Jniv.of iji. Library ■• ■i;vr> •ft;'- r>- 2C *. * ■ ' i.> jT ■ . V'. > 5 "V"' ■c •'. "• '.-V • \ '‘ V* 1 ,»^ -;■ • ^ ,i--:V***^'r ^-i:. . * , 1 ? “Ve.»ii>j^L*i i Vi -. ' -^^ • ♦ V ** ' S^ ^ V • • *? -IT*;- * Jy \ 't'-'i-'' &r-«; t!.; ’■ ”' ■ ■ ” • • V • • • » • . T», .V” r *, ^ - -?c^ >>»»3 i yi^- ■-f'fe' ' ■ * i r. A * ■'V • . V * ON THE CORRELATION OF TWO PLANES.* BY T. ARCHER HIRST, F.R.S. Definition and Determination of a Correlation* 1. A correlation is said to be established between two planes, when their points and right lines are so associated, that to each point in one of the planes, and to each line passing through that point, respectiv'ely correspond, in the other plane, one line and one point in that line. 2. It will be convenient to apply the familiar terms ijole and j^olar to a point and its corresponding right line. 3. It has been shown by Chaslesf and others, that such a correlation may be established, and that in one way only, when four points, in one plane, and their polars, in the other plane, are given ; provided always that no three of the points are collinear, and no three polars concur¬ rent. In fact, if Dj be the four points in the first plane, and Cg 1^2 ^2 fl^^ir respective polars in the other, then, being the polar of any fifth point Xj, it follow^s at once from the definition of a cor¬ relation that the pencils A^ (B;^ Xj^) and B^ (Aj Hj Xj) must be equi-anharmonic, respectively, with the rows (bo dg x^) and %''4''^’ 2 )' Hence, by a well-known method, if be given, the intersections of its polar x^ with Cg and and hence x.^ itself, are readily found. 4. From the definition of a correlation it also follows that if the polar of A| pass through A 2 , the polar of Ag will pass through Aj ; and similarly, that if the pole of lie on Og, the pole of will lie in Oj. Hence we may term Aj^ and Ag conjugate joints^ and and conjugate lines of the correlation. 5. That in any correlation two given points, or two given right lines, * Of the two questions, CorroLition and Ilomography, the first has heen selected fjr special consideration in this paper. With suitahle modifications, however, as shown in the foot-notes on pages 22 and 23, the results are all applicable to the second question. t “ Geometrie Superieuro,” art. 581. B 2 shall be conjugate to each other, is obviously to be regarded as one condition ; that a given point and line shall bear to each other the re¬ lation of pole and polar is, accordingly, equivalent to two conditions. Hence, and from art. 3, we may infer that eight conditions are neces¬ sary and sufficient for the establishment of a correlation between two planes.* 6. The problem to determine a correlation between two planes which shall satisfy any eight given conditions, is susceptible, in general, of a finite number of solutions. The determination of this number, when the conditions are of the elementary kind, single and double, described in the last article,—a necessary step towards the solution of the more general problem,—is one of the objects of the present paper. Systems of Correlations. 7. Two planes may obviously be correlated in innumerable ways so as to satisfy seven given conditions ; the totality of such correlations constitute what may be termed a system of correlations satisfying seven conditions. For example, if the polars % ^2 three points Pj were given, as well as two conjugate points and Ag,—a set of seven con¬ ditions which may be conveniently indicated thus : Pi Qi Rj All P 2 % »2 ^2) . then, in general, any line a^ passing through Ag may be regarded as the polar of Ai, and a correlation established, in the manner indicated in art. 3, which shall satisfy the eig^ht conditions, Pi Qi Ri Ai Ih 9.2 '"'2 ^2 and therefore, dfortiorf the seven given conditions (1). By giving to all possible positions, we obtain the several cor¬ relations of the system. We shall find it convenient to denote such a system by the symbol ( ). 8. It should be observed, however, that whenever «£ passes through the intersection of any two of the three given lines $'2 ’ 2 ’ vision alluded to in art. 3 ceases to be observed, and no correlation, in the ordinary sense of the term, can be established. 9. Such exceptions occur in every system of correlations. As * This also follows immediately from the “ expression analytiqne des figures cor- rclalives,” given by Chasles in the “ Geometrie Superieure,” art. 594. Expressed in trilinear co-ordinates, the result there arrived at is that \.pc + + U 2 Z = 0 will he the polar of if A 2 P'2 ’ Vg = ^Oj + tn^i + fiy-j I I'ai + + n*y\ I + w/Ai + n"yi, where the ratios of I, ni, n, I', n/, n', l'\ m”, n" are the eight constants of the correlation. 3 another example we may take the system / Pj \ \ P2 ^2 ^’2 ^2 / ’ wherein and a<^ denote given conjugate lines. The several correla. tioiis of this system are obviously obtained by giving to A^, the pole of fl' 2 ) all possible positions on ; but whenever comes into line with two of the three points the method of establishing a cor¬ relation described in art. 3 ceases to be applicable. 10. Instead of excluding cases such as those described in the last two articles, it will be found of the highest importance to admit them as exceptional correlations into the system satisfying seven conditions, and carefully to study their properties. The part they play in the general theory of correlation will be found to be strictly analogous to that played by degenerate conics in the investigations of Chasles on systems of conics satisfying four conditions.* Origin and Nature of Exceptional Correlations. 11. With a view of obtaining an insight into the nature and origin of exceptional correlations, we will first consider all the exceptional forms which a homographic relation between two planes may assume ; or, what is equally general, all the exceptional modes of putting two planes IT into perspective with each other. It is clear that, so long as the centre of perspective lies in neither of the two planes, the homographic relation between them is of the ordi¬ nary type; we have, therefore, merely to consider the cases where the centre of perspective lies, 1) in one of the two planes, and 2) in the intersection of the two planes. 12. If the centre of perspective be a point 2^^ in the plane IT^, and if cr' be the line in which the second })lane 0.' is intersected by llj, it is obvious, firsts that to every point M' of H', which is situated on a\ cor¬ responds, in ITj^, an indeterminate point on the line whilst to e^ery other point of 11' corresponds the point 2^^ itself; and, secondhj, that to the point 2^ corresponds a wholly indeterminate point of the plane iT, whilst to every other point of llj^ corresponds the intersection M' of 2j^M^ and a. With respect to the correspondence between the lines of the two planes, it is equally obvious, first, that to the line o' in 11' corresponds a wholly indeterminate line in 11^, whilst to every other line m' in II' corresponds the line through 2^ and (m'o') ; and, secondly, that to every line in Ilj, which passes through 2^, corresponds an indeterminate line in iT passing through the intersection (aqo'), whilst to every other line in corresponds the line o' itself. 13. If the centre of perspective 2^ lie in the intersection o' of the two * “ Comptos reiidiis d.s seances do rAcadcinic des Sciences,” 18G4. B 2 4 planes, Tli and FI', then to 2^, regarded as a point of either plane, cor¬ responds a wholly indeterminate point of the other plane ; to any other point of (t\ in either plane, corresponds an indeterminte point of the same line a in the other plane ; whilst to every other point, of either plane, corresponds the point 2^ itself. The lines of the two planes correspond in the following manner: to cf, regarded as a line of either plane, corresponds a wholly indetermi¬ nate line in the other plane ; to any other line through 2^, in either plane, corresponds an indeterminate line through 2^^ in the other plane; whilst to any other line, in either pl^ne, corresponds the line a itself. 14. Passing now from the perspective, to any other position of the two planes, we conclude that the liomograijldc correspondence between them may assume either of the folloiving two exceptional or singular forms. First. There may be a singular point in one plane, and a singular line in the other, whose respective correspondents are wholly indeterminate. To each point in the singular line will then correspond an indeterminate point in a determinate line passing through the singular point, whilst to each such line through the singular point will correspond an indeter¬ minate line passing through a determinate point of the singular line. It is of importance to observe that between the above-mentioned points on the singular line and the associated lines through the singular point there must, by art. 12, always exist an equi-anharmonic, or (1,1) correspondence. Secondly. In each plane there may be a singular line, and a singular point situated in that line, whose respective correspondents are wholly indeterminat-e. To every other point in a singular line will then cor¬ respond an indeterminate point in the other singular line, whilst to every other line passing through a singular point will correspond an indeterminate line through the other singular point. 15. If an exceptional homographic relation exist between two planes ITj and 11', and any ordinary correlation, such as that described in art. 3, be established between the latter, and a third plane fl^, it is evident that between the former 11^ and this third plane fig an exceptional cor¬ relation will exist; and vice versa., from any exceptional correlation, we can always pass, by means of an auxiliary ordinary correlation, to an excep¬ tional homographic relation. This principle enables us, readily, to deduce all possible exceptional forms of correlation from the results of the last article. IG. In so doing we must consider three cases, since the first case of the last article presents two varieties. First. The exceptional homographic relation between flj and fl' is such that in I lj there is a .singular point 2^, and in 11' a singular line F. The pole of cr', in the plane llo, being 2^, we .shall have, between lIi and flo, an exceptional correlation ivith singiUar points 2^, whose characteristic 1 roperties, easily deducible from art. 14, may be briefly described thus : The polar of a singular point is wholly indeterminate. The pole of every line tlirough a singular point is an indeterminate point on a determinate line tlirough the other singular pointy the two lines thus associated being al'ways corresponding rays of equi-anharmonic pencils. Secondly. The exceptional homographic relation between fTj and U' is such, that there is a singular line in and a singular point 2' in II'. The polar of 2', in the plane Fig, being 0 - 2 ? we shall now have, between the planes 11^ and llg, an exceptional correlation with singular lines and o-g, the characteristic properties of which, as deduced from art. 14, will be as follows ; The pole of a singular line is wholly indeterminate. The polar of every point in a singular line is an indeterminate line through a determinate point of the other singular line, the tivo points thus associated being always corresponding points of tivo equi-anharmonic rows. Thirdly. The exceptional homographic relation between 11^ and O' is such that in each of these planes there is a singular line, and a singular point situated on that line. Between the planes Oj and fig we shall now have an exceptional cor¬ relation with singular lines and points (in each line a point), of which the following properties are characteristic : The pole of each singular line, as well as the polar of each singular point, is ivhollij indeterminate. The polar of any point in a singular line, not coincident with the singular point situated therein, is an indeterminate line through the other singular point. An exceptional correlation with singular points is determined when the positions of those points, and of three pairs of conjugate lines pas¬ sing through them are known. In like manner, an exceptional cor¬ relation with singular lines requires, for its determination, a knowledge of the positions of those lines, as well as of three pairs of conjugate points situated thereon. Seven arbitrary conditions, of the kind de¬ scribed in art. 5, are necessary and sufficient, as we shall see, to deter¬ mine an exceptional correlation with singular points, or with singular lines. An exceptional correlation of the third kind, hosvever, that is to say, one which possesses singular lines and singular points situated therein, cannot, in general, be made to satisfy more than six such con¬ ditions. Hence it is that such exceptional correlations do not present themselves in the present paper; they have been described, solely, for the sake of completeness.* * Tt is well known, and has recently been again demonstrated by Schroter, (“ Journal f'iir die reine und angewaiidte Mathematik,” vol. 77, p. 105,) that two correlated planes can always be made to coincide in such a manner, that to each point the same line shall correspond, no matter to which of the two coincident planes that point be ascribed. The point and line, in fact, then become pole and polar relative to a conic. Now this conic degenerates to a line-pair, when the cor¬ relation has singular points ; to a point-pair, when it has singular lines; and to a line-pair-point, when it has singular points and lines. Hence the connection be¬ tween degenerate conics and exceptional correlations, alluded to hi art. 10. 6 17. The following properties of the two first kinds of exceptional correlations will be frequently referred to ; they are easily deduced from those already described. a. The pole of every line, in either plane, which does not pass through the singular point of that plane, coincides with the singular point of the other plane. b. The pole of every point, in either plane, which does not coincide with the singular point of that plane, is a determinate line passing through the singular point of the other plane. c. The polar of every point, in either plane, not situated on a singu¬ lar line of that plane, coincides with the singular line of the other plane. d. The pole of every line, in either plane, which does not coincide with the singular line in that plane, is a determinate point in the sin¬ gular line of the other plane. Ixelations between the Characteristics and Singidarities of any System of Correlations. 18. being an arbitrary point in one of two correlated planes, its polars, in the several correlations of any system, envelope a curve, w'hose class fx indicates the number of correlations of the system, in each of which the polar of passes through an arbitrary point Mg of the second plane. But since, in each of these correlations, (and in these solely) the polar of Mg will pass through M^ (art. 4), p will also be the class of the curve enveloped by the polars of an arbitrary point Mg in the second plane. Regarded as the number of correlations of the system, in each of which two arbitrary points M^ and Mg are conjugate to each other, p may be appropriately termed the class of the system of correlations. 19. In like manner, the poles of an arbitrary line, in either plane, lie on a curve whose order v may be termed the order of the system of correlations., since it indicates the number of correlations of the system in each of which two arbitrary lines and ??ig are conjugate to each other. 20. Between the characteristics p and r of a system of correlations, and the number of exceptional correlations which that system includes, relations exist precisely like those which have been established by Chasles between the characteristics of a system of conics satisfying four given conditions, and the number of degenerate conics included in the system. In fact, if TT denote the number of exceptional correlations with sin¬ gular points, and X the number of exceptional correlations with singular lines included in the system of correlatiuiis whose characteristics are 7 fx and V, then Hence we deduce and also yu = 2v — TTI V = 2fx —X ) \ + TT = + V X — TT = 3 (/U — y) Sju = 2X + 7r 7 = X + 27r 7 ( 1 ). ( 2 ), (3). 21. Since each of the relations (1) follows from the other by the Principle of Duality, a demonstration of the first will be sufficient. If and \ be any two lines in the first plane, and Mg any point in the second plane, we have to determine the number of correlations of the system (ft, v) in each of which the polar AgDg of the intersection passe's through Mg. By hypothesis, there are v correlations of the system in each of which the pole Ag of lies in an arbitrary ray of the pencil whose centre is Mg. If the pole Bg of in each of these V correlations, be joined to Mg by a ray m^, then to each ray will correspond v rays In like manner, to each ray will correspond v rays Wg, passing, respectively, through the poles of a-^ in the several correlations in which the poles of lie in m\. Hence there is a (r, v) correspondence between the rays and m^, and consequently there are 2r rays, with each of which an ^;^g and its corresponding Wa coincide. Although each of these 2v rays passes through the poles of tq and in one and the same correlation, they are not all polars of For instance, if and 2g be the singular points of one of the tt exceptional correlations included in the system, the poles of the arhltmrij lines and will coincide with Sg (art. 17. a.), and by so doing cause and 'W ^2 to coincide with MgSg. The polar of in this exceptional cor- • relation, however, will not in general coincide with this arhitranj line but with that ray of the pencil (Sg) which corresponds equi- anharmonically with the ray, through of tlie pencil (Sj), (arts. 16. first case, and 17. b.) Exceptional correlations with singular points being, clearly, the only ones in which the poles of a-^ and hi coincide, it follows, as indicated by the first of the relations (1) in art. 20, that the number /x, of correla¬ tions in which the polar of (ui?>i) passes through the arbitrary point Mg is less than 2y by the number tt of exceptional correlations in the sys¬ tem which possess singular points. 22. The number of exceptional correlations in a system satisfying seven given conditions being directly determined, the relations (1) of art. 20, written in the form (3), will enable us to deduce at once the characteristics of the system. I propose to apply this method to the determination of the charac¬ teristics of the fundamental systems, that is to say, of those systems which satisfy elementary conditions of the following types : 8 A given point shall have a given polar (2 conditions). A given line shall have a given pole (2 conditions). Two given points shall be conjugate (1 condition). Two given lines shall be conjugate (1 condition). The characteristics of the fundamental systems being thus deter¬ mined, it will be easy to deduce the number of correlations which satisfy any eight given elementary conditions. All that will be then requisite in order to enable us to ascend from elementary conditions to others of a more complex character, and thus to solve the problem of the correlation of two planes in all its generality, wdll be a more intimate knowledge of the properties of the curves of the order r, and of the curves of the class /j. which, as we have seen (arts. 18 and 19), are associated with each system of correlations. Enumeration and Glassification of the Fundamental Systems of Correlations. 23. a, /I, y, ^ being integers satisfying the condition 2a-h2/3 + y-(-^ = 7, Ave shall term (a/3yc) the signature of the system of correlations satisfying the following conditions : a points in one plane have given polars in the other. (d right lines in the first plane have given poles in the second. y points and ^ lines in each plane have given conjugates in the other plane. It is obvious that the systems of correlations whose signatures are (a /3 y ^) and (/3 a y ^) are identical. The first symbol, in fact, being regarded as descriptive of the data in the one plane, the second will indicate the corresponding data in the correlated plane. Now the total number of integral solutions of the equation 2«2/3-|-y-)-— 7 can be readily proved to be 52 ; hence there are twenty-six distinct fimdamental systems of correlations. 24. These systems may be arranged in six groups as follows : Group. Sig nature. I. . (3010) (0301) II. . (2110) (1201) Ill. C(2030) (0203) . 1 (2021) (0212) IV. i (1130) (1103) .1(1121) (1112) r (1050) (0105) V. . \ (1041) (0114) ((1032) (0123) 9 f(0070) (0007) VT J (0001) (0016) .(0052) (0025) ^(0043) (0034) 25. In each group, the systems are arranged in two columns, such tliat the data of each system in one column are correlative to the data of the system opposite thereto in the other column. Now to every correlation of the system (a [d y o) will obviously correspond, by the Principle of Duality, a correlation of the system (/> a ^ y) ; moreover, the class and order of the first system will be equal, respectively, to the order and class of the second system; and for every exceptional cor¬ relation with singular points in the former there will be an exceptional correlation with singular lines in the latter, and vice versa. It will suffice, therefore, to consider the systems contained in one only of the two preceding columns. Number and Nature o f Exceptional Correlations in the Fundamental Systems. 26. I proceed first to determine, directly, the number and nature of the exceptional correlations included in each of the thirteen funda- , mental systems whose signatures stand in the first column of the pre¬ ceding article. In doing so, two associated singular points of any exceptional correlation will always be indicated by the symbols and Sg, and two associated singular lines by and o-g. 27. Throughout this investigation it will be assumed that, between the positions of the given points and lines, no special relation whatever exists. 28. Let the symbol of the system (3010) be (SdO) / Pj Qj Rj Aj \ Vi's '•s Aj/' In this system there can be no exceptional correlation with singular lines. To prove this, I observe, first, that could not coincide with any one of the three lines p^ ^2? 5 ^*or if it did, then and Ri, as poles of and fg, would lie on (art. 17. d.) ; and Aj not being situated thereon (art. 27), would have (t^ for its polar, whereas by hypo¬ thesis this polar should pass through Ag; and secondly, that if there were a singular line o-g not coincident with any one of the three lines 9.2 '^”25 then the poles Pj^ of these lines would all lie on o-j (art. 17. d.), which is inconsistent with the assumed generality of the data by which the system is defined (art. 27). Again, if there be a singular point in the system, it must coincide 7 r =3 with one of the three points Pj Q| R^; for if it did not, p^ q<^ r^ would concur in iSo (art. 17. b.), which is not the case. Now if were coinci- 10 dent with P^, its polar, being perfectly arbitrary (art. 16.), might ob¬ viously be regarded as coincident with Moreover, and rg, as polars of and would then intersect in Sg 17- b.), and the polar of Aj would be manner all the'seven conditions would be fulfilled, and the exceptional correlation would be perfectly and uniquely determined (art. 16) ; the polar of any point Mj, for instance, would be theray ??i 2 , which passes through (^ 3 ^ 3 ) and satisfies the anharmonic relation (Qj R-^ Aj M^) = (^3 ^ 3 ) (^3 A 3 In like manner, 2^ might coincide with if Sg were coincident with (’’ 2 P 2 ) 5 ^^5 lastly, might coincide with R;,^, if 22 were coincident with Hence we conclude that m the system under consideration there are hut three exceptional correlations^ and that each of these has singular points. We have, consequently, by art. 20. (3), the values \ = 0, 9r = 3, = 1 , V = 2. 110 ) 29. The next system (2110) may be denoted by /Pi Qi ’1 ^i\ \P 2 ?2 ^2 ^ 2 / ;^=i Here cannot be coincident with o-^; for if it weve^p^ and ^ 3 , as polars of points not situated on cr^, would be coincident with cr^ (art. 17. c.) ; hence R2 must lie on (art. 17. d.) But if so, then, by art. 27, neither ^^2 nor can coincide with o-g, and as a consequence, Pj and Qj must lie on Again Aj^, not being on its polar, which by hypothesis passes through must be coincident with Hence the only possible correlation with singular lines is that in which these lines coincide, respectively, with Qi and ^ 2 ^ 3 ; that this correlation satisfies the seven given conditions, and is precisely determined by them, is readily seen. 77=1 Passing to possible exceptional correlations with singular points, it is obvious that R 2 cannot coincide with S 2 ; for if it did, P^ and Qp the poles of two lines p 2 and g '2 which do not pass through R 2 , would be coincident in 2 ^ (art. 17. a.), which is not the case. Hence we infer, by ai’t. 17. b., that must pass through 2 j, and as a consequence of this, —or rather of the fact that neither Pj nor can, under these circum¬ stances, coincide with 2 j,—that p^ and q^ must intersect in 23 . That there is one and only one exceptional correlation which has a singular point 23 coincident luith (Pgf/o), cind its associate 2 ^ situated in ?q, and that this correlation is precisely determined by the seven given conditions, is obvious on remarking that the anharmonic relation ^i(PiQHi^i) = (1^222)0^2^2^2^2) is the only remaining condition which 2 ^ has to fulfil (art. 16.). The point 2 j, in fact, is the intersection, with of the line which connects Aj with the only point A on P^Qi which satisfies the relation PlQl (PiQHi-^) = (p2Q2^2‘^2)‘ 11 For the system now under consideration, the equations (3) of art. 20 furnish the following values : — X = 1, TT = 1, /a = 1, V = 1. 30. The next system to be considered has the signature (2030), and (2030) / Pi Qi -^1 Pi Pi \P2 the symbol ^2 CJ* Neither y )2 nor y 2 can here coincide with 0-2 5 either did, the a=o poiars of A 2 B 2 C 2 would be coincident with o-^ (art. 17. c.), and therefore could not pass, respectively, through Hence we infer that if cr^ exist, and Qj must lie on it (art. 17. d.). But if were coinci¬ dent with Pj the poiars of Aj C;^ would be coincident with cr^ (art. 17- c.), which is inconsistent with the condition of their passing, respectively, through A 2 B 2 C 2 . Hence we conclude that there are no exceptional correlations ivith singular lines in the system. Again, if there be a singular point Sg, it must lie in one, at least, of 77=3 the two lines p<^ and ^ 2 » for otherwise Pj and would be coincident in (art. 17. a.). If ^2 were on p^, but not on q^, its associate Sj would coin¬ cide with Qj (art. 17.), and ^2 itself would, by art. 16., simply have to satisfy the condition Qi(PiAiBA) = SiiteAACa).(1). This is clearly possible, and in one way only (art. 29). In like manner, there is a second exceptional correlation for which 2 ^ is coincident with P;^, and ^2 lies on so as to satisfy the condition V, (QABA) = Sj (2ABA).(2). In every other exceptional correlation, ^2 must be coincident with (p 222)5 a,nd 2 ^ must satisfy the hornographic relation ^1 (PlQlAiB^C^) = (^ 2 ^ 2 ) (P2!l2^2^2^2)- Now it is well known that there is, in general, one and only one posi¬ tion for Sj consistent with this relation.* It is, in fact, the fourth intersection of two conics (S^') and (S/') circumscribed to the triangle AiBiCi, one of which (S/) passes through Pj, and is determined by the anharmonic relation (Sj) (P^A^^Bj^Cj) = (P2^2')(P2-^2^2^2)y and the other (S/") passes through Qj in such a manner that (S/') (Q.AiBiC,) = (^ 2 ^ 2 ) (( 72 A 2 B 2 C 2 ). The only exceptional case which can present itself arises when tlie above two conics (S/) (S/') happen to coincide. This case, however, is excluded by art. 27, since it implies the existence of a special relation between the positions of the given lines and points. * See Sturm’s excellent paper on the Problem der Projectivitdt und seine Anivendung auf die Fldchen ztveiten Grades., in the “ Mathematische Annalen,” vol. i. p. 5.33, wherein many of the auxiliary thGorem.s employed in tliis paper arc clahoratcly demonstrated. LIBRARY _ Uri'^RSITY OF IfPNO!?? 12 We conclude, then, that the system under consideration contains three exceptional correlations with sinrjular points^ and accordingly we have, as in art. 28, X = 0, tt = 3, ^ = 1, v = 2. ( 2021 ) A = 1 frr: l 31. We proceed to the system (2021) whose symbol is /Pi Qi Aj cA 9 .^ ^2 ^2 ^ 2 / If ^2 coincident with a singular line 0 - 2 , then would lie on (art. 17. d.), and at the same time the polars of Ag and B 2 would coincide with (Tj (art. 17. c.). But, since Q^, A^ and Bj are not in a line (art. 27), this is obviously inconsistent with the condition that these polars should pass, respectively, through Aj and Bj^. In like manner, ^3 cannot be coincident with (Xg ; hence, if this singular line exist at all, Pj and must lie on its associate (art. 17. d.), and the polars of Aj and Bj must coincide with o-g (art. 17. c.), that is to say, o-g must pass through Ag and Bg. This being clearly possible, and moreover the seven conditions being precisely sufficient to determine such an exceptional correlation,* we conclude that the system contains one^ and 07ily one, exceptional correlation ivith sinrjidar lines; the latter being coincident, respectively, with PjQi and AgBg. Again, it can be shown, as in art. 30, that if there be a singular point Sg, it must lie on one, at least, of the lines p^ $' 2 ' N^ow if Sg were situated on pg, but not on q^, then its associate would be coincident with Qj, and the pole of Cj would be coincident with Sg (art. 17. a.) ; but as this pole must be situated on Cg, (pgCo) is the only possible position for 2g. In like manner, if Sg were on q^, but not on ^g, it would neces¬ sarily coincide with ($' 2 ^ 2 ), and its associate with Pj. Moreover, it is evident that tbe seven given conditions suffice precisely to deter¬ mine one of each of the exceptional correlations here described. The only remaining position possible for (^ 272)5 which case the polar of Cg, necessarily a point on C|, would be coincident with 2^ (art. 17. a). The sole condition to be satisfied by this point 2^ on Cj is -1 (Pi Qi A-i Bi) (P272) (P2 72 ^2 P2) .; hence must be one of the intersections of by the conic (S^) which passes through Pj Aj B^ and satisfies the anharmonic relation (Si) (Pj Qj Aj Bj) = (poq^) (P2 72 -^2 ^ 2 )* Taking into account the two solutions of (I), we conclude that in the system under consideration there are four exceptional correlations ivith sin- (jular qjoinis. * The pole of any line for example, would be the point M 2 , on A 2 B 2 , deter¬ mined by the relation Ti Q) B’l (i m^) = ATIh M 2 ). 13 Tho equations (3) of art. 20 give us, in the present case, the values A = 1, TT = 4, fj, =. V = 3. 32. We arrive next at the system (1130), which is defined by the (iiso) / Pj iq A| V \ ^2 ^2 ^2 / Since in this case, cannot coincide with cr^^, (for if it did, the polars of PjAj^BjCj would be coincident with cr^, which is obviously impos¬ sible,) 2 17. b.). 17 Farther, if 2^ were not coincident with either of the points in which is intersected by and e^, then the poles of the latter lines would coin¬ cide, in with (art. 17. a.). This is clearly possible, and in one way only (art. 29) ; the necessary and sufficient relation to be satisfied by Sg being (d^ (Pj Cj^) = -^2 ^2 ^ 2 )- If Sg were coincident with then 2^, as pole of would be situated on e^, and satisfy the equation ^1 (^1 "^1 ^1 ^ 1 ) — (Pi ^h) (P2 ^2 ®2 ^ 2 ) 5 of this equation there are obviously two solutions (art. 31). In like manner, if were coincident with (then there would be two positions of 2^, on rZg, each of which would satisfy the necessary and sufficient condition — (P2 ^2)(P2 "^2 ®2 ^ 2 )* We conclude, therefore, that the system under consideration contains^ on the whole, six exceptional correlations u'itli singular gjoints. The following are the numerical values deducible from the above results: \ = 3, tt = 6, p = 4, v = 5. 37. The first of the four systems in Grroup VI. has tlie signature ( 0070 ) (0070), and is thus defined : /A, B, Cl Di El Fi GA \A 2 B, C2 D2 E 3 F2 Go/' There are here, as in art. 34, iw correlations with singular lines. Arro The necessary and sufficient conditions to be fulfilled by a pair of 7r=3 singular points is 2i(AiBiCir)iEiFiGi) = 22(A2B2C2D2B2F2G2) . (1), of which equation, as Sturm and others have shown, there are three solutions. This important theorem results, in fact, from the following considerations : In art. 34(1) it has been shown that, irrespective of Fg (which is an arbitrary point), there are on any line/g passing through Fg, two points Sg, to each of which corresponds a point Si, such that Si (Ai Bi Cl Di El Fi) = S 2 (A 2 B 2 C 2 D 2 E, FA . (2). It is obvious, too, that Fg itself counts as one such point; since one, and only one, point Si can be found to satisfy the condition Si (A, Bi C, D, El) = Fj (A5 B, C, D, E,). Hence, and from symmetry, we infer that the locus of the point Sg, to which another point S^ can correspond so as to satisfy the condition (2), is a cubic passing through the six points Ag Bg Cg Dg Eg Fg. The corresponding points Sj lie, of course, on another cubic, passing through Ai Bj Cj Dj Ej Fj, between whose points and those of the first cubic a (1, I) correspondence obviously exists. c 18 In like manner, the locus of each of the points S^, Sg, which satisfy the relation Sj (A;^ C;^ G^) = Sg (Ag Cg Dg Eg Gg) . (3) is a cubic passing, respectively, through the six points Aj E;^ Gj, and the six points Ag Bg Cg Dg Eg Gg. Exclusive of the five points Ag Bg Cg Dg Eg aud of anothcr point Sg, which may be termed the satellite of this group of five out of the seven points Ag Bg Cg Dg Eg Eg Gg, the above two cubic loci in the second plane intersect in three points 2g, which, with their associated points 2)]^? satisfy the relation (1). The excluded point Sg is that which corresponds to every point Sj on the conic (Ai^Bj^CjDj^Ej) in such a manner that Si (Ai Bi Cl Di El) = Sg (Ag Bg Cg Dg Eg).(4). It is, obviously, the fourth intersection of two conics (S 2 ) (S^) circum¬ scribed to the triangle AgBgCg, one of which passes through Dg, and makes (S^) (Ag Bg Cg Dg) = Ei (Ai Bi Ci Di), and the other passes through Eg, and makes (S 2 ) (Ag Bg Cg Eg) = Dl (A^ Bi Cl Ei). This point Sg is excluded for the following reason : Each of the cubic loci, in the first plane, determined by (2) and (3), cuts the conic (AiBiCiDiEi) once again; but under the assumption in art. 27 these two points of intersection will not be coincident, although by (2) and (3) each will have Sg for its correspondent. Hence, although Sg is common to the two cubics in the second plane, it cannot be regarded as one of the points Sg with which a point 2i is so associated as to satisfy the rela¬ tion (1). It is easy to see, however, that the satellite Sg, defined by (4), is the only one of the four intersections of the two cubics, in the second plane, which fails to lead to a solution of (1). We conclude, therefore, that the system under consideration contains three exceptional correlations ivith singular points ; accordingly, we have the values X = 0, tt = 3, /x = 1, v = 2. ( 0061 ) A=0 7r = 6 38. We proceed to the second system (0061) of Group VI., the symbol for which is /Ai B; Cl Dl El El gi\ \A2 B2 C2 D2 E2 F2 gj' Eor the reasons stated in art. 34, this system can contain no exceptional correlations with singular lines. If Si, S 2 be the associated singular points of any exceptional eorrela- tion, the condition Si (Ai Bi Cl Dl El El) — 22 (A2 B2 C2 D2 E2 E2) ( 1 ) must be satisfied. Hence, as we have seen in art. 37, Si and Sj must be corresponding points on two cubics ; one of which passes through Ai Bi Cl Dl El El, as w^ell as through the satellite of each of the six groups of five selected from these six points, and the other passes through 19 Ag B 2 C 2 D 2 E 2 F 2 , as well as through the six satellites similarly con¬ nected therewith. Besides satisfying the relation (1), however, it is to be remembered that if be not situated on ^ 1 , then S 2 must lie on ; and further, that if S 2 be not on g^^ then 2^ must be on ^ 1 , whence we conclude that in the system there are six, and only six, exceptioyial correlations with singular points, and that in three of these 2i is on gi, whilst in the other three •S 2 is on g^. We have thus the following system of values : X = 0, TT = 6, fx =2i, V = 4. 39. We pass now to the third system (0052) of the Group VI., the symbol for which is • VA2 B2 C2 D2 E2 /2 ^2/ Here again there are no exceptional correlations with singular lines. One condition to be satisfied by the singular points of every excep¬ tional correlation which the system may contain is 2 i (Aj Bi Cl Di El) =: ^2 (A2 B2 C2 D2 E2).( 1 ), in virtue of which to each point 2i will correspond, in general, one, and only one, point ^ 2 , and vice versa* Hence we infer that there can he hid one position for Si exterior to hoth the lines f and gi, for in such a case its associate would necessarily coincide with (/ 2 y 2 ) (art. 17. a.). That there will be one such position of S^ however, follows from the assumption in art. 27. In like manner, there ivill he one, and only one, exceptional correlation in ivhich Sj will he coincident with (/il7i), Cind its associate exterior hoth to /g and r/g. In all other cases, 2^ will lie on and therefore Sg on g^ ; or else Sj will be on g-^, and 2^ on f^. To decide how often the first of these cases will present itself, it will clearly suffice to determine the order of the locus of the points Sg which, by the relation (1), correspond to the several points 2^ of the right line/j. To do this we will first enquire into the number of points in which any line a^, passing through Ag, is intersected by the required locus. Now we have already proved that the locus of the point 2^ which satisfies the condition 2i (Aj Bj^ Cj E;^) = 22 (a^ ^2 ^2 ^ 2 ) is a cubic passing through B^^ Cj Dj E^ and having a double point at Aj (art. 34). Hence we may say that, exclusive of the point Ag, which has a perfectly arbitrary position on a^, there are three positions of on ^ 2 , to which correspond three positions of 2^ on /j, such that the relation (I) is satisfied. * The only exceptions to this arise when 2i (or 22 ) coincides with one of the five points in its plane, or with the satellite S) (or S 2 ) of these five (art. 37). C 2 (0052) A = 0 7r = 12 20 It is obvious, however, that this relation will also be satisfied when Hg coincides with A 2 , and with either of the intersections ofand the conic (Sj, through Ej, determined by the anharmonic relation (S.)(BiCiDiEi) = A2(BjC,D2E,). Hence, and from symmetry, we conclude that if 2 ^ describe any right line fi, the locus of the point ^ 3 , associated therewith by the relation ( 1 ), is a quintic which has a double point at each of the five points A 2 B 2 C 2 l) 2 E 2 . This quintic, ivhich has moreover a sixth doubte point (at the satellite S 2 of the above five points) corresponding to the intersections of f and the conic (AiBiCiDjEi), cuts in five points S 2 , to each of which cor¬ responds, by ( 1 ), a point 2 i situated on/i; whence we infer that there are five excepjtional correlations in the system under consideration, of which one singular point lies inf^, and its associate in g^. In like manner we conclude that there are five exceptional correlations of which one singular point lies in pi, and its associate in / 2 . Altogether, therefore, the system includes twelve exceptional correlations with singular points, but none with singular lines. Accordingly we have the following numerical values : X 0, IT ■= 12, ju = 4, V = 8. (0043) 40. AVe have now arrived at the last of the fundamental systems which need investigation. Its signature is (0043) and its symbol / Aj Bi Cl Dj 61 fi yA \A 2 B2 C2 B2 62 A ^ 2 / A =6 For reasons before stated (art. 36), every singular line cti must pass through two of the four points AiBjCiDi, and its associate o ’2 must pass through the conjugates of the remaining two; moreover, it is easily seen that the seven given conditions can be satisfied, and that in one way only, by an exceptional correlation whose singular lines have any one of the six pos.sible positions above indicated. Hence in the system there are six correlations ivith singular lines. „_i 2 The associated singular points 2 i, ^2 of every exceptional correlation in the system must, in the first place, satisfy the anharmonic relation 2 i(AiBiCiDi) = 22 (A2B2 021)2). The position of one of the two points being known, therefore, we have at once a conic on which the other must be situated. Now the point Sj must lie in one, ^t least, of the three lines eifgi, otherwise the poles of these lines would be coincident in 22 (art. 17. a.), which is obviously inconsistent with the condition of their being situ¬ ated, respectively, on e,, / 2 , and ^ 2 - Again, if 2 i lie on one only of the three lines e^ f g^, then its associate will necessarily coincide with the intei section of the conjugates of the other two, and 2 i itself will be 21 coincident witli one of the two points in which the first line is inter¬ sected by a conic determined by (1). Lastly, if Sj coincide with an intersection of any two of the lines ei fy gy, then will necessarily coincide with one of the two points in which the conjugate of the third line is intersected by another conic, likewise determined by (1). In every possible case, therefore, one of the associated points Si, must coincide with the intersection of two of three given lines, and the other must have one of two known positions on the conjugate of the third. Taking all possible combinations into consideration, and remember¬ ing that each one of them leads to two exceptional correlations satis¬ fying the seven conditions, we conclude that the system tinder consi¬ deration contains twelve exceptional correlations with singular points. We have consequently the following system of values : X = 6, TT = 12, jj, = S, V = 10. 41. The class, order, and singularities of each of the thirteen funda¬ mental systems, arranged in the left-hand column in art. 24, having now been determined, those of the remaining thirteen systems may, as stated in art. 25, be deduced by the Principle of Duality. The results may be thus tabulated: Group. Signature. (a/3y^) Characteristics. Singularities. T i (3010) . L (0301) TT i (2110) . I ( 1201 ) f (2030) ttt J . 1 ( 0212 ) ^ (0203) r(1130) TV d . I (1112) 1(1108) (1050) (1041) (1032) (0123) (0114) (0105) 1 2 1 1 1 2 3 2 1 2 2 2 1 2 4 5 4 2 2 1 1 1 2 3 2 1 2 2 2 1 2 4 5 4 o TT 3 0 1 1 3 4 1 0 3 2 2 0 3 6 6 3 0 0 0 3 1 1 0 1 4 3 0 2 2 3 0 0 3 6 6 3 22 Group. VI. Signature. Characteristics. Sing ularities (ciflyB) V TT A ^(0070) I 2 3 0 (0001) 2 4 0 0 (0052) 4 8 12 0 (0043) 8 10 12 6 (0034) 10 8 0 12 (0025) 8 4 0 12 (0010) 4 2 0 0 .(0007) 2 1 0 3 * Number of Correlations satisfying eight Elementary Conditions. 42. The number of correlations wliicb satisfy any eight elementary conditions may be readily determined from the preceding Table. In fact, if we indicate this number by the symbol where /3, 7, ^ have the same signification as before (art. 23), but now satisfy the equation 2a + 2/3-f = 8, we shall clearly have [a/lyr] egual to the class of the system of correlations (a/3y —1^), as ivell as to the order of the system of cor¬ relations (afoy d-l)- But the systems (o/Sy^) and (/3ay^) being identical (art. 23), and the class and order of the system (a/3y^) being the same, respectively, as the order and class of the system (/3a^y) (art. 25), it follows that [a/3y^] = [payB] = [a/3ay] = [/3acV] , in other words, that y and B as well as a and (3 are interchangeable in the symbol [apy^]. The following Table, therefore, gives the number * The meanings of the symbols being suitably modified, the results contained in this Table are at once applicable to the case of two homographic planes. For by the method employed in art. If), we have, corresponding to every system of corre¬ lations (a^yd) in two planes 111 , 112 , ^ system of homoyraphic relations' (a^yS) in the planes rij, n', which satisfy seven conditions of the following type : a points in the first plane correspond to given points in the second; 13 lines in the first plane correspond to given lines in the second ; y points in the first plane each correspond to a point on a given line in the second; and 8 lines in the first plane each correspond to a line through a given point in the second. Every such system of homographic relations coiitains exceptional ones. In tt of these there is a singular point in the first plane and a singular line in the second, whilst in A others there is a singular line in the first plane and a singular point in the second (art. 14). The points which correspond, in the several homographic relations of a system, to a given point in the first plane, lie on a curve of the order p in the second ; those which correspond to a point in the second plane, however, lie on a curve of the order v in the first. Simi¬ larly, the lines which correspond to a given line in the first })lane envelope a curve of the class v in the second, whilst those which correspond to a line in the second plane envelope a curve of the class y in the first. It is worth observing that if each point in the plane 111 he connected with every point in the plane n' which corresponds thereto, in the several homographic relations of a system whose characteristics are y and r, we shall have, in sjiace, a complex of the degree \x3-v — 'K\ir [art. 20 (2)J, of which n, and n' are singular planes. We have consequently twenty-six distinct comi)lexe8 associated with the several funda¬ mental systems of homographic relations satisfying seven conditions. 23 of solutions in all possible cases where the eight given conditions are of the elementary kind described in art. 22 : [4000] = 1 ; [3100] = 1 ; [ 2200 ] = 0 ; [3020] = 1, [3011] = 2; [ 2120 ] = 1 , [ 2111 ] = 1 ; [2040] = 1, [2031] = 2, [2022] = 3; [1140] = 1, [1131] =- 2, [ 1122 ] = 2 ; [1060] = 1, [1051] = 2, [1042] = 4, [1033] = 5; [0080] = 1, [0071] = 2, [0062] = 4, [0053] = 8, [0044] = 10.* 43. The only results in the above Table which cannot be obtained in the manner described in art. 42 are the first three. Of these, however the first is well known, and has already been alluded to in art. 3. The second differs in form only from the first; for whenever the poles of three lines are given, as in the first case, the polars of three points (the intersections of the given lines) are always known, and vice versa. With respect to the third result, it will be at once seen, on writing the eight conditions in full thus : Pi) Qi) '^1) ^1 1^2) 2'2) ^2) ^2 that they cannot possibly be satisfied by any correlation unless the an- harmonic ratios (ri5i)(PiQiri5j) and K 2^2 (P 2 5 ' 2 P 2 S 2 ) are equal to one another. The assumption of any such equality, however, would be in¬ consistent with that already made in art. 27, in virtue of which the positions of the given points and lines are perfectly arbitrary. * This Table obviously gives, also, the number of ways in which two planes may be rendered homographic (or put in perspective with each other), so as to satisfy eight conditions of the kind described in the note to art. 41. 24 Connexes determined hy Fundamental Systems of Correlations. 44. In art. 18 allusion was made to the series (doubly infinite) of curves of the class fi, and to the series of curves of the order v. which every system of correlations, satisfying seven conditions, determines in each of the two planes ; and in art. 22 it was observed that upon a more intimate knowledge of the properties of these curves must depend the solution of the general problem of correlation. Although unprepared, at present, to treat this wide subject exhaustively, I pro¬ pose, before terminating the present paper, to indicate briefly the more salient features of a few of the simplest series of curves of the above kind. 45. Before commencing to do so, I may observe that, according to the terminology employed by Clebsch in a very suggestive posthumous paper recently published in the “ Mathematische Annalen” (vol. 6, p. 203), each point of one of our two planes, and its polar in any correlation of a system, constitute an element o f a connex of the class y and order r;* whilst each point in the other plane, and any one of its polars constitute an ele¬ ment of the conjugate connex. As illustrations, therefore, of the Theory originated by the eminent geometer, whose loss to science is so uni¬ versally deplored, as well as on account of their application to the general problem of correlation, the well-defined conjugate connexes to which the several fundamental systems of correlations lead are well worthy of full investigation.f 46. Each curve of the class y may be termed the representative of the point by whose polars it is enveloped, and each curve of the order v the representative of the line of whose poles it is the locus. It is ob¬ vious that every representative of a point touches all the singular lines in its planeand every representative of a line passes through all the sin¬ gular points in its plane. C miex 47. Each of the systems (3010), (2030), (1130), (1050), and (0070) leads to a pair of conjugate connexes of the first class and second order. In each of these five systems, a point in either plane is represented by a point in the other plane, (in other words, every point of the plane has its conjugate, exactly in the same manner as the given points have,) whilst every line is represented by a conic passing through the three fixed singular points of the three exceptional cor¬ relations which each system includes. * A connex (/xi/), of the class ix and order v, is defined by a given relation (I. m O'* (o', y, zf — 0 between the coordinates x, y, z oi a. jDoint, and the coordi¬ nates 7], f of an associated right line. t I niay here mention that this investigation was in a tolerably advanced state eighteen months ago, when my studies were interrupted by other duties. Uncertainty as to when these studies may be resumed, has at length induced me, contrary to my original intention, to publish the present paper, before completing the whole enquiry. 25 We have here, in fact, the well-known case of the Quadric Cor¬ respondence of two planes, presented in its most general form; the three pairs of singular points being identical with the three pairs of Principal Points. The five systems above enumerated correspond, obviously, to the five different ways of determining a Quadric Cor¬ respondence by means of given Corresponding Points, Principal Points, and Principal Lines.* 48. The systems (0301), (0203), (1103), (0105), and (0007) all lead to connexes of the second class and first order. Every right line, in each of these systems, is represented by a right line ; in other words, every line has its conjugate, in exactly the same sense as the given lines have. Every point, however, is represented by a conic touching the singular lines of the three exceptional correlations which each system includes. In short, we have here a correspondence estab¬ lished between two planes which is, simply, the correlative of the ordi¬ nary Quadric Correspondence.f 49. Of the remaining sixteen fundamental systems of correlations, by far the simplest are the two whose signatures are (2110) and (1201), both of which lead to one and the same kind of connex of the first class and first order. In illustration of this I will consider briefly the first of the two sys¬ tems. Its complete definition is given by the symbol /Pi Qi ’"i ^i\ % P'2 ^2/ and, as shown in art. 29, it contains two exceptional correlations; of which one has the singular lines = P^ Qj and = Rg Ag, whilst the other has the singular points ^ 2 = (^ 2 ^ 2 )’ 5 latter point being so situated, on r-^, that ^1 (Pi Qi -^i) — ^2 (P 2 % P2 '^2)- Since the characteristics are /^ = I and r = l, it is obvious, first, that the polars of any point (say M^) in any two correlations of the system {e.g., in the two exceptional ones) determine, by their intersection, the point M 2 , through which all the other polars of pass ; and, secondly, that the poles of any right line in the two exceptional correlations determine the line upon which all the other poles of are situated. Now if be neither coincident with 2^ nor situated on its polar in the exceptional correlation with singular lines will coincide with the * See Rpye on Gcometrische Verwandschaften zweiten Grades, in the Zeitschrift fur Mathematik und Fhysik, vol. xi., 1866. t The fact that we have no distinctive name for this correspondence is doubtless due to the circumstance that no terms correlative to the very convenient ones quadric, cubic, &c. are yet in general use. A special case of the correspondence in question was briefly described in my paper “ On the Quadric Inversion of Plane Curves,” published in the Proceedings of the Royal Society, vol. xiv., 1865, and in the “Annali di Matematica pura ed ap])licata,” tom. vii., Roma, 1865. It was also distinctly referred to in the paper by Rcye above cited. Connex ( 21 ) Connex ( 11 ) 26 singular line o-g (17. a.), and its polar Wg, in the exceptional correlation with singular points, will pass through S 2 (17. b.), and be determined by the relation Sj (P^ Qj = Sg (jpg Rg ??^g). Consequently, every such point is rejpreserited, in the connex, by a point Mg = on the singular line Cg. If were coincident with 2j, however, mg would be perfectly indeterminate (art. 16.), and its repre¬ sentative Mg would, as a consequence, be an indeterminate point on erg. Lastly, if M;^ were situated on then, although mg would be deter¬ mined by (1), as before, the polar of M^ in the exceptional correlation with singular lines would be an indeterminate line through (art. 16.) ; and since the latter in one of its possible positions would be coincident with mg, the representative of M;^ must be regarded as an indeierminate point in m^* 50. The peculiar point hy q)oint representation to which we are led by the system of correlations under consideration is such, therefore, that 1) the singular point of either plane is represented by an indetermi¬ nate point in the singular line of the other plane ; 2) a point in the singular line of either plane is represented by an indeterminate point in a determinate line through the singular point of the other plane; 3) every other point in the first plane is represented by a per¬ fectly determinate point in the singular line of the second plane. From similar considerations we infer that the line hy line representation, to w'hich the present system of correlations leads, may be briefly described thus : 1) a singular line is represented by an indeterminate line through a singular point ; 2) a line through a singular point, by an indetermi¬ nate line through a determinate point on a singular line ; 3) every other line, in either plane, is represented by a perfectly determinate line through a singular point. 51. It will be observed that the representation just described is not, even in an exceptional sense, homographic ; since one of the character¬ istic properties of homographic correspondence is not fulfilled by it,—I mean that property, in virtue of which to a point in a line always cor¬ responds a point in the corresponding line. The representation above indicated, however, as well as that termed homographic, is to be reo-arded as one of the forms to which a connex of the first class o and first order may lead. 52. It is also worth observing that in the connex now under con¬ sideration the locus of the points which represent those situated in a given right line of either plane, breaks up into the representative of that line in the other plane, and the singular line of the latter plane ; * It is obvious that, in all the correlations of the system, except that which has sinpfular lines, the polars of a point on a singular line in either plane are coincident with a line which passes through the singular point of the other plane. 27 and further, that the envelope of the lines which represent those passing through a given point of either plane, breaks up into the representative of that point in the other plane, and the singular point of the latter plane. We have here, in fact, the simplest case of a general theorem which holds for all the connexes determined by the fundamental systems of art. 41, and which may be thus enumerated: The curve, of the order v, which represents any straight line, is always a constituent part of the envelope of the curves, of the class /u, ivhich repre¬ sent the several points oj that line; and, vice versa, the curve, of the class fx w^iich represents any point, is always a constituent part of the enve¬ lope of the curves, of the order y, ivhich represent the several right lines passing through that point.^ In proof of this, it will be sufficient to show that if and (Xg be pole and polar in any correlation of a system, then, in that same col relation, the tangent at to the representative of and the point of contact of with the representative of A^^, will also be pole and polar.f Now in the correlation immediately fol¬ lowing the frst, let a'g be the polar of A;,^, and A'^^ the pole of a^, in which case, of course, (ag af) will be the pole of Then, ulti¬ mately, as the latter correlation approaches to identity with the first, ((Xg df) becomes the point of contact of ^2 with the representative of A;^, and AjA'^^ becomes the tangent at A^ to the representative of a^. 53. The connexes, of the second class and second order, to which the Conne . . . . (22J systems (1121) and (1112) give rise, are essentially the same in cha¬ racter. We may confine our attention, therefore, to the first system, which has been denoted in art. 33 by the symbol /Pi Py Ai Bi CjX \P3 ^2 ^2 ^2 ^2/ and found to include four exceptional correlations. One of these has the singular lines a\ = Pj A^, ^ P 2 Ba; another the singular lines ^ 2 ^^ 2 ) of this line-pair lies on a fixed line s'^ passing through P 2 . ' Every four such lines m-i-, W 2 , n-i-> nf together with the singular lines 0 - 2 , 0-2 constitute six tangents of the conic which represents the arbitrary point (^ 1 ^ 1 ). 56. The degenerate conics of the first plane which represent points and lines in the second, have precisely similar properties, that is to say,— The points of the point-pairs lie on the singular lines 2 ^ 2 )) where 2'i, 2'/ are the intersec¬ tions, with Cl, of the conic (Si) which passes through PiQiAiBi so as to make the anharmonic ratio of these four points on it equal to {PtT) (P 2 qi A2B2). 59. The mode in which the conics and cubics representing special Connex ( 23 ; 30 points and lines degenerate, is very instructive. Instead of attempting an exhaustive discussion of the question, however, I must limit myself here to those cases of degeneration which most elucidate the general character of the representative curves. Every point C 2 on the singular line A 2 B 2 is represented, as we have already stated in art. 57, by a point-pair Ci, Ci. The point C'l is so situated on the associated singular line PiQi as to satisfy the condition PibJi (Pi Qi Cl Cl) = A 2 B 2 (^ 222 ^ 202 ) .(1) ; the polar of C 2 , in the exceptional correlation whose singular lines are PiQi and A 2 B 2 being an indeterminate line through Cl. The point Ci, through which the polar of C 2 in every other correlation of the system passes, is situated on the conic (Si) and fulfils the condition (Si)(Pi Qi A;^ B^ Cj) = (p2 g2)(P2 9^2 ^2 ^2 ^ 2 ) . (2)> as is at once obvious on considering the polars of C 2 in the two corre¬ lations whose singular points are 2l and and 21' and (i ?2 9 ' 2 )- 60. Between the points of the conic (S,), and those of the line A 2 B 2 , therefore, a (1, 1) correspondence is established, such that any two corresponding points C]^ C 2 thereof are conjugate to each other in the same sense as A;|^A 2 and Bj^Bg are by hypothesis. The system of cor¬ relations, in fact, would suffer no change if C^Cg were substituted in place of either of the latter pairs of conjugate points. 61. The conic, in the second plane, which represents any arbitrary point, sayAj, of the conic (Si), is a line-pair tohose double point coincides tuith the corresponding point Ag on AgBg. This follows from, the cir¬ cumstance that, in the system now under consideration, there are two correlations for which the polar of A^ coincides with an arbitrary line Og passing through Ag; in other words, there are two correlations which satisfy the eight conditions Qi5 h Pit (lit (^1t Bg, Cg In proof of this we need merely refer to the Table in art. 42, where it will be seen that [3011] = 2. 62. From the above we infer that every line in the second plane is represented by a cubic x\ which has a double point Cj on the conic (Si). For if C| be the point on (Si) which corresponds, by art. 60, to the point Cg in which AgBg is intersected by ^g, then, as we have just seen, there are two correlations of the system for which the pole of is coincident with Cj. The cubic Xi obviously passes also through the singular points Pj, Qj, 21, 21' as well as through the point Cl on PjQi, which, by (1) of art. 59, corresponds to Cg on AgBg. 63. Lastly, since two of the singular points of the second plane coin- 31 cide in obvious that every line a;^ in the first plane is repre¬ sented hy a cubicwhich has a double point at (^ 2 ^ 2)5 and passes, likewise, through the singular points (^ 2 ^ 2 )’ ( 2 ' 2 ^ 2 )’ through the point Q^,on A 2 B 2 , which corresponds, by (1) of art. 59, to the point Cl in which oc^ intersects P^Q^, and through the points Ti^and^^i 011 A 2 B 2 , which correspond, by (2) of art. 59, to the points and in which intersects the conic (Sj). 64. The remark in art. 61 is susceptible of generalisation in the form of the following useful Theorems, with the enunciation of which I will conclude the present paper : Theorem (i.) :— In a system of correlations (a/3y3), the curve, of the class [a/3(y+ 1)^] (art. 42), which represents either of two conjugate points A 2 , breaks up into the other, together with a point on each of the singular lines associated with those which pass through the former. The multi¬ plicity of Ag on the representative of A^ is [(a + l)/3 (y —1) ^], and that of on the representative of is [a (/3 + 1) (y —1) ^]. The num¬ ber of singular lines which pass through Aj is [a/3(y + l)^]-[(a + l)/3 (y-l) and the number of those which pass through A 2 is [a/3 (y+1) ^] —[a (/3 + 1) (y-1) ^]. Theorem (ii.) :— In a system of correlations whose signature is (a/3y^) the curve, of the order [a/3y(^ + 1)] (art. 42), which represents either of two conjugate lines a^, a^, breaks up into the other, together with a line through each of the singular points associated with those situated on the former. The multiplicity of a,^ on the representative of is [a (/3 + 1) y (^ —1)], and that of a^ on the representative of a^ is [(a-j-l)/3y — 1)]. The number of singular points situated on a^ is [«/3y(^ + l)]-[a (/3 + 1) y(^!-l)], and the number of those situated on a^ is [a/3y(J + l)]-[(a + l)/3y(a-l)]. May 2nd, 1874. •, ' A s