an absolute invariant. COOLIDQE S . 1 .
. only. Let us put /(a?) thrown back upon = log ^(a). Axioms IXVTII are compatible with the and with these elliptic. =e r (x) We thus get finally d = r Theorem hyperbolic. xvin The solution of this equation the preceding one. 36.274 PROJECTIVE BASIS is easily CH. or euclidean hypotheses.
and we there promised to return in the present chapter to a more extensive examination of this aspect of our noneuclidean geometry. Definition. as in previous chapters. cit. Theorem 3. Theorem 17. with curvature.8. when such a sum exists. and the measure of curvature of space.2. The best presentation of the problem in its general form. loc.* In Chapter II. we saw that the sum of the other two. or curve of minimum length between two A plane may be generated by a pencil of geodesies points. is greater than the distance of We thus come naturally to look upon a straight these latter.<<. distances from a point to . and in a space of ndimensions.CHAPTER XIX THE DIFFERENTIAL BASIS FOR EUCLIDEAN AND NONEUCLIDEAN GEOMETRY WE saw in Chapter XV. will be fcmnd in Schur. to Definition. to determine the nature of a threedimensiona] follows pointmanifold which possesses the property that every surface generated by a pencil of geodesies has constant Gaussian We must begin. a sufficient set of axioms. 27. A noiieuclidean plane is thus a surface of Gaussian curvature equal to  7v> A/" 1 This fact was also brought out in Chapter V. An assemblage of points shall be said to form a restricted region^ when their coordinates are limited merely by inequalities of the type 6 * <. = 1. The first writer to approach the subject from this point of view was Riomann. Theorem 30. line as a geodesic. through a point the geometrical simplicity of the plane may be said to arise from the fact that it is capable of oo 2 such The task which we now undertake is as generations. vol. * s2 . 1886. that the Gaussian curvature of a surface is equal to the sum of the total relative curvature.' Mathematische Annakn. : Any set of objects which may be put into one one correspondence with sets of real values of three independent coordinates z^ z^ z% shall be called points. Ueber den Zusammenhang der Raume constanten Riemannscheii Kriimmungsmasses mit den project iven Raumen. not collinear with any it.
. j = 1.276 THE DIFFERENTIAL BASIS FOR EUCLIDEAN I. a curve. As we have defined only those points whose coordinates are real. or to a smaller restricted region within the original one. a positive definite form for all real values of dz L dz^ dz^ 2 3 corresponding to points of the given . The expression shall be called the didaive element. between two chosen points along a curve shall be called the If length of the corresponding portion or arc of the curve. the expression length must be applied to that portion along which the integration was performed. 0j. s 3 real There exist nine functions i. or. II. This amounts to confining ourselves to the original region. 3 a^. We shall restrict ourselves to such a portion of the original restricted region that for no point thereof shall the discriminant of our quadratic form be zero. Definition. more simply. 2 2 . AXIOM AXIOM of There exists a restricted region. it is evident that the functions involved in the definition of a curve must be The definite integral of the distance element real also. and possessing the following properties is and all values of z^ restricted region. The curve whereon this arc lies shall be called a geodesic connecting the two points. . CH. The assemblage of all points whose coordinates are analytic functions of a single parameter shall be called an analytic curve. An arc of a curve between two fixed points which possesses the property that the first variation of its length is zero. 2. the curve pass many times through the chosen points. shall be called geodesic arc. and analytic throughout the restricted region. Definition. Limitation. Definition.
There will exist a single set of solutions corresponding to a single set of initial values for (z) and (/). then. Jordan. express a^ as functions of s. or These three equations are of the second order. vol. *<'*/ We have now a problem in variations.3 *= J A/ 2 V U simple >.are arbitrary.2. singularities may.^/^ 1 ' S' 3 H.2.xix AND NONEUCLIDEAN GEOMETRY 277 Let us begin by setting up the differential equations for a geodesic. 88. Let us write s is an analytic function of t with no in our region.3 . p. vanishes at the extremities of the interval the increments bz. by taking our restricted region sufficiently small.. Cours d^Analyse^ Paris. P we have *' the calculus of /1. 3 1.2.g. iii. hence the coefficients of each must vanish.3 288 =J 2 2 (^ *//'** + 2 ^ '/a*/)*T' f 52 ' i ^1. and write It is clear that We . .+ 2 y ^ *// ^' hence.* Let these be (0) and (Q * Cf. e. since Sz. ' Replacing ~ temporarily by z/.2. ^l. 18936. 2 3 ' /j = 2 ^%^ 0' cZ^. hence t is an analytic function of s.
Darboux. A real analytic transformation of a restricted region which leaves the distance element absolutely invariant shall be called a Definition. respectively.3 = * Cf. *. Two by a points of a restricted region whose coor amount may be connected a single geodesic arc lying wholly in a sufficiently small by restricted region which includes the two points. and write We shall take our restricted region so small that (4) shall be uniformly convergent therein. Given a geodesic through a point (5). 3 1. The shall be called the direction cosines of the geodesic at that point.278 THE DIFFERENTIAL BASIS FOR EUCLIDEAN Any point of such a geodesic will (3 CH. nected by a single geodesic arc lying entirely therein. be determined it by We have thus <fj > and r the length of the arc connecting ' with (z). 2 ("#ij . We shall from now on. loc. suppose that we have limited ourselves to such a small restricted region that any two points may be so connected by a single geodesic arc.* Theorem 1. . p. Now r the expression D (z z z n /^"T'V u (T^ r^^r^) has the value unity when ) l = 0. ii. cit. vol. three expressions congruent transformation. for all values for (z) and (z) Hence two points of the region may be conin the region. We may therefore revert our series.2. Notice that 1. dinates differ sufficiently small Definition. 408.
* It is clear that a congruent transformation will carry an arc whose variation is zero into another such. direction cosines there. with slight verbal alterations. AXIOM III. in order to avoid too close plagiarism. His article. Hence This expression shall be defined as the cosine of the angle formed by the two geodesies. the geodesies shall be said to be 'mutually perpendicular or to cut at right angles. though vitiated by a certain haziness of definition. for each surface of this sort may be generated in oo 2 ways by means of pencils of A We geodesies. shall later show that the choice of the name geodesic surface is entirely justified.2.xix AND NONEUCLIDEAN GEOMETRY 279 This is a positive definite form. When it vanishes.3 is au _ __ _ ds bs all is obviously an absolute invariant for congruent trans formations. those used by Woods. cit. Theorem 2. This comes at once from the fact that 1. are. shall be said to form a pencil. There exists a congruent transformation which carries two sufficiently small arcs of two intersecting geodesies whose lengths are measured from the common point. surface which they trace shall be called a geodesic surface. hence a geodesic * Our Axioms IIII. leaves nothing to be desired from the point of view of simplicity. It is also noteworthy that ho uses k where we conformably to our previous practice use  . into two arcs of equal length on any two intersecting geodesies whose angle is equal to the angle of the original two. In the present chapter we shall use a different coordinate system from his. set of geodesies through a chosen point whose Definition. are linearly dependent upon those The of two of their number. for the coefficients are the minors of a positive definite form. loc. The angle of two intersecting geodesies absolute invariant for all congruent transformations.
for in any surface we may transform congruently from a. in fact.e. It is now necessary to choose a particular coordinate system. and r const. = const. The square of the distance element will take the form i. into another such. Let us choose a fixed point (z) and a fixed geodesic through it with direction cosines (). by Axiom III. are indepensee. write for the square of any distance element can be put into the where fa is a function of * <f> and 0. form an orthogonal system for the same reason as before. r as coordinates of this point. we may transform congruently from one surface = const. . are not geodesic surfaces. cit. Let $ be the angle which a geodesic through (0) makes with the geodesic (f). we choose a geodesic surface determined by our given geodesic. We may. F. 160. that there will be no term in drd<t> or drdO. Bianchi. for. and another through (z). will be orthogonal to these radiating geodesies.ny two = const.* The surfaces <f> const. while 6 is the angle which a geodesic perpendicular to the last chosen geodesic and to (C) makes with a geodesic perpendicular to the given 9 geodesic surface. We may take <. into a geodesic surface. and the geodesic lines of space radiating from (z) and lying in this surface will be geodesies of the surface. for it is immediately evident that we might have defined a geodesic surface as generated by those geodesies through a point which are perpendicular to a chosen geodesic through that point. we have a geodesic surface. and we shall make use of one which will turn out to be identical with the polar coordinate system of elementary geometry. The curves r const. generating pencil. Finally.280 THE DIFFERENTIAL BASIS FOR EUCLIDEAN CH. The coefficients E. The coefficient G is independent of $ also. We For if we take = = = = dent of 0.. p. but the curves const. Differentlalgeometrie. It will also transform a geodesic surface into a geodesic. (5) in fact. perpendicular to the geodesies of the Let r be the length of the geodesic arc of () from (2) to a chosen point. geodesies through (2) into any other two making the same angle. 0.
There is a group of oc 3 congruent transformations which carry a geodesic surface transitively into itself. Theorem. p. and distinguish with care the two following cases p*0. Bianchi. Theorem 2. is an invariant of space which we may call its measure of curvature. 4. and is identical with the geodesic of the surface which connects those two points. be taken to stand for any geodesic surface. have = Here we must E cG.0 " "*" d0 Vds/ V ds Consider the geodesic surface $ =~ 5 which may. __ L dsj d ds JL "*" r ds J ~~ j 2 Lty Vds/ 4. equations surface These theorems enable us to solve completely our differential The Gaussian curvature of each geodesic (6). The on this surface will be * differential equations for a geodesic dsdsJ  2r Vds ds These are exactly equivalent to the combination of (6) and const.xix AND NONEUCLIDEAN GEOMETRY 281 (2) Let us at this point rewrite our differential equations in terms of our present coordinates A [dri . 153.. We shall denote this constant by p . indeed. ibid. 3.1 r^ (dd \ 2 r ds LdsJ 2 LD (ds) +2 ^( \ cQ\ + dr Vds/ c **? \~ds) /^ *r Vds/ 2 1 ' J d 1 . Theorem. Lastly. * 1 = 0. . The geodesic connecting two near points of a geodesic surface lies wholly in that surface. where cxirve c is constant. if we remember that two near points of a surface can be connected by a single geodesic arc lying </> = therein. All geodesic surfaces have the same constant Gaussian curvature.
by the equations preceding cfo 2 (6) = rfr + k* sin 2 2 y rC . Writing that this shall have Gaussian curvature rC" 7 .2.282 THE DIFFERENTIAL BASIS FOR EUCLIDEAN CH. G is now an easy task. 2. t. Hence.2. F. ^=*. T T giving eventually =1. f B cos . > Are get V 6 = A sin y care. It is clear to begin with that when Hence Again 1..3 But._d$ . 3 e. will be ^6 2 _ dr2 +6(r) d!</> 2 surface . The determination of the constants A. 3 <^^k *<r^ cos cos T= 1 2 2 ~ ~ 8m d<f>_d<t. The determination of our coefficients J5. 1. ^ 1. The square of the distance element for a geodesic = const.3 B requires a little J? ^ = 0. from (1) It 2.
(7) . make on Hence the surface r = const. we must find E'. surely Jess than unity. The surfaces r const. = A sin On As we saw a moment ago the other hand.. The differential equations for a geodesic curve of the surface 6 = const. These must be equivalent to those obtained from i. have Lastly.xtx AND NONEUCLIDEAN GEOMETRY 283 We proceed to calculate F. a constant everywhere. for 5= E vanishes with when But also I is A sin ^?r = an old integer. const..e. constant Gaussian curvature. will be /dr\ d /d Vrfs/ _ 1 ^0 2 /cZ(/>\ ' 2*r\ds) d ds = const. which v FQ the cosine of the angle which curves = is const. Hence ' = ds* = /^sin 2 = const.. we must have (6). Hence and sin 2 2 ^ [sin <^> d O 2 + d^8]. for each is capable of oo 3 congruent transformations into itself.. when (IS F =. and <f> = const. . 0. l<f) + B cos 0. there is is no dO term in ds 2 so that E= .. r and as F' but is not a function of it is Now when ~E > = 0.
we have <U* = dr + v* [sin2 2 4> dO* + d<f>*]. (^cZ o. Let the reader show that under the second case ~ = 0.284 THE DIFFERENTIAL BASIS FOR EUCLIDEAN CH. (7') now time to return to coordinates of a more familiar Let us write x ~k cos r 7* s #j = k sin v cos cos 0.) + ds* = d . ( But from our equations We thus get for the final form for our ' differential equation (9) (y ^ ~ ^+0 + cZs a i> Let the reader show that in the other case we have . It is sort. This is our ultimate form for the square of the distance element. 2 = & sin Y si i v (8) ' (dxdx) = 2 cits . find the differential equation of a geodesic.%ds*) = 0. a problem in relative minima To we have s( To determine A (icr) 2 = fc 2 .
for the protective theory is the same for a limited domain in all restricted regions. At bottom. have established our distance formulae three several times. formulae by proceeding from elementary geometry to trigonometry.: k V(xx) V(yy) . II. In Chapters IIV we took as fundamental the concepts point. if we replace our coordinates by homogeneous ones proportional to them ci ( X y\ " ^^ COS ~"~~~pr~i~ \ ) / ~ ' . Our task is now completed. and sum of distances. = alone. III are compatible with the euclidean hyperbolic and elliptic hypotheses. and with these d Theorem 5. laid down point the common and separation as fundamental. the essential feature of a geometrical system where the elements are points is the expression for distance. Axioms I. which would involve slight modifications). reached our analytic distance. and established the system of projective coordinates. In the present chapter we took as fundamental the concepts point and correspondence of point and coordinate set.^ Let the reader show that when = 0. such as we do when we first take up the study of elementary The Chapters VIXVII were devoted to analytic geometry.  285 Integrating 6 6 k* s = (xx) = (yy) = (zz). constructed project! ve geometry for all of our systems (except the spherical. and worked around to our previous distance formulae through grouptheory. We then introduced certain collineations called conyruent transformations. (yz) = 0. The essentials in our development were the distance element. upon the foundation which we had XVIII we took a fresh start. the geodesic curve. and the space conline . each time approaching the subject from a new point of view. S( x z )*+ (yy )*+( Z z )*.xix AND NONEUCLIDEAN GEOMETRY s . and then introducing a simple coordinate system. We We erecting a superstructure In Chapter established. (y) ci We to (x) have then for the length of the geodesic arc from 2 jfe cos = (xy) 9 or.
and the calculus of variations. to us. Cambridge. beyond a perad venture.. Which of the three methods of approach is the best ? To this question no definite answer may be given. formulae by means of surface theory. Let us not forget that. take a more abstract view. for that method which is best for one purpose is not. we are free at the outset to choose our line of approach.286 DIFFERENTIAL BASIS en. will. It also corresponded most closely to the line of historical development. large measure. . there is no answer to the question which method of approach is the best. f Cf. which to some readers might seem a trifle unnatural or forced. the fundamental importance of the sixparameter collineation group which keeps a conic or On quaclric invariant was brought into the clearest light. No. The first method depended upon the simplest and most natural fundamental conceptions. Moreover. even after cutting out a number of theorems. by assuming at the outset that space something whose elements depend in a definite manner on The modern tendency is to three independent parameters. full significance of the space constant. to look upon space. cit. f The battle is more than half over when the coordinate system has been set up. we were obliged to develop a coordinate system.* Our third and last method is. Russell. in preference. the other hand. but not essential as steps towards the ultimate goal. third edition. xix We reached our familiar stant. 3 is paid lor this directness. and exposed ourselves to being put down among those whom the late Professor Tait has stigmatized as ' That section of mathematicians for whom transversals and an. Tait. 372. in the last analysis. The determining choice among the three. as a set of objects which can be arranged in multiple series. The second method possessed the advantage of beginning with the assumptions which serve as a basis for the important subject of projective geometry metrical ideas were grafted upon this stem as a natural development. On the other hand it is the longest. best for another. the quickest and most direct and has the advantage of bringing out the It may. incomprehensible charm . are fortunate when. We * p. 309. or measure of curvature. in the end. integration. loc. be urged with some justice. interesting in themselves. 1890. we study pure mathematics to satisfy an aesthetic need. that too high a price has been . p. harmonic pencils have a. and presupposed a minimum of mathematical knowledge. as in the present case. however. An Elementary Treatise on Quaternions. necessarily. be a matter of personal aesthetic And this is well.
134. 154. 159. 67. of gravity of points. 94. 136. 171. 210. 150.  triangle. 178. 244. Chain congruence. 126. 150. 106. 281. 156. Archimedes.  196. 107. 134. 156. 49. 170. plane. 173. 177. 246. . of a pencil of complexes. Canal surface. Coaxal pencil of complexes. 34. 131. 124. 88. 202. 152. 28. 142. Birectangular quadrilateral. 212. 131137. 187. 159. 116. 231. of skew lines. radical of two circles. 232. 70. 116. 226. Asymptotes. 240. 102. 220. 134. 30. 32. 138. 156. 135. of a plane. 153. 135. 133. 159. 136. 279. 234. cosine. 157. 153. 105. 152. 130. Axes. 177. 151. Barbarin. 111. 146. Battaglini. 39. of halfplane. 67. 230. 176. Beck. 20. . auxiliary to conic. Bianchi. 104. of crosses. 119. 129. 105. 120. 110. 161. 135. of similitude. Bromwich. 99. Amaldi. 146. 175.of a conic. 109.  Aronhold. 158.of a triangle. coordinate. Bisector of an angle. 143. parallel. 31. 178. 119. 280. 113. Axis of a circle. 132. 107. of two planes. 97. . 157.  null. 143. 178. 6. exterior of a triangle. Axial plane of sphere. 117. 129. 205. 119.of a polygon. 87. Angles of a triangle. Area. reentrant. 31. 101. measure of. Centre of quadric. 136. Axis of a chain. Ceva. 211. 116. 39. 98. Clebsch. 135. 232. Angle. 137. 143. 116. 204. .of a conic. Asymptotic lines. 175. 179. 38. 203. 234. Amplitude of tetrahedron. 226. 226. 157. 149. 43.INDEX Absolute. 103. 143. 31. 155. 188. Bound Bound of halfline. 87. 31. 121. 143153. Amplitude of 172. Clifford. Cayley. dihedral. 102. 170. 124. straight. 24. 18. 95. Beltrami. 162. Centre of a circle. 103. 222. 148. Central conic. 146. 85. 159. 97. 134. 138. 102. 152. right. 136. 205. 128. 30. 180. of dihedral. 154. 210. 127. and exterior. 143. Circle. 176. 143. 129. 113. 136. . Clifford. 106. 102. 209. 188. 44. Bolza. 187. 178. 220. of a circle. 135. 251. 131. 99. Central quadric. 222. 136. Borel. interior 88. 212 213 Author. Coaxality. 181. 30. Collinearity. 157. 157. 88. 178. 127. 133. 116. Actual elements. 118. 167. 126. 103. 64. 206. 109. 154. 154. 135. 103. 134. 233. 15760.
176. 209. Congruent figures. 169. 124. 16. Dehn. 208. Discrepancy of a triangle. 114. 224. 90. 147. 75. Conic. 167. 38. 265. 28. 160. 251. isotropic. 92100. 39. Dannmeyer. 83. 110. 62. 146. 120. 66. 234. 164. of a point. 89. r harmonic. Distance. 52 69. 46.of lines. 15. 235. 232. Director points and directrices. 235. 268. Difference of distances. 195. 103. 257. 263. 54. 74. 67. ofquadric. 13. 208.' 91. 231. 154. Diagonals of quadrilateral. 83. 271. surfaces. 31. 285. Enlargement of congruent formation. 268. 138. 226. Equidistant curves. 69. 275. 247. 181. 205. . total relative. Concurrence. Connectivity of space. 141. 33. 73. 153. 121. 262. 200. of angles. 164. 285. 142. 74. 161. 229. 146. 252. 144 145. 24. 78. 149. general. 79. 36. 236. 35. 278. Congruence of distances. 196. 26.288 INDEX Cross space. lines of. 162. 125. 200. 34. 119. 187. 282. 247. Collineations. 238. 69.of a plane. 127. 86. 136. 235. 278. surfaces of zero. 201. trans law of. 187. Distance. 31. 189. 227. 281. 70. 17. Dupin. 261. Coplanarity. 272. 227. axiom of. Diameters of conic. 227. hypothesis. 37.from point to plane. 74. 278. Elliptic coordinates. 122. 43. directions on a surface. 90. 34. 156. 194. 17. Gaussian. 133. 24. of segments. 28. Division of segment. 29. 273. 82. 88. Distance of two points. 285. 46. 270. Comparableness of angles. 206. 130. 151. 91. 67. 67. 57. 38. Diagonal points of quadrangle. 201. 109. chain. 208. 14253. 118. 42. 275. 185. 33. Continuity. relative. 112. 73. 80. 264. 75. 167. 123. directed. space. 32. 141. Curvature of a curve. 143. 27. 138. 212. 82. 123. 199. 274. 260. 18. 218. 70. cosine. 226. 53. 87. 204. 259. 238. 72. of normals. 226. 123. 281. Cosine of angle. Cross. 90. 235. Coordinates of a line. 259. 215235. 261. to surfaces of Gaussian curvature zero. 156. 208. 38. 245. 194. 94. 283. 153. 36. Consistent region. 73. 239. of lines. 236. 105. 177. 269. of triangles. of skew lines. 117. 29. 150. 16. 227. mean Conformal transformations. Confocal quadrics. 46. 80. . 109. 264. 64. 111. 29. 206. 275. Ellipsoid. 253. 198. 204. 14. 168 r 169. 20. Engel. Conic. Confocal conies. 189 T 200. Cone of revolution. 205. 251. 159. 197. Edge of tetrahedron. 226. 70. 160. 237. 60. 78. Dunkel. Ellipse. Enriques. 225. Cross ratios. 110. 102. 40. 129. 266. 285. 279. 230. 222. 208. 188. 204. 70. in change of angles and sides of a triangle. 134. 79. 69. 280. 116. Desmic configuration. 264. 41.27684. 66. 148. of normals. of distance. 68. 148. 16. geodesic. Desargues. element. 25. 204. 89. . 35. 176. 279. 118. 249. 205. 237. 23. 28. 210. 265. 52. 205. 76. Darboux. 264. 73. 143. elevenpoint or line. 78. 272. Complex of lines. synectic. Cosines. 252. 170. 207. . 252. 132. 119. Density of segment. 207. T 239. analytic. 203. 223. 198. Conjugate diameters of a conic. 212. 108. 174. 146. Congruent transformations. of space. 254. 153. direction. 188.
15. 232. 47. 267. 222. 176. 23946. 289 72. 229. 91. 125. 243. 2833. 249. Gerard. 281. 159. 38. Manning. Left and right translations. 100. 146. 277. 132. Horocycle. 237. 98. 266. 37. 92. 235. 82. 164. 37. Von Frank. 147. fundamental onedimen sional. 266. 267. 144. congruence. 37. 75. Luroth. 270. 43. 75. 87. 186. Euclid. 17. 106. Layer of cross space. 241. 236. 68. 158. Less than. 38. 175. 240. 186. Infinitely distant elements. 209. 17. Fricke. 117. Graves. conies. 260. 24. 259. 145. 285. 271. Foci. 27481. Homothetic conies. 215. points and planes. 177. 168. 197. Focal cones. Imaginary elements. 89.INDEX Equivalent points. 46. 275. 242. 234. 221. 176. 210. 261. 274. 225. Liebmann. Klein. Intersection of lines. 156. 16. 267. 234. 174. 227. 215. 67. 153. 78. 151. 32. 160. 232. 175. 139. 268. 85. 129. Euclidean hypothesis. 15. 209. 168. 167. 118. 119. 92. 53. 210. 167. 99. 253. 249. 285. Lobatchewsky. 142. 284. 46. 167. Marie. space. 72. 252 253 Hilbert.  Kummer. Fubini. 78. 203. Halsted. 85. 35. 77. 176. 221. 142. 39. 86. Hyperbola. 13. 285. 34. 168. g< Lindemann. . 177. 84. 158. Evolutes. Frischauf. 'l3. 151. 16. 267. 153. 35. 244. 244. 97. 159. Initial point. 99. 257. and right parataxy. 261. 20. 254. 253. triangle. 107. 176. 143. 62. p measure of curvature of space. 145. 80. Forms. Fundamental region. 53. 208. 190. 230. ray. 24. 144. 22. 236. 42. 227. 146. 48. 276. 205. 79. 78. 239. 118. 279. 248. 201. 46. 208. 87. 50. 226. onedimensional forms. 169. Improper cross. 147. 193. 255. 120. 265. 280. 155. 142. 226. Involution. Extension of segment. 260. 231. 245. 99. 205. Left and right generators of Absolute. 169. Excess of a triangle. Horocyclic surface. 47. 73. Limiting points and planes. 75. 169. lines. 220. 15. Line. Lie. 78. 232. 174. Geodesic curvature. 47. Ideal elements. Hyperbolic hypothesis. 81. 125. 64. of space. 192. quadrics. Extremity of segment. 87. 281. 194. Indicatrix of Dupin. 245. Hamilton. 77. 189. 84. Ste. 79. 17. 205. Killing. Lobatchewskian hypothesis. Length of arc. 274. 152. Hyperboloid. Levy. 15. Isotropic curves. Fibbi. 252. 224 surfaces. 127. Harmonic conjugate. Greater than. lines. 177. 251. 30. 161. 124. 259. 209. 176. 46. Frenet. 86. Jordan. Joachimsthal. of planes. Left Halfline.*36. 222. surfaces. 219 separation. 267. Halfplane. Face of tetrahedron. Isosceles quadrilateral. 221. 146. 223. 17. 163. 231. 73. 34. Infinitesimal domain. 204. 261. 72. space. 34. a 252.
142. 85. 46. Menelaus. 279. 135. Polygon. 197. 155. 157. Salmon. 281. 78. 143. 99. 98. 235. 234. 174. 46. Right and left parataxy. 94. 67. 208. 239 46. 208. 53. . 253. 13. 253. in point. 50. 191. 220. Region. 13. 21014. 192. Products connected with a conic. 252. of geodesies. 189. d'Ovidio. Pieri. 156. 75. 243. 84. 38. 228. 20.217. 247. halflines. 125. 210. 268. 39. 260. 142. 20. 50. 27. 100. 252. 245. 55. Pseudoisotropic congruence. 206. Orthogonal substitutions. 113. left Peanp. Right and 245. 70. Quadrangle. Null angle. 208. 229. Region consistent. 149. 230. 266. Picard.224. 229. 46. Moment. Right angle. 193. restricted. of curvature of space. 183. 250. Normals to curve. fundamental. 137. 264. 225. 207. 139. 208. 144. 63. Parallel angle. 43. relative of two lines. 86. Meunier.215. 39. lute. senses. Ratios. 52.290 Measure of distance. Pencil of complexes. 14. 155. 34. 236. Quadrilateral. 46. 178. 21. 22. 201. 52. Paraboloid. 129. 266. 242. Motions. Pseudoparallelism 234. 79. complete. 275. 86. 233. of lines. 131. 267. 44. 7O. 225. 62. 217. Quaternions. 135. 235. 118. 50. generators of Abso 36. 105. 136. 124. 219. Perpendicularity. 99. 265. 73. 234. complete. 139. 224. 82. 210. 13. 151. 46. 184. Pythagorean theorem. Projection. 83. two rays. 197. 31. 223. 81. Minimal surfaces. 103. 24. 227. 115. 81. 139. 99. 95. 266. Projectivity. 143. 57. Parallelism. 132. 277. Orthogonal points. 107. 118. 138. 279. 138. 69. 28. 80. 13. 113. 101. 106. 247. 129. to surface. 237. 99. 30. Riemannian hypothesis. 34. 114.205. 197. 44. 125. 279. constant connected with conies. Right and 32. 109. 98. 260. 23846 Munich. 192. 224. INDEX 249. 67. Origin. Opposite edges of tetrahedon. 234. 259. 136. 112. 269. 220. 224. 87. 32. 145. Phi function. 64. 262. Revolution. 238. 193. Parabolic hypothesis. 235. Q7 Q& J/l. Middle point of segment. 86. 37. 259. 34. 229. Pasch. 194. distance. 112. 275. 78. 235. Parataxy. 49. 116. 150. Multiply connected space. 198. 137. 55. 13. 97. 29. surfaces Riccordi. 176. Principal points and planes. 139. 182. 234. Ray. sides of plane. left translations. 189. Pseudonormal congruence. 207. 51. 106. 222. 44. Russell. 235. Moore. 70. 275. 103. 235. 74. Plane. Rectangle. 276. Parabola. 110. 162. 82. 226. 74. Poincare. 278. 110. triangle. 118. 22. 234. Richmond. 134. 269. 99. 175. 43. 251. 204. Reflection in plane. 192. 45. 132. 86. 227. 143. Padoa. 170. 224. 33. Saccheri. 43. Scalene triangles. 253. Riemann. 35. 101. 115. 114. system of surfaces. 39. 254. 13. relative of Point. 45.219. Ratio of opposite sides of quadrilateral. of. 265. 49. 51. 53.
23846. 250. 239. 82. 76. integral. Sines. 207. 17. 35. of tetrahedron. 266. 19. 275. 20. Stephanos. 154. of quadrangle. of cone. 269. 24. 227. 267. 265. of triangle. Torsion. 194. 227. 46. 228. congruent. 73. Separation. 58. Sturm. 237. 262. 136. 271. 148. 142. 78. 175. of angles of a triangle. 254. Within a segment. Thread construction. 18. Transformations. 258. 21. 38. 43. 13841. 249. Segre. 73. Spheroid. 31. 63. 278. 145. 119. 91. 19. 182. 70. 21. Von Staudt. of quadrilateral. 315. 257. 247. 270. Volume. 59. Story. Vertex of angle. 252. Sine of distance from point to plane. 191. 174. 155. Voss. . 99. 45. of sphere. Sum of distances connected with a conic. Vahlen. 43. 15. 156. 251. 32. 92. 228. 273. 188. 85. Sum of angles. 184. 247. Tait. 246. 31. 34. 32. 29. 162. 69. 46. 129. of quadrilateral. 135. of quadrangle. 192. 286. 86. 187. 182. 239. of tetrahedron. Sides of angle. law of. 184. 252. 156. 62. Symmetry transformations. harmonic. 128. 125. 87. 18. 63. Sum of distances connected with a quadric. Study. 257. 74. 185. Young. 99. Segment. 240. Vailati. 247. 37. Veronese. 259. 229. 74. 291 98. 182. Strip. 253. 252. 245. Sum of two sides of triangle. 234. 100. 186. 32. 98. 20. Surface integral. 128. 25. 181. 62. 123. 13. 28. 257. Spherical space. Trirectangular quadrilateral. 248. 36. 185. 13. 64. 181. 177. 19. 172. Tannery. 34. 75. Weierstrass. 43. Veblen. 195. 252. Ultrainfinite elements. Within a triangle. 256. 260. 279. 182. 134. 31. 254. 70. Triangle. 162. Woods. 258. Stolz. representing. 18. 268. 127. Translations. 80. Tensor. 190. 20. 13. Spheres. 22. 170. Umbilical points. centres of. 1417. 203. 169. Similitude. Triangles. 126. 255. 92100. 116. 120. classes. 149. Tetrahedron. 32. 24. 86. Semihyperbola. 16. 74. of triangle. 279. Sphere. 83. 183. Stackel. 122. 248. 246. 89. Sense of directed distances. Synectic congruence. Schur. 30. 247. 30. Supplementary angles. 249. 13. Tangent plane to surface. 142. congruent. 175. 280. 233. Space. 15. 160. Semihyperboloid. Singular region. Staude. Terminal point. 93. Vertical angles. 108. 26. 23. Sum of distances.INDEX Schlafli. 185. Weber. 238. 19. 255. 24. 176. 35. 142. 183. 183. of description of involution.31. 245. 93. 19. 252.
M.OXFORD PRINTED AT THE CLARENDON PRESS BY HORACE HART.A. PRINTER TO THE UNIVERSITY .
10s. net. Sources for Greek History Wars. and the Prankish Empire. Vols. and the Lombard Kingdom. net. The Lombard VIIVIII. 5s. By E. F. 6d. Second edition. A Manual of Greek Historical Inscriptions. net. by the Author. 8vo. 8vo. STGSU). New edition. 2nded. A new edition. I and II. Is. CLAY. F. illustrating the history Empire. 2 ios. to A. A. TOZER. Hunnish. Invasion. Os. net. ed. The Native Nations ments to the beginning of Athenian Intervention. and Istria with Cettigne and Grado. 8vo.net From the CXXIVth Thirdedition. The Phoenician and Greek SettleVols. net. Hvo. Origines Kalendariae Hellenicae. The Ostrogothic Invasion. between the Persian and Peloponnesian Reissue. M. HILL. 6d. Eight volumes. 15s. With pi ales and illustrations. TWO volumes. 31s. 2 9s. Vol. Edited from posthumous MSS. F. L. M N. Crown 8vo. 1 Is. 133TO. Description of Ancient Greece. net. Crown 8vo. net. (is. HILL.RAWLiNsoN. 10s. H.146) and in part rewritten. Hvo. net. net. revised. 6d. Heraclius. : 2 2s. 8vo. 8vo. Sources for Roman Crown History. Dalmatia. Topography. Cramer's Description of Asia Minor. Bvo. net. Vol. 6vols. and Antiquities. Hvo. 2 9s. TOZER. OriginesKalendariaeltalicae. 7 vols. . lt>s.D. By E. revised by G. 4 4s. Mutiny. Three volumes. L VVI. revised. the Quarnero. Aetolia its Geography. EVANS. Egypt. Bvo. 6d. and the Herulian III. 1 4s. XI Is. By G. 1 14s. 1 12s. By H. net. The Barbarians. net. Two volumes. net. 8vo. ByG. Prankish Invasions. The Imperial Restoration. By G. from the death of Augustus to the death of 7s. 6d.D. The Visigothic. with many additions. Greswell's Fasti Temporis Catholici. us. net. Italy and her Invaders (A. 4 vois. I Tables and Introduction to Tables. GREENIDGE and A. 6s. SB. By A. III. 6d. Latin Historical Inscriptions.G. The Islands of the Aegean. net. C of the Early RUSHFORTH. IIIIV. J. By T. 4to. From the Tyranny of Dionysios to the Death of Agathokles. Epitome. and Vandal Invasions.c. The History of Sicily from the earliest times. J. 1H64. By the same author.C. or. 4to. net. Hvo.6d. 3 vols. 1 . and edited by H. By W. Clinton's Fasti Romani. Of TheodosiuS . A Manual of Ancient History. Seventy Years' Struggle with the Dynasty By T.CLARENDON PEESS BOOKS HISTORY Greece. With plates and maps. Royal 8vo. HODGKINT. 8vo. IGS. IIV in the second edition. Italy. . from the LVIth to the CXXIIIrd Olympiad. us. 8vo. net. WOODHOUSE. 4vols. etc Clinton's Fasti Hellenici. With maps and illustrations. 2 2s. IV. Olympiad to the Death of Augustus. 63s. Bvo. HICKS. 8vo. F. The Athenian and Carthaginian Invasions. Epitome. 1 1 16s. FREEMAN.JACKSON. Finlay's History Of Greece from its Conquest by the Romans (B. 8vo. J. 4to. B. by A.net.
By A.svo. I. Part III (Persian. and Primitive Linear Classes. Introduction to Greek Sculpture.BUTLER. II. Royal Bvo. 1 s. 16s. 72 illustrations in the text. large Herculanensium Voluminum isn. paper 1 Is* 1 Is. Hieroglyphic Royal Ho. Catalogue of the A. Engravings from the Fragments.) Royal Hvo. net. net (Mughal Emperors). A. 6d. Crown Hvo. with numerous collotype plates. Contemporary Dynasties in India). by V. net.Western Phrygia. 18s. By M. N. Archaeology Ancient Khotan. By J. Biasii Caryophm. 6s. I.The Cities and Bishoprics of Phrygia. E. J. Marmora Oxoniensia. Vol. Mediaeval. II. j3 3s. Sicily. 4Os. BOURDILLON). I. 4to. editae.) By w. 30s. illustrated. 7s. net. isss. By w. UFCOTT. B. L. Conquest High Egypt. (Oxford Fragments. The Lycos Valley and South. Cults of the Greek States. and the Demotic Tales of Khainnas. E. '2 3s. net (Sultans of Delhi. V. net Stories The Arab sources. 6d. 207 plates. ios. is. II. by H. Vol. (Published for the Trustees of the Indian Museum. LL. net. J. LE STRANGE. with 26 plates. net . Five volumes. With plates. linen. FARNELL. Archaeology in Schools. The of the Greek Vases in the Ashmolean By P. with descriptive list of antiques. net. o. Vol. Catalogue Of the Coins in the Indian Museum. net. Second edition. T. 6d. Vol. SCOTT. AUREL STEIN. GRIFFITH.M. 15s. 10s. De Antiquis Marmoribus. 32s. the Sethon of of the Herodotus. figures. South Indian. or Part I (Early Foreign Dynasties and Guptas). IIlIV. inscriptiones Graecae ad Chandleri exempla GUL. Partes n. net. R. 8vo. 7s. With 2 7s. MYRES and A MAX OHNEFALSCHRICHTER. By L. Catalogue Museum. TOD and WACE. 119 collo5 5s. l(>s. net . The Stone and Bronze Ages Hvo. Folio. in Italy and P. Fragmenta Herculanensia. 8vo. cur. net. ROBERTS.. net. and many other illustrations. Text. Miscellaneous). Royal 8vo. 8vo. PRET. tid. net. 2w. BUTLER. 8vo. 8vo. 6d. N. 30s. A Catalogue of the Cyprus Museum. M. Indian Government by M. net. from contemporary By G. Baghdad during the Abbasid Caliphate. 8vo. Ancient Coptic Churches of Egypt. 1791. sos. 6d. by H. Scripta Minoa. 8vo. I. IBs. N. ed. Portfolio containing seven facsimiles. Sparta Museum. Classical MYRES. Calcutta. By J. . Part I. and appendices. 1 Is. Par. Small folio. Vol. Vol. Royal 8vo. With eight plans. I. GARDNER. 10s. West and West Central Phrygia. Part II (Ancient Coins of Indian Types). net. Priests of Memphis. net . net. net. By L. By A. J. Hvo. SMITH. including the Cabinet of the Asiatic Society of Bengal. Vol. WRIGHT (the first section of Part II by Sir J. WRIGHT. J. net. IGs. 6d. By GARDNER and L. 2 vols. 10s. III. type and other illustrations and a map. :#s.ed. Vol. 3s. With eight plates. net. Ill. Of By A. in Detailed report of Archaeological explorations Chinese Turkestan carried out and described under the orders of H. By F. EVANS.
stiff The SaXOn Chronicles (7871001 A. Small 4to. 6d. 6d. le by C. edited 1 Is. L. WEAVES.). GEORGE. HUGHES. 3s. KINGSFORD. A II. appendices. by C. with the Annals of St. from J. KINGSFORD. STEVENSON. With an introduction by J. Chronicles of London. etc. EARLE. Edited by Crown 4to. 10s. In three volumes. Collection 12s. an historical essay. net. illustrating the condition of Church and State.English History Opera Historica. THOROLD ROGERS. Edited. 30s. S. Two J. The By Protests of the Lords. net. JOHNSON. net. STEVENSON. L.. edited by W. 6d. with supplementary . 14031458. Two glossary. G. edited by C. volumes. Crown covers. S tow's Survey Of London. 12s.C. METCALIE. 10s.B. H. Crown Svo. L. and and notes. with introductions. gilt top. K. B. 8vo. DialogUS de ScaCCailO. net. net. A revised text. insiuo. H. Historical Evidence. net. net. (id. E. C. I. The Crawford Small 4to. PLUMMER and 8vo. E. IHs. leather back. THOROLD ROGERS. PLUMMER. Small 4to. Small 4to. Handbook to the LandCharters. 5s. tid. Edited from the MS. Jewel. Introduction. EARLE.D. . W. Chronicon Galfridi Baker de Swynebroke. : Sources Two volumes. 7s. buckram. . . Small 4to. KINGSFORD. 6s. 3 Crown Svo. cloth. net. 1624 to 1874 8vo. Edited from the Twelfthcentury MS by F. Fortescue's Governance of England. with introduction. The J. Svo. edited by C. Edited by C. NAPIER and W. 8vo. 3s. H. Chronicle of John of Worcester. of the Saxon Chronicles Parallel extracts from the others. With illustrations and a map. R. IG. Revised Text. Vol. 12s. the Bodleian Library. 12s. CRUMP. The Alfred by J. 10s. by net GaSCOlgne's Theological Dictionary ('Liber Veritatum'): selected passages. The Song notes. with introduction Edited by A. 6d. EARLE. Text. Crown 8vo. net. 12s. Neot. edited. Of Lewes. MAUNJDE THOMPSON. PLUMMER. of early Charters and Documents. 6d. with introduction and notes. l()s. By J. Asser's Life of Alfred. now in Edited by A. Svo. and and index. 8vo. Ct Miracula Beati Olaili. By H. by Sir E. with a folding map of London in 1600 (by EMERY WALKER and H. cloth. Crown Vol. 6d. CRIBB) and other illustrations. net. with introduction and Extra fcap 8vo. leather back. C. Crown 8vo. C. X'2 2s. 2 vols. jl Is. notes.
based on Bumet's History of that of M. 12. 186976. The Constitutional Third edition. etc. with Fell's Life of Dr.H. net. GARDINER. Whitelock's Memorials. derived from Burnet's Memoirs. net. MADAN. 18s. FOXCKOFT. 6d. From 1625 to 1660. (17391810. 10s. JEI ios. etc Of Charters and Rolls preserved Calendar 8vo.w. svo. In three volumes. 6d. J. 1 5s. FIRTH.net. 6d. 9s. First Series.6d. By Svo. Memoirs of James and William. 16s. net. III. CLARK. Vol. History of Oxford A complete list of the Publications of the Oxford Historical Society can be obtained from Mr. tion. Small 8vo. 2s. Calendars. 8vo. W. C. hitherto unpublished. 6d. Ludlow's Memoirs. contained in the printed catalogues of the Bodleian and College Libraries. ua. From 1655 to 1657. Autobiography.. 16s. Hale.The Clarendon Press From the Eighth edition. 1468 "1610. to 1654.net. From 1649 14s. 6d. 10s. 8vo. II. R. 7s. I. 1 in the Bodleian Library. Hvo. (1625 1672. 6d.s.) 4vois. By S. 18s. and illustrations. svo. PHOTHF. With A Bibliography of Printing and Publishing notes. Bibliography Cotton's Typographical Gazetteer. 6d. Hakluyt's Principal Navigations. Select Charters and other illustrations of English Constitutional History. each 12s.HEwms.RO. F.) Ed. Time. Manuscript Materials relating to the History of Oxford . etc By Bishop STUBBS. 1625. in 11s. From 1523 to January 1649. 1 16s. preserved the 8vo. Two vols. Svo.) Ed. 2vols. 8s. Burnet's History of James II. 6d. Third edition. by OSMUND AIRY. I. i2s. ' LllttrelFs Diary. 7s. 6d. appendices. Life of Sir M. all Edited by H. Select Statutes and other Constitutional Documents of Crown 8vo. 8vo.' set down between the Years 1669 and 1696. 6d. Two volumes. Series of Charters. C. 8vo. The Early Oxford at Oxford. Vol. FKOWDE. A new edition. Documents of the Puritan RevoluCrown 8vo.) Aubrey's Brief Lives.A. Hvo. 8vo. 1 10s.) Six volumes. (Seep. ROIJTH. (lessieeo. 8vo. From 1558 to Crown the Reigns of Elizabeth and James 8vo. Dukes of Hamilton. Hammond. Supplement. . C My Own The Whitefoord Papers. 1902. By F. earliest times to 1307. net. Svo. MATJAN. By G. net. Edited from the Author's MSS by A. Vol. * Press. (l6781714. Calendar of the Clarendon State Papers Bodleian Library. Statutes.
net. S. G. STUBUS. . Development. the By A. P. R. Crown 8vo. 2s. by W. PRICE. with many illustrations. VI (Index). subjects. The Reign By William Rllfus Two volumes. Norman Conquest Extra fcap 8vo. 2s. By O. By E. 1 of England and . Durham. EDWARDS. Hampshire. 8vo. revised and History of the and Results. Seventeen Lectures and kindred By the same. CLARKE. EDEN. Three volumes. illustrated.) Berkshire. Vols. History of the By Of the same. in its Origin and By W. I. Stories History of Berkshire. by E. Demy Also in three volumes. 6d. 6d. od. 18671884. its Causes By Vols. crown Hvo. Edited by F. 6'd. 6d. 3s. enlarged. Constitutional History of England. S. III Norman Conquest E. net Oxford County Histories Crown 8vo. Perspective History Chart. LAMBORN. A. Essex. price 12s. WESTON. each Stories from LAM BORN. 6d. With 97 Crown School History Of England plans. By E. 8vo. Is. H. each (>cl Is. halfroan. G. 2 8s.Bishop Stubbs's and Professor Freeman's Books The 8vo. (Middle Ages). A Short First. out of print. each. Leeds and its Neighbourhood. LAMBORN. A. 1900. (Also in superior bindings. net. LIDDELU Others in preparation. 6s. G. School Books Companion BARNARD. to the death of Victoria. RAIT. 8s. 8s. Oxfordshire. Vol. By JOHN IRVING. etc. c. Library edition. Third edition. net. from the History of Oxfordshire. Also. and the Accession of Henry the 1 1 the same. C of England. FREEMAN. A. for junior pupils. Crown 8vo. 6'd. by H. II V (English edition) are U)s. 8s. Is. A. by F. M. to English History illustrations. ss. each. by F. on the Study of Mediaeval and Modern History Third edition. and others. 6d. With maps. and IV. A.
Villainage in England. net. BOASE. 5s. 8vo. The Welsh Wars military history. R.) 1 17s. for 1901. In two volumes. with Portrait. 8vo. 6s. William MARKHAM. I. B. Life and Letters.] 8vo. I. 18s. By in English Society in the Eleventh Century. VINOGRABOFF. of Clarendon. Is. TTE Edited by in Social and Legal History. A. being the Ford Lectures By C. W. RICE and Biographies By Ancient Britain and the Invasions of Julius Caesar. MACKAIL. a contribution to mediaeval By J. Six volumes. Edward Hyde. P. by W. 8vo. HOLMES. By L. (A. 5s. volumes. 8vo. Biographical Memoir of Dr. Post 8vo. 4s. STEBBING. MERRIMA*. f. 15361540. 8vo. 8vo. 8vo. Vol. A lecture by c. 9s. By C. Is. GROSS. index. ZUUJETA. Patronage under the Later Empire. FIRTH. by Sir CLEMENTS The Life and Works of John Arbnthnot.B. W. By ADOLPHUS BALLARD. Sir Walter Ralegh. with Index.C. . net. a contribution to British municipal history. H. 21s. 8vo. 12s. net. etc. cloth extra. net. An KEN. Oxford Studies By F. The Great Revolt 8s. net. Index separately. 8vo. etc. By the foundation of the Oxford Chair of Poetry. of Edward 8vo.D. <>d. 8vo. English Monasteries on the Eve of the Dissolution. 6d. and C. K. 15231535. 8vo. II. Vol. Essays By P. English Mediaeval History. A History Of England. 6'd. 8vo. PLUMMER. With two maps. Translated under the superintendence of G. 6d. Essays in English Mediaeval History. : The Gild Merchant C. 6 By G. E. VINOURADOFF. 6d. Earl 8vo. 16s. By Sir J. 1 By Two volumes.OMAN. Arch bishop of York. net Birkhead and Henry W. net. leather back. I . 16s. [Vol. KJTCHIN 3 3s. 16s. 6s. 8vo. Letters. net. Life and Times of Alfred the Great. 6d.net. MORRIS. 13991485. VON RANKE. principally in the Seventeenth Century. H. By ALEXANDER SAVINE. By Two Life and Letters of Thomas Cromwell. Lancaster and York. Index separately. Of 1381.Special Periods T. net. Markham. net. net. net. PAUL VINOGRADOFF. Is. notes. net. a Biography. Is. The Domesday BorOUghs. RAMSAY.
photogravure portraits. 14941610. Twovols. Already published Talavera. By F. illustrative (See p. European History Historical Atlas of GEORGE. 12s. History Of France. Cr. By C. 8vo. 2s. Ed. FISHER. 8vo. GEORGE. 6d. H. L. By I. Crown Bvo. net. Atextbook of European 3s. 6'd. SWIFT. 59. Vols. III (1795). net. Oblong 4to. FISHER. HEADLAM. plans. mainly in the By A. Vol. By A. III. HASSALL. net. The 17631815. F. L. Thiers' 6 maps. WARD. W. 17891791. II. History Of the Peninsular War. Crown 8vo. Documents of the French Revolution. 8vo. Genealogical Tables of Modern By H. Crown 8vo. L. Vol. 6s. Six lectures by H. Two lectures by A. B. (to 1453). etc. A Fall of the Old Order. 1809. II (1624). 8vo revised. F. Two volumes. Modern Europe. TANNER. 8vo. SMITH. 12s. I. : Two volumes. Crown Hvo. A. A textbook of European History. Two volumes. to Vol. 12s. W. net. Napoleonic Statesmanship 8vo. to Torres Vedras. 17891795. PLUNKKT. Speeches of the Statesmen and Orators of the French Revolution. with introductions and notes. Maitland. each. SMITH. The Life and Times of James the First of Aragon. 10s. MORSE STEPHENS. Memoir. 3s. to Corunna. with portrait. 12s. 6d. by A. stiff : a Short 2s. . 8vo. facsimiles. with maps. net MOSCOW Expedition. De Tocqueville's L'Ancien Edited. net. By L. net each. 5s. B. KITCHIN. 4s. 6d. 1 Is. with many maps. By WICKHAM LEGG. By E. 6d. Bonapartism. 14s. By H. Germany. boards. L. w. by F. 6d. : AngloChinese Commerce and Diplomacy nineteenth century. W. Frederick York Powell. By J. 7s. net. boards. SARGENT. Crown Bvo. Fourth (1904) edition. Regime et la Revolution. 8vo. Crown 8vo. 180910. net. with 10 maps and plans. OMAN. : be completed Vol. (35s. by G. Great Britain and Hanover. and portraits. A. 5s. David Binning MonrO 8vo. J. I . A Life and a selection from his Letters By OLIVER ELTON. To in six volumes. L.Life and Letters of Sir Henry Wotton. G. 18071809. Od. Cr. 6d.) History. UHUUHAHT. D. and Occasional Writings. with 8 maps. edited by H. The Renaissance and the Reformation. COOK WILSON. 8vo. With 21s. History. M. 6d. By G.
1905.C. Canada. IBs. dencies. 8vo. 35s. 6d. Bk. 6's. Aboriginal America.History and Geography of America and the British Colonies For other Geographical books. Geography of Africa South of the Zambesi. covers. 1906.M. 8vo. An inaugural lecture delivered on April 28. II. CornewallLewis's Essay on the Government of DepenEdited by Sir C. graphical. 8vo.G. British Colonies. LUCAS. K. Geographical. maps. BRAND (1909). 6d. 6d.O. The Discovery. P. Parti. paper Europe and her Colonies. Historical. Part II. by H. Historical Atlas. p. Introduction. 6d. Part II. With 13 maps. to date. 1891. 7s. 8vo. 3s. Part III (Geographical) in preparation. Is. Aboriginal America (concluded). Bk. P. Bk. net. EUERTON. 7s. 1763 1812. 1892.S.C. Historical and Geo9s. IV. East Africa. With eight maps. 6d. Vol. Vol. PAYNE. quarterbound. Crown 8vo.M.M. EGERTON. With 22 maps. 8vo. net. Geography of the 8vo. By Sir c. By J. lls. A Vol.G. ROGERS. Crown By the Hon. Australasia. 1903. V. The Ullioil Of 8vo. By sir c. The Canadian War With eight maps. of 1812. D. 2s. Colonies. 6d. GRESWF. By the same author. 1901. J. E.I/L. 6d. 27 maps. to date. H. I. I. net. The West Indian edition. I. In cheaper binding. II. EOERTON. Geography of the Dominion of Canada and Newfoundland. 12s. 6s. Part II. Second and brought up Vol. end of 1899 by H. 4s. 6d. by H. Also Part I. . Crown 8vo. 12s. 7s. Part II. K. 3s. K. Vol. Second Edition. Historical. Historical LUCAS. 6s. K. ATCHLJEY. By the same author. 4s. West Africa. By Sir c. R. VI. South and With eleven maps. Revised to the With five maps. Gd. Geographical. History of the New World called America. By E. 7s. 1898. II. (Origin and growth of the Colonies. see page 59. With maps. 6d.G. net. E. and brought up by 5s.M.G. By W. Vol. (id. revised Colonies. Also Parti. With ten maps. I. net.) With eight maps. History of the Dominion of Canada. Part I. P. 14s. 1903. 6d. New edition by H. edition. Crown Bvo. 3s. The Mediterranean and Eastern Second 1906. 6d. II. p. History of Canada. E. Vol. revised 7s. E. E. R. 8vo. EGERTON. 6s. STUBBS. South Africa. (Jd. LUCAS. Vol.C. with twelve by C.C. III. 1907. The Claims of the Study of Colonial History upon the attention of the University of Oxford. 6d. LUCAS.
COTTON. W. OSWELL. 6d. BIIADSHAW. C. Ill. Crown is 8vo. SETONKARR. II.G. J. By Sir C. CUNNINGHAM. By Captain L. By J. each Vol. By L. Alirangzib. net each. Ab The Earl of Auckland. Marquess of Hastings. 2s. By S. Mountstuart Elphinstone. Albuquerque. Lord William Bentinck. IV. Lord Lawrence.) Babar. By Captain L. Also in two vols. K. J. Earl Amherst. By MajorGeneral Sir O. VlSCOUnt Hardinge. By Sir W.Rulers of India Edited by Sir W. By D. net separately. D. By Sir W. By The Marquis of Cornwallis. Madhava Rao Sindhia. RITCHIE and II. BOULGER. 9 . T. and Tipu Sultan. Colonel MALLESON. KEENE. HUTTON. T HOTTER. By Major ROSS~OFBLADENSHUR<. SMITH. By Sir R. 4s. BOWRIN(. Sir Henry Lawrence. AITCHISON. Lord Clive. 1909. net. S. H. (There also a special Indian Edition. Abridged from the Rulers of India Vol. W. By Colonel MALLESON. TROTTER. . GRIFFIN. The Earl of Mayo. HUNTEK. B. MORSE STEPHENS. HUNTER. Sir Thomas Munro. Vol. Sketches Of Rulers of India. <>d. (>d. The Marquess of Dalhousie. James Thomason. LANEPOOLE. The Company's Governors . By ANNE T. Warren Hastings. By W. Viscount HA R DING L. The Marquis Wellesley. Vol. By By S. 2s. J. The GovernorsGeneral . S. By his son. EVANS. net each. COLVIN. By V. G.. Haidar H. By Sir A.. . TEMPLE. By Sir L. BIRNE Earl Canning. A. Akbar. General J. Second edition. by G. net. By W. MLEOD INNKX Clyde and Strathnairn. HINTEK. By Lieut. By J. Asoka. S. 7s. The Mutiny and After . John Russell Colvill. LANEPOOLE. Ranjit Singh. 3s. Dupleix. Crown 8vo. By H. By Sir H. By Colonel MALLESON. W. The Princes of India. I.
6d. Wellington's Despatches. cloth 4 4s. 6d. Brief History of the Indian Peoples. net. By V. Vol. A. Historical. By Sir J. HUTTON.. 1 4s. Treaties.. Is. Student's History of India. Second Edition. Economic. Alphabetical Gazetteer. Each volume contains a map of India specially prepared for this Edition. revised 8vo.C. III. Atlas. (Published for India in Council. W. 163539. P. Vol. By W. 10s.5 5s. With maps. XXV. The Oxford Crown 8vo. morocco back 7s. Eightyninth thousand. FOSTER. Reprints from the Imperial Gazetteer. By Sir JOSEPH HOOKER. 6d. net each.FOSTER. to 163. net . Bvo. 6'd. maps and 11 other illustrations. East Indies. By sir w. Med. Is. net each. VXXIV. covers. OWEN. b'd. 6d. entire work in 26 vols. 8vo. 6d. STRACHEY.svo. Introduction by By E. of 'The Indian Empire' separately. 10s. published by the Public Record S. net. The Government of India. 16249. 1907. 164043. net. net. and other and enlarged. The English Factories in India: Three Vols. IV. Two Vols. net each. the remaining 21 vols. H. Vol. plans. net. SMITH. and India. Hastings and the Rohilla War. Second edition. cloth 6s. J. Vol. cloth 15s. cloth 5 net. to the Muhammadan Conquest. price 15s. '2s.. Administrative.The 4 vols. morocco back 17s. morocco back 6 6s. with historical introduction and illustrative documents. 12s. Wellesley's Despatches. under the patronage of His Majesty's Secretary of State 161821. w. A sketch of its History and Organization. SAINSBUHY. net. Atlas. SMITH. each volume. 11s. Vol. 16223. I. illustrations. net. morocco back . Treaties. net . net. other Papers relating to Selection edited by S. 1 4s.5 have The Court Minutes previous been calendared in the Calendars of State Papers. 8vo. Vol. Second edition. Descriptive. Index. Med. Paper A Revised up to 1903 by W. II. to his Selection edited by OWEN. 3s. including the invasion of Alexander the Great. 8vo. A sketch of the Flora of British India. By Sir The Early History of India from 600 B.) Court Minutes of the East India Company. being a digest of the Statute Law relating thereto. A. 8vo. and other Papers relating Government of India. . C. 8vo. Paper The Indian Army.. B. ILBERT. J. (The six previous volumes of Letters received by the East India Company from its Servants in the East (16021617) may also be obtained. covers. Office. With 7 By v. 6d. Vol. HUNTER. XXVI.) 12s.
IV Asia.. 3s. Is. : The Dawn In one volume. Is. 7s. 21 UNSTEAB. The Oxford Geographies 6d. with Statistical Appendix by E. lOd. By EDUARD SUESS. Ed. Ill Europe. Is. Regions Of the World. Central Europe. 6d. Crown 8vo. F.D. I. net each. 2. VII The British Isles. Vol. Ed. net. 8vo. The Face Of the Earth. D. 6d. 72 maps and diagrams. HUGHES. J. Crown 8vo. 6d. tfd. with letterpress to each the maps printed by W. etc. Frontiers: Romanes Lecture(l907) by Lord CURZON OF KEDLESTON. POOLE. 6d. With Physiographical Introduction. net each . Prospectus on application. 5 15s. Geographical Memoirs under the general Medium 8vo. Physiographical Introduction (In the press. 3s. III. Others in preparation. HERBERTSON. at various prices from 30s. by A. halfpersian. u and in. By D. GEORGE. Vol. 2s. B. II In and About our Islands. 2s. Part 2s. Questions on the Senior Geography. 6d. n. Ltd. By ARCHIBALD LITTLE. By A. North America. J. and the whole edited by R. R. net. Questions and Statistical Appendix. K. Crown 3. (A. 8vo. Is. School Economic Atlas. By JOHN PARTSCH. 3. imperial 4to. to 35s. KIHK. India. By F. . 2s. volumes. RUSSELL. : : By F. from the Decline of the Roman Empire. editorship of H. 2s. 6d. Vol. Crown 8vo. By J. 2s.. Fourth edition. Is. Nearer East. JOHNSTON. The Oxford Geographies. 900). maps and With is. See p. and maps and diaWith Questions 117 Statistical (In preparation. Geography for Schools. W.I>. Ed.} to Vois.) Vol. The Preliminary Geography. Crown 8vo. together Relations of Geography and History. Vol. Is. J. In three 2 10s. Not sold separately. or in selected sets British Empire. in. The Far By East. 9001260). 20s. 92. or in single maps. 6d. BARTHOLOMEW. each Part Is. 6'd. (id. 4to. 166 grams. maps and diagrams. HERBERTSON. HOGARTH. net. M. 1 (to A. J. 90 maps. Appendix. 2. 6d. net. G. net . Britain and the British Seas. By J. 6'd. BEAZLEY. Sir THOMAS HOLDICH. 4s. Ed. of Geography. . TAYLOR. Introduction by L. MACKINDER. Vol. The Elementary Geographies. I. Is. net per volume. MACKINDER. with maps and illustrations from photographs. With two maps. L. I Physiography. G. & A. maps and diagrams.GEOGRAPHY Historical Atlas Of Modern Europe. E. 2s. By II. : : : Practical 27 Geography. By H. diagrams. With Physiographkal Introduction. The Senior Geography. II. By C. 6d. R. <>d. The Junior Geography. By H. II Modern 15s. LYBE. 3s.
edited by J. BEAU6's. LANEFOX PITTRIVERS. With an appendix on the Celtic Pony. By N. their With introduction by Language and Sir Folklore. by the late LieutGen. A. 12s. Hvo. 3 3s. net. Hindu Manners. and other Essays. A. 16s. RANDALLMAC:!VEII. Hvo. MYRES. 2 vois. net. Paintings. By R. net. Crown Bvo. Edited by R. Portfolio. L. A. with an Introduction by H. . ANNANDALE. with 21 plates. CHARLES ELIOT. B. Abydos. c. net. 6d. F. HOLLIS. net. net. 21s. HOLUS. CHARLES ELTOT. net. is' CHAJVJP. (A cranioiogicai study. net. Royal 8vo. 6d. net. TOKGUE. CODRINGTON. 6 lithographic charts. % Anthropology and the Classics. RANDALLMAC!VEK. The Nandi. svois. 14s. RHYS. 8vo. 6 d. net. Illustrated. with 6 collo49s. Studies in the Arthurian Legend. 8vo. Illustrated. K. BALFOUR. net. Celtic Folklore: By WALTER JOHNSON. BALFOUR. Crown 8vo. EVANS. On India Paper. 10s. W. their studies in their Anthropology 16s. FOWLER. in colour. Anthropological Essays honour of his seventyfifth birthday. 7s. With introduction by Language and Sir Folklore. I\. FolkMemory. By J. Copied by M. By ARTHUR THOMSON and D. H. The Evolution of Culture. 6s. and many other illustrations. G. H. The Ancient llaces of the Thebaid an anthropometrieal study. i2s. presented to EDWARD BURNETT TYLOR in Imperial 8vo. 25s. 8vo. Bushman With a preface by H. G. MARETT. 4s. : The By Earliest Inhabitants of D. Welsh and Manx. net. Imperial 4to. By A. net. (Jd. types. Bvo. six lectures J. and FolkLore. LANG. W. The Melanesians. MARSHALL. by F. By A. 8vo. i is. Iceland and the Faroes. The Masai. net. A. 6d. H.Anthropology Transactions of the Third (1908) International Congress for the History of Religions. and printed In a box. fid. MURRAY. 7s. c. by A. Illustrated. 8vo. JEVONS. Third edition. MYRES. Illustrated. L.. By J. RHYS. Translated and edited by H.
net. with an historical introduction by A. by J. 6d. I (fourth edition. Bentham's Introduction to the Principles of Morals and Legislation. L.LAW Jurisprudence Bentham's Fragment on Government. 6d. 6d. Hvo. 6d. Revised and enlarged by E. GREE^IDGE. Two volumes. 6d. MOYLE. H. net. By EHNEST J. Legal Procedure in Cicero's Time. Vol. No. Hvo. Fourth edition. Second edition. Crown 8vo. II. Is. Sixth edition revised. E. E. . A. 2s. The Elements edition. Extra fcap Hvo. JAMES BRYCE. SCHUSTER. Tenth 1906. sold in parts. GIUTETJER. B. 1. MARKBY. 6d. Translation 8vo. . with an introductory essay by E. H. E. S HARWELL. 1903). 8vo. 7s. 2. Part II. 1901. Studies in History and Jurisprudence. GRUEBER. 6s. E. title L Contract of Sale in the Civil Law. 16s. and translation. 4s. Institutes of Justinian. c. HOLLAND. Second edition. J. l(>s. 6d. Introductory Titles. 6d. edited as a recension of the By T. the Right Hon. Svo. GREENIDGE. and C. Jurisprudence. 12s. 190. Edited by K. 8vo. Part III. Third edition. Vol. 1907.C. 1907. 2). HOLLAND Also. J. : Law. los. place in Roman Svo. 6d. net. considered with reference to Principles of General K. 8vo. The Principles of Svo. 3s. MONTAGUE. H. Hvo. net. Public and Private Law.5. with an introduction to the study of the Corpus luris Civilis. Two volumes. WHITTITCK. 10s. Roman Law Imperatoris lustiniani Institutionum Libri Quattuor. Law of Obligations. (fourth edition. with introductions. Gai Institutionum luris Civilis Commentarii Quattuor : with a translation and commentary by the late E. 16s. By 1 5s. : GREENTDGE. Elements Of Law. Property Law. in paper covers Part I. C. Hvo. A. Part IV. by R. O'd. By E. I2s. By Sir W. commentary. By T. The of Gaius. MOYLK. Institutes 5s. 2os.ed. 6s. Select Titles from the Digest of Justinian. Institutes of LEDLIE Svo. 10s. HOLLAND. 6d. 1906).I. By A. By J. 10s. Family Law. of Jurisprudence. Bvo. Soira. POSTE. 11s. No. its Roman Translated by J. German Civil Law. : By T. The Roman Law of Damage to Property on the ' being a commentary of the Digest * Ad Legern Aquiliam (ix. net. 2s. By J. B.
1907. Parliament. 6'd. P. LATTER. all Cases illustrating the Principles of the By F. Third edition. Agency in 1910. by C. S. 6d. R. In two volumes. PLUMMER. Outline of the Law of Property.C. VON JHERWG. 8vo. By 8vo. Law in Daily Life. Sir F. . 6d. 6d. Crown 8vo. J. EVANS. Methods and Forms. 2s. ANSOK. 6d. with introduction. E. F. RADCLIFFE 12s. 10s. Edited by G. M. Y. Crown 8vo. ea. STUBBS.HAM. the reigns of Elizabeth and James Third edition. Hs. DIGBY. Vol. net. 6d. Calendar of Charters and Rolls. 6d. Modern Land Law. ILBERT. GOUDY. etc. Villainage in England. 10s. 6d. leather back. Law and J. i5s. and other Saxonic Documents. 12s. Hvo. I. 8s. 16s. Common Law. 8vo. EARLE. EDWIN CANNAN. Law of Real Property. Arranged and edited by W. Crown 8vo.6d. 6d.net Part II. Crown Hvo. I Crown Hvo. By E. Cases cited. P. Fifth edition. net. Vol. By J. 6d. 3s. ANSON. Welsh Mediaeval Law translation. MILES. 16s. Fourth edition. Twelfth edition. Constitutional by Documents of the Puritan Revolution. net. net. WRIGHT. 8vo. GWYER. Law of Principles of the English Law of Contract. illustrative of Documents other Illustrations of English Constitutional History. 16s. of the Constitution. L. Bodleian Library. 12s. C. Parti. GWYEK. I. 10s. Hvo. edited S. Constitutional Documents. net Text. 1908. The Management BlC. JENKS. with table of 1904. R. The Crown. net. net. of Torts. Text revised and edited.English by M. net. : by A. RALEIGH. Legislative 1901. from the earliest times to Edward I. By T. and its relation to Contract. 12s. 1909. Crown 8vo. By RUD. K. By VINOGRADOFF. Constitutional Select Charters and Eighth edition. By JOSEPH KING. Gd. GARDINER. W. 8vo. of Private Affairs. TS. L. A. T. 1 containing those preserved in the Ms. 1900. in the 8vo. 8vo. M. 10s. net. Essay on Possession POLLOCK and Sir R. selected and R. W. C. K. By Sir W.S. By sir c. 8vo. 6d. Handbook to the LandCharters. WADE the Laws of Howel the Good. BllIDGE. R. Translated with Notes and Additions by H.L 8vo. Od. revised Law and Custom By Sir w. II. net. Crown 8vo. Fortescue's Difference between an Absolute and a Limited Monarchy. etc. Select Statutes and other PROTHERO. Hs. Third edition. 8s. Introduction to the History of the By Sir K. S.
Sir C.ed. 6d. In time of peace. 1885. 8vo. With map. II. 8vo. 10s..International International 1909.C. ILBERT. being a manual of the LandTenures. The Indian Evidence Act. E. Hvo. with historical introduction and illustrative documents. Three volumes. Pacific Blockade. being a Digest of the Statute Law relating thereto. W. net. AngloIndian Codes. 8vo. '2 12s. Law Sixth edition by J. (1907. Second edition. JENKYNS. revised by T. 4s. ILHERT.I. halfmorocco. Law. Substantive Law. in paper covers. The LaWS Of War On Land. leather back. 2s.. net. By A. E. 12s. W. edidit T. E. 6d. supplement. 15s. K. Gd. Cloth. The Law of Nations.K. 2nd supplement. By the same. By Sir C. HALL.I. et d'Etudes sur ie Droit Coutumier de 1'Empire Ottoman. 8vo. BADENPOWELL. Twiss. net. ATLAY. 10s. K. K. with a preface by of the author. net. Hvo. revised and enlarged. E. and of the systems of LandRevenue administration. E. leather back. C. 3 3s. MAHKBY. Frowde). net. K. Hindu and Mahommedan Law By Sir the use of students.) Crown Hvo. By W. for An Introduction to 1906. vol. by Studies in International Law. 4 4s. LandSystems Of British India. P. 1898. 1 Treatise on the Foreign Powers and Jurisdiction of the British Crown. 15s. 2 17s. E. cloth. 6d. ios. HOG AN. 6d. Seas. Parts I (Vols* price per part. 6d. LandRevenue and Tenure in British India. H. with introductions and notes. 6d.S. HALL. 14s. net. Small quarto. 6s. 6'd. Edited by Sir C.E. with map.I.B. B. paper covers.C. Vol. P. net. 8vo. 8vo. II (Vols. P. Gd. 1905. 6s. By Is. In one 1 15s. MARKBY. ByT. lure Belli Libri Tres 1 Is. Par GEORGE YOUNG. net. 1877. HOLLAND. 1907. New edition. Cs.C. net.S. By B.C. HOLDERNESS.E. HOLLAND. 8vo. a collection of treaties and other public acts.E. with notes by Sir W. Corps de Droit Ottoman vols. : Ordonnances et Actes les plus importants du Droit Interieur. Second edition. W. 4 14s. 8vo. Lois. By T. 8vo. By Sir T. 3s. IVVII) can be obtained separately. Seven 8vo. . 1908. Part I. HOLLAND. net.I. 8vo. K.I. and in cloth. 5s. Adjective to 1891. LUCAS.M. vol. by WHITLEY 1 STOKES. The European Concert in the Eastern Question. 10s. and a CornewallLewis's Essay on the Government of Dependencies. By the late portrait H. 1st I.G. C. 6d. T. Gentilis Alberici de HOLLAND. Law. 1908. British Sir Rule and Jurisdiction beyond the 1902. C. Colonial and Indian Law The Government of India. I III) u Recueil des Codes. net. net (published by Mr. R&glements. 8vo. Edited. 6s.
BhmtSChli's Theory of the State. Lloyd's Prices of Corn in Oxford. M. By H. Second impression. Police. 8vo.nct. THOBOLD ROGERS. 8vo. 6d. Letters to Trower and others (isnisss). By w. 8s. ed. Cr. 8vo. RALEIGH. net. net. E. By w. paper. H. L. ROGERS. Crown 8vo. Edited by j. with introduction by Crown Adam 8vo. CANNAN. VII. I and II (125914001 8ks. By E. 3s. stiff Political Is. Extra fcap 8vo. First Nine Years of the Bank of England. BONAR and J.B.011582). the Romanes Lecture By J. Extra The Study of Economic History. 3s. 8vo. 6s. H. Vols. BRYCE. 2s. net. 8s. Is. net. 5s. 15831830. 7s. The Elements of Railway Economics. 6d. Revenue and Arras. L. Crown 8vo. Vol. Edited with introduction and notes by E. By Sir T. CANNAN. net. Third edition. C. A method of explaining some Theories of Pure Economic Science by diagrams. net. covers. Gd. 2s. By J. Vols. net. By A. and 17th Centuries. 7s.Political Science and science. 10s. CornewallLewis's Remarks Of SOme Political Terms. 6d. Edited by J. Sixth edition revised. Elementary Politics. HOLLANDER. CURTLEH. K England. D. 6d. Ill and IV (14. 1901. 32s. 8vo. T. 4s. (Jd. By PUKE. net. edition. 8vo. Economy political For Bryee's Studies and other books on general jurisprudence and Industrial Organization in the 16th By G. 8vo. cloth. 2s. By j. net. 7s. see p. and VI (15831702). Economy. . CUNYNGHAME. Economic Documents Ricardo's Letters to Malthus (isioisas). 6d. Translated from the sixth German edition. Treatise on the Geometrical Political Economy. Vols. net. in 12591793. of the Small Landowner. is. net. 32s. Crown 8vo. In two Parts (17021798). on the Use and Abuse New . net. Relations of the Advanced and Backward Races of for 1902. 8vo. Third edition. is. The Disappearance JOHNSON. 6'd. net. Mankind. 61. Being an elementary Elementary fcap 8vo. R. ACWORTH. Crown 8vo. RALEIGH. Smith's Lectures on Justice. V History of English Agriculture. 8vo. BONAK. THOHOUJ History of Agriculture The History of Agriculture and Prices A. 6d. UNWIN. H. net.