MATHEMATICAL PAPEBS
ffiotrtxm: 0. J. CLAY AND SONS,
CAMI3HIBGE UNIVEBSITY PBESS WABEHOUSE,
AVE MAMA LANE.
! DEIOHTON, BELL AND co.
j p. A. BBOOKHAUS.
Jforfti MAOMILLAK AND 00.
THE COLLECTED
MATHEMATICAL PAPERS
ARTHUR CAYLEY, So.D., F.E.S.,
BADLKUIAN PHOPHSBOK OP PURE MATHEMATICS IN THE UNIVERSITY OP CAMBRIDGE.
VOL. VI.
CAMBRIDGE :
AT THE UNIVERSITY PRESS.
1893
[All Mights reserved.]
ADVEETISEMENT.
THE present volume contains 33 papers numbered 384 to 41G published
for the most part in the years "1865 to 1872; the last paper 416,
of the year 1882, is inserted in the present volume on account of its
immediate connexion with the papers 411 and 415 on Reciprocal Sur
faces.
The Table for the six volumes is
Vol. I. Numbers 1 to 100.
II. 101 158.
III. 159 222.
IV. 223 299.
V. 300 383.
VI 384 416.
CONTENTS.
384. On the Transformation of Plane Curves ..... 1
Proo. London Math. Society, t, i, (186566), No. in. pp. 1 11
385. On the Correspondence of Two Points on a Curve ... 9
Proc. London Math. Society, t. i. (1865 66), No, vn, pp. 1 7
386. On the Logarithms of Imaginary Quantities .... 14
Proc. London Math. Society, fc. n, (186669), pp. 6054
387. Notices of Communications to the London Mathematical Society . 19
Proo. London Math. Society, t. u. (186669), pp. 67, 2020,
29, 6163, 10310']., 123125
388. Note on the Composition of Infinitesimal Rotations . . 24
Quart. Math. Jour. t. vm. (1867), pp. 710
389. On a Locus derived from T^VQ Conies ...... 27
Quart Math. Jour. t. vm. (1867), pp, 7784
390. Theorem relating to the four Conies tuhich touch the, same
two lines and pass through the same four points
Quart. Matli. Jour. t. vm. (1867), pp. 162167
391. Solution of a Problem of Elimination ....
Quart. Math. Jour. t. vm. (1867), pp. 183185
392. On the Conies which pass through two given Points an
tivo given Lines ......
Quart. Math. Jour, t. vm. (1867), pp. 211 219
v l CONTENTS.
393. 0/i the Conies which touch three given Lines and pass through
a given Point ....... , .
Quart, Math. Jour. t. vm. (1867), pp. 220222
304. On a Locus in relation to the Triangle . . .
L/ * * "
Quart Math. Jour. t. vm. (1867), pp. 264277
305. Investigations in connexion with Casey's Equation . ,
Quart. Math. Jour, t, vui. (1867), pp. 33d 841
3DC. On a certain Envelope depending on a Triamjk imcribcd in
a Circle
........ .
Quart. Math. Join. t, ix. (1868), pp. 3141 imd 17fi 170
397. Specimen Table M , <M (Mod. N) for any prime or composite.
Modulus .
.........
Quart, Math, Jour. t. ix. (1868), pp. 9690 ami pl,vte
398. On a Certain Sextic Developable and Sextic Surface conned
t/ieremth .
' ' ' *
Quail Math. Jour. t. ix. (1868), pp. I20.ua and 373370
890. On the Cubical Divergent Parabolas .....
Q^rt. Math. Jour. t. ix. (1868), pp. 18G189
W. On tlu, Cubic Ct. m inscribed in a given Pencil of flfe .^,,
Quart. Math. Jour. t. ix. (1868), pp. 210221
401. A Notation tf tjle foints md ^ ^ pasmts
Qa.t Math. J 011 , t Ix . (1868) , )M68 __ 274
2. On a Singularity of Surfaces
Quart. Mail, Jom . t . , x . (1868) pp jM!uas8 ' ' ' ' ia.1
3. On Pascal's Tlieorem
Quart. Maft . Jou , , I18 . . . , lgl
404. Reproduction of Euhr's Memfr of ,,,. rt
Solid Sody. . J 758 on the Rotation of a
th. Jou , t . IX .  ias
otvi , fe 1 p> HJ ^ . 147
CONTENTS. IX
FAQE
406. On the Curves which satisfy given Conditions .... 191
Phil. Trans, t, OLVIII. (for 1863), pp. 75143
407. Second Memoir on the, Curves which satisfy given Conditions ;
the Principle of Correspondence . . . . . 263
Phil. Trans, t. CLVIII. (for 1868), pp. 145172
408. Addition to Memoir on the Resultant of a System of two
Equations .......... 292
Phil, Trend, t. OLVIII. (for 1868), pp. 173180
409. On the Conditions for the existence of three equal Moots or
of two pairs of equal Roots of a Binary Quartic or
Quintic . . . . . . . . . . 300
Phil, Trans, t. OLVIII. (for 1868), pp. 577588
410. A Third Memoir on Skew Surfaces, otherwise Scrolls . . 312
Phil. Trans, t. CLIX. (for I860), pp. 111126
411. A Memoir on the Theory of Reciprocal Surfaces . . , 329
Phil. Trans, t. CLIX. (for 1869), pp. 201229
412. A Memoir on Cubic Surfaces ....... 359
Phil. Trans, t. OLIX. (for 1869), pp. 231326'
413. A Memoir on Abstract Gfeometry ...... 456
Phil, Trails, t. CLX. (for 1870), pp. 5163
414. On Polyzomal Curves, otherwise the Curves 77+Vl 7 t&c. = . 470
Trans, R. Soo, Edinburgh, t. xxv, (for 1868), pp. 1110
415. Corrections and Additions to the Memoir on the Theory of Reci
procal Surfaces 577
Phil. Trans, t. OLXII. (for 1872), pp. 8387
416. On the Theory of Reciprocal Surfaces ..... 582
Addition to Salmon's Analytic Geometry of Throe Dimensions, 4th ed.
(1882), pp. 593604
Notes and References ......... 593
Plates to face pp. 52, 122, 190
Portrait ......... to face Title.
CLASSIFICATION.
GEOMETRY :
Abstract Geometry, 413
Curves, 114
BeciprocHl Smfaces, 411, 416, 416
Cubic Surfaces, 412
Skew Surfaces, 410; Developable, 398
Singularity of Surfaces, 402
= Curve., ! pencil of aix K MS ,
ANALYSIS :
Hotatio, ls , 388
M cmoh . f
of
eight,, M enioi ' ' > 8, 40D
384.
ON THE TRANSFORMATION OF PLANE CURVES.
[JTi'om the Proceedwcjn of the London Mathematical Society, vol. I. (1865 1866), No. in..
pp. 111. Eead Oct. 16, 1865.]
1. THE expression a "double point," or, as I shall for shortness call It, a "dp,"
is to be throughout understood to include a cusp : thus, if a curve has S nodes (or
cloTuble points in the restricted sense of the expression) and K cusps, it is here regarded
as having 8 4 K dps.
2, It was remarked by Cramer, hi his "Theorie des Lignes Courbes" (1750),
that; ft curve of the order n has at moat (?&l)(ji 2), =s J (n 9 3n) + 1, dps.
3, For several years past it has further been known that a curve such that the
coordinates (fa : y \ z) of any point thereof are as rational and integral functions of
tlie order n of a variable parameter 6, is a curve of the order n having this maximum
mi rnboi 1 (?i !)( 2) of dps.
4 The converse theorem is also true, viz. : in a curve of the order n, with
(ra l)(?i 2) dps, the coordinates (so : y : z} of any point are as rational and integral
functions of the order n of a variable parameter Q or, somewhat lesa preciselv. him
coordinates are expressible rationally in terms of a parameter Q.
OS THE TJUHSFOKlttTION OF PLANE CUBVBS. fgg.
series of curves of the order n1, given by an equation V+9V = (\ eontaininrr ,,,
rbitrary parameter 6; any such curve intersects the given curvo in tl.o ,1,,, ,,
>.mt,, lg as two points in the S.S points, and in o e other point; hone,, w ' tl,,,,
' only one vamble pomt of intersection, the coordinates of HUH point, vii! ., ,,,', ,,,
:.tes of an arbitrary pomt on the given curve, are expressible mtio mllv in t, r
"*"
* = the foregoing theorem is tha
o expressible rationally in terms of a parameter 9 S
fet po ints ^succeed each other in ^^ AR ^" ^ T" V " ',""
tamed by g lv i llg to the parameter it s different real Un fron , " ,'"'
rve may be termed a , curve. "^ '"' ' lllui
inciation it , s necessary to refer to naee l<f! , Tr PP ' , " <mi ' lll!to th "
; Abelschen Functioned Or*, t J "( 8 7, 1 'TT '""' ^P' " Thll '' rio
I lei L1V  ^ a7 >: PP' Ho],,,,, Vlai> th(J ( ,, mil(!i(l . ilm
of any order
form (1, f)(i, , ) = . P aiamete " (6 ';), connected by nu (!(I , llUi , m
>0  rationally in terms of the p a mn rn IK \
a certain form, viz.: ' te ' S (fi ') O ' mn *<l by an o,,,,atioii of
fl = l, the equation is (1, t\t n ~fn , ,
j^'^erootof;^;,^: oT "' tho """"
"4 the equation is (1, py> a x,_ n v
i. the square root of i 6 'exU 'fltbn O f P" to " 10 * mo
" > 2, viz. ; '
5 odd, = 2/1 3, the equation is a tw, ,
that treating (f ^ , r,^' f r(1> '" =. and is besides such
represented Lyi 2 )f d J rteSmn CM '*^s, the curvo thoT'
that treating (f^' 1?' P" (1 ' '' ) " = . a "d is besides sue
" ^ &
of this, f , . , f
coordmates (. s , ; , ^ , f. place of (f, ,),
Possible rationaUy and hooge neously i
n teni
384] ON THE TBANSFOKSfATION OF PLANE CURVES. 3
of (f. % D connected by an equation of the form ( )* ( ^ = 0. Such an aquation,
treating therein (, 1;, f) as coordinates, belongs to a curve of the order 2/t, with a
/ituplc point at (f = 0, =0), a /*tupIe point at (77 = 0, = 0), and which has besides
(M2) 3 or (/* !)(/* 8) dps, according as D = 2/i3, or 2/*2. The coordinates (x.y.s)
of a point of the given curve are expressible rationally in terms of the coordinates
($''>}' ) of a point on the now curve ; and we may say that the original curve is
by means of the equations which give (to : y \ z} in terms of ( ; 7; : ) transformed
into the new curve.
11. A curve of the order 2/a may have ^(2/41) (2/t 2), 2^ s 3/t+"l dps;
hence in tlie new curve, observing that tho /t tuple points each count for Kffft
we have
In the caso jD = 2/i ~ 8,
Deficiency = 2/t fl
In the case D = 2/j, ~ 2,
Deficiency = 2/i 3 8/t + 1
/*"+
9,i 9 n
*""* fffM ' ' >Wj " J^/
Moreover for 7J = 0, tho transformed curve is a conic, with dps, and therefore wibh
deficiency =0; in' the case J)~l, it is a quartic with 2 dps, and therefore deficiency
= 2; in the caso D = 2 it is a quintic with a triple point =3, and a double point
= 1, together 4 dps, and therefore deficiency = 2. Hence in every case the new curve
has the wamo deficiency as the original curve.
12. The theorem thus is that tho given curvo of the order n, with deficiency D,
may bo rationally transformed into a curvo of an order depending only on the
deficiency, and having the same deficiency witli the given curve, viz.: D = 0, the new
curve is of tho order 2(=D + 2); = 1, it is of the order 4 (=. + 8); JD2, ifc is
of the order 5 (= + 3); and D>2, it is for D odd, of the order D + 3;
even, of the order .0 + 2. It will presently appear that these are not the lo
which it is possible to give to tho order of the now curve, Riemanu' r
ON THE TRANSFORMATION OJT PLANE CUIWES.
 of ( " i * ): H is con to
iew  and
[384
11
the s
toV* with deflcie 7 A which is by the
'^z^^^Htf eacl i
i / r, 1 w> 7 ?> 6; ~u. ine transformed
<leficiency D ^ ^ ^^ ^^
equations f fP  n . /? f i ^ ' J * , t0 each of fch
i ^ / ' " 5 ' * : ^ theie C01 'responds a ainde set of
of tlie
nf /fc . M
'
IT. If, however, the curves Po = ffn

curve is equal to this number.
I assume that we have k < n :
P though t , le same ii(*8)l pia i"t," ' ' f "* " ler
sections common, the equations bein/f fee onms have all their inter
the tan P + ^ + Jlo To mX t ***, "T" ' ^ " Manti<lal relati
pebble, TO ^ tike the c , v f P * (P^'o' f T*"? m ' Ve " ' W a
as many poiuta ns possiblej and it ""'3^ from T]^ m89 ' .r "" on > ff =
groateat possible number i s =*/,% + 31^2 i,f? , , ""J"* , J " St " lade ' that
the number of points m the eSvt ^0 rflTff'' T ^T'' ""^
8  3, K'^ 3^) 2, respectively. " """" ^ to
7 e ; 00 f de S ^e curve ff = with deficiency D, or with i(tf lh,)
i Ai = ?t 1, we mav asmmn flinf IK i. r ^^ '
of the order !, ^ pas f *** ^ '"fa* Curves P0 ( 0, J20
through the * (m  3n)  .D + 1 dps,
and through 2 + JD  4 other poiiitB,
together j^j^ T^ points of thfl
384] ON THE TRANSFORMATION OF PLANE CUEVES.
This being so, each of the three curves will meet the curve U=0
in the dps, counting as ?i a 3?i 2.D + 2 points,
in the 2)i + D~4> points,
and in D + 2 other points,
together ?t a )i points ;
whence the order of the transformed curve is D I 2,
20. In precisely the same manner, secondly, if k = n~ 2, then we may assi
that the transforming curves P = 0, Q = 0, R  0, of the order n  2, each pass
through the 4( a 3w)D + l dps,
and through n + D4 other points,
together ^(?i a ~?i) 3 points of the curve 7=0;
and this being so, each of the three curves will meet the curve U =
in the dps counting as n*  3n W + 2 points,
in the n+ D 4 points,
and in 7) + 2 other points,
together a  2??, points ;
whence the order of the transformed curve is also in this case = D + 2.
21. I was under the impression that the order of the transformed curve a
not bo reduced below D + 2, but it was remarked to me by Dr Clebsch, that in
case JD > 2, the order might bo reduced to D + 1. In fact, considering, thirdly,
ease & = w3, we see that the transforming curves P~0, (2=0, .# = of the o
, 8 may bo made to pass
through the (w 9  3)  + 1 dps,
and through JD  8 other points,
together (n a  8w)  2 points of the curve U = ;
and this being so, each of the three curves meets the curve U= 0,
in the dps counting as (?t a  3n) 2.0 + 2 points,
in the 8 points,
and in > + ! other points,
together n*  3 points ;
whence the order of the transformed curve is in this case =/)+!,
22. The general theorem thus is that a curve of the order n with deficiency
can be, by a transformation of the order nI or n2, transformed into a curv
6 ON THE TRANSFORMATION OF PLANE CUEVttS. [8g
the order + 2; and if D>2, then the given curve can bo by a transformation of
the order 3 transformed to a curve of the order .Z) + l: tho trans formwl curvo
having iu each case the same deficiency D as the original curve.
23. In particular, if jD1, a curve of the order n with deficiency 1, or with
^ (n s  3) dps, can be transformed into a cubic curve with tho aamo doluuoiioy, thud
is with Odps; or the given curve can be transformed into a cubic, Thin WIBO is
discussed by Olebsch in the Memoir "Ueber diqjenigou Curvcn doron G'ooi'diiuiton
elliptisehe Functioiien eines Parameters sind," Grelle, t. r.xiv., pp. 210271. And lu<
has there given in relation to it a theorem which I establish as follows:
21). Using the transformation of the order n 1, if besides tho 2H + .H 4(=: g w ;j)
points on the given curve U~0, we consider another point on tins curve, thon wn
may, through the (w 3 3) dps, the 2w3 points and tho point 0, draw a wrios of
curves of the order nl, viz., if P 0t Q 0t R , are what tho function* P, Q, M, become
on substituting therein for (ts, y, z\ the coordinates fa, y 0i *) of tin; givuu point
then the equation of any such curve will be aP + bQ + olt^Q, with tho relation
aP + bQo + cll between the parameters a, b, c; or (what ia tho same thing) olimimitiii"'
o, the equation will be a^R* 2\R) + b(QR, Q ^) = 0, which contain* tho Hind?,
arbitrary parameter a : b. In the cubic which is the transformation of tho givuu ourvi
we have a point O' corresponding to and if (f a , %] ,) bo the Gi.ordinatoH of this
point, then coire spoil cling to the series of curves of the order nl, wu havu a HOHCB
of lines through the point 0' of the cubic, viz,, the linos f + * + < with thu
relation a D + 6^ + cft = between the parameters; or, wluit is tho samo tliinif, w,
have the series o lines <& ffi) + 6 ^o) = 0, containing tho Hmnn inglo
parameter  I By determining this parameter, tho CUVVOB of tho ordor n~l, will
vlicMoiich .h * rder ^7 h t dp3 ' th 2?1  8 P 01 ^ ^ * 1*** ^
hich touch the pven curve ^=0; and the lines will bo tho tangent* to tho onbu
nm 1 ' tane " tS *"
'
theorem, viz.: p0m; mt 18 ' ATO have the following
* cency V 1 wo
curve, four curves of the order Ul  vifTl ^ draw S M to touch tho
where the ratio ft : 6 is determined bv .' T^ 9 ^ by ftn equation
~, ttmi ULUVBS oi uiue order n i' viV tl vuu\,n unu
the ratio a : 6 is determined by a certai^ 9 ^7" 7 ^ eqUati n ^ + ^'"0,
( wn the absolute invariant J*.J O f tho '1,^ f 8 4 ( ! ua1a ? 1 (*!* & ) 4l=0 i then
01 the quartic funchon, is independent of the
384] ON THE TRANSFORMATION OF PLANE CURVES. 7
positions of the "2 2 points on the curve U 0, and it is consequently a function
of only the coefficients of the curve U = 0, being, as is obvious, an absolute invariant
of the curve U= 0.
26. And, moreover, if the curve U= is by a transformation of the order n l,
by means of 2 3 . points on the curve as above, transformed into a cubic, then the
absolute invariant I s s J of the quartic equation which determines the tangents to
the cubic from any point 0' on the cubic (or, what is the same thing, the absolute
invariant S 3 = 2 lJ of the cubic, taken with a proper numerical multiplier) is independent
of the positions of the 2 3 points on the curve U=(), being in fact equal to the
abovementioned absolute invariant of the curve U=Q. The like results apply to the
transformation of the order 11 ~ 2.
27. Suppose now that we have D>2, and consider a curve of the order n with
the deficiency D, that is with (n a 3n) D + 1 dps, transformed by a transformation
of the order n 3 into a curve of the order Dl with deficiency D', then, assuming
the truth of the subsidiary theorem to be presently mentioned, it may be shown by
very similar reasoning to that above employed, that the absolute invariants of the
transformed curvo of the order JD + 1 (the number of which is =47) G), will be
independent of the positions of the D 3 points used in the transformation, and will
be equal to absolute invariants^) of the given curve 7 = 0,
28. The .subsidiary theorem is as follows: consider a curve of the order D + l,
with deficiency J), that is, with ^J) (D  1) D = $(& 3D) dps; the number of
tangents to the curve from any point ()' on the curve is (JD + l)D(7> 3D)  2,
=3 4kb 2, (this assumes however, that the dps are proper dps, not cusps,) the pencil of
tangents has 4>D  5 absolute invariants, and of these all but one, that is, 4>D 6,
absolute invariants of the pencil are independent of the position of the point 0' on
the curve, and are respectively equal to absolute invariants of the curve.
29. To establish it, I observe that a curve of the order D + 1 with deficiency D,
or with (J9 3 3Z)) dps, contains ^(D + l)(D + ^)^^(D^W) 1 ~4D+2 arbitrary
constants, and it may therefore bo made to satisfy 4D+ 2 conditions. Now imagine a
given pencil of 4Z*  2 lines, and let a curve of the form in question be determined
so as to pass through the centre of the pencil, and touch each of the 4.D2 linos;
the curve thus satisfies ID  1 conditions, and its equation will contain 4D + 2  (4D  1), 
arbitrary constants. But if we have any particular curve satisfying the 4.D  1 oonditu
then by transforming the whole figure homologously, taking the centre of the pencil
as pole and any arbitrary line as axis of homology, ao as to leave the pencil of linos
unaltered (analytically if at the centre of the pencil 0, y = Q, then by writing
<uu + (3y \ yz in place of z) the transformed curve still satisfies the 4>Dl conditions,
and wo have by the homologous transformation introduced into its equation 3 arbitrary
constants, that is, we have obtained the most general curve which satisfies the conditions
in question. The absolute invariants of the general curve are independent of the
1 It is right to notice that the absolute invariants spoken of hew, and in what follows, are not in gaucral
rational ones.
OE P LANE .
3 arbitrary constants introducer] 1 *], i
consequently functions of on y the cn.ffi 'T 8 '7 te sfo >*tion ; and they are
"'being eo.it is obvious that Lr^'t 8 ^ f" ^ f * D ~* "
of the pencil of _ 2 ,. at ^ " be respectively equal to absolute invariant,
curve of the order D + l i, =J(/+ nTfl ' f, he j* 50 '"' 6 ln * of the general
each of the dps, honee i,, the prLt oL *n, , ~ i^n '^ ^ " MdaoHon =1 ' for
406i and there are th, 4^7 a h !,7 ' * ( ^ + !) ( ^ + ^H*SI))8,
oql to absolute invariant the pet l^TT* * %' ' ^ f ">">
of the pencil, there a re 4D6, each o he ', t ^~ 3 abs Ulte """*
cmve, and consequently indepe, <l e f O f H ?" '" " absolllte "variant of the
d,ffe out planes, thail joilli ^ ^ e ^ d ransformed curve as situate in
point on the transformed curve we hnvf g " Wl m ' Ve with the oorrespondi lw
*): if the t,vo curves : %r;.; ol , s S 7 S , of ". Arming a scroll"
see ,on by the plane of the oriri lm l "! vn ; ' " res P eo "ly, then the eon pleto
d of ' Derating En.. J'ti , y l^! ,"? f ^ of the order
transformed curve i s made of ^^ &l* sect.on by the plane of the
inc.. Conversely, given a sc ^,, of ' *e order ', and of genei ^
tic od .
, c o
be.ng ln ,,. Olu , ves rf the tic od _ + sMtona ^
e _
of the other; but for t !lo ^ scroll of ,1 ' , "^ ta 8f otion e the one
. breaking up as above th <ndar + ' ; ' i> not possible to find
UN TIIK ('uKUKHl'nNUKXi'K OK T\Vi) I'nlNTrt ON A (IIMtVK.
A.ii .l/Mf/mdM/t'tif ,S'.,,!V/y. tul, 1. {iMli
V, II* i<l A'iil Itl, iNiiii }
VII.
fiiul
lllf '!
111 Mtiriii
M ti J*
I h .'I.'
t** 6?.*'"!i r
hit'.*' im l'6Utl"l (I s ', I) 4 '* II ul lll' 'lrl)M' I * 1
at I ',
I thcit tf
10 OX THE CORRESPONDENCE OF TWO POINTS ON A OUHVK [{J8. r J
HirvL' () (the equation of the curve @ will of course contain the coordinntGH of J*
us maim:ters, for otherwise the position of P would not depend upon tlmt of ./>),
I iiml that if the curve has with the given curve k intersections ut the point /*,
hon in the system of (P, P) t tlie number of united points is
'.vhenc. in particular, if the curve does not pass through tho point 1\ thon
"ibvr oi united points is = + ', as in a unicursal curve.
* The foregoing theorem is easily proved in tho particular liasu whom th X

y, ') m,d of the order in u . 'Tf fu oto " t tho ordur iu
' rf the,,, vanish, id.nBc.ny. o, b^* 1P '7 T BU h "^ "'"^ <lu '"
y.) = fc ,. 4 ' "* " =
,
of ft .8
o the n ooo,,,
. have ff=o, an el similarly if' rv / J" 1 *' ' } ' m tho
!"'. J wo have ff' = . The eo Z n ( !' ^ *? a '' " 10 "'''i^ of tho
wnluiatcs of the given point P ^7 ' COH8ld( g therein (,, ,, s ) , t ,, n
/, as L ters "
=
have ft,, a the case above SUHM8 .j"^^ a havmg a /{4l , 1)lo iui
1
"f h ,
but i T ? 0+ii the total "umber of 7,,, 7 ' W ' UOh m '
of
un e ;"v;d,tha t upon writing ( ,
are ~ iv ' "
ni
vJii Is, *'. 'i. i !<
*,( < ( s oj I * /. , f
MtlVr. / " It, Julcl ()(
tmtf
I "1
.Mn]si):i'i . S'i tie [!. tfi"it "I n rune il ill
ini,i m" lli mid i i n.'H M! llii" IMIV v, ifh <h
!!l" nmd'l J"mf i S'i !hl','
(D (.,i I t/,,
/'
(Sti .. d
in
* w i'i"i, 
12 ON THE CORRESPONDENCE OF TWO POINTS ON A CURVE. [385
tangents from F to the curve, that is, ct  n  2. Hence the number of inflexions is
(2)+4J} I =m + n4 + 2(2ro+2), =3(wm), which is right.
8. For the purpose of the next example it is necessary to present the fundamental
equation a = a + a'+2/ i ;D under a more general form, The curve may intersect the
given curve in a system of points F each p times, a system of points Q 1 each
7 times, &c, in such manner that the points (P, P'), the points (P, Q'), & c ., are pairs
ot points corresponding to each other according to distinct laws; and wo shall then
have the numbers (a, a, a'), (b, ft 0'), &c, belonging to these pairs respectively j via.
(1,1) are points having an (a, a') correspondence, and the number of united points
is =a; similarly (P, Q') are points having a (ft ff) correspondence, and the number
ot united points is = b ; and so on. The theorem then is
9. Investigation of the number ef double tangents :Take P', an intersection with
the curve of a tangent drawn from P to the curve (or what is the same thing P P'
"* "" Cm ' Ve); th Unikd P^ ta hero the "o nts "of
c ero the po nts of
ontact of the several double tangents of the curve; or if T be the number of double
"
r r 
s^wKvTsv 1
^ , v;> ana ^T, /y, ) to the points (P, P'),
2(a~a') + 2T/3 / 3'=
Moreover, n,m the last example the value of .._, is ,, D , ftnd fchfl
2r^^2(n0)JDj
but from above it appears that we have fiff (n  2)(m _ 3)) whence
ON Til 10 milWWI'ONI>l'JN(!K Oh 1 TWO PO1NTM ON A (HWVIO.
henre, il' (/i, ;>) he tlm iOmm<!lerinl,i<!M n' Mm nyHtein of (lOiiinn (4#), Mm number of tlm
nonien through ./' in '/ A l l!IU ' n ' these I'"" 1 with Mm tfiveu <mrvn I iiitnrriiutlJnn ut
/', mid (KinmtipmnMy k.fi, Mnrnover, eaeh ni' 1,1m (umiiH bi^idus nmntH Mm e.urve in
(2m 1) pointN, and (uiimnmmnMy a t .; a.' /L (2w,  I), jlnncii Mm rnnniila ^iviiH l'.ho
ininilici 1 of nnili'il
or, HM tliis may bn
': IUI I l'l I ill (tyi I'}.
llulp Mm HyHlinu !' rnnicM (l^) rniidiiim (iljt  1/) i(inlriiurH (t'Ofw/Hfls wfnmnuiit
1'iinli M' whir.h, vi'Kimlril us a ptiir nl 1 r.iiiii(!)(lonf> lini'M, nmnln tlm givnn r.nrvn in
nuiuniilt'iil, pninUM; thai, SH, UIM ]Hmil,]tuir in l.n bit niiimidiiniil IIH IL i
hlit! tfivi'ii c.iirvn in HI puiulH; mid Mmm in on thin lumouiit n t'nliu^
() in tlm number nl 1 Mm unil.id pniuln ; whounti, Ilimlly, Mm niinibm 1 f Mm
H (I./) (!) in M/WC inm. It in liardly nrnwary l,u remark Mini. il> IH iwmnimtl
Mm nmdiMiniH {$%} \\\\\ ntntUl>mim having im Hjmdiul ri'lalimi do M
II. An H linal example, HiipptiHi that Mm puinl, 7' uu a ^ivnu curvo of Mm inilnr
nt. and t.lm point. (,? nu a K' v "" nill ' Vl ' "' )lln nnll<r m '' ' iavn au t. ') nim'iiHpinHlounn,
ami ln, il, In ivipiinid hi linil tlit fla^t ul' Mn> mirvi' cnvolnpcd by Mm linn i'Q,
Take un iirbitravy pitinl O, jnin fUJ, and Ittl. Miw meet llm uni've m in /", then
(/', /") itit* puintN nu Mm eurvn m, Imvin^ a (Vet, ma') nurn i Hpniitlitin!o; in limt., l.n a
^iven pciHitiinu ul' /' I here (tuiTi'Mpuml ' piwilimiM "!' Q, and t,o eiL^h id' Lhese. ?
puKiMuiiH til 1 /*', lhat in, I,M 1'iieh pusitinn ul' /* there mnvNpund waf pnMiMntin nl' /";
and Himiltivly In each piiHiMnn nl' /'' there mm'spttiiit m' pnsiiiunH "I" I". The dill've
H in the nyHleni ul' Mm lines drawn IVmu itwh of llm ' puMitioiiH nl' Q to Mm
pnint U, henen Mm IMIIVC H ilnei imt JIUHH Mtniit^b 7 1 , mid we have /,:'(). Uniirm the
uiLinht'i' nl 1 tlm united puinls (/', ]"). Mind in, Mitt mnnher !' the liimw ./'^ whicih
thrungh tht pninl f>, in i 7itV I ?it', r thin ! the rliiHM til' tlm niii've envnlnped by
1JJ, 1 1. may bit remarked, I hut ii' Urn two enrve.s are tiurviH in Mpatie (plann ni nl'
dimhln rnrvatiirc). tlmn the liki? retmoning H(IWM that Mm numhor if tlr '" ..... '"'
whinli nittub a given lino in ja'HHt'o, tliitt. in, Mm tirdnr ol' Mm HCIM
by Mm Un /'(, in MHI' + ?'.
ON THE LOGARITHMS OK IMA<;,\.\  !V
10* O.f tlm ./.<
,"
[From the PIWQBC
1'P ^0 H Knit 
y ol' (ihu low
but wo may ah) fl p ca i f
Writing tluis
and similarly
wo have of course
We have
_ _ P KJ j' fl rt
moreover bo
^
a,. HO that wo Jmvd a^t^^y
i( > ON THE LOGARITHMS 01? IMAGINARY QUANTITIES.
mid similarly
log P = log r' + iff )
and
logp=:logp + f
Hence
log P  log P = log + i (0 ff ,/,),
so that, by what precedes, log P log P', if the chord 1"P, coiMuloral H drawn I,,
P' to P, cuts the negative part of the axis of . upward.,, fc = l (lg .,. 2lV . if ^
chord out, the negative part of the axis of , downwa,*, iO is = log 2i7r , a,,<l
in every other case it is = IOET
*
Consider the integral
path of the variable , ; in order fco ? , Iimi J *~ P > <' <V it nay dopcmd on bho
* 1 p rt vAvi\j4. uu yivti IjiiO llrtrnrii\ r. & ^
therefore fix the path of the variable , and I r T^ H ^ mlicaliion ' must
the right line P'P w f ' j 7 y tlUt " lff tho l )ath fco bo
gw nne j P. Wnte Mw ^ p/> ^ du
PI " ilUIVU , ! fflrtss / J (Vllr,M, ... . 'I . 1,
^ w
******>> ** *
is a, on g t , le right lino , , P (fclwt is from uio
coordinates whereof are 1 *o M H
*> yu, to the pomt
We have thus
'*"* ^! U
the path in each case beW a ririlt , fl
n , p g UG ^ ab Ve ' The indefinite integral ^^lo,^
and as u pa8ses f rom j t P . ^ / w Io tt '
P" there 18 no diseontinuity in the value of lo gw; tho
386] ON THE LOGARITHMS OF III AGINARY QUANTITIES, 15
It is to be observed that 6 has always a determinate unique value, except in
the single case y 0, us negative, where we have indeterminately Q ir,
It is further to be remarked that, taking A for the origin of coordinates, we have
6 angle asAP, considered as positive or as negative according as P lies above or
below the axis of is.
Starting from the equation.
P = re ie ,
we have similarly
P'rV ',
and
P ' u
~ ft 1 ?
P' r '
p
where </> is derived from  in the same way as Q from P t or & from P'.
Consequently
e i<00'<) l f
and therefore 6' <f> a multiple of 2??, say
06'$ = 2flwr,
and in this equation the value of in is determined by the limiting conditions above
imposed on the values of 6, &, <$. To see how this is, suppose in the first instance
that the finite lino or chord P'P, considered as drawn from P' to P, cuts the negative
part of the axis of as upwards ; P is then above, P' below, the axis of as ; that is,
6, 0' are each positive ; and drawing the figure, it at once appears that the sum
& + ( Q'}> that is Q 6', is a positive quantity greater than IT. And in this case the
angle </> will be equal to 2?r (Q &') taken negatively, that is, $ ~  {2?r (0 0')},
or ~ & iJ3 = 2vr. Bub, in like manner, if P'P out the negative part of the axis of
so downwards, P will be below, P' above, the axis of CD ; 6 and &' are here each
positive, and the figure shows that the sum + 6' is greater than ?r; and in this
case the angle $ is = 2?r ( + 6'} ; that is, we have & & <f> =  27r. In m
other case, (that is, if the ' chord P'P cither does not meet the axis of .u <w
meets the positive part of the axis of to,) 6 Q' and </> are each in absolut'
less than IT, and we have #' (= 0, So that w '' '  1? 
16 ON THE LOGARITHMS OP IMAGINARY QUANTITIES.
and similarly
, lg P' = log r' f i&\
and
ll> 15J llf>M<f.
J^ o A' 7^
Hence
.0 that, by what p.eede, logP^P', if the choi , d
* to P, cuts the negative pa,t of the axis of Hpwards , is .^ if
chord cut. the ^ w part of the ax , rf m
m every other case it is = loo L
& p '
or from telow to above ti o f PM P C m & m ab Ve * Wow
the logarithm chan g e S from +1 to ' "1 o f ". rf ^ " le imf P art '"'
Consider the integral
path of the variable ,; iu order to fL , ' ^ ^ '* may de P end on tlu!
' ta
Wrfh now ^ p ,, ( _ we lmye _
p; a,d it is easy to see tllat , the path of
the righfc Une x fo P
coorfinatea whereof are l, v . 0> to the point ) .
We have thus
'J* Ji '
the path in each case being a right line as abov e. The
and aa * passes from 1 to ~
386] ON THE LOGARITHMS OF IMAGINARY QUANTITIES. W
value of the righthand side is thus =log~. As regards the lefthand side, the
indefinite integral is in like manner = \ogz; hut here, if the chord P'P cuts the
negative part of the axis of to, there is a discontinuity in the value of log 2, viz.,
if the chord P'P, considered as drawn from P' to P, cuts the negative part of the
axis of as upwards, there is an abrupt change in the value of log z from  iir to + iV ;
and similarly, if the chord cut the negative part of the axis of at downwards, there
is an abrupt change from + iir to ITT; in the former case, by taking the definite
integral to be log P log P', we take its value too large by 2zV, in the latter case
we take ib too small by 2wr; that is, the true value of the definite integral is in the
former ctiso = logP  logP'2wr, in the latter case it is = log P  log P' + 2m, But
if the chord PP' does not cut the negative part of the axis of ?, then there is not
any discontinuity, and the true value of the definite integral is log P  log P', We.
have tints in the three oases respectively
P
jp
which agrees wilih the previous results.
It may bo remarked, that it is merely in consequence of the particular definition
adopted that there is in the value of log P a discontinuity at the passage over the
negative part of the axis of to ; with a different definition, of the logarithm, there
would bo a discontinuity at the passage over some other line from the origin ; but a
discontinuity somewhere there must be. For if, as above, the chord P'P meet ^the
negative part of the axis of IB, then forming a closed quadrilateral by joining by right
p p
MUCH the points 1 to P, P to P, P' to p r and p to 1 ; the only side meeting the
fdz
negative part of the axis of a> is the side P'P ; the integral J , taken through the
closed circuit in question, or say the integral
+v\ dz
h
has, by what precedes, a value in consequence of the discontinuity in passing from
P' 'to P; viz., this is =2t7T or == 2wr, according as the chord P'P, considered as
drawn from P' to P, cuts the negative part of the axis of at upwards or downwards ;
but) this value  2wr or + 2wr must be altogether independent of the definition of
the logarithm ; whereas if, by any alteration in the definition, the discontinuity could
bo avoided, the value of the integral, instead of being as above, would be = 0. The
foregoing value  2wr or +2tV is in fact that of the integral taken along (in the one
0. VI. 3
18 ON THE LOGARITHMS OF IMAGINARY QUANTITIES. [386
or the other direction) any closed curve surrounding the point 2 = for which the
function  under the integral sign becomes infinite : but in obtaining the value as
z
above, no use is made of the principles relating to the integration of functions which
thus become infinite.
The equation
log P = log r + id
gives
PIU ~ gm log P J.TO gi'iiiO
or say
(ea 4 iy} = r m e !me ,
where, m being any real quantity whatever, r m denotes the positive real value of r m ,
We have thus a definition of the value of (0 + fy)* and the value so defined may
be called the selected value. And similarly, for an imaginary exponent m^p+qi, we
have
r p
which is the selected value of (
It may be remarked, in illustration of the advantage (or rather the necessity) of
having a selected value, that in an integral fa, taken between given limits along
a given path it is necessary that we know, for the real or imaginary value of z
correapondmg to each point of the .path, the value of the function Z, and consequently!
Z is a function involving log, or.*, the indeterminate,! ess which present itsol
m these symbols (considered as belonging to a single value of ,) i S) P SO to
indefinitely multiplied, and fa is really an unmeaning combination of symbols,
by selecting as above or otherwise, a unique value of log* or & we rondo,,
function to be integrated a determinate function of the variable. '
387.
NOTICES OF COMMUNICATIONS TO THE LONDON MATHE
MATICAL SOCIETY.
[From the Proceedings of the London Mathematical Society, vol. n. (1866 1869),
pp. 67, 2526, 29, 6163, 103104, 123125.]
December 13, 1S66. pp. 67.
PROF, CAYLEY exhibited arid explained some geometrical drawings, Thinking that
the information might be . convenient for persons wishing to make similar drawings, he
noticed that the paper used was a tinted drawing paper, made in continuous lengths
up to 24 yards, and of the breadth of about 56 inches 0); the halfbreadth being
therefore sufficient for ordinary figures, and the paper being of a good quality and
taking colour very readily. Among the drawings was one of the conies through four
points forming a convex quadrangle. The plane is here divided into regions by the
lines joining each of the six pairs of points, and by the two parabolas through the
four points; and the regions being distinguished by different colours, the general form
of the conies of the system is very clearly seen. (Prof. Cayley remarked that it would
be interesting to make the figures of other systems of conies satisfying four conditions;
and iii particular for the remaining elementary systems of conies, where the conies pass
through a number 3, 2, 1 or of points and tor ' n "* "" A " f K """
0 XOTICES OF COMMUNICATIONS TO THE [387
algebraical sum of the distances of a point thereof from three given foci is (this
V,;H ^lecUfl for facility of construction, by the intersections of circles and confocal
<'uni<s). The ijiuirtic consists of two equal and symmetrically situated pearshaped
curves, exterior to each other, and including the one of them two of the three given
f'R:i, the other of them the third given focus, and a fourth focus lying in a circle
with the given foci: by inversion in regard to a circle having its centre at a focus
the two pearshaped curves became respectively the exterior and the interior ovals of
:i Cartesian. There was also a figure of the two circular cubics, having for foci four
iriveu points on a circle; and a figure (coloured in regions) in preparation for the
'"iistniction of the analogous sextic curve derived from four given points not in a circle,
March 28, 1867. pp. 2526,
_ Professor Cayley mentioned a theorem included in Prof. Sylvester's theory of
Imvation of the points of a cubic curve. Writing down the series of numbers
1 2, 4, o 7, 8, 10 11, 13, 14, 16, 17, &c,, viz., all the numbers not divisible by 3
then (repetitions of the same number being permissible) taking any two numbers of
he H*IC.,, we have m the series a third number, which is the sum or else the
ifiom.ce of the two number (for example, 2, 2 give their sum 4, but 2, 7 give their
ftore.ce o) and we have thus a series of triads, in each of wnich one number
elLruct I ,e!t tf loirT ***** *' **"* * ****** m & CuWc
rhCH ^ ^l?^ 1 ?^^ ^^l^"
'
.
H >. the third point g of f anf 7 th tc'ond nf ?? **!, *""' ^ fil ' 8t f them '
Wo have here L theo rem J't! t W o'l ^ *"", " " "f ****** of 4.
"f i Similarly, 10, 11, 13 ,, each J Ten /I L m t  7 ,", a ' S
It. 10, 17, 19, each of them by t l u , e e nslSnl } i ". tamaaA ^ two
tractions increasing by nnity L , ZB of ' 1 n; &e numhor of
that the.0 constactions, 2, I ^^ SLJ?* ""^ ^ ^ theorem is 
o point. Prof. Cayley mentioneTthat on , 7' b ' 5 al " OM md *e
p. 29.
O to
on a straight line. ' ^ f * heae P mte >=eing flxed , whie
387] LONDON MATHEMATICAL SOCIETY. 21
March 26, 1868. pp. 61 G3.
Prof. Cayley made some remarks on a mode of generation of a sibireciprocal
surface, that is, a surface the reciprocal of which is of the same order and has the
same singularities as the original surface.
If a surface "be considered as the envelope of a plane varying according to given
conditions, this is a mode of generation which is essentially not sibireciprocal ; the
reciprocal surface is given as the locus of a point varying according to the reciprocal
conditions. But if a surface be considered as the envelope of a qitadrio surface
varying according to given conditions, then the reciprocal surface is given as the
envelope of a quadric surface varying according to the reciprocal conditions; and if
the conditions be sibireciprocal, it follows that the surface is a sibireciprocal surface.
For instance, considering the surface which is the envelope of a quadric surface
touching each of 8 given lines; the reciprocal surface is here the envelope of a quadric
surface touching each of 8 given lines; that is, the surface is sibireciprocal. So
again, when a quadric surface is subjected to the condition that 4 given points shall
be in regard thereto a conjugate system, this is equivalent to the condition that 4
given planes shall be in regard thereto a conjugate system or the condition is sibi
reciprocal; analytically the quadric surface oaf + by* + cz* + dw = is a quadric surface
subjected to a sibireciprocal system of six conditions. Impose on the quadric surface
two more sibireciprocal conditions, for instance, that it shall pass through a given
point and touch a given plane, the envelope of the quadric will be a sibireciprocal
surface. It was noticed that in this case the envelope was a surface of the order
(= class) 12, and having (besides other singularities) the singularities of a conical point
with a tangent cone of the class 3, and of a curve of plane contact of the order 3.
In the foregoing instances the number of conditions imposed upon the quadric surface
is 8; but it may be 7, or even a smaller number,. An instance was given of the
case of 7 . conditions, via., the quadric surface is taken to be o# a } 6?/ a + c# 3 _+ dw* =
(G conditions) with a relation of the form
Abo + Boa, + Gab + ffacl + Gbd I Red =
between the coefficients (1 condition) ; this last condition is at once seen to be sibi
reciprocal; and the envelope is consequently a sibireciprocal surface viz., it is a
surface of the order (= class) 4, with 16 conical points and 16 conies of plane contact.
It is the surface called by Prof. Cayley the " tetrahedroid," (see his paper "Sur la
surface des ondes," IAQU/V. torn. xi. (1846), pp. 291 296 [47]), being in fact a homo
graphic transformation of Fresnel's Wave Surface.
[Prof. Cayley adds an observation which has since occurred to him. If the quadric
surface am? I faf + ex* + dm* = 0, be subjected to touch a given line, this imposes on
the coefficients a, I, G, d, a relation of the above form, viz., the relation is
A z lo + B*ca, + &al) + Pad + G*bd + H*od = ;
where A, B, G, F, G, H are the "six coordinates" of the given line, and satisfy
therefore the relation AJF + BG+ <7I/= 0, It is easy to see that there are 8 lines for
which the squared coordinates have the same values A*, B*, 0\ F\ G 2 , f/ a ; these
8 lines are symmetrically situate in regard to the tetrahedron of coordinates, and
22 NOTKWS Ol' 1 rOMMinVK'ATIONH TO TIM; I
moreover they lio in a hyporbolmd. Tim iuiulrin mirliuv, instnid ..!' Imin^ ilrliinul
above, may, it in cloav, bo dnfinod by thn i!iiiiviilniil. u lil.iuu>i n' fniii'liimf ( . m ')i ,,f , i"
8 given linos: that is, wo havo Um oiivrlnpi> nf a ipmdrin Miirliuv hnuliiiiir ,,(,[, ii(
8 given lines; thosu HHUH nob bring urliit.niry MUCH, bill liciii^ a K.Vdfrm il' u \vi'v liiMri'il
form. By what procedoH, tho nnvnL.pn i>t n .jiiai'Uc Hiiriinv. li. 'IIIHIU;I, linwovcr I'l'^r
in virtue of tho rohition AF + JH}* (!Jf ?>(), Uim \ i,u tnu^.r M piupn'^uuili,' nmfi,!!^
but that it rosolvon itsoU' into MK abdvciiioiil.i d liyp.'rbi.lni.l mKru i.win ' TrY
is, icatoi'ing tho oiiginal A, it, &,!., in p llt v n f / '/.' J . ,<T M ll ivrln.. ' !' i')",
Xic I Jka + (Jab + 2i\id \Qbtl \ Jfal  0, (which in in ijPi'irnil u h'tiahrdmitl), in wi,".*"
^1, A (7, /'', ff, // am tin) Hi]imnl (!oonliiml,oH uf u Im, (nr, wltiif i^ il,r ^i,,,, u^)"."
passing through tho givon lino and tlirnngh the HyimMulnVMlly''^!., 'u,' L'v'n nllMrii!',!!' J' 1
la, IHli.S. pp. ion, nil,
Frofiw Onyloy g,tvo an uooouni, h, i lr M.tin,/ ul' ,i M,,,,,,,. liv ,.. ((i , H(( ,.
Comus rmnnhohor ( on,ploxo ndo. Vonill^nHMimiMK ,,., Ktilrr',,!,,,!, H,,.,,J v u %
^ ....... ''"
) 
number of spaces, un,l .orUiu Huppl.n.nnlnry mnu.riH,, A' " J
1
.
H tho bottom bo also We,,,, away, ,,,, H i . t ' ' ^ ' , 7 ,"T"i " '' "" '' ........ "
supplomonhuy (1 nni.lilii!H ,., i,, ,,, ,,,,, ,T, .',, ' .' "'" , ..... ' ..... "' " f "'
The ohiof diniculty ,! i ..... ,, r ,,, ''., "'.' """ lv "' ' *'l(./..*")..
8l Wlomo,,tavy ...Jatta ^ J ,?, ..... '' '"" '" ' ........ "" ...... ""' ...... '' ......
December 10. 1808, ini l'
1P l
OlMorratioiw by I'rofoas
, 
I 1. 2, 3, au,l any ,, tl lr(!u i jT '," , '"' ""* >' liui " ..... >
io C T OS ' caoh ^ * Hi \ ;. T , ,; v ;r"' r *
*. o, 8, wo havu a tmnfo, vt io n ,, . ,, . i " y ' ,:' ''' " "' ,! hl ' "'"'I' 1 " !
system " '' ' *"A, . 1, ; ^, l,, u , illK U u t , lllV( , nM ,
Si o >
points 4, s, 0, rat i n >0 8implo ^ r "' 2 , "f," ' ^'" Ol " ro 'l 1 ""' 1 ^ having
387]
LONDON MATHEMATICAL SOCIETY.
23
Analytically, Cremona's transformation is obtained by assuming the reciprocals of
"ai 2/a 2* fa ke proportional to linear functions of the reciprocals of as lt y lt z l (of course,
this being so, the reciprocals of *' lp ft, ?i will be proportional to linear functions of
tho reciprocals of flj a , 7/ 3) z.j). Solving this under the theory as above explained, write
M
a , /> , c \
i i_
r T
a' 3
u6i i/i Z\
1
} =
. L J, /_
. =
1
: + + i
if
Honco n
QJB^O, &c., arc qnavtioH, or generally aQ.12, + ^R l P l + yP, Qt = is a quartic, having
throo double points (y^O, ^ = 0), (^ = 0, ^ = 0), (^ = 0, ^ = 0), and having beaides the
throo points which are tho remaining points of intersection of the comes (&0, ^0),
( Ji, = 0, P, = 0), (Pj = 0, Q! = 0) respectively ; viz., these last are the points
"1 'I 'I o
:L . . :  = ei hf : fy  id : dh (/e> &c. &o.
Tho double and simple points are fixed points (that is, independent of a, /3, 7). and the
fonnulm coino under Cremona 1 * theory. It is, however, necessary to show that if the
point* < 5', 0' are in a line, the points 1', 2', 3' are also in a line. This may be
rtono as follows :
Lot thoro bo throo pianos A, B, 0, and let the points of the first two correspond
hv ordinary triangular inversion in respect of the triangle , on the_plane A, and _/3,
oil the piano S. Lot also the pianos S, C correspond by ordinary triangular inversion
in respoot of the triangle ft on tho plane B, and 7 , on the plane lh " "^
Hpondonoo between A and is tho one considered, the points 188 to ming the
r,,.^',:;". r ^^ r^js, i~
A^nlnrmim nrnnP.vtioa musl) apparently belong to uiemonuB uw
interesting part of the theory.
i's "TJeber
t
ip" Grelle, torn.
but in the present Paper wwnwty proportional to
388.
NOTE ON THE COMPOSITION OF INFINITESIMAL ROTATIONS.
[From the Q mrtef ly J omd of Pw . e and
^ vm .
pp. 710,]
n the last Smith ,
^ COS
ion to Denote M infinitesinml
*X rigidly ^ in the point
388] NOTE ON THE COMPOSITION OP INFINITESIMAL ROTATIONS. 25
where I, m, n, p, q> r are constants depending on the infinitesimal motion of the solid
body.
Hence, first, for a system of rotations
&>! about the line (a lt b lt c,, f lt </,, h^,
0)3 ,, (tt a , y 3 , C a , J 2l (/ 2 , fly),
the displacements of the point (#, y, z}, are
Bo) = . ?/Sca> s2ow 4
Sy 2cw , + zZaw 4
and when the rotations are in equilibrium, the displacements (80, Sy, 8,2) of any point
(a), ?/, z) whatever must each of them vanish ; that is, we must have
2wa = 0, S<u& = 0, SMC = 0, Sa>/ = 0, "Za>(f = 0, Sw/i 0,
which are therefore the conditions for the equilibrium of the rotations w lf a) 2 , &c.
Secondly, for a system of forces
P l along the line (a,, &,, c,, / 1( (/j, /M,
* 3 )i Jl V*3) ^3) C a , _/ a , (/ 2 , ft fl ),
&c.
the condition of equilibrium as given by the principle of virtual velocities is
SP (al + Imi + on+fy + gq + kr) == ;
or, what is the same thing, we must have
which are therefore the conditions for the equilibrium of the forces P u P 3 , &c.
Comparing the two results we see that the conditions for the equilibrium of the
rotations WD w a , &c. are the same as those for the equilibrium of the forces P l} P a , &e. ;
and since, for rotations and forces respectively, we pass at once from the theory of
equilibrium to that of composition; the rules of composition are the same in each case.
Demonstration of Lemma, 1,
Assuming for a moment that the axis of rotation passes through the origin, then
for the point P, coordinates (as, y, z], the square of the perpendicular distance from
the axis is
= ( , y cos y + z cos /3) 9
+ ( x cos 7 . z cos ) a
+ ( to cos /S + y cos a . ) a ,
C. VI. 4
2ti .VOTE OX THE COMPOSITION OF INMNIT.ESLVIA1, IIO'I'A'J'IONH.
aid tho expressions which enter into this formula donoto (is fnllmvH ; via if l.lih Hll r]
^ lmt * at right angles to the plane through P ftnd t h axm nf mlul,i, m
Mil ""f n* : Iie rt Ular diStanCG f '
< Mirmtw oi Q refonwl to P as origin are
a cos 7 , _ g cos ffj
 A'cos/9+7/cosa
oach
' .V*, *) in place of ( an H
coordinate., (, J, c , f . n : '' ^' allt
" 4l <
cto .
of l emma 2.
gts ~
,"' "'" '''"' ........ ' ""'i''
to t ], ..inilind.],,,^ ( ,, u ,
n,l, ...... . l,v ,
I G f M .
Motion of the
389.
ON A LOCUS DERIVED FROM TWO CONIOS.
[Prom tho Quarterly Journal of Pure and Applied Mathematics, vol. vm. (1867),
pp. 778*.]
REQUIRED tho locus of a point which is such that the pencil formed by the
tangents through it to two given conies has a given auharmonic ratio,
Suppose, for a moment, that the equation of tho tangents to the first conic is
(HJ ay) (as by) = 0, and that of tho tangents to the second conic is (co cy) (at dy) 0,
and write
G (a
so that
write also
/,* / 
**~ A* lv *~A'
then tho anharmonic ratio of the pencil will have a given vali
(&/<;,)(& 70 = 0;
that is, if
or, what is the same thing, if
that is, if
 (B0}* =0,
^ 8 ON A LOCUS DERIVED PROM TWO CONICS. [389
where
are each of them symmetrical in regard to a, b, and in regard to G, d, respectively.
Let the equations of the two conies be
U =(a, b, c,f, (J , A>, y, 2) a = 0,
V (*',&', ef i f,g',kJa ll y l z^ = Q i
and let fa, ft 7 ) be the coordinates of the variable point. Putting as usual
(A, B> G, F, G, tf) = (i c / 2 , caf, i/ t , g} t af, kf~bg t fgch),
K = abo  af*  hf  c /i
the equation of the tangents to the first conic is
(A,B, 0, F, G,
where
X^Wfy, r^
and therefore
5l ,b sti t u ti, lg for 5 the vall>e _i (tf
tangents, an equation of the form aZ + 2hZ7+h7*n w l,i n i i ^ ,
to b e (Z7)(X~ & 7) = 0; that is, we ht e ^ " efa ' ^ taken
1 : ft+& : 06= a ; 2h : b;
an<l, in like , Mme , if the accented letters refer to the second code
1 : c f d : cd = a' :  2h' : b'.
Substituting for a, h, b their values, and for a' h' b' th
. ana 101 a, h, b the corresponding values, wo find
'
' a + b : ab
i c
0/aa
We then have
...Ja, A 7 ),
389] ON A LOCUS DERIVED PROM TWO CON1GS. 29
and similarly
( d)' = 4W. ..., /3, 7)'
We have, moreover,
(a + 6) (c + d)  2 (06 + cd)
= 4 (//7 3  ^7  Gf3>y
 2 (7V
 2 #
and substituting the foregoing values, we find
or putting for shortness
= (EG 1 + '0  2J!f, ., ., Gil 1 + (?'//  ^J"  ^'^, ., . Ja, ^3, 7 ) a .
the equation of the locus is
where (a, /3, 7) arc current coordinates. The locus ia thus a quartio curve havi
quadruple contact with each of the conies Z7=0, U' Q; viz. it touches them at th
points of intersection with the conic = 0, which is the locus of the point such ti
the. four tangents form a harmonic pencil.
The equation may bo written somewhat more elegantly under the form
30 ON A LOCUS DERIVED PROM TWO (.IONICS,
so that
A  2k ,.F=* mr
(BO ^ ,..) = (w)' jiff, H/J /,
and substituting these values the oqimtion in
which, if A, B, denote
a, p, 7
. /?, 7
i
fl, j0, fy
a > ft> 7
I , m, 11
f , m', n'
i , W, ?l
Xi tf'i ''
P, 9. r
p'. g', ?'
f, wi', n'
^. 2, '
}' J ,
i P> 7
a
a /3 > 7
a_{
i /3, 7
a, ft 7
_.
i, m, ?i
/' / /
f , m, n
^ fli, w
J?>> f/. ''
P. ff '
X. ?', *'
r, W', ft'
^'. '/, ''
, 7
respectively, (A + B + 0=0} is, in fact, tho
(2fi + l)M 9
or, what is the same thing,
that i
s
', m', *r'
,
7
jugate axes, the
we have K^abo, K'^
O.
ON A LOCUS DERIVED FROM TWO CONICS.
I suppose in particular that the two conies are
a? 4 inif 1 = 0,
fcho equation of the quurtic is
1 ma? +  1 _ a m ?  m  1 =
,, N ,, . .
or putting A, = jf/o/nrv5 tnis 1S
njiS _L 1
To fix the ideas, suppose that m is positive and > 1, so that each of the conies
in nu (illipso, tho major semiaxis being =1, and the minor semiaxis being = ,, ; .
3?or any roul valuo of Js the coefficient X. is positive, and it may accordingly be assumed
thai) X in poHitivo,
Wo have ..^..."t.. 1 ,> < 1, or the radius of the circle is intermediate between
m (m h I) in
liho aomiaxoH of tho ellipses, hence the points of contact on each ellipse are real points.
Writing for shortness
^  >
??i a + m
tho equation in
(a? + iwf  1) (ma? + f  1)  X (& + if a) 9 = 0.
For thu points on the axis of eo t we have
(fl)(flMJ a l)M< 1 a) B = 0,
that i
( m  \) & + { (1 + m) + 2X] a; 3 + (1  Xa fl ) == 0,
mid bhoucu
or, substituting for a its value, this ia
Eemarking tlmt tho values
are
32 ON A LOCUS DERIVED FROM TWO CONICS. [389
1 (m + l) a
and considering successive values of X; first the value \ = ,, =  ~, wo have
+ )
m/
m
1
m
or observing that
this is
^ = 0, or
\ w m
The next critical value is Xm. The curve here is
(a? + mf  1) (met? + fi\ m (& + ,*  a )a = Q
that is ' '
m (tf + # 4 ) + (l + m s ) a,' 2 ?/ 3  (m + 1) (a 3 + f) + 1
that is v ,/ y *
f m  I) 3 .ty + ( 2)w  m  1) (^ + /) + 1 _ , ?ia a = o,
or, substituting for a its value,
_ >
lmtfl&??li a8 _( m l)'(m
rt ,. . m( a
the equation is
.
m(m+iyt'
or, as this may also be written,
1
m H
in
or, what is the same thing,
in
389'
ON A LOCUS DERIVED FROM TWO OONICS.
33
which has a pair of imaginary asymptotes parallel to the axis of so, and a like pair
parallel to the axis of y, or what is the same thing, the curve has two isolated points
at infinity, ono on each axis,
linos
TKo next critical value in X = J(mtl) 3 ; the curve hero reduces itself to the four
,
(m
'
m
0;
and it in to be observed that when X exceeds this value, or say \ >(* + 1) 8 , the
cui'vo has no von! point on either axis; but when X=oa, the curve reduces itself to
(? I y 3 )"<), i.e. bo liho eirclo ro 3 + f  a. = twice repeated, having in this special
ciwo real pointw OH the two axes,
Ik i iww eauy to trace the curve for the different values of X, The curve lies
in ovory eano within the unshaded regions of the figure (except in the limiting coses
aftormuutiomsd); and it also touches the two ellipses and the four linos at the eight
points /i), at which points it also cuts the circle ; but it does not cut or touch the
four HnoH, tho two ellipses, or tho circle, except at the points k Considering X as
varying by HUCCUBnive stops from to oo;
A, = 0, tho curve is tho two ellipses,
\,
^^ M.!?JljA Vj the curve consists of two ovals, an exterior sinuous oval lying in the
\ W
four vegionH a and tho four regions 6 ; and an interior oval lying in the region e.
C. VI. 5
q *
ON A LOCUS DERIVED FROM TWO CONICS. Tggc,
1) 3
 there ,s still a s i oua oval M above>
aove> ^ .^^
\ w .
dwindled to a conjugate point at the centre.
; X "" ;
TO
; there is no interior oval, but only a
J
fc
, the onive becomes the four lines.
k>i(m+l) 2 , the curve lies wholly in tho f,,.
eonasting thereof of f olll . detaehed siuL ov a l s A TT  ^ ^ *" ^
!( + !). e^h oval approaches mm nea r' tnf ' ^ Y , lateS ' eSS from the
and inflate l ine . porti l w hieh temd " ' "^ W * ini1 f ed ^ the eide
And a X depart/ f fom the 1 tit T + , Z"" * V WhMl th OTal be
approaches more nearly to the circuil^ h PP '' 0a heS * "' each si
cont ^ 6eparates the 4
e crcu h
contains the dnnon. oval. ^ 6eparates the 4 ^giona 4 e , w hich
Finally, x = 0, the curve is t ], e ci role Wee repeated.
390.
THEOREM DELATING TO THE FOUR CONICS WHICH TOUCH
THE SAME TWO LINES AND PASS THROUGH THE SAME
FOUll POINTS.
[From the Quarterly Journal of Pure and Applied Mathematics, vol. vm. (1867),
pp. 162167.]
THE aides of the triangle formed by the given points moot; one of the given linos
in tliroo points, nay P t Q, 11; and on this same line we have four points of contact,
Bay A lt A tl A a , Ai\ auy two paii^i say A lt A t \ A B , A 4t form with a properly selected
pair, sn,y Q, R, out of the abovementioned three points, an involution j and we have
thua the three involutions
(A lt A; A aj Ai\ Q, .fi),
(A,, A s \ A it A, } H, P),
(A lt A,; A 2> AS\ P, Q).
To prove thia, lot ffi = 0, i/^O ^ G tno equations of the given lines, and bake for
the equations of the sides of the triangle formed Toy tho given points
b & + a y ~ab =0,
If oi + ' y  a' &' = 0,
the equation of any one of the four conies may bo written
Lab LW . L"of'b"
\j  ab b'os + a'y  a'V b"ai + a!'y 
and if this touches the axis of ic, say at the point co= a, then we must have
_^L JL' .. V/ = /f^") a ,
tvacKaiv a" () )(; a') (a;  a") '
52
Uti THEOREM RELATING TO THE FOUft CONICS WHICH TOUCH THIS
>r, aiming as we may do, #=('<)(") ('), this gives
L a =( ft op (a 1 a"),
MUW
manne, if the conic touch the axis of y, 8il y at the poi.,;
ifi =(b ^(fi'fi", '
and thence
p, ming 
P = a & (a'  ft ") (y j/^
f =a'b' (a,"  a )(b"~l)),
wo have ^V(a0( ~V),
and thenco
; * ;
wo have in like mamel .
 _
"' '
p , Putting
equation is '" ' ^
390] SAME TWO LINES AND PASS THROUGH THE SAMS TOUH POINTS. 37
and "by attributing the signs + and  to the radicals, we have, corresponding to the
four conies, the equations
(a  ,) V (X) + (a 1  ffl ) V (A") + (ft"  a.) V (X"} = 0,
 (a  ,) V (X) + (a'  a a ) V (A") + (a"  a 3 ) V (X") = 0,
(a  .) V TO  (a'  a,) V (T) + ( a "  ,) V ( JT") = 0,
(a  Oi) V (X) + <'  ,) V (*')  ("  <0 ^ (A) = 0,
where a n 3( a 3) a., aro the values of a for the four conies respectively,
Eliminating a" we obtain the system of three equations
(2a  ,  a,) V (X) + (a,  ,) V (A") + (a,  ,) V (A r ") 0,
foi  ) V (A') + (2a'  ,  ,) V (A r/ ) + (a 3  ,) ^ (.Y")  0,
(a, + a  a,  a.,) ^ (A r ) + (a, } a 3  a,  a,,} V (A r/ ) + (a, + 4  a,  a 3 ) V (^") = 0,
and then oliminating the radicals we have
, a 
a, + 4 ~ a 
==0,
which is in fact
1, . +'j act,'  0,
as may bo verified by actual expansion; the transformation of the determinant is a
peculiar one.
The foregoing result was originally obtained as follows, viz, writing for a moment
a V (X) + (if V (*"') + " V (X") = ,
tho four equations aro
 a, <I> = 0,
 <E> = 2
these give
(a,
2 (a  B
38 THEOREM RELATING TO THE I'OITR CONICS WHICH TOUCH THE
From the last equation we have
(a,  ,) fc = 2 (0  a V (X)  a' V (Z')J  2a 4 {>  V (X)
= 2(i^)*2(a
that is
or substituting for J(X), *J(X') their values in terms of <]>, we find
which way be written
that is
T
or
ft Of.,
that is
or finally
which is a known form of the relation
I ~1 _ ft .
V~ aa >
o + ft', acs' = 0,
gives the involution of the quantities a, a'; ,, 4
We have in like manner
1 t . ,, . .
0,
and
''olutions of the systems < a "j a . a
[390
, w
and < a; 1( 3 ; . a4
39
390] SAME TWO LINES AND PASS THROUGH THE SAME FOUR POINTS.
It may be remarked that the equation of the conic passing through the three
points and touching the axis of m in the point ; = ia
(a  a)* (of  a") 6 , (ft'  ? (a"  a) V (a"  ) 3 (a  o') b" _ Q
" + ' ''' ^ ' '"'
and when this meets the axis of y we have
>0.
yb " r yV ff6
Hence, if this touches the axis of y in the point y = {3, the lefthand side must be
 (a ~ a) 2 (ft'  a") + , (ft'  a) 3 (a"  a) + j (a"   ) 3 (a  a') (y  /3) s
, ft a )
and equating the coefficients of  a , we have
i/
( a _ a ). (a'  a ") + ^ (a'  ) 3 C^"  a) + j/ ("  a ) a ( a ~ a ')
P (a _ ) (a'  O + ^ (a'  )* ("  a) + ^ (a"  ) (  a j
ft
_a
or what is the same thing,
fa  ,f (rf  a") + (  ) ("  a) +
s ft t( '
"
= 2/3 (ft  a) 3 ('  a") + ~ ('  ) 9 ("  )
which gives /3 in terms of a, that is ft, ft, A, A in terms of ,, a a , ,, 4 respectively.
Cambridge, 30 November, 1863.
391.
SOLUTION OF A PBOBLEM OF ELIMINATION.
[From the Quarterly Journal of Pure and Applied Mathematics, vol. viu. (1867),
pp. 183185.]
IT is required to eliminate en, y from the equations
a > b , c , d , e
a' , b' , G , d', e'
a", b", G " , d", e"
This system may be written
= 0.
if for shortness
Or putting
we have
2/ 4 = SXe;
 ft + X a j &c.
 A
"* fV
391] SOLUTION 01? A PROBLEM OF ELIMINATION. 41
or, what is the same thing,
X (6 + Ao) + X' (6' + &0 + A." (&" + Aw") = 3
X (o + to*) + V (c' + fctf) + V (c" 4 Ad")  3
X (d 4 Are) + V (d f + fre 7 ) + X" (<?" + to") = ;
and representing the columns
a b> a' b f , a" I",
b c, b' c', b" c",
c d, c' d f , c" d",
d e, d 1 e', d" e",
ty
1, 2, 3, 4, 5, 6,
each equation is of the type
X(l + fc2) + V (3 + fc4) + V (6 + fc6) = 0.
Multiplying the several equations by the minors of 135, each with its proper sign,
and adding, the terms independent of k disappear, the equation divides by k t and
we find
X 2136 4 V 4135 + X" 6135  ;
operating in a similar manner with the minors of 246, the terms in k disappear, and
we find
X 1246 + V 3246 4 X" 5246 = ;
again, operating with the minors of (146 + 236 + 245 + &24C), we find
X {1236 + 1245 + k (2146 + 1246)}
+ X' {3146 4 3245 4 k (4,236 4 324G)}
+ X" {5146 4 5236 4 k (6245 4 5246)} = 0,
where the terms in Is disappear, and this is
X (1236 4 1245) 4 V (3146 4 3245) 4 X" (5146 4 5236) = 0.
We have thus three linear equations, which written in a slightly different foi
X 1235 4X' 3451 4 ^" '
X (1236 4 1245) 4 X' (3452 4 3461)
X 1246 4V 3462
and thence eliminating X, X', X", we have
1235, 1236 + 1245,
3451, 8462 + 3481,
5613, 5614 + 6623,
C. VI.
42 .SOLUTION OP A PROBLEM OP ELIMINATION.
which is the required result. It may be remarked that the second and third
are obtained from the first by operating on it with A, A 3 , if A ~ 2^ + 4<8 S f (
say the result is
1235
(1, A,
3451
5613
= 0.
In like manner for the system
if the columns are
then the result is
a? , afy, ay, ohj\ ay*, f
a , b , o t d , e , f
ft' > &' , c' , d' , e' , f
a", 6", c" , d", e" , /"
ft'", &'", c'", d"', e"' , f"
a b, a'V, a"b", a,'" V",
b G, V c' } 6"o", b'"o'",
c tZ, c' d', c" d", c'" d'",
d e, d'e', d"e", d'" e'",
*/> e'f, e"f"> e'" f",
= 1,2, 3,4, 5, G, 7,8;
(1, A, JA
where
12357
34571
56713
78135
=0,
43
392.
ON THE CONIOS WHICH PASS THROUGH TWO GIVEN POINTS
AND TOUCH TWO GIVEN LINES.
[From the Quarterly Journal of Pure and Applied Mathematics, vol. vm. (1867),
L pp. 211219.]
LET ro = v0 be the equations of the given lines; s = the equation of the
line ioining the given points. We may, to nx the ideas, imagine the implicit constants
so determined that + tfM0 shall bo the equation of the line infinity.
Take ffmyO.flnyO as the equations of the lines which _ by their inter
section with = determine the given points. The equation of the come is
y (m) + \i (n)J V (wj) = a + y V () + 7*
or, what is the same thing,
(co  my) ((o  ny} + 2 (w + y V (*)) y* + y V = >
so that there are two distinct series of comes according as ^ (vm) is taken with the
positive or the negative sign.
The equation of the chord of contact is
& + y V (w) + y* 0,
which meets ,0 in the point { ^y^nn)^ , = 0} that is in one of the ^centres of
the involution formed by the lines (, = 0, y0). (flW0. ^0). It to be
observed that the conic is only real when im i* positive, that is (the lines ^ and points
being each real) the two points must be situate in the same region or in opposite
redon3 of the four regions formed by the two lines: there are however other real
cases e. K . if the lines a0, y0 are real, but the quantities m 7, are conjugate
imageries; included in this we have the circles which touch two real lines.
b ^
ON THE CONICS WHICH PASS THUOUGH TWO GIVEN POINTS [3Q2
f V (m) + V ()} V W = + y V (mn)  7 (a + y),
which gives two coincident points, that is the conic is a parabola, if
that is (1 " 7) {A/ (wm) * ?)  * (V W + V ()},
or 7 9  7 {1 + V 0)) == i {V (m)  ^ (n))
7 = HI + V (m) V {(1 + m) (1 + n )J],
where it is to be noticed that
y == i [1 + V (m) + V {(1 + m) (1 + 71)}]
is a positive quantity greater than V(Mi), say 7 =^
7 = HI + V (mn)  V {(1 + m) (1 + )}]
is a negative quantity, say T = g, ff being positive.
The order of the lines is as shown m fig. i, 8ee plafce facing p _
7 =  ro to Y =  3 , curve is ellipse; 7as _ fff parabo]a
7  ff to p, curve is hyperbola; 7 ^ pambola Pj)
7=p to 7 = OT, ellipse.
Resuming the equation
the coefficients are ~ ' *+ TV 0,
(. & ' c. / ^ /O  [1, wi, 7 9 , 7V(m), 7,
and thence the inverse coefficients are
(4, 5, a, j, t?, jj) =
or, omitting a factor, the inverse coefficients are
W. a, a, j ffl a,. [ 0) a,
Considering the line
the coordinates of the pole of this line ar
are
392] AND TOUCH TWO GIVEN LINES. 45
or (what is the same thing) introducing the arbitrary coefficient k, we have
kx + 7/i  v v* (inn) 0,
ky + 7\ v = 0,
kz~\*/(iwi)  /* g" y(m)  V ()1 3 v =
the first two equations give
k ; 7 : li^A W (*)) : v \V V(w) ffl ) :*#/*&
that is
/t\V(m)l ~"iy V (*") 0)
,
or, substituting this value of 7 in the third equation,
lAW(m)l* , w (ffw)} + ^W ^ 0>
Xw/*y l ^ (Gy*j(mi
that is
(m)  V ()}' I ( ~ 2/ V (
+ * {  ^ V (*)! y {/i  X V (wwi)} = 0,
which is the equation of the curve, the locus of the polo of the line X0 + /*y+M*=0
in regard to the conic
(<a  my) (a ~ny) + 2{ + y*/ (ma,)} i* + 7^ a = 
In particular, if A, = /i = y = 1, then for the coordinates of the centre of the conic,
we have
to . y : 2 = 7 + V(<) = 7 + 1 = V (> + * + jj^ W ( m ) ~ V ()) 8 i
and for the locus of the centre,
(ffl^i{V(m)V())' + (y>l^
so that the locus is a conic, and it is obvious that this conic is a hyperbola. Putting
for greater simplicity
{oy =A,
co  y V (ron) = Y>
z Z,
the equation of the curve of centres is
Z B . HV (m)  V ()]' + XY i 1 + V (*)] + YZ {1  V (n)} = 0,
or, writing this under the form
Y[X {1+ V (mft)} + J? {1  V (>))] + i IV (*)  V (n)) a X* = 0,
46 OX THE COMICS WHICH PASS THROUGH TWO GIVEN POINTS [Q
the equation is
YQ + X* = Q,
where
 V (
these values give
{1  V (m)] * = [\l (m)  V ()} Q + 2 (1 + V (wi)) ^,
or, what is the same thing,
(1  V (mi)} w=  V ( mn ) X + Y,
[I ~ V (mn)} z = 2 (1 + V (mn)J Z + { ^ ( w )  V ()}'
whence also '
{1 + V (')} JT + 27+ (^(m)  V (
or the equation of the Hue infinity is
{1 + V(0) Z+ 2F+ {V(m)  V()}' Q = 0,
bo apKed to
e asymptotes
7Q + X = 0.
In fact we have identically
asymptotes of the oouio\B=oH, n w ( . Ue n " ty> PUte iu "vidonoo the
fa' a Wntnl S *' /* 7^ in the pl ao o of ^
, and . y. that is , ' * *
(u
392] AND TOUCH TWO GIVEN LINES.
that is
 4/3y&' (a s
47
 (a 3 + 4/37/c') ( 2/Vc'a  a?;*) 2  40y F 1 (OHO 4
or, what is the same thing,
_ 43# (> + W) (&V + 2/2) = {20tf oj + 2/3 a /,/y  (2/37/0' + a 3 ) ^} a
 (a 2 + WO (2^'<w  ^) 3  *^^ C
which, when tw + /5i/ + 7^ = is the equation of the HUG infinity, puts in evidence the
asymptotes of the conic #" + # 0.
Now writing X, 7, Q in the place of en, y, z\ tt = l, and == (1 + V (")). = 2 >
y={V(ft) V()l a we navo
 16 [{1 + V (w)) a + 8 (V (0  V ()i a ] (7Q + ^)
= [4 (i+ v ()) ^ + sy  (4 iv w  v ()) + (i + v (flw
 [(1 + V (m)" + 8 [V ()  V ()] s ] [4^  (1 + V <*)} Q] a
 16 [{1 + V <ro)l ^ + 27 + [V (0  V ()} 3 Q] 3 .
and the asymptotes are
= + V(l + V (WMI)}" + 8 IV C" 1 )  V ()} a [4A r  {1 1 V (mn)} Q}.
At the centre
4 {1 + V ()} A r + 8 F  [4 (V (w) ~ V ()} 2 + {1 + V <)} 9 ] Q = 0,
4Z{l + V(m)}Q ! =0.
but the first equation is
{1 + V (wn)) [4X  Q {I + V (*n){] + 8 F  4 (V (0  V (n)) 9 Q = 0,
so that we have
4A'  {1 + V (flw)J Q, 2 7 = {V (0  V ())' Q.
the first of these is
2 {V (TO)  V ())' (  ff)  {1 + V (ift)l* (*  y)  (1  ) * = >
and the two together give
2X {V (m)  V ()!'  {1 + V (W3i)] F = 0,
so that we have
2 {V (m)  V (*)!" (  ^ )  [i + V ()$ {^  ff V ())  o,
to determine the coordinates of the centre.
48 OX THE CONICS WHICH PASS THROUGH TWO GIVEN POINTS
The equation of the chord of contact is
} I 70 = 0,
which for 7=1 is parallel to y = and for y = V(0 is parallel to <y=0. Jlut
coordinates of the centre are
a : y : z =  7 + V (mn) ;  7 + 1 : ^ ( wm ) + l + _. (^ ( m ) _ ^ ( n )}^
which for 7 = 1 give
V(w) H 1 + i{V (*)  V())'i 2 + 2 V * : 2im
and for 7 = V (ni) give
a = 0,
if : *lV(w) : V
y (tttu
the cent,, !, J
+ y V (m) + 7'^ = 0,
where, writing for *, y , , the coordinates of the ^^ w(j
+ 2 v () + 4 y (m) + 1 + {V (m) _ ^ (B)} ,J = Oj
(mn)
7(1 +
that is
or, what is the same thing,
77 =
and consequently 7' = 7 only for 7==0 ,
TJ.
the fixed point Lx /^ \ ft con ' es P on{lll ig positions of the chnvrl nf o rtll *. rt . n
tL _ ^u pome (a + i/V(m) = 0, *=,( i * ^  cl ot con(; acli thmuwh
see fig. 2 in the
392] AND TOUCH TWO GIVEN LINES, 49
Chord of Contact. Centre,
01\. P a , at infinity on hyperbola,
OL(z = Q). L, (z = Q, a,'i/ = 0).
OH. 0, the lino joining this with being always
behind 00.
OP i. P., at infinity on hyperbola,
X {' + y V (mi') = 0] .  ' X (m = 0, y = 0).
OG (parallel to y = 0). G (on line # = 0).
077 (purallol to ^ = 0) and so back to H (on line ffi=0) and so on to
I have treated separately the case */(mn) = I.
Consider tho conies which touch the lines y a 1 = O f yM = and pass through
the points
Tho aquation is of tho form
f  a? + Je(t *? = $>
and to determine ft, AVG havo
c a
1 o" 1 + /c (1 )' J = 0, and therefore k ^  r a .
(.* ~ a )
Tho (Muixtion thiiB bocinnes
(1  ) 3 (f  ffiS ) + c a (  ) = >
thiiti in
(1  a) 3 y a + (c a  (1  a) 3 } ffi 9  2c 3 aa; + c a a 9 = 0,
or IIH this may bo written
{r&fi 1 a /i3/yS C\ _ \a
'^^f^o
Honoe tho nature of the conic depends on the sign of c G (la) a , viz. if this b&
poflilivo, or between the limits l + o, lc, the curve is an ellipse,
(Bcoordinate of centre ;
c (i a)
which is positive,
o (1  a)
a'semiaxis =
ysemiaxis
Tho coordinate of centre for a = 1 + c is = + oo (the curve being in this case a pai
P!> and for 1  o it is also = + oo (the curve being in this case a parabola P 3 ).
coordinate has a minimum value corresponding to a=V (1 o)> viz  fehis is ~i I 1 +
C. VI.
50 ON THE CONICS WHICH PASS THROUGH TWO GIVEN POINTS, &0.  ,'JU'J
Hence as () passes from 1 + c to V(l~ c3 ). the coordiimfce of tho eontn; pus^n
from x to its minimum value {1 +*/(! c a )j; iu the passage wo hnvo a= rrivJiiff
the coordinate =1, the conic being in this case a pair of coincident linos (; 1)' J =. ; O.
And as (2) passes from the foregoing value Vflc 2 ) to 1  o, tho (inordinate of (In 1
centre passes from the minimum value ^{1 + ^(1 c a )} to GO.
The curve is a hyperbola if a lies without the limits 1 + c, I  c,
.'(.'coordinate of centre  ZJ^. (
which has the sign of a,
0semiaxis = ~~ ..L...T /
ysemiaxis =
semiaperture of asymptotes =
' [ " A/  (i );
^l^r 1C ^.= 0. (Parabola), but inereases as l illoroiu(oa
coordinate of centre is =
lc, coordinate of centre
s
u ' " >, 0,
tho hyperbola being in this case the pair of lines * *,'
a re
.. . c')}, and then as a !>/ ,7 a mnximum l"itiv viiliu.
, i. . "'ii'H us a passes irom //i _ ^ <. ..
centre diminishes from HlV(l<fll n T ~ C I cooiflinatp i>l
negative, the Hnes y ^^ = n ^f.^,1/,.,, , i8 to bo Barker! that a 1
51'
393.
ON THE CONICS WHICH TOUCH THREE GIVEN LINES AND
PASS THROUGH A GIVEN POINT.
[From the Quarterly Journal of Pure and Applied Mathematics, vol. VIIL (1807),
pp. 220222.]
CONSIDER the triangles which touch three given lines; the three lines form a
triangle, and the lines joining the angles of the triangle with the points of contact of
the opposite sides respectively meet in a point S: conversely given the three lines and
the point S t then joining this point with the angles of the triangle the joining lines
meet the opposite sides respectively in throe points which are the points of contact
with the throe given lines respectively of a conic ; snch conic is determinate and unique.
Suppose now that the conic passes through a given point; the point 8 is no longer
arbitrary, but it must lie on a certain curve ; and this curve being known, then taking
upon it any point whatever for the point S, and constructing as before the conic
which corresponds to such point, the conic in question will pass through the given
point, and will thus be a conic touching the three given lines and passing through
the given point. And the series of such conks corresponds of course to the
points on the curve.
52 OX THE CONICS WHICH TOUCH THEEE GIVEN LINES, &O. [""SO 3
the condition in order that the conic may pass through the given point is a J & + G == O>
and we thus find for the curve, which is the locus of the point 8, the equation
{(*)
or, what is the same tiling,
the rationalised form of which is
= 0.
This is a quartic curve with three cusps, viz. each angle of the triangle in a
and by considering for example the cusp (?/ = 0, # = 0) and writing the equation uiitlor
the form
o? (y  zf 2a (yz* + y z z) + if& = 0,
we see that the tangent at the cusp in question is the lino s/,3=0; that is, fclio
tangents ^at the three cusps are the Hues joining these points respectively with tins
given point (1, 1, 1). Each cuspidal tangent meets the curve in the cusp counting 1 IXH
three points and in a fourth point of intersection, the coordinates whereof in tho Oiuscs
of the ^ tangent y* = 0, are at once found to be a : y : z^ 1 : 4 : 4, or say thia IH
the point (1, 4, 4); the point on the tangent xto = Q is of course (4, 1, 4), and thai*
on the tangent *y0 is (4, 4, 1). To find the tangents at these points reHpoctivoly,
I remark that the general equation of the tangent is
that is
X . Y . Z
or for the point (1, 4, 4) the equation of the tangent is 8Z+r+^0, or
8 + y + ^0; that is the tangent passes through the point , = 0, or+y+^O, boi
the pomt of mtersection of the line , with the line . + + ;. ^hioh IB t
e ne , wt the line . + y+ . 0 hioh IB tli
a 4 Sr/fT^'JV' 1} " r ^' d fc the U ^ *
o tl haimoni ' it ' ^ 1' 'n } T^^ ^ to)I h th ^ ints
ot the harmonic lino a + y + g = Q mt h the three given lines respectively,
of
* " "'
r
without much difficulty. ' w Sles P 10 *^ may be effected
Plate I.
$5 face p. 52.
394.
ON A. LOCUS IN RELATION TO THE TRIANGLE.
[From the Quarterly Journal of Pure and Applied Mathematics, vol. viir, (1867),
pp. 204277.]
IF from any point of a circle circumscribed about a triangle perpendiculars are
kit Tall upon the aides, tho feet of tho perpendiculars lie in a line; or, what is the
Htimo thing, tho IOCUH of a point, such that the perpendiculars let fall therefrom upon
tho Hides of a given trianglo havo their feet in a. line, is the circle circumscribed
about iiho triangle.
In this well known theorem we may of course replace the circular points at
infinity by any two points whatever; or the Absolute being a pointpair, and the
terms perpendicular and circle being understood accordingly, we have the more general
theorem expressed in tho same words,
But it is loss easy to sec what the corresponding theorem is, when instead of
being a pointpair, the Absolute is a proper conic; and the discussion of the question
affords some interesting results.
Take (so = 0, / = 0, s = 0) for the equations of the sides of the triangle, and let
tho equation of tho Absolute be
(a, &, o, /, g, hfco, y, *0 2 = 0,
then any two linos which are harmonics in regard to this conic (or, what is the
same thing, which are such that the one of them passes through the pole of the
other) are said to be perpendicular to each other, and the question is:
Find the looiiB of a point, such that the perpendiculars let fall therefrom on the
sides of the trianglo have their feet in a line.
Suppoain* as usual, that the inverse coefficients are (A, B, 0,1, &, 5),
K IB the discriminant, the coordinates of the poles sides respee
54
ON A LOCUS IN RELATION TO THE TBIANGLK
[394
(A, H, G), (H, B, F), (G, F, 0}. Hence considering a point P, the coordinates of which
are (x, y t z), and taking (X t Y, Z) for current coordinates, the equation of the perpen
dicular from P on the side X = is
X, 7, Z = 0,
A, H, G
and writing in this equation A'=0, wo find
( , AyL
for the coordinates of the foot of the' perpendicular. For the other perpendicular
respectively, the coordinates are
and
(BxHy, , BzFy},
(CxGs, OyFz , o ),
and hence the condition in order that the three feet may lie in a lino is
or, what is the same thing,
, AyEss, Az~
CsaGz, CyFz,
= 0;
that is
( ^  OH) & + 5
f
the,ofore a cubic.
and /f' for the discriminant
or as th l8 may also be written
, the equation i s
^
394] ON A LOCUS IN RELATION TO THE TRIANGLE. 55
that is
r 2 f A nn nTT\ A B C'l /A B G \fa y z\
vffrftffi (ABO  FGII)  rv a  M ~ rrT a *F+ ( Jf lit + 7v ^ + W ^) ( B? + O? + /// )
and the cubic will therefore break up into a Hue and conic if only
2 ABO
TFG'H.' ( ~~ ' ~ j^a ~ (yi ~ p3 ~ '
and it is easy to see that conversely this is the necessary and sufficient condition in
order that the cubic may so break up.
The condition is '
n = zFQ'ii 1 (ABO Fan)  ACPIP  BHW  ow a = o,
we have
A A' + ]&' + GO' = 3vl56'  AF*  5G a  C// a , = 7f + 2 (ABO
and thence
a j'<?7/' (^^' + J5J3'
that is
tt = A G'H' (Q'ff  A'F)
so that the condition il is satisfied if If = 0, that is if the equation
(A,B, 0,F, ff.SJfc fl.0^0,
which is tho lineequation of the Absolute breaks up into factors; that is,, if the
Absolute be a pointpair.
In the case in question we may write
(A, B, G, ff, G,
that is  . ,
(A, B, 0, F, G, J/) = (2aa', 2/3/3', 277', 0</ + 0<y>
whence also, putting for shortness,
(/  /3V> 7'  y' a > a $' " a '^^ = ( x> ^ v ^'
we have
(A 1 , B', 0', F', G', #') = (^> /*"' ^
and also
JTf = 0, 2 (X^a  FGH) = AA' + BB'+ CO', =  2
50 ON A LOCUS IN RELATION TO THE TRIANGLK
The original cubic equation is
J? + 77V) ay& + eufayz (py + vz} + pfipza (\cv \ vg) 4 77'^ (Xft. 1 +
mid this in fact is
(aofayz + pfi'ttxte + w'vwy) (X# + p,y I MS) = 0,
The equation \x\ py +vz = ft is that of the lino through tho two pointn which
constitute the Absolute; the other factor gives
s
= 0,
which is the equation of a conic through the angles of the trianglo (to = 0, y = (), z =s ()),
and which also passes through the two points of the Absolute; in fact, writing' (, /3, 7)
for (a;, y, z) the equation becomes a/3y ('X + ftp + y v ) = 0, and so alw> writing (', /3', y')
for (.B, y t 2) it becomes a'/9V (oX + ^ + 7^) = 0, which relations nro identically maiw fusel
by the values of (X, /*, v). Hence we see that the Absolute being a pointpair, tlm
locus is the conic passing through the angles of the triangle, and tho two points of
the Absolute; that is, it is the circle passing through the angles of tho triimglo.
But assuming that 1C is not =0, or that the Absolute in a proper conic, thu
equation fi = will be satisfied if
AFG'ff + BQH'F + CHFG' + PG'H'  0,
we have F, ff,H' = Kf,Kg, Kh respectively, or omitting the factor K\ the equation
becomes
AF<th + BGkf+GHfg + Kfgh = Q,
which is
/VA 3  bctfh*  coA 8 / 8 4 a6/ a (/ a + 2oio/^A = 0,
or, as it may also be written,
tho l r e "^^i anbstitutiug also for ^, ^
the values AT/ % /a, the equation of the cubic curve is
2 (a&c /jA) X yz + Ay, (hy +
and the transformed form is
we have
so that the foregoing condition
_
a/*
394]
ON A LOCUS IN" RELATION TO THE TRIANGLE.
0* *?/ P
being satisfied, the cubic breaks up into the line 7.+  + 7 = 0, and the conic
/ 9 h
'
A
G
It is to be remarked that in general a triangle and bhe reciprocal triangle are
in perspective; that is, the lines joining corresponding angles meet in a point, and
the points of intersections of opposite aides lie in a line ; this is the case therefore
with bhe triangle (# = 0, y = Q, 2 = 0), and the reciprocal triangle
(uai i hy f gz = 0, ha> + 6y +fz = 0, gio +fy + cz = 0) ;
and it is easy to see that the line through the points of intersection of corresponding
sides is in fact the above mentioned line ?4^ + ?0. It is to be noticed also that
f y h
the coordinates of the point of intersection of the lines joining the corresponding
angles are {F, G, H). The conic
9
is of course a conic passing through the angles of the triangle {& = (), y 0, 2 = 0);
it is not, what it might have boon expected to be, a conic having double contact with
the Absolute (a, b, o, /, y, Ajw, y, zf.
I return to the condition
=0
'
abo c(/' fl by 1 * c/t 3 fyli
this can be shown to be the condition in order that the sides of the triangle
( 0, y = 0, z = 0), and the sides of the reciprocal triangle (eta; + % + 0*  0, /we + fy +/* = 0,
r/ft> +/?/ H c# = 0) touch one and the same, conic; in fact, using line coordinates, the
coordinates of the first three sides are (1, 0, 0), (0, 1, 0), (0, 0, 1) respectively, and
those of the second three sides are (a, h, g\ (h, b, /), ({/, f> o) respectively ; the equation
of a conic touching the first three lines is
^ + + 0.
and hence making the conic touch the second three sides, we have three linear
equations from which eliminating L, M, N, we find
which is the equation in question.
0. VI.
58
ON A LOCUS IN RELATION TO THE TRIANGLE.
= 0,
We know that if the sides of two triangles touch one and the same conic fclmir
angles must lie in and on the same conic. The coordinates of the angles are (1 ' O ())
(0, 1, 0), (0, 0, ]) and (A, ff, G), (H, B, F], (ff, ff t G) respectively, and the anglon 'will
be situate in a conic if only
111
A' IP
111
H' B' F
1 1 1
G' F' G
an equation which must be equivalent to the last preceding one; this is easily voriliiil
In fact, writing for shortness
V =
we have
1
1
1
1
I
I
a'
A'
ff
'
A'
H'
G
I
1
1
I
I
I
h>
b'
7
H'
B'
F
1
1
i
I
1
I
ff*
/'
G'
ff'
G
S.
and the second factor is
But
 aGH (AF + Iff) + A.FKBG + AFgOff,
= AF(a6ffI
so that the second factor is
which is
" ^ (A* 
+
~  L _ JL 1 , 2
o . "
394] ON A LOCUS IN RELATION TO THE TRIANGLE.
so that we have identically
 ABCffWH* Q = Kalcf^h V ,
and the conditions V =0, D=0 are consequently equivalent.
The condition
L L_JL_1 + A =0 "
abc a/ 3 If/" ch* fgh '
is the condition in order that the function
may break up into linear factors; the function in question is
/ bo ca ab
(<^. c '7 7 j
whichiH
i A B , 0\
so that the condition is, that the conic
(a, o, c, /, </, /iji. ft; 1/> z) "T \^j:J g h /
(which is a certain conic passing through the intonation, of the Absolute
A n Is ~* i it t . 
(, fc o t /, g t Kt*. y, *0, and of the locus conic ^++^0) 1 be a
pair of lines. Writing the equation ef the eenic in question under the form
6c
the inverse eoefficients A' t B', G', F, ff, ff of this conic, are
Ale Boa Gab _^jr  a G
""
i, B"  r<  trF ' G : S. Hence, if in regard w i.uia u  
so that wo have M .(r "' ^ = ), and join the corresponding
form the reciprocal ef the tmngte (.Oj 0,j J, J ^ ^ ^
angles of ^e t,o t *. ^^^^ fa recipvocal in reg ,d
It is to be noticed that the conic
60
N A LOCQS IN RELATION TO THE TRIANGLE. [394
contains the angles of the reciprocal triangle, and is thus in fact the conic in winch
are situate the angles of the two triangles. For the coordinates of one of the aiielon
of the reciprocal triangle are (A, H, (?); we should therefore have
which is
or attending only to the second factor and writing
GH=Kf+AW )
the condition is
Kfgh + APgh + SGhf+ GSfg = 0,
or substituting for If, A, B, G t F, a, H their values and reducing, this is
~__4.__
aio a/" Ziy" c/i' fgh~
t t if the trian f b i suoh that the re >
tuaugle he m a oomo (or, what j. the same thing, if the sides touch a
conic) then the cubic locus breaks up into the line  + 8+^0, which is the linn
inTtt otic* "' 3 f intei ' Se ti0n f the ^% des of the two triangles, a,ul
S

which is the conie through the angles of the two triangles.
its alrie^rrfTh MiaeS ; g! T , a C0ni (the Abs lute ) to constraot
' S
in a cic
in re Md to the conic, Ho
394] ON A LOCUS IN RELATION TO THE TRIANGLE. Gl
I suppose that two of the angles of the triangle are given, and I enquire into
the locus of the remaining angle. To fix the ideas, let A, B, G he the angles of the
triangle, A' t B', C' those of the reciprocal triangle; and let the angles A and E be
given, We have to find the locus of the point G': I observe however, that the lines
AA', BB', GG' meet in a point 0, and I conduct the investigation in such manner
as to obtain simultaneously the loci of the two points and 0. The lines C'B', G'A'
are the polars of A, B respectively, let their equations he 3 = 0, and i/=0, and let the
equation of the line AB be 2 = 0; this being so, the equation of the given conic will
be of the form
(a, b, c, 0, 0, fc
I take (a, (3, 7) for the coordinates of and (a?, y, 2) for those of G'; the
coordinates of either of these points being of course deducible from those of the other.
Observing that the inverse coefficients are
(6c, co, ab h 2 , 0, 0, oA),
we find
coordinates of A are ( b,  h, 0),
B (h, a, 0).
The points A' and B' are then given as the intersections of AO with 6"^'(y=0) and
of BO with C'S'Cw^ )') we find
coordinates of A' are (Aa + && , fcy),
5' ( , act + Aft A?),
Moreover, coordinates of 0' are (0, 0, 1),
C^ { w it %\
<J W !/ z )'
The six looints A, B, G, A', ff, G' are to lie in a conic; the equations of the
MA M AT* AY 4 670 aX + hY=Q, Z=Q, and hence the equation of a
lines G.O., dx*, Aj> ^ ie /tArw _ '
conic passing through the ,points G', A, B is
L M ft^Q
ST+Af + WC + 6F ^
Hence, making the oonio pass through the remaining points A', * 0. we find
62
ON A LOCUS IN RELATION TO THE TRIANGLE.
[304
and eliminating the L, M, N, we find
1
a
1
h '
A + 6/9
1
1
A
b '
ct 4 A/9
1
1
Ay
aa + h
y* hie + by '
z
= 0,
or developing and reducing, this is
h cwB
A Aa + 6y
1 a + A/9 _ 1 Aa + & _
As? + 6 ~ '
We have still to find the relation between (a, ft 7 ) and (a;, y, *); this is obtained
by the consideration that the line A'ff, through the two points A', 7/ tho coordinntOH
of which are known in terms of (a, ft 7 ), is the polar of the point 0, tho coordinates
of which are fa y, z). The equation of A'B' is thus obtained in the two forms
and
+ (Aft! +
and comparing these, we have
: y : z = a.
or what is the same thing
chy
(ate + %) (Jig + by)
eliminate
where
_
/m& ^ A a
or, completing the elimination,
wtioh is a quartic curve having a node at each of the oi
points
394] ON A LOCUS IN RELATION TO THE TRIANGLE. 63
that is, at each of the points B t A, 6'. The righthand side of the foregoing
equation, is
=  (ab  7i 3 ) (ha, aft,
so that the equation may also be written
(am + hyf (hx + %) a + eAV f oo; a + i/ + f xy) = 0,
\ "'I
Secondly, to eliminate the (as, y, z) t we have
obh*Y 2 1 w
_
h ah<t + l3 b C '
where
7_ _ chf
or, completing the elimination,
(a&  A 3 ) <% 3 = (A6,  06, Aa$aa + A/3, Aa +
 (06  /i a ) A f ao? + 6^8" + ^ *P
\ ii
that is
fa, 6, o,0, 0,^/3,7)^0.
\ ft A /
Writing (a 1 , y, *) In place of (a, /9, 7), the locus of the point is the conic
Ub,o t 0, 0, f]Uy, *Wo,
\ /t A /
which is a conic intersecting the Absolute
(a, 6, o t 0, 0, AJ, T/, ^) 3 = 0,
ali its intersections with the lines ffi=0, i/ = 0, that is the lines C'B' and (TA 1 .
In regard to this new conic, the coordinates of the pole of C'B'(v = Q) are at
once found to be (~h, a, 0), that is, the pole of G'B' is B ; and similarly the coordi
nates of the pole of Cf'A'fo0) are (b, A, 0), that is, the pole of ^ IB A. Wo
may consequently construct the conic the locus of 0, via given the Absolute and the
points A and B, we have O'A' the polar of , meeting the Absolute m two points
L, th\ and O'ff the polar of 4 meeting the Absolute m the pomts (b, and b,) the
lines O'A' and 0'*' meet in G'. This being so, the required come puses through the
points ou .. fti. &, the tangents at these points being Ah. **,& Bb> JP^^
eight conditions, five of which would bo sufficient to determine the conic It * to ta
remarked that the lines 0'S t O'A! (which in regard to the Absolute are h epotai of
A, B respectively) are in regard to the required come the polars of B, A respectively.
The conie the locus of being known, the point may be taken at any point
of thie conic, and then we have A' as the intersection of O'A w.th AO B as h,
interaeetion of O'S? with BQ> and finally, G as the pole of the line AB in legaid
64 ON A LOCUS IN RELATION TO THE TBI ANGLE. [[394
to the Absolute, the point so obtained being a point on the line G'O, To each position
of on the conic locus, there corresponds of course a position of G\ the locus of
is, as has been shown, a quartic curve having a node at each of the points G' t A, 7?.
The foregoing conclusions apply of course to spherical figures ; we see therefore that
on the sphere the locus of a point such that the perpendiculars let fall on the aides of a
given spherical triangle have their feet in a line (great circle), is a spherical cubic. If,
however, the spherical triangle is such that the angles thereof and the poles of the sides
(or, what is the same thing, the angles of the polar triangle) lie on a spherical conic ;
then the cubic locus breaks up into a line (great circle), which is in fact the circle
having for its pole the point of intersection of the perpendiculars from the angles of tho
triangle on the opposite sides respectively, and into the beforementioned spherical conk.
Assuming that the angles A and B are given, the abovementioned construction, by
means of the point 0, is applicable to the determination of tho locus of the remaining
angle G, in order that the spherical triangle ABO may bo such that the angles and
the poles of the sides lie on the same spherical conic, but this requires somo further
developments. The lines Q'E>, G'A' which are the polars of the given angles ^ J1
respectively, are the cyclic arcs of the conic the locus of 0, or say for shortness tho
come 0 and moreover these same lines O'ff, G'A' are in regard to the conio O,
the polars of the angles S, A respectively. If instead of the conic we consider tho
polar conic it follows that A, B are the foci, and G'A', O'ff the oompondimr
duectrices of tiie come 0'. The distance of the directrix G'A' from the centre of tho
conic measuring such distance along the transverse axis is clearly  00"  distance of
' '
hn
boft'tfn T Vei ;r T 1S =9 ; that i8 ' the C0nic & is ft co described
0* l! ' , " aUSV r ^ (01> SUm r dUteMM of *" fci diBtomooH)
90. Considering any tangent whatever of this conic, the pole of the tangent is n
^ ^ ^ f interSeCti011 " *' P^onS lo
ta tiofl  ' P^on
from the angles of the spherical triangle on the opposite sides; hence to
(viz.
i , ~ " "" v jjuino. luiiujyitj ne in a snhor
mly to construct as before the conic 0' with th f A H "
SArrjs t j^^^i'itrr*!* ^
f the conic 0'. It i s moreovpr nW ^\ opposite sides respectively, is a taragoiit
 10 uiuieuvei Cieai, tnat ffivon n fnionnln ^ nn i ,1
i question, if with the foci AS d niangie^^^a having the property
7 in like manner with the foci A *P nn^r^ 88 aX1S ==9 W desci ' ibe a conic, ctnd
?, 0' and the same transverse axis\ve 1 h T tr ^ SVOrse axis ' aud with th foci
onics will have a common tan^nt the Tol! wL P Tu^ f fln that th threb
action of the perpendiculars fmm +i, i * wneieoi will be the point of infcer
B spectively, P IpentllCUlaiS flom the "^ of the triangle AEG on the opposite siclo*
395.
INVESTIGATIONS IN CONNEXION WITH CASEY'S EQUATION.
[From fcho Quarterly Journal of Pure and Applied Mathematics, vol. vm. (1867),
pp. 384341.]
IN 1,1 paper road April 9, I860, and recently published in the Proceedings of the
Royal Irish Aoad&niy, Mr Casey has given in a very elegant form the equation of a
pair of oh'oloH touching each of throe given circles, viz. if U = Q, F = 0, TP = be
bho equations of the three given circles respoctively, and if considering the common
(iangonlM of (F = 0, lf=0), of (1F=0, U=Q), and of (ZJ^O, 7=0) respectively, these
oommoH tangents being Huoh that tho centres of similitude through which they
roHpuotivoly posH Ho iu a lino (vis;, the tangents are a.11 three direct, or one is direct
and the other two arc inverse), then if /, g, h are the lengths of tlio tangents in
intention, bho equation
belongs to ti pair of circles, each of thorn touching the three given circles, (There
mil, it IH olenr, four combinations of tangents, and the theorem gives therefore "
aquations of four pairs of circles, that is of the eight circles which touch the
givun circles.)
Generally, if Z70, F=0, WQ are the equations of any three curves of the same
order n t and if/, g, h arc arbitrary coefficients, then the equation
is bhal; of a curve of the order 2n, touching each of the curves 7=0, F=0, F=0,
7i q timos, viz. it touches
= 0, at its n* intei'soctions with gV W=0,
0, VI.
66 INVESTIGATIONS IN CONNEXION WITH CASEY'S EQUATION.
If however the curves U=Q, V=Q, W=0 have n common intersection, then tho
curve in question has a node at this point, and besides touches each of the throe
curves in n"l points; and similarly, if the curves U = Q, F=0, W=0 have k common
intersections, then the curve in question has a node at each of these points, and bosidos
touches each of the three curves in ?i 3 k points.
In particular, if tf~0, V=0, W = are conies having two common intorseotioiiH,
then the curve is a quartic having a node at each of the common intersections, ami
besides touching each of the given conies in two points; whence, if tho coefficients
f, ij, h (that is, their ratios) are so determined that the quartic may have tnvo mtmt
nodes, then the quartic, having in all four nodes, will break up into a pair of conicw,
each passing through the common intersections, and the pair touching each of thn
given conies in two points; that is, the component conies will each of thorn touch
each of the given conies once. Taking the circular points at infinity for the coinmon
intersections, the conies will be circles, and we thus see that Casey's theorem i in
effect a determination of the coefficients / g, k, in such wise that the curve
(which when U=Q, 7=0, F = are circles, is by what precedes a bicirculnr queutin)
shall have two more nodes, and so break up into a pair of circles.
The question arises, given ZT0, F0, F = 0, curves of the same order n, it in
required to determine the ratios / : g : h in such wise that the curve
may have two nodes; or we may simply inquire as to the number of the sots of
values of (/:</: A), which give a binodal curve, V (fU) \ V (gV) + V (h'W) = 0.
^ I had heard of Mr Casey's theorem from Dr Salmon, and communicated it toffotlwi
v u o !lT" lg considerati M to Prof  Cremona, who, in a letter dated Bolocrim.
March 3, I860, sent me an elegant solution of the question as to the number of tho
binodal curves. This solution is in effect as follows :
^n Um *' F== ' lf = of the samo ***** ; oonaidor the
pom ( g, A) and corresponding thereto the curve fU+gV+hWV. As long as tho
rl^l' f\? 7> ^ curve /^+^+^0,will not have any node, an ,l
in order that this curve may have a node, it is necessary that the point. (/ V A)
shall he on a cortam curve S; this being so, the node will lie on a curvo > tho
Jacobs o the curves U, 7, W ; and the curves J and S W ill correnp d to
urve * , , a llode at sorae one P^nt on the
cuiye J. and conveisely, m order that the curve fU+ ff V+hW=0 may bo a curvn
havmg a node at a given point on the curve J t it JLJ that the ^n (/> 7 I)
shall be at some one point of the mirvn 5 1 TK^ / i T l V **' '
=
a binodal curve having a
395] INVESTIGATIONS IN CONNEXION WITH CASEY'S EQUATION. 07
each of the corresponding points on J ; and each cusp of S corresponds to two coincident
points of /, viz. the point (/ g, h) being at a cusp of 2, the curve fU + yV+hW = Q
is a cuspidal curve having a cusp at the corresponding point of J. The number of
tho biiiodal curves fU+gV+hW=Q is thus equal to the number of the nodes of 2,
and tho number of the cuspidal curves fU + gV +h 17 = is equal to the number uf
bho cuHpa of 2. The curve 2 is easily shown to be a curve of the order 3(l) 3
nncl class 3n(l); and qua curve which corresponds point to point with J, it is a
owrvo having the same deficiency as J t that is a deficiency = i(3ft4)(3 r >); we
liavo tlionco tho PHtokeriau numbers of the curve 2, viz. :
Order is = 3(ttl) 9 ,
Class = 3n(nl),
Cusps 12(ttl)( n2),
Nodes = a(nl)(w2)(3'Snll),
Inflexions = 3(nl)(4n5),
Double tangents = f (!)( n2)(3' + 38).
!
also
( W , it ir lg thonin n + l for )
p
, and
th
Ping now to the pvoposod question
of tho same ordor i and we may confer
the curve VCW + VW + V W = <>.
of the orfer 2, having rf contacts wit]
na the point (/, g, K) i
o ir o
'
' Hch ; 3 a cuvvo
g iven curves ff, F, If. As leng
g ^ ^ orfer ^
he point (f. . shall lie on a
68 INVESTIGATIONS IN CONNEXION WITH CASEY'S EQUATION. [395
point of /. The number of the binodal curves fi is consequently equal to that of
the nodes of A, and the number of the cuspidal curves fl is equal to that of the
cusps of A ; we have consequently to find the Pliickerian numbers of fcho curve A ;
and this Prof. Cremona accomplishes by bringing it into connexion with the foregoing
curve S, and making the determination depend upon that of the number of the conies
which satisfy certain conditions of contact in regard to the curve 2.
Consider, as corresponding to any given point (/, g, h) whatever, tho coni
r + + 2~ which paases blll ' ou gh tlll ' ee ftxecl points, the angles of tho triangle
* J J
A > = 0, y = Q, z = Q. For points (/ g, h) which lie in an arbitrary Hue Af+J3g+ Oh = 0,
the corresponding conies pass through the fourth fixed point & : y : z = A : Ji : (J.
Assume for the moment that to the points (f t g, A) which lie on the foregoing ourvu
A, correspond conies which touch the foregoing curve Then 1". to the points of
intersection of the curve A with an arbitrary line, correspond the conies winch piuw
through four arbitrary points and touch the curve S; or the order of the curve A
is equal to the number of the conies which can be drawn through four arbitrary
points to touch the curve 2; viz. if m be the order, n the class of 2, tho number
of these conies is =2i + , or substituting for m, n the values 3(nl) 3 and 3(wl)
respectively, the number of these conies, that is the order of A, is = 8 (wl)(3 2).
2. To the nodes of A correspond the conies which pass through three arbitrary points
and have two contacts with 2, viz. if m be the order, n the class, and # tho mimbor
of cusps of S, then the number of these conies is =*(2wt + )' 2m 5nfye t or
substituting for m, n their values as above, and for K its value = 12 (71 t)(w 2)
the number of these conies, that is, the number of the nodes of A, is found to bu
= f (n  l)(27?i 9  63tt + 23n + 16).
3". To the cusps of A correspond the conies which pass through throe arbitrary
points, and_lmve with 2 a contact of the second order; the numbei of these (m,
as above) a B + ^ or substituting for and K their values as above, the number
of these comes that is the number of the cusps of A, is  8 (  1) (7 B  8) Wo
have thence all the Pluckerian numbers of the curve A, viz. these arc
Order 3(rcl)( 82) f
Class = "
Nodes  f (n  1) (2to  63 ? i 3 + 22u + 16),
Cusps = 8(wl)( 7w8X
Double tangents =  ( n  1) (18ft. _ 36?i a + lQn +
Inflexions =12(nl)( n _2)
: d f vmr s r^rr^ ^i
numbera include the result that the number of the binodaf curie!
s
16),
395]
INVESTIGATIONS IN CONNEXION WITH CASEY S EQUATION.
69
The proof depended on the assumption, that to the points (f, g 3 h) which He on
/ )
the curve A, correspond the conies  +  +  = which touch the curve S ; this
so y z
M, Cremona proves in a very simple manner : the points of / correspond each to each
with the points of S, or if we please they correspond each to each with the tangents of
S. To the Cm (n  1) intersections of J with any curve a (viz. V (fU) + V" (gV) + V (hW} =0)
j* i
correspond the Gn(n 1) common tangents of S and the conic *~ + + = ; if fl has
x y z
a node, two of the Gn (n 1) intersections coincide, and the corresponding two tangents
will also coincide, that is Jl having a node (or the point (/, g, h) being on the
curvo A), the conic touches the curve 3J. But it is not uninteresting to give an
independent analytical proof, Write for shortness
dU = Ada) + Bdy + Cdz,
dV^A'dto + B'dy +G'dz,
and lob (to, y, z) bo the coordinates of a point on J, (X, F, %} those of the corre
sponding point on 2, (/, ff> It) those of the corresponding point on A, Write also
for shortness
tiG'B'G, QA'C'A, AffA'B^P ; Q : R t
thon wo have
AX + BY +CZ 0,
A'X+B'Y
giving
giving
A , B ,
A 1 ,
G'
=
u )
A' +ff +G' =0.
^1" B +B" +0" =0,
= 0, which is in fact the equation of the curve /; and moreover
A", J3", Of
, Y : Z~P : Q : jK, to determine the point (A r , 7, ) on S; and
h
U '
vr
or what is the same thing, f : ff : h = P*U : Q*V : JZ'Tf, to determine the point
(/> </> A) on A  Treating now (/ (/, h) as constants, and (A r , F, ) as current coordinates,
the conic ^ + 4 + 0, will touch the curve S at the point (P, Q, B), if only the
X y &
70 INVESTIGATIONS IN CONNEXION WITH CASEY'S EQUATION,
equation of the conic is satisfied by these values and by the consecutive values
P + dP, Q + dQ, R + dR; or what is the smne thing, if we have
L + + A =0
P * Q * R '
that is
:4:4= Q(iB .BdQ : B&P  PdR : PdQ  QdP.
If the functions on the righthand side are as U : V : W, then these equations give
f'.g:h = P*Ui V : R*W t
that is (/, g, h) will be a point on the curve A. It is therefore only necessary to
show .that in virtue of the equation /=0 of the curve J, and of the derived
equation d/=0, we have
QdRMQ : MPPdR : PdQQdP=U : 7 : F.
Take for instance the equation
V(QdR  EdQ)  U(WP  PdE) = 0,
that is
dR (UP+VQ+ WR)  J ( raP + 7cZQ I WAR)  0,
and this, and the other two equations will be satisfied if only UP 4 VQ + WM = 0,
WdR=Q', we have, neglecting a numerical factor,
W=Ck +&$+&'*,
whence, attending to the values of P, Q t R, we have
hence also
so that
if only
:11 U, iNvrMTi.mHiNM IN n.NNKxmx WITH CAH.UY'H EQUATION
UIN . ~ j
lltll  ..ulnlifuiiiiK I'"' /'> V. ''.'"'. ./I', r/H' ihoir vain,*, fch lofthand side is _ri
whii'li ^ "'. hii" "^ I" 11 ' m 'l l " ls "' iu(i IIIOVIM!, mid C / A\ ' ~'
(lll ,,, rnrv, A. V/ ' ^ A) 1S a l Kjl!lt;
l. IM >., In* .M.M..I. ilmi Hi.. ln ,nrvr, >;. A arn K() .> mo tricnlly connected through
,!, I),,,,,, m 1iih,iiy II.IHI'. in lnll..wi.: vi/.. hikni); HH nxim Iho sides of the
i; (ltll ,i l, v (If" ')'>'' I.MIM!''. tlin, ^iuiiiii; Jh.ui any pnim, (/ ) t \ of A
'" Iil1 " fl'WI ^l^+fe^O, and finally tho
jnvr.," .'u.r ''. I ^1 ' M. v,hi.<l, l.y whal pi.Tnl,^ i; t )nliuH 2 in the point corro
(iiKni'lm:! I" 'I*' 1 fi''""'!"! I"""' ,/', '/. /O "' A: mid iimvomoly afcarting with an assume. I
. :,,! ,,,, >" v,.' !'A(, (It. 'iii t  ' I ' i ; I' wliinh unrtHiw tiliTOugh the angles of the
I .> If O
Mi.uu;! .ul 1 .4".^ i X " '1" ir'.iim'.l ji..iul,; tlm IIIVDIHO line / h flfy + / = ; the
luiiuini< piit' (' )  ] ( ) "' ' !l ^ J) "'""i illld lilmll >' lll(! > vurHO point (/. ff, A), which
will 1*' H "" 't" 1 ''tni..' e \. *h" j*"iM* t'Mii.'i.jimuliiii.f in llu'r iiMHUiucd point on the curve S.
72 [396
396.
03ST A CERTAIN ENVELOPE DEPENDING ON A TRIANGLE
INSCRIBED IN A CIRCLE,
[From the Quarterly Journal of Pure and Applied Mathematics, vol. IX. (ISCfH),
pp. 3141 and 175176.]
CONSIDERING a triangle and tho ciroumHcribod circle, and from any point of tho
circle drawing perpendiculars to tho sides of tho triunglo; tho foot of tho throo
perpendiculars lie on a lino; and (regarding tho point an a variable point on tho
circle) the envelope of the line is a curve of the third olusn, having tho lino infinity
for a double tangent, and being therefore a curve of tho fourth order with throo miHpH,
see Steiner's paper "Ueber erne boaondoro Curve drittor ICloaao und vierton GrndiiH,"
Grelle, t. Lin. (1857), pp. 231237, which contains a series of very beautiful geometrical
properties.
Mr Greet', in a paper in the last volume of tho Journal, hot) expressed tho
equation of the line in a very elegant form, vk if a, ft 7 are tho perpendicular
distances of the point from the sides of tho triangle; A, tt, tho angles of tho
triangle; (X, ^ *)(, ^r, ^~); and (X, F, Z) certain cuiTont oooixUimtoH.
viz. these are the perpendicular distances from the sides, multiplied by sin A tan .4,
, sin G tan G respectively; then the equation of tho line is
where the parameters X, /*, v are connected by the equation \ tan A + ^ tan S + v tan G  0,
or say by the equation
39(5] ON A CERTAIN ENVELOPE DEPENDING ON A TRIANGLE &0. 73
Wo havo a cubic equation in (X, /*, v) with coefficients which are linear functions of
(X t Y, %)> and the required equation is that obtained by equating to zero the
reciprocal of this cubic function, the facients of the reciprocal being the (a, b, c)
of the linear relation ; the reciprocant is of the degree G in (a, b, c) and of the
degree 4 in tho coefficients of the cubic f auction, that is in (X, 7, Z), But I remark
that tho equation in (A,, p, v), regarding these quantities as coordinates, is that of a
cubic curve having a node at tho point \~fj.~v t or say the point (1, 1, 1); the
corresponding value of Xa + /tb + vc is = a 4 b + c, and the reciprocant consequently
contains the factoi 1 (a + b I c) 9 , or dividing this out, the equation is only of the degree
4' in (a, b, c). The equation of the curve thus is
  j 1   ruoip. [X\ (\  /t) (\  P) + Yp (p  v) (/*  X) + Zv (v \)(v /*)] = 0,
~"
being of the degree 4 in (a, b, c), and also of the degree 4 in (X t Y, Z\ that is,
treating (X, Y, %} an current coordinates, the envelope is as above stated a curve of
tho fourth order.
A symmetrical method for finding the reciprocant of a cubic function was given
by HUHBU, HOU my paper " On Homogeneous Functions of the Third Order with Three
Varmblofl,'" dumb, and DitU, Math, Jour., vol. i. (1840), pp. 97104, [35]; the
developed oxpreHnion there given for the reciprocant is however erroneous ; the correct
value ia given in my "Third Memoir on Qualities," Phil. Trans., vol. CXLVI. (1856), see
the Tublu 07, p. (H4, [144] and we have only in the table to substitute for (, ij t }
the quantitiuH (a, b, o), and for (a, ft, c, /, g t h, i, j, Is, I) the coefficients of the
cubic function of (X, p t v) t vk multiplying by 6 in order to avoid fractions, these are
( a, b t G, f> (J, It, , j, 1& > l )
{OZ, G7, 0^, 27, 2, 2J, 2^, 2Z, 27, X+Y + Z)
rospcctivoly. The Hubstitution might be performed as follows, viz. for the coefficient of
u', wo havo
76 ON A CERTAIN ENVELOPE DEPENDING ON A.
where the reciprocal in question may be calculated from the before mentioned tulili
67, viz, multiplying by 3 in order to avoid fractions, the coefficients of the table arc
(, b, a, f t (j t h, i, j, k, I )
= (0, 0, 0, o, a, I, b, o, a, ab c)
respectively, and for the facients (, ?j, f) of the table we have to write (to, ;/,
The expression of the reciprocant is
== 6W 4 cVty + aW + &o.,
and dividing by (aHjr + *)* we have the equation of the envelope in the form
6 W + C 9 ay + aWz 1 + &c. = 0,
which must of course be identical with the former result
boyst (cy l)zf + cazta (as co)) 4 abxy (bat ay) 8 + &c. 0.
Instead of discussing the curve of the third class
it will be convenient to mite (0, y, z) in place of (, y t f), and discuss tlui (nuvf
of the third order, or cubic curve
U = aic (y  zf + ly(z~ (u) a + os (to  yf = 0,
which is of course a curve having a node at the point ( = y z) t or nay at tlt^
point. (1, 1> 1), and having therefore three inflexions lying in a line. The equation <*
the tangents at the node is found to be
a (y  *) fl + b (s  ic) s + G(&yf = 0,
that JSj at the node the second derived functions of V are proportional to
(&fo, o + tt, a + l, a, b, c).
The equation of the Hessian may be found directly, or by means of tho
No. 61, in my memoir above referred to. It is as follows:
(6 + c){a(b
+ (o +a){6 (c
+ ( + &) (c (a + 6) + 2a&)
 (So?
2cct +
a s 6) + Qabc] aye = 0.
^IIM.I.I: i\.i iiiui.it IN A rimi.i:. 77
> '>* '.' ' ' : 5 Ar, ?!(. ! . , in ',, I,
i u)i(siiit I)HI iiiiiiH'iirul I'uHur 'I 1 ,
// ' 'M / '^ '!./'> ; ! /I. '//'
/, .'. A'. AM . v, 'f tl.
/ /M'l" 1 ! I/).
/ Mi^ 7 I *V).
' M!*I' t Ui,
Ml" ""llt (1, 1. H. Hi'' lllV.illll
'Hi' 'i""Ji'l ilvlivril rilll''!i"UH inV
 c!i"!" "*' ill 1 niiii' i il. in ijis
78 ON A CERTAIN ENVELOPE DEPENDING ON A [396
I write
X = y ~ z, Y z cc, Z = $ ij,
so that we have identically
and that the equation of the tangents at the node is
Z + &P + c^ = 0.
I write also for shortness
F'='2F + I+L, I'=F+
the equation of the cubic is then
U = asoX* + &7/F + GzZ\ = 0,
and that of the Hessian is
HU = X* (Lfo + F'y + I'*) + F' (J' K + L y + G'z) + Z* (H'x + 7r'y + L
Now observing that we have
lf i  3 (G +J
/v+scM=/v
we find
+ SJtf Z7  Z a (/T 
which shows that the function I2U + 3MU is a cubic function of yz t .%!&> y,
decomposable therefore into three linear factors; and the equation HU+3MU = Q, is
consequently that of the three lines drawn from the node to the three inflexions of
the cubic (or the Hessian). We know also that the Hessian of the three lines is the
pair of tangents at the node^), viz. that regarding any. one of the variables X, Y, Z
as a linear function of the third of them (in virtue of the equation X+
then that the cubic function of X, Y, Z has aA r2 + &7 3 + cZ' 1 for its Hessian,
1 Tailing as the canonical form of n uodal oubio I/=K 8 fj/ B +6toya = 0, then we have HU=x 3 \y a Zlxyz = Q ;
3 =0 ia the equation of the lines from the node to the inflexions, and the Hessian of the binary oubio
is^icy, where ,ri/ = is the aquation ol the tangents at the node. We obtain as the only linear functions
of U, HU wMoh are ilecGinpontible, a?+# 3 and xyz, the equation .1^2=0 gives #i/ = which belongs to the
tangents at the node or else s=Q, which is the equation of the line through the three inflexions : this lino
ia BO obtained n little further on in the toxt.
396] TRIANGLE INSCRIBED IN A CIRCLE, 79
It in interesting to verify this ; I write Z   X  7, the cubic function then
assumes the form
whoro (a, /3, 7, &} have the values presently given.
The Hessian is
(27  20 s ) A 3 + (S  7) . 2AF + (2/38  2f ) F',
or writing 2ZF = # 3  X 3  Y\ this is
= (2a<y  2/3 fl  S + fly) A' 3 + (2/3S 
We find after some easy reductions,
ct= K'+ F f , =3 (a + c) (no
(& +c ),
(6 +c ),
^S tt'+ J' r = 3 (& + o)(6o + Jlf),
and hence
aS  /9y =  81 (a + c) (6 + c) {(c + Jlf ) (60 + JW)  a& (a + c) (6 + c)},
where the expression in ( } is
= (06 + 2ao + 60) (a& + ao + 26o)  a& (a& + ac + 6c + c a ),
^ c (a6 (3ft + 36) + 0(6 + 2a) (ft + 26)  ft6 (a + 6 + c)},
= 2c ((t + 6) (&o + ca + a6),
and thoreforo
a g _ ^ 7 = 1 6 2 (6 + c) (c + a) (a + 6) Mo ;
the othor coefficients may be similarly calculated, and omitting the merely numerical
factor, we have
Hessian =(6
which is right.
I write next>
80 ON A CERTAIN ENVELOPE DEPENDING ON A
or writing azY, y = z\ X, this is
[396
+ #> [K'X3'Y },
we may determine , so that the cubic function of X, Y, 2 contains the factor
aX + bY'+cZ~; writing Z = XY, then
Contains the factor Quotient is
X s ( K'+ F'
JP7  H' + 2ff'
)
(a + o) X s
3(6c+Jlf)F.
We have seen that
whence the quotient is, as above stated,
=  3 (ac H j]/) X + 3 (6c + JJf ) 7.
Comparing the coefficients of X*Y, we have
<& = ( H' + W + F' + /') + 3 ( + ) (6c + jtf) _ 6c (ao + M}>
(a + c) (6 + o) + 3 (a + C ) ( ffl & + ac + 26c)  6c (a6 + Sao I bo),
that is * = 12Hf; and the same value would have been obtained by comparino the
coefficients of ,YP. Hence SUWU divides by a^ + iP + c^ the quotient being
,. , . 
which is
or, finally it is
and we thus have
=  3 {(bo + M) + (ca + j) y + (06 + jlf ) #} ,
so that the three inflexions are the intersections of the cubic curve by the line
(ftc + JO* + (oa
It may be noticed, that if we write
bate + oay + ohz ~  Mu,
by+ cz= v
TRIANGLE INSCRIBED IN A CIRCLE. 81
then os, y, z will be as
: (c  a) {(23f ca) 6u}
and substituting these values in the equation
aai (y zf + ly(z a;) 3 + cz (to  y)* =
of the cubic, we have a cubic equation for the ratio (u : v); and thence the values
(a 1 , y, z) for the coordinates of the inflexions.
It may be added, that wo have
=  3 (oX* + & y + c# a ) {(bo + Jtf)co + (ca + M)y + (ab + M) z\
which is the equation of the cubic expressed in the canonical form.
Pp. 175 179. Effecting the process indicated p. 73, but writing for greater con
venience (to, y, z) in place of (X, Y, Z\ so that the substitution to be made is
( a , I , G , / , (j , h , i , j , k } I )
= (6*, Qy, Qg t 2# 20, 20, 2s, 2a, 2
respectively (where I have corrected a misprint in the formula as originally given) I
find the equation of the envelope to be
tyz (y z)* a 4
I fans (z ~ wf b :
yf c' 1
b 3 c
c 3 a
1 4?/^ (?/ a + z* 4 3fl>y  2y* + 6) a a b
bc a
ca s
4 4no (" 4 a 3 + 3ay  2m; + 5y^) ab a
+ y(y* Vttfg  2i/ 3 ffl f i/s a + 3SaJi/2 + ^ + 12s a a; + ^ ca
+ e (z*  22 a a;  2^y + zffl 9 4 38wy^ + zf + 12^^ 4 12#2/ a ) a a b 3
4 2y^ (llaj a 4 7/ a +  %* 4 24#2/ + 24^aj) a 2 bc
4 %m (lly a + 2 3 4 &  2#w 4 24^ 4 2%) b 9 ca
4 2y (ll^ a + a 4 2/ a ~ 2y + 24w 4 24y#) c s ab = 0.
C. VI.
82 ON A CERTAIN ENVELOPE DEPENDING ON A TRIANGLE &C. [396
The function on the lefthand side is the quotient by 48(a + b + c) a of the
sextie function, Table 67 of my third Memoir on Qualities, [144] ; the foregoing quotient
was calculated without using the coefficient of the term in a a b u c 3 (grfZ 2 ) of the table,
but by way of verification, I calculated from the table the term in question, and found
it to be
( + y + zY
 2 (a + y + zj (ya 4 w + ay)
and this should consequently be equal to the coefficient of a 3 b a c a in the product of
(af b + c) a into the foregoing quartic function of (a, b, c) that is, it should be
+ y (f
+ z (z*  2s 9 *  2z*y + zaP + SSsoyz + zf +
+ tyz (1 lffi a + f + z*  Zyz + 24a,'y f
m (U f + z? + a?  2^a;
+ a 9 + f  Ztcy
which is accordingly found to be the case.
397.
' SPECIMEN TABLE jfcf = o6" (MOD. N) FOB ANY PEIME OR
COMPOSITE MODULUS.
[From the Quarterly Journal of Pure and Applied Mathematics, vol. ix. (1868),
pp. 95 96 and plate.]
IF N be a prime number, and a one of its primitive roots, then any number M
prime to ff, or what is the same thing, any number in the series 1, 2, ... tf 1, may
be exhibited in the form jtf e a (Mod. tf) ; where a is said to be the index of M m
regard to the particular root a. Jacobi's Gmon Arithmetics (Berlin, 1839), contains a
aeries of tables, giving tho indices of the numbers 1, 2, 3...JT1 for every prime
number N less than 1000, and giving conversely for each such prime number the
numbers M which correspond to the indices 1, 2. ,., (tf1) (Tabute Numerorum ad
Indices datos pertwentiim et Indicwn Nmiero dato correspondent^. A similar theory
applies, it is well known, to the composite numbers; the only difference is, that m
order to exhibit for a given composite number N the different numbers less than J?
and prime to it, we require not a single root a, but two or more roots a b,... and
that in terms of these we have Jfa'P ... (Mod.JV). For each root there xs on
index A (or say the Indicator of the root), such that a" el (Mod. N}, A being the
least index for which this equation is satisfied; and the indices *&,.., extend ^ from
1 to A B respectively; the number of different combinations or the product AB... t
being precisely equal to <j>(N), the number of integers less than N and prime to it.
The least common multiple of A. B..,, is termed the Maximum Indicator and repre
senting it by /, then for any number Jtf not prime to Jf> we hav.e Jf'sl(Mod. iY), a
theorem made use of by Gauchy for the solution of indeterminate equations of the
first order. Thus JV=20, the roots may be taken to be 3, 11; the corresponding
exponents are 4, 2 (viz. 3^1 (Mod. 20) 11 si (Mod. 80)), and the product of these
is 8, the number of integers less than 20 and prime to it; the series [go to p. 86]
J. X l
84
SPECIMEN TABLE M s Ct a b ft (lIOD. N) FOR ANY
[397
SOS.
1
2
3
1
5
6
7
8
10
11
12
13 :
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
'29
30
HOOTS
1
2
3
2
5
3
3,5
2
3
2
5,7
C
3
2,11
3,7
10
5
10
3,11
2,13
7
10
5,7,13
2
7
2
3,13
10
7,11
IND.
1
2
2
4
2
2,2
6
4
10
2,2
12 j
C
4,2
4,2
16
6
18
4,2
6,2
10
22
2,2,2
20
12
18
6,2
28
4,2
M. I.
1
2
2
4
2
G
2
G
4
10
2
12
6
4
4
16
G
18
4
6
10
22
2
20
12
18
6
28
4
1
3
2
4
2
G
4
6
4
10
4
12
G
8
8
1G
G
18
8
12
10
22
8
20
12
18
12
28
8
1
2
1
3
1
___.
1
3
2
5
1
2
1
4
6
0,0
1,0
0,1
1
2;
fi
1
1
3
2
t
0,0
1,0
5
8
10
fl
1
B
0,0
1,0
2,0
0,0
1,0
1,1
10
11
4
7
1
17
5
16
2
0,0
1,0
0,0
1,0
2,0
1,1
4
2
8
20
1C
IB
0,0,0
1,0,0
1
7
2
8
3
1
2
6
0,0
1,0
2,1
11
27
22
18
0,0
JL
3
7
1,1
8
4
3
3
g
7
3
0,1
1
7
8
1,1
3,0
0,1
5
9
14
2
4
12
15
3,0
3,0
1
6
21
2
0,1,0
8
6
8
1
16
3
in
M
5
1,0
2
G
4
2
2,0
6
10
2,0
8
18
14
4
2,0
M
10
5
2
1
1
2,1
1
6
1
11
ll 1
11
4
0,1
3,0
13
5
6
0,1
6,0
8
1,1,0
16
5
13
1,1
2 it
0,1
12
6
15
3
14
9
i\.
13
8
3,1
3,1
12
4
13
1,1
0,1
3
12
0,0,1
19
8
0,1
2
,1,0
14
2,1
3
11
7
14
If
16
2,1
2
7
6
13
11
3,1
IV
1C
8
14
4,0
10
4
4
1(1
17
3
8
3,1
5,1
7
17
1,0,1
13
10
15
4,1
7
1,1
18
4
15
U
1!)
2,1
4,1
6
0,1,1
8
7
12
5,0
Ifi
20
3,1
7
12
9. ft
21
5
10
12
9
10
22
11
17
14
ti
23
1,1,1
11
2
11
5,1
24
8,1
24
10
4
2i>
li
10
4,0
H
26
9
lii
27
3,0
23
28
14
29
2,1
307]
PBIMJfl Oil COMPLEX MODULUS.
85
81
2
!)!)
Bd
3fi
80
37
88
39
40
41
42
3
44
15
46
7
48
49
60
3.7
,ir>
3,10
a
a,o
,10
'B
a
2,14
11,21
5,1!)
28
j),21
2,2tt
5
B
,7,17
3
3
20
HO
H,2
10, a
10
12, a
i,a
an
1H
12,2
1,2,2
40
6,2
42
L0,2
12,2
22
16
1,2,2
42
UO
ill)
H
10
10
10
la
im
18
12
4
10
12
10
12
22
40
4
12
20
Hi
'20
24
la
ao
18
24
Hi
40
18
ia
20
24
22
1C
10
42
20
ia
ID
;u
21)
"a'fi "
i
41
an
j
0,0
1,0
LA.
2,1
s.o
0,0
1,0
8,0
0,1_
2,1
n,o
0,1
1
5
0,0
1,0
11,0
2,0
0, 1 '
8,0
10,
0,0
1,0
11
Id
22
1
1
4
0,0
1,0
2,0
0,0
0,0,0
1,0,0
20
13
12
22
0,0
1,0
ay
17
'\C>
K
0,0
1,0
8,0
0,0
1,0
2,0
16
1
30
18
14
17
2
38
14
16
1
0,0
,0,0
2(5
1
10
29
1
1
2
a
4
5
11
a
a,i
'.)
28
88
82
12
"(V
no
lit
If
8r>
'2
1,1
1,0
8,0,1
2,0,0
1
89
88
80
8
14
7
11)
14
2
9,1
2,0
1,1
8,0
19
10
,1,0
27
30
12
13
10
11
33
ao
Ifi
2
7
8
!)
10
a
7
'28
lit
it
7,0
1.1
0,1
0, 1
11,1
7
'1
El
9,0
1,1
8,1
_.
B,l
11,
10, 1
11,1)
7,0
a,i
5,1
d.O
12
17
JL
10
11
8
7,0
0,1
0,1,0
1,1,1
11
27
81
25
87
6,1
0,1
(i
11
40
4
22
2,1
il.O
1,1
1,1
2,1
9
14
17
27
82
8
22
86
,1,0
,0,1
8
17
11
12
13
14
16
id
17
18
19
20
IH
i
H
SB
ii
"17
11
m
10
10
~"c
u
2H
a?
1C
"in
J,0
fi,0
'i;T
(1,1
M
11,0
M
_
112
4,0
0,0
H, 1
i, i
M
M
Id
ia
ifi
10
15,0
0,1
1,1
a,n
H
li
7
25
28
' ao"
17
21
1
2
4,0
2,1
5,1
11_ L 0_
8,0
10,1
0,0
2,1,0
24
88
10
9
81
14
21)
80
18
4
5,0
4,1
!IO
Ifi
81
20
41
8,1
1,1
4,0
9,0
fl,0
7
16
28
42
20
29
81
0,0,1
1,1,1
ao
25
28
U6
80
19
14
0,0,1
1, 0, 1
1,1
2,0
24
D
20
a
10
0,1
fi,o
(1,0
5,1
11,0
18
2
10
11
89
10
31
0,1,1
2,0,0
24
88
87
10
16
13
21
22
28
21
25
7,1
4,1
7,0
i)
in
M
, 0,1
0,0
6,0
'4,1
81
...._
ad
ao
Id
15
10
8
11
1,1
4,1
a, 0,0
2,0,1
17
G
11
7
28
28
10
18
19
21
8,1
87
1
25
19
8,0
4,1
0,1
8,1
10,1
4
18
38
8
43
19
3,0,0
IV
8
18
14
8
2U
27
28
2<)
30
0,0
5,0
an
o
H
'!)
8,1
5,0
. 7,1
.L
1 35
27
10
4
10
ay
18
B7
15
la
10
0,1
5,0
8,1
Al
7,1
0,1,1
1,1,0
2,1
82
27
21)
13
12
7,0
7,1
8,0
5,0
10,0
a
20
5
12
46
20
9
2,1,0
1,1,0
7
4
41
9
4
81
82
88
84
35
8,1,1
2
82
85
4,0
28
86
26
15
38
4,0
8,1
0,1
7,0
21
8
4
24
13
21
16
1,0,1
12
82
19
84
23
7
18
8U
37
38
30
40
41
42
43
44
45
KB
11,1
8!)
8,1,1
40
21
41
8,0
18
21
0,1
8,1
7,1
12
5
26
40
87
41
 7
a, o, i
8,1,1
16
6
8
31
12
5
_43
43
5,1
44
6,1
if>
l:
i
22
5
11
45
47
48
49
47
2,1,1
48
21
49
10
60
50
86
SPECIMEN TABLE M = tfW (MOB. N) &C.
[397
(from p. 83] of these is in fact 1, 3, 1, 9, 11, 13, 17, 19, each of which is expressible
in the required form, viz. IsSMl , SSMl , 7 = 3M1, &c. (Mod. 20): the maximum
indicator is 4; viz. I 4 si, 3* si, 7* si, &c. (Mod. 20).
The table pp. 84, 85 gives the Indices for the numbers less than N and prime
to it, for all values of JV from 1 to 50; the arrangement may be seen at a glance;
of the five lines which form a heading, the first contains the numbers N\ the second
the root or roots belonging to each number N, the third the indicators of these roots,
the fourth the maximum indicator, the fifth the number </> (N). The remaining lines
contain the index or indices of each of the $N numbers M less than JV and prime
to it, the number corresponding to such index or indices, being placed, outside in the
same horizontal line. For example, 30 has .the roots 7, 11, indices 4, 2 respectively ;
the Maximum Indicator is 4, and the number of integers less than 30 and prime to
it is 8 ; taking any such number, say 17, the indices are 1, 1, that is, we have
17 = 7 1 . II 1 (Mod. 30).
The foregoing corresponds to the Tabulce Indicum Numero data oorrespondentium of
Jacobi ; on account of multiplicity of roots there does not appear to be any mode of
forming a single table corresponding to the Tctbulcs Numworum ad Indices datos perti
nentium; and there would be no adequate advantage in forming for each number N
a separate table in some such form aa
Roots
3 11
No a.
1
1
11
1
3
1
1
13
2
9
2
1
19
3
7
3
1
17
which I have written down in the form of a table of single entry ; for although
(whenever, as in the present case, the number of roots is only two) it might have
been better exhibited . as a table of double entry, when the number of roots is three or
more ifc could not of course be exhibited as a table of corresponding multiple entry.
87
398.
ON A CERTAIN SEXTIG DEVELOPABLE, AND SEXTIC SURFACE
CONNECTED THEREWITH.
[From the Quarterly Journal of Pure and Applied Mathematics, vol. ix. (1868),
pp. 129142 and 373376.]
I PROPOSE to consider [first] the sextic developable derived from a quartic equation,
viz. talcing this to be (a> &, o, d, e\t> 1) 4 = 0, where (a, b, o, d, e) are any linear functions
of the coordinates (so, y, z, w\ the equation of the developable in question is
(ae  4>bd + 3c 3 ) 3  27 (ace  ad 2  & a e + 2&ccZ  e 3 ) 3 = 0.
I have already, in the paper " On a Special Sextic Developable," Quarterly Journal
of Mathematics, vol. vir. (1866), pp. 105113, [373], considered a particular case of this
surface, viz. that in which o was =0, the geometrical peculiarity of which is that the
cuspidal edge ia there an excubo quartic curve (of a special form, having two stationary
tangents), whereas in the general case here considered it is a sextic curve. There
was analytically the convenience that the linear functions being only the four functions
, b t d, e, these could be themselves taken as coordinates, whereas in the present case
we have the five linear functions a, &, o, d, e.
The developable
88
ON A CERTAIN SEXTIC DEVELOPABLE
[398
viz these equations are really equivalent to two equations, and they represent a curve
f"tho fourth order which is an excuboqnartic. We may m fact find the equations
of the nodal curve by assuming (a, b, o, d, e& 1)' to be a perfect square, say to
avoid fractions that it is = 3 (d? + 2/3* + tf, then we have
which equations as involving the two arbitrary parameters a : ft : 7, give two equations
between (a b c d, e), and we may at once by means of them verify the above
mentioned equations of the nodal curve. It also hereby appears that the nodal curve
is as stated an excuboquartic curve; viz. we have between a t b, c, d, e a single
linear relation that is a quadric relation between a, /3, 7, and this equation may be
satisfied identically by taking for a, /3, 7 properly determined qvmdric functions of a
variable parameter 6; whence a, b, c, d, e are proportional to quartic functions of the
variable parameter 0, or the curve is an excuboquartic.
The equations of the nodal curve may be presented under a somewhat different
form; viz. the cubicovariant of (a, b, o, d, e$t, 1)*0 being
9nc' J  QlPe
5abe + loacd  IQlfd
 We
Ibbce
ae
*, 1) =
say this function, multiplied by 6 to avoid fractions, is
(a, b, c, d, e, f, g^t, I)",
that is
9ac 3 ~
d = 3 (+ ad* 
e = 2 (+ ade +
f = 1 (+ e 9
g = 6 (+ 6e 3  Scrfe + Zd s ), '
then the equations of the nodal curve may be written
a = 0, b = 0, c = 0, d = 0, e = 0, f=
= 0.
AND SEXTIC SURFACE CONNECTED THEREWITH. 8&
It may be mentioned that we have identically
ae  4bd + 3c a = 0,
af  3bB + 2cd = 0,
ag  9ce + 8d 3 = 0,
bg  3of + 2dc = 0,
bf ~ 4ce + 3d 3 = 0,
and moreover
a6bf + 15ce10d a
= 6(bf
so that fcho equation of the developable may be written in the form
ag6bf+loce10d 9 = 0,
or in the more simple form
each of which puts in evidence the nodal curve on the surface.
The nodal and cuspidal curves meet in the points
ft b _ *L 
b~~ c~d e'
being, as it is easy to show, a system of four points. The four points in question
form a tetrahedron, the equations of the faces of which may be taken to be # = 0,
7^ = 0, = 0, w = Q; and the equation of the surface may be expressed in this system
of quadriplanar coordinates.
We introduce those coordinates ab initw, by taking the quartic function of t to be
(a, b, c, d, e$t, l)' = w
that is, by writing
ftas
b 
G =
e =
Observe that (t lt U, t B , t^ being any constant quantities, we thence have
e  dS<i + o2i*>  VSt&tt
= tf (a i)(o
0. VI.
(70(7 0(7 0(7 ftt)
<,)( *0(8<8)(8*4).
12
90 ON A CERTAIN SEXTIO DEVELOPABLE [398
and thence in particular
e  a + Ga  ay + a 7 = 0.
viz. this is the linear relation which subsists identically between (a, 6, c, d, e), the five
linear functions of the coordinates (a 1 , y, e t w).
Starting from the above values of (a, I, c, d, e), we find without difficulty
J = (a  /3) s 09  7)* (7  afmjz + (a  ) 3 (/3  S) a (S  a ) a e&
+ (a  7 ) 2 (7  S) 2 (S  a) 9 aww + (/3  7 ) a (7  S) 2 (S
but we thus see the convenience of introducing constant multipliers into the expressions
of the four coordinates respectively, viz. writing
IB (/3yS) 9 a/,
where for shortness
or what is the same tiling, taking the quartic to be
(ft, &, G, d, e$t, l)^^
we find
J^ (X> V') 3 (a;
I  (XV V) [V (a)W 4 2/'2') + X (T/'W' + /(B 1 ) + v (s'w 1 + a>y )},
where for shortness
or writing
we have
and the equation of the developable is thus
{V <>W + yV) + ft' (y'w 1 +*'') + / ((c'w' + y)} 3  27X>V (a/y'y I y W + ^Vw' + co'y'wj = 0.
Observe that J=0 is a cubic surface passing through each edge of the tetra
hedron, and having at each summit a conical point; J=0 is a quadric surface passing
through each summit of the tetrahedron, and at each of these points the tangent
398]
AND SEXTIC SURFACE CONNECTED THEREWITH.
91
plane of the quadric surface touches the tangent cone of the cubic surface: to show
this it is only necessary to observe that at the point (of = 0, y' = 0, z' = 0) the tangent
cono in y'z' 4 dot H as'y 1 ~ 0, and the tangent plane is XV 4 p'y' I vY = 0, and that
these touch in virtue of the abovementioned relation V (V) + V (X) + V (v') = 0 I*
follows that ou the curve of intersection, or cuspidal edge of the developable, each of
the summits is a cuspidal or stationary point, that is, the cuspidal curve has four
stationary points; this agrees with the character of the curve as given, Salmon "On
the Classification of Curves of Double Curvature," Comb, and Dull. Math. Jour. vol. v.
(1850), p. 39, via. the character is there given
a = 6, m = 6, n = 4, r = 6, g = 3, h = 6, a = 0, /3 = 4, as = 4, y = 6,
(/3 4, that is, there are ns stated 4t stationary points).
To find the equations of the nodal curve, instead of transforming the equations
as given m terms of (a, 6, c, d, e\ it is better to deduce these from the equation
of the surface ; viz. if there is a nodal curve, we must have
: BJ :
: 18 J
Writing these under the form
parameter C 1 ), we have
XV + /
\V +
0, &c., where ^ is regarded as an arbitrary
h
I ^'
i' + ff
'^O = 0,
w 1 ) = o,
which equations (eliminating ff) mnst be equivalent to two equations only.
I remark that the first three equations may be regarded as a set of linear
equations in 1, w', V, 8'w' >, and determining from them the ratios of these quantities,
we have, suppose, , , n
1 : _ 10 : 6 :  ffw = A : Jl : JJ)
whore
A'
*V,
*W,
a/
G =
The value of ff i in fact =^, that is, instead of the fool
', \ve have really four determinate equations.
92 ON A CERTAIN SEXTIC DEVELOPABLE [398
"We have thence AD BG = and (substituting In the fourth equation) A (\'a + fjfy' + vs?)
f C (y'z + #V f aty') = ; each of these equations must contain the equation of the
cone having (#' = 0, y 1 = 0, = 0) for its vertex, and passing through the nodal curve.
The two equations are of the orders 6 and 4 respectively; aud as the curve is a
quartic curve passing through the vertex iu question, the equation of the cone is of
the order 3. I have not effected the reduction of the sextic equation, but for the
quartic equation, substituting for A, G their values, this is
 (XV + fify' + Jaf ) [XV (y  z'} + /*y (2'  of} + v'z* (z  x 1 )}
+ (^ + M + ay ) [xv w  *') + /*y (*'  ao + M W  y'}1
+ fo*V (y 1  z'} + vW (z 1  x'} + XV (a/  y 1 }} (of + tf + /) = 0,
which is easily reduced to
XV  a' + 'z r + ^ + e'
+ y'^' ( ^ 3 + y'z 1 ^ + a;y ) (x 1  2/')
+ ^V [(of + y' + O (2/Y I sV 4 ,y) I
+ v'\' [(a f + y' + z f ) (y'z* + sfot 4 'i/') +
+ XV [(aC + y + 2') (2/V + *V + /2/') + x'y'z 1 } (of  y 1 } = ;
and I have found that this is transformable into
2 [si v (V) + y' v M + *' vcoi x [> v vc^'x^y  w'^o ^ v(/*o(^ ^ v ) *y vc^xxv  ^o
V (/)} (V (')  V (V)} (V (V)  V (/*'))] = 0,
viz, the two functions are equivalent in virtue of the relation V(X') 4
or, what is the same thing, they only differ by a function (a/, ?/', .s') 4 into the
evanescent factor X /a + /jf 3 + 1/ 3 2/*V 2y'X' 2XV The function in { } equated to
zero is therefore the equation of the cubic cone.
I do not atop to give the steps of the investigation in the above form, as the
investigation may be very much simplified as follows: by linear combinations of the
four equations in co', y' t sf, w', &', we deduce
( X'  jtt' + y') (^ + w'  x 1  1/') + W (fffij  *W) = 0,
( V + pf + v 1 } (d + y j + z' + <of) + W (tfd + atof + x'y' + ts'w' + y'w' + z'w') = 0.
Hence writing
X = X' X v', at = w' + of y' ^,
fj, X' + /i' i/, y = w' a;' + 1/' /,
v = X' /*' + j/, 2 = w' a;' T/' + ^',
w = / + / + y 1 + /,
398] AND SEXTIC SURFACE CONNECTED THEREWITH, 93
wo find
X/i = X' 2 / 3 + y' 42X>',
and thence
pv 4 yX + X/i =  (X' a + ^' 3 + i/ 2  2/*V  2j/X'  2X>')> = >
that is
1 4. 1 4. 1 0
, H  h  = U,
X /i i/
the relation which connects the new constants X, p, v* Moreover
yz~MW = 4i (tfz 1 afwf),
zx yw = 4 (tf a/  y'w'},
xy zw k (afy 1  z'w'),
3w a  fl) a  f  &  8 (i/d + t/a/ + w'y' + a/v/ + y'w' + z'w'}>
and writing for greater convenience 6 j,, the equatioiis are transformed into
ccw  yz ,
yw m,
Ovz =zw ay,
where
.
X p v
iz. those equations, eliminating fl, give the equations of the nodal curve.
From the first three equations eliminating 6, we deduce
/*)===
or, as these equations may be written,
which equations, from the mode in which they are obtained, are it is clear
to two equafcionB only. Using the fourth equation, and ehmmatmg * by
therein for ^ ^ fo their values from the first three equations, we find
94 ON A CERTAIN SEXTIC DEVELOPABLE [398
that is
8^ + tf + ^ + rf2wf2* +  +
J V so y z
or, what is the same thing,
sty* (3H/ 5 4 x~ + f 4 &)  2w fy"^ + sV + tff) = 0,
we have to show that this is in facb included in the former system, for then the
four equations with 6 eliminated will it is clear give two equations only.
Observe that the former system may be written
(p  v} ifz* F (v  X) sPa? + (X  j^) off = 0,
fir ( ti^fi ___ i
and that we have thus to show that substituting for w the value 10 =
fj,
in the equation
anje (3w a + 3 + f 4 s a )  27y (i/V + sV I a; a i/ a ) = 0,
the result is
(}j.  v) fz* + (v  x) 0V + (x  it) a?y = o.
The substitution in question gives
3.' W ^)' (a , ) _ y
(p vfyz J ^ J ' (/i y w ^ x
thafc is
3a: a (^tj/ Q  vsff + 0*  v) a i/ 2 s a (a? + y 8 + ^ a )  2 (/i  v) (^f  v&) (f# \ 0V + ttPif) = Q,
which is in fact
_ pty (y* _ ^ ^ _ gfy + 2/iWC 9 (^ _ z *y _ uV (y  &) (af>  if) = 0, '
that is, throwing out the factor y' i zP t it is
? (f  z*)  vW (a; 3  f) = 0.
But in virtue of the equation ~H  & =0, we have
1 \ j, v
and the required property thus holds good,
398] AND SEXTIO SURFACE CONNECTED THEREWITH.
We thus see that the equations of the nodal curve are
the nodal curve is thus the partial intersection of the two cubic scrolla (skew surfaces)
(ft  v) wyz = to dnf  v&), (v  X) wgai = y (v&  ^ a ),
viz taking A B t ' 0, D to be the summits of the tetrahedron th 3 faces whereof at*
taking A Qf ^ has ^ fop ^ nQdal C f
rgt d\,cU BA CD fo, gene,ato,; the aaooud has JU) to .nodal taeto. ^
for a shmle directrix, ^D, 0J) for generators; the surfaces intersect in the line AM
tL; S lint S) twice/and the le OD; the order of the residual curve, o, nodal
curve of the developable, is thus 9 (2 +2 + 1), =4 as it should bo.
I remark that the equation
(^y)^ a +^^)^ a 1(^^)^ 3 = '
is the equation of the cone having its vertex at the point A (00, 2/ = 0, *0), and
.
ier' of linos tomgh two points which p through a give, pomt not on the
curve is in fact = 3.
It remains to introduce the coordinates <. y, ., ) * the equation of the
developable. "We have
1 '  2
and thence / / / \ a / rt ,
V = (w  ')  (y 
giving
andsimilarly
8 (^' + off) ^tf
Moreover
= (fp _ ^ _ _
+ y +
96
ON A CERTAIN BEXTIC DEVELOPABLE
[398
Consequently
' (yV + 2'' + afy'}  (to 4 <e + y 4 z)
x { 3w 3  2*y (0 4 y + *)  ffi"  f  ** 4
atyY  (w +  #  *) (w  4 y  #) (w   y  s),
 w 9 ( 4 y
+ rfVi3 _L rt/3 _L_ 3 ^_ 3f4Q i nfiff __ &ff& _, ^/yi ^^ />!4l3 4 fl^it J.
i^ T ^ ~r y* y A &ui it iti^/ * Ui ij p
Putting for shortness
^^aj+2/4xr, V = a 2 4 1/ 2 4 s a  %yz 
the two expressions are
toiiS
4 m  2p f  V
or obsei'ving that  pV is
we have
4tti 11 .
4 cs 3 4 y s 4 a 3
4
that is
'^' + wV' 4
4
 2w (p a + V ) 4
4 Z/ a 4 ^ 3 )
(a? \f
4
Moreover
8 {V (a/w 1 4 2/V) 4 /*' (tfof 4 //} 4 / (z'w 1 4 *V')
4 1/ (w 3  ffi a  y a 4 s 3 ),
  (X 4 /* 4 y) w 3 4 ?w a 4 /i2/ a +
398] AND SEXTIC SURFACE CONNECTED THEREWITH.
and we have
/* = 2 (7  S) (a  8) (a 
v2<
whence X^y = SX'/^V
Hence finally X, ^, v denoting as just mentioned, and therefore satisfying
_ 4. i: 4. L = o, the equation of the developable is
\pv {w 3  w (a? + ;?/ 2 I ^ a ) + Iwyz]* f 108 {(X + /i + v) w  Xte 3  pif  vsP}* =
(say this is \jj,vT*\ 108$ a = 0), and this surface (which has obviously the cuspidal curve
jS' = 0, T 0) has nlso fcho nodal curve
aP (/My 8  P) w a (vz*
~  ^^
X
I will show <)> posteriori that this is actually a nodal curve on the surface. Intro
ducing an arbitrary parameter $, the equations of the curve may be written itt supra,
= anu yz,
6vz zw cey,
2(9 (X I /* + v) w = Sw 3 a?f z
and wo have thonco, OB before,
Hence
(X + /i I v)
w (a? + if + z*} +
(X + ft + v) vf Xffl 3 iw/*
Honce writing
= 0.
= >
& . 2 (X + p + v) w  (3w a  3  f  &} 0,
0. VI. 13
ng ON A CERTAIN SEXTIC DEVELOPABLE [398
substituting for its value = , and attending to the significations of S and r l\ we have
which are in fact the conditions to be satisfied in order that the point (as, y, z, w}
may belong to a nodal curve of the surface Vi'2*+ 10SS*= 0.
It is to be noticed that the coordinates of the before mentioned four points of
intersection of the cuspidal and the nodal curves (beinff as already mentioned stationary
points on the cuspidal curve) may be written SB, y t #, w = (l, 1, 1, 1), (1,  1, 1, 1),
(1,1,  1,1), (1.L 1. !)
We have 'thus far considered the developable, or torse, the equation of which is
[X' (x'w 1 + yV) + p' (y'w 1 + z'x 1 ) + v' (z'w + 'y')Y
where VM + V(/*0 + V("') = 0; 01, what is the same thing, writing a, 1, o, in place
of V (V), V (/*0. V CO respectively, the torse
where ft + 6 + c = 0.
Inverting this by the equations of, y', z', w'^, ~, , , we obtain a s
surface
{a 9 (ano + yz) + 6 1 (yw + B) + o 3 C^w + 0i/)! 3  27 a 6V a^w (ffl + y + z V iu? = 0,
where o + 6 + o = 0; which surface I propose [secondly] to consider in the present; paper.
The surface has evidently the singular tangent planes # = 0,^ = 0, 2 = 0, w == 0,
each osculating the surface in a conic, that ia, meeting it in the conic taken fchrice, viz,,
AJ = 0, in a conic on the quadrio cone cfyz + iPyw + rftew = 0,
= 0,
= 0,
w = 0, a> + & 3 2ffi + o^y = ;
398] AND BEXTIC SURFACE CONNECTED THEREWITH. 99'
and it has also a cuspidal conic, the intersection of the plane K + y + g + w*=Q with
the qiwdric surface
ct 3 (xw + yz) + & a (yw 4 sue) + c 3 (ziu + 2/2) = ;
it may be observed that the four conies of osculation are also sections of this surface.
The surface has also a nodal curve, the equations of which might he obtained by
inversion of those of the nodal curve of the sextic torse above referred to; but I
prefer to obtain them independently, in a synthetical manner, as follows:
Take a, /3, y arbitrary, and write
_ A = (b  c) a. + bfi  cy , F = by  c/3,
.B (c a) /3 + cy a j G = cet ay,
 G ~ (b  c) 7 + an  6/3, ' H = a(3  la,
M = (6  c) + (c  ) jS + (a  &) y,
Q ==a 9 (&o)a + & a (<>3) + tfl ( a  & )r :
then it is to be shown, that not only the equation of the surface is satisfied, but that
also each of the derived equations is satisfied, by the values
a : y : * : w^aAGHQ : WHFQ : cGFGQ : abofGHM;
oticli of the quantities A, B, G, M. Q is linearly expressible in terms of ff, G, H,
which arc themselves connected by the equation aF + bG+oH=Qi the foregoing values
of ar, y> 2, w aro consequently proportional to quartic functions of a single variable
parameter, say F+G; and there is thus an excuboquartic nodal curve,
To establish the foregoing result, we have
+ cF =0,
aG + bff + cG  0,
aA + bB + cG = 0,
aAGH +bttBF
aff^BT +6J0 1
aM.ff.ff + 6"^J3^ + fl"C^ff =  abc 1*3 ( a
vhioh are all of them identical equations; but as to some of them the verification is
rather complex. ^ _ 2
100 ON A CERTAIN 8EXTIO SUIIPAOIC &0.
Hence we have
at  y i g = Q (aA Gil + blllU' 1 ~ o(M.l)
Jind thonco
f [Q (a. \
Moreover
and
( + y + ^ + w) B  27 (ak)' 1 (AMMWMQy ().
Again
I c'^) w
a H.  . 7 ' .. J
+ 6 a w + c^;i/ = ttbolWHQ* (<tIl(W+ WAG I o/t.71/7)
. %Q (a ( /3 I 7)' J ,
and thence
(t a (,'tfiy 4 y#) I // J (yw I sis) I t:' J (;/
and the two oquntions inavkod (#) verily bho oiiiabion ol' tho
To verify the derived oquiifcionH, writo fin 1 it, iiiitiuouli 1* *= tt* (i/a *\ into) *\ b* (sta [ yw)
+ c a (a 1 )/ + am), HO (ihali the equation of tho Hiirfuco in 7 );1  ffltPbWGi/sw (u> h y I I w) 11 ^ 0,
and the derived equation with roHpoot to w in
JL^L h .. 2
7 J <?/ ; a  y I e I w '
or substituting for P and flj + y + + w thoii valucH, this IH
and similarly for y, e, find w. In parliieular, iioiiHidoring tho dorivod equation in nwpuob
to iw, this is
a*tv + 6 a
.
and wo have as before
which is thus verified; tho verification of tho durivod oquationn for y, ^ w can bo
effected, but not quite so easily.
Tho existence of the oxouboquaibio nodal ourvo is thus established.
399,
ON THE CUBICAL DIVERGENT PARABOLAS.
[From tho Quarterly Journal of Pure and Applied Mathematics, vol. ix. (1868),
pp. 185189.]
NEWTON ' reckons five forma, viz. these are tho simple, the complex, the orwiodal,
the aonodal, and the ouspidal, but as noticed by Murdoch, the simplex has three
different forma, the ampuUate, tho neutral, and the oampaniform. We have thus the
8 forma at once distinguishable by the eye.
Pllloker has in all 18 species, the division into species being established or completed
geometrically by reference to the asymptotic cuspidal curve (or asymptotic semicubical
parabola), and analytically as follows, via. writing the equation in the form
^> = <D83oflj + 2d,
the different species are
.simplex, f  a?  3c + 2^1 ^ ampullate,
i/ 8 = a 3 3a 2rfj" ' campnuiform,
f = a? + 2d, neutral,
3 a, 18 2c^, campaniform,
f = ?" + Scffl f 2rf,
)( I/a = ? + 3caj,
/ == ) 3 4 3oaj  2cZ,
complex, i/ 2 = a 8  3ca) + 2rf g
*
acnodal, i/ 3 = of 1  3cj  2c V (c),
crunodal, y = "  Sea; + 2c V (c),
cuspidal, 2/ 3 = fl' 3 ;
102 ON THE CUBICAL DIVERGENT PARABOLAS. [39&
but of the simplex species, there are five which are to the eye campaniform, and the
three complex speeies have with each other a close resemblance in form.
I remark as regards the simplex forms, that the tangents at the two inflexions
meet in a point R on the axis, and that the ampullate, the neutral, and campaniform
forms are distinguished from each other according to the position of R, viz. for the
ampnllate form, R lies within the curve, for the campaniform form R lies without the
curve, and for the neutral form, R is at infinity. It is to be observed, as regards the
complex forms, that here R always lies without the curve, between the infinite branch
and the oval.
The further division of the simplex and complex forms so as to obtain the 7 + 3
species of Pliicker, may be effected by considering in conjunction with the point R
a certain other point 7 on the axis ; it is to be remarked that excluding the
inflexion at infinity the cubical divergent parabola has in all eight inflexions, two real
and six imaginary, viz. the inflexions lie by pairs on four ordinates, or if ID be the
abscissa corresponding to an inflexion, as is determined by a quartic equation; this
equation has always two real and two imaginary roots, each of the imaginary roots
gives a pair of imaginary inflexions ; one of the real roots gives a positive value for
i/ 3 and therefore two real inflexions, the tangents at these meet in the abovementioned
point R on the axis; the other real root gives a negative value for y a and therefore
two imaginary inflexions, but the tangents at these meet in a real point on the axis,
and this I call the point /, It is clear that for each of the four pairs of inflexions
the tangents at the two inflexions meet at a point on the axis, so that if X be the
abscissa of such point, then X is determined by a quartic equation; two of tho roots
of this equation are imaginary, the other two roots are real, and correspond to the
points R and / respectively. .
The equation of the curve being as above
y" = '3aB + 2cZ J
then the coordinate m belonging to a pair of inflexions is found by the equation
ar6ca.' a l8cfa;3c a = 0,
or what is the same thing,
(1, 0, c, 2d, 3c>S>, 1) 4 = 0,
(the invariant I is =0, and hence the discriminant, = 2V/ 3 , is negative, or tho roots
are two real, two imaginary, as already mentioned) : the corresponding value of X is
easily found to be
y _ x 3 + 3ctc  4d
3(a a c) '
and we thence obtain
3oZ 4  4fLY 9  6c a Z a + WodX  (c 3 + 4>$)  0,
or what is the same thing,
(80, d, c*, Sod,  <?  ^Z, 1)* = 0,
ON Tllli (HWKJAL DIVERGENT PARABOLAS. 103
lor Uin (.!(ua(;ii)n in A r ; the qiiadrinvariant I is =0, and hence the discriminant,
 27./' J , in imgtitivu ; that w, the roots are two real, two imaginary, as already
iimiitioiiod.
Miu Himplux forms, first, if o = 0, thou for the ourvo
f = ? + %d t
il iippimi'H that 74 lion nl; infinity, I within. tho ourvo; and for the curve
f ;"  2d,
that A' lia withuiili tlui ouvvu, / at inlinity,
.It Curtilmr appcavH Uvuli \vlusn (Z = 0, or Ixir liho ourvti,
ft, / lio nipudiHtiiuttr frotn Uio vwtox, R without, / within tho ourvo,
]niutt! in tlui nui'vn
7/aa W + atvr + 2rf,
wiiMjn, when (/, ~~ 0, Uio pointM jf(, I aro oquidiHtant from tho vortx, and for c = 0, the
point. H IH at inlinity, it in nany to iufur by wmliinuity that tho points 7J, / lie R
l', / within tin? imvvn, / liuin^ noiiror to tint vortox.
Anil miniltirly in \>\w c.uvvo
l?/ a =; l Hft';2(i F ,
that Mm pninlH Jt, I lit: ./i without, ./ within tho (iiirvo, R boing naaroi to tho vortox.
Ajriiiti, in ihn I'tuvi'
;/ J ^rt;' l ;itoH2^
fiiutn, in thii uiirvn f * u? \ l\w + $d, R IH without, / within tho ourvo, and an o
ln'iioini'H rc(), 74 IIUBMIIH oil' Ui iniiiiity, it unpoaiH that a having ohnnguil its sign, or
lor Llm (iiirvo now in ipiOHtiou, H liitvin^ pnmud thwnigli inlinity, will bo tdtuato within
Lhu mirvti; that in, ,/i', / lit! uiuih of thoni within tho curvo.
And Himilavly for Llin mirvo
yj=/M3c,2d,
it appi'inH that H, I lio wich without thu ourvo.
co, Ibmlly, lor tlio Hiinplox i'oniiH, wo havo tho 7 apcoioa of Pltickor
f M flu  3ou  2rf, c 3
x ainpulhito, ,/i, f within the curve;
2/i ? Sow 2(2, c u
ix (iainpaniform, 74, / without the curve;
simplex neutral, J within tho ourvo, JK at infinity;
104 ON THE CUBICAL DIVERGENT PARABOLAS.
f = a*%d t
simplex campaniform quasineutral, R without the curve, / at infinity ;
y s = fl! 3 +3cflj + 2d,
simplex campaniform, R without and further from, J within and nearer to the curve ;
simplex campaniform equidistant, viz. R and I are equidistant from the curve,
R without and I within ;
f = a + Sea  2d,
simplex campaniform, R without and nearer to, J within and further from tho curve.
Passing to the complex forms, suppose for a moment that a is the diameter of
the oval and J3 the distance of the oval from the vertex of the infinite branch ; the
equation of the curve then is f = as (as  a) (<o  a  /3), or changing the origin so as to
make the term in & 1 to vanish, this is
y (as
or, what is the same thing,
3 ) a  fa (  ) (2a + jS) (a + 2/9),
or comparing this with y 3 = ar 1  3c + 2d, fZ is =+, or , as </3, o = y9, >/3, or say
as the oval is smaller, mean, or larger ; viz. the magnitude of the oval is estimated
by the relation which the diameter thereof bears to the distance of the oval from
the infinite branch. In the case d = 0, or for the curve y* = x*~ Sea; it appears (as
for the corresponding simplex form ?/ 3 = $ + Sea 1 ) that the points R, I arc equidistant
from the point 0, which is in the present case the middle vertex, or vertex of
the oval which vertex is nearest to the infinite branch. As the oval diminishes, so
that the curve becomes ultimately acnodal, 7 remaining within the oval ultimately
coincides with the acnocle ; and as the oval increases so that the curve becomes
ultimately cruuodal, R remaining between the oval and the infinite branch; ultimately
coincides with the crunode; and it hence easily appears by continuity that for
a smaller oval I is nearer to, R further from the middle vertex ; while for a larger
oval, J is further from, R nearer to the middle vertex. Hence for the complex forms
the species are
smaller oval, I nearer to, R further from the middle vertex;
^ ofBotSt
mean oval, R and J equidistant from the middle vertex ;
?/ 3 = a; 3 3ftB2(Z,
larger oval, J further from, R nearer to the middle vertex: and the division into
species is thus completed.
Gwribridge, June 16, 1865,
400.
ON THE CUBIC CURVES INSCRIBED IN A GIVEN PENCIL OF
SIX LINES.
[From tho Quarterly Journal of Pure and Applied Mathematics, vol. ix. (1868),
pp. 210221.]
WE have to consider a pencil of six lines, that is, six lines meeting in a point,
and a cubic curve touching each of the six lines. As a cubic curve may be made
to satisfy nine conditions, tho cubic curve will involve three arbitrary parameters ; but
if wo have any particular curve touching the six lines, then transforming the whole
figure homologoitHly, tho centre of the pencil being the pole and any line whatever
the axis of hornology, tho pencil of lines remains unaltered, and the new curve touches
tho six linos of the pencil ; the transformation introduces three arbitrary constants,
and fcho general solution is thus given as such homologous transformation of a
particular solution, To show tho same thing analytically, take (<e = 0, y = 0, 2 = 0) for
the axes of coordinates, tho lines &=(), i/=0 being any two lines through the centre
of tho pencil, HO that tho equation of the pencil is (*$>, #) a =0, then if <b(x, v. *)=*0
is the cquati
10 (J ON THE CUBIC CURVES INSCRIBED IN A [400
in fact, comparing the two sides of this equation, we have each of the seven coefficients
of the sextic equal to a function of the seven quantities V( C )> W (c), &V(c)j
&V(c), j, J> ft so tliat conversely, these seven quantities are determinate (not however
rationally) in terms of the coefficients of the given sextic. And when the sextic is
expressed in the foregoing form, then it will presently be shown thafc we have
(a, h, k, 6]fa y} 3 +3z(j, i,/3fc 2/) a + ^ 3 = 0,
or, what is the same thing,
(a, 6, o,f t 0, A, 0,j, fcjflf, 2/,) 3 = 0,
as the equation of a cubic curve touching the six given lines ; and by what precedes,
it appears that this may be taken to be the equation of the general cubic curve
which touches the six given lines. On account of the arbitrary constant c, it is
sufficient to replace s by oa: + j3y + .&, or, what is the same thing, to consider = as
the equation of an arbitrary line, but without introducing therein an arbitrary multiplier.
To sustain the foregoing result, consider the cubic
(a, &, c, /, g, h, i, j, /c, Zjffi, y, 2) 3 = 0,
then in general if 4 = (,$, y, zf, = ($>> y, *)'(, /?, 7), C^^&a, y, 2) (a, /3, ry) Q ,
^ = (>>$. a > &> 7) 3 i * ne equation of the pencil of tangents drawn from the point (, /3, 7)
to the curve is
A*D*  GABOD + 4vi0 3 + ^"D  35 3 a a = 0,
but wi'iting for shortness
(,$<*, y,gy = (A' t ff t (?, J)'J1, #)',
so that
A' = (a, A, fc, 6 Jflj, 2/) a ,
^ O'.t/J*,ff)'i
0" ((/, i$ffi, ?/) ,
D'= c
then for the tangents from the point (# = 0, #=0), writing (a, /3, *y)(0, 0, 1), we have
A = (A' > S, G'.D'^l, zf,
B  (ff, G', jD'Jl, *),
and thence the equation of the pencil of tangents is
AW  QA'S'G'D' + *A'C' & + 4>B' a D'  3J3' 3 a /a = 0.
Hence for the curve
(a, I, o,f, 0, k, 0,j, k, Z$tf, 2/) 3 = 0,
400] GIVEN PENCIL OF SIX LINES. 107
we have g*=Q, i' = 0, and therefore C" = 0; the equation of the pencil of tangents is
J l' a D' 3 +4#' 3 D'=0 ) or throwing out the constant factor D', and then replacing A 1 , B\ D'
by their values, the equation of the pencil of tangents is
o((a, h, k, frjffi, yyp + qO*, 1,/Jia, 2/) 3 ] 3 = 0,
which is the beforementioned result.
The coefficients V(c), &V(0), &V(o). ^V(c), j, I, f, or (as we may call them) the
coefficients of the cubic curve, are, it has been seen, functions of the coefficients of
the given sextic (*$#, y) 8 ; hence the invariants S and T of the cubic curve are also
functions of the coefficients of the sextic, and it is easy to see that they are in fact
invariants (not however rational invariants) of the sextic. To verify this, it is only
necessary to show that the invariants S and T are functions of the invariants of the
functions A/( C ) (> /*> k, &$X Z/) a and (j, I, f$x, y)"; for if this "be so, they will be
invariants of the function
!>(, h, k, l>lx, ?/) s ? + 4[(j, l,f$x, yf}\
that is of tho soxtic. Wo have in fact the general theorem, that if P t Q, R,... be
any qualities in (a;, y, ,,.), and <(P, Q, It,...) a function of these quantics, homogeneous
in regard to (iv, y, ...), then any function of the coefficients of $, which is an invariant
of the quantica P, Q, .R, ... is also an invariant of 0.
Considering for greater convenience the function
(a, h, k, &$a, y) s
in place of V (o) . (ft, h, k, b^ic, y)\ the invariants of the two functions (a, h, k, &][#, y) 3
and (j, I, f$fo, y)' J are as follows:
D = a 9 b  Gabhk + 4a/c 3 + 4&A 3  3W,
V fj~l\
@ = j (bh  k") + I (lik  ab) +f (ah  h*\
R = + 1 a 3 / 8
+ 6
 G
 6
+ 12
+ l
 B
+ 12 bhjl*
 6 6/c a 2
+ Q
 18
+ 9
 8
108 ON THE CUBIC CURVES INSCKIBED IN A [400
viz, Q, V tire the discriminants of the two functions respectively, and 0, R are
simultaneous invariants of the two functions, R being in fact the resultant. The
corresponding invariants of the functions V(c) . (a, h t k, 6}J>, y}\ and (j, I, /Joj, y}* are
obviously c 2 D, V, c@ and cR.
The values of 8 and T are obtained from the Tables 62 and 63 of my "Third
Memoir on Qualities," Phil. Trans, vol. CXLVI. (1856), pp. 627647, [144], by merely
writing therein g=*i=Q. It appears that they are in fact functions of c a D, V, c
and cR; viz. we have
8 = V 2 +c,
T = 8 V 3 + c (4R + 1 2 V 0) + cTJ.
The invariants of the sextic (*$>, y)\ if for a moment the coefficients of this sextic
are taken to be (a, b, c, d, e, f, r/), that is, if the sextic be represented by
(a, 6, c, d, e,f, 0]j> ( T/) 9 are the quadrinvariant (=af/ G6/+ looe 10d), Table No. 31
and Salmon's A., p. 203 ( l ), the quartin variant, No. 34, and Salmon's B., p. 203, the
sextinvariant No. 35, and Salmon's C., p. 204, and the discriminant, which is a function
of the tenth order ^a s g s +&c. recently calculated for the general form, Salmon,
pp. 206207, say these invariants are Q,, Q,, Q g and Q 10 . These several invariants
are functions of the abovementioned expressions c a D, V, c and cR\ whence, con
versely, these quantities are functions of the four invariants Q 9 , Q 4 , Q tl Q n  and the
invariants S, T of the cubic curve, being functions of c 3 D, V, c and cR, are also,
as they should be, functions of the invariants Q a , Q t , Q a and Q n of the sextic pencil
(*", y) 8 
To effect the calculation of Q a , Q t and Q Si I remark that inasmuch as by a linear
transformation, the quadric (j, I, f$ce, 7/) 2 may be reduced to the form 2%, and that
the invariants of (a, h, k, b^so t y) 3 and 2% are
Q = a &s _ Qabhk + 4a/c 3 + 46A*  3/iVc 3 ,
V =  Z 2 ,
=  I (ab  We),
.K =  8l 3 ab,
hence, writing j = 0, /= 0, and writing also c=l, we may consider the sexfcic
[(oj A, /c, o]j_A', y) 3 ] 3 + 32i a a; 3 7/ 3 ,
that is
( 2 , ah, i(Za/c + Sh*') ] ^ (ab + 9M + 3 Ql s ), ^(26A+3A?), i/c, 6Jw, y) fl ,
the invariants whereof are found to be functions of the last mentioned values of
D, V, , _K; to pass to the given sextic (*#, y) , put equal to
c [(ft, h, k) oQco, 2/) 3 ] 3 + 4 JY j,
have only to consider D, V, 0, R as having their beforementioned general values,
pft the coefficient c by the principle of homogeneity.
to Salmon's Lessons Introductory to the Modem Higher Algebra (Second Edition 1866).
1885, the values are given, pp. 260 2G6. '
400]
GIVEN PENCIL OF SIX LINES.
109
As regards the discriminant Q ln , this as already remarked, has been calculated for
tho general form, but for the present purpose it is easier, by dealing directly with the
form [(a, k, k, 6Ja?, 2/) B ] a + 32J B afy s , and then interpreting D, V, 0, R and restoring the
coefficient o as above, to obtain the discriminant Q,o of the function
[c (a, h, k, 6$X 2/) 3 ] a + 4[(j, /,/][&', i/) 3 ] 3
in the required form, as a function of c a D, V, c, cR.
I find after some laborious calculations
O 9 = 10 No. 31 = c 3 (
+
) 4 = 10000 No. 34= c j {
Q s = 1000000 No. 35 = c" { +
9 a
40 jR
288 V
256 V 3
99 D 3
400 RU
2304 V0D
8640 3
12800 .ft V
82944 V 2 2
1 4608 V 3 D
20480
+ 147456
65536 V"
7992 D 3
72000 RQ*
145152 V@^D
622080 3 D
160000 ^D
691200 RV0D
3456000 R 3
3815424 V a @ a D
36080640 V 4
635904 VD a
C+ 33177600 .RV 2 3
+ 4669440 ,BV B D
+ 217645056 V 3 3
+ 23003136 V J D
+
+
110
ON THE CUBIC CURVES INSCRIBED IN A
+ (? (+ 8192000 J^V 3
11010048012V 4
509607936 V 2
1^+ 14155776 V e D
+ G J 62914560 12V
452984832 V 7
134217728 V
[400
= multiple of discriminant
. 4 64
48
to which may be joined
QJ= c 1 [ 81 D a
720 RU\
5184 VD
1GOO R 3
f 23040 R V
+ 82944 V 2 3
^+ 4G08 V ft D
+ c J 20480 1V 3
\+H745G V J @
+ 65536 D
and
: C"
+ C B ''
180
1120
+ 2880 VD
 8640 @ B
o a f 1600 E a
1+ 10240 RV
270 D 3
720
51840
77760
400]
GIVEN PENCIL OF SIX LINES.
in
r  23200 ]&*n
 5356SOR0VD
4 432000 R&
 1762560 V6 2 D
4 4510080 V 1
+ 17280
f  64000 R 3
 1382400 _R 2 V0
 5806080 .RV 2 3
+ 30720 JRV'Q
+ 3317760 V 3 3
 1105920 V^D
 204800 JJ 2 V 3
39321fiOEV*.
The foregoing values of S and T give
+ c 3
+ 24. VfciD
64.0"
16 J$
+ 96 RV
 48 V 2 4
so that
Qio.
cltt 3 '
and therefore
[oR (V
which are interesting in the theory.
112
ON THE CUBIC CUEVES INSCRIBED IN A
[400
Wehave
and if by means of these values we eliminate o and c a Q, we obtain Q 2 , Q 4 , & and Q,
as functions of S, T, V and aft Choosing instead of Q 4 and Q, the combinations
Q<?Q t and Q a ~8Q/, and forming also the expression for the combination Qa(QtQd,
we have tints the system of formula?
= 9 2 T
 432 S
 72T
_ 72
+ 86*
+ 144
+ 128 VScR,
+ 272*
 4212 PViSf
+ 6588 T a V 3
+ 252 T'cB
 7776 TO 3
 16848 TVS
+ 3456 2'
 2592
 1296
 1824
 448
+ 544320
461376
+ 74304
+ 15552
 10368
 10728 V 7 S
 3264
 1536
+
81
972
612
108
 3888
 5184
 11052
 2016
288 2'V"
352 yV a
64
 77760 V
+ 153792 V
 1728
+ 29376
+ 26996
+ 576
+ 1088
+ 512
4:00] GIVEN PENCIL OF SIX LINES. 113
and, as mentioned above,
Q 10 = c 3 R 3 ( 2*
The juatmentiouod value of Q 10 should, I think, admit of being established a priori,
and if this be HO, then the substitution of the values of 8 and T in terms of c 3 D,
V, c, c.R, would be the easiest way of arriving at the beforementioned expression of
Q 10 in terms of these same quantities. The calculation by which this expression was
airived at., is however not without interest, and it will be as well to indicate the
mode in which it was effected.
Calculation of Q w ,
Wo liavo to find the discriminant of
o[(a, k, k t 6]J>, yf]'+Wl*atof.
Consider for a moment the more general form P 2 f 4<Q 3 , then to find the discriminant,
wo have to eliminate between the equations
.
dy dy
these aro sabiBfiod by the system P0, (2 2 = 0, and it follows that if R be the resultant
of tlio equations JP = 0, Q 0, then the discriminant in question contains the factor
Ji a . For tho other factor we may reduce the system to
pdP ( dt}^
_
da) dy dy
Now writing Q  2?a;y, these equations become
dx
, ^_ L
X dot ^ dy
tho resultant of which is = I 3 into resultant of the system
dco
dP dP _
'm~" <s dy '
15
0. VI,
114 ON THJ3 OUJUC OU11VJW INK01UKUI) IN A [400
but in virtue of bho second (Munition, wo Inivo
. , / dP dP\ ,. (IP
**(*<& + * dyrfo'dy'
which reduces tho flrnt u(iiiH.i<m lit)
a/y''/VM^wy^o,
'' (A'W (/!/
or omitting tho factor Sy, to
v' ( "'.72JW <>,
w /
rf/ J
Honco, writing P=V (o) . (, A, /:, /'$. ,'/)", and fc]wlbro ^H V (o) . (, A, ^'H ,'y)' J ,
ditto yl, tlin two (iqiiatioiiH boconio
(a, A, %, ])w
tho second of whicli IH morn hiiuply wiibtiju
Honco, ioHtoring tha factor #', and iitwi to iivditl fractimm iuUwhioing tho i'uclov Hit*,
tlio rcHultunt of Iho two oquatioiiH in
8i(tUI [HZW lo(((, h, Q>, !)(/*, A', ftj*. 1)}.
where II donotoH tho producit of tho iucitfun comtHinuiding to Iho tliroo routH ,'c,, ?,,,  a
of tlio oqiintion
(, A, ft, tjfli, l^wO,
or what is th< Raino tiling,
fc fl H Arc 1J ~Awj 6
so that tho aymmotric functions iiru In bo fontid from
Tho required disovimiimnt i the foregoing , wsnultunb inultipliod by ft, or wiy by
c 3 Ji a , that is tho discriminant Q lo i
if for shortnoHS wo write
ft(a, A, A5, I) 1 , (a, *,
400] GIVEN PENCIL OF SIX LINES. H5
and when the symmetric functions have been expressed in terms of the coefficients,
the result is to be expressed as a function of Q, V, , R by means of the values
c a D = ft a & + 3 + 4&fc"  Gadhk 
V =  Z 3 ,
c @ =  I (ab hk),
cR =  8l 3 ab.
Thus, for instance, the first term of the result is
= c a # 3 . 8#a
which ia
which is a term in the beforementioned expression for Q w
401.
A NOTATION OF THE POINTS AND LINES IN PASCAL'S
THEOREM.
[From the Quarterly Journal of Pure and Applied Mathematics, vol. ix. (1868), pp. 268
TAKING six points 1, 2, 3, 4, 5, 6 on a conic ; let A, B, 0, D, E, F, G, H, I, J, K, L> M, N,
denote each a combination of three lines, thus
12. 3*. 56 = ^
13 . 45 . 62 = J3
14. 66. 23 =
15.62.34 = _D
16. 23. 45 = #
13 . 46 . 25 = Q
14 . 52 . 36 =H
15 . 03 . 42 = I
16. 24.53 = J
12.3G.45 = tf
13 . 42 . 56 = L
14.53.62 = ,/Y
16.64.23 = ff
16.25.34 =
then any hexagon formed with the six points may be represented by a combination of
some two of the letters A, B, &c., viz, the three alternate sides are the lines repre
sented by one letter, and the other three alternate sides the lines represented by the
other letter: for example, the hexagon 123456 is AE\ and so for the other hexagons.
Any duad AE thus representing a hexagon may be termed a hexagonal duad ; the
number of such duacls is sixty. Each Pascalian line may be denoted by the symbol
of the hexagon to which it belongs; thus, the line which belongs to the hexagon
is the line AE,
I form the following combinations :
IMO.DHJ each involving all the duacls 12, &c. except those of 123 . 456,
DE&.BNO
ELM, BO J
HLff.CGI
EH.JKN
AEH.OKO
AMN.ODF
AQJ.ELO
ABI.DKL
Q'KM.BFH
124 . 356,
125 . 346,
126 . 345,
134 . 256,
135 . 246,
136 . 245,
145 . 236,
146 . 235,
156 . 234,
401] A NOTATION OF THE POINTS AND LINES IN PASCAL'S THEOREM. 117
and also the combinations :
AEQMI involving all the duads 12,13, &c,,
ASHJN
BOFIO
ODGJK
DEFRL
KLMNO
which I call respectively the tenpartite and sixpartite arrangements. It is to be
remarked that (considering IMO . DHJ as standing for the six duads IM, 10, MO
JJIf, DJ, HJ> and so for the others) the tenpartite arrangement contains all the
tiixfcy hexagonal duads : and in like manner, (considering AEGMI as standing for
tho ten duada AS, AQ, AM, AT, EG, EM, El, GM, GI, MI, and so for the others)
tho aixparliite arrangement contains all the sixty hexagonal duads.
Tho GO Pascalian lines intersect by 4's in the 45 Pascalian points p, by 3's in
20 points g and in 60 points A, and by 2's in 90 points m, 360 points r, 360 points t t
360 points 2, and 9 points w.
The intersections of the Pascalian lines thus are
i<5 p counting as 270
20 fir
60
60 A
180
90m
90
360 r
360
360 1
360
360 z
360
90 w
90
1770 = ^60.59,
and the intersections on each Pascalian line are
Bp counting as 9
1 g a 2
3 A . 6
3m &
12 r 12
12 1 12
12* ,. 12
3w ^_
59.
118 A NOTATION OF THE POINTS AND LINES IN PASCAL'S THEOREM. [401
For the tenpartite arrangement, any double triad such as ABI .DKL gives 15
intersections; 10x15 = 150; and any pair of double triads such as ABI .DKL and
AEH.GKO 'gives 36 intersections; 45x36 = 1620; and these are
(Off 60o
10 x ' J J
45 x
9m
90m
150
4 A
270 p
180 A
8r
300 r
8*
360 j
82
3602
.2*
90 w
1620
rfo
1770.
For the sixpartite arrangement any pentad such as ASHJN gives 45 intersections;
G x 45 = 270 ; and any two pentads such as A.BEJN and AEQMI give 100 inter
sections ; Ifi x 100 = 1500 ; and these are
6x
15
30 A
180 h
15m
A.K.
90m
270
' Irtf
40
18 p
24 r
60 g
270
300 r
24 1
360 1
24,0
3600
w
10f>
90 w
1500
J.UV
j_(_/u \j
1770.
I analyse the intersections of a Pascalian line, say AE, by the remaining 59
Pascalian lines as follows:
Observe that ^i? b&longs to the triad AEH, the complementary triad whereof is
GKO\ it also belongs to the pentad AEIMG. We thus obtain, corresponding to AE,
the arrangement
ESS
HA
HE
B N J
F L $
IMG
KG
401] A NOTATION OF THE POINTS AND LINES IN PASCAL^ THEOREM. 119
viz. HAENJ, is the pentad which contains HA, the arrangement of the last three
letters B Y N, J thereof beiug arbitrary ; HEFLL is the pentad that contains HE, but
lihe lust three letters are so arranged that the columns HBF, HNL, HJD are each of
bhem a triad, IMG is then the residue of the pentad AEIMG, and KOO is the
complementary triad to AEH t but the arrangement of the letters IMG, and of the
letters KOO, are each of them determinate; viz. these are such that we have BFICO,
NLMKO, JDGGK, each of them a pentad.
And this being so we derive from the arrangement
20 All, EH\
3m KG, KO, CO;
Qh AI, AM, AG; El, BM t EG;
12* IS, IF, MN, ML, GJ, GD ; HB t HP, HN, EL, HJ, HD;
Qp A'B, AN, AJ; EF, EL, ED; BF, NL, JJ);
12 r GB, GF, GJ, G3)\ OB, OF. ON, OL; K$, KL, KJ> D t
Ut FL, FJ), LD; BN } BJ, NJ; IG t 10; MX, MO; QK, GO;
3w IM, IG, MQ\
59"
via, bho line AE in question meets AH, EH each of them in a point g\ KO, KO, GO
each in a point m; and so on. By constructing in the same way an arrangement for
each of the lines AS, &c., we find the nature of the point of intersection of any
two of tho lines AS t AE, AH, &o. ; and we may then present the results in a table
(soo Plato), which shows at a glance what is the point of intersection (whether a
point 0, m, li, z, p, r, t, or w) of any two of the Pascalian lines.
I further remark that representing the 45 Pascalian points as follows :
12 . 84 = a
13.24=0
14.23=?^
15.23 = s
16.23 = 7/
12.35 = 6
13.25 = A
14 . 26  n
15.24 = *
16. 24 = #
12.36 = c
18 . 26 = i
14 . 26 = o
15 . 26 = u
16. 25 = a
12 . 45 = d
13 . 45 = j
14.35=2
15.34 = y
f fl QA ~ fi
12 . 46 = e
13. 46 = ft
14.36 = 2
15.3P
12.56=/
13.56 = 2
14 . 56 = r
15 '
23 . 45 = e
25 . 34 = X
34 . 56 = p
23 . 46 = C
25.36 = /i
35.46 = o
23.56 = 7?
25 . 46 = v
36.45 = r
24 . 35 = 6
26 . 34 =
24 . 36 = i
26 . 35 to
24 . 56 = K
26 . 45 = TT
120 A NOTATION OT? THE POINTS AND LINES IN PASCAI/S THEOREM.
the sixty hexagons and their Pascalian lines then are
AE 123456 12.45 23.56 34.61
AH
125634
12.
63
25
.34
56.
41
cXr
EH
145236
14.
23
45
.36
52.
61
mra
GK
123654
12.
65
23
.54
36.
41
feq
GO
143256
14.
25
43
.56
32.
61
npy
KO
125436
12.
43
25
.36
54.
61
ft/iS
AM
126534
12.
53
26
.34
65.
41
bffr
AG
125643
12.
64
25
.43
56.
31
G\l
AI
124365
12.
36
24
.65
43.
51
CKV
EG
132546
13.
54
32
.46
25.
61
tt
DF
126435
12.
43
26
,35
64.
51
a^
FL
124653
12.
65
24
.53
46.
31
f&k
DL
134265
13.
26
34
.65
42.
51
ipt
BN
132645
13.
64
32
.45
26.
51
keu
BJ
135426
13.
42
35
.26
54.
61
ff"*
jar
153246
15.
24
53
.46
32.
61
fay
GK
125463
12.
46
25
,63
54.
31
<w
KM
126354
12.
35
26
.54
63.
41
faq
10
152436
15.
43
52
.36
24.
61
Vfti
MO
143526
14.
52
43
.26
35.
61
fr
EM
145326
14.
32
45
.26
53.
61
mTTf)
El
154236
15,
,23
54
.36
42.
61
STZ
AN
123465
12,
,46
23
.65
34,
,51
eyv
AJ
124356
12,
35
24
.56
43.
61
fa/3
AB
126543
12.
54
26
.43
65.
31
d&
DE
154326
15,
32
54
.26
43.
61
S7T/3
EL
132456
13,
45
32
.56
24.
61
fa*
EF
123546
12.
54
23
.46
35.
61
dfr
CD
143265
14.
26
43
.65
32.
51
ops
GF
123564
12.
.56
23
.64
35.
41
.ftp
401] A NOTATION OF THE POINTS AND LINES IN PASCAL'S THEOREM. 12 L
GG
132564
13.
56
32.
64
25
.41
l&
GI
142365
14.
36
42.
65
23
.51
qxs
Mtf
146235
14.
23
46.
35
62
.51
mtru
GJ
135246
13.
24
35.
46
52
.61
ffaa
SI
136245
13.
24
36.
45
62
.51
ffru
DG
134625
13.
62
34.
25
46
.51
flue
LM
135624
13.
62
3o.
24
56
.41
Mr
FI
124635
12.
63
24.
35
46
.51
0&03
EH
136254
13.
25
36.
54
62
.41
hro
m
125364
12.
36
25.
64
53
.41
GVp
FO
125346
12.
34
25.
46
53
.61
avy
LO
134256
13.
26
34.
56
42
.61
hpz
DK
126345
12.
34
26.
45
63
.51
aw
KL
124563
12.
60
24.
63
45
.31
fv
SO
134526
13.
52
34.
26
45
.61
AfS
NO
152346
15.
34
52.
46
23
.61
wy
KG
132654
13.
65
32.
54
26
.41
leo
OJ
142356
14.
35
42.
66
23
.61
P*y
JK
124536
12,
53
24,
36
45
.61
1*8
KN
123645
12.
64
23.
45
36
.51
eeiv
DH
143625
14.
62
43.
25
36
.51
00,
HJ
142536
14.
63
42.
36
25
.61
pta
HL
136524
13.
52
36.
24
65
.41
hir
EN
146325
14.
32
46,
25
63
.51
BF
126453
12.
46
26.
53
64
.31
DJ
153426
15.
42
53.
26
34
.61
LN
132465
13.
46
32.
65
24
.51
GM
135264
13.
26
35.
64
52
.41
IM
142635
14.
63
42.
35
26
.51
GI
136425
13.
42
36.
25
64
.51
0. VI.
122 A NOTATION OF THE POINTS AND LINES IN PASCAL'S THEOREM. [401
Each Pascalian point belongs fco four different hexagons; viz. a to the hexagons
KD, KQ, FD> FO; and so for the other points, thus:
, 0) * (D
6 (A, K)(M, J) V (0, N)(J, 0)
G (A, F}(H, 7) * (#, 0)(7, L)
d (A > ff}(B,M!) (E t J)(G, 11}
e (A, J0(ff,'tf) /3 (4
/ (a, L)(K, F) 7 (
ff (B, G)(J,V) S (5
h (B,L)(H, 0) e (S,K)(0,N)
i (J),M)(G t l) r (a.^XJ 1 , G)
j (E,1C)(G,L) TJ (
k (B t L)(F,N) e (
I (A, G)(S, G) t (H, K)(J,L}
m (E, N)(H, M) K (A, G}(I, J)
n (0, M)(G, 0) \ (A, D)(G t H)
o (B,D)(0,ff) p (G, 0)(/,^)
p (Q,H)(F,.J) v (F,N)(H } 0}
ff (Q t M)(I t K) f (A, 0)(B t M)
r (A,L)(H t M) (B t ty(F, J)
s (V,E)(D,I) TT (D, M)(E,K)
t (J,L)(D,N) P (0,L)(D t 0)
u (B, M)(I, N) a (G, N)(J, M)
v (4,0 )(#,/) r (B,ty(B,I)
w (D,Jf)(ff, /O
I have constructed on a very large scale a figure of the sixty Pascalian lines, and
the fortyfive Pascalian points, marking them according to the foregoing notation; but
the figure is from its complexity, and the inconvenient way in which the points are
either crowded together or fly off to a great distance, almost unintelligible.
AB Al AO AH Al AJ JIM AN BC BF BH Bi BJ Bti BO CD Cf CC Cl CJ CK CO DE Of OC OH DJ BK DL EF
A
A
A
A
A
A
A
AN
B
B
B
B
B
B
B
C
c,
G
C
C
c
c
D
D
DJ
OK
DL
Ef
EC
N
/
EL
EM
Fl
FH
n
fO
Gl
t> P h ff & P h P P
ft ff 7l h p r t P X.%,
z x t t r T r t t 7
t % x, t r t T p z x.
h z w w t r r r r t
X ff Z X Z T i Z p p
if z h it t t r z p 2
t z t z p m m t t 7
if z w 7i % m m r r t
p h ff p 7ip 7t,p h y
r t r T r t t mm j
mTii p t z x,p r t p
t T z r p p % t r
w m % t p p z r t x
r t r l r r mm z
t r p r z z p t r 7
T T t m t r r ? r f
r r r m r t t r r t
p li r z t p z I r t
t w t p T % p r t 2
p % & h j? b p r p
p h phffhppj
hyp p_h p ft t ~*
ff J\ h p ph p x %
7i p y 7i p p h z t
p h h /j It ji fft'f
h p p h p h ff {, t
1> r p t z x r t /
p p r i z t r f h
y k p P h z h t irx.
li z t h z w I w p ff
y z x z g z % % hJi
h t x. w h t H ff p
h t t \y z ir z hp p
prrtztp&hl
r r t r T 1 nnnp x
t 7' r ? t r m m h /
j> t $ r z x t r p t
z, t z r p p t r A 1
z r Z t pp r t ff z.
P h it ji p x, r t x.
r t t z r P % r t x,
t w r t r r 7 m in x,
z Z7 p t z p i f r i
p z m r m t {, r r /
x h ?ti t m r T r r t
h P P fj 7i p n h p z
P Ji x g z z x. z z p
h p x x h. t w w L p
P /i p z z r t Irx,
h p t z w z h w t i
p p i> h t n % 7i x
7ip h p z z x % g
fyjj li. ff?> f r t x,
p h p g t \Y t H %.
p t z r t ffhpft
z if z t w g p 7i P
T z x> T f li ji a h
t Iv % t trp 7b ff p
% z ff x. x. 'h p h p
T 7/1 t. rn r t r r p t
t m, r m t r r ? h w
rpzztptrrt
T t r t r T m m z f
r z p p t z t r t r
6 ZPP r z r f ji x
1 p z z j' p r t x p
m r f 7 mi T r t r
m, t r t m r r r r t
P P t z z 7* t r t p
r t x p %. h _p Ji p //
^ it z z h j} h p h.p
t r r mm p z z p t
Z t p r t g ff z z x,
T r t m. m h. z h x w
p r z t r p z t r z
z. r p t r h z w t, h
r m, t r r h z w t, w
t, m r r r p Z t r f,
z z r r i z jy p z r
ff ( z w t w h x, r
y r z t r I z P t
t r h y hp p h J
Z z h p h j} p h ft
wt ffp p h hp x
T h hp ff p h w
w t p Tvh ff ~\7i.p t
h zp p kp h y t
Xp Ji, 7i jo h p ff tr
r t fv h z, w t t ff
t /i fi z if x if t r r
zgzffZK&xrz
%1l W X: 7l t \Y t t %,
r p tzt T zpmt
t Kw zwt h x,7n,r
z z tpp r t r wTi,
z. z T pp t r t z ff
r tmt T m r T t p
t T m T t m. r r n h
Xwh fhzwtx,t
r t ?'7tim x. p p %, t
p p t t r t't T r ?'
pp r r t p z Zp 7
f r 7n ? r t f: r r m
r t 7m r r r t tm
z h t tw% h x, lit
y z &x, zzp r tp
z t 7' p zzp r t r
xwt zhzh t wt
'PP r t z % ff ff x.
f r ff z, ff x>x, x x>#
m mh tr %,'kx, t w h
r tp x,x,t r r t ff
r7*7i;wx,wt x, fa Ji,
r r p t %, t r p x. p
t wxp p t r r t ff
' t wfat 7i,xis wfc
p Kwli, t w t X, bli,
zfrtp r t r p z,j?
X'X,X,71T} T t & 7 ;
CK
GU
HJ
Hi
HN
IM
, 10
JH
JN
% Zffzz & z %. zr
tp r t zxp tm
t wht w z?i %.7'm
wxh t 7i,t wzz
tzrprtpxrx,
'%tJitwxbtz.
% ww t ht h zt t
x t trp r zp ww
r z t r pp x,x, r
t %wt hzhzt
t pp t r w t & %,?i
m r t r r wt h z w
nt r r r t r p z t
it hw t zp t r p
P r t zp t Immr
i t wh z r rmmt
rp r_p x, r t %p r
tTi,tz,Ji,tw%,Jtst
trpzpwtwth
Y t hhx. t r t r p
vt x, t kz, kw t r
7i x, z t h t wft, % T
z_P x, r p r t z# t
t r % x, z ff ff z x, t
"rwwtK&xhw
' r t t r p %,p %, t
P t r tmt r r x,
'. h r r m m r r x,
*% t t iy X< hh Km,
'.p 7' T t Z p Z 7} 7ft.
KM
KN
KQ
LM
IN
LO
MN
MO
NO
A
t r 7 r 7ttr t t ? r
r t z r t pp x. r m
r P t r % zpp r
7nr mr r t t % t,
z. t pt r p x.mr
Prxrtzprp
r 7n r r m t t t %
zzxzzgffrr
t t r z r p z t I
r r t& t Z p z, x,
fl * A6AH Al AJAHANBOBFB
'7)i r t t zp t r t
nr r t t i r zp t
r xj3 z t 7 t r z
t P Zp % z z z &
r 7nt t. r 7' r r m
r t jo z, r r mmr
t r z p p z r t r
P t p xminr r r
X r xp r t p % r
xx,g gr t r tp
VX>% ZX,XX,ffff t
P zp r x, r t p x 7'
px r r t p %p x,rn
9&rp t r tp x, r
I t ztp tr zp x,
It t t r zp zp
zj? t x r t r zp X,
t t r Tiirr r t t r
%p r r rtmnt t p
zpmrmr r t t T
H BI BJ BH BO CD OF C6 01 <V CK CO BE OF 06 OH OJ OK DL f
V!
EG EH El EL EM Fl F H F L FO Gl QJ GKGMHJ HL HN IN 10 JK JN KL KM Ktf KO LM LN LQMNMO NO
txzrtxx
wx.
ff x
& &
% t
X w
x, t
r t
x, A.
is
r t
& h
P K,
7n m
t r
T
p P
p
T t
^ r
r t
w x.
A ff
rtx,
t
mnv TV
t r *
r t h
m i)t X
r r w
r r t
t w %,
_p 7t, t
r j> j3
t p p
r t r
m t r
>n T t
& t p
p t %,
p r z
X, r p
t T r
. r
r t
trim,
r r
r r
t r
t T
T t
r t,
m m
ff
t
t
w %,
%'X.
K p
z r
li t
p
%, & r
t w p
r t TJ
r
z ft t
% %, %
p ft x
r t g
t w g
r t %
t m, r
r m t
y h p
x. w %
ff Z %
x. &
xxr
%. t r
w t
P &
r r
r r
h, p
w t
t r
iv t
x,w
p h
t
trrz
x. p r k
h % t w
x Ji %,
wt x,
fop p
b T p
w t r
tr t
z p i,
7t, & r
\TJ fa
p fa
Jl p
p ll
t t r h
Kwi t
P li p
7i p Ti,
Z X ff _
t 7i> x w t p
I if z It x. p
r 't % %p 7i p p
& t h '  
w A it
X, t X, p W
PJ? Tip
tip r
'P r r
p p t r I r r mm p
pripx.7 t 7nrTp
fatwxktmrrh
p p T t r r p t r r
rjy r x. p r p % p m
t r p p x. 7i 1 z h p
t r 7 r r m, r r T t
T t m r r m b r r r
P r P x T t p x T
% r t xp r m>t t &
r rnbt t mr t t r
rmr t> t t r % p y
% % % ff p p t p x, t
r % r p x, r % p xm
r t p p % x, x, ff ff r
t p t Zp r r t t t
t r &%p %r %p %
r t t t p %p r
X, r t w t
ff j> h z r t
x, r t Tij} li
x t w t r t
g & % h t w
z x, t
w t r
w t %,
t r t
r
T p
X,
7i t $ x.
% w % li
p %,
t r x, r t t
r mm x % z t
ft
m r t
r b r
r x y
r r t
m t r
r p X
t _p X,
t: ZP
r r p
r m r
m m
r t
t 7'
r r
r r
ir b
t r
p Ji, T t t w
r t p h r t
b ir T t p fi
x, li jj % X, h.
p X* Z
t r r
b
%
r t w t
min, z. %. w
mm._p h t
w A x t T r p x. h X,
t X. p r T T & h %. jo
% %, x x, w t t r is r
t t r xw t r t r
h 7bp X t r r w
X' t r ff Jip niTn %, %
JbW t ff x x m>H p
wit x, x X p T 7" h %,
t p x. h % 7~ r z p
r r t t w t %. x, m m
p p P ff h
p P h P t T
p r t p hp
t t w r t r
r P & t h, x,
1 % h T x, p
r T t r t T
rm7ii_p li p
m r r t h z
m r r r %j?
p p for mm,
W t t r
t 7 r
fijp m m
% p r r
fa & r T
p h m m
r t T t
p & r T
%. h T T
fop t 7'
?ti %,1vmr 7 1
rp y,mr r
hJi xp p %
P 3tp fox, A
in, r h 73
Xp h X
A x x, p
mmp 7i>
r rp X.
9 r r
"?* 7'
~\ff
f/
T r
ff
t p X p P %>
x ff ff r t t
z p % x. z P
kpxrbt
r x. p r p %
r t t % ff ff
mttrpz
p ZP t Zp
r%pmb t,
r t t t x,p
t, t p X,
p x. p %
t ff ff
xpp
p x. t t
t ( & P
xpt t
p XX f>
y ff t t
x, P
r r t
in % p
r t r
m,p x,
T t r
r r I
t p X
t % p
%, m r
t, r p
r x, p
r % t
mz b
r % r
r z z.
7ti x. r
t ff p
t ff
t T t
x, r t
%
r m t
m ?' ?
r
t t p
t t x<
1s t
p r r
t % t
r p r
t t x
r r
t r
%.
r t
t T
p
x, p
r P
T x,
t r
X.
t r
%
in t
r r
xp
m r
7fo t
T P
T z,
? r
t
% p r r % t t r r ?n
z r r p t t x m r r
z z t t jo r r r r m,
x r p r t % IT m r
x t x t r jj r T in r
ffp p JJ X X, X t f' t
ff X & XP P P t t t
t T m. T % t x. T p r
B
E
G
H
At
AJ
AM
N
BO
BF
BH
Bt
BJ
BfJ
BO
ffj
GK
CO
Of
DF
DO
0H
DJ
OK
p x, T r X 7* T* t Tn
Is T 7' m x, %, t p r r
r tmr X, 7* p t X r'
P X. t t ff p p %, x, t
xP t t ff % X.P p t
r r in t p T % r % t
Imr r t %. % r r p
p t t X p p ff t X X'
x. t t, p x x. ff t P P
r X T p t /W 7' t & T
t x. % t r r 7n,p T r
P ff P P % & t x, % t
x y x.
%p t t
7' r x, 7 r
P t % t p ff p
%. t P t x ff
t x p t
x t p % t p
p t x. i,
pjy t
t m t
x t x
t p
ff %.
ff p
t,
h, ft, fa tih ft' w w
fa fo h w w fa h w
h w k w A w h
7i 7i A Ji
li ff
h
7i 7i w
hw h ir~Ji
h ww?i< Jt
w h h~w h
wh
EF
EG
EH
Et
EL
Fl
FH
'A
FO
8f
BJ
OK
*flf
HJ
HL
HN
/M
to
JK
KL
KM
KN
KG
LM
LN
LO
NQ
H 1 ELEMFI FH FL FO f I 8J GN OM NJ HL Hff IM iO MJNKL KMKN KO LI* LN lOMfJ MO NO
402.
ON A SINGULARITY OF SURFACES.
[From the Quarterly Journal of Pure and Applied Mathematics, vol. ix. (1868),
pp. 332338.]
A SURFACE having a nodal line has in general on this nodal line points where
the two tangent planes coincide, or as I propose to term them "pinchpoints." Thus,
if the nodal line be the curve of complete' intersection of any two surfaces P = 0,
<3 = 0, then the equation of the general surface having this curve for a nodal line is
(a, 6, eJP, Q) 2 = (where a, b, c are any functions of the coordinates), and the pinch
points are given as the intersections of the nodal line P = 0, Q = with the surface
= 0. Consider the case where the nodal curve is a curve of partial intersection
represented by the equations
P, Q>
= 0, or say by the equations > = 0,
P', Q',
r = (viz. p, q, r denote the functions QR'Q'R, RP'R'P, PtfP'Q respectively),
and consequently we have identically
(P, Q>&l!t.p, ff. )<>,
(P 1 , Q', fi'&j, * Oo,
or what is the same thing, (\ p) being arbitrary,
(XP + ^P', *Q + t*Q', \R + ?&$$, V, r).
The general surface having the curve in question for its nodal line is represented by
the equation ^, '
(a, 6, c,f, g t h$p t q, ') a = G,
(where (a, 6, o, /, g, Ji) are any functions of the coordinates), and it is easy to see that
the condition for a pinchpoint is the same as that which (considering p, q t r as
coordinates and all the other quantities as constants), 'expresses that the line
> ') = 
1 fi O
J. U ~~""tj
124 ON A SINGULARITY OF SURFACES. [402
touches the conic
(a, b } c,f, ff, /t)(p, q, r) s = 0,
viz. A, B, G f F, G, H being the inverse coefficients, 4 = &c / 2 , &c., this condition is
(A, 11, 0, F, G,
or what is the same thing, the pinchpoints are given as the common intersections
of the nodal line p~Q, 5 = 0, r=0 with each of the three surfaces
(A t S t C t F, &, J/)(P, Q,tt)* = 0,
(A, B, C, F, G, S)(P t Q, R) (F, Q', R 1 } =0,
(A, J9, 0,F, G, H) (P', Q', JS') a = 0,
these last three equations in fact, adding only a single relation to the relations
expressed by the equations
2> = Q, ,7=0, r0.
If the functions P, Q, R, P', Q', R' are linear functions of the coordinates, then
the curve (p 0, 2 = 0, r = 0) is a cubic curve in space, or skew cubic ; and if
moreover (a, b, c, /, g, h] are constants, then the equation
(a, b } c, /, g, /i}J>, q, r) a = 0,
belongs to a quartic surface having the skew cubic for a nodal line : this surface is
(it may be observed) a ruled surface, or scroll. With a view to ulterior investigations,
I propose to study the theory of the pinch points in regard to this particular surface ;
and to simplify as much as possible, I fix the coordinates as follows:
Considering the skew cubic as given, let any point on the cubic be taken for
the origin; let #=0 be the equation, of the osculating plane at 0; y = that of
any other plane through the tangent line at 0; s = t that of. any other plane through
0, not passing through the tangent line ; and w = that of a fourth plane ; then
the equation of the cubic will be
y, z
==0,
or what is the same thing, the values of p, q, T are yw Z*, zy new, and &g T/' J
respectively. And conversely, the cubic being thus represented, the point (# = 0, i/ = 0,
2 = 0) may be considered as standing for any point whatever on the skew cubic; the
osculating plane at this point being 'as = 0, and the tangent line being <w = 0, y = 0.
For the purpose of the present investigations, we may without loss of generality write
w~l; and for convenience I shall do this; the values of p, q, r thus become y 2?,
yz ES, fez i/ 3 , and the equation of the surface is
(ct, ft, c,/
402] ON A SINGULARITY OF SURFACES. 125
At a pinchpoint, we have
(A, B t G,F t G, 3) fay, if = 0,
(A t B, G,F, G,S)fay t *) (y, z, 1) =0,
(A, B, O.F, G,H) &*, 1) 9 0,
and hence the origin will be a pinchpoint if = 0, that is, if abltf Q. This however
appears more readily by remarking, that the equation of the pair of tangent planes
at the origin is
(a, 1), e t f, (J, K$y t as, 0) a = 0,
or what is the same tiling,
(a, h, &fe, #) a = 0;
the two tangent planes therefore coincide, or there is a pinchpoint, if only 6/i 3 = 0.
By what precedes, it appears that if we wish to study the form of the quartio
surface, 1, in the neighbourhood of an arbitrary point on the nodal line; 2, in the
neighbourhood of a pinchpoint; it is sufficient in the first case to consider the general
Kiir face
(, 1>> o,f t g,
in the neighbourhood of the origin; and in the second case, to study the special
surface for which ab  h* = 0, or writing for convenience a = 1, and therefore 6 = 1$, the
surface
(1, M c,f, g>
in the neighbourhood of the origin.
Considor first the surface
(a, 6, c,f, ff,
A plane through the origin is either a plane not passing through the tangent line
(a,0 = 0), and the equation ^=0 will serve to represent any such plane; or if it
pnaa through the tangent line, then it is either a nonspecial plane, which may _be
represented by the equation y0; or it is a special plane: viz. either the osculating
plane > = of the nodal line, or else one or the other of the two tangent planes
( A IVy aO a = of fche surface  I G0n8ider thei ' eft)re the secfclolis of the surlace
by these planes *  0, y = 0, w0, (a, k, bty, ^) 2 = respectively.
Section by the nonspecial plane 2 = 0,
The equation is
(a, i>,o,f,gJ>&y>~>fy^>
which represents a curve having at the origin an ordinary node the equations of the
two tangents being (a, A, Vfy. tf0, vk these are the mte^ctrons of the two
tangent planes by the plane 2 = 0.
126 ON A SINGULARITY OF SURFACES. [402
Section by the nonspecial plane through the tangent line, viz, the plane y = 0.
The equation is
(a, &. G,f, g, K$r& t a?, fl#) a =0,
or what is the same thing,
ba?  ZfaPs + Shies* + cV  2i/#3 3 + ass 1 = 0.
Writing as usual a* = ,4 z* J &c, we have
and since ub fi 2 is by hypothesis not =0, ^i has two unequal values; we have at
the origin two branches as djS 3 f B^ + &c., ,'C = .A 2 s 9 + .5 2 2 3 + &c., having the common
tangent <c=0 (viz, this is the tangent a;~0, y~ of the nodal curve), and with a
twopointic intersection of the two branches, that is, the point at the origin is an
ordinary tacnode.
Section "by the osculating plane so = 0.
The equation is
(, &, o, f, g, K$$  z\ yg t  yj = 0.
Wo may write y = z* + Az> 1 + Ssa,, we at once find ^ = 3, and then
(a, &, c f /, ;x,
that ia
(a, h t
A has two unequal values, and the "branches through the origin are
viz. the branches liave the common tangent line ?/ = (the tangent = 0, 2/ = of
the nodal curve), but in the present case a threepointie intersection.
Section by one of the tangent planes (a, h, b^y, a;) 3 0.
Writing y^tno), and therefore (a, h, &$m, 1) 9 = 0, the equation is
(a, &, c, /, (/j h^mxz' 1 , tsmztc, aw mV) f = 0,
which represents of course the projection of the section on the plane z = 0, at  0, but
which (since there ia no alteration in the singularities) may be considered as
representing the section itself. Developing, the coefficient of a? is am 3 + 2Am + b, which
is e= 0, and the equation becomes
a + cm 4
+ 2 [Am" + (6  ff) m /) a^ I ' 2m 9 (/w  c) _
+ 2{om + ft) CM? + (6 + 2^) m 3  2/m +
+ 2 (1m  g) ax 8
+ a ,3 4 = 0,
so that the curve has at the origin a triple point, the tangent to one branch being
the line = (the tangent a? == 0, y = of the nodal curve),
402] ON A SINGULARITY OF SURFACES. 127
Consider next the surface
(1, A 9 , c,f, ff, A&*, $**> awJ^0,
being as already remarked, the general surface referred to a pinchpoint as origin.
Section by the nonspecial plane z 0.
Tho equation is
(l./^c./tf./^,*, 2/70,
whore, attending only to the terms of the ^lowest order, we find (1, h, A'Jff, .) a = 0,
that is (y  /(A') a = > showing that the origin is a cusp.
Section by the nonspecial plane through the tangent line, viz. tho plane
Tho equation is
(1, h\ Otf.g.Ki** , awyO,
or what is the same thing,
= 0)
that i
, .
(ha + 2 a ) a  2> 2 s + ca;V  S^a'S 3 = 0,
writing to *. + ^. find at onoo , = i and then ,1' = , *at tho
branch* a,e to^i^i whence we have at the origin a c U3 p of the second
ordov or node cusp.
Section by the osculating plane #=0.
The equation is . n .
(I, h* t o t f,ff,K$#*. **>& = *>
siting y^W + ^ ^ easily find ?*, and then
* A ** diaannear of themselves, the terms in * give
whore the terms in *, and ^ +4 diaappeai
jl" + 2jy/i = 0, and the branches are
via. there is a cusp of a superior order.
Section by the tangent plane y~ha.
Tho equation is
12 g
ON A SINGULARITY OF SURFACES.
[402
representing the projection on the plane of m. Developing, the equation is
2 (/ gh) a? (Me z)
+ 2* 1 =0,
and there is at the origin a triple point (= cusp + 2 nodes) arising from the passage
of an ordinary branch through a cusp; the tangent at the cusp being it will be
noticed the line = 0, that is the tangent = 0, y=Q to the nodal curve at the
pinchpoint.
The results of the investigation may be presented in a tabular form as follows :
Nature of Section.
Plane of Section.
Nonspecial .
Ditto, through tangent line
of nodal curve.
Osculating plane of nodnl curve.
Either of the two tangent
planes.
The single tangent plane.
Origin, an ordinary point.
Node,
Tacnode = 2 nodes,
Triple point, one branch touch
ing the tangent of nodal line.
Origin, a Pinchpoint.
Ousp.
Nodecusp, = node + cusp.
Triple point, = cusp + 2 nodes ;
the cuspidal branch touching
the tangent of the nodal line.
I have not considered the special cases where one of the two tangent planes, or (as
the case may be) the single tangent plane of the surface coincides with the osculating
piano of the nodal curve.
403.
ON PASCAL'S THEOREM.
[From the Quarterly Journal of Pure and Applied Mathematics, vol. ix. (1868),
pp, 348353.]
I CONSIDER the following question : to find a point such that its polar plane in
regard to a given system of three planes is the same as its polar plane iu regard to
another given system of three planes.
The equations of any six planes whatever may be taken to be A r 0, Y=0, J?=0,
U^O, F=0, F=0, where
X+ Y+ Z+ U+ V+ 17=0,
and so also any quantities X, Y } Z, U t V, W satisfying these relations may be regarded
as the coordinates of a point in space ; we pass to the ordinary system of qnadriplauar
coordinates by merely substituting for 7, W their values as linear functions of
X, Y, Z } U.
This being so, the equations of the given, systems of throe planes may be taken
to be
and if we take for the coordinates of the required point (, y t z, u, v, to), where
0+ y+ + + + w = 0j
aw 4 ly + GZ +j% f gv + hw = 0,
then the equations of the two polar planes are
, ,
a? y z u v w
o. vi. 17
130 ON PASCAL'S THBOHEM.
[403
respectively, and we have to find (a, y, ,, u> v> w ) t 8uch thftt these fcwo e tions
raay represent the same plane, or that the two equations may in virtue of the linear
relations between (X, F, Z, U> V, W) be the same equation.
The ordinary process by indeterminate multipliers gives
1
 + X + (j,c = 0,
*nd we have the beforementioned linear relations between fo y . ,, u> v> w} . these
last are satisfied by the values
if only
in fact, ^ satisfying this equation, the relation
+ # + # + w
is obviously satisfied; and observing that we have
.
ft atf* " > f6
we have J
+fti{ffv + hw
BO that the relation + ff , + *.0 is also .atisBed. Substituting the
the six equatiom oontahiing 7c  x  * wm be ai1
403] ON PASCAL'S THEOBEM. 131
The coordinates of the required point thus are
1 1
1 c0' /(?' <j6> liQ
where
 4
i n T
a 6 ' 6  c  f6 g6 h
and, the equation in being of the fourth order, there are thus four points, say the
points Oi, S , O s , 0,,, which have each of them the property in question,
It will be convenient to designate the planes X =Q, Y~Q, j?=0, U=Q, V~=Q,
W = Q as the planes a, b, c, /, g, h respectively; the line of intersection of the planes
X = 0, F=0 will then he the line ab, and the point of intersection of the planes
X = 0, Y~0, Z=Q the point abc\ and so in other cases.
I say that from any one of the points it is possible to draw
a line meeting the lines af.bg. ck (1),
ag . bh . of (2),
ah , bf . G(j ' (3),
af.bh.cg (4),
aff.bf.ch (o),
ak.bg .of (6),
and consequently, that the finir points are the four common points of the six
hyperboloicls passing through these triads of lines respectively.
In fact, considering d as determined by the foregoing quartic equation, and writing
for shortness
(c6)Z = 0, (h6)W=H,
so that
the equations A+J?~Q, B+G = Q, (7 + JJ=0, are equivalent to two eqiiations only, and
it is at once seen, that these are in fact the equations of a line through the point
meeting the throe lines af, bg, oh. respectively,
1 1
The equation A + F 0, is in fact satisfied by the values X : U=  ^'; j 7.,
(t O J C7
and by X Q, U=0; it is consequently the equation of the plane through and
line (if' ' " _i_ /3 (I Jo tlio omicifinn fiF +.ITO rilnnw f.lirniin'Vi O n.TiH f.lid
132 ON PASCAL'S THEOREM. [403
bg; and 0+B"=0 is the equation of the plane through and the line e/t; and the
three equations being equivalent to two equations only, the planes have a common
line which is the line in question.
The equations of the six lines thus are :
(1) ,
(2) ,
(3) .
(4) A+F =0,
(6) A + G=Q,
It is further to be noticed, that if in any one of these systems, for instance in tho
system A + F Q, B + G = 0, G + ff==0, we consider as an arbitrary quantity, then
the equations are those of any line whatever cutting the lines of, &</, c/t; and hence
eliminating 9, we have the equation of the hyperboloid through the three lines
of, fy, ch ; the equations of the six hyperboloids are thus found to be
ax +fu _ by + yv _ cz + hw
V / _LI W + fl 2 + W '
ass + gv _ by + hw __ 02 +/H
/0 v cw + fey % +>
W
( . . cfl! +/w _ byjhw _ cz + gv
^ J "+ir ~ 'w+w ~ ~^^
,_, ax + ffv by+fy c^ + /wy
(o) = =
" '  y +u
/f\ wr _ by + gv _ c^+fu
w/ ~~
respectively; the equations in the same line being of course equivalent to a single
equation,
For each one of the six lines we have
(A,B t 0)(.F, ff, H)
in some order or other, and it is thus seen that the six lines lie on a cone of the
second order, the equation whereof is
X 2 + S 3 + ^  .F 3  G B  H 3 = 0,
403] ON PASCAL'S THEOREM. 133
Consider now the six planes a, b, c, f, g, h, and taking in the first instance an
arbitrary point of projection, and a plane of projection which is also arbitrary the
Hue of intersection ab of the planes a and b will be projected into a line ah', and
the point of intersection of the planes a, b, c into a point abc; and so in other cases.
\Ve have thus a plane figure, consisting of the fifteen lines ab, ac, ...gh, and of the
twenty points aba, abf, ...fgh\ and which is such, that on each of the lines there lie
four of the points, and through each of the points there pass three of the lines, viz.
the points abc, abf, abg, abh lie on the line ab ; and the lines bo, ca t ab meet in the
point abo, and so in other cases, If now the point of projection instead of being
arbitrary, be one of the abovementioned four points 0, then the projections of the
lines of, by, ch meet in a point, and the like for each of the six triads of lines;
that is in the plane figure we have six points 1, 2, 3, 4s, 5, 6. each of them the
intersection of three lines as shown in the diagram,
l = af.bg,ch,
Zag.bh. c/,
3 = all . bf , eg,
4 = af . bh . eg,
5 = ay bf . ch,
6 = ah . bcf . of,
and these six points lie in a conic, It is clear that the lines af t ag t ah; bf, bg, bk;
cf, G(j, ch are the lines 14, 25, 36; 35, 16, 24; 26, 34, 15 respectively.
Conversely, starting from the points 1, 2, 3, 4, 5, 6 on a conic, and denoting the
lines 14, 25, 36; 35, 16, 24; 26, 34, 15 (being, it may be noticed, the sides and
diagonals of the hexagon 162435) in the manner just referred to, then it is possible
to complete the figure of the fifteen lines ab, ac,...gh and of the twenty points
#fco, abf,...fffh, such that each line contains upon it four points, and that through each
point there pass three lines, in the manner already mentioned,
Of the fifteen lines, nine, viz. the lines af> ag, ah ; bf, bg, bh ; of, eg, ch are, as has
been soon, lines through two of the six points 1, 2, 3, 4, 5, 6 ; the remaining lines are
bo, ca, ab ; gh, hf, fg. These are Pascalian lines,
bo of the hexagon 162435,
ca 152634,
ab 142536,
gh 162436,
hf 142635,
bg * 162534,
134 ON PASCAL'S THEOREM. [403
which appears thus, viz.
line 6c contains points bcf , Icy , Ich ,
= bf . cf t if/ . c(/ t l)k .ch,
= 35.2G, 16.34, 24.15;
s
that is, l)c is the Pascalian lino of the hexagon 162435 ; and the like for the rest of
the six lines.
The twenty points ale, abf, . . .fyh are as follows, viz. omitting the two points
abo, fgh, the remaining eighteen points are the ' Pascalian points (the intersections of
pairs of lines each through two of the points 1, 2, 3, 4, 5, 6) which lie on the
Pascalian lines fie, ca, ab, gh } Jif, fy respectively ; the point ale is the intersection of
the Pnscalian lines be, ca, ab, and the point fgh is the intersection of the Pascalian
lines f/h, hf, fg, the points in question being two of the points P (Steiner's twenty
points, each the intersection of three Pascalian lines).
"We thus see that \ve have two triads of hexagons such that the Pascalian HUGH
of each triad meet in a point, and that the two points so obtained, together with the
eighteen points on the six Pascalian lines, form a system of twenty points lying four
together on fifteen lines, and which points and lines are the projections of the points
and lines of intersection of six planes ; or, say simply that the figure is the projection
of the figure of six planes.
It is to be added, that if the planes are a, b, c> f, g, h, then the point of
projection is any one of the four points which have the same polar plane in regard
to the system of the planes , 6, c, and in regard to the system of the planes / g, h,
The consideration of the solid figure affords a demonstration of the existence as woll
of the six Pascalian lines as of the two points each the intersection of three of
these lines.
404.
REPRODUCTION OF EULER'S MEMOIR OF 1758 ON THE
ROTATION OF A SOLID BODY.
[From the Quarterly Journal of Pure and Applied Mathematics, vol. ix. (1808),
pp. 361373.]
EULER'S Memoir "Du mouvement de rotation des corps aolides autour d'nn axe
variable," Mem. de Berlin, 1758, pp. 154193 (printed in 1765), seems to have been
written subsequently to the memoir with a similar title in the Berlin Memoirs for
1760, and to the "Theoria Motua Oorporum Solidorum &c.," Rostock, 1766, and there
are contained in the firstmentioned memoir some very interesting results which appear
to have escaped the notice of later writers on the subject; via, Euler succeeds in
integrating the equations of motion without the assistance furnished by the consideration
of the invariable pla/ne. In reproducing these results I make the following alterations
in Euler's notation, viz. instead of cc, y, 2 I write p> q, r\ instead of Ma*, Mb*, Mo 2
(where M is the mass) I write .A, B, 0, these quantities denoting the principal
moments, and in some equations where the omission or j., ' " ' " ar '
vnallir immofnviol T nnlfn A R H i f.lin nlnnn /.
136 REPRODUCTION OF EULEB's MEMOIR OF 1758 [404
and introducing the auxiliary quantity u such that du = pqrdt, we have
$ = +
r" = (+2
where 51, 3i, < are constants of integration, and thence
f=
V {(51
where bite integral may without loss of generality be taken from w=0; u, and
consequently jp, g, r, are thus given functions of t\ and it is moreover clear that
31, 93, ( are the initial values of p", g s . T*. We have also if w be the angular
velocity round the instantaneous axis
Euler then assumes that the position in space of the principal axes is geometrically
determined as follows, viz, (treating the axes as points on a sphere) it is assumed
that the distances from a fixed point P of the sphere are respectively I, in, n, and that
the inclinations of these distances to a fixed arc PQ are respectively X, p, v. We have
then the geometrical relations
cos 3 1 f cos a TO + cos a n = 1 ;
* f . cos I , , cos in cos ?i
sin (u, y ) = . : . cos ( u, v ) = : : .
sramaintt ^ ' smmsinw*
'  """*,, oos( ! ,X) =  co8 ' !cosZ
Bin n Bine x ' smresin
sin I sin m r sin I sin m '
whence also
__ cos X cos n sin X cos I cos m
Sin ill i J  ;
sin I sin m
= s !?^ CQS T1 ~ GOS ^ cos ^ c _ s w
sin j sin m '
cos X cos m + sin X cos I cos n
sn v =
sin I sin t
am X cos m cos X coa I cos ?t
CQSV =
sm i sin n
404] ON THE BOTATION OT A SOLID BODY. 137
The geometrical equations connecting the resolved angular velocities jy, q, r with
the differentials of I, m, n, X, p, v are
dl sin I =dt(q cos n  r cos m), d\ sin 3 1 = dt(q cos m + r cos n ),
dm sin m = dt (r cos ( ^cos?fc), dp sin 3 m = dt (r cosn 4jtf cos I ),
rf sin n = fZ( (p cos m q cos 2 ), rfy sin 3 n eft (p cos f 5 cos m).
Multiplying the equations of motion respectively by cos I, cos m, cos 11, and adding,
we obtain an equation which is reducible to the form
d (Ap cos I + Bq cos m + Or cos n) 0,
whence integrating
Ap cos 4 Bq cos m + (7r cos n = 2),
2) being a constant of integration. One other integral equation is necessary for the
determination of the angles I, in, n. The expressions for dl, dm, dn give at once
p dl sin I + q dm sin m + r dn sin n = 0.
Instead of the arcs I, m t n, Euler introduces a new variable , such that
v = p cos I 4 q cos in + r cos n ;
by means of the last preceding equation, we find
civ = dp cos I + dq cos m + dr cos n,
and then, substituting for dp, <fy, d? 1 , their values,
(L cos 2 Jf cos m. Ncasn\ ,
......   .......... ' 4* I C"">
ff
from which the relation between u and , is to be determined. "Wo have
cos a I + cos 3 m + cos 3 n 1,
jfljp cos + Sq cos wt + C 1 ? cos n = 3),
p cos 1 4 # cos m + 7' cos w v,
which give cos I, coam, oosw in terras of u, v t the resulting formula* contain th&
radical
)
5y 4 6V)J J
which for shortness is represented by VlOOl We bnen havo
* ~
COS t to'Jaa j,2 i JMi
4 CMoti (ifp  Zr B ) +
C. VI.
cos m  = I?A*<?i* 4 M*B*i*p ^'N^G^f
(MBf  LA.(f) + XBrti (g a  Jtfp 1 ) + ^Vgpg V{()]
18
138 . REPRODUCTION OF BULER'S MEMOIR OF 1758 [4Q4
and substituting these values in the differential equation
~  ^ CQS ^ , AfjMam Ncosn
du p q + r
the equation to be integrated becomes
Now substituting for p, % r their values, we h
ve
and writing for shortness
9[+
= K,
where K =
substituting these' values and observing that
the radical of the formula becomes
Vl()l = V (ST 2iJf J^ffitt + SiD'iJIf jYw  3>ff + 23) JV 
and the differential equation becomes
+(0)
which can be reduced to the form
Kdv   CTJ n r ^^T,
404] ON THE ROTATION OP A SOLID BODY. 139
Euler remarks thai) as the righthand aide of the equation contains only the
variable u, the solution will he effected if we can find a function of ,, a multiplier
of the lefthand side; he had elsewhere explained the. method of" finding such
multipliers, and applying it to the equation in hand, the multiplier of the lefthand
side, and therefore of the equation itself, is found to be J^T^lWWGu' V wlmt 1S
the aame thing
Multiplying by this quantity, the righthand side may for shortness be represented
by dU, so that
(H  iLMNFu} </ (ff) dit. ___
dU ~ (K2LMNGu).J {(ZLu + SI) (2Jlf + S3) (Sflfo + )} '
and U may he considered as a given function of u, or what ia the same thing of t.
As regards the lefthand side, attending to the equation K = E& F*, the radical
multiplied into V (<?) may he presented under the form
and consequently the lefthand side becomes
(K  ZLMNGhi) Gdv + LMNG (Gfv  3XF)
(7f  iJKffff tt) V{(<?  ^ a ) (^'  ZLMNQu)  (tfv  WJf'fl '
which putting for the moment K*LMNG~p> Gv$>Fq, ffS) 3 / 2 , becomes
_J$$lJty. the integral of which is am 1 !; henco resfcoring. the values of p, q, /,
p*j(f*p*<?y w
the integral is
.
sm
Hence considering the constant of integration as included in U, or writing
(H ZLMNFit) V (ff) rf
we
have for the required integral of the differential equation
sm 7(e^
whence also
TW^'
and
~~ ~" " =cos U,
182
140 HEPKODUCTION OF EULER's MEMOIR OF 1758 [404
so that the value of the original radical is
, ,, .. ^(QVWKZZLMNGU)}
V i(    coa (/.
Substituting iu the expressions for the cosines of the area I, m, n, these values of t)
and the radical ; the formulas after some reductions become
sm +
ff ~&7(I{  ZLMNGu)
.
ff ff V (K  2IJIT jy fft*) V (ff) V
   _ ___ _  ___ . . _.. ___ .
ff ff V (tf  2MjVff tt ) V (ff) V (JST  ZLMNGu) J
where for shortness _p, (jr, r are retained in place of their values V (2Z + 9t)' V (2Jlf  93),
The values of ^, ?ji, being known, that of \ could be determined by tho
differential equation
7x (U(qcos in + z cos n)
IvA* "" """  ; r^5   ,
sin" I '
and then the values of /*, i/ would be determined without any further integration;
but it is better to consider, in the place of any one of the principal axes in particular,
the instantaneous axis, which is a line inclined to these at angles , (3, y, tho cosines of
33 Cf 1*
which are, i,  (if as before o> a = i > + g' + j*). Considering the instantaneous axis
as a point of the sphere, let j denote the distance OP from the fixed point P, and
tf> the inclination OPQ of this distance to the fixed arc PQ, We have
cos j = cos a cos I f cos cos m f cos 7 cos }
sin j cos {/> = cos a sin 2 cos X + cos /3 sin m cos /* + cos 7 sin ^ cos y,
sin j sin = cos a sin I sin X + cos /3 sin m sin /4 + cos 7 sin n sin y,
a { A _ JA _ COS 7 COS TO  COS ff COS 7^
V W
_ _
smisinj ' V W si n ; sin j
cos 0*  <) = CQ3 ^ CQSmcos ?' cos f tt  ^  cos cos K. cos .y cos ^
^ W sinmsinj > cos (/^</0  sinmsin / 
cos (v  A) = ""rCMftcoaj cog , , , cos ff cos I  cos a cos m
sinwsmj ' ^ ^ sih 7i sin j  '
404] ON THE ROTATION OF A SOLID BODY. 141
so that X, /i, v are determined in terras of j and (ft. These expressions give
dcf} = a . 3 . {cos I (qdr  rdq) + cos w, (rdp pdr) + cos n (pdq  qdp}},
7 sm 3
which is reducible to Euler's equation
p (jlf <  JV33) cos I + q (Mi  m) cos m + r (i93  Jl/21) cos
and thence, substituting for cos I, cosm, cosra their values, and observing that
Af (JlfE  #33) + #2 a (OT  ZS) + C? 3 (Z93  JlfSI) =>(H
BGp* (m  N$? + 04Q'
LA (JlfgJVffl) +MB
the equation becomes
7/
 1* 4 dt =
where it is to be remarked that
3
 ( ff  5) a ) (If  ZLAfNQu) sin
Now
the differential d$ can be expressed as a fraction, the numerator whereof is
sin U
<u)
and the denominator
(ff  2) a ) ^ a + g (/f  ZLMNQu)  22)^ V (<?  > a ) (^  ZLMNGu} sin Z7
 (g  2) a ) (/^  ZlMNQu) sin 3
To simplify, write
> 9 ) = A,
142 REPRODUCTION OF EULER's MEMOIR OF 1758 [4Q4
the numerator is
and the denominator
W + Gs  23X?fts sin U ~ AV siii a
which, observing that /t^ff3) 3 , is equal to
(*7i  35s sin U? + <?s a cos 3 U,
and we have
the integral of which i
s
where S is the constant of integration, or substituting fo r A 8 their values, the
equation is
tan (A + g W ^V (g  3> 3 ) ~ S> sin g V (JC
cos Oy iff (K%LMNGu)}
It way be added that
cosj = v = _ _ u sn
and therefore
r>f\<3 n *~^ LV "^ } \ ~ j&i/jriii Crt'jf sin (_/
UUd / " j ~~^ ti. ,^ ,, /j
Euler remarks that the complexity of the solution owing to the circumstance that
tae nxed point P is left arbitrary; and that the formulas may be simplified by taking
this point so that tfSO, and he gives the far more simple formula corresponding
to his assumption; this i s in fact taking the point P in the direction of the normal
TO Miff inVfintt'" mlrunji ,>J J.1, _ li n , .... J
pane, and the resulting formula are identical with the ordinary
for the solution of the problem. The term invariable piano is not used by
or' T;M 5"? ^ OOaoatA La S Ean * rtl " Essai "" le P^leme do trois
corps, Pros de I'Aoad. de Berlin, t h ix., 1772.
To prove the beforementioned equation for (fy; stai'ting from the equations
COS j = COS CE COS / + COS /? COS m h COS ry COS 71 =  ,
a
sinj cos tj> = cos a sin i cos X + coa j3 sin w cos /* + cos 7 siu cos v
sin j sin < = cos a sin i cos X + cos sin m sin ^ + sin 7 sin n sin v,
we have
cos j cZj cos $  sin/ sin
404] ON TEE ROTATION OF A SOLID BODY. 143
the second term is
cos \ cot I (a cos n r cos in}
to
cos
cos fj, cot m (r cos I p cos n )
to
4 cos v cot 11 (p cos m q cos I ),
to
and the third term is
+ sin X cosec I (a cos i 4 r cos n )
a
j sin ft, cosec wt (r cos n +p cos )
to
4  sin i> cosec n (p cos 2 f q cos TO),
to
Hence the second and third terms together are
pa f , cosZcos?i coswioosw . cosm, . ooa 3 ^
^ COS X  ^i  COS U,  ;  + Bin X r T + Sill /* r I + <W!.,
co \ sin 2 sin w sin f B
_pq ( cos X sin n cos (y  X) + sin X sin n sin (y  X) ^
~ "w + cos ^ sin n cos (/A v) + sin p sin ?t sin (/*  c)J
7 . f cos X cos (y X) + sin X sin (v X)
= L ^ am ft ^ , C , .
w [+ cos /* cos (/t v) + si
cos{X + (y X)}
** sin n ( cos v + cos c) + &c,, = ;
Q)
we have therefore
cosj dj cos $ sin j sin < (2$
=  sin a da sin 2 cos X  sin /3 d{3 sin i cos /t  sin 7 dy sin n cos y,
q . . r . .
= d . sin 2 cos X 4 d  . sin ?n, cos w 4 ft . sin n sin w
to a> w
1
= 4. _ ( s i n l C os X o5p + sin m oos /4 dg 4 sin 71 cos y ar)
_ ^ (ain2oo8\p + sinmooa/4g' + sinwcosyr)
144 REPRODUCTION OP EULER's MEMOIR OF 1758 [404
Hence therefore
sin j sin Add =  cot j cos Ad ~
a
~
(sin I cos X dp + sin in cos /* dq + sin n cos v o
ittu / " 7 \ I *
t ^ t^sin ( cos A, i + sin m cos pq + siun cos y r)
, I/
=  cotj cos (f>  (cos I dp + cos m dq + cos n dr)
dto
+ co\,j cos ^ j (p cosl + q cos m + r cos n)
  (sin / cos X dp + sin m cos ^ dq + sin n cos v di
dto , ,
+ ^ (sin i cos X , ^j f sm m cos p , q
+ ^ K eot J cos ^ cos ^ + sin Z cos X)p + &c.J.
But we have
.
sm Z sm j *
cos a
 COt ?,
sm t smj J)
siu (X  4) = cos Y cos m  cos ff cos n
sin? sin j
and thence
cos \ = cos ((X  0) + 0)  C03 ^ ^ cos g ~ cos ^ C08 ?') ~ "P ^ (oos 7 cos m  cos ^ cos n)
sin { sin j
whence also
cot j cos $ cos Z + sin Z cos X
= siSj f cos * cos ^ + cos ( cos  cos ; cos j)  sin A (cos 7 cos m  cos /3 cos n)},
sln7 l cos a cos ^ ~ sm (cos 7 cos m  cos cos )},
GOS ^ ~ siu ^ ( r cos m  ? cos '01
404] ON THE ROTATION 7 OF A SOLID BODY.
Hence the expression for smjsin0d< is
=   r . fi ?J cos <i sin <A (r cos m w cos n)} dp + ,. .1
l r
+ j . Ho co8d> sin A (? cos m  <; cos n)l + ,..]
' LU r
 __; . [a) do) coa </> sin {(?' cos m </ cos H) rfjj + ...}]
. <u 2 cos (i = sin j sn
Bin j r
or finally
1 , 1 Slllrf) r , . , . n I
di = ~v ^ [(' cos w  g cosii) dp + *c.J,
r a L J 
that is
sin i sin
J (w smj
1 C (r cos t gr cos ?t)
w sin"j I + (p G os n  r cos 1} dq
[ + (f? cos Z > cos wi) di 1 J
which is the required expression for d<f).
Recapitulating, A, B, G, p> q, r denote as usual,
L = G , M = , N = > dupqrdt,
r = ^ (( +
SI i 23 +
su
so that
C, VI.
146 REPRODUCTION OF EULEJl's MEMOIR OJT 1758 &0. [404
sm
G
COS ft = __
" cos
sm
cos ? = sin U
_p cos ^ + ^ cos in + r cos n
ZLMNGu)} sin [7],
tan (d, + $) =
le angles which _ determine the position of the body are thus expressed in
)l w, which is given as a function of t hy the fort
where p, q, r denote given functions of w.1
405.
AN EIGHTH' MEMOIR ON QUANTICS.
[From the Philosophical Transactions of the Royal Society of London, vol. GLVII.
(for the year 1867). Received January 8, Bead January 17, 1867.]
THE present Memoir relates mainly to the binary qnintic, continuing the inves
gations in relation to this form contained in my Second, Third, and Fifth Memoirs
Qualities, [141], [144], [156]; the investigations which it contains in relation to
qu antic of any order are given with a view to their application to the quintic. j
the invariants of a binary quintic (via. those of the degrees 4, 8, 12, and 18) i
given in the Memoirs above referred to, and also the covariants up to the degree
it was interesting to proceed one step further, viz. to the covariants of the degree
in fact, while for the degree 5 we obtain 3 covariants and a single ayaygy, for t
degree 6 wo obtain only 2 covariants, but as many as 7 syzygies; one of these
however, the syaygy of the degree 6 multiplied into the quintic itself, so th
excluding this derived syssygy, there remain (7  1 =) 6 syzygies of the degree 6. T
determination, of the two covariants (Tables 83 and 84 post] and of the syzygies
the degree 6, occupies the commencement of the present Memoir. [These eovariai
41,,, ,,:* M nn A iff nf 4V.rt r,nnm. 1 J.9 " TnWnc nf n  : " L
148 AN EIGHTH MEMOIR ON QUANTICS. [405
I then apply it to the quintio equation, following Professor Sylvester's track, but so
as to dispense altogether with his amphigeuous surface, and making the investigation to
depend solely on the discussion of the bioora curve, which is a principal section of
this surface. I explain the new form which M, Hermite has given to the Tschirn
hausen transformation, leading to a transformed equation the coefficients whereof are
all invariants; and, in the case of the quintic, I identify with my Tables his cubi
covariants f (<r, y) and fa(x, y). And in the two new Tables, 85 and 86, I give the
leading coefficients of the other two cubicovariants fa (at, y) and fafa, y), [these are
now also, identified with my Tables]. In the transformed equation the second term (or
that in ^) vanishes, and the coefficient SI of & is obtained as a quadric function of
four indeterminate* The discussion of this form led to criteria for the character of a
quintic equation, expressed like those of Professor Sylvester in terras of invariants,
but of a different and less simple form; two such sets of criteria are obtained, and
the identification of these, and of a third set resulting from a separate investigation,
with the criteria of Professor Sylvester, is a point made out in the present memoir.
The theory is also given of the canonical form which is the mechanism by which
M. Hermite's investigations were carried on. The Memoir contains other investigations
and formula in relation to the binary quintic; and as part of the foregoing theory of
the determination of the character of an equation, I was led to consider the question
of the imaginary linear transformations which give rise to a real equation; this is
discussed in the concluding articles of the memoir, arid in an Annex I have 'given a
somewhat singular analytical theorem arising thereout.
The paragraphs and Tables are numbered consecutively with those of my former
Memoirs on Qualities. I notice that in the Second Memoir, p. 126 we should have
No. 26 = (No. 19)' 128 (No. 25), vk the coefficient of the last term is 128 instead
of 1152. [This correction is made in the present reprint, 141, where the equation
is given in the form Q f = G*  128Q.]
Article Nos, 251 to 25*. The Binary Qmntio, Govariants and Syzygies of the degree 6.
251. The number of asyzygetic covariants of any degree is obtained as in my
Second Memoir on Qualities, Philosophical Tmnsactiom, vol. CXLVI. (1856), pp. 101126
[141], viz, by developing the function ' ' '
_
(1 *)(! aa) (1  o&) (1  a?z) (1  ate) (1 _ aft) '
as shown p. 114, and then subtracting from each coefficient that which immediately
precedes it; or, what is the same thing, by developing the function
(1 *)(! aw) (l^)(lai85)(l
which would lead directly to the second of the two Tables which are there given:
the Table is there calculated only up to *, but I have since continued it up to *
so as to show the number of the asyzygetic covariants of every order in the variables
up to the degree _ 18 in the coefficients, being the degree of the skew invariant, the
highest of the irreducible invariants of the quintic. The Table is, for greater
convenience, arranged in a different form, as follows :
405]
AN EIGHTH MEMOIR ON QUANTICS.
149
Table No. 81.
Table for the number of tho Asyzygetic Covariants
of any order,
la 13 u
to the degree IS.
ID 10 17 18
o
i
1)
1
2
3
1
1
o
1
u
1
1
1
2
3
3
4
i
3
1
(}
2
1
4
g.
u
i
it
2
3
1
1
2
3
1
5
7
S
3
4
2
1
4
3
7
t!
11
l>
4
5
1
1
2
3
5
11
S
10
13
5
6
1
1
4
a
7
7
2
11
18
6
j
1
a
4
G
8
10
13
10
7
9
10
11
12
'3
14
15
16
'7
iB
'9
no
21
J3
25
26
27
28
ay
3
1
1
1
1
2
1
2
1
1
I
3
2
2
a
2
1
1
1
2
4
3
2
3
2
2
1
1
1
4
G
4
G
4
4
3
8
2
2
1
1
u
a
7
G
7
5
11
G
3
3
7
7
S
8
8
7
8
U
5
G
II
10
8
11
11
1)
10
8
I)
U
7
U
11
11
13
is
13
12
12
11
11
11
!)
11
10
15
13
1G
14
17
11
10
13
14
12
13
14
10
17
10
13
18
19
17
17
15
LI
17
111
21
19
23
SI
24
21
20
22
111
10
20
24
24
26
20
27
2(1
27
25
25
18
18
24
2S
27
32
20
33
30
29
20
23
27
2U
32
33
30
35
37
SO
87
36
18
211
23
13
J2
39
37
42
JO
4t
10
41
3
9
10
ii
'3
'4
'5
16
'7
18
'9
20
11
=2
23
24
35
26
28
29
3
3'
33
34
35
30
37
38
39
40
A
1
2
1
1
1
3
2
2
1
1
5
3
3
2
7
7
5
5
3
12
10
7
7
15
13
12
10
10
18
19
10
10
13
23
20
20
17
31
27
28
24
25
35
30
20
40
42
33
30
35
32
33
34
35
36
37
38
39
40
4'
1"
13
44
i
2
1
3
2
G
G
7
7
13
10
10
1t
31
21
29
28
31
42
43
41
45
46
4,
1
2
*
5
10
13
17
23
31
47
4 7,
48
1
3
5
7
11)
17
21
27
48
49
49
.
8
7
14
27
50
So
I
S
10
18
5'
5 1
2
5
13
22
52
53
o
3
7
17
53
S3
L
6
1ft
22
54
54
1
2
7
14
55
55
1
3
10
18
56
2
5
iy
57
57
3
7
17
SB
58
i
B
10
59
59
1
2
7
14
fa
60
1
3
10
61
61
2
S
13
f,*
6a
3
7
63
63
1
5
10
64
64
1
2
7
65
65
1
8
10
66
66
2
I!
67
67
3
7
68
68
1
5
69
<x>
1
2
7
70
70
1
3
7'
7 1
2
5
72
73
3
73
73
I
G
74
74
1
75
75
1
3
76
76
2
77,
77
1
3
78
79
79
So
1
1
2
So
8t
81
a
83
82
83
83
i
84
84
1
i
"5,
Bt
87
87
8S
Bt
89
i
89
90
90
150 AN EIGHTH MEMOIR ON QUANTICS. [405
[In regard to this table No. 81 it is hardly necessary to notice that for any
column with an even heading the numbers of the column correspond to the even
outside numbers, while for any column with an odd heading the numbers of the
column correspond to the odd outside numbers. The table is in fact a table of the
differences of the numbers of the o/table, 142 ; thus in this table writing down
cols. 5 and 6 and in each of them forming the differences by subtracting from each
number the number immediately below it, we have cols. 5 and 6 of the table No. 81, viz, :
1 1 2 3 6 7 9 11 U 16 18 19 20 col. 5, 1312 of /tnblo.
1011222 2 3~3 2 1 T col. 6 of table No. 81.
1 1 2 3 5 7 10 13 16 19 23 25 29 30_38_J3 col. 6, IB of tftnblo.
TTTl 2 2 3 2~"T 3 4 2 4 ~1 2 col. 6 of table No. 81.]
252. The interpretation up to the degree 6 is as follows;
[In the following Table No. 82 as originally printed, the heading of the fourth
column was "Constitution. Nos. in ( ) refer to Tables in former Memoirs except (S3)
and (84) which are given post," and the covariants were referred to by their No/
accordingly.] J " UHt
405]
AN EIGHTH MEMOIR ON QU ANTICS.
Table No. 82.
151
Degree.
Order.
No.
Constitution. Notation is the alphabetic notation of Id3,
A the quintic itself, ZJ, G qitadricovnrifmta,. &c.
^:;r
1
viz. the abaoluta constant unity.
1
5
1
A
Hf.
2
10
1
A*
6
1
C
JHf.
"
2
1
E
N.
3
15
1
A*
11
1
AG
9
1
]i>
XT
)f
7
1
AS
5
1
E
N.
99
3
1
T>
ff.
'4
20
1
A 1
16
1
j\?Q
)9
H
1
AF
H
12
2
G a A 3 J1
99
10
1
Ah
8
2
AD t EG
>9
6
1
I
N.
Jl
4
2
//, J?
N.
9>
1
G
N.
5
25
1
A D
99
21
1
AG
99
19
1
A*F
99
17
2
A*JJ, AG*
I)
16
2
A z E t GF
,,
13
2
A*, ABG
99
11
2
AI + J3F ' GE
S.
9>
9
3
AW, All, CD
19
7
2
BM> L
N,
99
6
2
AG, BD
91
3
1
K
iV,
1
1
J
N.
6
30
1
A 6
)9
26
1
A 4 G
9J
24
1 
A S F
99
22
2
A 3 G 2 , A*B
1)
20
2
A OF, A*J$
18
3
W + 4C 3 + AWAiBG =0
S.
91
16
2
A{AI+ J3F ~ GE} =0
S',
1)
14
4
6AG& l$F 4C a + AW =0, AW 3
S.
12
3
AL H SDF ~2CI =0, ABE
S.
9)
10
4
WG +12ABD AW+W =0, CH
S,
8
2
AK + 257 32)22 =0
S,
6
4
A J j 2JS// j5 3 GG 9J? a =
S.
19
4
1
N
ff.
9)
2
1
M
ff.
152 AN EIGHTH MEMOIR ON QUANTICS. [405
253. For the explanation of this I remark that the Table No, 81 shows that we
have for the degree and order one covariaut; this is the absolute constant
unity ; for the degree 1 and order 5, 1 co variant, this is the quintic itself, A ; for
degree 2 and order 10, 1 covariant ; this is the square of the quintie, A*\ for same
degree and order G, 1 covariant, which had accordingly to be calculated, viz. this is
the co variant C'; and similarly whenever the Table No, 81 indicates the existence of
a covariaut of any degree and order, and there does not exist a product of the
covariants previously calculated, having the proper degree and order, then in each such
case (shown in the last preceding Table by the letter j\ r ) a new covariant had to bo
calculated. Oil coming to degree o, order 11, it appears that the number of asyzygetic
invariants is only = 2, whereas there exist of the right degree and order the 3 com
binations AI, BF t OE\ there is here a syzygy, or linear relation, between the
combinations in question; which syzygy had to he calculated, and was found to be
as shown, AI+BFCE=Q t a result given in the Second Memoir, p. 126. Any such
case is indicated by the letter 8. At the place degree 6, order 16, we find a syzygy
between the combinations A Z J, A*BF, AOE\ as each terra contains the factor A,
this is only tho lastmentioned syzygy multiplied by A, not a new syzygy, _and I
have wiitten 6" instead of S. The places degree 6, orders 18, 14, 12, 10, 8, 6 indi
cate eacli of them a syzygy, which syzygies, as being of the degree 6, were not given
in the Second Memoir, and they were, first calculated for the present Memoir, It is
to be noticed that in some cases the combinations which might have entered into
the syzygy do not all of them do so ; thus degree 6, order 14, the syzygy is between
the four combinations ACD, EF> SO 3 , A*H t and does not contain the remaining com
bination A Z S. The places degree 6, orders 4, 2, indicate each of them a new covariant,
and these, as being of the degree 6, were not given in the Second Memoir, but had
to be calculated for tho present Memoir.
2134. I notice the following results:
Quadrinvt. QH = 3<?,
Cubinvt. QH . = H +54GQ,
Disct. (o5 + #M) = ((?, Q, BUfa, /3) a ,
Jac. (3, H) = &M t
Hess. 3D =JV,
the last two of which indicate the formation of the covariants given in the new
Tables MNo. S3 and N No. 84: viz. if to avoid fractions we take 3 times the
covariant D t being a cubic (a, ...} 3 (x, 2/) 3 > then the Hessian thereof is a covariant
(a, ...)"(*> 2/) s j which is given in Table, M No. 83; and in like manner if we form
the Jacobian of the Tables S and H which are respectively of the forms (a, , ,) a (w, y) 3 ,
and (, ..) e (X $> tn ' s i 8 a covariant (a,..) 6 (#. yf, and dividing it by 6 to obtain the
coefficients in their lowest terms, we have the new Table, N No. 84. I have in
these, for greater distinctness, written the numerical coefficients after instead of before,
the literal terms to which they belong.
405] AN EIGHTH MEMOIll ON QUANTICS.
The two new Tables are :
Table No. 83. M = (*$, y}\ See 143.
Table No. 84. JY = (*$>, i/) J . See 143.
153
Article No. 255. Formula for the canonical form arc 5 4 by rj + cz* = 0, wAere a; 4^42 = 0.
255. The quintic (, b, c, d, e, /$>, 2/) 5 may be expressed in the form
r B 4 sv 5
where , w, w arc linear functions of (to, y) such thab + D + w = 0. Or, what is the
same thing, the quintic may be represented hi the canonical form
where 0+y+* = 0; this is = (c,  C)  c , o, o, 6  oj, Z/) B > d the different
covariants and invariants of the quintio may hence be expressed in terms of these
coefficients (a, b, c).
For the invariants we have
Q = J = 6 9 c 2 i c 9 ft 2 + ft a 6 3  2a6c (a + b 4 c),
Q = 7f = 2 & a c a (60
If = / = 4ft B &V (6  c) (c  a) (a  i).
[Observe that throughout the present Memoir, the invariants, instead of being called
Q Q ~U, W are called 7, J, K, L, viz. the /, J, K, L in all tliot follows denote
the invariants, and not the covariante denoted by these letters in 142, 143, Moreover
D is used to denote the invariant Q', which is in fact the discriminant of the qnmtic.]
Hence, writing for a moment
a +b + c =p t and therefore J =<f4/jr,
be + ca+ab~q K = r*q>
abc =r & = r *>
(a 6)9(6o)(oa)'=j)Y 4 ? s ^ v + 18 ^ m " 2l7)l0 '
and thence , , (,,
73 = iQr w  45 3  4p* i 4 ISp^ 1  27r"),
= 16r 10 (pY  4g 3  42^' 4 ISpqr  27r a },
20
154 AN EIGHTH MEMOIR ON QUANTICS. [405
that is,
which is the simplest mode of obtaining the expression for the square of the 18thic
or skew invariant / in terms of the invariants J, K, L of the degrees 4, 8, 12
respective!}'.
If instead of the invariant K of the degree 8 we consider, the invariant D [~Q'
as before mentioned] of the same degree, this is
Q' = D = {Jw + cW 4 a*  2a&c (a 4 b 4 c)) 2  128 a i 2 c 3 (bo + ca 4 ab),
d 4
.and we have also the following covaviants ;
B = ( ao, ab ~ac 6c,
== bcyz 4 ca^ffl f abxy.
= ( ttc, Sac, Sac, a6ftc~ic, 86c,  36c,
 bciftP + cu&t? + a6a#y.
Z) = (0. ubc, aba, 0^, yy* aba ast/2,
4 ( oac 2 4 otc a  5a&c) j*i/
4 (  10ac a 4 lO&c 2  Sale) afy 3
4 (  10c a 4 'lOfcc 3 4 2a6c)/y
4(  5c 9 4 S6c a 4
4 ( afr  c 3 4 6c 9 4 6\) 4
(fc  o) a 2 * + (o  a) &y 4 (a  6) c^ s
 aJc (i/ *)(* ) (a;  y) (yz +MB + xy],
Article No. ^Q. Expression of the 18thio Invariant in terms of the roots,
256. It was remarked by Dr Salmon, that for a quintic (a, b, c, d, o, / $, i/) B
which is linearly transformable into the form (o, 0, c, 0, e, 0$>, 2/) B , the invariant 7
is = 0. Now putting for convenience y = 1, and considering for a moment the equation
then writing herein ~r for , the transformed equation is
405]
where
AN EIGHTH MEMOIR ON QUANTICS.
155
ny
, &c.;
P lm/3 1 7 lm/y
hence m may be so determined that ' + </ may be =0; viz. this will be the case
if jfl + 7  2j)ijS7, or m^J^. In order that S' + e may be =0, we must of course
have w, = "t 6 , and hence the condition that simultaneously j3'47 /= and S' + e' = is
2oe
jQ I X 1 I
".J.7 ~ "Jj.5 ; that is, (& + 7) Se  7 (S 4 e) = 0. Or putting IB a for a and /3,
2/37 2oe
7 , &c. for /3, 7, &c., we have the equation
(a,  a) (  /3) (0  7) (0  8) (*  e) 0,
which is by the transformation asct into ^^y^ changed into
(dj  a') (IB  0') (as  7') (a  S') (as  e') =
(where a' = a), and the condition in order that in the new equation it may be possible
to have simultaneously /S' 4 7'  2a' = 0, 8' + e'  2a' = 0, is
or, as this may be written,
J
1, jS + 7, /?7
1, S+e, Se
=0.
IIenoo writing ts + tx' for a, the lastmentioned equation is 'the condition in order that
bho equation _ / . .
(?  ) (  0) ( ~ 7) (*  S > (  e ) =
may be transformable into
J
whoro jfl' + y^O, S'+e'==0, that is, into the form a, (? ^) (?  8*) = 0. Or replacing
i/, if wo have
(a, 6, o, rf, e./jfrtf, y ) fl = 
then the equation in question is the condition in order that this may be transformable
into the form (*', 0, c', 0, e', <%, T/); that is, in order that the 18thic invanant I
mav vaiiiah. Hence observing that there are 15 determinants of the form in question,
and that any root, for instance , enters as rf in 8 of them and in the simple power
a in the remaining 12, we see that the product
1, 2 , a 3
^0 2
1, S+e, Se
156 AN EIGHTH MEMOIR OX QUANTICS. [405
contains eacli root in the power IS, and is consequently a rational and integral
function of the coefficients of the degree 18, viz. save as to a numerical factor it is
equal to the invariant /. And considering the equation (a, . .$#, 7/) fl = as representing
a range of points, the signification of the equation 7=0 is that, the pairs (y9, <y) and
(, e) being properly selected, the fifth point a is a focus or sibiconj ugate point of
the involution formed by the pairs (/3, 7) and (S, e).
Article Nos, 257 to 267. Theory of the determination of the Character of an
Equation; A utciliars ; Facultative and Nonfacultative space.
257. The equation (, b, c,.,$>, #) = is a real equation if the ratios a : b : c,..
of the coefficients are all real. In considering a given real equation, there is no loss
of generality in considering the coefficients (a, b, c..) as being themselves real, or iu
taking the coefficient a to be = ] ; and it is also for the most part convenient to
write # = 1, and thus to consider the equation under the form (1, 6, c..,]a;, 1)" = 0.
It will therefore (unless the contrary is expressed) be throughout assumed that the
coefficients (including the coefficient a when it is not put =1) are all of them real;
and, in speaking of any functions of the coefficients, it is assumed that these are
rational and integral real functions, and that any values attributed to these functions
are also real.
258. The equation (1, b t c...fa, 1)<^0, with a real roots and 2 imaginary roots,
is said to have the character ar+2@i; thus a quintic equation will have the character
or, & + 2t, or r + 4i, according as its roots are all real, or as it has a single pair, or
two pairs, of imaginary roots.
259. Consider any m functions (A, E, ... If) of the coefficients, (m= or <n). For
given values of (A, B, ... 1C), non constat that there is any corresponding equation (that
is, the corresponding values of the coefficients (b, c, ...) may be of necessity imaginary),
but attending only to those values of (A, B,...K) which have a corresponding equation
or corresponding equations, let it be assumed that the equations which correspond to
a given set of values of (A, B,,..K) have a determinate character (one and the same
for all such equations): this assumption is of course a condition imposed on the form
of the functions (A, B,,..K}> and any functions satisfying the condition are said to
be "auxihaiu" It may be remarked that the n coefficients (6, c, ...) are themselves
aimliars; in fact for given values of the coefficients there is only a single equation
which equation has of course a determinate! character. To fix the ideas we may con
sider the auxiliars (A, ,..,%) as the coordinates of a point in ^dimensional space, or
w.space.
Any given point in the mspace is either "facultative," that is, we have
ang thereto an equation or equations (and if more than one equation then by
wna* precedes these equations have all of them the same character), or else it is
" nonfacultative," that is, the point has no corresponding equation.
261. The entire system of facultative points forms a region or regions, and the
entire system of nonfacultative points a region or regions; and the mspace is thus
divided into facultative and nonfacultative regions. The surface which divides the
405] AN EIGHTH M13MOIR ON QUANTICS. 157
facultative and nonfacultative regions may be spoken of simply as the "bounding
surface, whether the same be analytically a single surface, or consist of portions of more
than one surface.
2G2. Consider the discriminant D, and to fix the ideas let the sign be determined
in such wise that D is + or ' according as the number of imaginary roots is
= (mod. 4), or is = 2 (mod. 4) ; then expressing the equation D = in terms of the
auxiliars (A, #,.,../(!"), we have a surface, say the discriminatrix, dividing the mspace
into regions for which D is +, and for which D is , or, way, into positive and
negative regions.
263. A given facultative or nonfacultative region may be wholly positive or wholly
negative, or it may be intersected "by the discriminatrix and thus divided Into positive
and negative regions. Hence taking account of the division by the discriminatrix, but
attending only to the facultative regions, we have positive facultative regions and
negative facultative regions. Now using the simple term region to denote indifferently
a positive facultative region or a negative facultative region, it appears from the very
notion of a region as above explained that we may pass from any point in a given
region to any other point in the same region without traversing either the bounding
surface or the discriminatrix ; and it follows that the equations which correspond to
the several points of the same region have each of them one and the same character;
that is, to a given region there correspond equations of a given character,
264. It is proper to remark that there may very well be two or more regions
which have corresponding to them equations with the same character; any such regions
may "be associated together and considered as forming a kingdom ; the number of
kingdoms is of course equal to the number of characters, viz. it is =(+ 2) or
( + 1) according as n is even or odd ; and this being so, the general conclusion
from the preceding considerations is that the whole of facultative space will be
divided into kingdoms, such that to a given kingdom there correspond equations having
a given character; and conversely, that the equations with a given character correspond
to a given kingdom. Hence (the characters for the several kingdoms being ascertained)
knowing in what kingdom is situate a point (A, J3, ... /), we know also the character
of the corresponding equations.
265. Any conditions which determine in what kingdom is situate the point
(A, B,.,.K) which belongs to a given equation (1, 6, c, .."&, l) ft =0, determine
therefore the character of the equation, it is very important to notice that the form
of these conditions is to a certain extent indeterminate ; for if to n. given kingdom
we attach any portion or portions of nonfacultative apace, then any condition or
conditions which confine the point (A, B t ... K) to the resulting aggregate portion of
space, in effect confine it to the kingdom in question; for of the points within the
aggregate portion of space it is only those within the kingdom which have corre
sponding to them an equation, and therefore, if the coefficients (&, c,.,.) of the given
equation are such as to give to the auxiliars (A, B, ... K) values which correspond to
.a point situate within the abovementioned aggregate portion of space, such point will of
necessity be within the kingdom.
158 AN EIGHTH MEMOIR ON QU ANTICS. [4Q5
266. In the case where the auxiliars are the coefficients (&, c, ...), to any given
values of the auxiiiars there corresponds an equation, that is, all apace is facultative
space. And the division into regions or kingdoms is effected by means of the
discrhninatrix, or surface D = 0, alone. Thus in the case of the quadric equation
(1, cc, y$ff, l) a = the mspace is the plane. We have D = a?y t and the discriminatrix
is thus the parabola ^^ = 0, There are two kingdoms, each consisting of a single
region, viz. the positive kingdom or region (tfy = \) outside the parabola, and the
negative kingdom or region (a?y = ~) inside the parabola, which have the characters
2s and 2z, or correspond to the cases of two real roots and two imaginary roots,
respectively. And the like as regards the cubic (1, a, y, z\Q, 1)' = 0; the mspace is
here ordinary apace, D = ~ 40"# + 3^y + fayz  % 3  z\ and the division into kingdoms
is effected by means of the surface D = 0; but as in this case there are only the
two characters Br and )+ 2i t there can be only the two kingdoms D = + and D = 
having these characters 3r and r + 2i respectively, and the determination of the
character of the cubic equation is thus effected without its being necessary to proceed
further, or inquire as to the form or number of the regions determined by the
surface =0: I believe that there are only two regions, so that in this case also
each kingdom consists of a single region, But proceeding in the same manner, that
IB, with the coefficients themselves as auxiliars, to the case of a quartic equation, the
TOspn.ce is here a 4 dimensional space, so that we cannot by an actual geometrical
discussion show how the 4space is by the discriminafcrix or hypersurface D = divided
into kingdoms having the characters 4?, 2r + 2 4i respectively. The employment
therefore ^ of the coefficients themselves as auxiliars, although theoretically applicable to
adequation of any order whatever, can in practice be applied only to the cases for
which a geometrical illustration is in fact unnecessary.
2G7. I will consider in a different manner the case of the quartic, chiefly as an
instance of the actual employment of a surface in the discussion of the character of
an equation; for in the case of a quintic the auxiliars are in the sequel selected in
such manner that the surface breaks up into a plane and cylinder, and the discussion
is in fact almost independent of the surface, being conducted by means of the curve
(Professor Sylvester's Biooni) which is the intersection of the plane and cylinder.
Article Nos. 268273. Application to the Quartw equation,
26S. Considering then the quartic equation (a, b, c, d, e\0, 1)< = (I retain for
symmetry the coefficient a, but suppose it to be 1, or at all events positive), then
if /, J signify as usual, and if for a moment
S = a?d  3a6c + 26",
X =
we . have identically
+ *) a* 9 (6" 
405] AN EIGHTH MEMOIR ON QUANTICS. 159
(see my paper, " A discussion of the Sturmian Constants for Cubic and Quartic
Equations," Quart. Math. Journ., vol. iv. (1861) pp. 7 12), [290]. And I write
a = b' 2 uc,
269. I borrow from Sturm's theorem the conclusion (but nothing else than this
conclusion) that (a 1 , y, z) possess the fundamental property of auxiliare (that is, that the
quartic equations (if any) corresponding to a given system of values of (x } y, z) have
one and the same character). The foregoing equation gives O&y s?z y 3 a square
function, and therefore positive ; that is, the facultative portion of space is that for
which Qn) s if a?z y s is =+. And the equation
is that of the bounding surface, dividing the facultative and n on facultative portions
of space.
270. To explain the form of the surface we may imagine the plane of coy to be
that of the paper, and the positive direction of the axis of s bo be in front of the
paper. Taking z constant, or considering the sections by planes parallel to that of &y t
z~Q, gives f (QoP  y) ~ 0, via. the section is the line y. = 0, or axis of ts twice,
and the cubical parabola y~a?.
* =
(non.fao.)
T(non,fcto.)
V s /
z = +, the curve ic 3 = has two asymptotes y = ^vz, parallel to and equi
jy z
distant from the axis of a, and consists of a branch included between the two
parallel asymptotes, and two other portions branches outside the asymptotes, as shown
in the figure (2 = +).
1S
. the curve f
AN EIGHTH MEMOIR ON QUAKTICS. [ 405
has no real llsympt ot e , and consists of a single
branch, resenrb.in, in its appearance the cubical parabola as shown in the Hgnvc
Taking ., as constant, or considering the sections by planes parallel to that of
the equation of the action is ,, which i s a cllbical pm , bolil)
facultative rerions ^4' /?
0). and if (., y } } ' ' ' Wl e "' one of the regions
is in the region B, the character is 4r,
A '> >, 4i,
e co r on, 7; to obtain
* + ,* + J ff+ includes the whole of facultative region
is, (a, y f ,) being eac h positive, the character is 4rj
*==
' IV I
= , 2,=  I Delude each a part and together the whole
of facultative region A' t
405] AN EIGHTH MEMOIR ON QU ANTICS. 1.61
that is, z being +, but (cc, y) not each positive, the character is 4i;
include each a part and together the whole
'of facultative region G,
v =  , y =  )
s = , as =  , y = + does not include any facultative space,
that is, z being , the character is 2r2i; and the combination of signs z = ~,
,%== , y = + is one which does not exist.
The results thus agree with those furnished by Sturm's theorem ; and in particular
the impossibility of z =  , = , y = + appears from Sturm's theorem, inasmuch as
his combination would give a gain instead of a loss of changes of sign.
Article Nos. 274 to ^S^Determination of the characters of the quintic equation.
274. Passing now to the case of the quintic, I write
J = G,
K = Q,
T _ 77
jj u,
I = If;
vi. J is the quartinvariant, K and D are octinvariants (D the discriminant) i is
12thic invariant, and J is the 18thic or skew invariant. Hence also > J, V, i LJ
arc invariant of the degrees 4, 8, 12 respectively; and forming the combinations
2Z L J* ^ D =J
I assume that (* y. *} are auxiliars, reserving for the Concluding articles of the
pioaont memoir the considerations which sustain this assumption.
275. The separation into regions is effected as follows :Wo have identically (see
ante, No. 256)
or putting for K its value = T ^(e/ 3 ~D), this is
+ DJ ( 4J + 61
o. vi.
162 AN EIGHTH MEMOIR ON QUANTICS. [405
or writing as above
_ 2" L  / 3 D
*'  T 3 ' ^75*
whence also
i ^ 2ni
1 + * jT,
this is
2 M * d 80 +*)!}
+ 36 l+*'J.l + **
or, what Is the same thing,
Ta
= <(&, y) suppose,
276. Hence alao writing s~J, we have
saf> fa y)==2.2 33 ~= + ,
or the equation of the bounding surface may be taken to be
x$ (0, y) = 0,
that is, the bounding surface is composed of the plane = 0, and the cylinder
<t>(> 2/) = 0. Taking the plane of the paper for the plane ^0, the cylinder meets
this plane in a curve <f>fa i/)=:0, which is Professor Sylvester's Bloom: this citrve
divides the plane into certain regions, and if we attend to the solid figure and instead
of the curve consider the cylinder, then to each region of the plane there correspond
in sohdo two regions, one in front of, the other behind the plane region, and of these
regions m solido, one is facultative, the other is nonfacultative (viz. for given values
of (*, y], whatever be the sign of (, ff ), then for a certain sign of g t gffa y) will
be positive or the solid region will be facultative, and for the opposite sign of
^toy) will be negative or the region will be nonfacultative). It hence appears that
we may attend only to the plane regions, and that (the proper sign being attributed
to *, that is to J) each of these may be regarded as facultative. It is to be added
that the discnnimatnx is in the present case the plane y = 0, or, if we attend only
to the plane figure, it is the line # = 0; so that in the plane figure the separation
into regions is effected by means of the Bicorn and the line y = 0.
AN EIGHTH MEMOIR ON QUANTICS. 163
277. Be verting to the equation of the Bicorn, and considering first the form at
infinity, the intersections of the curve by the line infinity are given by the equation
y s (2ya))^0, viz. there is a threefold intersection if = Q t and a simple intersection
2?/# = 0; the equation ?/ s = indicates that the intersection in question is a point
of inflexion, the tangent at the inflexion (or stationary tangent) being of course the
line infinity; the visible effect is, however, only that the direction of the branch is
ultimately parallel to the axis of as. The equation 2yaf = Q indicates an asymptote
parallel to this line, and the equation of the asymptote is easily found to be
1y _ g; + 5 = Q.
278. The discussion of the equation would show that the curve has an ordinary
cusp; and a cusp of the second kind, or nodecusp, equivalent to a cusp and node;.
toho curve is therefore a unicursal curve, or the coordinates are expressible rationally
in terms of a parameter </>; we in fact have
whence also
279. The curve may be traced from these equations (see Plate, fig. 2, where the
bloom ia delineated along with a cubic curve afterwards referred to): as tf> extends
from an indefinitely small positive value e through infinity to I e, we have the
branch of the curve, viz.
(ft = e> gi voa 'V=GQ, i/ = ~oo, point at infinity, the tangent being horizontal,
ft = cc, gives / = 1, ?/==+, the nodecusp, tangent parallel to axis of y,
$ = _ 2, gives to = 0, y = 0, the tangent at this point being the axis of at,
( f> = ~l G t gives i = oo, a/ = +, point at infinity along the asymptote;
and as c/> extends from ,v;~l+e to ?*, we have the lower branch, viz.
164 AN EIGHTH MEMOIR ON QUANTICS. [405
281. The form of the Bicorn, so far as it is material for the discussion, is also
shown in the Plate, fig. 3, and it thereby appears that it divides the plane into three
regions ; viz. these are the regions PQR and S, for each of which < (#, y) is = ,
and the region TU, for which </>(#, y) is = + ; that is, for PQR and 8 we must have
J~ t and for TU we must have J=+. Hence in connexion with the bicorn, con
sidering the line y=Q, we have the six regions P, Q, R, S, ?', U. It has just been
seen that for P, Q, R, S we have J=~, and for T, U we have /=+; and the
2"i  ,/ s D
sign of ./ being given, the equations ce = j , y = ^ , then fix for the several
regions the signs of 2"Z / 3 and D t as shown in the subjoined Table ; by what
precedes each of the six regions has a determinate character, which for R t S, mid U
(since here D is =) is at once seen to be 3rj2t, and which, as will presently
appear, is ascertained to be or for P and r+U for Q and T.
282, We have thus the Table
P, D= + t J=~,
U,
so that we have the kingdom or consisting of the single region P, the kingdom
r + 4t consisting of the regions Q and T t and the kingdom 3r + 2i consisting of the
regions R t 8, and U.
283. For a given equation if D is =, the character is 3r + 2i; if D = +, J = +,
the character is j + 4i; if D = + , J=~, then, according as M/ 3 is =+ or is
, the character is 5r or r+4&, But in the last case the distinction between bho
characters or and r + 4i may be presented in a more general form, involving a
parameter /*, arbitrary between certain limits. In fact drawing upwards from the
origin, as in Plate, fig. 3, the lines #2y = and tc + y = Q, and between thorn any
line ^ whatever an/(,y0, the point (at, y), assumed to lie in the region P or Q, will
lie in the one or the other region according as it lies on the one. side or the other
side of the line in question, viz. in the region P if te + fjaj is =, in the region Q
if a? + py is = h But we have
and J being by supposition negative, the sign of M  J* + pJJ) is opposite fco that
of + ^y. The region is thus P or Q according to the sign of ^LJ^^JD' and
completing the enunciation, we have, finally, the following criteria for the number of
real roots of a given quintic equation, viz.
If D =  , th e character is 3r + 2?,
If D = +, J+, then it is r+ 4
405] AN EIGHTH MEMOIR ON QU ANTICS.
165
.But if D I, J=~ } then /* being any number at pleasure between the limits +1
2, both inclusive, if
2"i  J a + fiJD = + , the character is 5<r,
2H4, The characters 5? of the region P and r + 4i of the regions Q and T may
be ascertained by moans of the equation (a, 0, c, 0, e, 0&0, 1) 5 =0, that is
tlioro in always the roal root # = 0, and the equation will thus have the character or
or rM'f! according as the reduced equation a# 4 + 10c0 a + 50 = has the character 4? 1
or <l<t, ,l.li in clear that (a, e) must have the same sign, for otherwise 8* would have
two real values, one positive, the other negative, and the character would be 27  + 2i.
And (it; (?) having the same sign, then the character will be 4?', if 8 3 has two real
positive values, that is, if e 5o is = , and the sign of c be opposite to that of
a and e, or, what is the same thing, if ce be =~; but if these two conditions are
not HtitiHfuul, Chen the values of 0* will bo imaginary, or else real and negative, and
in oithor oasti the character will be 4r.
2Hr>. Now, for tho equation in question, putting in the Tables 6 = rf=/=0, we find
J= 16 OG (tie + 3c 3 ),
J ;I = 2 ia ce s {2 (ae  c 2 ) 4  o 2 (oe + 3c 3 ) s j
2 ia ce 3 (ae  oc 2 ) (Stfe 8 + a c 3 e a + 8oo*fl + 5c").
Wo hiivo by supposition 73 = +, that is, ae = + ; hence J has the same sign as ce;
whoncn if ./'= + , thon also 00 = +, and the character is 4i; that is the character of
the rogion T is jh4f. But if J = , then also ce = . But ae being = + , the sign
of 2.  /" w the name as that of ce (ae  6c 2 ), and therefore the opposite of that of
tie  (V : hcncn D =s h , J"  , the quartic equation has the character 4?' or 4i
according an 2 n /yi/ is =4 or =. Hence the region P has the character Br and
tho nsgion Q tho character r + 4i ; and the demonstration is thus completed,
Article Nos. 286 to 203. HEHMITK'H new form of TSOHIRNHAUSEN'S transformation, and
application thereof to the quintic.
2H(i M Hormite demonstrates the general theorem, that if f(a, y] be a given
cpuintio of tho th order, and <j>& y) any covariant thereof of the order 2, then
wnwidoring the etjuation /(, 1)0, and writing
166
AN EIGHTH M.RMOIU ON QUANTI08.
[405
(where //(*, 1) is the derived function of /(, 1) in regard do ;), thtm eliminating
a 1 , we have an equation in z, the coefficients whoroof tiro nil of them invariants of
287. In particular for the qaintic /(,*, y) (, 6, (', rf, e t ffcai, jry)", if
ftfa y), &0, y), <M (t '> y), <M<i'', y)
arc tiny four covariant cubics, writing
2 = 
(vix. the numerator is a covuriimt ubio involving bho iutlutunniiiititi! iMmllUjinntK
t, j u, i), w) than, in the tiiuisformod oquatittn in z, kho cooillcioutH uvo all ul' liluini
invnriants of fcho given quiutio. ContUioting tho invostigntion by IIKMUIH ol' a oorlitiiu
canonical form, which will bo rofonod bo in tin.! soquol, ho lixtw tho wgnirioation o(' IIJH
four covariant cubica, these being roMpeetivoly oovarmnli culmw of tho tlogrouK ,'{, " ( 7,
and 9, defined as follows; vk startiupf with tho form
3
f>  ;'A.
* i
6, v ,
c , d ,
u , d
d , o
>' > f
8A 8(^ ^ 0, tfjfa y y, or (JU
and considering also the quadrio covwiaiit
(. A 7?*, ?/)',
then ti, ^ a) ^ 3 , 0,, arc derived from the form
(A, B, 0,
ia, we have
, .#, Cf,
1 5, 6',
whero ^(flv y)J and [0,( ffi , y) ) ft ro tlio functions originally eallod by him fcfe ) and
^(w. y): those uliamatoly HO called by him are 9 <V
405]
AN EIGHTH MEMOIR ON QUANTICS.
167
whore fa (a, y) is the cubico variant ( 27.4 3 D + QABC  25 3 , . .$, </)' of fa (x, y),
= (~'3A, 5, <7, 3]J>, y)\ ut sivprcb,
The covariant fates, y) has the property that if the given quintio (,..,][#, y) 5
contains a square factor (k + mi/)' 2 , then fafa y) contains the factor Ix \rny: [fa (as, y)}
and [fa(x, y}\ are covariants not possessing the property in question, and they were
for this reason replaced by fa (as, y} and fa (as, y) which possess it, viz. fa (as, y) contains
the factor latmy, and fa(ai, y) contains (kv + myj 1 , being thus a perfect cube when
the given quintic contains a square factor.
288. The covariants fa (as, y) and fa (as, y) are included in my Tables, viz. we have
fa ( a! , y } =  87) of 142, 143,
(observe that in K the first coefficient vanishes if it, 0, b 0, which is the property
just referred to of fa(a), y))', the other two covariants, as being of the degree 7 and 9,
are not included m my Tables, but I have calculated the leading coefficients of these
covariants respectively, viz.
Table No. 85 gives leading coefficient (or that of ,' 3 ) in fa(sv, y), and
Tablo No. 86 gives leading coefficient (or that of a; 3 ) in fa (to, y) [and by means
thereof we have the values of the covariants in question],
The coefficients in question vanish for a 0, 6 = 0, that is, ^ 3 (ai, y) and fa(ss t y)
then each of them contain the factor y\ if the remaining coefficients of fa(to t y) were
calculated, it should then appear that for a = 0, 6 = 0, those of a*y, esf would also
vanish, and thus that fa (a), y) would be a mere constant multiple of y s .
Table No. Ho [= leading coefficient of 16J5J
aV* + 1
fl^V 8
 1
fl& 3 (// a + 64
&v a
f ,9 f nf9 i ~\ f\
ft ft J 11
tMcdf*
 94
rtiVy  54
frto/  144
cWf 32
ft tJCG /
+ 86
& a c a y la 48
6V +135
ttV '+16
ft Oft C/
+ 106
fliWe/ + 184
6*0 V + 108
a*bd#
 96
i a ce 3 135
6 9 c^y + 288
V/' 2
+ 63
ab*d s f 272
CflC J i5U
rcWfi/
188
fli a ( 2 e 2 + 243
bWe + 80
rtw'
+ 32
a&oy  06
Wtfdf  360
oVSy
+ 60
+ 68
aficVy + 212
a6c a f/c 8 + 148
69 t 1 O
OS + lOO
6 2 c a (? 2 e + 360
a d%
 36
abod^e 412
Fed 4  160
abd 5 + 144
6c B / + 108
rttf 1 ^  36
6c 4 c?e 180
a u ) o
f(C fi 'iO
ic 3 d 3 + 80
flC 3 (Z 3 e + 124
ftc ft 48
+ 32
+ 415
+ 1119
+ 1294
168
AN EIGHTH MEMOIR ON QUANTICS.
Table No. 86 [= leading coefficient of S'].
[405
atcej* + 9
cMef*  9 1 rtW/ 3 4 120
6V ;J  576
i B / ;i + 192
aWf 3 + 21
tfbcdf 3  162
W/"  21
fl&W + 672
b*cef 2  1440
rtWe/ 3  7S
:i 6c&y a + 99
aW/ 3 H 480
o6V/  359
6"(Z 4 / 2  192
V/ + 48
n^e/ 2 + 309
aWcdef 2  2160
6 !1 cV' + 3456
We 3 /  1080
%&'/ + 12
aWfcfi 3 / + 1023
H6 B crfV a  864
6V ' + 2025
n %^  240
W/ B + 120
abWf+ 3094
iVW/ a + 2592
a 3 c 3 / a  81
a6w/ 1053
& 3 ce 4 '  3915
6W/' + 3546
aWdef + 1026
aWa* + 1314
aPWy + 528
frcd'af 4 5280
W/  768
rt a ic s e/'  1863
aJWfl 3  45
^crffl 5  13500
J crf a / 2  738
a*lc*d*f* + 253S
afiV^'  2592
bWf  4800
aW/  5G4
?l<?d*f + 2340
rtfiW/  9747
ftW + 7800
ft ; 'ct& 4 3 1066
s 6cV + 672
alMPef 8496
6V/ S  648
W + 756
tfbctPef + 2820
6Vf/e ;1 + 26610
iWe/  14040
W  69G
WV _ 7812
a&txPf 4 8544
/> J cV + 3075
6^y  3024
a&ctPe  16650
6W/' H 9120
2 6cZV + 457 S
oJ'rf'a + 720
6 3 cW + 16350
ftV<J/ a  324
(tbo s f' + 972
WC&B  19200
aW/ f 3b88
6c 4 rfe/ + 24048
W + 4800
Vrf*e/  8748
6c 4 fi s  4464
fiVff/ 1 + 4860
a a c a rfe :i  4800
6c :1 (/ ;1 /'  15984
6W/ 1  3240
aW/ + 4248
flfiflWe"  30108
6 2 cW  8100
V(/V f 14520
abcWe + 35088
6cW ;! c + 9000
tPecPc 11448
aicrf"  8640
W  2400
a rf + 2592
oV  7776
atftPf + 5184
j atMe* + 12960
] fluWe  14400
I fflc 3 ^ + 3840
+ 7S 3258 +41253 124716
[The values thus are fa (at, y) = lGBJUJ}G; fa fa $ = S'.]
2S9. The equation in g is of the form
+ 68640
where D is the discriminant of the quintic and 91, 53, <, 2) denote rational and
integral functions of the coefficients (a, 1>, c, d, a, /). And the covariants fa fa y),
fa fa y\ fa fa y), fa fa y) having the values given to them above, the actual value
of 21 is obtained as a quadric function of the indeterminates (t t u, v, w) t viz. this is
 [D,/ 3  QBDto  D (A  IQAty v*] +D[~ JS a + W,uw + Q(B$~ IQADJ v^
where _D,= 25^5 + IGG, these quantities, and the quantity _A r (= DflQABfy
afterwards spoken of, being in the notation of the present Memoir as follows;
A = J (== G}>
= OZ
D = D
A = 9 (16Z  JK\
N = 1152 (18Z 9  JKL ~ 10).
405] AN EIGHTH MEMOIR ON QUANTICS. 169
290. If by establishing two linear relations between the coefficients (t, it, v, w)
the equation Sf = can be satisfied (which in fact can be done by the solution of a
quaclric equation), then these quantities can be by means of the relations in question
expressed as linear functions of any two of them, say of v and w\ and then the next
coefficient 33 will be a cubic function (v, wj 1 , and the equation 33 = will be satisfied
by means of a cubic equation (v, wj l = 0, that is, the transformed equation in z can
be by means of the solution of a quadrie and a cubic equation reduced to the trinomial
form
( 3)
and M. Hermito shows that the equation [ = can be satisfied as above very simply,,
and that in two different ways, viz.
291. 1. 21 = if
Dtf  QBDtv  ( A  104fl ) v n * = 0,
Btt*  W t uw  (9BD  IQADJ itf = 0,
that ifi, N denoting as above, if
292. 2. Writing the expression for 9[ in the form
A (i 3  7?w + Wuw  IQADuP) + ED (IQAv' 1  Qtv ' +
then 9( = 0, if
i a  Du 2 + Wuw  WADi^ = 0,
IQAv*  Gtv  u* + 9 Dw* = Q.
These equations, writing therein
 '
V2
become
 f>A 7 2 + 8VOTF+ U 2 + 1QA UW+ (25^ 9  9) Tf = 0,
the first of which is satisfied by the values
T = pW~U, F=pF + ij7;
and then substituting for T and V, the second equation will be also satisfied if only
a = BA
O. VI. , 22
AN EIGHTH MEAIOIR ON QUANTICS. [405
Article Nos. 29S to 295. HERMITE'S application of the foregoing results to the deter
mination of the Character of the qitintic equation.
293. By considerations relating to the form
D (D,  WAS) iff] + D[~ Bit + W t uw + QBD
M. Hermite obtains criteria for the character of the quintic equation /(JB, 1) = 0.
294. If D =  , the character is 3r + 2i, but if D = + , then expressing the fore
going form as a sum of four squares affected with positive or negative coefficients,
the character will be 5r or 2 + 4i, according as the coefficients are all positive, or
are two positive and two negative. Whence, if N denote as above, then for
D = + , JV=, A = + , 5 = , character is 5?,
J> + , jp^, BA 1>
anc ' r character is r + 4ij
J) = + ,JY = + j
and further, the combination = + , N = , ]) l =  ) = + cannot arise (Hermite's first
set of criteria),
295. Again, from the equivalent form
which, if <a, of are the roots of the equation 90 a  1040 + .D = 0, is
then by similar reasoning it is concluded that
JJ = + t 25 A* 9.Z) = + , A = , N~~ t character is 5r,
(Hermite's second set of criteria.)
405] AN EIGHTH MEMOIR ON QUANTICS. 17 1
Article Nos. 296 to 303. Comparison with the Criteria No. 283: the Nodal Cubic.
296, Tor the discussion of Hermite's results, it is to be observed that in the
notation of the present Memoir we have
A = J,
B If =
A = UL  JK = fa (2" L  J 3 + JD),
N = 1&&JKLIP
* foa OS! 7" 3 I A TT ( Ta T\\ f Tn
or, putting as above,
" Li J' is i ji e i i " " i L/
JP .. ^ ^ = j aiiu tiiiereiore i 4 is ~~y^~ > * 2/ fs
i/ / <y j
we havo
A = J,
= ,7 . {T/"  3?/ a + Bay 4 9# 2 + HT/ + 100).
It thus becomes necessary to consider the curve
<$> (a), y) = y 3  &f + tooy + 9* 3 + lly + lOai = 0,
the equation whereof may also be written
i 5 = (y  1) V25~^9y.
297. This is a cubic curve, viz. ib is a divergent parabola having
lino 9; + 4y + 6 0, and its ordinates parallel to the axis of
a node at the point #1, y + l, that is, at the node
curve is thus a nodal cubic; we may trace it directly fro
to be noticed that qti& nodal cubic it ' is a unicursal curve
therefore rationally expressible in terms of a parameter ^;
we in fact have
81(0 + 1)= ^(^8),
whence also
172 AN EIGHTH MEMOIR ON QUANTICS. [405
298. We see that
\]r = <x>, gives x = co , (/ co , point at infinity, the direction of the curve
parallel to axis of <v.
^=9, a; 0, #=0, the origin,
A/r = S, a' = 1, t/ = fl ) the node, tangent parallel to axis of ?/.
^ tfi =$&7> y = 2 $?~> tangent parallel to the axis of y,
^ = 4, # JgijS, y = %, tangent parallel to axis of ss.
^p 0, (i; = 1, ^ = 41, the node.
i/r = 16, .i; = 7Gff, y = 41f, the cusp of the bicorn.
A^ = GO, i = co, i/ = GO, point at infinity, direction of curve parallel to
axis of IK.
299, The Nodal Cubic is shown along with the Bicorn, Plate, fig, 2; it consists
of one continuous line, passing from a point at infinity, through the cusp of the
bicorn, on to the nodecusp, then forming a loop so as to return to the nodecusp,
again meeting the bicorn at the origin, and finally passing off to a point at infinity,
the initial and ultimate directions of the curve being parallel to the axis of at.
300. It may be remarked that, inasmuch as one of the branches of the cubic
touches the bicorn at the nodecusp, the nodecusp counts as (4 + 2=) 6 intersections;
the intersections of the cubic with the bicorn are therefore the cusp, the nodecusp,
and the origin, counting together as (2 f 6 f 1 =) 9 intersections, and besides these the
point at infinity on the axis of a, counting as 3 intersections. This may be verified
by substituting in the equation of the cubic the bicorn 0values of a and y. But
to include all the proper factors, we must first write the equation of the cubic in the
homogeneous form
(9w + 8y + 52) 3 z  (y  *) a (25*  9y) = 0,
and herein substitute the values
: y i * = (0
the result is found to be
6<  9)*  (20 + 3) a (4f + 40 a I 180 + 27)} = 0,
that is
and considering this as an equation of the order 12, the roots are = 0, 3 times,
= 2 J 1 time; = , 2 times, and = oo, 6 times,
301, The cubic curve divides the plane into 3 regions, which may be called
respectively the loop, the antiloop, and the extra cubic; for a point within the loop
or antiloop, ^(a t y) is = , for a point in the extra cubic $(&, y} is ==+. If in
conjunction with the cubic we consider the discriminatrix, or line y~Q, then we have
in all six regions, viz. y being = +, three which may be called the loop, the triangle,
405] AN EIGHTH MEMOIR ON QUANTICS. 173
and the upper region; and y being =*, three which may be called the right, left,
and under regions respectively; the triangle and the other region form together the
antiloop.
302. It is now easy to discuss Hoimite's two sets of criteria; the 6rsfc set becomes
== \ =  Jas + =+, t(. !/) = > character or,
, , . ,.
character  + 4u
?/ = +, tfl = +, J( + ff) = ' f(^2/) =  cannot exist.
Referring to the Plate, fig. 4, which 'shows a portion of the cubic and the bicorn,
then 1 the conditions ?/ = +,f (to, y} =  imply that the point (a>, y)_is within the
loop or within the triangle of the cubic; the condition yl= brings it to be
within the triangle, and for any point within the triangle we have + # = , whence
also the condition J> + ff) + becomes J=J hence the conditions amount to
/ =  (to ) within the triangle; but by the general theory (ta t y), being within the
triangle, that is, in the region P or T, if /=, will of necessity be within the region
P; so that the conditions give J , (, y) within the region P; the corresponding
character being 5r, which is right.
2 y = + ty (a y) =  t the point (x> </) must be within the loop, or within the
triangle; if (*, y} is within the loop, then yl + , + yl, and the condition
J(tfl)(+y) = + becomes J, that is, we have J =  and (, y) within the loop,
that is, in the region T. And again, if (w, y] be within the triangle, then yl=,
*, + +, and the condition J(y 1) (0 + ff) =+ still gives /; but /=, and
<, ) within the triangle, that is, in the region T or P, will of necessity be in
the region T;' so that in either case we have /=, (, y) ^ the region T, which
agrees with the character r + 4i.
3 +, t(*. 2/) ==+ > (" *) is in the Upper regi n 
or 2'; if (a, y} is in the region Q, then of necessity / = , and if in the region 1,
then of necessity /== + , that is, we have
/=, (a, y} in the region Q, or
/=+, (aj, y) in the region P,
which agrees with the character r + 4i.
And it is to be observed that the portions of T under 2 and 3' respectively
make up the whole 'of the region T, and that 3 relates to the whole of the region &
so that the conditions allow the point (a, y} to be anywhere in Q or A which is
right.
ysa+i ^( Xi y ) = _, (a, y) is in the loop or the triangle, and then 2/l = +
sai i , ,
implies that it is in the loop, whence a> + y = + , and the condition J (*
becomes / = ; we should therefore if .the combination existed have J=, (*, y)
within the loop, that is, in the region T; but this is impossible.
AN EIGHTH MEMOIR ON QUANTICS. [405
303. Hermite's second set of criteria are
y+> 2/ = +> </=, $(%>!/} = , character or,
y = +, ^y = +, J=, ty(x,y) = + }
y = +, 2g.~y = +, J=+ t  character r f 4i.
y = +J ^fy=, J
1. If y=+ t $*(&; y} = , then the point (a?, y) must be situate within the loop
or within the triangle; and recollecting that at the highest point of the loop we have
y = a p t the condition ^^s=+ is satisfied for every such point, and may therefore be
omitted. The conditions therefore are /=, (a?, y) within the loop, that is, in the
region T t or within the triangle, that is, in the region P or the region T\ but for
any point of T the general theory gives J=+, and the conditions are therefore ,7=,
(flf, y) within the region P \ which agrees with the character or.
2  2/ = +. if 1 OB, #) = +> that is, (IB, y) is within the upper region, that is, in the
region Q or T \ and tyi_0 = +, (., y ) ^ill be within the portions of Q and 2' which
lie beneath the line y = ^; but J=~, and therefore (0, y) cannot lie in the region
T\ hence the conditions amount to /=, (#, y) within that portion which lies
beneath the line y ~ *$. of the region Q.
_ 3  ff = +, ^y = +, (, y) lies beneath the line y = ^, viz. in one of the
regions P, Q or T; but J= + , (x, y) oannot lie in the region P or Q; hence the con
ditions give /=+, (as, T/) within the portion which lies beneath the line y = fy of the
region T.
4  2/ = +. ? ttat is, (a;, ?y) lies above the line y = ^, and therefore in
one of the regions T or Q; and by the general theory, according as (a;, y) lies in T
or in Q, we shall have /= + or J=~, hence the conditions give
J = t ( X} T/) within the portion which lies above the line y = ty, of the region Q,
J = f, (a;, y} within the portion which lies above the line 2/ = ^f, of the region T,
2 a , 3 D , and 4, each of them agree with the character r + 4i, and together they imply
J"==^(<e, y) anywhere in the region Q, or else J=+, (#, T/) anywhere in the region T,
which is right.
Article Nos, 304 to 307". HERMITE'S third set of Criteria; comparison with No. 283,
and remarks.
304. In the concluding portion of his memoir, M, Hermits obtains a third set of
criteria for the character of a quintic equation ; this is found by means of the equation
for the function
4D (0 a  00 W.  00 (04  0,)
of the roots (# fl , lt 6 B , 8 S> 4 ) of the given quintic equation (a, b, o, d, e, /0, 1) B = 0.
The function in question has 12 pairs of equal and opposite values, or it is determined
405] AN EIGHTH MEMOIR ON QUANTIGS. 175
by an equation of the form (tt a , 1) 13 =0, which equation is decomposable, not rationally
but by the adjunction thereto of the square root of the discriminant, into two equations
of the form ( 3 , 1) G = ; viz. one of these is
+ 10 (a+3
+ " [i(aVA) 9 +A]
u a d
+ w* [I (a 4 VA> + A] A
+ a (a3V'A)A 3
+ A 3 = 0,
and the other is of course derived from it by reversing the sign of VA, I have in
the equation written (a, d) instead of Hcrmite's writing capitals A, D ; the sign
of the term in u 9 instead of +, as printed in his memoir, is a correction communicated
to me by himself. The signification of the symbols is in the author's notation
a = 5V1,
d 4. 6" (.4.0^ A),
A = 6>0,
whence, in the notation of the present memoir, the expressions of these symbols are
a  o 4 /,
A = 5"D.
305. From the equation in u t taking therein the radical VA as positive, M. Hermits
obtains (d<0 a mistake for d>0) the following as the necessary and sufficient con
ditions for the reality of all the roots,
A = +, a 4 3 VA = , d = +, character 5r
(Hermite's third set of criteria).
306. It is clear that a+3VA =  is equivalent to (a =  and '"
we have a 3  9A = 6" (125^9.0), so that these conditions for the chi
Now, writing as above,
& 2"i/ 3 D
these are i/ = +, J = ~, ty* &=>+> 2/ = ; the conditions </
(, y) is in the region P or the region Q ; and the condition
176 AN EIC4HTH MEMOIR ON QUANTICS. [405
line # y lies between the lines # + ?/ = 0, a 2?/ = 0, and so does not cut either
the region P or the region Q) restricts (tc, y) to the region P ; and for every point
of P y is at most =1, and the condition IJA y = + is of course satisfied. The con
dition, 125 J a 9D = +, is thus wholly unnecessary, and omitting it, the conditions are
= 0, character 5r,
which, being an admissible value of /*, agrees with the result ante, No. 283.
307. It may be remarked in passing that if 12345 is a function of the roots
(ft 1 ,, #, a' 3l A' 4) d' B ) of a qiuntic equation, which function is such that it remains
unaltered by the cyclical permutation 12345 into 23451, and also by the reversal
(12345 into 15432) of the order of the roots, so that the function has in fact the
12 values
a, = 12345, # = 24135,
cr a = 13425, ft = 32145,
a 3 = 14235, # = 43125,
4 = 21435, # = 13245,
a B = 31245, #=14325,
/= 41325, # = 12435,
then <$ (a t /3) being any unsym metrical function of (a, /3), the equation having for its
roots the six values of <(, j8) (viz. <(a u #), 0(o a , #)...< ( BJ #)) can be expressed
rationally in terms of the coefficients of the given quintic equation and of the square
root of the discriminant of this equation. In fact, v being arbitrary, write
, iIIa {*(, 0)1, Jf=II 8 [<;(# a)},
then the interchange of any two roots of the quintic produces merely an interchange
of the quantities L, M\ that is^
L + M and (Z Jlf ) + &(&' * a , # 3 , es tt a' n )
are each of them unaltered by the interchange of any two roots, and are consequently
expressible as rational functions of the coefficients; or observing that $(aj lt K. I} # 3 , es it a; B )
is a multiple of VD, we have L a function of the form PfQVI)'; the equation
i = 0, the roots whereof are v = < (a, t #) ...v = ^ (a a , #), is consequently an equation
of the form P + Q*/D = Q > viz. it is a sextic equation (*Jw, 1) B = 0, the coefficients of
\vhich are functions of the form in question. Hence in particular
i(? = 12345 = (a,  a; a ) a (x,  to,)* fa  a>$ fa  fl ) 3 (d? B  ^) a
is determined as above by an equation (*][?, 1) G = 0. Another instance of such an
equation is given by my memoir " On a New Auxiliary Equation in the Theory of
Equations of the Fifth Order," Phil, Tmns, vol. OLT. (1861), pp. 263276, [268],
405] AN EIGHTH MEMOIR ON QUANTICS. 177
Article Nos. SOS to 317. HEEMITE'S Ganonioal form of the qirintic.
308. It was remarked that M. Hermite's investigations are conducted by means
of a canonical form, viz. if A(=J, =G) be the quartinvariant of the given quintic
(a, I, c, d, e, f$as, yf, then he in fact finds (X, F) linear functions of (a 1 , y) such that
we have
(a, 6, c, d, e, ffa, y)* = (X, & Vfc, \'Te, p', V,Y, 7)
(viz. in the transformed form the two mean coefficients are equal; this is a convenient
assumption made in order to render the transformation completely definite, rather than
an absolutely necessary one) ; and where moreover the quadricovariant E of the trans
formed form is
or, what is the same thing, the coefficients (\, /t, V&, V&, /*', \') of the transformed form
are connected by the relations
V  V Vft + 3ft = 0,
\>  V Vft + 3ft = 0,
the advantage is a great simplicity in the forms of the several oo variants, which
simplicity arises in a great measure from the existence of the very simple covarianfc
operator ~ . ~ r (viz, operating therewith on any oo variant we obtain again a covariant).
rtA d j.
309. Reversing the order of the several steps, the theory of M. Hermite's trans
formation may be established as follows :
Starting from the quintic
(a, b t o, d, e,
and considering the quadricovariant thereof
((a, /3, 7) are of the degree 2), and also the linear covariaut
Pa + Qy
, Q) are of the degree 5), we have
and moreover
(, A 75ft P)' V*
viz. the expression on the left hand, which is of the degree 12, and wind
an invariant, is = 0, where G is (ut suprti)
C. VI.
AN EIGHTH MEMOIR ON QU ANTICS. [4C
The Jncobian Df the two forms, viz.
P ,
aQ0P) +
is a linear covariant of the degree 7, say it is
and it is to be observed that the detenninaub PQ'P'Q of the two linear forms
2 (a, A 7$Q, P) 3 , that is, it is = 2(7.
310. Hence writing
whence also
the determinant of substitution from (X, Y) to (T, U) is =2, that from (2\
(a;, y) is T7t%0 t = , and consequently that from (X, Y.) to (ai, y) is 1.
We have
.42 12  E7 S = {(/3 3  407) (Pw + Q#) 2  (Fat + Q'j/) 3 } ;
1 <v
or putting for P', Q' their values, this is ~ ^ into 4 (a, /3, y^Q,  P) 2 (a* 3 + 2/3a;r/ +
that is, we have
A 2* a J7 fl = s# a + jSwy + 7#' 2 j
and we have also
AT 1  tf 3 = i VI [(,Y + F) 2  (A"  F)] = VI AT,
consequently
a a + /3a;y +>yy* = A 2" J ~U z = ^AX F.
311. We have
405] AN EIGHTH MEMOIR ON QUA.NTICS. 179
so that, pausing a moment to consider the transformation from (a, y) to (T, If), we
have
(a, b, e, d, e,f~$a, t y}^ ~^(a t b, c, d, e,flQ'TQU, 
' , c  d . e  f y > ^) r ' suppose,
where (a, b, c, d, e, f) are invariants, of the degrees 36, 34, 32, 30, 28, 26 respectively;,
it follows that b, d, f each of them contain as a factor the 18fchio invariant J, the
remaining 1 factors being of the orders 1G, 12, 8 respectively.
312. That (a, b, c, d, e, f) are invariants is almost selfevident ; it may however
be demonstrated as follows. Writing
[$} 9t + 263 ft + 3c9 rf + 4d9(, + 5e9/, = 8 suppose,
{aS) v } = 569 + 4c9 ft + Md c + W d + fd c , = 8,
then Ptc + Qy, P'<B+ Q'y being oo variants, we have SP = 0, 8Q*=P, 87 J ' = 0, SQ' = P',
whence, treating T, U as constants, 8 (Q'T QU) = FT PU,B( P'T +PU) = Q. Hence
S(a, b, G, d, e,
= 5 (a, 6, c, d, e&Q'TQU,  P'T+ PU? .(P'T+PU)
+ 6(0,6,0, d, J ) 4 ( P'TPU)
+ 5 (6, c, d, e, /J Y 0,
the three lines arising from the operation with S on the coefficients (a, b, c, d, e, f)
and on the facients Q'T QU and P'T + PU respectively; the third line vanishes of
itself, and the other two destroy each other, that is,
8 (a, I, c, d, ifUQ'TQU,  PT + P /)* = (), and similarly
8, (a, 6, c, d, e,fHQWQU, PT
or the function (a, I, c, d, e,f&Q'T~QU, PT + PU)*, treating therein T and U as
constants, is an invariant, that is, the coefficients of the several terms thereof are all
invariants.
813. The expressions for the coefficients (a, b, c, cl, e, f) are in the first
obtained in the forms
a= 2(i +6^
e
f B  2 (i 7  SJI/'O) ^l" 1
180 AN EIGHTH MEMOIR ON QU ANTICS,
where, developing SI. Hermite'a expressions,
[405
ni =
24J/=
24' =
3JJf =
A'B + 1
/V/  1
,U?7 + 1
7+1
A 6 C' 2 + 1
A 3 C  1
(77 + 6
A 5 JF + 6
AB^  3
A 4 BG  24
jt^C' + 12
A 3 & + 9
s + 24
X 3 C a  39
AJPG + 9
ABC 1 + 108
C 11 + 72
and substituting these values, we find
3Ga =
36b =
36c =
36d =
36o =
30f=
A J H + 1
/I'J/  3
A S J1 + 1
ABI  3
A*JJ + 1
7J/  3
A B C* + 1
4C/  24
A S C + 1
C 1 /"  12
A*G i 1
^l ^ 3 + 6
yi J 5 3 + G
^i 8 JS* + 6
A*J3C  39
^5(7 27
A'JIC  1C
A*& + 9
yl a j5 3 + 9
yl/J s + 9
vi 3 ^  54
yPC 2  42
ylC a  30
A*J3C  36
JIG* + 144
WG + 36
^l^C 2 + 288
C 3 + 1152
I have not thought it worth while to make in these formulae the substitutions A ~ J,
B=^K, CSL + JK, which would give the expressions for (a, b, c, d, e, f) in terms
of J, If, L.
314. Substituting for (#, i/) their values in terms of (X, Y), we have
(a, 6, c, d, e, /$>, y?
and by what precedes
this gives
and thence
v, v,
7) s suppose,
= 1204
405] AN EIGHTH MEMOIR ON QUANTICS. 181
the lefthand side is a linear covariant of the degree 5, it is consequently a mere
numerical multiple of P + Qy, and it is easy to verify that it is = 120 (Pa + Qy).
(In fact writing b=d = e = Q, the expression is (3<fld tf s afdjfi^ (aa? + lOcafy 2 \ff), and
the only term which contains x is a a / 3 .3i / s 9 a! 3 ,10ca? J y 3 = 120 a c/' a .iB ; but for &=tZ=e = 0,
Table J gives Px = a?cf z , and the coefficient 120 is thus verified.) But Pa + Qy
VO V(7
is j^r (X + Y\ and wo have thus Av Av f = :=., whence not only v = v' ) v/e
v 4 v ^
6'
suppose, but we have further fc = TT=^, a result given by M, Hermite.
V A 6
315. Substituting for v v f the value V/j, we have
(a, 6, c, d, e, /$>, ?/) fl
*
and we havo then aaP + fSwy + 72^ "JAXY, viz, the lefthand side being the quadri
covariant of (tt, &, Cj d, e, / jfls, 2/) B j the equation shows that fche quadricovariant of the
form (\ /j,, V&, V, ^', X'JX, F) 5 is =^/A.XY, and we thus arrive at the startingpoint
of Hermite's theory.
316, The coefficients (\ /t, V&, ^, /i', X') of Hermite's form are by what precedes
invariants; they are consequently expressible in terms of the invariants A, B, G (and I).
M", Hermite writes
XV==f/ f w' = h,
and he finds
or, what is the same thing,
SAB+O , _AB + , G
' ~' '
which give g, h, k in terms of A, B, G, and then putting
A = (9/c 2 + Uhk  ghy  24Aft>, == ^
(the equation J B = X 7 A is in fact equivalent to the beforementioned expression of 1
182
AN EIGHTH MEMOIR ON QUANTICS.
[405
in terms of the other invariants), the coefficients (X, /*, //, X') are expressed in terms
of y, h, k, that is of A, E, G, viz. we have
72
24
18*0"  9A (.7 + 1GA) + (0  IGfc)
1 6A A  /i  VA,
these values of (X, p, ///, X') could of course be at once expressed in terms of (./, K, L) t
but I have not thought it necessary to make the transformation.
317. It has been already noticed that the linear covariant (0, = P#hQ?/), was
= VI (Vifc, \lltQX, 7),
it is to be added that the septic covariant (P'as + Q'y) is
= VJ'(V, V/^Y, 7),
and that the canonical forms of the cubicovariants ^i (,*, y), &c. are as follows :
^(X, 7) =\'A(^ SVA, SVfc, XP'. 37.
^(Z, 7) = ^(^, V,  Jk.pfS.X, Y)\
(X, Y)} = A> fa  3 VE, 3 V& (  X5^, r),
vo>
3( V&
 3( V&
5/i, V&,
+ 3 (8 VI" Vfc  96 ( VP + /*/ Vft  2/i'fe )),
or, as the last formula may also be written,
(X, Y) = A/I f (tfff  53/i + HO/0 /*  64V V/c),^
405] AN EIGHTH MEMOIR ON QUAXTICS. 183
It is in fact by means of these comparatively simple canonical expressions that
M. Hennito was enabled to effect the calculation of the coefficient VI.
Article Nos. 318 to 326, Theory of the imaginary linear transformations which lead to
a, real equation.
318. An equation (a, b, c, .][#, y) n = is real if the ratios a : b : c, &c. of the
coefficients are all real. In speaking of a given real equation there is no loss of
generality in assuming that the coefficients (a, b, c, .,.) are all real; but if an equation
presents itself in the form (a, 6, c, ...][#, y) n ~Q with imaginary coefficients, it is to be
borne in mind that the equation may still be real ; viz. the coefficients may contain
an imaginary common factor in such wise that throwing this out we obtain an
equation with real coefficients,
In what follows I use the term transformation to signify a linear transformation,
and speak of equations connected by a linear transformation as derivable from each
other. An imaginary transformation will in general convert a real into an imaginary
equation ; and if the proposition were true universally, viz. if it were true that the
transformed equation was always imaginary it would follow that a real equation derivable
from a given real equation could then be derivable from it only by a real transfor
mation, and that the two equations would have the same character. But any two
equations having the same absolute invariants are derivable from each other, the two
real equations would therefore be derivable from each other by a real transformation,
and would thus have the same character; that is, all the equations (if any) belonging
to a given system of values of the absolute invariants would have a determinate
character, and the absolute invariants would form a system of auxiUai's.
But it is not true that the imaginary transformation leads always to an imaginary
equation ; to take the simplest case of exception, if the given real equation contains
only even powers or only odd powers of so, then the imaginary transformation a : y
into ia : y gives a real equation, And we are thus led to inquire in what cases an
imaginary transformation gives a real equation.
319. I consider the imaginary transformation a : y into
(a + W) + (o + (M)y : (e +/) + (g 4 hi) y,
or, what is the same thing, I write
co = ( + U) X + (c 4 rfi) Y,
and I seek to find P, Q real quantities such that Pu+Qy may be transformed into
a linear function RX + 8Y t wherein the ratio E : S is real, or, what is the same
thing, such that RX + SY may be the product of an imaginary constant into a real
linear function of (X, Y). This will be the case if
184 AN EIGHTH MEMOIR ON QTJANTICS. [405
that is if
which implies the relations
01; what is the same thing,
and if the resulting value of P : Q be real, the lastmentioned equations give
(ctff  ce) &  (ah + bg e/ de) + bhdf= 0,
and being known, the ratio P ; Q is determined rationally in terms of 6.
320. The equation in & will have its roots real, equal, or imaginary, according' sis
(ah + bgof c?e) a  4 (ag  ce} (bh  df),
that is
ct'fc 8 + W(f + tff* + d*tf
 Zahbff  2oAo/ Wide  2bgof
+ kadfyh 4
is =4, =0, or =.; and I say that the transformation is subiinaginary, neutral, and
superi magi nary in these three cases respectively. In the subimaginary case there are
two functions P&s + Qy which satisfy the prescribed conditions ; in the neutral case a
single function ; in the superim aginary case no such function. But in the lastmentioned
case there are two conjugate imaginary functions, Pffl + Qy, which contain as factors
thereof respectively two conjugate imaginary functions  UX + VT,
321. Hence replacing the original ce r y t X, Y by real linear functions thereof, the
subirn aginary transformation is reduced to the transformation cs \ y into JeX : K, where
k is imaginary ; and the superimaginary transformation is reduced to at + iy : co iy
into k(X+iY) : (X iY}> where k is imaginary. As regards the neutral transformation,
it appears that this is equivalent to
cs = (a + bi) X + (G + di) 7,
7,
with the condition = (aK + &/) a 4>agbh, ~ (ah  l)g}\ that is, we have ah bg~Q, or
without any real loss of generality ff = a, h = b, or the transformation is
ss(a + bi) X + (c + di) Y,
2/= (a+bi)7 t
that is, K. : y**X+kY ; 7, where k is imaginary.
5 J AN EIGHTH MEMOIR ON QUANTIC3.
of thf LJ he rigiUal eqUati n aftei ' aUy iml transformfltion ^ofi ^ till an equation
(a, ...J^ #) = Q ;
and if we consider first the neutral transformation, the transformed equation is
(a, ...QX+kY, 3 7 ) = 0j
tins is not a real equation except in tho case where k is iml.
tn T ^ A th !i  aU P erii ; m inai T transformation, starting iu like manner from
(a, ... &#, 2/) )l =0, this may be expressed in the form
(a + /3i, y + Si, . . . , 7 _ S a _ ^ j ffl _ h ^ , G _ ^ n = ^
viz. when in a real equation (, y) we make the transformation a, : into
a' + iy : 0iy, the coefficients of the transformed equation will form as above pairs of
conjugate imaginaries. Proceeding in the lastmentioned equation to make the trans
formation +;/  aiy into k(X+iY) : XiY, I throw k into the form
cos 20 + i sin 2<, = (cos <f> + i sin 0)  (cos ^  i sin <^)
(of course it is not here assumed that 4, is real), or represent the transformation as,
that of + ty :  *y into (cos + ,' s in <,!,} (A' + iY) : (co S  i sin </,) (,Y  i Y) ; the trans
formed equation thus is
+iy), (ooa ^4' sin
The lefthand side consists of terms such as (A' 2 f F)2* into
(y + K) (cos ^ + i sin s # (,Y + iY) + ( 7 _ Si) ( cos ^
viz. the expression last written down is
= (7 cos s0  S sin s <) ({ jf + 1 F) s I ( A r  i
sin ^ + 8 cos
and observing that tho expressions in [ ) are real, tho transformed equation is only
real if ( 7 cos_^  8 sin . 8 & + ( 7 sill S{/> + fi oos s</)) bo realj that  S  n oi , der fchafc fchfl ^
formed equation may be real, we must have tan ^= real; and observing that if tansd,
be equal to any given real quantity whatever, then the values of tan A are all of them
real, and that tan A real gives cos A and sin A each of them real, and therefore also A.
real, it appears that the transformed equation is only real for the brans formation
+ ;r) : (cos A  i sin A) (X  iT),
wherein ^ is real; and this is nothing else than the real transformation * : into
X OOB^  1 sin A X sin A + Y cos A. Hence neither in the case of the neutral trans
formation or in that of the superimaginary transformation cau we have an imaginary
transformation leading to a real equation y
' VI  ' ' 24
186 AN EIGHTH MEMOIR ON QUANTICS. [405
324. There remains only the subimaginary transformation, viz. this has been reduced
to at : y into kX : Y, the transformed equation is
17=0,
and this wilt be a real equation if some power & of * (p not greater than ) be
real, and if the equation (,..$*. 2/) = contain only terms wherem the index of a
(or that of ) is a multiple of p. Assuming that it is the index of y winch is a
multiple, the form of the equation is in fact (^, 2/*) w =0, (nmp + a), and the
transformed equation is ,Y(MP 7^ = 0, which is a real equation.
325 It is to be observed that if p be odd, then writing &=K (K real) and
taking k 1 the real jith root of K, then the very same transformed equation would
be obtained by the real transformation at : y into Jc'X : F; so that the equation
obtained by the imaginary transformation, being also obtainable by a real transfor
mation, has the same character as the original equation.
326. Similarly if p be even, if K be real and positive, the equation &P = K has
a real root jf which may be substituted for the imaginary k t and the transformed
equation will have the same character as the original equation ; but if K be negative,
say /f = l (as may be assumed without loss of generality), then there is no real
transformation equivalent to the imaginary transformation, and the equation given ^ by
the imaginary transformation has nob of necessity the same character us the original
equation; and there are in tact cases in which the character is altered. Thus if p = 2,
and the original equation he x(tf, f) m = Q, or (a?, py=Q, then making the transfor
mation re : y into iX : F, the transformed equation will be X (X\  Y*) m = or
(X*,  P)" 1 = 0, giving imaginary roots X* + of* = corresponding to real roots s?  aif = 0.
Article No. 327. Application to the availiars of a quintic,
327. Applying what precedes to a quiutic equation (a,. . . .J0, 2/) B =0> tu is after
any real transformation whatever will assume the form (of,. . .Jo 7 , 2/') 6 = 0; f 1 ^ ne on ty
cases in which we can have an imaginary transformation producing a real equation of
an altered character is when this equation is (a', 0, o', 0, e' t OJJV, i/') 5==0 (cf not =0), or
when it is (a 1 , 0, 0, 0, d, OjJV, yJ = Q, viz. when it is a/( a '' 4 + 1( Wy a + 5y) = 0, or
*(oV 4 + 5e'y' 4 )=0, In the latter case the transformation to', y' into X\/ l\Y gives the real
equation Z(a'Jf 4 5eT 4 )=0. I observe however that for the form 0, 0, 0, e', OJX 2/) 4 ,
and consequently for the form (a, ... $<B, i/) B from which it is derived we have
J = ; this cose is therefore excluded from consideration, The remaining case is
(a', 0, o', 0, e', 0][a)', y 1 } 6 = 0, which is by the imaginary transformation of : y' into iX : Y
converted into (a', 0, c', 0, e', OjZ, 7) 5 = 0; for the first of the two forms we have
J' = 16a'cV a ) and for the second of the two forms J"= 16a'oV 9 , that is, the two values
of J have opposite signs. Hence considering an equation (a, I, c, d, e, /}[ffi, /)" = for
which J is not = 0, whenever this is by an imaginary transformation converted into
a real equation, the sign of J is reversed ; and ifc follows that, given the values of
the absolute invariants and the value of J (or what is sufficient, the sign of J), the
405]
AN EIGHTH MEMOIR ON QUANTICS.
187
different real equations which correspond to these data must be derivable one from
another by real transformations, and must consequently have a determinate character ;
that is, the Absolute Invariants, and J, constitute a system of auxiliars.
ANNEX. Analytical Theorem in relation to a Smart/ Qituntw of any Order.
The foregoing theory of the superimaginary transformation led me to a somewhat
remarkable theorem. Take for example the function
or, ns this may "be written,
(a, b, c$ir, + k, 1  Aw) 3 ,
ft or ( c, 26,
26, 2  2c,
ft,  26,
a
26
c
a 3 c, 26,
oj 26, 2ft  2c,  26
1 ft, 26, c,
then the determinant
c, 26, a
26, 2ft  2c,  26
ft, 26, o
is a product of linear functions of the coefficients (a, b, c); its value in fact is
=  2 (ft + c) (ft + 2W + ci a ) (a  26i + a a ), =  2 (a + c) [(ft  c) 2 + 46 9 ].
To prove bhis directly, I write
a a
ct' J ,
and \ve then, have
c' re 4
( c, 26, ,
(26, 2ft 2c, 
( a, 26,
c,
26,
26)
26, a 1, 2 , 1
2a2c, 26 i, , i
' ^V, fr , ~~ ' J\t , 1
n ? ^ ^9 o 9^1 n / ? 3 \
V J i t, t y, ^, u, ^t ) } \i t k, i> )
2m' f 06',  2j'c' 2, ,  2i
ft',, 26', c' : 1, 2 , 1
242
188 AN EIGHTH MEMOIR ON QU ANTICS.
whence observing that the determinants
1,
2 , 1
i\ 21*
i a
,
"
W j ^^ CH
2i, ,
a
&,
 2i a , i 2
1, 2 ,
i
are as 1 : 2, we have the required relation,
c,
2ft,
26,
26
c
=  2'6V, =  2 (a + c) {(a ~ c) 2 + 4i 3 }.
It is to be remarked that the determinant
2 , 1
, i
, taken as the multiplier of
c,
26,
26, 2a2c, 26
a,  26, c
[405
is obtained by writing therein = 6 = C( =1; and multiplying the successive lines
thereof by 1, #, ^ (1, ^ 1 are the reciprocals of the binomial coefficients 1, 2, 1), the
proof is the same, and the multiplier is obtained in the like manner for a function
of any order; thus for the cubic (a, fr, Cj dp + 0, 1 &),
= a. 3
a; 2
iw
1
rf> Sc, 36, a
3c, 66^3c^ J 3ft 6c t 36
36, SrtGc, 86St?, 3c
a, 36, So, (2
the multiplier is obtained from the determinant by writing therein o = 6 = o = d=l >
and multiplying the successive lines by 1, fa tf, $, viz. the multiplier is
1, 3, 3, 1
i, *, , i
i\ ~P, i\ i?
and the value of the determinant is found to be
 9 ((a
405] AN EIGHTH MEMOIB ON QUANTICS. 189
But the theory may be presented under a better form ; take for instance the cubic,
vis;, writing ' and j for x and k respectively, we have (a, b, c, d$ky + lx, Iy  k) a
i/ ...
d, 3c,  36, a
3c,  66 + 3d, 3a  So, 36
36, 3a6c, 66  3d, 3c
f a, 36, So,
a bipartite cubic function (#][/<;, l) s (x> 2/) 3 ! an( ^ tne determinant formed out of the
matrix is at once seen to be an invariant of this bipartite cubic function.
Assume now that we have identically
viz. this equation written under the equivalent form
(a f , 6', o', d'\X, Y) 3 = (, 6, c, dQX + Y, i (X  F)) 3 ,
determines (a', b', G', d') as linear functions of (o, 6, c, d), it in fact gives
of = (a, 6, c, dl,  i) 3 = a  36t 4 3ci fl  di 3 ,
&' = (, 6, c, c^l,  t) a (1, t) = a  6i ci 2 4 dt 3 ,
o' = (a, 6, c, d$l, i) (1, *) a ~a4' &i~ ci a di 3 ,
then observing that % + fo? i (fy  few) = ( *y) (T ^ + 0. we liave
(a, 6, c, d$7c2/ 4 lx, Iy kaf^ = (of, b', c', d'$ (a) 4 iy) ( ik + 1), $ (so  iy} (ilc + I)) 3 ,
and if in the expression on the righthand side we make the linear transformations
V2,  ik + 1 = k' V2,
iy~ iy' 2, ifc + i =  2 f
which are respectively of the determinant 4 1, the transformed function is
= (a', 6', c', d'&V, iyy,
that is, we have
(a, b, c, djty + to, ^feBy = (tt', 6', o', d'Jfe'a/, 2Y) 3 .
190 AN EIGHTH MEMOIR ON QU ANTICS.
The lastmentioned function is
[405
k'*
I"
. 36'
d 1
and (from the invariantive property of the determinant) the original determinant in
equal to the determinant of this new form, viz. we have
 d t
3o,
36,
a
So,
Qb +
3d, 3a 
Gc,
36
36,
8a
Gc, G&
3d,
3c
a,
3&,
8c,
d
[(a  3c) 2 + (36  rf) a ] [(a + c) 3 + (6 + rf)' J ] ,
which is the required theorem. And the theorem is thus exhibited in its fcrur
connexion, as depending on the transformation
(a, ...TJX y) n = (',
ADDITION, Tift October, 1867.
Since the present Memoir was written, there has appeared the valuable pnpor by
MM, Clebsch and Gordan "Sulla rappresentazione tipica delle forme binarie," Anmdi
di Matematica, fc, I. (18(i7) pp. 23 79, relating to the binary quinfcic and sextic. On
reducing to the notation of the present memoir the formula 95 for the representation
of the quintic in terms of the eovariants a, /3, which should give for (a, b, c, d, o, f)
the values obtained ante, No. 312j I find a somewhat different system of values; viz.
these are
36a =
36b =
36o =
36d =
36e =
35f
A 7 + 1
*/I 4 /  1
A + 1
*A 3 I ~ I
A S J1 + 1
*^' J /  1
A 6 C + \
A*BI 3
A B C + 1
ABI  3
^(7. + 1
ASf 3
A 6 JP + 6
*AGI + 24
jl'J? 2 + 6
*CI + 12
A 3 JP + 6
A*SC  39
A 3 $C  27
/ 9 B(7  15
A*B 3 4 9
^J 1 + 9
AJP + 9
A*C*  54
^(7 2  42
AC* ~ 30
A*JPC 126
*AffC 90
'^C  64
AJJC* + 288
flC 1 '' +144
G a +1162
where I have distinguished with an asterisk the terms which have different coefficient*
in the two formulas, I cannot at present explain this discrepancy.
Plate IH
Mc/.Z.
The lower cusp of (he Bicomis drawn out of its true fosilion,
which ismuchfitttheroff(dorigtfieasyji/j)Me.,l}iec06rcliiiates
in fact are x=76j$, y^<!4i (tlwcoordiiudts of the. upper
or nod&cusp'beiny 1, 1.
\*
,0.,
Ccuyley's Papers. VI
406]
406.
ON THE CUBVES WHICH SATISFY GIVEN CONDITIONS.
[From the Philosophical Transactions of the Royal Society of London, vol. CLVIIL (for the
year 1868), pp. 75143. Received April 18, Read May 2, 1867.]
THE present Memoir relates to portions only of the subject of the curves which
satisfy given conditions; but any other title would be too narrow: the question chiefly
considered is that of finding the number of the curves which satisfy given conditions ;
the curves are either curves of a determinate order r (and in this case the conditions
chiefly considered are conditions of contact with a given curve), or elsi3 the curves are
conies; and here (although the conditions chiefly considered are conditions of contact
with a given curve or curves) it is necessary to consider more than in the former
case the theory of conditions of any kind whatever. As regards the theoiy of conies,
the Memoir is based upon the researches of Chasles and Zeuthen, as regards that of
the curves of the order ?, upon the researches of De Jonquieres: the notion of the
quasigeometrical representation of conditions by ' means of loci in hyperspace ^ is
employed by Salmon in his researches relating to the quadric surfaces which satisfy
given conditions. The papers containing the researches referred to are included in the
subjoined list. I reserve for a separate Second Memoir the application to the present
question, of the Principle of Correspondence.
List of Memoirs and Works relating to the Curves which satisfy given conditions,
remarks.
De Jonquieres: "The'oremes generaux concernant les courbes gdomdtriques planes
d'un ordre quelconque," Liouv. t. vi. (1861), pp. 113134. In this valuable memoir is
established the notion of a series of curves of the index JV; viz. considering the curves
of the order n which satisfy in( + 3)l conditions, then if N denotes how many
there are of these curves which pass through a given arbitrary point, the series is
said to be of the index N,
In Lemma IV it is stated that all the curves G u of a series of the index JV can
192
ON THE CUBVES WHICH SATISFY GIVEN CONDITIONS. [406
be analytically represented by an equation F(y, &)=$, which is rational and integral
of the decree N in regard to a variable parameter X : this is not the case ; see
Annex No. 1.
Chasles: Various papers in the Gomptes Rendw, t. LVlli. et seq. 1864 67, The
first of them (Feb. 186*), entitled " Determination du nombre des sections coniques
qui doivent toucher cinq courbea donnees d'ordre queloonque, on satisfaire a diverses
autres conditions," establishes the notion of the two characteristics (/t, v) of a system
of conies which satisfy four conditions ; viz. /t is the number of these conies which
pass through a given arbitrary point, and v the number of them which touch a given
arbitrary line. The Principle of Correspondence for points on a line is established in
the paper of Jime July 1864. Many of the leading points of the theory are repro
duced in the present Memoir. The series of papers includes one on the conies in space
which satisfy seven conditions (Sept. 1865), and another on the surfaces of the second
order which satisfy eight conditions (Feb. I860).
Salmon: "On some Points in the Theory of Elimination," Quart, Math. Journ.
t. vn. pp. 327337 (Feb. 1866); "On the Number of Surfaces of the Second Degree
which can be described to satisfy nine Conditions," Ibid. t. vm. pp. 1 7 (June 18(i6),
which two papers are here referred to on account of the notion which they establish
of the quasigeometrical representation of conditions by means of loci in hyperspaco.
Zeuthen: Nyt Bidrag... Contribution to the Theory of Systems of Conies which
satisfy four conditions, 8. pp. 1 97 (Copenhagen, Cohen, 1865), translated, with an
addition, in the Nouvelles Annalss.
The method employed depends on the determination of the linepairs and point
pairs, and of the numerical coefficients by which these have to be multiplied, in the
several systems of conies which satisfy four conditions of contact with a given curve
or curves. It is reproduced in detail, with the enumeration called "Zeutheu's Capitals,"
in the present Memoir.
Cayley: "Sur les coniques de'termine'es par cinq conditions d'intersection avec une
courbs donne'e," Gomptes Ecndus, t. LXIII. pp. 912, July 18G6. Results reproduced in
the present Memoir.
De Jonquieres: Two papers, Gomptes Rendtis, t. LXIII. Sept. 186G, reproduced and
further developed in the "Me'moire sur les contacts multiples d'oidre quelcouquo des
courbes du degre* r qui satisfont a des conditions. clonne'es de contact avec une courbo
fixe du degre' m; suivi de quelques reflexions sur la solution d'un grand nombre do
questions concernant les proprie'te's projectives des courbes et des surfaces alge"bricues,"
Qrelle, t. LXVI. (1866), pp. 289 322, contain a general formula for the number of curves
& having contacts of given orders a, 1), c, . , with a given curve U m , which formula
is referred to and considered in the present Memoir,
De Jonquieres: Recherches sur les aeries ou systemes de courbes et do surfaces
algtSbriques d'ordre quelconque ; suivies d'une reponse &c. 4. Paris, Gauthier Villars,
1866 (').
1 The foregoing list is not complete, ami the remarks aro not intended to give even a sketch of the eon
tents of the worts comprised therein, but only to show their bearing on the present Memoir.
406] ON THE CURVES WHICH SATISFY GIVEN CONDITIONS. 193
Article Nos. 1 to 23. On the quasigeometrical representation of Conditions.
1. A condition imposed upon a subject gives rise to a relation between the
parameters of the subject; for instance, the subject may be, as in the present Memoir,
a plane curve of a given order, and the parameters be any arbitrary parameters con
tained in the equation of the curve. The condition may be onefold, twofold,... or,
generally, kfold, and the corresponding relation is onefold, twofold,... or Afold accord
ingly. Two or more conditions, each of a given manifoldness, may be regarded as
forming together a single condition of a higher manifoldness, and the corresponding
relations as forming a single relation ; and thus, though it is often convenient to con
sider two or. more conditions or relations, this case is in fact included in that of a
/cibld condition or relation. In dealing with such a condition or relation it is assumed
that the number of parameters is at least =&; for otherwise there would not in
general be any subject satisfying the condition; when the number of parameters is
Ic, the number of subjects satisfying the condition is in general determinate.
2. A subject which satisfies a given condition may for shortness be termed a
solution of the condition ; and in like manner any set of values of the parameters
satisfying the corresponding relation may be termed a solution of the relation. Thus
for a &fold condition or' relation, and the same number k of parameters, tho number
of solutions is in general determinate.
3. A condition may in some cases be satisfied in more than a single way, and
if a certain way be regarded as the ordinary and proper one, then the others are
special or improper: the two epithets may be used conjointly, or either of them
separately, almost indifferently. For instance, the condition that a curve shall touch a
given curve (have with it a twopointic intersection) is satisfied if the curve have
with the given curve a proper contact; or if it have on the given curve a node or
a cusp (or, more specially, if it be or comprise as part of itself two coincident curves) ;
or if it pass through a node or a cusp of the given curve : the first is regarded as
the ordinary and proper way of satisfying the condition ; the other two as special or
improper ways ; and the corresponding solutions are ordinary and proper solutions, or
special or improper ones accordingly. This will be further explained in speaking of
the locus which serves for the representation of a condition.
4. A set of any number, say <a, of parameters may be considered as the coordi
nates of a point in wdimensional space ; and if the parameters are connected by a
onefold, twofold,.,, or kfold relation, then the point is situate on a onefold, twofold,.,,
or Ar fold locus accordingly ; to the relation made up .of two or more relations corresponds
tho locus which is the intersection or common locus of the loci corresponding to the
several component relations respectively. A locus is at most wfold, viz. it is in this
case a pointsystem. The relation made up of a Mold relation, an 2fold relation, &c., is
in general (k + 1 4 fee.) fold, and the corresponding locus is (& + I + &c.) fold accordingly.
5. The order of a pointsystem is equal to the number of the points thereof,
where, of course, coincident points have to be attended to, so that the distinct points
of the system may have to be reckoned each its proper number of times. The locus
c. VI. 25
194 ON THE CURVES WHICH SATISFY GIVEN CONDITIONS. [406
corresponding to any linear jfold relation between the coordinates is said to be a jfold
omal locus; and if to any given /cfold relation we join an arbitrary (a A) fold linear
relation, that is, intersect the ftfold locus by an arbitrary (w  k) fold omal locus, so as
to obtain a pointsystem, the order of the &fold relation or locus is taken to be
equal to the number of points of the pointsystem, that is, to the order of the point
system. And this being so, if a ftfold relation, an Jfold relation, &c. are completely
independent/ that is, if they are not satisfied by values which satisfy a less than
(k + l+ &c.) fold relation, or, what is the same thing, if the fcfold locus, the Zfold
locus, &c., have no common less than (k + 1 + &c.) fold locus, then the relations make
up together a (jfc + 1+ &c.) fold relation, and the loci intersect in a (k \ 1 + &c.) fold
locus, the orders whereof are respectively equal to the product of the orders of the
given, relations or loci. In particular if we have A+Z + &C. o>, then we have an
wfold relation, and corresponding thereto a pointsystem, the orders whereof are
respectively equal to the product of the orders of the given relations or loci.
6. A Mold relation, an Zfold relation, &c., if they were together equivalent to a
less than (k + 1 + &c.) fold relation, would not be independent; but the relations, assumed
to be independent, may yet contain a less than (& + Z + &o.)fold relation, that is, they
may be satisfied by the values which satisfy a certain less than (k + I} &c.) fold relation
(say the common relation), and exclusively of these, only by the values which satisfy
a proper (fc + 1 + &c.) fold relation, which is, so to speak, a residual equivalent of the
given relations. This is more clearly seen in regard to the loci; the &fold locus, the
Jfold locus, &c. may have in common a less than (k + 1 + &o.) fold locus, and besides
intersect in a residual (k 4 1 + &o.) fold locus. (It is hardly necessary to remark that
such a connexion between the relations is precisely what is excluded by the foregoing
definition of complete independence.) In particular if & + J+&Q. = to, the several loci
may intersect, say in an (a) j) fold locus, and besides in a residual wfold locus, or
pointsystem. The order (in 'any such case) of the residual relation or locus is equal
to the product of the orders of the given relations or loci, less a reduction depending
on the nature of the common relation or locus, the determination of the value of
which reduction is often a complex and difficult problem,
7. Imagine a curve of given order, the equation of which contains w arbitrary
parameters ; to fix the ideas, it may be assumed that these enter into the equation
rationally, BO that the values of the parameters being given, the curve is uniquely
.determined. Suppose, as above, that the parameters are taken to be the coordinates
of a point in dimensional space ; so long as the curve is not subjected to any
condition, the poiut in question, say the parametric point, is an arbitrary point in the
CD dimensional space; bub if the curve be subjected to a onefold, twofold,,., or /cfold
condition, then we have a onefold, twofold,,., or Mold relation between the parameters,
and the parametric point is situate on a onefold, twofold,... or &fold locus accordingly:
to each position of the parametric poiut on the locus there corresponds a curve
satisfying the condition, that is, a solution of the condition. In the case where the
condition is wfold, the locus is a pointsystem, and corresponding to each point of
the pointsystem we have a solution of the condition ; the number of solutions is
equal to the number of points of the pointsystem.
406] ON THE CURVES WHICH SATISFY GIVEN CONDITIONS. 195
8. Considering the general case where the condition, and therefore also the locus,
is Mold, it is to be observed that every solution whatever, and therefore each special
solution (if any), corresponds to some point on the Mold locus; we may therefore have
on the Mold locus what may be termed "special loci," viz. a special locus is a locus
such that to each point thereof corresponds a special solution. A special locus may of
course be a pointsystem, viz. there are in this case a determinate number of special
solutions corresponding to the several points of this pointsystem. "We may consider
the other extreme case of a special Afold locus, viz. the Mold locus of the parametric
point may break up into two distinct loci, the special Mold locus, and another Mold
locus the several points whereof give the ordinary solutions: we can in this case get
rid of the special solutions by attending exclusively to the lastmentioned Mold locus
and regarding it as the proper locus of the parametric point. But if the special locus
be a more than Mold locus, that is, if it be not a part of the Mold locus itself, but
(as supposed in the first instance) a locus on this locus, then the special solutions cannot
be thus got rid of: we have the Mold locus of the parametric point, a locus such
that to every point thereof there corresponds a proper solution, save and except that
to the points lying on the special locus there correspond special or improper solutions.
It is to be noticed that the special locus may be, but that is not in every case, a
singular locus on the Mold locus.
9. Suppose that the conditions to be satisfied by the curve are a Mold condition,
an Zfold condition, &c. of a total manifoldness =ro. If the conditions are completely
independent (that is, if the corresponding relations, ante, No. 5, are completely indepen
dent), we have a &fold locus, an /fold locus, &c,, having no common locus other than
the pointsystem of intersection, and the number of curves which satisfy the given
conditions, or (as this has been before expressed) the number of solutions, is equal to
the number of points of the pointsystem, or to the order of the pointsystem, viz. it
is equal to the product of the orders of the loci which correspond' to the several con
ditions respectively; among these we may however have special solutions, corresponding
to points situate on the special loci upon any of the given loci; but when this is
the case the number of these special solutions can be separately calculated, and the
number of proper solutions is equal to the number obtained as above, less the number
of the special solutions,
10. If, however, the given conditions are not completely independent (that is, if
the corresponding relations are not completely independent), then the Mold locus,
the 2fold locus, &o, intersect in a common (o>j)fold locus, and besides in a residual
pointsystem. The several points of the (w  j) fold locus give special solutions in fact
the very notion of the conditions being properly satisfied by a curve implies that the
curve shall satisfy a true (k + 1 + &o.) fold, that is, a true cofold condition ; the proper
solutions are therefore comprised among the solutions given by the residual point
system, and the number of them is as before equal to the order of the pointsystem,
or number of the points thereof, less the number of points which give special solutions:
the order of the pointsystem is, as has been seen, equal to the product of the orders
of the Mold locus, the Zfold locua, &o,, less a reduction depending on the nature of
252
19G ON THE CURVES WHICH SATISFY GIVEN CONDITIONS. [406
the common (a>j)fold locus, and the difficulty is in general in the determination of
the value of this reduction.
11. In all that precedes, the number of the parameters has been taken to bo o> ;
but if the parameters are taken to be contained in the equation of the carve homo
geneously, then the parameters before made use of are in fact the ratios of these
homogeneous parameters ; and using the term henceforward as referring to the homo
geneous parameters, the numbers of the parameters will be =o + 1.
12. I assume also that the equation of the .curve contains the parameters linearly ;
this being so, the condition that the curve shall pass through a given arbitrary point
implies a linear relation between the parameters; and the condition that the curve
shall pass through j given points, a jfold linear relation between the parameters. It
follows that the number of the curves which satisfy a given /cfold condition, and besides
pass through m k given points, is equal to the order of the ftfold relation, or of the
corresponding jfcfold locus ; and thus if we define the order of the /cfold condition to bo
the number of the curves in question, the condition, relation, and locus will be all of
the same order, and in all that precedes we may (in place of the order of the relation
or of the locus) speak of the order of the condition. Thus, subject to the modifications
occasioned by common loci and special solutions as above explained, the order of the
(k f I + &o.) fold condition made up of a ft fold condition, an 2fold condition, &c., is
equal to the product of the orders of the component conditions; and in particular if
k + 1 + &c. = a), then the order of the wfold condition, or number of the solutions thereof,
is equal to the product of the orders of the component conditions.
13. The conditions to be satisfied by the curve may be conditions of contact with
a given curve or curves. In particular if the curve touch a given curve, the para
metric point is then situate on a onefold locus. It is to be noticed in reference horobo
that if the given curve have nodes or cusps, then we have special solutions, viz. if
the sought for curve passes through a node or a cusp of the given curve ; and each
such node or cusp gives rise to a special onefold locus, presenting itself in the first
instance as a factor of the onefold locus of the parametric point ; this is, however, a
case where the special locus is of the same mauifoldness as the general . locus (ante,
No. 8), and is consequently separable; throwing off therefore all these special loci, we
have a onefold locus which no longer comprises the points which correspond to curves
passing through a node or a cusp of the given curve; the onefold locus, so divested
of the special onefold factors, may be termed the "contactlocus" of the given ourvo.
To each point of the contactlocus there corresponds a curve having with the given
curve a twopointic intersection, viz. this is either a proper contact, or it is a special
contact, consisting in that the sought for curve has on the given curve a node or
cusp, or (which is a higher speciality) in that the sought for curve is or contains as
part of itself two or more coincident curves (ante. Nn. RY Tn n. nnint. nnwil an
406 ] ON THE CURVES WHICH SATISFY GIVEN CONDITIONS, 197
the order r which touch a given curve of the order in and class n, the order of the
contactlocus is = n 4 (2r  2) m.
14. If, then, the curve touch a given curve, the parametric point is situate on the
contactlocus of that curve. If it touch a second given curve, the parametric point is
in like manner situate on the contactlocus of the second given curve, that is, it is
situate on the twofold locus which is the intersection of the two contactloci; and the
like in the case of any number of contacts each with a distinct given curve. But if the
curve, instead of ordinary contacts with distinct given curves, has either a contact of
the second, or third, or any higher order, or has two or more ordinary or other contacts
with the same given curve, then if the total manifoldncss be =k t the parametric point
is situate on a Afold loous, which is given as a singular locus of the proper kind on
the onefold contactlocus ; so that the theory of the contactlocus corresponding to the
case of a single contact with a given curve, contains in itself the theory of any
system whatever of ordinary or other contacts with the same given curve, viz. the
lastmentioned general wise depends on the discussion of the singular loci which lie on
the contact loons, And similarly, if the curve has any number of ordinary or other
contacts with each of two or more given curves, wo have here to consider the inter
sections of singular loci lying on the contaotloei which correspond to the several given
curves respectively, or, what is the same thing, to the singular loci on the intersection
of these contactloci; that is, the theory depends on that of the contactloci which
belong to the given curves respectively.
15. Suppo.se that the curve which has to satisfy given conditions is a line; the
equation is ats \by + cz Q, and the parameters (a, b, c) are to be taken as the
coordinates of a point in a plane. Any onefold condition imposed upon the lino
establishes a onefold relation between the coordinates (a, b, c), and the parametric point
is situate on a curve; a second onefold condition imposed on the line establishes a
second onefold relation between the coordinates (, 6, c), and the parametric point is
thus situate on a second curve; it is therefore determined as a point of intersection
of two ascertained curves. In particular if the condition imposed on the lino is that
it shall touch a given curve, the locus of the parametric point is a curve, the con
tactlocus; (this is in fact the ordinary theory of geometrical reciprocity, the locus in
question being the reciprocal of the given curve ;) and the caae of the twofold condition
of a contact of the second order, or of two contacts, with the given curve, depends
on the singular points of the contactlocus, or reciprocal of the given curve ; in fact
according as the line has a contact of the second order, or has two contacts with the
given curve (that is, as it is an inflexion tangent, or a double tangent of the given
curve), the parametric point is a cusp or a node on its locus, the reciprocal curve: this
is of course a fundamental notion in the theory of reciprocity, and it is only noticed
here in order to show the bearing of the remark (ante, No. 14) upon the case now
in hand where the curve considered is a line.
16. If the curve which has to satisfy given conditions is a conic
(a, b, o, f, ff, Jigai, y, *) = 0,
we have here six parameters (ft, b t c, f, g, Ji) f which are taken as the coordinates of a
11)8 ON THE CURVES WHICH SATISFY GIVEN CONDITIONS. [406
point in 5 dimensional space. It may be remarked that in this 5dimensional space we
have the onefold cubic locus ale  a/ 3  Iff  c# t 2/</A, = 0, which is such that to any
position of the parametric point upon it there corresponds not a proper conic but a
linepair; this may be called the discriminant locus. We have also the threefold locus
the relation of which is expressed by the six equations
(Jo_/ = 0, ca(f = Q, aiA a = 0, ghaf=Q, 1tflg = Q, fg G h=*Q) t
which is such that to any position of the parametric point thereon, there corresponds
not a proper conic but a coincident linepair. I call this the BipointIocus( 1 ), and
I notice that its order is = 4 ; in fact to find the order we must with the equations
of the Bipoint combine two arbitrary linear relations,
(* ~$a, b, c, f, g, A)  0,
(*'$, 6, c,/ g, A)i=0;
the equations of the locus are satisfied by
a : & : c ; / : g : h = a a : {3* ; 7' : 7 ; yet : a/3
(where : /3 : 7 are arbitrary) ; and substituting these values in the linear relations,
we have two quadric equations in (a, /?, 7), giving four values of the set of ratios
(a : /3 : 7) ; that is, the order is = 4, or the Bipoint is a threefold quadric locus.
17. The discriminantlocus does not in general present itself except in questions
where it is a condition that the conic shall have a node (reduce itself to a linepair) ;
thus for the conies which have a node and touch a given curve (?ft, n\ or, what is the
eame thing, for the linepairs which touch a given curve (m, n), the parametric point is
here situate on a twofold locus, the intersection of the discriminantlocus with the con
tactlocus. It may be noticed that this twofold locus is of the order 3 (n + 2m), but
that it breaks up into a twofold locus of the order 3?i, which gives the proper solutions;
viz. the nodal conies which touch the given curve properly, that is, one of the two
lines of the conic touches the curve ; and into a twice repeated twofold locus of the
order 3m which gives the special solutions, viz. in these the nodal conio lias with the
given curve a special contact, consisting in that the node or intersection of the two
lines lies on the given curve. By way of illustration see Annex No, 2. But the con
sideration of the Bipointlocus is more frequently necessary.
18. Suppose that the conic satisfies the condition of touching a given curve; the
parametric point is then situate on a onefold contactlocus (a, &, c, /, g, A)* = (to fix
the ideas, if the given curve is of the order m and class n, then the order q of the
contactlocus is = ?i+ 2m). The contactlocus of any given, curve whatever passes
through the Bipointlocus; in fact to each point of the Bipointlocus there corresponds
a coincident linepair, that is, a conic which (of course in a special sense) touches the
given curve whatever it be; and not only so, but inasmuch as we have a special
1 In framing tho epithet Bipoint, the coincident Huepair is regarded as being really a pointpair: aeo
jwit, No. 30. iii
406] ON THE CURVES WHICH SATISFY GIVEN CONDITIONS. 199
contact at each of the points of intersection of the given curve with the coincident
linepair regarded as a single line, that is, in the case of a given curve of the mth
order, m special contacts, the Bipoiutlocus is a multiple curve on the corresponding
contactlocus.
19. If the conic has simply to touch a given curve of the order m^ and class n 1} then
the order of the condition (or number of the conies which satisfy the condition, and
besides pass through four given points) is equal to the order of the contactlocus, that
is, it is =n 1 +2m 1 . If the conic has also to, touch a second given curve of the order
m* and class a , then the order of the twofold condition (or number of the conies
which satisfy the twofold condition, and besides pass through three given points) is
equal to the order of the intersection or common locus of the two contactloci; and
these being of the orders w, + 2^ and 2 + 2m 3 respectively, the order of the intersection
and therefore that of the twofold condition is =(n, + 2m 1 )(nj + 2m a ). But in the next
succeeding case it becomes necessary to take account of the singular locus.
20. If the conic has to touch three given curves of the order and class (m^ %),
(Mm a), (m a , ih) respectively, we have here three contactloci of the orders nj^^m,,
s+2ma, n 8 + 2m 8 respectively; these intersect in a threefold locus, but since each of
the contactloci passes through the threefold Bipointlocus, this is part of the intersection
of the three contactloci; and not only so, but inasmuch as they pass through the
Bipointlocus 7^, m a , m, times respectively, the Bipointlocus must be counted m 1 9n a m 3
times, and its order being =4, the intersection of the contactlocus is made up of the
Bipoint reckoning as a threefold locus of the order ^TH^BO and of a residual three
fold locus of the order
= iWas + 2 (njngWia + &c.) 4 4t (?? a THjm, + &c.)
and the order of the threefold condition (or number of the conies which touch the
three given curves, and besides pass through two given points) is equal to the order
of the residual threefold locus, and has therefore the value just mentioned.
21. In going on to the cases of the conies touching four or five given curves,
the same principles are applicable; the contactloci have the Bipoint (a certain number
of times repeated) as a common threefold locus, and they besides intersect in a residual
fourfold or (as the case is) fivefold locus, and the order of the condition is equal to
the order of this residual locus; but the determination of 'the order of the residual locus
presents the difficulties alluded to, ante, No. 10. I do not at present further examine
these cases, nor the cases of the conies which have with a given curve or curves
contacts of the second or any higher order, or more than a single contact with the
same given curve,
22, The equation of the conic has been in all that precedes considered as con
taining the six parameters (a, b, o, f> g, Jt) ; but if the question as originally stated
relates only to a class of conies the equation whereof contains linearly 2, 3, 4, or 5
parameters, or if, reducing the equation by means of any of the given conditions, it
200
ON THE CURVES WHICH SATISFY GIVEN CONDITIONS. [406
can be brought to the form in question, then in the latter case we may employ the
equation iu such reduced form, attending only to the remaining conditions; and in
either case we have the equation of a conic containing linearly 2, 3, 4, or 5 parameters,
which purametera are taken as the coordinates of a point in 1, 2, 3, or 4dimensional
space, and the discussion relates to loci in such dimensional space. This is in fact
what is done in Annex No. 2 above referred to, where the conies considered being
the conies which pass through three given points, the equation is taken to be
fi}3+gz.v+hxy = Q, and we have only the three parameters (/, g, A); and also in
Annex No. 3, where the conies pass through two given points, and are represented by
an equation containing the four parameters (a, o, c, A): I give this Annex as a ^ some
what more elaborate example than any which is previously considered, of the application
of the foregoing principles, and as an investigation which is interesting for its own
sake. See also Annexes 4 and 5, winch contain other examples of the theory. The
remark as to the number of parameters is of course applicable to the case where the
curve which satisfies the given conditions is a curve of any given order r; the
number of the parameters is here at most = ()' + 1) (r + 2), and the space therefore
at most r(rh3) dimensional; but we may in particular cases have <u + l parameters,
the coordinates of a point in a dimensional space, where w is any number less than
ir(r + 3).
23. I do not at present consider the case of a curve of the order r, or further
pursue these investigations; my object has been, not the development of the foregoing
quasigeometrical theory, so as to obtain thereby a series of results, but only to sketch
out the general theory, and in particular to establish the notion of the order of con
dition, and to show that, as a rule (though as a rule subject to very frequent exceptions),
the order of a compound condition is equal fco the product of the orders of the
component conditions, The last mentioned theorem seems to me the true basis of the
results contained in a subsequent part of this paper in connexion with the formula*
of De Jonquieres, post, No. 74 et seq. But I now proceed to a different part of the
general subject
Article Nos. 24 to 72, Reproduction and Development of the Researches of
CHABLES and ZEUTHEN.
24. The leading points of Ghasles's theory are as follows: he considers the conies
which satisfy four conditions (&), and establishes the notion of the characteristics
(fi t v} of such a system, viz. p, = (4X ), denotes the number of conies in the system
which paas through a given (arbitrary) point, and v t ==(417), the number of conies in
the system which touch a given (arbitrary) line. We may say that p is the parametric
order, and v the parametric class of the system.
25. The conies
CO, (.'./X O//). ('//A (////)
which pass through four given points, or which pass through three given points and
touch a given line, &a, ... or touch four given lines, have respectively the characteristics
(1, 2), (2, 4), (4, 4), (4, 2), (2, 1).
406] ON THE CURVES WHICH SATISFY GIVEN CONDITIONS. 201
26. A single condition (A') imposed upon a conic lias two representative numbers,
or simply representatives, (a, ) ; viz. if (4#) be an arbitrary system of four conditions,
and (& v) the characteristics of (42), then the number of the conies which satisfy
the five conditions (X, 4>Z) is ~ ap, + /3v.
27. As an instance of the use of the characteristics, if X, X', X", X"' t X"" be
any five independent conditions, and (a, /3), ... (a"", /3"") the representatives of these
conditions respectively, then the number of the conies which satisfy the five conditions
(X t X', X", X'", X""} is
= (1, 2, 4, 4, 2, 1J, )(', ') (a", /3")('",
viz. this notation stands for laVY"""+ 2Saa'a"a"'/3"" ... 4
28. In particular if X be the condition that a conic shall touch a given curve
of the order m and class n, then the representatives of this condition are (n, m),
whence the number of the conies which touch each of five given curves (m n) .
(m"", n"") is
= (1, 2, 4, 4, 2, IJn, m)<X m') (", m")(n'", in"')(n"", m"").
29, A system of conies (4/Y) having the characteristics (ft, v), contains
2y  n linepairs, that is, conies each of them a pair of lines ; and
2^~y pointpairs, that is, conies each of them a pair of points (coniques
infimment aplaties).
aO. I stop to further explain these notions of the linepair and the pointpair;
and also the notion of the linepairpoint.
A conic is a curve of the second order and second class; qu& curve of the second
order it may degenerate into a pair of lines, or linepair (hut the class is then =0):
qua curve of the second class it may degenerate into a pair of points, or pointpair
(but the order is then = 0). The two lines of a linepair mny be coincident, and
we have then a coincident linepair; such a linepair (it must I think be postulated)
ordinarily arises, not from a linepair the two lines of which become coincident, but
from a proper conic, flattening by the gradual diminution of its conjugate axis, while
its transverse axis remains constant or approaches a limit different from zero; the
conic thus tends (not to an indefinitely extended but) to a, terminated line^); in other
words, the tangents of the conic become more and more nearly lines through two fixed
points, the terminations of the terminated line; and these terminating points, which
continue to exist up to the instant when the conjugate axis takes its limiting
value =0, are regarded as still existing at this instant, and the coincident linepair
aa being in fact the pointpair formed by the two terminating points. Similarly the
two points of a pointpair may be coincident, and we have then a coincident point
* A line is regarded aa extending from any point A thereof to B, and then in the same diraition, from B
through infinity to A; it thus consists of two portions separated by those points; and considering either portion
as removed, the remaining portion is a terminated lino.
o. VI. 26
202 ON THE OUSVES WHICH SATISFY GIV.EN CONDITIONS. [4QG
pair; such a pointpair (it must in like manner be postulated) ordinarily avisos, not
from a pointpair the two points of which become coincident, but from a proper oonio
sharpening itself to coincide with its asymptotes, and so becoming ultimately a pair
of lines through the coincident pointpair; and the coincident pointpair is rcgai'dinl
as being in fact the linepair formed by some two lines through the coincident point
pair.
31. In accordance with the foregoing notions we may with propriety, and it will
in the sequel be found convenient to speak of a pointpair as a line terminatoil by
two points on this line, and similarly to speak of a linepair as a point torminutnil
(that is, the pencil of lines through the point is terminated) by two lines through thn
point.
32. If in a pointpair, thus considered as a line terminated by two point** 1.1m
two points become coincident (the line continuing to exist as a definite lino), or, what
is the same thing, if in a linepair thus considered as a point terminated by Mvo
lines, the two lines become coincident (the point continuing to exist as a doliiiitn
point), we have a linepairpoint ;" viz. this is at once a coincident linepair ami a
coincident pointpair; it may also be regarded as the limit of a conic the axnw of
which, and the ratio of the conjugate to the transverse axis, all ultimately vanish :
it may be described as a line terminated each way at a point thereof, or as a point
terminated each way at a line through it. The notion of a lino pairpoint flrt
presents itself m Zeuthen's researches, as will presently appear; but it may bo nobmuil
here that linepatrpoints, and these the same linepairpoints, may present bhomsolv
among the 2i//* linepairs, and among the 2^p pointpairs of the system of couias 4 A".
88. Returning to the foregoing theory of characteristics, I remark that the flnwln
mental notion may be taken to be, not the characteristics fa ) O f the conies whidi
satisfy four conditions but in every case the number of the conies whieh satisfy fivo
conditions. Thus for the comes not subjected to any condition, wo may consider tho
syin DO is
OU (::/), (A//), (:///), (.////),
denoting the number of the conies which pass through fivo given points, or which
Ten nZ S rt g 'T P mtS a ' ld t0ll0h a giV6n HM < *" " ^ioh touch v
given lines i these numbers are respectively
= 1, 2, 4, 4, 2, l.
So for the conies which satisfy a given condition X, or two conditions 2X, .... or five
conditions 6Z, we have respectively the numbers
406] ON THE CURVES WHICH SATISFY GIVEN CONDITIONS. '203
where the X, 2X, c. belong to the symbols which follow: read (X::), (Z/./), &c.,
or, as we may for shortness represent them,
X", j/", p'" t a"', T'
P", *"> p", *"
fi , v t p
tl , V
viz, ^the single condition X has the five characteristics (p"', ... T'"), ... ; the four
conditions 4<X, the characteristics (p, v) as in the original theory; and the five
conditions 5X a single characteristic p, .
34. We thus see the origin of the notion of the representatives (a, ) of a
single condition X; for considering the arbitrary four conditions 4>Z t the characteristics
whereof are . (p t v], and assuming that the single characteristic, or number of the conies
(X, 4<Z), is =a^, + /3y, and taking for (4#) successively the conditions
00. (*./), (://), (///>. (////).
having respectively the characteristics
(1, 2), (2, 4), (4, 4), (4, 2), (2, 1),
we have
^'"la + 2ft
i/"  2a 4 4/3,
p"' = 4a + 4/9,
o'" 4a + 2/3,
that is, the characteristics (/", i/", /)"', <r r ", r'") of a single condition X are not
independent, but are represontable as above by means of two independent quantities
(a, /9); or, what is the same thing, we have
which being satisfied, tho representatives (a, /3) are given by
35. I find that a like property exists as to the characteristics {/*", y", p" t a fl ) of
the two conditions 2Z, viz. these are not independent but are connected by a single
linear relation,
X'f i/' + /" o^O.
This may be proved in the case where the conditions 23T ore two separate conditions
(X t X'}\ viz. let the representatives of these be : (a, 0), (a', ^') respectively, then
26 2
204 OX THE CURVES WHICH SATISFY GIVEN' CONDITIONS. [400
combining with them the three arbitrary conditions X", X'", X'" having respectively
the representatives (a", /3"), (a'", #"'), (a"", ""), we have the general equation
(X, X', X", X'", T' ") = (!, 2, 4, 4, 2, !$, /3)(', /?')(", /TXa'", /3"')("", /3"")j
taking herein
successively, and observing that the representatives of (.) are (1, 0) and those of (/)
are (0, 1), we thus obtain for (^", ,/', p " t ff "), characteristics of (X, X'), the values
/' = (!, 2,4$, /3) (a, '),
p"=(4, 4, 2J, )(', ff),
"(*, 2, !$, /9)/9'),
(viz. /i" = laa' + 2(^' + R ^) + W, &&), and these values give identically
2/i"3i/' + 3 /) "2<r":=0,
which is the foregoing equation. And I assume that the theorem extends to the
case of two mseparable conditions 2X, but in this, case I do not oven know vrhoro
the proof is to be sought for.
The characteristics &, v >, p ' } O f the three conditions 3X are in general indopondoufc.
the oonditioii
X and t '
A, and (p, v ) the characteristics of the conditions 4,2, then
(X, *Z) = a fJ , + /3v )
this is the most convenient form of the theorem bnf /* o\ i * ,
of the characteristics ('", ,'" ,'" ,'" ^0 H' , ( ' ^ kn Wn ftlllotlona
effect an expression for (X lz) in 'terms of \l T ^ Z ^ ^ nfltio11 ifi in
F ^, w; in teims of the characteristics of X and 4 reapootivoly.
There is, similarly, an expression for (9Y i?\ * n ,
<W < P', O of M (satisfying Kto X^V + i; 11 T m it amCtOL " b i C9
(ft P, /)) of 2JT, viz. we have t"+l/r0)aiid the elmractoristics
e ^ h ^ d t t t;::rfVr^ " ara  io
3^ three separable conditions Z" ?'" 7>J' ' f } res P ectlvel y> and tlio
(a"", /n respectively; we have, b 1't ^presoatatives (, ^
406] ON THE CURVES WHICH SATISFY GIVEN CONDITIONS.
and with these values the function
(X, X\ Z", Z'", Z""), =(1, 2, 4, 4, 2, l
205
is found to be expressible as above m terms of (& v, p), (&', v ', p', </) ; but I do
not know how to conduct the proof for the inseparable conditions 2X and 3Z.
37. It may be remarked by way of verification that writing successively
W = (.'.), (:/>,<//),(///>,
that is,
0*, v, p) = (l, 2, 4), (2, 4, 4), (4, 4, 2), (4, 4, 1),
we have in the first case
(2AV.)= ip' + i* 7
and similarly in the other three cases,
(2Z //) = ,',
38, Let O, y, p, ff ) be the characteristics of
(X, i/, p', o') the characteristics of 2X (M'~f / + /?'
for (2JK", 3^), writing successively for 3^
, (/*$ y 4f p  o = 0), and
' = 0). Then in tho formula
and
(2J5T ), characteristics (^ y,
we obtain expressions for the characteristics (2Z, 2^.) and (2Z,
eliminating from the formulas, first the (a, o') and secondly the
may be expressed in two different forma as follows:
of (2X, 2.2T), viz.
/), each of these
i (p"' i pv)
' +
IPP'
7 +
206 ON THE CURVES WHICH SATISFY GIVEN CONDITIONS. [406
the two expressions of the same quantity being of course equivalent in virtue of the
relations between (p, v, p, <r) and (X, v\ p' t a') respectively.
The characteristics of (X> Z}, (X, 2#), (X, 3Z) are at once deducible from the
beforementioned expression up + $v of (X t 4>Z).
39, Zeutheu's investigations are based upon the beforementioned theorem, that
in a system of conies (4), characteristics (p, v), there are 2^ v pointpairs and
2i> ft, linepairs. If in the given system the number of pointpairs is ~\ and the
number of linepairs is = vr, then, conversely, the characteristics of the system are
v ~
And by means of this formula he investigates the characteristics of the several systems
.of conies which satisfy four conditions (4.X) of contact with a given curve or curves,
viz. these are the conies
<1,1) (!)(!), (1,1) (1,1), (1,1,1X1), (1,1.1,1),
(2)(1)(1) , (2)(i, 1) , (2,1)(1) , (2, 1,1),
(2) (2) , (2, 2)
()(!) , (3, 1)
(4)
where (1) denotes contact of the first order, (2) of the second order, (3) of the third
order, (4) of the fourth order, with a given curve; (1)(1) denotes contacts of the first
order with each of two given curves, (1, 1) two such contacts with the same given
curve, and so on. A given curve is in every case taken to be of the order m and
class n, with S nodes, cusps, r double tangents, and i inflexions (m^ n^ B lt K lt r lt tij
?tt a , n, foe., as the case may he). The symbols (1), &c. might be referred to the
corresponding curves by a suffix'; thus (l) m would denote that the contact is with a
given curve of the order m (class n, c.) ; but this is in general unnecessary,
40. In a system of conies satisfying four conditions of contact, as above, it is
comparatively easy to see what are the pointpairs and linepairs in these several
systems respectively; but in order to find the values of \ and or, each of these point
pairs and linepairs has to be counted not once, but a proper number of times; and
it is in the determination of these multiplicities that the difficulty of the problem
consists. I do not enter into this question, but give merely the results.
41, For the statement of these I introduce what I call the notation of Zeuthen's
Capitals. We have to consider several classes of pointpairs and the reciprocal classes
of linepairs. A pointpair may be described (ante, No. 31) as a terminated line, and
a linepair as a terminated point; and we have first the following pointpairs, viz.:
A t line terminated each way in the intersection of two curves or of a curve with
itself (nocle),
J3, tangent to a curve, terminated in a curve, and in the intersection of two
curves or of a curve with itself. .
406]
ON THE CURVES WHICH SATISFY GIVEN CONDITIONS.
207
G, common tangent of two curves, or double tangent of a curve, terminated each
way in a curve.
D, inflexion tangent of a curve terminated each way in a curve :
and the corresponding linepairs, viz. :
A', point terminated each way in the common tangent of two curves or the double
tangent of a curve.
B', point of a curve terminated by the tangent of a curve, and by the common
tangent of two curves or double tangent of a curve.
G', intersection of two curves, or of a curve with itself (node), terminated each
way by the tangent to a curve.
.D', cnsp of a curve terminated each way by the tangent to a curve :
all which is further explained by what follows ; thus in the case (1) (1) (1) (1),
 (Ik (!)!, (Ik 0k> tne value of 4 is given as Smim s .w(aft 4 (=3m 1 am B ffi,). Here
A is the number of the pointpairs terminated one way in the intersection of any
two mi, m 2 of the four curves, and the other way in the intersection of the remaining
two m 3 ,m, of the four curves. But in the case (1, 1)(1)(1), =(1, l) m (l) W (!),,, the
value of A is given as =Bm i m 3 + m f rn 1 .mm it Here A denotes the number of the
pointpairs, which are either (7%?%) terminated one way at a node of TO, and the
other way at an intersection of m,, m 2) or else (mm, . mit^) terminated one way at an
intersection of in, w^, and the other way at an intersection of m, m a : and so in other
cases.
4<2. This being so, we have
G
(1,
4
D =
1Hi /_ 3 fHjTO^mj
2
A. = Sjtt^ij . 7lg?t4 (=3 WiWa?isU4 ),
,=(!,!) dk dk
+ wm 1 .mm a ,
1
^U. = T)ii?Za 4 HHj . W)J 3 ,
+ Sn a 7?h
2
J9 = TWl a ?2 2 + TWiijJl!
2) m, + mwij, (ft  2) ?ft,
+ ?m] (m 2) a + mis (m 2) ?^
1) + mtn$ii (m 1)
+ Whm,(l)+wn a m a (l)
2),
+ ?i iigTn (?i 2),
4
G' = S^ns
^+nn a(W 2) W ,
+ tfwih (n 2) WB + mm a (ft 2) 7
3
D'= wi.nl
208
ON THE CURVES AVHICEL SATISFY GIVEN CONDITIONS.
[40G
, ix =<i, iva, iv,.
^.= 88, + wmij (mm,! 1),
S = 8?u (mi  2) + Sjn (m  2)
(n 2) (TOT I) + mm,, (n, 2) (7*1 1),
= T . J flli (ffl, 1) + Tj . 4 TO (m  1)
+ K! (m ~ 2) (nil 2),
D i . ^ TO! (mi 1) 4 (i .  m ( m ~ !)
TT,
2)
, (m
(^
(n
1) +
+
2),
(2)
m, a) <i),
(1, 1, !)(!) (1,1, 1) M (!),,
4 = Swini,,
5 = S (n  4) ii + S?h (m  2),
+ 0wi (w  2) (t 3),
(7 T (m 4) 1! + 7Mi! . ^ (m 2) (m 3),
4) tii + T7! (?i 2),
= 8 ( n  4) ?i, + mm, 4 (  2) (?
(1, 1, 1, 1), =(1, 1, 1,
= i8(S~l),
= 8 (n  4.) (m  4),
E'=
43, Secondly, we have the point paiis :
E, tangent to curve from intersection of two curves or of a curve with it?
(node), and terminated at the point of contact and the last mentioned point.
J 1 , tangent to a curve at .intersection with another curve or with itself, and
terminated there and at a curve.
(?, common tangent of two curves or double tangent of a curve, terminated ftl
one of the points of contact and at a curve.
D, ut supr&.
lining cusp of a curve with intersection of two curves or of a ciu'vu
If, and terminated at these points.
/, line from cusp of a curve touching a curve, and terminated at the cusp and
at a curve.
J, Inflexion tangent of a curve, terminated there and at a carve :
406]
ON THE CURVES WHICH SATISFY GIVEN CONDITIONS.
209
and the corresponding linepairs, viz.
E', point on a curve in common tangent of two curves or double tangent of a
curve, and terminated by this tangent and by tangent to a curve.
F, point on a curve in common tangent of this and another curve or in double
tangent of this curve, and terminated by this tangent and by tangent to a
curve.
D', ut suprcl,
H', intersection of inflexion tangent of a curve with common tangent of two
curves or double tangent of a curve, and terminated by these lines.
/', intersection of inflexion tangent of a curve with a curve, and terminated by
this tangent and by tangent of a curve :
and this being so,
E ~ n .
F = minj , m s + mtn. 2 . in l}
D =
H~K.m l m 2 ,
E = Sji,
F ~ m . mi (mi ~ 1),
G = ntii (mi 2),
I = ??! (wij 2).
(2,
F = mm, (m  2) + 2Sm,,
H =
/ M K (n  3) mi + ?ii (m  2),
/ = wn lt
C. VI.
E' = in ,
F ~ mi! , m,
E' =
(j  2),
' = , in, (h  1),
/' =
(HI 2).
^nik (n2) +2TW,,
1 ' = mm, (n 2) H
H'
210 ON THE CURVES WHICH SATISFY GIVEN CONDITIONS.
(2, 1, 1), ={2, 1, l) w .
E = & (n  4),
[406
Q = 2 T (m  4),
= t4(m3)(m
H= $ K ,
I = *(n 3) (
J = tm ~3.
= r(m4>),
= 2T( 3),
4),
H = IT,
I' = i(ro3)(n4),
44. Thirdly, we have the pointpairs:
K, common tangent of two curves or double tangent of a curve, terminated at
points of contact.
L, line from cusp , of a curve touching a curve, and terminated at cusp and point
of contact.
M, line joining cusp of a curve with cusp of a curve, and terminated by the two
cusps.
N, inflexion tangent terminated each way at inflexion, viz. this is a linepairpomt,
0, cuspidal tangent terminated each way at cusp, viz. this is a Unepairpoint :
and the corresponding linepairs :
K' t intersection of two curves or of curve with itself (node), and terminated by
the two tangents.
L' f intersection of inflexion tangent of a curve with a curve, and terminated by
the inflexion tangent and the tangent at the intersection.
M', intersection of inflexion tangent of a curve with inflexion tangent of a curve,
and terminated by the two inflexion tangents.
N' } = 0, Unepairpoint as above.
0', N } Unepairpoint as above:
which being so, \ve have
(2) (2), =(2) M (2) n!l .
L =^1, + ^^
M=KK i .
9
K'
= mm lt
3
L 1
= mi +
1
M'
= iti.
406]
ON THE CURVES WHICH SATISFY GIVEN CONDITIONS,
211
(2, 2), =(2, 2) m .
L = *(n3),
M = $K(K1),
N =L,
=K.
K' = 8,
L' = t(mS),
K ,
0' =
45. Fourthly, we have the pointpairs :
P, tangent of a curve at its intersection with another curve or itself, terminated
each way at the point of contact linepairpoint.
Q, common tangent of two curves or double tangent of a curve, terminated each
way at one of the points of contact linepairpoint.
J, ut supra.
M, cuspidal tangent terminated at cusp and at a curve:
and the corresponding linepairs :
P', = Q, linepairpoint.
Q f , ~P, linepairpoint.
J', lit 'suprh.
R!, inflexion of curve terminated by the inflexion tangent and by tangent to a
curve : '
which being so, we have.'.
P = mm,,
2
T)/
Qwh ,
2
Q = m??^,
/ = , ,
5
i/' = ICUi,
R^Km, .
4
jy _ tn _
, IV.
P = 2S,
2
<8'2T,
Q =2r,
2
P' = 2S,
J = t (m  3),
S
r/ /
JR = K (m  3).
4
Ji'= *(
46. ^And lastly, we have the pointpairs N t (linepairpoints) and the linepairs
N', 0' (linepairpoints), ut suprti, and
JV = ,
0"t
272
212 ON THE CURVES WHICH SATISFY GIVEN CONDITIONS. [406
47. Where in all cases the central column of figures gives the numerical factors
which multiply the corresponding capitals, thus we have
X = 2y  /* = A + 25 + 4(7,
ar = 2iy =:;i' + 25' + 4a'
for (1,
X = 2y/i = J 4 + 25+4(7+3.0,
OT = 2/i y = 4' + 25' + 4(7 + 35',
and so on.
48. The elements (m, n, S, K, r, t) of a curve satisfy Pltioker's six equations, and
Zeuthen uses these equations, m a somewhat unsystematic way, to simplify the form
of his results.
It is convenient in his formulas to write 3w+4, = 3 + , =a, and to express every
thing in terms of (m, n, a), viz. we have for this purpose
2S = m 2  m + Sn3a,
2T = ?t a STO w3o.
But I make another alteration in the form of his results; he gives, for instance,
the characteristics of (1, 1) (1) (1) as
V = V m y y
li = fi'" w^m, + /*" (??!,n a +
V
where
X =2m( m+ n8)+ r, =(1, 1,. ),
/=/ =2/1 ( m + 26) + 2T, =(1, I:/ ),
X" = " = 2 (2m + B  5) + 28, =(1,1 //),
i/" = 2w(m+ 3)+ S, =(1, I///),
viz. the four components have really the significations (1, 1..) se t opposite to them
respective ly ^ and accordingly, instead of giving the formula for tho two characterisbics
ot (1, 1)(1)(1), I give those for the four characteristics (1, l..V & c . O f (1, 1) thus in
every case obtaining formula which relate to a single curve only. Subject to the last
mentioned vanation of form I give Zeuthen's original expressions in Annex 6; but
here m the text I express them as above in terms of (m, , a), viz.
49. We have the formula
(1)
406] ON THE CURVES WHICH SATISFY GIVEN CONDITIONS. 213
0, 1)
: /)  2m a
 m  2  3a,
, 1, 1)
win" + n 8  2wi a
n 3  4f m  ^ n + a ( 3m  Bn + 20),
(1, 1, 1, 1)
m a 
 ?t
I ^ m s n + wi 3 ft 3 + mn'
+ i
(2)
(.'.)= ,
(///)= ;
(2,1)
( : ) = 12m + 12?i + (2m + n  14) a,
( /)  24i  24 + (2m + 2w  24) a,
( //) = 12m + 12n + ( m + 2w  14) a ;
(2, 1, 1)
( )  24m 3 + 36mn + 12?i 3  168m  168n + ( m 2 + 2?n
( / ) = 12m + 36m + 24n  168m  168n + a (iwt" + 2m
(2,2)
( )  27m + 24?i  20a + fa\
( / ) 24m + 27n  20a + a 3 ;
(3)
( : ) =  4m  3w + 3,
(./) 8m  8n + 6a,
214 ON THE CURVES WHICH SATISFY GIVEN CONDITIONS. [406
(3, 1)
( . ) =  8m a  12wn  3?i 3 + 56m 4 53n + a (6m + 3n  39),
( / )   3m 3  limn  S?t 2 f 53??t + 5G + a (3 4 6  39) ;
(4)
( )   lOwt  8n 4 6a,
(/) =  8m I0+ 6a.
50. By means of the foregoing formula) I obtain, as will presently be shown, the
following formula for the number of the conies which satisfy Eve conditions, viz. :
(4, 1) =  8m 3  20)ftn  S* 4 104i 4 104n + a. (6m + 6n  66) ;
(3, 2) = 120wi 4 120/1+ a ( 4m 4 78) +3tf;
(3, 1, 1)  f m 8  lOm'H  10mn  f n* 4 ^ m + H6m + ifi  434i  4S4n
(2, 2, l)
4 a ( 8m  Sn 4 327) 4 a a (i 4 ^w  12) ;
(2, 1, 1, l) = 6m + 30m 9 H + 30wm 8 +0.*l74m s 348mtt
+ a (ft m 3 4 wte + inw 1 + ^ w '  yi wi 2  26nm  * n* + fifa m + 4fi u _ 960)
4
1 mw ?i 3 + ^m 9 4 23m + ^nAfimafiw + 486)
51. I observe that by means of the abovementioned expressions of (X, 4<Z) and
(2Z, 3^), the foregoing result^ other than those for (5), (4, 1), &c., may be presented
in a somewhat different form, viz. we have
where (.) denotes (4Z .), (/) denotes (4^/) f and so in other cases, the understood
terra being 3Z or 2Z, as the case may be.
1 In my paper in the Comytes Rendvs, I gftvo eiToneously the oooffioients 
ON THE CURVES WHICH SATISFY GIVEN CONDITIONS. 215
(82) (2) =
+ (//)<***"+)
+ (///)( m+fftfa);
3m  3 + a ( J m + ft + 2)}
, 1,1) =
^m^7i
( m+ t 1));
in all which fonnulea it is to bo recollected that we have
('.)* /) + (//)(///) o,
to which may be joined
where a, & are the representatives of the condition (#), and whore (4>X) is to bo con
sidered as standing successively for (4), (3, 1), (2, 2), (2, 1, 1), and (1, 1, 1, 1), the
values of (43T) and (4A7) being in each case given by the foregoing Table.
52, The formulas are very convenient for the calculation of the numbe
conies which satisfy five conditions of contact with two given curves ***
example, (8), =(3) tl . denotes the condition of a contact of the third
given curve (wj), then writing for symmetry (2) m in place of (2), we have
= a ( 4m, 4?t,
53. To obtain the foregoing expressions of (5), (4, 1), (3, 2), (3, 1,
(2, 1, 1, 1), and (1, 1, 1, 1, 1), I assume that the given curvo breaks
curves (in, n, ) and (m' f n', a'), or, as we may for shortness express it, in
m and m'.
216 ON
We have then
THE CURVES WHICH SATISFY GIVEN CONDITIONS. [406
f
e 'this contact with the curve '. Writing tins under the form
observing that (5), is a Action * <* .. X 1 that
functional equation ^(w+m 1 , + < + )*(w, , ) <M
< (m, n, a) =
i, c ro arbitrary constants; but as the solution should be symmetrical in
re. to t, , we have A or the solution is * (, , ) = < + ) +
54. Similarly we have
vi, the C o,ics .vhich have ,vith the aggregate curve m + rt ^.^te ^
made up of the conies which have the two contacts 4 and 1 w tl the one cu 01
vUh the other curve, or the contact 4 with the one curve and the contact 1 with
Ihf ottr curve, The expression on the righthand ride is a known flmotion of ( n. ),
(m f , n, a') ; hence the form of the functional equation is
and any particular so.ution of this equation being obtained the general ^ s^tion in
found by adding to it the term am + fc^ca. Assuming that the partial Boloi on
ymmetlal in regard to ( n), then the term to be added a, before = a(m+n) + 
And similarly far (8, 2), (8, 1, 1), fe; that is, in every CaS e we have a solution con
taming two arbitrary constants a, c, which remain to be determined.
55. Now in every case exce.pt (5) w the number of intersections of the conic with
the curve is > 6 (viz, for (4, l) m and (3, 2) m the number is 7, for (3, 1, 1} and
(2 2 1} it is 8 and for the remaining two cases it is 9 and 10 respectively); hence
if' the riven curve be a cubic, the number of conies satisfying the preserved
conditions is 0; and since a cubic may be the 8^^^^ * V 10 ?
cuspidal cubic, we have the three cases (, n, )(8, 0, 18), 8, 4, 18), and (3, 8, 10)
We have thus in each case three conditions for the determination of the constants
a, o ; so that there is in each case a verification of the resulting formula.
56 In the omitted case (5) W1 when the curve m is a cubic, the theory of the
conies (5V is a known one, viz. the points of contact of these conies, or the "sextachc
points of the cubic, are the points of contact of the tangents from the points ol
inflexion; the number of the conies (6) m is thus (8)*, viz. in the three case*
406] ON THE CUBVES WHICH SATISFY GIVEN CONDITIONS. 217
respectively it is = 27, 3, and 0. Hence for determining the constants we have the
three equations
9 + 18c = 27,
7ahl2c = 3,
(j(ti10c= 0,
which are satisfied by (( = 15, c = 9, and the resulting formula is
(5)15?H,16n + 9.
In the particular case of a curve without nodes or cusps, this is (5) = 12n  15m,
= m(127M27), which agrees with the result obtained in my memoir "On the Sextactic
Points of a Plane Curve," Phil. Trans, vol. CLV. (1865), pp. 545 oVS, [841].
57. The subsidiary results required for the remaining cases (4, 1), &c. are at once
obtained from the foregoing formulas for (4#) (1), (32) (2), &c.; for example, we have
with like expressions for (3, l) m (!), , &e.,
(S) w (2) ffl . = t  8m 
fa' ) ( 8m 
+ (i' a  ^j,') ( 3m 
with like expressions for (2, l) w (2) m . ( (2, l) m (l, !),, &c. &c.
58. Calculation of (4, 1), We have
(*, l)w  (4, 1)  (4, l) w . = (4), a (l) w . + (4) m  (1) M ,
=  16mm'  20 (mw' + m'ri) 16nri+ 6 (' + a'n) + 6 (cmi f + V),
the integral of which is
(4, l) m =  8m? .20wOT  87z, a + a (m + n) + a (6m + 6n + c).
The particular cases (m, n t a) (3, 6, 18), (3, 4, 12), (3, 3, 10) give respectively
= 252 + 9ft + 18c,
0= 64 + 7a I 12c,
0= 36 + 6a + 10o,
satisfied by a = 104, 0=366,
0. VI. 28
18 ox THE UUBVJJS WHICH SATISFY GIVEN CONDITIONS.
W. Calculation of (3, 2). We have
<*. 2) lll+)i ,  (3, 2),.,  (3, 2),,,, = (3) w (2),,, + (3),,, (2) w ,
=  4 (ww' + m'a)  4 (710' + n'ct) + fla / :
'.0<i
(8, 2),,, = a (in + n) + a ( 4t  4u + c) + Sa 11 ,
' '
324 + {) + 18c = 0,
. . , .
the Integra! w
, ,
ami, as before,
. , = ,
tinl by o 120, c =  78.
CO. For the calculation of (3, 1, 1) we have similarly
(. I, 1U,H3, 1, IX. (8. 1, ^^.(S
cs, IV
value, so obtained
pt that thc terms in
d the auh.equeot eases
In the Ft ,,nt ca.se
thc more
i
"* 8101111 f (3 ' ^ ^'
* ^ fc "
and T f g Ulg ex P ra8slon
! thp T " " ^ + )
} '^ equabonfl which
"'"
"'"
r (8, ", I >.
382 + 80 + 180*0,
The remaining cases are (2, 2 1W9 i i
}> ( ' lf X '
and
 1. 1). Wo have
(2),, + (2,
by flaa
Again,
406] ON THE CURVES WHICH SATISFY GIVEN CONDITIONS. 219
and
1680 + 6 + 100 = 0,
satisfied by a = 1320, c=960; and finally,
(i, i, i, i, iw(i, i, i, i, i) m (i, i, i, i, i )m , = (1 1( lt i), Jt (i)
Wld Hi.i,i)mai
 30618 + QO<E+180c = 0,
 .14094 + 70fj+120c=0,
 10692 +60+ 1200 = 0,
satisfied by 10ft = 6818, 10c = 4860, that is, a = upi t c = 486.
62. The contacts of a conic with a given curve which have been thus far considered
are contacts at unascertained points of the curve; hut a conic may have with the given.
curve at a given point thereof a contact of the first order, the condition will be denoted
by (2); or a contact of the second order, the condition will be denoted by (3), and so on.
It is to^be observed that the conditions (2), (3), &c. are sibireciprocal, the contact at a
given point of the_curve is the same tiling as contact with a given tangent of the curve;
but if we write (1) to denote the condition of passing through a given point of the
curve, this is not the same thing as the condition of touching a given tangent of the
curve ; and this last condition, if it were necessary to deal with it, might be denoted
by (1), But I attend only to the condition (1). The expressions for the number of
conies which satisfy such conditions us (1), (2), &c. are obtainable in several ways,
G3. (1) When the total number of conditions is 4, the question may be solved
by Zeuthen's method, viz. by determining the linepairs and pointpairs of the system
4>Z> with the proper numerical coefficients, and thence deducing the values of the
characteristics (4#.) and (4#/). A few cases are in fact thus solved in .Zeuthen's work.
64. (2) By the foregoing functional method, It is to be observed that there is a
difference in the form of the functional equation, and that the general, solution is
always given in the form, Particular Solution f Constant, so that there is only a single
constant to be determined by special considerations. To take the simplest example, let
it be required^ to find the number of the conies (3) (1, 1) : writing for shortness in
place hereof (1, 1), or (in order to mark the curve (TO) to which the symbol has.
reference) (1, l) m , let_ the curve (m) be the aggregate of the curves (m) and (m').
Regarding the point 1 as a given point on the curve (m), that is, an arbitrary point
in regard to the curve (m'), we have thus the equation
where the righthand aide is known j and .so in general the form of the functional
equation is always 0(m + m')^(m) given, mine, that is,
+ ri, a + a.')  <j> (m, n , a)  given function of (m, n, a, nf r n f , o!)\
282
220
ON THE oravffl waroH SATISFY GIVEN CONDITIONS.
these cases
(1. !)+,'  (I, l) M = B
whence (I, !)*= + 2m + const. = (I, !)() and
replaced
CM, l) =
be  mes 
[ 406
have m the first of
 4,
_
(I, T, I, I, 1) ^ , + 2wi  8.
and
2,1,1
2 1 1
eoiidein
(2, 2, l) B =
the fcnr given points on
have
reducn,
Jf ^ V 2 * boide, that is, if
U eoincide > there is uo further reduction, but we
1 I mjte indifferently (l)M il\
J wi.j, (i..j
nr /..i\ i
or (:;1). ftuc i so
406] ON THE CUHVES WHICH SATISFY GIVEN CONDITIONS. 221
GO, The expressions involving a single (I) may in every case be reduced by the
foregoing method to depend upon other expressions; thus we have
)(!,!) =(!) 2(2)
)(!, 2) =(2) 3(3)
(I 1,1) =(1, 1) 2(2, 1) ,
( Z )(T, 1, 2) =(!, 2) 2(2, 2) 3(1, 3),
(1, 1, 1, 1)=(1, 1, l)2(2, 1, 1),
_ (1. S) =(3) 4(4)
(1, 4) =(*) 5(5)
&c.,
where, comparing for example the equations for (2) (I, I, 2) and (2#)(1, 1, 1), it will
be observed that in tho first case tlie contacts 1, 2 of tho symbol (1, 1, 2) successively
coalesce with the point I, giving respectively 2 (2, 2) and 8 (1, 3), the exterior factor
being in each case the barred number, whereas the second case, where tho contacts 1, 1
of the symbol (1, 1, 1) are of the same order, we do not consider each of these symbols
separately (thus obtaining 2(2, 1)42(1, 2), =4(2, 1)), but the identical symbol is taken
only once, giving 2(2, 1). Thus wo have also
(1, 1, 1, 1, 1) = (1, 1, 1, l)2(2, 1, I, 1),
(57. Tho value of a symbol involving (2), say tho symbol (&Z) (2), is 'connected
with that of (3/); but as an instance of the correction which is sometimes required
I notice the equation
(2, 1, 1, 1)(1, 1. l/)IiO2)03) + i(n2)(3) + 3(3 l 1, 1)42(4, l),
which I have verified by other considerations.
68, We obtain bhe series of results :
(1)
( ") = 1,
(/./) = 2,
( = //) = 4,
('///) ==4,
(////) 2 J
(1, 1)
(.'.)= + 2w2,
( : / ) = 2 + 4m  4,
(.//) = 4<n f 4??i 4,
ON THE CURVES WHICH SATISFY GIVEN CONDITIONS.
(1.2)
(:)= 8,
A 1, 1)
( : )2i a +2iH+^i s Ow^i+ 8 fa,
12  3,
8  3a ;
(T, 3)
( ) = 4m3H4 + 3a 1
( / )=* Sm8B 4 + G;
(1,1,2)
( . )=
(///) = !;
(2.1)
( : ) = 2m+ 4,
( ' / )  2m + 2/i  6 ]
( // ) = m + SH  4 ;
(2,2)
(2, 1, 1)
( ' )=
^( / ) =
(3)
(0 = 1,
406] ON THE CUEVES WHICH SATISFY GIVEN CONDITIONS, 223
<3, 1)
( ) = n + 2m  6,
(/) ~2n+ m6;
(4)
() =1,
winch are the several cases for the conies which satisfy not more than four conditions,
and
69, For the conies satisfying 5 conditions, we have
(5) =1, *
(3, 2)  9 + a t
(3, 1, 1) =  W " + 2nm + n * _ y m _ j^ 4. 27 _
(2, 3)
(2, 2, 1)
(2, 1, 1, 1)
(, 4) =
(I, 1, 3) =
(1, 2, 2) = 27m + 24)i + 27  23 + a 3 ,
(T, 1, 1, 2) #wH30m + *ftt"^w
+ (w 2 + 2nz, + i  27m 
(T, 1, 1, 1, l)^w 4 + ^w a ft + m 5( rt 1l + i7
 JLWI'  6m aR" + i + + 150
70, The given point on the curve to which the symbols I, 2, &c. refer may be
a singular point, and in particular it is proper to consider the case where the point
is a cusp. I use in this case an appropriate notation; a conic which simply passes
through a cusp, in fact meets the curve at the cusp in two points ; and I denote
the condition of passing through the cusp by l'/d ; similarly, a conic which touches
the curve at the cusp, in fact there meets it iu three points, and I denote the
condition by 21 J 1/cl, 21 are thus special forms of 1, 2, and the annexed T indicates
the additional point of intersection arising ipso facto from the point 1 or 2 "being a
cusp. Similarly, we should have the symbols 3/cl, 4/el, 51; bufc it is to be observed
that ab a cusp of the curve there is no proper conic having a higher contact than
ME SATISFY
2*1; thus if the symbol contains JvT n, ,
to
* ,
" 6il
n m, f ,
seo Annexes Kos. 4 and 5. J OTS f the Cilae f the ""s
itself h ^ to ^
we have " e cin ' ve 1S an ordinary point. For exampk
in simply ol
which only diffpr fmm th
r fion, the e x pl , ssiom with
in place of " + 2 '" ~ 3 ' 2 " + 4 " 1 ~ 6. * + *  0, 4 m + 2  S
2. We have
(1*1)
('/) = 2,
(///) 4,
406] ON THE CURVES WHICH SATISFY GIVEN CONDITIONS. 225
(1*1, 1)
(.'.)= +2m3,
(//) = 4. + 4m  6,
(///)== 4w + 2m3;
(1*3, 2)
( : )= a4,
(1*1, 1, 1)
( : ) 27W 9 + 2mn + n a &$+ 13
( / ) = 2m" + 4m + ii 3  8i  7n f 18
( // ) = m s + 4nm F 2?i a  4m  8?i 4 12
, 3)
( . ) = _ 4, m _ 3 n _ 5 4. 3 Hj
1, 1, 2)
4m + 8
, 1,1,1)
(2*1, 1)
( : ) =s 2m +  5,
(// )= m+2n4;
, 2)
226 OX THE CURVES WHICH SATISFY GIVEN CONDITIONS.
(2*1, 1, 1)
( ) = wi a +2ift + 
a fm
78. The remainder of this table, being the part where tho symbols () ami (/)
ot occur, I present under a somewhat different form as follows :
(8*1, 2)
(3*1. I, 1)
(2, 3)
(2,2,1)
& I. I.I)
= 0,
= 0,
= 0,
= 0,
(2*1, 3)
f2*l 9 i\
\ A. i , & t Jl
(2*1, 1, 1, 1)
I 4)
=o,
1.1..) 
eh ange of a cusp into an i fl n ^
h as fti ; but ; have a : o n :,; t
contain this new symbo l.
the investigation* oontamo.l in the
^ ' ld o
theroforo ^ ivo **
nece8s ^'y to consider tho formula)
fl/
o,, rv
tho
e
of n
th
orders
406] ON THE CURVES WHICH SATISFY GIVEN CONDITIONS,
conditions; viz. the number of the curves O 1 ' is = /*(a+ 1) (6 + 1) (o + 1)... into
+ [rm  (a + 6 + c . .) p  2]'~ a (06 + ao + bo . .) [D] 2
where the curve t/'" is a curve without cusps, and having therefore a deficiency
D = (in l)(m2)8; the numbers ft, b, G,.. are assumed to be all of them unequal,
but if we have of them each = a, fi of them each = b, Ac., then the foregoing
expression is to be divided by [a] a [/3] p . , . ; and p denotes the number of the curves G r
which satisfy the system of conditions obtained from the given system by replacing
the conditions of the t contacts of the orders a, b, c, &c. respectively by the condition
of passing through a + b + c . . . arbitrary points. In order that the formula may give
the number of the proper oitrves C" 1 which satisfy the prescribed conditions, it is
sufficient that the ir(r + 3)(tt fi + c . .) p conditions shall include the conditions of
passing through at least a certain number T of arbitrary points : this restriction
applies to all the formula) of the present section.
7n. I will for convenience consider this formula under a somewhat less general
form, vise. I will put p = 0, and moreover assume that the r (r + 3) ~ (a + i + c . .)
conditions are the conditions of passing through this number of arbitrary points ;.
whence /* = 1.
, We .have thus a curve G 1 ' having with the given curve ff m t contacts of the
orders a, b, G,. respectively, and besides passing through r(i % + 3) (ft + 6+c..) arbitrary
points; and the number of such curveH is by the formula =(+ 1) (b + l)(c+ 1),,.. into
+ [rm  (a + 6 + c . .)  I]' 1 (a +6 + c . .) [DJ
+ [rm  (a + b + c . .)  2]'" (at + ac + &c . .) [D] a
+ [rm (ft + 6 + a . .) t ] (ct6c ... )
where, as before, in the case of any equalities between the numbers a, 6, o, ..., the
expression is to be divided by []*[$]''....
76. I have succeeded in extending the formula to the case of a curve with
cusps: instead of writing down thp general formula, I will take successively the cases
of a single contact a t two contacts a, b, three contacts a,, b t c, &c, ; and then denoting
the numbers of the curves G 1 ' by (o), (a, b), (a, b t c), &c, in these cases respectively,
I say that we have
(a) = (a + 1) (rin  a"
a
292
228
ON THE CURVES WHICH SATISFY GIVEN CONDITIONS.
(a, b) = (ft + 1} (b + 1} f [rm  a  ftp
"40<
6
' 0(6
' "i
fVf' J
L K ,\
 [So (a + 1) (6
[ m  a _fi_ p
+ [ m _ ft  & _ _ ip ( ft
I [rm  a  6  o  2]' (
+ (t
[m _ f( _ b _ c _ : , ?
+ [ ~  &  c  2] 1 (a
I uo + 60) [J3]"
(a + 1) f [ m  a _ & _ c __
1+ '
J i .1
^r
' (, i, o, rf), putting therein fo ,
+ i + . + I.^ 4 +
mte dow
and also the formula for f/i /,
"* for (a, i, ,
lor the combinations of fa b r rJ
W have (a ' '' ' *
_
putting
(. b, c, rf), ( a> 6 f ) ) ( aj 6 j and ^^ ^
[rm  p
406]
ON THE OUEVES WHICH SATISFY GIVEN CONDITIONS.
229
. (a +
f [rm  a.  1] 3
+ [rm  a  2] a ' [7)] 1
3 [rm (t 3] 1 /3' [7J] a
alod
(a, 6, o, d, e) =
[>m a ]"
+ [?w  a  I]' 1 a [U] 1
S [rm  a  2] 3 [D] 3
f [?'m   3] a 7 [D] 3
+ [ m _ a _4 ( ]ig [jf)].>
 e [D] G
+ [ rm a  2J 3 a' [D] 1
+ [rm ~ a  4] 1 y [D] 3
+ S' [!)]' J
/ [ m __2]> ^ ][*]
J [
 aficde .
[?'t  a 4]'
230 ON THE CURVES WHICH SATISFY GIVEN CONDITIONS.
7S. In all these formula thera is, as before, a numerical divisor in the case of
any equalities among the numbers a, b, o, &c. And D denotes, as before, the deficiency,
viz. its value now is D = %(ml)(m$)% K \ or observing that the class n is
w a w2S3, we have D= $nm + l +fa or say ^3m + ^n^ = 1 + A if
i\ =  DI + H + ^K.
79. It is to he observed with reference to .the applicability of these formula
within certain limits only, that the formula* are the only formula) which are generally
true; thus taking the simplest case, that of a single contact a, the only algebraical
expression for the number of the curves G>' which have with a given curve Z7* a con
tact ,,f the order , and besides pass through the requisite number 4r(' + 8) ' of
arbitrary points, is that given by the formula, viz.
(ft) = (a + 1) (rin a + aD)  ttK ,
Considering the curve 0* and the order r of the curve G * given if
s

but the e is no algebraic function of a which would give the number of
tllC iMYinPi 1 nni'uoo f f r or. ,.,li i i . ".". 5<vu unu llUHlUQr OI
ioe. cuives C as well beyond as up to the foregoing limiting value of a
f
of the n 4 
given points, a, therefo e ,,o ol, , f 7 " " illaid<lnt line l ir *8* *ho
the nibe/of th S> ^ I 2 ^""7 f"^ ^ ^ """^ ' i
coincident linepair hfvin, at elVoflf,' , . hna J inin these P ointe P
contact therewith, that I ha 1 , "J ""T* ^ the ^ our a%coial
with tbe given curve; if the numW n f*/ m ~. 1 A'2) ways three special contacts
case any line whate e th ol ,Tlh 1 ^ P iUlS " 1 ' ttoo in the fl *
whatever, .^dec, as . coS^nt Hne^ C^ , ^ "^ '"" ^ U " e
curve; and so in general there i, 7 , pe lal oolltaolls with the given
)vMch vah,e the cUenf 7 1LT u"", V!*, """"^ '' giV611 P """ fo "
imi^per curves 0', and for valued nfeS to iff ^v d9temillBta nnmbar f
mfimte series O f irapl . opei , Olu . veg ' mei101 . to ll the oonditoons may be satisfied by
De Jonquieres has determined the mini , ^ oomiderati "8 as these that
to which the conditions shou d ^T I 7 "1 ? ' th numbOT of arbit 'y Point S
I "for for his investigation and it ts n "?' " le f0 uto ^ be W^lo :
remark that in t g e ou whT^l n P T* P '? XVH " ld XVUI f his >".
can be corrected so L to riveT '' i Ilr0p " S luti n8 ia fcite  *e
give the number of proper solutions by simply
ON THE CURVES WHICH SATISFY GIVEN CONDITIONS.
231
406]
subtracting the number of the improper solutions: but this is not so when the improper
solutions are infinite in number; the mode of obtaining the approximate formula is
here to be sought in the considerations contained iu the first part of the present
Memoir; see iu particular ante, Nos. 8, 9 and 10.
HI. The expressions for (a), (a, b}, &c. may be considered as functions of rm, 1+A,
and K, and they vanish upon writing therein ?m=0, A=0, = 0; they are consequently
of the form (rm, A, /O' + tVm, A, ) s + &c., and I represent by [a], [a, b], &a the several
terms (rm, A, K)\ which are the portions of (a), (a, i>), &c. respectively, linear in rm, A,
and K. The terms in question are obtained with great facility ; thus, to fix the ideas,
considering the expressions for (a, b, c, d),
1. To obtain the term in rm, we may at once write D = 1, = 0, the expression
is thus reduced to
and the factor in { } being =rm[rm~a I] 3 , the coefficient of rm is
which is
=  (a + 1) (b + 1) (c 4 1) (d H !).(+ 1) (a I 2) (a + 3).
2. To obtain the term in A, writing rm=Q, = 0, and observing that
&c. give the terms A, A, A, 4 '2 A,  6A, &c. respectively, the term in A is
1) C [l] s a. 1} A
4
+ ft
+ Y
+ 28
S. 2
(a + 2) (a + 3)
A.
3. For the term in K S writing r = 0, D ~ 1, and observing that []', [] a , [] 3 ., [J*
give respectively the terms K, , 2, 6/c, this is
(a + 1) (6  1) {[ a  2] 2 + [ a  3] 1 ").!
SM(al) ([_*_3] l + a'"}. 2
. + a&od . 6 .
oqo
ON THE
nre
where the terms in (  nr
that is,
(+2)( + a) and 
respectively; whence the whole expression is
(a+3)
, 6 abed
the expression multiplying ( + 2)(+ 3 ) j,
K,
and we have moreover
Ihe series of fonmil
* attain from completing the reduction
aa s
+ ( + 1)
O.K,
and coeff . of .
.
C0effloi * of  expressed in term,
406]
ON THE CURVES WHICH SATISFY GIVEN CONDITIONS.
 /3
 7
23
233
G
where a =
, [a, b> o,
( Se
2S
6e
(c + rf + e + 3) (a 4 1) (6 + 1)
(6 + o + d + e + 4) (a { 1)
 ZAabcde
rf+e, jS = &c., , . , e =
A
 2 cd (c 4 d + 2) (a + 1) (I +1) ( + 3)
( + 4)
K,
(a +
where a =
83. The complete functions (a), (a, &),.(> 6. c), &c. may be expressed by moans of
the linear terms [a], [a, I], [a, b, c], &c. as follows, viz, we have
(a)
(a, b, c) =
M ra
k 6],
(, 6,0, *) W[6][o]M
5, d]
+ [o][6, o, q
+ [a, 6, o, d],
c. vi.
30
34 OX THE CURVES WHICH SATISFY GIVEN CONDITIONS.
and sti on; this is easily verified for (, &), and without much difficulty for (a, b, a)
but in the succeeding cases the actual verification would be very laborious.
S4. The theoretical foundation is as follows. Writing for greater distinctness (<0
in place uf (it), we have ( ( t) M to denote the number of the curves O' which haw
with a given curve 17'" a contact of the order a, and which besides pass through
ir(r + 3)tt points. Let the curve U m be the aggregate of two curves of the orders
>n, >: respectively, or say let the curve U m be the two curves m, in', then we have
()m+w' = (). +(tt) m ;
u functional equation, the solution of which is
* ()w [].
where [a] is a linear fnuction of , ,, *, or, what is the .samo thing, of ,, A, AT.
;r " when the coefficients  "' w * >~
Similarly if (, J) m denote the number of the curve, G" which havo with the
of h
pomts  then lf the given
where {() (6),,,.} is the number of the pnrwm (^ win'ni, i LI
order a and with ,' a contact of rt P S! , , 1 hflTO Wlth m a ntflfit o( ' thc
points; and the lil , S ^ 1 en' nn * T ^""^ ^ ^^ + >" *
which are not too great ^^ of , "T'^' bnt ** Tnllie8 f ftnd
order, of ^^ ^
and thence the functional equation
(. aW*(, i) H1 ( ff) fi
But [a], rtl &c . being l ilieai . functions of % ^
, , M+*M+M m .,
and thence a particular solution of th, ,n f  
general solution is therefore q l n 1S at onoe seen to be [a] m [b] M ; the
. W)
present "de^^ A, . Henee, assuming fo, fc ho
formula (, 6) = [ a ] [i] + W uM be fo ^ d to be fc 6] , TO \ ave the
406] ON THE CURVES WHICH SATISFY GIVEN CONDITIONS. 235
The investigation of the expression for (, 6, c),,, depends in like manner on the
assumption that we have
and so in the succeeding cases; and wo thus, within the limits in which these
assumptions are correct, obtain the series of formula) for (a, &), (a, b, c)....
85, It is to be observed in the investigation of (a, 6) that if a~b t the two
terms [a] m [&], and [],[&], become equal, and the equal value must be taken not
twice but only once, that is, the functional equation is
and the solution, writing ^ [a, a],,, for the arbitrary linear function, is
0, a) m = ^ [a] m [a] m + % [d., a] m ,
in which solution it would appear, by the determination of the arbitrary function, that
[a, a] has the value obtained from [a, b] by writing therein b = a. Writing the
equation in the form
(a, ft) H]
and comparing ' with this equation for (a, b), wo see that [a, b] is not to be considered
as acquiring any divisor when b is put = a, but that the divisor is introduced as a
divisor of the whole righthand side of the equation in virtue of tlie remark as to
the divisor of the functions (a, b), (a, 6, c).,. in the caso of any equalities between
the numbers (a, 6, c...). This is generally the case, and the foregoing expressions for
[., b], [a, b, c], &c. are thus to be regarded as trno without modification even in the
case of any equalities among the numbers a, 6, c..,,
86, To complete according to the foregoing method the determination of the
expressions for (a), (a, 6), . . , we have to determine the linear functions [a], [a, b], &o.,
which are each of them of the form fni + gk + kic, where (/, g t A) are functions of r
and of a, b, &c.; and I observe that the determination can be effected if wo knofr
tbe values of (cr), (a, 6), &o. in the cases of a unicursal curve without cusps 'and
with a single cusp respectively. Thus assume that in these two cases respectively
we have
(a) (rt + l)(m a),
(a) ~ (a + 1) (rm a) a.
Waiting first A = l, 0, and secondly A = l, = 1, we have
(a + 1) (i w  ft)  a =ftn g + h,
whence
/=(afl)r, 0(a + l)a, Aa,
302
36 *>N" T!K CURVES WHICH SATISFY GIVEN CONDITIONS. ~4G(5
giving the foreg.iing value
[]  (a + 1} it + (a + 1) fflA  .
Siiuilfirly, flu two (.ontacts assume that we have in the two cases respectively
(, 1) = (ti + 1) (fi + 1) [ m  a  b]\
String h,i from the formula fo >] = (, 6)  [a] 0] /fo +, A + A., and
MU^., IU 'Jy A =  1, = (j, and A =  1, * = 1, we have
( + 1) (6 + 1) [m  fl  6f  (( + 1) ( m _ fl)) ( (i + 1} (m _ fi)) =/m _ ^
( + l)(Hl)[m6p_ (0(6 + 1) + fi(a+1) j [m _ n _ fti1]1
 ( + 1) (m  a)  a} {(b + 1) ( m _ 6) 6 j ^y^  ff
the fir,t of which, putting therein a + ft, aftft ft fc onee redueed to
; wheilce h
87. The actual calculation of fa 4 d w,,,,i,i , , , .
l>uo,,t terms still more so; but it i' s 1 /, , .' labmo , and that of Oho
the foregoing values, assuming them to be J!! u } pnu<riple a PP lia8 ' and >at
for , uuicursal ourv, withou^cusp," that ' W UU be btabad if "^ know,
the number of contacts a, I, \ ,, .
that '^ a " d fol a
w, with a si
ollap ,
. that the liminution of ( o 6 , . ,
( '' C ' )oCCaslmet "=y the single cusp i
ON THE CURVES WHICH SATISFY GIVEN CONDITIONS. 237
88. Consider a unicuraal curve U m , and a curve & having therewith t contacts
of ^ the orders a, b, c, ... respectively. The coordinates (at, y, z) of any poinfc of the
umcursal curve are given as functions of the order in of a variable parameter 0;
and substituting these values in the equation of the curve G 1 ', we have an equation
of the degree rm in 0, but containing the coefficients of G f linearly; this equation
gives of course the values of 6 which correspond to the nn intersections of the two
curves. Hence in order that the curve G r may have the prescribed contacts with U m ,
the equation of the degree rni in 8 must have t systems of equal roots, viz. a
system of a equal roots, another system of b equal roots, &c. ; this implies between
the coefficients of the equation an ( + b + c ...) fold relation, which may be shown to
bo of the order (u + l)(6 + l)(e+])...[rm(a + & + c . ..)]'; and since the coefficients
in question are linear in regard to the coefficients in the equation of the curve C f ,
the order of the relation between the lastmentioned coefficients has the same value ;
that is, the number of the curves C" 1 which have the prescribed contacts with the
nnicursal curve U m and besides pass through the requisite number of given points, is
= (a + 1) (b + 1) (r, + 1.) . . , [rm  (a + b + c . ..)]'.
Hi). The reduction in the case of a cusp appears to bu caused as follows :
Consider on the curve lf m a points indefinitely near to the cusp, and let the condition
of the curve C/ having the contact of the oth order be replaced by the condition
of passing through the a points; that is, consider the curves O 1 ' which have with the
curve U m (t l) contacts of the orders b, c, ... respectively, which pass through the a
points ori the curve U m in the neighbourhood of the cusp, and which also pass
through the requisite number of arbitrary points. The number of these curves is
=s(6 + l)(c + l)...[?wa(6+o + ..)] i  1 (the term rnta instead of rm, on account
of the given a points on the curve: compare herewith Do Jouquieres' formula con
taining nn~p\ Kach of these curves, in that it passes through a points in the
neighbourhood of the cusp, will ipao facto pass through a + 1 points (viz, a curve
which simply passes through the cusp of a cuspidal curve meets the cuspidal curve
there in two points, a curve which touches the cuspidal tangent meets the curve in
three points, &c,), and be consequently, in an improper sense, a curve' having a contact
of the ath order with the given curve U m , I assume that it counts as such curve
a times, and this being so, we have, on account of the curves in question, a reduction
= a (6+ 1) (c H l)...[n?i  (a + b + c ...)]'"'. We have in like manner for the curves
passing through b points in the neighbourhood of the cusp a reduction b(a 4l)(o + l)..,
[nn (a + b 4 c ...IT" 1 . &c.. and hence when the ffiven unicnrsr 1 ~ T7 "" w ' 
 SR OX THE CTJBVES WHICH SATISFY GIVEN CONDITIONS.  .f
* a ..mi. (tlint i* j=2) first in terms of in, n, K) ami finally in limns oC m, .
(a ~;j ( , .f A  ., s nVtvc). The results are
for ,='2, that is, ourvo O 1 ' n onnio.
2,jy + i,
B+
4w+ (IH+ 8
[1J 2. 2A A
i] aiir A 2
(! 4n,if 12A 3
HJ= i>j,.n 20A 4(r
W i;^..i. SOA U K , = _ 18ffl+ lfin +
(1 1J  UV;i!  20A+ IK
' "'" ( ' A " i " 16/f = 12iit B0
~ Jx = 56m  BSit 
4Gx \ = 140m  180 
144w
82+
'JlH
HH
[1. 1,1,'J]= W)Gj m _ 2(}7
Till 17 i^n
on. i each ca,e equal.)
 In the case of the
(1 ::)
(2 ..)
10,.
628+ aH%
=  1184m + 1864M+ 780,,
=  1188,,,+ l8fifl B+
i2wii m,, ... , a
fi(im. ,,., ;n, u
HOWH 132 Kir,
WJHI. dH .f() r ,
=  82wt
=  88m
=  118dm lOWwi
 118w* lllfl,. H
= 820mh SOlOw ail
= lOOfiflHH lOlifif!,;.
 10018m
that in th iftHi
, tho cooffioio.itH of
cases i
 3fl+ 3fl
(5)
406]
ON THE CURVES WHICH SATISFY GIVEN CONDITION'S.
239
2 (1, 1 ..) = (2i + )'
4m n 3a ;
(1, 2 :) = (2m 4 ft) a
(1, 3 .)
(I, 4)
2 (2, 2 )
(2, 3)
G (1, 1, 1 :) =
(2m + ,) ( fan  3ft 1 Sa)
56m + 49 39;
(2m + n) ( 10m  8ft + Go)
140rn + 122 84
[(m  3) ( 12m  Gn + 60)] (= 
3) (4i + 2)]) ;
a (~ 4m  3n + 3a)
144m+126n90a
[24m + Qn + (n  12) o] (=  [6r + (  3) ]) ;
 4m  n  3a)
78a
2 (2, 1, 1 .) =
2 (3, 1, 1)
2 (2, 2,
n) (law* i^. i.<*an i w u
336m 336ft 4 288a
 [2a (m  2) (m  3)] (=  [6ft (m  2) (m  3) + 2* (m  2) (m  3)]) ;
= (2m + iif ( 4m  3?i + 3a)
4 2 (2m + n) (56m + 49w  39a)
+ ^_ 4 nt _ w _g)( 4m 3ft + 3a)
1184m 1094ft + 786a
'31Qwi226ft~l;
(=
+ (2m + n) (64m 4 48?t  40a)
11887ft 1110ft 820
S(2?i + 10m 38)
4 4 (6m 2  41m + 69)
+ T (8m  32)
OX THE CTBVES WHICH SATISFY GIVEN CONDITIONS.
'406
2* (1,
fa (
+ IS ( wt  4) r
+ 3) (4* + 2*)
+ 4 (2w + w) ( 32i  58. + 78a)
+ 2208m +26102858a
LiiGfl (t2)( t 3)
(=[(m2)(m3
14m
(2, 1, 1, 1} =
3 (2m + w )
68n + 78s
10650m + 10656?i8016
1040m + 24  2256)
where the correction is
(m4) '
 (?B  4)
.+ * (+16n06)
406] ON THE CURVES WHICH SATISFY GIVEN CONDITIONS.
120 (1,1,1,1,1)= (Sm + w)"
+ 1 (2m + n) s (~ 4m  n  3a)
+ 10 (2m + tt)( 32m  68 + 78a)
+ 1 ( 4m  w  3a) ( 32m  58?i + 78a)
+ 5 (2m + n) (2208m + 2610  2358a)
241
102912m
31m 6 +
 810m 4 
 235m 3 
+ 10690m 3 +16080nm +960
+ 2970m 3 + 900m
15630m 
^+ 28440
a a (135m 540),
where the correction is
= (rn~4)f
 186m 8 979m a + 6774m *) ,
+ n (70?ft 8  180m 3  1750m + 9060)
/~210m fl h2130m7110'
180tt m1
which is
=  ( m _ 4) / 31 (28 + 8/c) 3 + 110 (28 + S) (2r + 3t)
+ (^m 3 + 1142m + 3174 )( * t)
+ (14m ~638 1524 )(2S + 3/c)
+ ( 390m I 110?i + 4272 ) (2r + 3t )
+ ( 210m 3  180mw + 2130m+ 990?i  7110)
but I have not sought to further reduce this expression, not knowing the proper form
in which to present it.
92. The question which ought now to be considered is to determine the corrections
or supplements which should be applied to the foregoing expressions (a), (a, 6), &c,, or
to their equivalents [], [ti] [6] + [a, &], &o. in order to obtain formulae for. the cases
O. VI. 31
242 ON THE OUBVE8 WHICH SATISFY OIVHK OOMDITIOJffl. [ 40
analogous to the formute I which "t L Vf H ^ M "" tra sfo ti<
not considered. 2 P ' "' thls ls a question which I have
which
A,x No.
ana ly tical
Annex No, l
to in the notice of D E JoNQUlto , memoil . of x

of
i t i
' Wo b the
'', L lse
linearly and homogenco,,^
,
f

+ 8 T 1
these "efflaientB so
param6tera ' Ass me '
""' " le m ' Ve should f '
of the
pam.netes
the pammetevs (,
but ?t is oonvolL
as to reduce as far as nossible
number of the
.series (that i s ,
botwaai the
coordinates of a
question must be
shall pass through a
(m that expressed b o a r
therein as belonging To ft. Z" I !' t f m ' Ve "W^B the
curve parses thrfugh a gTvo.fpoint'Z'r 'v **" " mtats)! that is ' when
are given a s the LelS P 7l t iS l^ T^ f ^ P"
number of the curves i US by an omal mefM lo s; the
u
on a
noinf
t .
thlll & tak S parameters to be th.
C T ""r^.^ th ^' ">
Moreover ' the "dition that the cnrvo
^ pmmetm a HneM '
,,,. ., .
406] ON THE CURVES WHICH SATISFY GIVEN CONDITIONS. 243
of the order n containing linearly and homogeneously the w I 1 coordinates of a certain
(&) 1) fold locus of the order N. It is only in a particular case, viz, that in which
the (to  1) fold locus is unicursal, that the coordinates of a point of this locus can
be expressed as rational and integral functions, of the order N of a variable para
meter 8 ; and consequently only in this same ease that the equation of the curves
C n of the series of the index iV can he expressed by an equation (*$, y, z) n = 0,
or (* JflJ, y, 1)" 0, rational and integral of the degree A r in regard to a variable
parameter 6.
If in the general case we regard the coordinates of the parametric point as
irrational functions of a variable parameter 8, then rationalising in regard to 6, we
obtain an equation rational of the order N in 0, but the order in the coordinates
instead of being = n, is equal to a multiple of ?z, say qn> Such an equation represents
not a single curve but q distinct curves C n ., and it is to be observed that if we
determine the parameter by substituting therein for the coordinates their values at a
given point, then to each of the N values of the parameter there corresponds a system
of q curves, only one of which passes through the given point, the other q 1 curves
are curves not passing through the given point, and having no proper connexion with
the curves which satisfy this condition,
Returning to the proper representation of the series by means of an equation con
taining the coordinates of the parametric point, say an equation (*$#, y, I)" 1 0,
involving the two coordinates ((&, y), it is to be noticed that forming the derived equation
and eliminating the coordinates of the parametric point, we obtain an equation rational
in the coordinates (so, y\ and also rational of the degree N in the differential coefficient
CL1I
~; in fact since the number of curves through any given point (# , y ) is =N t the
differential equation must give this number of directions of passage from the point
(a) , 2/ ) to a consecutive point, that is, it must give this number of values of ~~ , and
(lit)
must consequently be of the order N in this quantity.
Conversely, if a given differential equation rational in <o, y, p, and of the degree
ft?/
JV in the lastmentioned quantity ~, admit of an algebraical general integral, the
curves represented by this integral equation may be taken to be irreducible curves,,
and this being so they will be curves of a certain order n forming a aeries of the
index JV; whence the general integral (assumed to be algebraical) is given by an equation
of the abovementioned form, viz, an equation rational of a certain order n in the
coordinates, and containing linearly and homogeneously the tu + 1 coordinates of a
variable parametric point situate on an (to 1) fold locus. The integral equation
expressed in the more usual form of an equation rational of the order N in regard to
the parameter or constant of integration, will be in regard to the coordinates of an
order equal to a multiple of n, say =qn, and for any given value of the parameter
will represent not a single curve G n , but a system of 5 such curves : the first
mentioned form ia, it is clear, the one to be preferred.
31 2
244 ON' THE CURVES WHICH SATISFY GIVEN CONDITION'S. [40G
Annex No. 2 (referred to, No. 17). On, the linepairs which pass through three given
points and touch a given conic,
Taking the given points to be the angles of the triangle formed by the linos
(a= 0, y= 0, 3= 0), we have to find (/, y, h) such that the eonic (0, 0, O,/, g t hfa, y, 2f = 0,
or, what is the same thing, /^ + ^' + % = 0, shall reduce itself to a linopair, and shall
touch a given conic (1, 1, 1, \, /*, v \, y, *)=*0. The condition for a linopair is that
one of the quantities/ g, h shall vanish, viz. it iafgh = Qi the condition for the contact
of the two conies is found in the usual manner by equating to zero tlio discriminant of
the t^ifr + 8fy(p+0g?( v+ ehy + 2(\+ef)(r + $g)( v + 0h)**(a, b, G>
snppo.se ; the values of a, b, c, d being
= ~ I (/" I
c= v
Hence Considering (/ (J) h) as the coordinates of the parametric point, wo have blio
discriminantlocus a = 0, and the contactlocus
a?d* + tac 3 + Wd  36V  Gabcd = 0,
and at the intersection of the two loci a = to fMij Q^\ A , ,
t ^i , , ' U) y t* 0ct ~ c ) = 0, eouationa brenkinn mt
" 8yStem (a = ' 6 = } "^ imd ^ S ^ m O/^lsko; 'tha 2, of
= 0, /H 5" + A'  A  2HF  = 0,
^
noticed above thisto' " /l = ' "^ / a + ff a 2X/ = 0. A.
The second system is
=o,
""  + (* ~ ti g + (/A  ) A)' = 0,
<*, s the second equation may also be mitten,
.Jffi ( tou tr .
"(*+)0. in which the Bne *v + 0 t^C ^ ' ^ ha hel ' S the Unfl P odr
and the like if , . 0, or if h = . T^ Jy te m 1 ^ 0m ( ' '' '' X ' "' "^ '
LMS system it has been seen occurs only once.
406] ON THE CURVES WHICH SATISFY GIVEN CONDITIONS. 245
Annex No. 3 (referred to, No. 22). On the conies which pass through two given points
and touch a given conic.
Consider the conies which pass through two given points and touch a given conic.
We may take #=0 as the equation of the line through the two given points, and
then taking the pole of this line in regard to the given conic and joining it with the
bwo given points respectively, the equations of the joining tines may be taken to be
X**Q and F0 respectively. This being so, we have for the given points (A r =0, #=0)
and (F=0, #=0) respectively, and for the given conic
aX'+ &P + 2hXY + c^ 2  ;
and since the required conic is to pass through the two given points its equation will
be of the form
= 0,
where (x, y, z, w) are variable parameters which must satisfy a single condition .in order
that the lastmentioned conic may touch the given conic. The condition is at once
seen to be that obtained by making the equation
bo
considered as a cubic equation in X, have a pair of equal roots ; or if we write
=  ax*  by* cz* + 2fe (icy zw),
D =
then the required condition is
Hence the conic
wX* + 2a; YZ 4
satisfies the prescribed conditions, if only the parameters (as, y, z, w) satisfy the last
mentioned equation, that is, if (to, y, z t w) are the coordinates of a point on the sextic
surface represented by this equation.
The surface has upon it a cuspidal curve the equations whereof are
A, B, a =0;
5, (7, D
'240 ON THE CUftVES WHICH SATISFY GIVEN CONDITIONS. /tQ<>
this iriiiy be; considered as the intersection of the qnadric surface Ad ./&() ami UK?
cubic .surface AD SC'= 0; and the cuspidal curve is consequently a soxfiic.
The surface has^alao a nodal curve made up of two conies; to prove thin I writo lor
shortness & = /<~V6, ^ = h + ^ab; the values of A, , 0, D then are
A = ScA'AJ] ,
S ~  kL\w 
G =  ax 3  b
and it is in the first place to be shown that the surface contains the conic
is : y : z : w =
/a '
whore 6 is a variable parameter. Substituting these values, wo have
and hence
val.es which Sa ti sfy identic^ the equation rf
fco subsfcitute
M'.W:M = Q>fr.M t Z
then the derived equation
s
)aa0
406 J ON THE CURVES WHICH SATISFY GIVEN CONDITIONS. 24
that is,
It I 14
tb I OL
or finally
 k (A  3.Bfc + mG  I<?D] = 0,
which is satisfied by the foregoing values of A, B> G, D; hence the conic is a nodi
curve on the sextic ; and by merely changing the sign of one of the radicals \/o, ^l (an
therefore interchanging Jc, lc^) we obtain another conio which is also a nodal curve c
the surface, that is, we have as nodal curves the two conies
\a:y:z:wt=0/bi 6*fa, : 1 : kP + j, and
at : y : s : w^B^/b :tfVa : 1 : ^5* + ^.
It is to be remarked that each of the nodal conies meets the cuspidal curve in tv
points, viz, writing for shortness ** r \/~ C , . , = r \/  for the interae
K v /C'i KI v 1C
fcions of the first conic wo have
x : y : z \ w = Va : V& : 1 c  and =  @
and for the intersections with the second conic
K ; y : z : w~%^& >. ~%^b: .1 : r and =^
248 ON THE CURVES WHICH SATISFY GIVEN CONDITIONS. [40(>
of the sextic surface leads to the following values (agreeing with thoau obtained from
the formnlro by writing therein m = n = %, a = 6), vk
(1 ;:) =6, = 27/1+71,
(1, 1 /.) = 4, = 2w + Zmn + $n*  2m  fri  {fa,
(2.'.) = 0, = ,
(3:) =4, =4m3ft+3a.
I remark that the section by an arbitrary plane is a sextic curve having cuaps and
4 nodes; it is therefore a wrimmd sextic; this suggests tho thoorom that bhu rn.xl.u
surface is also W ^d, viz. that the coordinates aro oxpioaaiblo lutionally in toniiH of
two parameters; I have found that thia is in faet tho cruse. In doing fchi 8 Uioro is
no loss of generahty m supposing that a6ol; and aiming That thin IK .,,
and puttmg also 1+A^, 1 + 4 = ^ and therefore 37^^ + ^ w h ^ vo
The equation of the sextic surface being, as before,
AW
3^0  C>A80J) 0,
.1 say that this equation is satisfied on writing therein
e (, 0) are arbitrary, fc f act
theae
. ^
whence, being arbitrary, We hftve
, Cf, J?^ j)t
  [* co
406] ON THE CURVES WHICH SATISFY GIVEN CONDITIONS.
viz. the equation (A t S, 0, D^ca, 1) 3 = 0, considered as a cubic equation in o>, has the
twofold root w = a, that is, we have the above relation between (A, B, 0, D). Whence
o\ i _ \ a
also writing sin^ = ^ _, cos< = , the equation of the surface is satisfied by
the values
 : z : w
/ 2
= A/ ^ (1  /c,a)
V ri.^
: 2   (1  \) + 2  r
\ 7 \ Is
or the coordinates are expressed rationally in terms of a, \.
Annex No. 4 (referred to, Nos, 22 and 71). On the Gonics which touch a cuspidal cubic.
In tho cuspidal cubic, if a = be the equation of the tangent at the cusp, y =
that of the line joining the cusp with the inflexion, and # = that of the tangent at
the cusp, then tho equation of the curve is f=a?z; the coordinates of a point on the
cubic are given by as : y : z=>\ \ : 6 s , where 6 is a variable parameter; and we have,
at the cusp #^co, at the inflexion = 0. In the cubic, m = = 3 ) a (= 8n + K) = 10.
Considering now tho conic
250 ON TIIF, oimvKM wiiirn HATIMIA* <;ivrv OI\HMHI\M. (int.
ivtid wo lliim obtain Uio mjimtiioii ol' lh*< runi^'l Ixnr, in Hi l>itn
((to '\llli I Hi:') 1  2V In.,'  ^ftfil \nt AV I '>'  H.
lioh is a oniifnld locim of Mm unlnr li. It lii!loW! llin? r hut,^
(UI, I ;.) '1. "Kivimif! with iK'l. 1 i n ,,., ;i.
whioh in a oiirlultl lntntH of Mm nnlcr ^; ir it! .MHT i'..].fiM i,,r
(Id, I :/)> IU, ti^ivnitiK witli iKJ,  I..M.
Tli is a iimUiir of Hitmn ililllmilty IM nlmiv rlmr v,< hit',.
(![, I //) :i M, i^ivt'iiiK with tUl. 1 * .
but 1 iirmiitt'il to oniidl, UIIH, 'H M , rciiinrl{ih K (bit I .M ,,MI
of Itm valuu (, I . //) , IK ;
"Wn liavi) thn noxl.ii! lomin
and ciiiiiliiiufd tluirnwilh f,wo
(* }, s , fc, /, ac..,:i,^ tl f M/,/ ,
,tah into,,!, in a thr^lol.l I.H...H ,,r rlin ,, ( , r ,*, lt
'mtun, IIH rt ol M { ,, hn ,,! IV , .
thrao turuiH, lcavin K a msidnal
W may iiuaginn (,1m ,,,. ,, ( R( ,. f/ ft ,^,,,^^.^1 ,
four ciwnliimluH, and HII iiidui'n iht* iu,,1,1.,',./ rJ t 't!' Hlw
i ii,vui,u ,jp j)tMuh?ni lnnn i" '  >
HlOllttl H]HM't. Wn luiVM Uli
a (Icivulopaliltt nnrf, wv tir ^ tf
(a0, 60, Ucfl^Mfh wl'n'li, '" ' l t ! U ' lrtk! l ' nrvi> I sin K Uirwuyh M. IWM
\ v, ^ w, oca . MUJ, \vintsi! art( ]l((lU j , _ * t
of tlioflo pomb counts ihwo Untw mnur,, M,,, n,,^^ ! ' !"
uurvo, tiio nnmbor of
it in to b
406] ON THE CURVES WHICH SATISFY GIVEN CONDITIONS. 251
is situate on the nodal line of the torse, and that the quartic curve touches there the
sheet which is not touched by the tangent plane a = 0; for this being so the _ quartic
curve touching one sheet and simply meeting the other sheet meets the torae m three
consecutive points, or the two points of intersection count each of them three times.
The torse has the cuspidal line
fif = ae46rfl3c a = 0, T = ace + Zbcd  ad? fre e a = 0,
and the nodal line
a
and the equations of the nodal line are satisfied by the values (a = 0, &0, 3ceU 0)
of the coordinates of the points in question. To find the tangent planes at has
points, starting from the equation ff27P0 of the torse, taking (A, S, 0, J), Z}
as current coordinates, and writing
=
then the equation of the tangent plane is in the first instance given in the form.
SSslMWO, which writing therein (a  0. 60 B Cfl W0) assuine. as lt shouM
do, the form = 0; the lefthand side is in fact foun^ Mo
Proceeding to the second derived equation, tins is JSW+2S0S) 
or substituting the values of the several terras, the equation is
the terms in BO, BJ>, 0* vanish identically, that in * is (4836 .
.Ie7(toW)i. wWoh also vanishes; hence there remain only the terms
by A, giving first the tangent piano ,1 = 0, and secondly the other tangent plane,
+ ^.27o" =0
Taking the equations of the quadrio surfaces to be
(\, p, v, p, o, Tja', 6", o&, 8062*. M
(V, ^.^ p',^,
252 ON THE CUBVES WHICH SATISFY GIVEN CONDITIONS,
tlie equations of the tangent planes are
p (3oE + SeC 4dD) + a (eA  SdB) + T (dA  12cB) = 0,
"406
in all which equations we have 3ce 2tZ J =0; and if to satisfy this equation wo write
c : d : e=2 : 3/9 : 3^, then the equations of the tangent planes become
($ (Aft  8B) + 8 (30/9'  4D/9 + 27?) = 0,
(3GB  4D/3 + 2) + (<r/3 + T) (,4/9  8.5) = 0,
or the three tangent planes intersect in the line .A#8# = 0,
which completes the proof.
Reverting to the sextic locus,
(ae + 4bd  3c 3 ) 2  27 (ace + Zbcd ~ ad?  fre  c 11 ) 2 = 0,
considered as a locus in 4dimensional space depending on the fivo cu.nclinatiw
(a, b, c, d, e), this has upon it the twofold locus
ae  4bd + 3c 3 = 0, ace 4 26ccZ  CM?  b*e e 3 = 0,
say the cuspidal locus, of the order 6, and the twofold locus
'"' ;6 3 ), S(adbc), ae + 2M3c a , 3 (fie erf), 6(ced a ) =0,
say the nodal locus, of the order 4; there is also a threefold locus,
a, b, G, d ~0,
&, c, c, e
say the supercuspidal locus, of the order 4. We thence at once infer
(1*1, 2 ;) = 6, agreeing with (jl, 2 :) = a4,
[1 1 ,\ = A ,', r  .
' '' ' " ti**) 1, J :) = 2)/i a + 2m)i + i)i B ~ 8??;
' }
but I have not investigated the application to the symbols with ./ or //.
the cusp,
! = of the
cuspidal
s
ON THE CURVES WHICH SATISFY GIVEN CONDITIONS. 253
isumo the equation of the conic to be (d, 3&, 0, 0, a, ;}c.';, y, z)=0. The equation
1 the contactlocus then is
K. till is is a developable surface, or torse, of the order 4, and we at once infer
(2*1, 1:) = 4, agreeing with (21, 1 ;) 2m + n 5.
will nhow also that we have
(2*1, 1 /) = 6, agreeing with (2*1, 1 /) = 2m. + 2w  6,
ml
(2*1, 1 //) = 5, (2*1, 1 //) m + 2n  4.
'ho condition that the conic may touch an arbitrary line KOI + /9y + 70 = 0, is in fact
finch, considering therein (a, &, c, rf) as coordinates, is the equation of a quadric surface
lassing fchrough the conic  0, 4M3c 2 =0; the quartic torse also passes through
hiH conic; hence the quadric surface and the torse intersect in this conic, which is
,f tiho order 2, and in a residual curve of the order 6; and the number of the
ionics (2*1, 1 /) is equal to the order of this residual curve, that is, it is =6.
If the conic touch a second arbitrary line ' + &y + /* = 0, then we have in like
imunur t.ho quadrio surface
(0, ft, f(4M3c 3 ), fac, fftft, OJ[', ff, 7) 2 = 0;
ihat in wo have the quartic torse and two quadric surfaces, each passing through the
onie rt0 ^cZBc a 0, and it is to be shown that the number of intersections not
in this conk' IK  5 The two quadric surfaces intersect in the come and in a second
rmic this second conic meets the torse in 8 points, but 2 of these coincide with the
noint'(t0 6=0 0=0, which is one of the intersections of the two comes (the point
^Tn Lo 1 = in in fact a point on the cuspidal edge of the torse, and, the conic
passing through it, reckons for 2 intersections), and 1 of the 8 point, comcide 8 with
tho other of tho intersections of the two comes; there remam therefoie 821, 5
in tor sections, or wo have (2*1, 1 //) = .
Annex No. 5 (referred to, Nos. 22 and 71). On the Gonics which ^^f^^
of the cuspidal cubic be fc!/'0 (*0 tangent at c
1( Hue joining cusp and inflexion; equation satisfied by
and let the equation of the given conio be
Z7=(a, 6, o,f, g,
254 OX TEE CUBVES WHICH SATISFY GIVEN CONDITIONS. [40G
then writing
 c^ + 2/0 1 + 2t/0 3 + 6# 2 H 2A0 + c,
thy equation of a conic having with the given cubic ab a given point (1, #, ^ a )
contact of the second order, and having double contact with the given conic, is
=
\t,
viz. in the rational form this is
a, y, z
Ve, i, 0, &
(Vey . i, 30 s
(V)" . . 60
V, i, 0, ^ 3
(V@y . i, 3^
. . 6*
viz. this is
(VI)" .
0,
or developing and multiplying by * this is
or, what is the same thing,
and substituting for its value, this
a)
^
, = ,
s
406]
ON THE CURVES WHICH SATISFY GIVEN CONDITIONS.
255
The coefficients of the powers 16, 15, 14, 13 of 8 all vanish, so that this is in
fact an equation of the twelfth order (*p, 1)" = 0; and putting, as usual,
(fcc/ 3 , ca f f, abh*, gh<tf, hfl>ff,fgoK)**(A, B t 0, F, G, H),
the equation is found to be
+ SOoff
I 'i'
 IftJttU
6^  20/C
 36eS)
 22W?,
+ lOAffJ
+ 16/Ar + * 0ft "^
+ 6hF]
lOcff"
 ISO/iff
/to
Oa '
+ 20M. \
+ 40/F
 12rf,
60/B
a + 33AJ5\ ~ 5;tC/
 0*
w
= 0,
here the form of the coefficients may be modified by means of the identical equations
(A, H, fffa, kg) = K.,
(H,B,
(ff, J 1 ,
There is consequently a conic answering to each value of given by this equation, or
we have in all 12 conies.
In the case where the given conic breaks np into a pair of lines, or say,
(a, &, c, /, ff t lifa, y, 2) B 
then, writing for shortness
we have
(A, B, G t F, <?, H) = (^, P, &, YZ, ZX, XY).
256
ON THE CUBVES WHICH SATISFY GIVEN CONDITIONS.
[40(5
Substituting these values, but retaining (a, 6, c> f, g, h) as standing for their values
0.2XV, &c., the equation in Q is found to contain the cubic factor 2X0 J  3 Y&* + %,
where it is to be observed that this factor equated to zero determines the values of
6 which correspond to the points of contact with the cuspidal cubic of the tangents
from the point (X, Y, Z}, which is the intersection of the linos Tuw + /*# + y* = 0, ami
' i'*=0; and omitting the cubic factor, the residual equation is found to be
icA
12cF
8A
12/F
106AT
* 3//F
:*M
3 17AF
: ! s
ia^
where the form of the coefficients may be modified by means of the identical equations
The equation is of the 9th order, and there are consequently 9
cones.
Annex No. C (referred to, No. 48).tfontei fl r l with the variation referred to in the
teat, ZEm'HEN's/oms/or the cAaraofemtfos of the conies which mfofy four conditions.
(1)
(1, 1)
2m,
= 2m(
2 (2m
T,
(1. 1, 1)
 tt '  80m 
84m 
+ 3n  26)
406] ON THE CURVES WHICH SATISFY GIVEN CONDITIONS. 257
+ 4 (m 2  11m + 28) T 12 (n a  llw + 28) S
+ 2 (m 2  llm + 28) T 14 (?i 2  11 + 28)
+ (4 (n  4) (w  4)  1) (8 + 2r) + S a 1 2r a } ;
(2)
(..)= 3m +t,
(.//) = 2 (3m
(2, 1)
( . / ) = 2 (3m 4 (m + n  12) + 24 (m H ft),
( // ) = 3 (m" + 2mn  10m + 4n) + (m + 2  14.)
(2, 1,1) , x
. = (2m + n7)(6T + (i3)}
5) + T) (3m + *  36)
+ ((n  m) (m +  5) + $) (3m + 1  36)
2 2^
n 3")
(.) 2(4
(/)= 2( 3
(4)
. = 107i 10m + 6K= 8m
0. VI.
258 ON THE CURVES WHICH SATISFY GIVEN CONDITIONS. [406
.. Annex No. 7 (referred to, No. 93),
In connexion with De Jonquieres' formula, I have been led to conaidm tho following
question.
Given a set of equations :
a = a (viz. b = b, c c, &c,),
ab = ab /viz. ac= ao t &c., and the like in all tho aubHuquunt equations
+ ( ll)a.6\ + (H)a.c,
abe = abc
+ ( 12) (a. be + 6.00 +o.a&)
+ ( 111) a.b.o,
abed = abed
+ ( IS) (a. bed +&c.)
.cd +&c.)
+ (1111) a.l.o.d,
and so on indefinitely (where the ( ) is used to denote multiplication, and ab, aba, fed.,
and also ab, abc, &c. are so many separate and distinct flymbolH nob oxpixifmiblo in
terms of a, b, o &o., a, b, c &c.), then we have conversely a sol; of tu
a (viz. 6 =b, c=c &c.),
/viz.ac= ac c., and the like in all tho Kubsoquont cquationw
+ [ 11] a.b\ +[ll]a.c,
abc  abo
+ [ 12] (a . be + b . ac + e , ab)
, +[ 111] a.b.o,
abed  abed
+ [ 13] (a. bod +&c.)
+ [ 22](ab.cd +&c,)
+ [ 112](a.b.cd
+ [1111] a.b.o.d,
40G] ON THE CURVES WHICH SATISFY GIVEN CONDITIONS. 259.
and so on; and it is required to find the relation between the coefficients < ) and
[ ]; we find, for example,
[11] =  (11),
[12]  
= 3
(111),
 (13),
[22] =  (22), . 
[112] 2 (IS) (12)
+ (22) (11)
 . (112),
[1111] = 12 (18) (12) (11)
+ 4 (13) (HI)
. .  3 (22) (11) (11) :, '
+ 6 (112) (11)
(1111) j
and it is to bo noticed that, conversely, the ooofflotenW ( ) are given in torms of I he
Efficients [ ] by the like equation, ,ith the very same ~al m
fact from the last set of equations, this ia at once seen to be the case as feu as
(112) ; and for the next term (1111) we have
 4 M [3 [12] [11] [111]) + 4 [13] [HI]
+ 3 [22] [11] [H] +(3~fi > 3 [6][11][U]
[11]
\ +[22] [11]
I [112] J
having the same coefficients 12, +4, 3, + 6, I as in the formula for
in terms of the coefficients ( ); it is easy ; to infer that the property hold
generally.
To explain the law for the expression of the coefficients of either set in t
260
ON THE CURVES WHICH SATISFY GIVEN CONDITIONS.
'400
the other set, I consider, for example, the case where the sum of the mimbora ia tho
< ), or [ ] is = 5 ; and I form a kind of tree as follows :
13
11111
22 112 1111 H 111 12 HI 12 11 11
111 12 11 H 11
11
H
Jl
II
[14] =  (14),
[23] =  (23),
[113]= 2 (14) (13)
+ (23) (11)
(113),
[122]= (14) (22)
+ 2 (23) (12)
(112),
[1112] =  6 (14) (13) (12)
 3 (14) (22) (11)
+ 3 (14) (112)
 6 (23) (12) (11)
+ 3 (113) (12)
+ i (23) (111)
+ 3 (122) (11)
 i (1112),
40 G] ON THE CURVES WHICH SATISFY GIVEN CONDITIONS. 261
[11111] = + 60 (14) (13) (12) (11)
 20 (14) (13) (111)
+ 15 (14) (22) (11) (11)
 30 (14) (112) (11)
+ 5 (14) (1111)
+ 30 (23) (12) (11) (11)
 10 (23) (111) (11)
 30 (113) (12) (11)
+ 10 (113) (111)
+ 10
 i (11111).
To form tho symbolic parts, we follow each branch of the tree to each point of
its course : thus from the branch 113 we have
(113) belonging to [118],
(113) (111) [111H],
(113) (12) [1112],
(113) (12) (11) [11111];
viss. (118) belongs to [113]; (113) (111), read 11 (3 replaced by) 111, belongs to [11111];
(113) (12), read 11 (3 replaced by) 12, belongs to 1112; (113) (12) (11), read 11 (3
replaced by) 1 (2 replaced by) 11, belongs to [11111].
And observe that where (as, for example, with the symbol 122) there are branches
derived from two or more figures, we pursue each such branch separately, and also
all or any of them simultaneously to every point in the course of such branch or
branches ; thus for tho branch 122 we have
(122) belonging to [122],
(122) (11)
^
(122) (11) (11) [Hill].
Similarly for tho branch 23 we have
(23) belonging to [23],
(23) (111) .. [ m2 ]>
(23) (12) [122],
(23) (12) (11) (same as infrcl) [1112],
(28) (11) (111) " [Hill],
(23) (11) (12) (same as siqml) [1112],
(23) (11) (12) (11) [Hill]
262 ON THK OU.RVI'M WHICH HATIHKY (IIVKN ('O.NIHTln.VM.
We than obtain thu Hyuiholio parts nf t.hn Hcvrral rNprrwinnM I'm' >, 'J!I.. , . 1 1 1 1 1 1 1
respectively: tho sign of ouch tonn IH + or  nrruriliittf in lli,< iuii.il, n ,!' lai.n., in
( ) if* ovoii oi' odd; lilinn in tho oxpnNHmii lur 1 1 1 1 I 1 1, lh,< ,.,ni ( MK IM) { I'.!) ( 1 1 )
has four factor*, mul in UuTofnro .)., ,ho turn. (I !:)< li!)( 1 1 In., ilm, lii,.,, fl . m ,l i,
theroforo .
^ Tho numorioal (wflidionlH am ohtainnd UN liillmvM. Thtiv j ,.,. ..... ,,.1, (iuior
donvcHl from tho oxprctwi.m in [ ] un Mm lufMmn.l mi] ( , ,,r ,,,, nprnlinii; vj,:, (;,.
[11111J, wliioh contaiiiH llvu oijual HjinliolH, UHH furfur in t , ;!. ;i . ., ,",, . >i n , n' ( . J.H
which contaiiifl bhiiHi uiinul HyinlxilH. il, IH I . .  ( (! ; (U1 , , ...... , ,,;,', n ^.^
such as Li.1222] ooiitninin^ two equal symlmls, un.l i.h,.,, nj.iul ^ml,..!,, lit' 'd.iur
woud bo i. 2.1. 2. 8, 12, and HO in ,.!. H i,nilnr ,,,.). lu m,v !,,, ,. ihn rU, .
hand ado cil th u uquttfci,,,,, w , IIUts(l , i(l . a  h , (ill ,. Hl( . h ls , ,, , ' ivhj1 , ( l ,, II , airi ],,
equal flymbos .miltiply by i; fur a fnntnr ], ll!f (  , ,, vv  )irll ,,, ....... ,
symbols, m^ by ii( and nn. And in i\ w ,,,, win,,, u ,, tlH . .',, ,x,,, ,,
m ( 122 )(,,) or (gs)(l8)< , , vMtl }} (tr( , N imin . IHI ( ; ' I' 
'
(
bo taken aooonnt of HU , h U.no thai, it n,mn,; n, wlmi in ,,,. ,,. ltim , .MHV ,,,
oooffioioiit obta,nod nbovo in tho fi,r ra.h ,,.,.,,, , tll , , ..... ,!,:, l]l(lu ,
 above to bo multiplunl by thn nn.nbor of ,h, ,,,,, ,., ,1, L n
cxumplo, talcing in ordor tho s,voml hrnm of U,,, ,M,,,SM ( ,,, fll , ,.,,, ,,'
Actor iB0. and tho Hovoml ooollini.nl.s a,v ' ' '' "' (fIIIIIM " h
'1204. 1 20. A. II O.J. 1.1. ^..,
nL:
the whole InorenHocI by imUyuJlo; vii fi l "'" 1 ""' i " tl >'" ttlllitl1 '"**^.l u,,!
[ 3 *lPl[n.uiia2] 1 [nia],r mn i
the sums of tho oooffioionlH aro
and wo have ~ ' ' '" 2 ' "'* a ' " (i . +24 n*ipfrLivlv.
1 + (l) '
^. i" 5" r r
partitioned in tho forms 1.4,23 Mm I m> n " f W " P '" whioh lhtt i bo
theorem is Uwa a theorom in tho ft^UoJ. If ^ {^y. nd tho iMHnenUonod
" .....
264 HIICON 1 1 MKamnt UN 'nu; i.tn"
i *
numbor of tho propni 1 solutions, 1 IIH tJiiw inmll i lu il> himiiiM ,; ;<>',M t . lM im > n, h
CMC tho oxpruHHinn of tho Sni]ilt'tiiiml. ; l.hr r\n.> .inn M. ..hiuin. .1 . ;( !i m .,ia. 5 ., ,
bo tujcounliod for m\<\\\y omm^li, and Mm l<ini\vlf<I^i ! tli.< ulil vu., ,( rlum v,iil
bo a convonionli IHISJH lor ull.drior ii
in'c for pniufn in a lin>
th papor in tho dtitiijtton ,/iVW//,v, June July I Hill. ivt.>nr.l (.. ,., .; Kn y M, ;IJI . FJI
in oxtiindtid to unicnmil (inrvrw in a pup,.r ,,t ih, w.i.i. <,<n. ..,, M.utli Miitj
nun on a
lit^aUon (lu
not lio tho MVHO of u oiilvu of K ivt'ii .lulirirnry 1) r..i r .t<.>i.'. ih .,> ,,,,, , ,,l
a (Hirvu uf tho onln w will, H W ..)( H( jj) rt ' M)l ., ,,.,,,,1^ n..^^"'.^
coimlniiiliUH (:, ?A j) im > imii..r.i..i,,il to , (l ii, t , )l(  ttll , ;,..,.,, Il . (ill ,, l> ti ,
_... i i\  * '
1, but
, pp 4M (1% ,,,!, my M H.r.in i,
ml .StoiV//. No. Ul., (Vi. l.sii.'i. ;IH.
HATIHFY (HV.UN CONDITIONS. 2G5
JJii, Siippusi! Unit Mm nHTifMpowlintf points aro P, P' and imagine that when P
givmi tin c.niTtinpiiMiliiig puinU P' an tlm inliwioc.binnH of him givon curve by a curve
(l.hi' i>iiinl>itni ut 1 Mu! I'lirvn H will of (ionrw! contain lihu eoowUimtus of P as
TH, fur nUiinyim' llir position of .!" would not clopond upon Llmt of P). I
l.liitt. if Mil* nwvn H Ims with !.lui givon uni'v*) & intciwocbions at tho point P,
I hen in I. he n t vt.i'in tit' (mints ( P, /") Uun immbnv of nnil.iul point* is
win 'lint 1 in puH'i HI lit r il' (.In curvi* ( H ) ilntiH nol; pawn Mivnugli thu point P, then the
iniiiiliiM 1 nl' uiiilrd jininlH in a . ft', UH In l;hn IMSU of a iniicnii'Hal cui'vo, (I have in
Mii< pup'T nl' April IHlilj H.IIDVP niH>rri!(l to, provuil MUM Muiorani in tho particular case
wlii'i'i 1 tin 1 /, iiili'i'm'il.iiinM IL(. MKI jmlni. /' tnko {ilmm in conKoipionon of the curve f 1 )
liiivin.iv H /.'l.npli' pninl nl /', liul, have unl gone, intn thu inoro diflioulfc invsbigabion
Hi] 1 Mu nisi* wlicri' llir k inl.ui'si'rlioiiH jirisd wholly or in part from a oontaot of tlm
iiirvi' H, ni 1 any hrumh nr hnuii'ln's lhovnf, widh 1,1 HI givtiu nuvvo at /'.)
!Hi, . i .<i In iiltMirvnl thai. Um gintiMiil noLion of a nnitiMl poiuli is as follow*;
tnkiii^ Mir ptiiiit /' n!, niniliiin on thn jfiviiii itnrvu, tlm (surv H hus at thin point b
iiiliT.'ii'i'liiins \vilh l.hi ^ivi'ii rurvn ; l.lm niiuaiiniig iniovHodliionH aro liliu cori'CHpoiiding
p u iiilM /"; if Ihr a tfivi'ii psit.hm of /' onn or inoro tif tlm points P (some to
miiiftili* willi /', llint. in, il' I'm 1 ihi' given jiosiMtiu uf P tho ourvo f) has at bhis
point niniv limn / inhiMi'itioiiN with tint givnii urvn f tluni tliu point in quoHtiim in
it nnili'il ]'iitl,
il infill HI IliHl. Mi^lil. uppi'iu 1 thrtt. if for n tfivmi position of P n iiuinbor a, Jt, . .
nr / uf tin' pninl* /'' Mhnnl.l nuiiii !,u roinnidu with /', tlmn that tlio point in question
Nhimltl ivcLuii, fur a. .'1, ... nr ,/ (ns Mm nine niiy !') unitdil points: but thin is not
w.. Thin !N pnliiis HI, niHily HIKIM in tho mwt of a iinimnwiil ourvo; taking tho
t .,,iml.j,iii i.r i'.,nis...n.lrnn hi In. (& t \) a (0\ l) (t '0 ( fchun wo luw fl3a' united points
rurrrHpi.n.linx I" llio vuluiK nf I) which wilisfy tins uiiniiliion (0, l)*(9, l^'^O; if thw
cipmUitn hiw it /Uiplc i'..,,. lh\. llw point /' which lumwuiK to this valuo X of the pani
nirtir IK iv.knn'.'.l im / unitr.l pmntK. Jint HLurtin^ from tlm ouuatioii (0, !)(0', !)' 0,
if mi wriiiiiK in MIIH fijinilinn fl=\, tho ivnulting o<p.uition (X, \.) a (0' t iy0 has a
root (;, X. il. K.II..WH ilu.l tin. tHiiutinu (0, \Y(0, 1)"' hn a root = \, and that
Ih,. p,.ml whirh IKIIIIIKH lit U*i. viilmi tfX w a unitod point; if on writing m tho
iini. a. Ih rrK.ill.iHK ^jmition (X, 1) <fl'. !.)' ImH u j tuple root fl' = X, ^
HI./. /* Lhl, HH ..piation (fl, l)(tf, :i)"'0 hoH a jtuplu root 0**K> nor oon
itiiil. ihi ]iuiiiL nnnwrnng to flX in anywiHO rockoiiB us j uiutod pomta.
26G SECOND MEMOIR ON THE [107
gives a point which reckons as j united points. But if 6= A, gives fchu jfol<l rnul
^ = \, this shows that the line 6K has with the curve j intoraoetioiiH at fcho pninl,
tf = ^ = X; not, that the line = 0' has with the curve j intoraoctionfi tit fcho point, in
question.
08. Reverting to the notion of a united point as a point P which in fmoh tlmt mm
or more of the corresponding points P 1 come to coincide with P\ in tho oiwtj whom
P is at a node of the given curve, it is necessary to explain fclwfc tho point P nnisl
be considered as belonging to one or the other of fcho two branches through fchu nod.',
and that the point P is not to be considered as a united point unless wo hnvn tin
the same branch of the curve one or more of the corresponding points V oominL' Lu
coincide with the point P. If, to fix the ideas, k = l, that is, "if'tho cuivo H simply
pass through the point P, then if P be at a node the curve pooum through tli'c
node and has therefore at this point two intersections with tho givcm curve ; hub iJu
second intersection belongs to the other branch, and the node is "not a united point,
m order to make it so, it is necessary that the curve 8 hould at tho undo tuiicih
the branch to winch the point P is considered to belong. The thing app.a.H vniy
clearly m the case of a unicursal curve; we have here two values 01 x fl ,,
b * r
bianch of the curve; and m the equation of correspondence (6, l)*(0', :!)'(),
tfA, we have an equation (X, i).^. i)^ satisfied by ^' = V bub iirt by 0'~~X and
^^^i 1X= ! " thus , nofc " twted b ^ *' val  ^^at Jl
a node qitd node is not a united point,
V V
v i = x fo r> and the equati<m
o ,
.mited point, the cp 1; r ^ t "T "" the
'
., u
to be rej e od , he > '
along with any other spec also tiomJ " P""" 1 W "^ illchl ' lu thu "
!" and instead ^^^ "^ ^ " Sup, ,lu
 write _' + Sup p. = g/,/).
oxamples in whfah 1;ho , Mll , vo
JF ^ WitK * "^ "' "
joinng them pass e s Lugh" a I ed 1 H, ^^/T P into 8Uflh thftt the linu
of contact of the tangents througho tW ?.' t " ^ P UltS wU1 be tho P" 1 ^"
be equal to the o.ass of the c. The el' IT Ti t *" I " litod ''' "' !11
the given curve a single intersection * P h h '" h " e ftP ^ih haa with
corresponding to a given position of P are' tlT /C = L The l )ointa ?"
^ith the eurve, that is, W e have .*?? ^"T"* 8 " l ~ 1 ^^ of O.P
, and m hke manner = m _ j. E(loh of tho
407] CURVES WHICH SATISFY GIVEN CONDITIONS. 267
cusps is (specially) a united point, and counts once, whence the Supplement is =tc
Hence, writing n for the class, we have n+2(m 1) + n  27), or writing for 2_D its
value = m a 3Hi + 22S2, we have 31 = m 2  m 2S SK, which is right,
101. Investigation of the number of inflexions. Taking the point P' to be a
tangential of P (that is, an intersection of the curve by the tangent at P), the united
points are the inflexions; and the number of the united points is equal to the number
of the inflexions. The curve is the tangent at P having with the given curve two
intersections at this point; that is, &=2; P 1 is any one of the TO 2 tangentials of
P, that is, a'=w2; and P is the point of contact of any one of the 11  2 tangents
from P' to the curve, that is, a=2. Each cusp is (specially) a united point, and
counts once, whence the Supplement is =*. Hence, writing t for the number of
inflexions, we have
t(m2)(tt 2) + /e=4D;
or substituting for 2D its value expressed in the form  2m + 2 f K, we have
t = 3 3? + K,
which is right.
102. For the purpose of the next example it is necessary to present the funda
mental equation under a more general form. The curve may intersect the given
curve in a system of points P', each p times, a system of points Q 1 , each q times,
&c. in such manner that the points (P t P 1 ), the points (P, Q'\ too. are pairs of points
corresponding to each other according to distinct laws; and we shall then have tho
numbers (a, a, a 1 ), (b, /3, $'}> &c., corresponding to these pairs respectively, viz. (P, P') arc
points having an (a, a.') correspondence, and tho number of united points is = a;
(P, Q 1 ) are points having a (/3, /3') correspondence, and the number of united points
is b, and so on. The theorem then is
p ( a _ a _ ') + q (b  j3  /9') + &c. 4 Supp. = 2M),
being in fact the most general form of the theorem for the correspondence of two
points on a curve, and that which will be used in all the investigations which follow.
103. Investigation of the number of double tangents. Take P' an intersection .of
the curve with a tangent from P to the curve (or, what is the same thin# P, P'
cotangentials of any point of the curve) : the united points are here the points of
contact of the several double tangents of the curve ; or if T he the number of double
tangents, then the number of united points is =2r. The curve is the system of
the 2 tangents from P to the curve; each tangent has with tho curve a single
intersection at P, that is, & = 2; each tangent besides meets the curve in the point
of contact Q' twice, and in (m  3) points P' ; hence if (a, a, a') refer to the points
(P, Q^), and (2r, , /3') to the points (P, P'), we have
From the foregoing example the value of aaa' is .**. In the case where
342
2G8
SECOND MEMOIR ON THE
4Q7
tlie point P is at a cusp, then the n2 tangents becomo bho ,'! tungtmtH from
the cusp, and the tangent at the cusp; hence the curve tmsots tho givon urvu in
2 (  3) + 3, = 2w  3 points, that is, (>i2) + ( 1) points; this dooH noli pmvti (&,,
Xo. 90), but the fact is, that the cusp counts in the Supplement (w  1) tiniGH, unil tint
expression of the Supplement is = (n  1) /c. It is clear that wo huvu /3 = fi'~ (. ~ B)( m _ ;}),
so that the equation is
8D
that is
 2 (  2) (m  3)
or substituting for 3D its value 2*+2+ and reducing, thin is
2r = n a + SHI  10a  3,
which is right.
104 As another example, suppose that the point P a K ivon ,,, , (ih ..... ,[
* and the pomt Q on a g, curve of tho order ' havo an (,, ') uomH,,,,,,,!,^,.
Hi let ,t be reqmrecl to Hnd the ola SS of the ourvo envelope,! by h, lin, 'J '
arintar pomt 0, join 0, and let this ,neet tho cnrvo. in /"'>'
i ^
a given line is ='+'
+ "'
s 
r (
" 10 Iill0a P( 2 whioh '""
by
103 to m .
arbitrary.
g .veu Olll ,e, (3^)( 2) ancl (3Z) ,
cond, 10 , 1S , and besides have with the rive,
( the case may be) two eontaets tClst
p, 2, and then 6nally (5), (4, 1) f ,
have w,th the given eurve a oontao 'of t
d also of the fa. o^,.., flye contaot
ftm ,
" iOS whioh satis fy
"""T f th S8 OIld 1> 1 '' '
md on with the condition.
numban! of tho >"' which
contact of t ho TouHh
407] CURVES WHICH SATISFY GIVEN CONDITIONS. 269
10G. As regards the case (42) (1), taking P an arbitrary point of the given curve
7)i, and for the curve H the system of the conies (42) (T) which pass through the
givun point P and besides satisfy the four conditions, then the curve has with
tho given curve (42) (I) intersections at P, and the points P' are tho remaining
(2w 1)(42)(T) intersections: in the case of a united point (P, P'), some one of the
Hystem of conies becomes a conic (42) (1); and the number of the united points is
consequently equal to that of the conies (42) (1); we have thus the equation
((4Z) (1)  2 (2m  1) (4,2) (l)J + Supp. (4,2) (1) = (42) (1) . 2D.
107. It is in the present case easy to find cl priori the expression for the
Supplement. 1. Tho system of conies (42) contains 2(42 )(*#/) pointpairs^); each
of these, regarded as a line, meets the given curve in in points, and each of these
points is (specially) a united point (P, P'); this gives in the Supplement the term
m{2(4)(4<#/)j. 2. The number of the conies (42) which can be drawn through
a Gimp of tho given curve is =(4?); and tne CU8 P is lies pect of eaca ^ tnese
(ionics a united point; we have thus the term K (4t2>),_ and the Supplement is thus
= m{2(420~(4#/)}(42.). We have moreover (42) (1) = (42), 2D = H2m + 2 + ;
and HubBtitnting tliese values, we find
which in right.
108 It is clear that if, instead of finding as above the expression of the
Supplement, the value of (4^(1), (^) +mW), hod been taken as known, then
the equation would have let! to
Supp W (1) = m I 2 (*% ')  (*% />! + * (4 ^ ^ ;
and this, as in fact already remarked, is the course of treatment employed in the
remaining eases. It is to bo observed also that the equation may for shortness be
in tho form _ 2 (2m _
+ Supp. (I) = (1)22);
viz. tho (42) is to be understood as accompanying and forming part of e;
and the like in other cases.
109. Wo have the series of equations
(42) ((1) (1) (2m !)(!) (2* 1)1
= ( 1 ) LL)
+ Supp. (1)
. The oxp,ossicm . pointpai, b ** as equivalent to ana standing ta that of a o*u
BOO First Memoir, No, 30.
SECOND MEMOIR ON THE [407
(32) 2) (2) (2m 2) (I, 1)
4 Sl 'PP (2) = 2(2)27;;
(32) 2 ((2) ft l}(2)(2,2)}
+ {2(1, !)(!, l)(2, tt 3)ft I) (2m 3)}
+ Supp. (1,1) =(Ij 1)2/);
W f(3)(3)(2m3)(T, 2)j
+ SPP.(3) = ;i(IJ)2/);
(22) 2{(3)(2,l)(2 l l)J
+ 1(2, l)(2, 1)(2 W *4)_(T  1, 1)2)
22) 3 {(3) ft 2)(3)(23))
+ !(1. 2)(T 1 2)(2 M 4)(T l 2)(2fl l 4)J
+ Supp, (1, 2)
'
(1,
+ Supp. (l t i, i)
, ,
(2) [(4)(4)(2*4)(T l 3)J
+ Supp. (4)
= 4 (4) 27; ;
(Z) 2 {(*) (3,1) (2, 2))
+ ((3. l)(3, l)(2 B 5)(l, 1, 2)}
+ S PP (3, 1)
2 M3 , I),.
)(2,
+ Supp. (2, 2)
+ Supp. (2, l,
_
+ Supp. (i, 3) /;
(1,8)
971
407] CURVES WHICH SATISFY GflVEN CONDITIONS.
(Z) 3 {(3, 1)(1, 1, 2) (3, 1) (2m 6)}
+ 2 {2 (2, 2)(l, 1, 2) (2, 2) (2m 5)}
+ {2(1, 1, 2) (I, 1, 2)(2w6)(l, 1, 2) (2m 8)}
+ Supp. (I, 1, 2) = ^' l ' 2)2Z>:
(Z) 2 {(2, 1, 1)  (I, 1, 1, 1) 3  (2, 1, 1) (2m  6)]
+ {4(1.,l,l 1 l)(I 1 l,M)(27)(T>l,l ) l)(2 7)}
+ Supp. (1, 1,1,1) ^' L ' *' ; " *
110. I content myself with giving the expressions of only the following supplements.
Supp. (4Z)(1) =w[2(.)(/)] + <0
Supp. (3) (2) = *w. [2 ( : )  { / )] + i* (' /)
Supp. (3^) (1, 1) = ( 2 w wl ~ 3 " 3 
 2m I
Supp. (22) (3) =im[S (.'.)(:
Supp. ( 2) (5)  + 6 (2 + 20,
where a, 6 are the iopreaQnbativoa of tho coudition Z.
It may be added that we have in general
Supp. (2)(4X) =Supp, (4Z) + & Su PP' (4Z/),
where (4Z) stands for any one of the symbols (3), (3, 1) ....(I, 1, 1, 1).
111. The expression of Supp. W (I) ^ ^ ee " oicptainod ^A, No. 108 Tlmb
of Supp (82) (2) may also be explained. 1. The pointpairs of the system of comes
W ^reg^tng each "pointpair a lino, are a eat of linos eave toping a aarvo; the
ass T his eurve is equal to the number of the ifc* wlneh pass through au
SSnJ point, that is, as\t first sight would appear. *T the ^;/^^
the system (3^0, or to a<8*:)(B*./>' it is, however, neceeaary to ndm i that
number of distinct lines, and therefore the elass of the curve, IB emeha f of Ihu
lTp(8*0(/)]; Hch being H0 ,the number of the point p an,(^) wboh, ^dcd
as lines touch the given curve (of the order m and class W ) is 4[2(S^:)(3^/}].
The pi f contact of any one of these Hues with the given curve ,s (specially) a
umted point, and we have thus the term * ft pS(8Z:)(W  /)] of the Supplement
r The number of the conies (32) which toueh the given eurvo at a given oup
thereof, or, say, the conies (32) (M), is *(8*./), d the ousp is in
of these conies a united point; we have thus the remaning term 4*
Supplement
272
SECOND MEMOIR ON THE [407
Article No, 11 2 fco 135.^ Mmi fo
contact with a (jiven Curve.
^ qURti nS ' WMCh : firat '"' " >' I call their
First equation:
{(5) (6) (2m 5) (I, 4)}
+ Sl W ^ on
= o (5) W.
Second equation :
26)(4 l l)(2, 3)}
+ ((4, l)(5, l)(3m0)(l, 1, 3)}'
+ Supp. (4, J.)
^ =4(4, 1)27J,
Third equation :
3 {(6) 2 (3, 2) 2(8, 2)}
+ {(3, 2)(3 f 2) (2t G)  2 (1, 2, 2)}
+ Supp, (3, 2)
3(3, 2)21).
Fourth equation:
2 {(4, 1)2<3 J 3, l)(2, 2, 1)}
+ {(3. 1,1) (3, 1, l)(2 t  7) (1,1,1,2)}
+ Supp. (3, 1, l)
' 3(3, 1, 1)2^.
Fifth equation:
4 [(fi)& 3) (4, 1)}
+ R3, 2) (2, 3) (2m 6} (I, 1, 3)}
+ Supp, (2, 3) 
' =2(2, 3)2D.
Sixth equation:
3{(*. l)2(3, 2, l)2, (3, 1, l)}
+ 2[(3 l 2)(2. 2, l)(2, 2, 1)}
+ (2(2, 2,1) ft 2,1) (2m 7) 2 (1,1,1, 2)}
+ Supp. (2, 2, 1)
Q ^ 2(2,2,1)25.
Seventh equation:
2 {(3, 1.1) 3(^1, l, 1) 3 (2, 1,1,1)}
+ {(2, 1, 1, 1)< 2 1, 1, 1) (2m 8) 4(1, 1, 1, 1, 1)}
+ Supp, (2, 1, l, l)
* =2(2, 1, 1, 1)27).
407] CURVES WHICH SATISFY GIVEN CONDITIONS.
Eighth equation :
+ {(4, 1)  (T, 4) (2m  5)  (I, 4) (2m  5)J
+ Snpp. (1, 4) = (1. 4) 2.D.
Ninth equation :
4 {(4, !)(!, 1,3) (4, 1) (2m 6)]
+ 2 [(3, 2)(l, 1, 3) (2, 3) (2m 6)}
+ {2(3, 1, !)(!, 1, 3)(2m7)(l, 1, 3) (2m 7)}
+ Supp. (I, 1, 3) = (T, 1. B)2D.
Tenth equation ;
3 {(3, 2) 2(1, 2, 2)(3 f 2) (2m 6)}
+ {(2, 2, 1)  (T, 2, 2) (2m  7)  (T, 2, 2) (2m  7)}
+ Supp. (1, 2, 2) =(1, 2, S)8D.
Eleventh equation:
3 ((3, 1, !)(!, 1, 1, 2) ft 1, l)(2w7)]
+ 2{2(2, 2, 1)2(T, 1, 1, 2)(2 ] 2, l)(2m7)f
+ {3(2, 1, 1, 1)(T, 1, 1, 2) (2m 8) (I, 1, 1 ? 2)(2m8))
+ Supp. (1, 1., 1, 2) =(1, 1, 1, 2)2.0.
Twelfth equation :
2 {(2, 1, ], 1)4(1, 1, 1, 1, l)(2, 1, 1, l)(2m8)}
1 {5(1, 1, 1, 1, !)(!, 1, 1, 1, l)(2m0)(T, 1, 1, 1, l)(2m9)J
+ Supp. (I, 1, 1, 1, 1) = (I, 1, 1, 1, I) 2D.
113, I alter the forms of theao equations by substituting for 2Z) its value
= n 2m+2 + /c, and by writing for the expressions with (1) their values,
(1, 4) = (4) 5 (5), &c.,
and except in the terms {Supp, (6) (5)}, &c,, by writing for K its value Sn + a,
The resulting equations, if the Supplements were known, would serve to determine
the values of (6), (4, 1), &c. j but I assume instead that the last mentioned expressions
are known (First Memoir, No. 50), and use the equations to determine the Supple
ments, or, what comes to the same thing, the values of the terms in { } which
contain these Supplements, We have thus the twelve reduced equations, with resulting
values of the supplements,
C. vi. 35
274
SECOND MEMOIR ON THE
[407
114. First equation :
(5)
+ {Supp. (5)~x
(that is, we have
15m
3m
8m +
+ 10w +
4
Supp.
and so in the subsequent cases, the equation gives the vahio of bha town in
which contains the Supplement).
115, Second equation:
2(5)
, 1)
+ (Supp. (4, l)*(4, 1)}
+ (4,
(1, 3)
6m a 
 30m 30ft + a ( .1,8)
ft 3 + 104m + 104ft + a ( 6m + Oft  66)
+ 18m + 9n+ a( 3m  0)
i a  36m 30n fa (3m 8ft hi 8)
'i a 56m 53n + a ( i$m JJj;,  30).
I stop for a moment to notice a very convenient verification of fcho term in f };
putting therein a = 3H, the term is ' Jl
18m + 9n + (9m?i
," 1 "," 1 ' * 8. ^ who, any hfchor tormH outor
= m " wnl=B '= 8 ' te  ""> iuo IB ise iH
above by sirap , y ,_
term is of the proper form P S +
question is material in the seoll
of the numerical verification ' '
116. Third equation:
,_ oeffloiot .aloulablo an
,7 ^ * " = wh(m ""
C mplete reducti " '> tho form it,
' the P oint he & the mko
3(5)
(3,2)
{Supp. (3, 2)* (3, 2)}
, 2)
 45j?t
+ 120m +
+ 15m
 36m
( 4m
Verification is 15 + 3 (1 . 2  7) = Q,
{)7 riHLVKM WHICH MATISKY UIVKX CONIH'ITONH.
1 IV, Knurl ll rijimtinil ;
MM. I, 11
I Snpi. (II. I. I) *( 'I I. hi
I (M, I. h(4m I :l/i Zn)
Vni)ti'iili"M
H, Kiilli
275
 Km" (It
I i^ll' 1 w
I a I, fan 1
'itn<
A i 4 iaMHvi V 7)a..:.o.aM..a(i.4.(VH)a..as
?* i* "
,f HHJ  !(
hi. SiMh
:u4, n
, 4 W i*  HOifiH  a4 lt I !H2t h !1 1 2
! 4 240/u h 240/1
I * .torn' 4 lOHmw 4 4Mn  IliMlw  MOu
01
(1)
(
CD
(I)
(W
352
276
SECOND MEMOIR ON THE
+ 18m+18198)
 8 SB 156) + * (
 lfi  lfi + 654) + tf (
30)
6)
+ 50m + W n  276)
120. Seventh equation
= =
Verification is
;
2(3,1,1)
(2, 1, 1, 1)
{Supp. (2, 1, 1, 1) K (2 1
(i)
(2)
+
(2, 1, 3, l)(2t + jj a)
W + W*+toM
(3)

(i, i, i, i)
* fl*^^!^
{9)
(1)
<*)
(3)
<*)
w
 3m .  Kteft,  20 re  W + 109ms + 282m , t + mn , _ 86Sm _ 8flgB
6rf + 30^ + 8 (h,. + 6^  174 m .  Si8m)l _ U4tf + 1320m
 l* 5m= fa 3_ 20m ._ 5mn+ y n , + 8Qm+ 8to
wv^i 8B ^ fl . + mH 37))w+ na _ 150m _ 76a
+ 2 + l i+ Sn!ft , + K> + ilraS+ 8tol + lfare ,_ 382m _ 40g/j
(1)
(2)
(3)
W
W
f C
( . + ^, + ^ . /
^+l 2 , m+3 ^ 69m _ 69?i+582)+aa(
~ 9)
^ a
^
a "2flmn,^ w + afam + aiia B .060) + (_^
mf)i + 28)
19m + fw 54)
407] CURVES WHICH SATISFY GIVEN CONDITIONS.
121, Eighth equation :
277
= =
5(5)
+ (4, 1)
+ fSupp. (I, 4)* (I, 4)
4)
 76m 75)i + ( 45)
 8m 3  20 wn  8n? + 104w + 104n + a (6m + 6?i  66)
+ 1m 4)i + a ( 1)
10m
a ( 6m + 10)
+ on (6)
Verification is 14 + 3.1=0.
122. Ninth equation :
= =
4(4, 1)
+ 2(3, 2)
+ 2(8, 1, 1)
+ jSupp. (I, 1, 3)* (I, 1, 3)
U(m2)(2(.l ) 8) (/I, 3))
, 3)
(4 (4, l) + 2(2, 3))
3m 3
 32m 3 
 20mn a  3i s + 109m 3
32?i 3 (
109n s
16m"  SOwin
+ 3m s
8mn s
+ 24m + 24?t  264)
278 SECOND MEMOIR ON THE
123. Tenth equation:
= =
[407
3 (3, 2)
+ (2, 2, 1)
+ fSupp. (T, 2, 2)* (I, 2, 2)
U(ro2){2(.2, 2)(/2, 2))
+ 3(3, 2)
24m 3
+ 360m + 860w <i)
24n 9 468m 40Hw ca>
64m
: ( Urn  12  234) + tf (
(~ 8m 8n
(3)
W
(5} +( 20w
(B)
Verification is + 33 + 3 ( 3 . 2  13) + 9 . 2 = 0.
124, Eleventh equation
3(3, 1,1)
4(2, 2, 1)
3(2,1,1,1)
= =
*)( 1,1,2)
+ (3(5, 1.1) + 2 (2,2,!))
 ff*'
(1)
(.5)
(a)
407]
CD
<*)
CURVES WHICH SATISFY GIVEN CONDITIONS
+ 848m+ ^tpw 3  1302m ~1302 + a(
96m 3 + 21Gmn + 9Gn a  1872w1872n + (
1044m  522n a + 3960m + 3960?i + a
,[3 2?)l
19Gm a 
56ft fl  392m 392i + a  ?"  2m  mn"
(7)
279
(8)
(3)
(0
(6)
(S)
(7)
(1)
(2)
(3)
w
w
to)
(7)
 ^m ^w+ 873)! a 2 (  ^)
 32m  32 + 1308) + a' J ( 2m + 2?t  48)
+ 358m + 368n 2880) + a"(fm$n+ 84)
Vorifioation is 4 (  ft  6  3) + 2 (V  ^ + 18)  2  191
125. Twelfth equation:
2(2, 1, 1,1)
+ 6'(1. I. 1. l > a )
+ f Supp. (1,1. 1,1, !)*(!.
= =
+ )(>!, 1,
2)i(2, 1, 1, 1)
(3)
(6)
(Q)
280
(i)
HTCflOND MKMOMl ON T1IK
j 12 W i !l .
V"
(fl) + MwM tym'H I
111 H'
:l/Sm u  fi2wm
I "]()// .( a (
n
+ 8 (f A I i + 1. I A) H h ( ii  .1  ;j) 4 I ( H  (jf.) 2  M) , II (( !]  , j) 4 ,.  ;t . a J . (( .
12fi. Ib will b nbfioiT(l thai, in bhn ni^lali HII.I lulhmiuK itUit.i.H ( vi/. tln
rojii tho expression of Uio Supplmn.mL tumtiuuH Mm Hymlml (1). I ] m v inHiitli..
ilong wilh tlio Hiipplomunt within l;hu ( ], tl,,, tonns (i!) a(.4)t/+)I *r v
too ara (mU) int., n.uuhnv ,,f p,,!,,!^ (4), to,.: tl,i 'w f.,r n.un, ,(',,, )m  v
t Hiinplifioa tho calculation, l.th fnu thn Hyuiuuarical form inul,,' wliit.h l,h, .
ion pi^onfc bljomHulvcH in thu H.v.val .quulionH, ,! !.,,,. t  1M ( ,
Kmni. in .juoHUoi,. (te, tormH bcnug imai. ,n,iUi^ w f R miwbor ()
y Auth^ fchoory known in torn w O f tho CJapi,:,^. It in u. 1 ....ticn) Mn (, r r V
tion, to U Uie nyntom to wldoh tlio (Japitul^ l,ul<mg, w<, ,h,,i,,M. h v
zt r ; m ; 1 ^11 T,^/* ^ u. "
upp. (2, 1, I, 1), bho Capitals belong to tho Ryatom (1, 1, i. i).
407]
CURVES WHICH SATISFY GIVEN CONDITIONS.
281
CO
(2)
(4)
0)
127. Referring to Nos. 41 to 47 of the First Memoir, for convenience I collect the
capitate whieh belong to a single curve, giving the values in terms of m, n> a as follows.
(1, 1, 1, 1)
'11  8 (n  4) (m  4)  %m a n
8)i a + m
+ (fm 2 + fm 6 + f)
 2m 3  fm tt ji + 4mu 2 + 10m 2  14m  16?i 2  8m
+ a(
8 + 2m s  ^m^  fmn. 9
+ a(i 2 +*m ' In);
 18m
m + 6).
(2, 1, 1)
(3)
(8)
(2)
(1)
(2)
(5)
ff =
[= t . i (i  II) (w  4) supra] ;
J =
+ a(
+ (
+ 2m 
n+ 6);
+ a( 3m +9);
mw" + 8m 3  m4n 9 32m+
+ a (  3m + 12) ;
mu
 3m  4?i + 12) ;
3).
(3)
(2)
CO
(2,2)
L =(n8)
Jbfi(*l)
JV*
~K
0. VI.
_ 8n a
4m^ + a( I);
i);
i).
+ (
36
282
(2)
(2)
(5)
(4)
(3, 1)
P =
Q
[=
(4)
[= tcsuprti],
SECOND MEMOIR ON THE
+ ( 3);
+ (  3);
+ a (m  3).
3m+ct (first equation).
a(2mG)
+ On
( 8 econd equation).
6JiT=
9)
[407
2)
15m
ff(?i7) (third equation).
(fourth equal
407]
CURVES WHICH SATISFY GIVEN CONDITIONS.
283
7i s 4 8m n  3 (fifth equation).
= 3?>m 2 4 24??i 2  3mn  12n a  96m 4 12?t + a (
 9m + SG)
4 / 3m7i 2 4 9?/m 4 12 3 36?i 4 (fli
44J" =  12m 3 + 36m +a(
i. gjp Qti~ 4 18ft + o. (
.w3m4w4 lii)
4m 12)
2?i 6)
12??i 2 4 Ginn G?(' J 60 i Qn + H (?j.
in  8m 27i + 30)
(sixth
equation).
7^ ^mft  2m 3  flm* + 4w' + 10'  1'
to16 2  8m +C
ifcft (U
.[. 4^' _ yjftp + g??!, 3 ??t a ft  9'wt a 72wi 3 4 !
)m 4 20?i s 4 160m S
!0 (2)
+ 4D  Gi ;i + 42i a
72m
(31
ioi"
1 fill ( 1)
Tl' 9?"
4 X/ ~ ^
... " a ""
(i)
(8)
7i s 20 i 2 i
4 Gm46)t~24)
4 27m  60)
 14m 4 24)
n+ 8)
? 4 19m 4 An 54) (seventh equation).
2P
+ J'
a (eighth equation).
(2, 3) = 4m4?i 6
1.2(5, 1)= 2m 427?. 12
 2?ft  2rt  18 + 3a (used
= 2*71" +2w16n+(
^
(ninth equation),
362
284
3Z
+ m
~w
 o
*(2, 2, 1) =
2F
 ff
22)'
J"'
0)
(2)
(it
(5)
(3)
(7)
(8)
W
(10)
WJ
SECOND MEMOIR ON THE
( 3a 9)
a(6n l) + s
 2)
 1)
+ S3B + (S13) + a (tenth equation).
[407
4" fl
3n + a ,
80
+ m 2 f m)l+18)ia _ SttI91n
(i)
(a)
(3)
(i)
(5)
(e)
(7)
(a)
(8)
(10)
(ii)
(m + ftjy.)
(eleventh equation).
407]
I.l, 1)=
CURVES WHICH SATISFY GIVEN CONDITIONS.
285
(0
2D
&
is) + 36m + (
+ 18 + a(
10m 10ft + 40)
+ 7m  12)
n a + S 6)
4 Hi 2  mw + fw 3   m ~
129. We have consequently, by means of the results just obtained,
Supp. (5) = (5)
(twelfth equation)'
+ N
Supp. (4, 1) = K ft 1)
42J + J2
Supp. (3, 2) = (3, 2)
+ 67iT + i
Supp. (3, 1, 1) = *(3, 1, 1)
+ }} + J5 +
Supp, (2, 3) = (2, 8)
+ Q
Supp. (2, 2, 1) = (I, 2, 1)
+ 3& + I +
Supp. (2, 1, 1, 1) *(2, 1, 1, 1)
3 J" 4 J*
(first equation).
(second equation),
(thh'd equation).
(fourth cauation),
(fifth
(sixtl
(sevi
SECOND MEMOIR ON THE r , Q
(I, 4) = , I>4 _
 i + 0 ( eighth equatioil)
(I, 1,3)  *(1.1.8) + 2.8 *a
, . .
, (ninth equation).
(I, 2, 2) = 2 2 )
 /, ,,
/T , (tenth equation).
. (1,1,1,2)= .(1,1,1,2) + , (2, 2,1)
BJ)
+wyw ,
Observe that. (eleventh equation)
ffjup.o, 3^/ +8 ^^ +7/+s//
to o^ the fom of te _
(twelfth equation).
and therefore a l so ^ 3) = (W ' 3) + ^ 3)  (SI, 3)J,
* 3) = K (aZ, 3) + * [(2, 3)  (2S ( 3)J,
the second term IY2 3)(Is\ q^7 u
The remark app.ies to all the ^veTn.^^ te S f Zeuth ' s Capitals.
tan 8 much as (M) = (JS 1 ^ T^' nly as re S arda *> t four of them,
which are th us expressible by means of M!l ^ rigi ' W ' te S K " ( 5 ' L ^
fornn,!. (First nJoi,. No,. /9 HT? 8 ' '* ^ the  si *' of the
= A: = Deferring to
(first equation).
(second equation).
407]
K (3, 2)
(3, 1,
CURVES WHICH SATISFY GIVEN CONDITIONS.
= K ( 9 1 ) = (3 (n 3) + K  1 + 1)
= 3i I 2M +
287
to
= K (w a + 2mw + i?i a  J .f m  tyn + 27  fa)
~H\ 21 + D 1 + J'
\ix. K~ I H =^i 3  iw +4
K 1 . 27 = 2mn  6m  8 + 24
(third equation),
(fourth equation).
(2, 3) (2JS, 3)
*(2, 2, 1) (2ia, 2,
, 2,
3)
/cTi i i n j 71'
= K {&K\.) L, Jt *) "r Jy
4)
 (1^4)40
(2, S) + 2(5, 1)^
* (1*1,1,3) + * (2^,3) +*(3)
, 3)
, 1, S) + S*<a,
(1 2, 2) * (F*!, 2, 2) + * (3 (n  3) +  1)
(  ' ' '
+ (2*1, 2, !) + *( 3)
1, 1, 2) + 2* (2*1, 1, 2)
(fifth equation).
( sixth eq ation )
(seventh equation).
(eighth equation).
(ninth equation).
(tenth equation).
,. ,
equatzon).
goo
SECOND MEMOIR ON THE r ^
"(1, 1, 1 1, I) + *(2 f 1, 1, 1)
' lleferrmg to
181. Henoc, substituting in the ex p, eS siou S of the sevo.al Supplements, we h ave
Supp. (5) = Q
+ N
(nrst equation).
Supp. (4, 1) 
(second equation).
. (3, 2) = 8i + 2 jtf +0
/4.T ,
(third equation),
f f a
(fourth equation).
(fifth equation),
,. ,
(sixth equation),
. (5, 1, 1, !) = , fe lflfl) + Jy
+ ^ + 4(7 + 4J? + JX
(seventh equation).
Supp. (1,4) . {1S I 4) +
+ (mf)(4J\T4.0)
"^ , . ]pl
(eighth equation).
 (2, 3) = ^2^ 3)
+ Q
. 2,1) =
2P 0
(ninth equation).
(tenth
107 I (MIHVIW WHICH HATIHl'Y (MVMN CONDITIONH. 289
>. (I, I, I, 1!) " (!*!, I, I, 2)l2/c(yi, 1, 2)
i (M ... y,)(;tA; r ;t7<'.. (itfia/H .//i a/i r ,/)
 aw.. a/ 1  <;  ,\./H y/'i ft/ jf >> 27/ ^
(nlovtmlih oquatinn).
Hupp, (I, I, I, I, I) *(Ul t I, i, I, I)..SK(5J1, ], 1, 1)./)'
Hm4j)(vl..'J/H.4r;.1"!U))
..jUlj7fjU~2/>7>'.
(twolfbh equation).
Ilia, limn* Ihmlly, nii'ivly rnllirtiiiK lhn tnrnm, wo hivvo liho following oxprcHBionfi
til' (iln Sllpplt'liM'llhl ill til'' IWi'lVn Pipmtinim I't'SJll't'.l.iviily,
Stipp, t,M ' A f i " ('"'Mb uqiialiiou),
Hupp. (4, I)  '.(./ I a/M.r (widow! (!(]iuition) (
, (II. H) it A'  4A^/.MiV..:iO (Uiil (Hiuation).
, Cl ( L U ' " I /'' f /' t'^' I '//laM:Ul^>'l2./' O'nui'lilt oquation).
.Sup)., tl HI ArJ*l. 'I) M<> C Iiftl1 <I"".on).
, 'A '*;, I)  *(;!*'!. 1!. IH:UH'M'4./>!U' (Hixtli oipmbion).
, ttf. 1, 1, Ct ..... tv!vl, 1, I, l)'h/M"KM'4/H<2/J' (mwrnth o(imtion).
i. ll. +)  *tUI. 4) I (4m.7)A'4.(^ 1)0 (^^ liquation).
, {I, I. M)  MU'I, I, :J)M!tal, rt)
t U!'" (ij/ I 'Kawroyl'(">w10)./H'('tirt
I (Sim ^.S)A'.Htm./,t(mMJ)J/.( t 2//^0)A r l('m!)}0 (Umfcli o
Hupp (I, 1 ( t. S)  *U*, 1. I. UH>'M^'< 1. 8)
i , i a,,, .. Ti) /I I (Hi*  ) A' I 1'*''* ^ !I ) ^'t' (' ;t M lfl > f/
IH ( m  1 ) // < t  (!i  1 ) t H ( w  1 r ') '' 'I' !1// (ulovonth
Hupp, o, i, i, i, n *(i
), (twolfth oquation).
m 1 rocnll Urn ^imirk. .iHto. Nu. ISO. thai lu uuth ^imUon tho Capital bolong
Iho .plorn ubtaiiHxl by .HmmwhiiiK Ui bnitinl number by un.ty and ro.nov.ng iho
; (4) fur Uw llwi wiuuiimi. (. 1} for Urn cil, > "
a vi.
290
SECOND MEMOIR ON THE
forms of the
hibit how the
Supplement arises, whether from prope? , 6XI 3 , act
or from pointpairs (comcMent];"^ ,)"];?" r'T^' " tOUOUng a < " CU3 P>
terns Imepairpoints). Thua fo instarl , "" noludm g <> f in those
pl, n ati n s ,K rat P Mem^C o 4^rT ( ^ + ' **"* '
described a, "inflexion tangent teLinaI l^
number of the Knepairpcinl del ib d a "oli,
< P ," or in what is here the a onl ! T ^^ te ilMt ^ each
Hnepair each inUoxion tangent anTeaeh ^T/ W ' hare as o
of the firat equation, ,vhen g the ^^ point 1= T '7*' ReVerti " t0 the 8
the curve e is the conic (6), hay:n e with * P ? genm ' al f the ^
Beting it in the 2m _/; it a ' 1 r VG * lnteraect; ^t P, and
 8 becomea the coincid nt 1^paT 1 r ^ " " '" M '
number of intersections at P is therefore f 1 > taDgent talren twioe .
united point. Similarly, when the ' t ~ ' \ ^ m& xim is tllerefora (pe
oomcident linepa u . formed by the tanC L n ? * T' "* U1Te 9 beco  s
P therefore =6, and the ousp i Z s f, ^ T^ * 6 mnlbar f ^eotions at
tota number of special united foin s  K if T"^ & Um f ed " int: ^ thus the
result, Sivpp. (g).y + 0. * ~ + ' : "P 8 * Wlth th ^ Agoing A posteriori
13*. Or to take another example; f or the fifth equation we have
Supp. (2, 3) = (21, B ) + Q.
" " d "* ^nt terminated
puipoae, ,ve have each double talent as a .1 ! IT a WPto for the present
of its points of contact, and also a a coincident " ^ " "^ * ^ e
* pomts of colltact , Eeve the Ze on T;r m reSpe0t ' the other of
IB a pomt in general on the given curve th!T n T^"' when the P int ^
toudnng the curve at P, and "hav ng Teddet \ h'i, " ""f Sy8tem f W ^ ^
S1 f a ^ eaoh =nic the number of inte L f p" . nte0t f thfl thM "toi
rp ras s r p is t r 2(5  3)   th  ; (2i; s )ii the t tow mber ^
pomfB P. Suppose that the point P ; talrm f f if. 2 (2 ' 3 > "iterseotions are the
angent; of the (2, 3) conica, 1 (I ^1" ^, t f 6 ,, p0lM f oontaot of a double
formed by the _double tangent taken LTe and .iv?, HT ^ C in ident ""P*
the remammg (2,3)l conics are pl . OB er '!!,?  , thel ' efore * ^'ersections at P,
at P, or the total number cf inteZt loT f^^' ^ ""ersections
s a gam of 2 intersections. As remarked Wo 96 t l ' ] + 2 intersections; or there
that the pomt in question is to be consWeS' TV J^' f **>** *&
I do not know how to decide d pnori whlh r i^. H (8Pe ia%) 2 mited P '^
points or as 1 united point, but it is &?* b V, e arfed being 2 united
1 umted pomt; and as the points in lesL a + Tf ed M ^ ( 8 P edal W *f
double tangents, we have thus the numbef T O f t , ^ P intS f onta t &
the point P is at a cusp, all the <f T? P UEited P<)ints  A ? ain ' whe "
(2, 3) comes remain proper conics &,,)& 3)
OQ1
407] CURVES WHIOH SATISFY SIVEN CONDITIONS.
First Memoir, No. 73), but each of these fend conic touching the cuspidal *
with the given carve at the cusp not 2 but 3 intersects, so that the to al
of intersection at P is 8 (Si. 3), =3(2, 3), and there is a gam of (2, 3)(2l, <)
intersections. Kach cusp counts (specially) as (tol, 3) united pomts and togothei the
cusps count as .(Si. 3) united points; we have thus the total number K (2*1, 8) +
of special united points, agreeing with the expression, Supp. (2, 3) = *(!, V + V
135 As appears from the preceding example, or generally from the remark, ante,
No. f I tve not at present an^ d ^ method of ^determining the proper _ca 1
multipliers of the Capitals contained in the expresses of he severa Supplemcn s
these conies becomes a coincident linepair ; this regarded
the given curve () ordinary interseckons ( a number, . 4 at most
the contacts which the line may have with the curve) ; for each th P
taken as a position of P, one of the conice which make up the curve becomes
Sd^t Unfair, and kcre are in respect of this conic two cUo at
instead of one intersection only. We have thus m respect of .the paiUcu a>
"
ttvrs ,rr
thi, bu" being unable to <&.. the results th any degree of com
pleteness, I abstain from a further discussion of them.
292
[408
408.
ADDITION TO MEMOIR ON THE EESULTANT OF A SYSTEM
OF TWO EQUATIONS.
f *? f y<d *** f m ' TO ' 
, pp. 173180. Received August 6,Kead November 21, 1867.]
mtkout
tables fo, options in he fo
is the standard form m nuanZ Thl V
I rt 1
"'
'
oaloulried.
Hssultant of
ooeffic: te . ^
r
meam whereof the present tables were
Table (3, 2).
Eesultont of
( 6 _ ^
( P , , 4 ;
408] ADDITION TO MEMOIR ON THE RESULTANT 03? A SYSTEM &0. 293
Table (4>, 2).
Resultant of
(ft, 6, c, d, e^oi,
Table (3, 3)*.
Resultant of
(a, 6, o, (Zftflj, 2/) 3 ,
(p, g, r, ajflj, y) 3 .
* N.B. la the corresponding
table o the memoir, there is an
error in the signs of the last
two terms ; they should be
294
ADDITION TO MEMOIR ON THE
Table (4, 3).
Resultant of
[408
408]
RESULTANT OP A SYSTEM OP TWO EQUATIONS,
295
Table (4, 4).
Kesultanfc of
(a, b, c, d, e$>, y)\
(p, q, r, s, tjaf, y)'.
296 ADDITION TO UKUOtB ON THE IlESULTANT OP A SYSTEM, & C . [408
Table (4, 4) continued:
408]
297
n. vi.
38
298
ADDITION TO MEMOIR ON THE
Table (4, 4) concluded ;
[408
408]
RESULTANT 01? A SYSTEM OF TWO EQUATIONS.
299
300
409.
January 9, 1868.]
[IT is remarked, Proc. B, Soc. vol xvrr n <m H, t ^ ,
I had in foot i n regard to the 1 ' P .' 31 *' that the ab title is a misnomer:
to the ro f' ed "* th twofoltl relati <
^ t ?f 8efeld ^
"" W0uld ha
" On the
of ec l ual ts of a binary quartio or
f
of
systems of equalities
"P> of the
.
re e 1 uati < )na
kno !
" 7 =i ^
""" the1 ' f the two
" llrtioM
U " less 
to the rootsystems
the rootle
conditions for the existence O f
quintic,"]
In oomideiing the conditions for the
between the roots of an equation > *
imposition of relations A "ret o'n
^0, or it i, say Wol d> o^tc b"
as regards onefold relations, the theory of
F=0 i, a Ration co~d e d Jf
relation satisfied if, and notTttfi er , , , la
relations is satisfied. The 1 ke not ot f "' ""
vk, the eompound relation I a eTation ,
the other of the two eomponeut a n s t t T
fro,n any further di sous8 ion P of h e tt orv of r ' PU T rP Sdy "^ at P resenfc
for the existence of given system, rf 1 posito. I say that the conditions
furnish instances of and, ompo Z in fT f' betWe n the r 0ts of an
and its first.deri.ed functio? n , ' J SY T?" ^ "" ^^
&stderiyed functions in regard tc x ' * ^ Same tUn & that
Of course,
relrtion
it ia a
,
as
409]
ON THE CONDITIONS FOR THE EXISTENCE &C.
301
unless, there is satisfied either the relation for the existence of three equal roots, or
else the relation for the existence of two pairs of equal roots; or the relation for
the existence of the quadric factor is compounded of the lastmentioned two relations.
The relation for the quadric factor, for any value whatever of n, is at once seen to
he expressible by means of an oblong matrix, giving rise to a aeries of determinants
which are each to be put = ; the relation for three equal roots and that for two
pairs of equal roots, in the particular cases n = 4 and n = 6, are given in my " Memoir
on the Conditions for the existence of given Systems of Equalities between the roots of
an Equation," Phil. Trcms. vol. OXLVII. (1857), pp. 727731, [150] ; and I propose in the
present Memoir to exhibit, for the cases in question ?i=4 and n = 5, the connexion
between the compound relation for the quadric factor with the component relations
for the three equal roots and for the two pairs of equal roots respectively.
Article Nos. 1 to 8, the Quartic.
1, For the quartic function
(a, &, c, d, eJiB, y}\
the condition for three equal roots, or, say, for a root system 31, is that the quadrin
variant and the cubinvariant each of them vanish, viz, we must have
= ace ad* 
 c 3 = 0.
2, The condition for two pairs of equal roots, or for a root system 22, is that
the cubico variant vanishes identically, viz. representing this by
we must have
(A, B, 60, 10D, BE,
A = a?d 
+ 26
 Oao 8
0= abe  Sacd + 2b 2 d
E =  ade+
F =  ae*  Zlde + 9o 9 e
(? =  6e" + Bode  2d 3
= 0,
C6 a o = 0,
==0,
=0,
= 0,
Qcd* = 0,
= 0.
3. But the condition for the common quadric factor is
0,
o 36,
So, d
6, So,
3d, e
ttt
36, So,
d
&,
3c, 3d,
e
and the determinants formed out of this matrix must therefore vanish for (/, J) = 0,
and also for (A, B, 0, D, E, F, (?) = 0, that is, the determinants in question must be
syzygetically related to the functions (/, /), and also to the functions (A, B, 0, D, E, F, ff).
302 ON THE CONDITIONS FOR THE KXI8TJ3NOK
4. The values of the determinants are
1234 = 3 x
1235 = 3 x
1245 =
13453 x
2345^3
3 <v 4 1
oVe  1
a e a  1
fl&e 9  1
aW  3
affe  1
nice + 4
firf a + 1
2f+9
orfa H *!
a(l j (i ~ \
alcd + 14
rrc 3  9
if I 3 3
flcrf 3  9
Wo i 1
u(J tf " O
bet fa i 14
&d  8
4V + 6
b^cd + 2
fid 3 + 8
/l/i/yl i O
iVi/tt "I" ^3
t\ t\ . ')
oV J i ti
5. The sjrzygetic relation with (I, /) i s g i von b y means of bl
. 36 , 3c , d
or, as this may be written,
"here //T u the Hessian Qf
6  That is, we have
o mit ft, firet
409]
O3? EQUAL ROOTS OE A BINARY QUAE.TIO OR QUINTIC.
to each other they will also be = ,= The equations may then he written
ao  6 a , . ad  bo, ae + 26cZ  3c 5 , be  cd, ced* = Q,
a , 26 , 6c , %d , e
and the ten equations of this system reduce themselves (as it is very easy to show) to
the seven equations
f A T? /** n T? ii 1 n\ n
LA, Jj, u ; JJ, Mi, J! , Cr) = U,
which, as above mentioned, are the conditions for the root system 22.
8. It may be added that we have
A Jl C D E F G
.1234 =
c
~4&
+ 3a
.1235 =
c
SI)
+ a
=
d
3o
+ a
1245 =
~ e
i 'id
3 C
=
 o
+ Gc

=
 d
+ 3c
 b
. 1345 =
~ e
+ 3d
~ c
=
 e
+ 3c
 i
.2345 =
3e
4 4d
C
where it is to be noticed that the four equations having the lefthand side ~ 0, give
B : G : D : E ;, F proportional to the determinants of the matrix
d, 3c, . , a
e, . > 60, . , a
d t 3c, 1)
~e , , , + So,  b
the determinants in question contain each the factor o, and omitting this factor, the
system shows that , 0, D, E, F are proportional to their before mentioned actual
values,
Article Nos, 9 to 16, the Quintio.
9, For the quintio function
(a, b, G, d, 0,/$<B, 2/)Y
the condition of a root system 41 is that the covariant, [#=] No. 14, shall vanish,
A = 2 (ae  4>bd + 3c a ) = 0,
B= afSle f 2cc2 =0,
viz.. we must have
10. The condition of a root system 32 is that the following covariant, viz.
, = ] 3 (No, 18) a (No, 14)  26 (No. 15) a ,
304
409] ON THE CONDITIONS FOR TEE EXISTENCE &O. 305
12. The conditions for the common [cubic] factor are
a, 4&, Co, 4d, e =0,
a, 4&, Gc, 4dl, e
b, 4o, Qd, 4e, /
b, 4c, Qd, 4<e, f
the several determinants whereof are given in Table No. 27 of my "Third Memoir on
Qualities," Philosophical Transactions, vol. OXLVI. (1866), pp, 627 647, [144].
13, These determinants must therefore vanish, for (A, B, (7) = 0, and also for
(SI, S3, ... S, 9W) = 0, that is, they must he syzygetically connected with (A, B t G}, and
also with (9f, S3, ... 8, 9ft). The relation to (A, B } 0) is in fact given in the Table
appended to Table No. 27, viz. this is
(? x +J9x 4 A x
1234 =
H Ga 3
 12 oft
4 16 ac  10 i a
1236 =
+ 6 ab
 2 ao  10 6 a
4 6 flrf
1236 =
~ 2 ao I 8 6 a
4 G ad  18 6e
 2 rf/* 4 8 e a
1245 
4 18 ac
 6 ad  30 be
4 8 e 4 10 &2
1246 =
+ 126c
1 4 as  4 bd  24 c 3
+ 4 ie 4 8 erf
1345 =
1 24 ad
 8 aa  40 bd
i 4 a/4 20 60
1256 =
 1 00 + 4 bd + 3 o 3
4 1 a/ 1 4 6 6*3  18 cd
_ 1 bf + 4 ce 4 3 rf 3
2346 =
+ 20 ae i 40 bd  30 c 3
 80 be 4 20 erf
4 20 bf 4 40 ce 30 rf 3
1346 =
1 4 rtfi I 8 fo* + 6 o 2
 36 cd
4 46/4 8 CO + Grf 3
2346 =
1 4 rt/n 20 fto
~ 8 &/  4 oe
4 24 o/
1306 =
H 4 60 4 8 erf
4 4 bf  4 cc  24 t a
4 12 do
2356
4 8 bf \ 10 flc
 G </  30 rfe
+ l& df
1456 =
4 G ao
4 G of  18 cfo
 2 df+ 8 e 3
245G =
+ G of
 2 rf/  10 e a
4 6,/
8456 =
+ lGdflQ G *
i2*y
4 G/ 2
14, Between the expressions 91, 35, &c., and 1234, 1235, &C., there exist relations
the form of which is indicated by the following Table :
306
991l
9QSS
99fit
9181
QffiTJ
9*Sl)
SWll
982T
to
&
ON THE CONDITIONS FOR THE EXISTENCE
"a
13 X
X
Q MS
13
s
o 13
V.
o
("1
M) 1*1
^ a S5f
tn ^ ""
O v_> %.s
+ '
S*
o ^3 
409] 01? EQUAL ROOTS OF A BINARY QtJARTIC OR QTJINTIO. 307
15. Assuming the existence of these relations, we have for the determination of
the numerical coefficients in each relation a set of linear equations, which are shown
by the following Tables, viz. referring to the Table headed c9l, b, a(S, a . 1234, [first of
the seven tables infrcl] if the multipliers of the several terras respectively be A, B, G, X,
then the Table denotes the system of linear equations
A + 3.B + 330 4 0^ = 0,
3 A +0B 1020 16^ = 0,
fee.,
that is, nine equations to bo satisfied by the ratios of the coefficients A, S, G t X s
and which are in fact satisfied by the values at the foot of the 'Table, viz.
A : B : G : X = + 66 : 11 : +1 : + 6.
There would be in all fourteen Tables, but as those for the second seven would
be at once dcdncible by symmetry from the first seven, I have only written down the
seven Tables; the solutions for the first and second Tables were obtained without
difficulty, but that for the third Table was so laborious to calculate, and contains such
extraordinarily high numbers, that I did not proceed with the calculation, and it is
accordingly only the first, second, and third Tables which have at the foot of them
respectively the solutions of the linear equations.
16. The results given by these three Tables are, of course.,
G6c8t 11 #8+ !< + 6(1,1234 = 0,
330 cm + 110 o  65 &< + 9 a5>  105 a . 1235 = 0,
+ 260478676 eSt
 617359490 $3
+ 144200810 cS
+ 9666911 6JD
+ 9090786 a
 721004050 o . 1234
+ 90914176 &.1235
160758675 a, 1245
+ 11559296 a. 1236 = 0.
It is to be noticed that the nine coefficients of this last equation were obtained
from, and that they actually satisfy, a system of fourteen linear equations; so that the
correctness of the result is hereby verified.
392
308 ON THE CONDITIONS FOR THE EXISTENCE
17. The seven Tables are
First Table.
. 1234
a'bf
+ 3
+ 33
a?ce
+ 3
 102
J
aW
216
+ 3
abe*
+ 21
+ 135
+ 1
abcd
12
 144
+ 120
V
16
+ 480
+ 9
ab'd
+ 30
150
+ 8
atfc
+ 50
+ 240
300
 6
b 4 c
25
150
06
 11
+ 1
Second Table.
+ G
ffl. 123ft
6. 1234
+ 10
 390
+ 156
100
 600
+ 1600
+ 126
1000
Third Table,
+ 33
102
216
+ 135
+ 120
+ 480
+2664785761617359490^:
65)
~_
 90
 195
+ 10
+ 360
 390
1500
+ 900
+ 1800
+ 155
.+ 225
+ 100
 600
+ 1600
1500
+ 125
1000
+9656911
+ 9090785
fl.1234
 16
+ 36
+ 16
162
+ 96
+ 80
 60
. . ~.
721004050 +
6.1235
.184ff
l . 1*3

=* flj
+ 10
 4
i G
. yj
+ 24
20
, ',;
00
t Iri
+ 00
+ 4
,, i ft
84
+ 90
 10
24
80
+ 64
+ 60
40
+90914175
1G07B8G76
4119S9:
409]
OF EQUAL K.OOTS 03? A BINARY QUARTIC OR QUINTIC.
309
Fourth Table,
ft a5 tZ.1234 o. 1235 ft. 1236 6.1245 a, 1246 a. 1345
v
+ 3
i 3
 114
+ 4
a z ldf
12
+ 33
 90
 264
+ 6
 6
 4
 24
W
I 21
 195
 990
+ 16
 4
+ 64
V/
10
+ 10
+ 468
_ 4
24
+ 24
a a orfe
 144
 102
 390
+ 1320
 16
+ 24
+ 24
208
# 2 c 3
216
+ 1080
+ 36
+ 144
aftv
+ 60
+ 155
+ 360
+ 900
i 4
22
+ 6
+ 24
rt&Ve
+ 30
+ 135
1500
 2700
+ 16
 6
26
20
 40
abcPe
+ 240
+ 100
i 900
+ 900
_84
+ 16
96
+ GO
alotP
+ 120
 600
+ 1800
 600
152
24
+ 96
 40
ao s d
i 480
+ 1600
+ 96
+ 64
Wf
24
+ 225
+ 16
b a ce
 150
+ 125
+ 60
10
+ 90
&W
 150
1500
+ 80
80
&W
300
1000
 60
40
/JO
Fifth Table.
a c.1234 (2.123B c.1236 0,1245 i. 1246 5.1346 a. 1256 a, 2346 a. 1346
+ 3
 19
+ 1
+ 21
+ 33
 114
 608
+ 4
 2
+ 16
 144
+ 10
 90
+ 537
 4
+ 6
 6
16
+ 20
36
102
 195
 245
 16
+ 16
+ 16
 80
16
 216
~ 390
+ 1740
+ 36
+ 24
+ 16
+ 60
+ 36
+ 30
+ 106
 264
 245
+ 4
4
 24
16
 80
 16
+ 135
 990
1700
+ 16
 4
+ 64
+ 240
1 240
+ 360
+ 468
+ 1740
22
+ 6
24
+ 24
+ 60
j 36
+ 120
+ 100
 1500
+ 1320
2000
162
84
 6
26
+ 24
 208
 860
 20
 600
+ 1080
+ 600
24
+ 144
+ 9GO
+ 480
+ 900
+ 600
+ 96
+ 16
96
+ 960
+ 1600
+ 1800
 400
+ 64
+ 96
320
 150
+ 225
+ 900
+ 16
+ 24
~ 150
+ 125
2700
+ 80
+ 60
20
 40
300
i 900
 60
10
+ 90
+ 60
 1000
1500
 600
40
80
 40
cfibcf
tflcdf
ttbcde
abd*
<tc s e
310
ON THE CONDITIONS 3TOR THE EXISTENCE
h*22&&
409] OF EQUAL BOOTS 03T A BINARY QUARTIO OB. QUINTIO. 311
And the remaining seven Tables might of course be deduced from these by writing
(/, e, d, G, b, a) instead of (a, &, c, d, e, /), and making the corresponding alterations
in the top line of each Table.
18. The equations 91 = 0, 95 = 0,...., 9)1 = consequently establish between the
fifteen functions 1234, 1235,... 3156 a system of fourteen equations, viz. the first and
last three of these are
1234 = 0,
1235 = 0,
160758675.1245
+ 11559295.1236 = 0,
+ 11659295.1456
160758675.2356 = 0,
2456 = 0,
3456 = 0.
To complete the proof that in virtue of the equations ?l = 0, S3 0, ,., 931 = all
the fifteen functions 1234, 1235, ,,. 3456 vanish, it is necessary to make use of the
identical relations subsisting between these quantities 1234, &c. ; thus we have
a, . 1345 + 4& . 1245 + 6c . 1235 + 4d , 1234 = 0,
1 . 1345 + 4c . 1245 4 Gd, 1235 + 4e . 1234 = 0,
which, in virtue of the above equations 1234 and 1235 = 0, become
a. 1345 + 46. 1245 = 0,
I, 1345 + 4o.l245 = 0,
giving (unless indeed ao ~ & a = 0) 1245 == 0, 1346 = ; the equation 1245  then
reduces the third of the above equations to 1236 = 0, and so on until it is shown
that the fifteen quantities all vanish.
410.
A THIED MEMOIR ON SKEW SURFACES, OTHERWISE SCROLLS
[From , the Philo^Moal Tn^tions of the Royal Society of London, vol. OLIX. (for
the year 1869), pp. 111_ 126 . Recelved May g0j _ I
others! CI ' ^7 V S SUp P lemeQt ^ to my" Second Memoir on Skew Surface,
e th t ; ' P ' ' ' an relates als '
eS K ' er ' " ' Uartic sorolls ' " Pointed out to
in th8 Fesent Ifcmoirj and
are
Quart* Scroll. Xinth Species, S(l,), aM a triple directrix line.
64. Consider a line the intersection of two planes, and let the eauation of tho
r nr n ; he or ^ 3 ' that f the sec nd p' e tain L:r y
paiameter 6, the equations of the two planes may be taken to be
(P, 1, r, s%0, iy = 0, (, Jfl, 1) = o,
Hebe, die
410] A THIRD MEMOIR ON SKEW SURFACES, OTHERWISE SCROLLS. 313
where (p, q, r, s, u, v) are any linear functions whatever of the coordinates (#, y, s t w).
Hence eliminating we have as the equation of the scroll generated by the line in
question
(p> !?> n s'ti.v* ) 3== o,
viz. this is a quartic scroll having the line u=0, v=Q for a triple line; that is,
the line in question is a triple directrix line.
55. Taking & = (), y = for the equations of the directrix line, or writing u=a! t
vy, and moreover expressing (p, q, r, s) as linear functions of the coordinates
(x, y, z, w), the equation of the scroll takes the form
and we may, by changing the values of & and iv t make the term in (,, i/) 1 to be
where the arbitrary constants a, , 7, S may be so determined as to reduce this to a
monomial fee 1 , /cafy, or ]ca?y\
5G. The coefficient k may vanish, and the equation of the scroll then is
z (* ]5>, y) 3 + w (*'>, y)' = 0,
or, what is the same thing, it is
(#5>, y)*(z, w) = Q,
viz. the scroll has in this particular case the simple directrix line 20, w0, thus
reducing itself to the third species, $(1 3 , 1, 4<), with a triple directrw line and a single
dvreotirito line. It is proper to exclude this, and consider the ninth species, S(1 3 ), a&
having a triple directrix line, but no simple directrix line.
57. The scroll $(1 3 ) may be considered as a scroll 8(m, n, p) generated by a
line which meets each of three given directrices; viz, these may be taken to be the
directrix line, and any two plane sections of the scroll. The section by any plane is
a quartic curve having a triple point at the intersection with the directrix line;
moreover the sections by any two planes meet in four points, the intersections of the
scroll by the line of intersection of the two planes. Conversely, taking any line
two quartics related as above (that is, each quartic has a triple point at ita i
section with the line, and the two quartics meet in four points lying in a line]
lines which meet the three curves generate a quartio scroll $(1 8 ). This appears
the formula
8(m, n, p) %nmp <wn fin yp (Second Memoir, No, 5);
we have in the present case
in = 1, n 4, p = 4, a = 4, # = 3, 7 = 3>
and the order of the scroll is 324 1212, =4, that is, the scroll is a qi
scroll; there is no difficulty in seeing that through each point of the line there
C. VI. *Q
314 A THIRD MEMOIR ON SKEW SURFACES, OTHERWISE SCROLLS. /'
three generating lines, but through each point of either of the plane quavtir.s ,jiilJ
single generating line; that is, that the line is a triple directrix line, bul> m*)'*
the plane quarries a simple directrix curve.
58. We may instead of the section by any plane, consider the section by a '^
through a generating line, or by a plane through two of the three iranoral.!,,,, I*'
which meet at any point of the directrix line; if (to consider only the most iiJ
case) each of the planes be thus a plane through two generating lines, the ,.*' ^
by either of these planes is made up of the two generating lines, and of a rrH
passing through the directrix hue; the directrices are thus the line and two ,><
each of them meeting the line; vo have therefore in the foregoing formula
m*=l, n = 2 t p = 2, a = 0, /9 = 1, 7 = 1,
and the order of the scroll is 822, =4 as before.
Quartic Scroll, Tenth Species, Qfy wit k a ***** skew mhio met twicB % ^
generating line( 1 }.
^ ^ bteHleotion of '" P 1 ^ and let tho oqmti.m ,,f ,m
a vamble
fo ?, 'F, 1)' = 0,
line the skew cubic determined hy
= 0.
^::r
scroll as described above. ' d ^ th 8Cro11 is ^"^quently a qn ftl . t i
, In,
^ rii
2 indicting that it 1. me twice by flBflh ia a nod(l1 (") "ne on tho aorolJ, tbo o
410] A THIRD MliJMOIR ON SKEW SURFACES, OTHERWISE SCROLLS. 315
60. The coordinates (a;, y,, z, w) may be fixed in such manner that the equations
of the skew cubic shall be
j, y, z
y, z, w
or, what is the same thing,
each of the equations $</ f>(t = $ ) rq' r'q = 0, pr' p'r = is then the equation of a
quadric surface passing through the skew cubic, or, what is the same thing, each of
the functions prfp'ty, 1'$' ~ r '<l> 2 }r '~P >r ^ s a linear function of yw z", zy xw, wz y n \
and the equation of the scroll is given as a quadric equation in the lastmentioned
quantities. It will bo convenient to represent the equation in the" form
(H F, G, 1$ A F, Crftmu z 2 zti anu xz~y^ = Q,
or, writing for shortness
which letters (p, q, r) are used henceforward in this signification only, the equation
will be
via. this is a quadric equation in (p, q, ?), with arbitrary coefficients.
61. Comparing with the result, Second Memoir, Nos. 47 to 50, we see that in
the particular case where the coefficients (A, B, G, F, G, H) satisfy the relation
AF + BG+GJ{=Q ) we have the eighth species, 8(1, &), with a directrix line and a
directrix skew cubic met twice by each generating line. We exclude this particular case,
and in the tenth species consider the relation AF + BG + GR=Q as not satisfied, and
therefore the scroll as not having a directrix line.
62. I consider how the scroll may bo obtained as a scroll 8(m*,ii} generated by
a line meeting a curve of the order in twice and a curve of the order n once. The
first curve will be the skew cubic, that is m3; the second curve may be any plane
section of the scroll; such a section will be a quartic curve k"""" +.IIVOA nnr^
at each intersection of its plane with the skew cubic. ^
cubic, and a plane quartic meeting the skew cubic in tnree poiuw, BIMIU ui buem a,
node on the quartic, then the scroll generated by the lines which meet the skew
cubic twice and the quartic once will be a quartic scroll. In fact (see First Memoir,
No. 10, [339], and Second Memoir, No. 5) the order of the scroll is given by the formula
8(irP, n)9i([m] B + Jf) reduction, = 16  reduction. And in the present case the
reduction arises (Second Memoir, No. 4) from the cones having their vertices at the
intersections of the skew cubic and the quartic, and passing through the skew cubic.
Each cone is of the order 2, and each intersection quA double point on the quartic
gives a reduction 2 x order of cone, = 4 ; that is, the reduction arising from the
three intersections is =12; or the order of the scroll is 1612, =4. .
402
1 THIRD MEMOIR ON SKEW SUBFACES, OTHERWISE SCEOLLS. [410
6 lnr e t hZ;h illStead V 1 " ^"A * [ ^ in gene '' a1 ' *** ^ <^n
and of 1 Z I*" 61 ' g \ 6; the SGCti0n iS hm ' e made u ? Of the ***
and of a p ane cubic passing through each of the two points of intersection of

r;~r, *""""" <" "* " > "  >. *.
{instead of bdonrimr f n ,!,, ? . , '' e P"rtilar case where the line
P1 Per lnVOl " tl0n) meet3 a ^ *e lo
., , , , '*; ;i r d!iiates "
involution is rap esses that the hue shall belong to an
(4,S, 0,f, G,
or Mo irs: S'^i s ^ ^ 
^ ^Y : T tt ' V* : ^ a'^ : a* & , p ^ , ^ _ .,
, , , ^ "^ : ^^ : *> W*o : W '. w : AB'^L
and denoting ather of these seta of equal ratios by
: & : fl ; /
then ( fll ft. Cl /, fl , , () Bntis/y identicall the z . ' a ' h >
the line. "Jf^hCftu, and are said to ba the six coordinates of
410] A THIRD MEMOIR ON SKEW SURFACES, OTHERWISE SCROLLS, 317
We have to determine the locus of the line (a, 6, c, f, g, h) the coordinates
whereof satisfy the relation
and which besides meets the skew cubic yw~^ = 0, yz w 0, fcs ?y 3 0.
The equations of the skew cubic are satisfiec by writing therein
x ; y \ z : w = I : : f 3 ; t 3 ;
and hence taking 6, <jy for the parameters of the points of intersection of the line
(a, &, o, /, (/, h) with the skew cubic, we have
1, 0, 6*, 0>;
1, <j>, # #
as the coordinates of two points on the line in question ; whence forming the
expressions of the six coordinates of the line, and omitting the common factor </> 6 t
these are
(", 6, . o, /, </, ft) = 0$, (8 + 0), 1, 6* + fy + # ty (6 + </.), ^^,
and hence the condition of involution gives between the parameters 0, < the equation
(A, B, G, F, <?,
Moreover the coordinates of any point on the line in question are given by
m : y \ z : w = l + m : 10 \imft : ffi + nuj) 2 : W 3 ti( a ;
and writing as above p, q, r = yw z i ) yz xw, ccz f, we thence find, omitting the
common factor (0 d>) a ,
^j : q : r = ^ :  (0 + </>) : 1 ;
and eliminating ^^), ^ + <^, we at once obtain
(A, B t 0, F, G, fl^pr, qr, r*, fpr, pq, ^ a ) 0,
oi'j what is the same thing,
(S t F t 0, if, jlJ'.GSi), ff ,r)'0
as the equation of the scroll generated by the line in involution which meets the
given skew cubic twice,
Reciprocal of the Qmrtic Scroll
67, I propose to reciprocate in regard to the quadrio surface B + 2/ a + s s + i a =
the foregoing scroll
6 6
(J2, ff, 0, ^AF, ff Ip, ffl r)* = 0.
If the coordinates (a, 6, o, / g, h} of a line satisfy the condition of involution
(A t B, G, ff, G, ffya, 6, c, /, </, A)0,
THIBD ME
C01lditioil
a,l l c, f, g , h) = .
Tim reciprocal of the beforementioned skew cubic r
vol. vii. (18BC), pp. 8782, [872]
" "
equation of the osculating plane is
f
Pneter whereof is i, the
1, 2, ff
1 , t
or, what is the same thing, the equation
IF =0,
s
Hence fo, the H,e ,Meh is the intersectimi rf
F)=O
oraitbing . the common factor
we have thus between the parameter, ft ^ the relati(m
410] A THIRD MEMOIR ON SKEW SURFACES, OTHERWISE SCROLLS. 319
let p be the third root, then we have
y
6 + $ + p = T ,
w
6<}>.p = ^
and thence
(X, F, Z,
Substituting for + <j> and ^ their values in terms of p, we find
Fp [7 a  ZX  pXY]  GpX (7 p,Y) 4
or, what is the same thing,
from which and the equation
(x, F, ^
we havo fco eliminate p,
70. "Writing for shortness
( ., H t G, A^X, Y,Z,
(JJ,
(^,5, 0, .$ )==S,
and therefore .y+ jQF+y# + 8F0 : the two equations are
SF = 0,
Writing the first equation in the form
multiplying by p, and reducing by the other equation,
/3(p"X  pF) paX ~ 7 F = 0,
or, as this may be written,
320 A THIBD MEMOIR ON SKOT SURFACES, OTHERWISE SCROLLS. [ 4TO
From this and the preceding equation we deduce the values of p'A'pF^ ttd
pXY\ viz. writing for shortness P p ^ awi
we find
p'X/>
or, what is the same thing,
whence also
and thence
and we have therefore
E F ,
r
(Mepeiident of p) a
1. I fab any one of these equation, for instanoe
^3 Tr = I/jr
or, what is the same thing,
t s m tt to , obtain f the r
we h at f ' m the values f A 7, S, if only + 5ff + Off aot = 0,
410] A THIRD MEMOIR ON SKEW SURFACES, OTHERWISE SCROLLS. 321
and substituting these values, the equation in question becomes
qi' ( Ba t A/3  HS)
r fl ( GetAy ffS)
This becomes
.Ar (q/9 + ivy) = Tlr ( pa) =
 Bqra
+ ff (pr  q a ) a + F (pr  q 2 ) a
+ If { qrS 4 (pr  q a ) 7}
viz. the whole equation divides by a; and, omitting this factor, the equation is
.Apr I .% + Gi a + #(q a  pr)  #pq + f/p s = 3
or, what is the same thing, it is
(H t ff, O.B.AF, <?$p, q, r^ = 0,
where I recall that we have
p, q, r~/3S~7 a , fty ctS, y j9 a ,
a. j3, 7, S being linear functions of the current coordinates (X, Y, Z, W), viz. we have
=( , , H, G 3 A$X t Y, Z> W), '
p = (H, ., F, J5J ),
72. It thus appears that when AF \BG + OH is not =0, the reciprocal of the
scroll
(H, F, 0,3, A F, G^p, Q t r^ 
has an ecr nation of the verv same form.
322 A THIRD MEMOIR ON SKEW SURFACES, OTHERWISE SCROLLS. [410
equation contains the constant factor AF + BQ + CH, so that throwing thia out, tlm
reduced equation will be only of the third degree in the coefficients.
74. The transformation is a very troublesome one, but I will indicate tho stops
by which I succeeded in accomplishing it. Each of the functions (p, q, r) in a mimlrir
function of (A', 7, Z, IF), say,
P = (, b, c, d,f, c/, h, I, m, ftpT, Y, 2, W}*,
q = . . 5 ),
'=(">  I );
we have to form the value of
viz. representing this for shortness by
Ca> 6, G, d, f, g, h, I, m, if
',..
t" ..
me cootncient of Ji* is ' "
(B. F, 0, B.AF, fftot, a", ., M ', , ),
that of X'T is
(H, F, C, S, A f,
of the t, ms ,Y, ^. 7 , fe in
viz. this fa P '' = lfe '' ' <l ' f < ff ' >'' l  m ' "W. Y, Z,
HO I A THIRD MNMOIll ON HKKW HUUIJ'ACIIW, OTU13KW1WK .SOHOLLS. 323
Eitiil Mm viilmvi of lltn (inulliimmdH <*, />, ... wluuh outor into iho Inrnnilu. 1 am ^ivmi by
mirnmi itl 1 Lliii I'tillowinj,; viilucH nf p, i, r; vi, UICBO urn
A* !' # IP r# &Y A'r A' ir ,nr #ir
P . . / yi //. ... /' ij , . (u<\  <? t ... tw, /iV' T , /u/; AM, v^,  ,mrQ;A r t r, ^ H ; X J ,
I  ti* van iii/'Vr'  af/^ .2r/.A'
Hl t IHl, <!(!, JUl, Ml, Aft, Alt, A*. AH, A(! f .
Ml// 17'^ 7'7/ >\ lift ///'' M'V' 1 Q 
.,. p* (,'//
VTi, AM un iiiMliinrp til' lltn culniiluMnn nf a nin^lo innn, l.liu t'n^lliiiotil nf A' 1
(//. A 1 , (/, /.', A *!<\ '
vi'/. thin in
// (/i //  (/'y
ill,. wlin) hnti in UMIN (AF+IHl \Vll) 11*. vi/. ilmni is tin* liiilm f \F}' Jltt\<<l/i
tt.H llli'lllinllf'il illtnVC,
7(1, TliinwiiiK mil lln* finfur in in.hi.k.n, AF\> IHt '\>Wl, lh" iMulifin ttl' llu>
l HtT'lll 1M 1'nltltll It" 111 1
I A' B l' .
lXX ,
^.VMI' t  \\AU1I 
>f.A"*r* . 2/v/+^<//
tf
Iltf* 
1 '
412
324 A THIRD MEMOIR ON SKEW SURFACES, OTHERWISE SCROLLS. [410
+ XY 3 2FGH
+ XY*Z AFE  BGH  SF*H + 2W
+ AT'IF" 2ABH UFG  BFH + 206H + 8F*Q
+ XYZ*  AFG 2BFff+BG*OGH
+ XYZW AF 8AF*UBG + ACH  2B*H + J3FG + oOFIf ~ 2CG*
+ XYW> 2A*B  3^F+ 2AGG + &Q + BOH  QG'FG
X&W 2ABF2AOG 2BCB 8BF*+OfQ
X2W* A*G+AB t
XW S 3ABG&
BFHF'G
+ Y*W AF* + Ml  2CFH  F*
ABF
+ YZ*
+ YZ*W ACF+ 2&F BQG 
4 YZW*
+ YW* A&
GF*
But the option may be TO , itton
n
r ( a^S + 2af
Z ( a 7 S  23'g
that of VlPiB^arof^^ 6 c 3 efficienfc of
J'ff, that of 7& haiml^r* ^ ^~ gain ' the coefflci ^ of
410] A THIRD MEMOIR ON SKEW SURFACES, OTHERWISE SCROLLS. 325
or, what is the same thing,
as may be verified by actual substitution of the values of the coordinates.
78. By what precedes, substituting for p, q, r their values in terms of a, & y, 8,
it appears that we have the remarkable identity
(H, ff t C> B, AF,
t 7 ( a/3S 4
+ 2 ( a.yS 
1 4 T7( aS 3 
79. In the case above considered of the tenth species, $(3 2 ), for which
not = 0, the three forma of the reciprocal equation are of course absolutely
equivalent to each other. The first form has the advantage of putting in evidence
the fact that the reciprocal scroll is also of the tenth species; the other two forms
do not, at least obviously, put in evidence any special property of frho reciprocal scroll.
Reciprocals of Eighth Species, 8(1, S 3 ), and Ninth Species,
80, If AF+BG+ (7// = 0, then the equation
(ff.^a, JJ, 4JF,
is a scroll of the eighth species, $(1, 3 a ). The first form of the reciprocal equation
becomes identically = 0, on account of the evanescent factor AW \BG\ OH, but the
second and third forms continue to subsist, and either of them may be taken as the
equation of the reciprocal scroll. Taking the third form, and calling to mind the
significations of (a, /9, <y, S), viz,
H t G,
, W),
ff,
a, .
it is to be observed that = 0, /3 0, iy = 0, 5 = are the equations of four planes
passing through a common line, viz. the line whose coordinates are (A, J3, 0, ff> G, H) t
and the equation thus puts in evidence that this line is a triple line on the reciprocal
32G A THIRD MEMOIR ON SKEW SURFACES, OTHERWISE SCROLLS. [410
scroll; that is, the reciprocal scroll is a scroll of the ninth species, 5 (1 s ). Or stating
the theorem more completely: For the scroll, eighth species, S(l, 3 3 ),
(H,F, C,E,A~F,~G\^^ r)s = 0,
generated by a line meeting the line (F, G, M, A, B, G), and tho skew oubio p =
q0, r=0 twice, the reciprocal scroll is of the ninth species,
X ( =5 + 3807  2/3 3 )
+ Z ( 7 S
+ T^( S^
having for its triple line the reciprocal line (A, B, G, F, (9, //).
81. It should of course be possible, starting from the equation
(*P r , 7)<+^(*;pr ) 7)3+ IF (^ Y) r)>o
<> "> 
Hence writing for a moment
Illl .l A Tllllth .MKMiUH ON HKKW mNlKAOIW, OTHKIUVJHH HOUOLLy. 327
thai in
y i r n ii' ( r IK  %*) i AM* < A'#. ro i a I'^'A' IK,
II, liy I hr <<iiuui<ui nl' I, In; Miii'ull ;
llllil \Vtl l.llllH HIT I hul. IM I'ljimiinll (if Jli! IVI'lpl'lK'ul HLil'llll 1M
( //HI  y) (.I*; //) ( i/o .r^if t,. 0,
"i" ' ;11 .V M* ' ' 1"' (l  v 'w '' ' H ' L "Will ' s '(l. "I J ) Kdnnmlii'il by a lino nnuM,iiig Llui lino
", '" t), ami I In* cnliiiT ritmi p 0, n^ I), i" i() Uvinn, Tho <!iimtitm M nliviuiiHly
iiii'linlnl in tin ^'iui'Vnl 'ijiiiil imi
Nil. lu'lui'iiiu^' ID (In* ^finnil IHH nl' tlm nrull, diV/ArfA w/wxjiVw, (S'tl, H a ), il. IH
ti'n)irr in hliuw ^riiiiit'tfii'iiilly hn\v it i iluil iln ivrijimnil in n ftornll, /i/'/i^t N/JIK;M,W,
S'U 1 ), Ciiiihitlir in dif Mn.ll iS'{l, H') uy 1'lmm ,hrmij.jli 1,1m tliri'ttirix lino; t.hiH
rniiiuiiirt ltu'i'f ^.'iM'UUinti liitiM nl' Iliti wrull, vi/. Lilian iii'd Mm Hitli' ol' l>hn trian^lt)
litiu'f Ity llir llti'if [inintH j' iiih'ivu'iHi'Mi mf Uii; ilarti' \vi(li lhn Hkmv t.niltio: huniMt
in iliD i'iTiir..i'ii lit^inv \vi* hnvi* a ilirritlrK Mm* niuli i,(, til. I'm;!) iuint. nt' il. l.linvd
ai< llii'fi* i,;rn<')nliutj lini'n; llnil. in, wn Imvn u ni p nll N(l*) wil,li u Iripln tlii'tirdnx linn,
I '"iivi'ivu'ly, iti in linj< wirli [lip wmll ,S'{ '), rurh pliuii' llnmigli llin l.rijilu (liviail^rix linn
itjt'i'f i (lit* siri'iill in tlti^ HIM' llir.'i linn's ami in a Him(lu ijnin*vuling linn; wlii'min
(Inn* i 1 * in !li< ri'i'ipnu'jtl nt'tl H aitiiln tlin'rlrix linn; but. in unlcr l,ti wlinw l.lial
i) ii n ).ii,.ll ,s'( ( ;i j ). wi liavn yrl i.u H!MW Ihut Uni'i* in, us n ntnlul ilinmlrix, u
rtkiw niliii* nifl (Avii'r liy twli ^rii'Malin^ linn; thin iuiplii'M tlnif,, iiMiiprutiully, in l,fin
wii'cll .V(t 3 ) nii'li grin'mliiit; lint* IM lite mtriwiHitni nf (AS'tt OHtinUrin^ (>liuu'H of u
M^i'sv ftiliit! (liMtj^int jiliiiMH nf a titiuliti (UIHC), wh mittli fitiiuo cotihiiuiug l,v"
liiKn M' tin' wntl! a jfi'nini'trirjtl proprfty whh'h IM far from nbvimiH;
iu ill Ncritll, niititt hjtirii'x, XflC), wlu'n; the i ..... : ...... ..... '
, flu* pr<.iniy iJnit. I'iM'li jiohrnniiig Hint in u linn
t' !i!mbi IM ih' )riipiMly ili'tt fiivli line is jtten UMI .MI.., ( ,, W *. ...... ......... .. ........ n
i'H trf a nkiw (riihit 1 i MI*, wluit SM iht* Hnmn lliiiiXi iw liingont ]ilam!H of u tjuartic
Ainimus, J/y IM, IHOO.
iho loagoiiitf Moinuir ww* wriltou I runiMvistl from ProfcBwor Cromoiia a
I altar (foiiod Milan, Nnvoinbor 2U, 180K. in which (btwiduH tins ninth and ItmtH pooioa
'roiiidn i> (l above!) ho refbm t two oilier tfpuaiu* of ijumtio Morollu. Ho tromnrkfl that
328
A THIRD MEMOIR ON SKEW SURFACES, OTHERWISE SCROLLS. [410
*
Deficiency
No. of
species,
Nodal curve,
Bitangent torse
Corresponding
to niy species,
1
r 3
2 3
10
2
?/ 2 + J?j
^s + 5,
7
3
fl, 3
JTa+JP,
 (say, 12)
4
R
5,+^
^i 3
(say, 11)
p = Q
6
^sf l
^a^
2
5
7
r a
W
8
8
^! 3
^
9
9
^l 3
7?j' 3
3
10
J?,'
^
6
pl
11
A+A'
J?, + A'
1
12
^
#
4
where T 3 denotes a skew cubic, 2, a torse
conic, K, a qlla drio cone, A, R, S diff ent
or three
u:
QH ^ n ^ /
line" R* R*
' Xl Mi
I have not yet examined the two new
viz these aw
,
' f ' "
twice
moT, mn j *i
mentioned m this enumeration:
411]
411.
A MEMOIR ON THE THEORY OF RECIPROCAL SURFACES.
[From the Philosophical Transactions of the Royal Society of London, vol. OLIX. (for the
year 1869), pp. 201229. Received November 12, 18GS, Head January 14, 18GD.]
THE present Memoir contains some extensions of Dr Salmon's theory of Reciprocal
Surfaces. I wish to put the formula on record, in order to be able to refer to them
in a "Memoir on Cubic Surfaces," [412], but without at present attempting to com
pletely develope the theory.
Article Nos. 1 to 5. Extension of SALMON'S Fundamental Equations.
I, The notation made use of is that of Salmon's Geometry, [2nd Ed.] pp. 450459,
[but reproduced in the later editions, see Ed. 4. (1882), pp. 580592], with the
additions presently referred to; the significations of all the symbols are explained by
way of recapitulation at the end of the Memoir. I remark that my chief addition to
Salmon's theory consists in a modification of his fundamental formulas (A) and (B) ;
these in their original form are
a (n 2) K H p + 2<r,
b (n  2) = p + 2/3 + 87 f 3t)
o (n  2) 2<r i 4/3 + 7
6 (n ~ 2) (n  3) = 4/c + [ab] \
c(?i2)03) = 6A + [c]H
where
[ab] = ab  2p (
[ac] ~ac  So,
[bo] be.  3/3  2'
a vi.
330
A MBMOm ON THE THEOBY OF BEOIPBOOAL SURFACES. [ 4U
*
to node, these will be
but these present themselves
cip rMa l equations.
respectively,
fonnl betje *"*
6', cuicnodes,
S, binodes,
j , pinchpoints,
%, closepoints,
0, offpoints,
t", cnictropes,
f , bih^opes,
j' , pinchplanes,
%', closeplanes,
0', offplanes;
[06] = ab  2p j,
[ac] = ac So  Y
A>
i
abore
; e r
reci P ro cal term
*>,
in th,
M ^eir vahes, the
which replace the original formula (A) and (B).
1 This addition to the theory is in fnnt a'
from tlie t af UB  L e i
411] A MEMOIR ON THE THEORY OF RECIPROCAL SURFACES. 331
5. For convenience I annex the remaining equations ; viz. these are
a'= n(n 1) 26 3c,
/ = 3ji (n  2)  66  Bo,
the equations
(jr=&s&2/c3y6i,
7 = c a  c  2/t  3/3,
(q, r in place of Salmon's R, S respectively); the equation.
a = a j
and the corresponding equations, interchanging the accented and unaccented letters, in
all 23 equations between the 42 quantities
n, a, 8, K ; b , k , t, q, p , j ', c, h, ?, or, 6, % ; /3, 7, ; S, (7,
', a', S', *'; 6', k', if, g', p', j'; c', h', r', o', 0', K ' ; (3', y, ' ; ', 0'.
Article Nos, 6 to 12. Developments,
6, We have
(a b c)(2) =(/c~ J /3~5)6 J Q4 7 3,
 8&  ISA  G (60  3J3  27  i) ;
and substituting these values of S, in the formula
' = a (a  1)  2S  3,
and for a its value, = n (w 1) 26 3c, we find
n'  ?i O  l) fl  n (76 4 12o) h 46 a h S6 + 9c a + 16o
 Sk  ISA + 18/3 + 15W 4 1 a* _ fl*
332
A MEMOIE OK THE THEOBY OF BEOIPHOOAL STOFACK . [4U
Writing in the first of these '~2 = (_!)_
other ^ by mMM of the vaiues rf ft .
'=: 2042 + * ff _2
(Salmon's equations (D)).
W hich
to the
q , , .
8.  * and reducing the
eq ,, ati on,
,
which, writing thereia a = n(nl\9h * i
(n 1)  26  3c , and * = 3n (n  2)  6&  So, becomes
a' = 4n (  2)  86  lie  2/  3 X '  2C"  45'.
' * ^ iproea! of ,.
of the e generated by he tanj 11 T* ^ f th S P ilMcie torse
". what i, the same th Sg t i tfe n, T ? ' f"' 6 SeCti " f the
pla D e section; that i s , it s ea'n I 7 , ' ! Pm tle ^ mes which to oh the
spinode curve and the plane ZtiL m fl ? f. f P iutS f ^<* of the
The spinode curve is in fact fo ' "', J' ^ " ^ <nd " f the s P inode
interseotiou of the surface by the He ^"Jl * f"?'' Y"" 1 ' ^"g" 1 ^"
a curve of the order ~ ^^'^ "
~ t * order a redaction + U . a, S in g
Hessian surface ns the fee in the , "' ! ," """ S8me thing ' that
cuspidal curve taken 1! time s a e ,1 w h oh T 1 i"? *, aken " tim6S ' md in the
also as appears post, No 'T , , amved at b y othe1 ' me , and
p 
.
pmohpomt, a closepoint, a cnicnode arrl 7 I i ' ' '
must signify that the surface d 1 ff 60 *
which are not regarded as belon"L t !h "
itaelf that for tL reciprocal of'T Lh
reciprocal of a binode it is a lL ? * thls curve ls a
Pinehpcint it is a ,i ue c^nti 2 CCd f ti '^ fc ,
hue counting 3 times. fche reci Pcal of a closepoint, a
reol P ro ls of a
Geometrically thi s
" ertain
at Ce S ^ est
ic, and for the
411] A MEMOIR ON THE THEORY OF RECIPROCAL SURFACES. 333
9. It is clear that p' will in like manner denote the order of the nodecouple
curve.
10. I express in terms of
n, 6, c, h r k, & 7, j, 8, x , O t S
such quantities and combinations of quantities as can be so expressed. We have
rt = ft' = n( 1) 26 3o,
K' = S (n  2)  66  So,
4* = 127i 4 c (fin  6)  Gc 3  % + 30  2&
24(  ( Sft + 8) 6 4 (15n  18) c + 86 a  18c a  2 (8ft  ISA) + 20/3  167 + 4>j 4904 6#
q = & a  b  2&  87  6, (i auprk),
r = &  c  2A  3/3,
2o = c (n  2)  (4/3 + 7)  0,
8p = (IGn  24) & 4 ( 15m + 18) o  86 s + 18c 3 4 2 (8fc  18A)  9 (4/3 + 7)  4j  90  6^,
8 = Sn (n  1) (n  2) + 6 ( 32 + 56) + c ( 17n + 40) + 86 a  18c a
 2 (8A  ISA) + 1^ (4/3 + 7) + 4j 4 170 + 6^ 4 85,
2S  (n  1) (  2) (n  3) + 6 ( fai 2 + 20/t  24) 4 c ( 6w a + Ion  18) 4 126o + 18c a
+ (8/c  18/t)  9 (4/3 + 7)  0(9 + 20,
8ft' = 8n (51  l) a + ( 32 4 40) & + ( 21 4 30) c + S6 a  18c a
 2 (8/c  18/0 + 21 ( 4 ^ + 7)  12 J + 21  !8%  160 24J5,
c' = 4re (n,  1) (n  2) 4 ( 16i + 28) b 4 ( 10a + 26) c 4 4& a  9c"
 (8/c  ISA) + 10 (4/9 + 7)  4j + 100  6^; ~ 6C  8B,
f =  a + ' (n'  1)  So', c' supra),
W  9S' =  lift (  2) 4 a (n'  2) 4 226 4 30c, (', a supril),
0'==c'(n / ~2), c' supra),
4/c'  3 (i 1 4 3/3' + 27')  2p'  / = ( 4 y + G) 6' + 26' a , (w' f 6' suprii),
6/1  2 (*' + 3/3' 4 27')  So'  tf = ( 4n' 4 6) c' 4 3o' a , (ri, o' supri) ;
[or in place of either of these,
8/c'  ISA'  4/>' + Do'  2?" 4 3 X ' (2f  3c') ((n'  2) (n'  3)  a}, &', c', a supra)},
p' + 2/9' + 87' 4 3i' = &' (n'  2), I)' supra),
3 r ' + 2i' 4 ^  off 7  &  40' = c', (c' supra),
(twentythree equations, being a transformation of the original system of twentythree
equations).
334 A MEMOIR ON THE THEORY OF RECIPROCAL SURFACES. [411
11. Forming the combinations 4i + 6r, 2USq + l8r (the last of which introduces
on the opposite side the term +m), we obtain Alices
te + Or = c (5n  12)  5 y ~ IS/3 + 30  ^
equations which are used post. No. 53.
,*
' lt" e ? r;? ttt f diff T !t points f
on face by any piaue
oint of fche line
f  B
z R, ,, , ,? on * T face by any piaue * hrough the Hue tai s tfe i;
*Z' A, '? ,? e r ? oint of fche line ne and the same ta g^ pi.
hen the sect um of the .uvfaoo by the tangent plane oontams the line at least twice
li tl VT 3 '' le lme " fo '' Mi ' if thMe times the line is far
turned a Z,';'; P "T* " le t0ml " " nl " U " e ^ like m ' ^
* ra '' oscl ' kr a ^
tl '
trned a Z,';'; P "T* " le t0ml " " nl "
vm be oo,* ra '' oscl ' kr a S^ plane. The* epithets, S erola r> torsal and oscular,
will be convenient m the sequel.
Article Kos. 13 to 39. Explanation of the New Singularities.
I proceed to the explanation of the new singularities.
13. The cnicnode, or singularity 0=1, is an ordinal , conica i . ; . ,
the tangent plane we have a proper quadricone.
= ' s
cnot is P conteot ; " say rather the
oauta.e !s . touch.ng a surface, not at a single point, but along a conic.
the te 3 aV U Tr aVlg "'Y 110 * 01, and the reciprocal surface having
ctr:tS~ d SocTZetr? 6 of the cnicnode sk direoti ns f
touching it at x points ^ The , k^ nf t " lWS ^ f "" onicta) P e conio '
twice and in a S T , the C ' 11Ctl ' pe meets the surface in * nic
twee and in a les dual curve which touches the conic at each of the six iioints It
would appear that fee six contacts are part of the notion ,f the cnictrope "
that ae idu7cuLT e !' ave ; surfao with a conio of i 3lme contaot . i> *
on o rix t me" v at al l f lnteree tl0n the P 1 * f c does not touch the
asmte "" ' Collic
and here the conic = 0, F=
8lven as ^e s of intersection of the Uvo couos
411] A MEMOIR ON THE THEORY OF RECIPROCAL SURFACES. 335
17. For a surface having the cnic trope G'=I, the Hessian surface passes through
the conic, which is thus thrown off from the spiuode curve ; or there is a reduction
= 2 in the order of the curve, which agrees with a foregoing result.
18. The binode, or singularity B= 1, is a biplanai node, where instead of the proper
quadricone we have two planes; these may be called the biplanes, and their line of
intersection, the edge of the binode, The biplanes form a planepair.
19. The bitrope, or reciprocal singularity B' l, is the plane of pointpair contact;
but this needs explanation,
20. Consider a surface having a binode, and the reciprocal surface having a bitrope.
We have the bitrope, a plane the reciprocal of the binode ; in this plane a line, the
reciprocal of the edge ; in the line two points, or say a pointpair, the reciprocal of the
biplanes : these points may be called the bipoints. There are in each biplane three
directions of closest contact; the reciprocals of these are in the bitrope three directions
through each of the two points. The section, of the reciprocal surface by the bitrope
is made up of the line counting three times (or fche line is oscular), and of a curve
passing in the three directions (having therefore a triple point) through each of the
two bipoints, The bitrope contains thus an oscular line ; but it is part of the notion
that there are on this line two points each a triple point on fche residual curve of
intersection.
21. We may however have on a surface an oscular line without upon it two or
any triple points of the residual curve of intersection. Such a surface is Mtv \ Ny 3 = Q ',
the intersections of the line aa = 0, y = with the curve = 0, JF = will be all of
them ordinary points, The reciprocal surface will have a binode, but there will be
some special circumstance doing away with the existence of the directions of closest
contact in the two biplanes respectively, I do not at present pursue the question,
22. For a surface having a bitrope B' = 1, it appears from what precedes, that the
oscular line must count 4 times in the intersection of the surface with the Hessian ;
for only in this way can the reduction & in the order of the spinode curve arise,
23 The pinchpoint, or singularity j = I, is in fact mentioned in Salmon; it is a
point on the nodal curve such that the two tangent planes coincide, or say it is a
cuspidal point on the nodal curve, If, to fix the ideas, we take the nodal curve to be a
complete intersection P = Q, (3 = 0, then the equation of the surface is (A t B, <7JP, Q) a =0
(A t B, G functions of the coordinates); we have a surface A0J30, which may be
called the critic surface, intersecting the nodal curve in the points P  0, Q = 0,
^6'J5 a = 0, which are the pinchpoints thereof; or if there be a cuspidal curve,
then such of these points as are not situate on the cuspidal curve are the pinch
points : see my paper " On a Singularity of Surfaces/' Quart. Math. Joiwn. vol, IX. (1868)
pp. 332338, [402]. The single tangent plane at the pinchpoint meets the surface
(see p. 338) in a curve having at the pinchpoint a triple point, = cusp + 2 nodes, viz.
there is a cuspidal branch the tangent to which coincides with that df the nodal
curve; and there is a simple branch the tangent to which may be called the cotangent
335
A MEMOIR OS THE THEORY OF RECIPROCAL SURFACES.
particuiav case
he ode, ,0 CllS1>klaI bl ch >' and
the cider n  2, the tangent to which is the cotangent,
' Mdl1 "' 1 ' 1 si
[411
he
of
the
let
Plane fa " H
. tl, ai
' H P
, is in fact a torsal plane touching
twi e aud in a "H'X'l
e reaSOn that the am by the
, Secti " h at ^ * ident
nocle  oou P le P' a wit1 ' two coincident nodes. The plane
" conple aue in a HM * p"** thi g h ^ ^ to,, inng
o the ,  " P1 fig, the reciprocal
of the pnohplaue la tln.s a point of the nodal curve, and is a pinchpoint' the
' ' ? "".P^point is the reciprocal of the point P  th tang "to
:;; d e i ;: t reeipi r :f the H r " that is  f the tai ^ nt <* p * *
, and the cotangent at the pinchpoint is the reciprocal of the torsal line.
2a. There is in this theory the difficulty that for a surface of the order n the
l plane meets the residual eurve of intersection in  2 points 7 a d if
oo  virtue of h only
ono of the (2) pomts of intersection of the torsal line with the residual lane
onm, a po.nt of the nodecouple curve of the reciprocal sul I, ft
f
a proof.
of A rn ft / rf / 7, / v
01 p t a, di /, g h> t> m> B)
i', Oil + 2l^r
fu
aw
' See Salmon, p. 2ig, wlMM it ia only Btaw[ ^ th() HflgBian ^^ ^^
functions
of
the Hoe.
411]
A MEMOIR ON THE THEORY OF RECIPROCAL SURFACES,
337
and representing this for a moment by
A, H, Cf, L = 0,
H s B, F, M
G, F> G, N
L, M, N, D
then in the developed equation
D (ABO AF*  BG 2  Off* + 2FGH)
~(BG~F\ GAG\ ABH\ GHAF, HFBQ, FG  GH^L, M, JV) = 0,
observing that 0, F> M, N, D are of the first order in x, y, the only terms of the
first order are contained in B ( DQ Z  G# + 2NGL) ; and since G, D, N are of the
first order, we obtain all the terms of the first order by reducing B, G, L to the
values 2i//, G, D ; viz. the terms of the first order are
=  2^ (C*d + D*c 
Hence the complete equation is of the form
 2i/r (G*d + D*c  ZCDn) a; + (so, y}* = 0,
or, what is the same thing, <]> + f*p ~ ; the Hessian has therefore along the line
<c = 0, y = Q the same tangent plane a = as the surface; or it touches the surface
along this line ; that is, the line counts twice in the intersection of the two surfaces,
28. If instead of the right line we have a plane carve, say if the equation be
axj) + P 3 ^ = 0, then the value of the Hessian is a;<3> + P 1 !' = (viz. the second term
divides by P only, not by P 3 ), so that, as before mentioned in regard to a conic of
contact, the surface and the Hessian merely cut (but do not touch) along the curve
{o~0, P = 0, To show this in the most simple manner take the equation to be
a' + P 2 = 0; let A', B', G', D' be the first derived functions of fa and (A, S, G, D),
(, &, o, d, f, g, A, I, m t n) the first and second derived functions of P ; then if in the
equation of tho Hessian we write for greater simplicity = 0, the equation is
' +
Ph + AB t C
" + P(I + AG, 1
7 + PZ +^ID
PI +& ,
P/ + 8G,
Pm + 5.0
Pf+BG,
Po + 0* ,
P?t + OD
Pm + BD,
Pft + OD,
Prf + ^ a
= 0.
D' + PI + AD,
The equation contains for example the term
 (D 1 + PI + AD)* {P a (bo /*) + P (bG*
 2/BO)},
dividing as it should do by P, but not dividing by P a ; and considering the portion
hereof  D"*P (W + oB 3  2/BG), there aie no other terms in D' B P which can destroy
this, and to make the whole equation divide by P s ; which proves the required negative,
0. VI. 43
338 A MEMOIR ON THE THEORY OP RECIPBOCAL SUltPAOJilS. [411
29. For the offpoint or singularity = 1; this is a point on tho cuspidal oumi
at which the second derived functions all of them vanish. In further explanation hwnf
consider a surface U=Q, and the second polar of an arbitrary point (a, # y, S) ; vi/.
this is (a9. + ^9y + 79 / + S3 w ) s yO l or say for shortness A a /"=0, whom tho ttonniciuhl.*
of the powers and products of (a, ft y, 5) are of course the second dorivod functions of //;
this equation, when reduced by means of the equations of tho cuspidal oiirvo. may <uMnirn
a factor A, thus assuming the form A(aP + /9Q+ 7 ^S6') 3 =0, and if HO tho inJiormxitimiH
of the cuspidal curve with the second polar (20 + 0, if, as for simplicity in Hitpjm.s.nt,
there is no nodal curve) will he made up of tho intersection* of tho cuspidal cmrvo
with the surface A = 0, and of those with the surface P + /9Q + r , &Sf o inctti
twice; the latter of these, depending on the coordinates (a, ft y, S) of bho arbitrary
points, are the points a each twice; the former of thorn, or mtorsootioMH of tho rnspidid
curve with the surface A=0, are the points tf, or offpoints of tho cuspidal !.,.
1 there _ is a nodal curve, the only difference is that tho offpoint* aro H.ioK uf thr
above points as do not He on the nodal curve.
instance of the mor in which this nia^nlariU' iuy
a+ ^ 3 = ' Where ** dofi" of th fi,nti,L m 
* if * e th rdor f HW " rfMu 
W; and we e t
 f Uu
CUSP"'"' ~ " " " " " 8l  ll 'hicc
equation of th
and if fe an ; ri Sco"Z's tho = fo = A ^T " ( "' '"'
:'*il), then
\
has the cuspidal conic = o, Sx' + an n ,,
has tho offpoint s OT = , j, = , W + ln. (l S C ( 0rain 8'/" ldor tho fmm M + OQ'O)
(. 0, y . 0, w = 0) eaoh twice ; 9 ^ ' * ' 8 ' thB l )oilrta (0. 0. U 0).
But siting the same equation in tho form
whore (4 ' O* &! + j!/'  ^ _ ,), = _
411] A MEMOIR ON THE THEORY OF RECIPROCAL SURFACES. 339
it appears that there are also the three cuspidal conies y 3 2a? = 0, xfl ssw = 0. Reducing
by means of these two equations, the equation of the second polar is at first obtained
in the form
= ;
but further reducing by the same equations and writing for this purpose y = WJB (<u 3 = 2),
the equation becomes
(4, (Ja;, 0flj a 5fl) a (3o) s Ay6Aa!), 2a;A  sAw  wAs 8 )" = 0,
that is
a; 3 [20 (3o> 2 A?/  6Aa;) + 3 (2#Aic  sAw ~ wAs)]'  0,
and we have thus the offpoints e?~Q t if 2a 3 = 0, a?ffiu0, in fact the before
mentioned two points each G times ; and the complete value of 9 is & (4 h 12 =) 16 ;
viz, the offpoints are the points (03 = 0, j/=0, = 0), (# = 0, y = (\ w0) each 8 times.
On account of this union of points the singularity is really one of a higher order, but
equivalent to = 16.
I am not at present able to explain the offplane or reciprocal singularity 0' =\..
33, As to the closepoint or singularity % = 1. I remark that at an ordinary point
of the cuspidal curve the section by the tangent plane touches, at the point of contact,
the cuspidal curve : the point of contact is on the curve of section a singular point
[in the nature of a triple point, viz. taking the point of contact as origin, the form
of the branch in the vicinity thereof is y a .'== 0, whore y = is the equation of tlio
tangent to the cuspidal curve], such that the point of contact counts 4 times in the.
intersection of the cuspidal curve with the curve of section. At a closepoint the form
of the curve of section is altered ; viz, the point of contact is hero in. tho nature of
a quadruple point with two distinct branches, one of thorn a triple branch of the form
y s = a?, but such that the tangent thereof, y = 0, is not the tangent of the cuspidal
curve; the other of them a simple branc!^ the tangent of which is also distinct from
the tangent of the cuspidal branch : the poiit of contact counts 3 H times, thnt is
4 times, as before, in the intersection of the cuspidal curve and the curve of section.
The tangent to the simple branch may conveniently be termed the cotangent at the
closepoint ; that of the other branch the cotriple tangent.
34). We may look at the question differently thus : to fix tho ideas, let tho cuspidal
curve be a complete intersection P = 0, Q = ; the equation of the surface is
(A, B, G^P, Q) S =*Q, where ACB 2 = Q, in virtue of the equations P = 0, Q = of the
cuspidal curve, that is, ACB* is =MP + NQ suppose. We have (as in the investiga
tion regarding the pinchpoint) a critic surface .4 (7  I? a = 0, this meets the surface in
the cuspidal curve and in a residual curve of intersection ; the residual, curve by its
intersection with the cuspidal curve determines the closepoints ; the tangent at the
closepoint is I believe the tangent of the residual curve. Analytically the closepoints
are given by the equations P = 0, Q = 0, (A, B, G^N, Jlf) a = 0. It is proper to remark
that if besides the cuspidal curve there be a nodal curve, only such of the points so
determined as do not lie on the nodal curve are the closepoints.
432
340 A MEMOIR ON THE THEORY OP RECIPEOOAL SURFACES. ("4 1 I
35. I take as an example a surface which is substantially the same as one which
presents itself in the Memoir on Cubic Surfaces, viz, the surface (1, w t anj$M*iey t *) g ^0,
having the cuspidal conic w n ~xy = Q, 2 = 0. Since in the present case' AOl&Ll\ w <>
have JI/1, iV = 0, and the closepoints are given by P=Q, Q = Q, = 0; that; is, fchiv
are the points (2 = 0, w=Q, = 0) and (2=0, w=0, y = 0).
36. I first however consider an ordinary point on the cuspidal curve, or conic wa wy
 ; the coordinates of any point on the conic are given by IB : y ; g : w = 1 Q*  () (f
where $ is an arbitrary parameter; we at once find fl+ y  Qtg \ 2 W ) =0 for '] M !
equation of the tangent plane of the surface or cuspidal tangent piano at fcho
f " ng to M the mtersection of this
i *i.
and thence
and the equation thus is
or, what is the same thing,
viz. this is
.
or reducing, it is . ;
b 8 p  ? ' *  rhe
the conic, and in the vicinity of tH ' ^ , Maumed I 3oin(i 0,
bch of the for m . ?** 


411] A MEMOIR ON THE THEORY OF RECIPROCAL SURFACES. 341
= 0, that is in the line <e = 0,w = three times, and in the line ce~Q, w+22=0
(that the section consists of right lines is of course a speciality, and it is clear that
considering in a move general surface the section as denned by an equation in (w, z. ?/),
the line w = represents the tangent to a triple branch w 3 =^j&c., and the Hue
?y + 22=0 the tangent to a simple branch); these lines are each of them, it mil be
observed, distinct from the tangent to the cuspidal conic, which is cs = Q, 2 = 0. And
similarly the tangent plane at the other of the two points is 7/ = 0, meeting the surface
in the cuvve y = 0, w 3 (w h 2#) = 0, that is in the line y = 0, w = three times, and in
the line y = 0, w h 22 = 0.
38. The closeplane or reciprocal singularity %' = 1 is (like the pinchplane) a torsal
plane, meeting the surface in a line twice and in a residual curve; the distinction is
that the line and curve have an intersection P lying on the spinode curve ; the
closeplane ia thus a spinode plane; it meets the consecutive spinode plane in a line
p passing through P, and which is not the tangent of the residual curve. In the
reciprocal figure, the reciprocal of the closeplane is on the cuspidal curve, and is a
closepoint; the reciprocal of the point P is the cuspidal tangent plane; that of the
line p the tangent of the cuspidal curve; that of the tangent of the residual curve
the cotriple tangent; tliat of the torsal line the cotangent.
39. The torsal line of a closeplane is not a mere torsal line ; in fact by what
precedes it appears that the surface and the Hessian intersect in this line, counting
not twice but three times, and it is thus that the reduction in the order of the
spinode curve caused by the closeplane is =3.
Article Nos. 40 and 41. Application to a Glass of Surfaces,
40, Consider the surface FP* + GRQ S = 0, where / p, g, r, q being the degrees
of the several functions, and n the order of the surface, we have of course
/+2p~0 + 2?' + 30.
There is here a nodal curve, the complete intersection of the two surfaces P = 0, R = ;
henco 6=pr, k = %pr (p !)(? 1), =i& (b p r + 1); t = 0', whence (q) = pr(p + r 2).
There is also a cuspidal curve the complete intersection of the two surfaces P = 0, Q ~ ;
hence o=pq, h =* ^pq (p  1) (q  1), o(o p q + 1); whence (r)=pq(p + q 2): I
have written for distinction (g), (r), to denote the q t r of the fundamental equations.
The two curves intersect in the pqr points P = 0, Q = 0, R = 0, which are not
stationary points on either curve ; that is, /9 = 0, 7 = 0, i =pqr.
There are on the nodal cnrve the j = (f+g)pr pinchpoints .F= 0, .P = 0, _ft = 0,
and = 0, P = f R = Q. There are on the cuspidal curve d=fpq offpoints ^=0,
P = 0, Q = 0; and there the gpq singular points G = 0, P = 0, Q = 0. I Hnd that these
last and also the points each three times, must be considered as closepoints,
that is, that we have # = (g + 3f)pq.
342 A MEMOIR ON THE THEORY OF RECIPROCAL SUBFACKS. ["411
41. We ought then to have
b (n ~ 2) = pt
c (n  2) = 2<r + $ ;
the first two of which give p, and then, substituting thoir values
The first of these is
2 = 2^ + 2 ,. + S?+/ + , /j . ^ +/) + (2) , + ;j
and the second is
** + 8 +
 that the equations are satisfied.
Article No. 42. The Flecnoclal Curve
/:
SWtsr^'=."r5j?
Article Nos. 43 to 47 Stirf
3c . '
an 
r Slde of the
A MKMOIU OK TUi; TIIKORV <IK KKl'M'lWUA I,
iH'dci ul' MM Miirlimi' in ,lm ..^in, Tim twu UIIIVPH inh'ixriH in in ]ininlH mi rtn axis
nn 'l ' n in'* '> ninlM. Irinniii^ J(HI U () puiiH til' pninUi, niMmh' MVinnmlnntlly mi
(ppiiNil.1' Midi'M nl' 1,1m IIMM; Lilian ttiHl. >;rfi'ml.< k(ut?*m) riivlnt. nixlll! ciim( nil Mm
Miirlimn; Mm uiidrM tfi'imnih' fi rmtliH, wliirh arc imilal rumi n Mm NIII limp. ami Mm
niNpH jfi'iii'i'ati' K riivlrn. rnnpiilal iiirvcji mi Mm Hiirlimi'. Tlmiv im m 3  in Sifi  JU 1
iiriUvt nf pliin<> i>iiti.iit*t. nimmpiiiiiliitj; in Mn ptniic nirvu In ilm l:tin.>nU iMi'n'tnlii'uliir
A ''I'" ii\ii*. I'lif'lt '!' flu iii pmn:i on Mm xin jlivi'M in !,lu' riiirlii ..... i nur n*
(inui)(iiiiu;v) limvi; iiml \v.' IIHVP Mm.* t.wn *!<< iudi ul' / liiui, mitli thai ulmif* flm
linen M' tiuli jH'l, III Miirliii'M is liuicliiul Ity nn (inui^iiuu'V) ini'iiilinn jilium; VJK. llna;
an 1 ill.' riirulnr plitnrM .1 !/// (), .r ..///:, jmnninj.f llir.iiirjli elm nxjs. I MI
nillmnl. !ilnipini( I" nlimv i) fhal. llii'H.' 'Jw limM urn lintL imi j' Inn y' 4 Elm! ii.
Mny I'Hi'h rcttnftt .!n> itnmr M' Mit< n>iiM>il> nuvc liy M('). Tlic iiUl\inii>< in'
.'h' J ' lim tiiS MA pjivlin wliii'h ^iiimMhih* flit Npinmli ciirvn uii Ilir fiininuv.
I !. Atnl WP ran Miuu vnily Mm). I In rmupli'in iuti'mii'iinn n' MIP fiiirl'ai'n
ilin lli'iiniini in fiiml' 1 up in mvMiiluttn' \villi Mn f t'>n.'i(niii!; Mtriirv ; vi/
t hd>T ul' MM i I'ai'i'  'J,ni t
( )nlT i'l' I ltciiui M*^/;j '),
whi'iiin ni'diT ill inli'nii'rliuii lli/H . . \fltn
Nntliil I'uiA'p, .J(/M a wift iinl.i. S titii.i .H/n ? . SF  ItiiS
<!itHpii(ul iui'v.'. A ciivl(. II fiit ..... i . \>\>K
tin'h'rt nl nintur!, m j  m  *JJi   ?U, l!//i' 'Jj Iri < liv
I'iili'H 'Ji , .'I liiiiii r Urn
l"i. \\ r n limy Ity n ^imilur iTii^niiiuj Jinw Mint lln< MiiiTitr amt Mm ll*'imnit H
Mt'i'l, in flu' imtlat itirvi' ItiKdi U:J liiius. nut! in lli>' ciuipi'lul nnv.' tukni *,!'('
ami nin^'iph'iiMy Mmf, Mn unltT M' Mi r..iilnul iiil'i^rniiMii i,r HtrntnJii tnrv.^ in
.(! In Ull '^/t 'J7r.
Ti flli'i't. (JiiM, nliHcrv Mini, jil any pMiut, \vht'v.r nf H .jiiiuhiir nmla*"' ilic ijintfriil,
plitim iiH'utM lln* Niiriiu,' in M jtuii' ui' liuiii, linn it, in u cuiv* liuvini; at Mir pinl i.t
t it< tnuli 1 with an iallrxir.ii MII each lnMiiili, m 1 miy, H U.H'iHMit,'. rjuHs,{iim in
ptiuii' lixiii'c a t'mir having its *'*! n MH tht< uxi* .!' iMiufi.in ami itt '" :
riiltiiit Ilmi'f"\vilh, iiml Mm trmmf huviji)r with tin* caiv.) .(' tin? niil>'i o I. IN
ittiitn al, any juiinl /'; iln jmiui 7* gimml*^ H rircl*, MIK!' Hun.
Hrdp thu MiiHiit't in imfiiltitnl liy a fiuulrii' MirtJiro it' iv*'"l"ii"
\\w inuntlian iwM'iiiiiiM Imv., a f'HU'j>niniir matact; tlm ri
Mio surJUi! a lU?llmMiinln t'ircli'; uutt I a^miin Miat it
nrolo. HUIICO iT itiu niiinlufi' nf tin 1 poiatw r In fl, w.* luivr *n the mirlm** f
HuHcianodo circloti, **%& HrciiiHlw riruliw, lliitl w, a nVi'autl" tarv.' nf tfiu unlur ift
I wish lo Hhw ihnt wet lmv j? wfiin.<ni
1 Observe iliai Uus trm in m cant te t;tii riil of in a difffrcni umiin#r, by nny Bticrallioit **f
H itucE 11 io which Urn prw^nt ii
nHmufi n***!
344
A MEMOIR ON THE THEOKY OF RECIPROCAL SUUFAOJM.
4*3. The problem is as follows: given a curve of tho order m with
K onsps; it is required to find the number of the conies, contm on a
ami an axis coincident in direction with this line, which have with 1;]io
n 4pomtic intersection, or contact of the third order. This may be w .lvu
f fomnhe contained in my "Memoir on the Curves which mbisfy given
m /. vol. CLVIII. (1868J, pp. 75144; see p. 88; [400].
Taking a  = for the given line, the conic (a, 6, c, f t CL AT),* jyj
. s cent, oa the given line and an axis coincident thomw th tf ! y 1L'
^"tmg these two conditions by 2Z ( it is easy to see that wo Imvn
)=2, (2,T.//)=2,
$
given
mvini
<! b
will
" .
" '
f/i  ]
H run!
!j m .
nim'
mns
'
!,..
' "
* thon
,
where *, ft % S denote {2,Y..), (2Z/) rST./A t9Y im   ,
TO ^.o 1. 2, 2, 1 respecti ely, a, d whl "' ( ^ ^ "" "' " W '"'^^ >*'
respectively. ^' wil01 ? ft ' / 9  7 denote (M/;), (^./j,
a w (loni(IH ih ^
contact of the third order with a l e ' I 7 S do oto ^ liU<,!i uf u
<. ' > ( the order, the class 4Z\ " , (l ? S ' *i or a( V with a W ivun CM,,,,,
2S3, aa3 + #) ) th.m wo Imvo
(3: )4OT
Order of fl ecilo d e 8Urf
0Ior of intersection
curve, Hfli
Cuspidal curve
of contact
of contact
. w
411] A MEMOIR ON THE THEORY Off HEOIPUOCAL SURFACES.
Article Nos. 48 and 49. The Flecnodal Torse.
4)8. Starting from
22&' I 27c' = 6 (66' + 8c')  7 (26' \ 3c')
= G (3n' a  6n'  )  7 (it' 3  '  S)
345
that is
11* W 226' 27 c' = &n'  7S + 6*.
I find
tt'(ll?j/24)22&'27c'
= w (?i ~ 1) (llw ~ 24) + b ( 69n + 96) + o ( 04 + 156) + 266 3 + 87c a
 G2& H4A+ 141/3  D4y + 77*+ 3y' + %  15(9  45^  IOC 9.8.
49. For a sinface of the order ?i without singularities this equation is
n f (lift'  24)  226'  27c'  (n  1) (11  24) ;
to explain the moaning of it, I say that tho reciprocal of a floonodo is a flccnodal piano,
and vice versd; the reciprocal of the rtecnodal torse of tho surface n (viz. the toise
generated by the floonodal planes of the surface) is thus tho flecnoclal curve of the
reciprocal surface n'; and the class of the torse must therefore bo equal to the order
of the curve. The flecnoclal torse is generated by tho tangouts of tho surface n along
the curve of intersection with a surface of tho order lln24; the numbor of tangent
planes which pass through an arbitrary point, or class of the torso, is at once found
to be (nl)(ll ft 24); for the reciprocal surface the order of tho flccnodal curve
w by what precedes '<11' 24)  22i>'  27c'; and the equation thus expresses that
tho order of the curve is equal to the class of tho torso,
Article No. 50. The general Surface of the Order n without Singularities.
50. In the general surface of the order n without singularities, wo have
n = n,
K = n (n  1) (n  2),
& =0,
h =0,
4. , . A
2=0,
' ' 44
' m WP
U (i
A MEMOIR ON THE THEORY OJ? RECIPROCAL SURFACES.
[411
c =0,
A =0,
* =0,
o =0,
C'O,
/?0,
d =o,
7 =0,
i =0,
'* n {*!)',
' (!),
*'3(B~2) 1
/j'= /i(ji
/=0,
c' in (n 
r 2
^0,
A "j
** u,
2) (3ft 4),
7' = 4^(71 2) (.,
t' =0.
^
'
41
A MKMOIIt ON 'I'lIK 'I'lll'XWS" ()]' ItKril'ltOdA!. HlMtKAtHW.
rlinli! Now. ill lii (!!'. liu't'stt'iiuiuni of FnrntHlu fm ('I'.
fil. Tlu< valim //' a{Hii)(llH..ai) I'm* n
by Sulimm liy intloju'iidriil, j;mnii'Irinil rniiM
\villi,.iii
, vi/. he ul
iiuinlmi' !' inf,i ( rsir.innn nl' f.hn Npiimtli* iMtrv.t (nnlrr  I// (:!}) ly .m llr
nl' rim tinlor Lin :H.
fiy. Tin) vttliin M' /J' nilt.'il. Itn uhliiiiutliK' in tint cicic nl' 11 siirlMm with Minifiilui illrM,
anil 1 liitvii linen Inil lu ruiirliul^ Mini, we Imvc
I'liliriu 1'llUt'Uutl (>",,/, (f, ( y, f? ( //, (',,/", /'', x> fr . 'O.
ImL I liiivn luil, ynf, cuiiipli'lt'ly iti'du'iiiiiii'il I hi* ruifliiiinli ol' lltn MIMMIV ttniMliun, T
'iil ItirnitiliL in I In iimti nf 11 hiirKfi ...... f I In 1 nnl.i' i* williniii itii
X> (; > M* '''/ "' A:' ( '" W tllt ' 11 lt!
(V, //, */', o', r', ft', y, I' hiivin^ tint vuliitH in tltt* tl'ivx"i"K Tultlc). ll wiw liy ii^
i'ur # tin t'xin'nniuii f flin Wti lurm Iml with iuiliMnritiiiitaii iMiriliripru^ niitl
tldtonniniiijf (luwi in Hiwh winn Iliut tlu ivriprtHtjil MiuiiMn itlxmlt! In an iil
that tint liiri%'(iin^ runnnlu (ur fl' wiw firrivnl lit,
fill, 1 IIHNH1UU
(;U ~ 7/1 MA/
(,/J  A'} H ^r
h linciu 1 funotion (i, j, 0, ^, (7 t /{ t
where it if* to bo runmrkocl that, in virtuo of Ihw mu
the coofltaumbi of thu form tiro ronlly nrbitrnry: I i
which lad mo to write JDmUQ, 303.
348 A MEMOIR ON THE THEORY OF RECIPROCAL SURF AC EH.
54. Forming the reciprocal equation
0= 2n'(n'2)(ll'24)
(?/?' Hy'If
+ linear function (^ j', ff t ^ G", ]]', i t j, 0, x , 0, .//),
and substituting herein the values which belong to the surface of tho onlor n
singularities, we should have identically
= 2 (  1)= ( n  2) (V
 12) [^l
 2) [D tL ( n _
or dividing the whole by ( n 2), this
0= 2 9
+ 2 (3  4)
o o
'J A MKMtMIt, ON TIIK TIIKOllY (IK ltK< 'N'HlMIA I, Ml) IlKACIW.
vix. M 1(! nqimf.imiH am niul vcrl.imlly tlmviiwnnk Tin I'usl,, HIT ..... I, ami Miinl <' 4 iml,inns,
" ml ''' ln H1 "" "'' ''I* 1 I'unHli MIII! liff,h, nil ;i vu tlm M umi! mlal.Hm, Kl*.! .'I/I /..();
MICTM urn (iuiim'ipn'iil.ly, iunhinivn ul' (Jiui, live iii<lit]Huiiliinl, mltil.iohH. My innibininif (Jin
cqimliioiiH HO IIH l.o Minipliiy Ihr nunibn'H, 1 liinl lluMc in Im
ivl >li^\>(! m) r(),
7 .i /M.^A; . UA'I. ;w. 47*1 o,
Jii/l /M.,M/> . N/y  i;.:^ MI
5(5, I fniiiul, us pivHrnMv t.i.'til,inu,'.l, A 111), // iM'. (! (..!.; vuliir^ wli
(an ,]uiy slmnld tin) MH mrntiil ii{iiiLtinii ; ami MMMI ii^iiniiii^ D IK) :unl
wn htivo A':^. f/ ..'HM, 7/..., a ., Hi / ,... HIS; HIM) ).).,. InrnniUi  H
MiiH'iir Iniu'liun (/, j, D, y, (', H. *",/", /)', y*. r",
lit. piiiiui.HH mil, (iiitiblinif lhc ili<h>niiiiuvti ..... if the ciiil1iiMt>nls ui' Mu< lin.iU' li
ill,. ^.ui. m  iht<..r.t m rlmt. il' tin,,'
in 'it '
'liWH / which IH ...upln MI. /i. /^Mipl .,n . ;[ 7 4n..* <tu ,.. Ihh fit., numb, , ,.*
Mm poiul.H ui' inti'iHt'nt.iiiii nf tin tlivcn BiirftMttH in
'p  HI {/7/A ') vat f  n
Apply l.luM tn thu ta.sn ..!' M Miirliin nl' lli, nnl.r M with a nmlnl riinv ,,f
ninl clnsn r/, inirmMin^ ibt' ,l!c^nii tiiul D.rnnilui HiirliinH, vv Jmvn
Onlir. PiiH.iiiK tltmiixli (ft, y), li
'] H .(,
H 2t 1 1
number of inUM'Ki.TLiuim in
ihnt JM
350 A MEMOIR ON THE THEORY OJ? RECIPROCAL SUKPAHI'M. [4 1 1
and the value of /3' is one half of .this,
= 2 (  2) (1 1,,  24)  (110  272) 4 I g.
I have not succeeded in applying the like considerations to the cunpidnl oui'vo.
08. As regards the general theorem, know (Salmon, ,,. 274) that if two Hurli,*
of tho orders ft partially mtersect in a curve of the ordor m nml ohm r
besides , a cm , ve of tho m , der m , ota .
.
of the order ,; the surfaces vt "1 ' ^ Jmglm tl '''" *.,
residual eurve rf the ote /:1 ,/" , T ""' " *""'"' '""' '" "
points, and thence the Lf JL. JT in fll "' IU1 " ^ '" " ( "" ..... """>
M a si,np,e eurve on each of t L /T "tl? M '" 'T '"" Hi """
iOr] points, W hence ' " """' ..... " ""^ "'

in a residual curve of the order ^ fl "'M"" "' tll( ' flllm> "' ^ '''.
,,^) points ;w henee the fee 8 : f ;r!; ^ '"? " 11!Uta t!l " "'" P "
to curve being a _ t k ^ ^ "^ moo, m p(la  mafi) , f ^ , .....
*B enrve fl times and in a residua cu of tl ,' '^ C " "" >'' "'' ' ..... "'
surfaces meet in mve ' tl10 I' p  0m, whoimu !;!, ,, Ivn
^(^/3m) a [ m(( , + /3/ ,_ 2/3 ,
pointoj and equating the two values, we have
Lastly, if the curve m be
nd in a re
of
. a
of
thl , 00
the
value of ^' by
 (life 303)
A MKMMIH ON TIIM TllKOIt.Y OK ItKOI J'itOOA L HHIt I'ACKS. IJIi 
mill (nliHcrvinjr Unit, llir Tallin uf SiiitfiiluriU'H in my Memoir tin <!iil>i< Surliinw writ
iihluini'il wilJioul, (Jin nitl of lln< fitrmulit mm in iiiuMl.inu) I unili'iivuiir hy iiii'iuis f Mm
rcHiiltH Mmivin riniLiLiiiril In litnl Mir viilnt'M id' Mm miluimvM roHHrirnlM ft, if, a; \, /i, n,
A', ,</', .'<'', A.', /i', i'.
(H). Km a rnliin mirJiuc n<'.\, unil I'nr n nuliin titnI'm.n willmul. Mimjulru' linrs (in
litnl, liir nil Miti I'uscH cxiupt, ,h' milm; Nrrolln XXII inn! X X 1 11 ), tlin liinuiilii in
Tahli', and rHrnvil ,n hy ,]u< utlixi'il ruiimn nmiihci*, Ilir
r,.,.. A.
;U. i/ Hi,/,
ftl.2A V,
iM am
(I)
(ID
(!!!)
(IV)
(VJ)
(VIII)
(IX)
(XIII)
(XVI)
(XVII)
(I ^ 51.  h  '.'</ > V
(I i,". ". ,;.... ,'J/',
vhii'li uvc nil HiUJHlini if nnly
A 2I.
// h Mil': :,:;((;
// . if' ra 1 M.
01. .If WP upply tit Mm name MinfimcH lint rn^ipi'iu'itl i*iiiuiinit I'nr /#, in, what, IM
I.HI MUIHI thing, iipply ihu uriginal rtinttitin to thi niipitMinl mii'liir's, IIM ^ivcit hy
, wi* IKIVH
Mm uppMV ami Itiwrr liuU'tM uf I In* Tttliln uf Si
r Hnvti'fi uf t'liuiiMuiiN, vi/. thin in
HA' +
10i* I 7
0*.117Kfi~n75(i2VX,
GUI 7 Ot r iH4 A'~f/'
0^ afi4 U522..W3X.
OM. 3024 81442^101*
Om 14.M8 1385 2fc'/ 
518 504~4A'6X.
0^ 383 322 A'2/
0 54 So %/,
(1)
 r ', (li)
IMS, (1U)
!>, (IV)
45. (VI)
12. fVirn
120. (IX)
47, (XUT)
1K (XVI)
01, (XVH)
H (XXI)
352 A MEMOIR ON THE THEORY OP .RECIPROCAL
all satisfied if only
II
hence if 
^
c/' + ty = 88,
9 +</' 18,
* = 1.
I remark however that the cubic scroll XXII or XXIII gives
= 54 (330 272) 2 (M V),
^'
a
' ""
IWH tho
f
section
numerical contration.
G3. Combining the two sets of results, we find
A = 24,
ff =</>
(0 =(C
""
f iul,,
n(; (,h,
h' = 5,
and the formula thus
IS
 272)6
A MIUI01K (IN TIIK TIIKiWV OP HKdil'lUKJAL M
f.<> (Inl.iinmiH! I Iii'iii, I'm rhc iv/i>^m ! Imt ,',, ,,n ,1 ,
j ... ' , , ""; llm1 ' l " nl! "I I'linin wii luivii
' . #' ' *V ' " : " I1H '" f l IN a v.'i;y ivniarhiibln rrliU.tnn,
"I wlimh 1 d.i init jii'ivrivr any .? ftn'tiri rcii.sou.
I. inn
a r>:t
() ,,...
' h ;, ; u 
wriUoII in lh) ,
vain. ..!'//,
nn.l
>n nil 2
Nr.. tin (u (IS. IttWl
i!mi WM lmv (  l,,l,w,vn
V /'../'I ', A. i', rr.
./', ,,'./; ;'. A'. ,.'. (r ',
", Vl/. Illl'f!!' illV
" ; (',
"' 11(11 I) 'J/. He,
A 1 ' " tilt (;i  i!) llfi Ni: t
(n .
(ii  8
fl)
ii)
7 f. fl.
C, VI,
45
354
A MEMOIR ON THE THEORY OP BEOIPHOOAL f
a' ('  2) ('  3) = 2 (S'  C") + 3 (oV  So'  tf) + 2 ('/,' _ g^' _ /),
together with the equations for /? and /9'.
66. The symbols signify as follows; viz.
n, order of the surface.
a, order of the tangent cone drawn from any point to tho ma'Auai,
S, number of nodal edges of the cone.
*, number of its cuspidal edges.
b, order of nodal curve.
k, number of its apparent double points.
t , number of its triple points.
q> its class.
G, order of cuspidal curve.
A, number of its apparent double points.
r, its class.
ere cuspidal oiim is mot
number of closepoints.
f
atatiolmiy
ou
, nnmbe, of i n ts ra eotio lls , statiollaly points
number of cnicnodes.
class of the smfaoe.
of f
41 '
A MUMCIIH UN Til 13 TI1I*WV O' HKOI I'ltOOAl, HUltRAOKH.
3 fir,
tf', lllltlllllT nf il,H illlli'XMHIN.
//, nliiHH (if )imi'"i'nup]i' j.nrs
//, limilliiT (if i(.fi nppiivt'lil.
/,', tminlii'i' M' JIM Uiplr piru
f/, IH unli'i 1 ,
//, dl'tliT n' Ilinll'l'mipli'
('', I'lllHH ti' HJtlHHM
fi', nuitilirr nf iti* ii
ll. ul. lr panr.,.
0'. imitiltrr ul' MI'IJMIII,
ami Hpinniln ti.rnrH, MUHinnurv
<'M'nii,ln ,,nrtO.
/.f, iniitiltiT ..I 1 riminion
'l iJlP HjiilMllIi' (pi
7', uuiiiW of rnMininii piitii.H, siitlimmrv plmu N , ( " ij..t<'M'nii,
i". iiiuiiliiT !' t ,i)iiiMfii pliuii^ lint, Nliitiiiimrv plumM nf .. idler
//', imiiili.T iij' ItitrnpfM .(' Mtii'1'in'i'.
('', iniiiiliri' dl' \l* inirfrnos.
(IK. It. is \ wri \\ y n.r,K,mv l .v.Nili Mml lt Hl j .....  (1 plan,. \ tt H tannin, planr
m.'.hn K Mi,. HUfin.i. ,u a nnv, Imvi.,^ m, ,,, ,, t i Hl . nf , ...... art , u H,,!,,,.,!,, Ml . IMM1 . ,,),
Iho ,ii,,.... niiv... An,! Hiniliirlv i* u..i!,,.ifnii,l,, plui..< in u tlniili
plum, nr plum* iiMnliux ...... irliicn in a nirv, !mvm K two ..... ,. K; ,, ( . Hlw lnp nf
tl.p ]ilnii,.!i IN thn M.U.Mnpl,. t( ,, w . ( lm , ,],  I(I . MH , (f lh(1 lujtH (( . (1((Im . ti ..... ni]( ^
'oupln inirvr; i,lm olh.r I.n.iM mud.' .,* tt f ,, n , nil Mxlui.,,,1 j,, th,, ,,..,(; A,..'
AlllU'iHtX, Aayttrt M, IH(HI.
AH in Uu* (.hiirv ' 1'iirvrn. . in ihut nf .S.iHlu.^, l.lt.iv r< rttitrtiii I'm
Lha tiwUT, 1'lr.KH, &,. uitil Hin^iilniiiit'K wliicli hnvw thn wunic vallum in Mm orh
Uw ivfipnwnl xiiroM n^piutivcly ; for cuiivoniiMirn I r.p^Hmt. any muh Mr
nuwiM f ilu. Hynili.,1 v vi& ^ ,,./,,..., ^ v , K , 1u(tH ( , a . ((1 nnii . linl , (((
JH otjiutt l tlm Huind liiiicrii,u (N'. ', &',.) nf tint uuiuMitod litttm Hy wh.it
WO IltlVb mV. (ll , d j t jr( IWWHVep t ., Mr Ujlll ahy fniu . it(Ii (jf th(( nnu( ,. ( , nt(l
whioh IH 0, or which in .tjtml t Hynmun.runl niuotinn nf miy nf fclu mcfi
minoutiiitucl Ictiow, ..r lu n funciinn of a. i S ; for itiHUuioo, frutri tlm mpu
No. .1 wo Irnvg :i''HNi', nml thuno 3 o*9u'.# '. tlmt iXflnc
45
356
A MEMOIR ON THE THKOKY OF JlKOiJ.MtOCAL HUUIMOMH.
and from one of the equations of No. 11 we have n 2(7 4#/e  o li/'^y;. :n\')i' <!,
we have thus the system of eight equations,
V
a (n  2)  K + B  p  20
b (n  2)  p  2/9  3 7  34
o(n2)2ff4 7
, v
Y
, V
. V
.. V
2^2/3 + /3 + St H j
3r+ c So S4
or if from these we eliminate , ^ o, then the system ul' (ivo n<iui,i
<.nN,
V
2)
ing Memoir:
"H ,,mi,l,,,. ....... ,
fl , m a
to
and thence
of
l+4W .,
_ M) 6 + (12
preceding
, cttei ,, W() fll)ta , ,
and mldin^ lit'ivl.u rh
wo linvr
or, wluit, iw I)MI name tiling,
il' fur
q,...^ . 10,,.
) . it . ,, , . o ( ,. Wn ,.
i ,s/; aq/ .. , ;IOV ... as,,
i  Wj n.s A 
iH win, M,. viu,. ni'/r, NH,
)^ V' 1 " l&;~ia# ..... V?
4)
11:17
4 1 1*2//
I itit link nuuinpl In v.nlj ihiN ,ti]itiai..ti, l.uf ] will purliully vnrily u nHi.lt
ilcMluuiblu from it,; vix. if*' i tlm likr iumlim, ,,r Llm a.rcntu*! kttum, tliun wo liuvo
<r> *!/ n n' t
II (4.^  4* 40) i + 8y 4 1 2<ty + 2247? + 306*  tf
whcr
358
A MIOIB ON
and IT is the Hke function of the accented letter, And this being so,
TO should h ave
 48 C  48.3
or, as this may be written,
13on  48c 48/9  13y + H = S.
We have
26ftlS C +0i7j 8^ + ^40
and multiplying by 4 and adding, the equation to be verified i
1)to C 8)18(^ + r ) + n + + 28
But we have from the Memoir
s
which reduces the equation to
n + iia^ 1  46^ + 110 300 648 = 2;
or substituting for II its value, this is
that is ( ^^
an equation which is aatiafiecl if
and
^720, or else
41 a
1 1 f~t i
A MKAIulll UN UinW!
'Ynln (I, /V,,Ym, v ,/ (( V,,/ VY,,i,./,W ,./*,/, /,>,// X nn ;, f]f (J/  AHf/(
J'r I*!!". ,.,,. II :(!(}. lt,r,ivv, NHVVM.W IS. inns,: H,, W 1 .
n  ^ ;(j , ,
y U. lH(i.]
.. ,,, ll( , , u Jt n,,.,,^,.,, Hupp.M.H^Iui'y l,u f.lml.
y I il.f.H NliiiiHi. (> , ( , UiMirilHUi,,,, t .fSiirfa.VH ,,r tit, Thinl t)n!,. r iiil'u Kpinii<H
i i,t,iviii.. I., iln, ,,1,.,,,.,, ..... 4il ..... u..' ..I' Sin KM ,ir P.m,^ ,MH Mi,, jvalilv i*f Mtnir
.111., /'A,/. 7V,,.v. v,,l run. (IK (i ; !( , ,,,, )i;i JH. Mm M,,. !,} ..... , (1 f ,, ln " M , IIlltli . is
lll.Mrl.1. I ,b,r K H 1( ! Mll,^! !,,  lt . nlrilliuir (liu^.I) . IMJHIH lihjj Hi! ( t ,' Vuli, V ut' llu
!*'>, Hll.iitliii^ .ri.ly [. i!<r tlivimMh JitfM lim.iityhvM, .u UH I j.ivf' f . I., ivrlimi il,)
v,. t iy Iln... t ,.M,H .IrpM.iir.i? MI. ill., miimv ,,1' i,, Mi^nihirili.^. And I uU ..... I <, (,!.
n.^h.Mt v.iy iitii.'h ..it tir,'..iuir ..I* ll... li K lu . I,, .,l.luiu.'d in M^VMI.' t.. Lttn l.lm.ry
U.'riupul SIII,IO M . 'IV .M..m..ir ,,.(rnv ( l In InniiNlirH in liirl, a slnro nf ittulorialH
I'
lN ,lti.,.n,*. iiii^ii.irl, ,,,H l i;n,, (pllltiully nl" n,i(.,.tl.ly ^t'Vinll'll) tll< t'lim.il>U
lmn'.i ...!. liimfr, ,,r I In H ,vriul r.M^ nf nil.i. Mnrlmps, nr whui in tin niuitt* Iliing
r<iiiulitim in ihiiiii.'1'.t.inliimtt'*. u* i t . Ht'VMial MiflhiTs i.nliM : ........ 1
360
A MEMOIR ON CUBIC SUBFACSES.
[412
WXZ +
WXZ + (X + Z) ( F 3 
fa, 6, c ,
7X a  # = 0,
P + Y (X + Z+ W) +
WXZ+(a, 6, c, d][Z, F) 3 0
IFJ^ + (Z + } f 7  jf ) = 0,
) 6, c, d^X, 7) 3 =0,
0,
 0,
Y 3 = 0,
=o,
IV =1220,,
VI =12 7? f
VII =123 fl ,
VIII =1230',,,
IX = 1'? 213
x =12^0;,
XI =12 A,
XII = 1217,,
XIII = 12_0 3 2a 3j
XIV =12B e (7 a ,
XV =1217
XVI =124<7 2 ,
XVII =1223,^
XVIII =12 A 20,,
YTY 1 9 n r<
.A.J.A. = liJ 2j fl Ujjj
XX =12f7 g ,
XXI =1235 8)
XXII = 3, 5(1, 1),
XXIII = 3,5 (171),
2. Where C' s denotes a conicnode diminishing the class by 2 .# 5 4 J? j9 a
biplanar node diminishing (as the case may be) the class by 3/4, 5, or 0; and
,nu i "^P^^' node diminishing (as the case may be) the class by 6, 7, or 8.
The affixed explanation, which I shall usually retain in connexion with the Roman
number shows therefore in each case what the class is, and also the singularities which
cause the reduction: thus XIII=12 3 2C 2 indicates that there is a, biplanar node,
A dimmishmg the class by 3, and two conicnodes, 6\, each diminishing the class
by 2; and thlls that fche lass is 12 _ 3 _ 2 . 2j =5 _ Ag ^ case / xxn ^
XXII, these are surfaces having a nodal right line, and are consequently scrolls, each
of the lass 3 vi, XXII is the scroll 5(1, 1) having a simple directrix right ,i no
disinc from the nodal line, and XXIII is the scroll 5(0) having a S imple directrix
righ, hne coincient_with the nodal line: see a, to this my "Second tt i
Skew Surfaces, otherwise Scrolls," Phil. Trans, vol. CLIV. (1864), pp. 559577,
3. The nature of the points tf a , tl S tt S t , S 6 , U 6> U r> U 6 r
C'(=a 2 ) is a conicnode, where, instead of the tangent
quadnc cone.
on
^
a proper
412] A MEMOIR ON CUBIC SURFACES. 361
In 7? 4 , the tangent plane is distinct from each of the biplanes;
In B s , the tangent plane coincides with one of the biplanes; we have thus an
ordinary biplane, and a torsal biplane :
In B u , the tangent plane coinciding with one of the biplanes becomes oscular; we
have thus an ordinary biplane, and an oscular biplane.
U (= V a , U 7 or UH) is a uniplanarnode, where the qimdric cone becomes a coincident
pianopair ; say, the plane is the uniplane. It is to be observed that there is not in
this case any edge. Tho uniplane meets the cubic surface in three lines, or say " rays,"
passing through tho uniplanarnode, viz.
In f/n, the rays are three distinct lines ;
In U 7t two of them coincide:
In t/o, they all tbreo coincide.
4. To connect these singular points with the theory of the preceding Memoir, it
in to bo observed that they are respectively equivalent to a certain number of the
cmicnodoH (7((7 a ) and binodos B ( B s ), viz. we have
G = G
B[ = ' B,
B, = 2(7,
B, = G+ B,
(B a = 3(7,
(U t = 3(7,
U 7 = 2(7+ B,
U s = C + 27J.
5 I tako tho opportunity of remarking that although the expressions cnicnode and
binodo properly refer to the simple singularities G and B, yet as 0,= 0, 0, is properly
Hpokon of as a onionoOu, and we may (using the term binode as an abbreviation for
biplanarnodo) apeak of any of the singularities B 3 , B tt BB< as a binode. Thus the
auriaco X = 12 #,(7, has a binodo 3< and a cnicnode a a ; although theoretically the
binodo ., in equivalent to two cnicnodes, and the surface belongs to those with three
cnicnod^ or for which (7 = 3, I use also tho expression unode for shortness, instead
of uniplanarnode, to denote any of the singularities U, t U 7 , U 6 .
6 Tho foregoing equations (substantially the same as Schlafli's) are Canonical
forms; tho reduction of the equation of any case of surface to the above form is not
a way obvious. It would appear that each equation is from its simplify in the foi n
be t adapted to the separate discussion of the surface to which it belongs; there is the
Si^Lg. that Aquations do.t *^(^^*^ <^
the Hurfacea they ought to do so) lead the one to * . F2 _ ftZa _^ )==0
^jLiw^rsrTKS.S'.^iK.: ^
form a tueory or tne qutii
the geometrical theory of the surfaces to
means i"j n c*iu, " _ . .
they respectively belong, and the imperfection is not material ^
0. VI.
362 A MEMOIR ON CUBIC SURFACES. [412
7. I have used the capital letters (X, Y t Z, W) in place of Sohltifli's (m t y, z t w) t
reserving these in place of his (p, q, r, s) for planecoordinates of the cubic surfaces,
or (what is the same thing) pointcoordinates of the reciprocal surfaces; but I havo
in several cases interchanged the coordinates (X, 7, Z, W) so that thoy do not in
this order ^ correspond to Schlafli's (at, y, z, w) : this lias been done so as to obtain a
greater uniformity in the representation of the surfaces. To explain this, let A, ./?, 0, D
be the vertices of the tetrahedron formed by the coordinate planes A = YZW, B=%WX.
G=WXY, D = XYZ\ the coordinate planes have been chosen so that dotcrmmato
vertices of the tetrahedron shall correspond to determinate singularities of the surface.
8. Consider first the surfaces which have no nodes .B or U. It is clear that;
the nodes O a might have been taken at any vertices whatever of the tetrahedron
they_are taken thus; there is always a node G, at D; when there is a second node 0,'
this is at 0, the third one is at A, and the fourth at B. *'
9. Consider next the surfaces which have a binode , #, or # (j . t h is ia
taken to be at D, and the biplanes to be X=0, <)(') (the edge being theruforu
JJB) viz. m 5. or B ti where the distinction arises, Z = is the ordinary biplane
^O the torsal or (as the case may be) oscular biplane. If there is a second nodo'
th ls of necessity lies in an ordinary biplane; it may be and is takon to bo in tho
biplane A =0, at C, I suppose for a moment that this i s a node G.. It is only
when the hnode is B 3 or S t that there can be a third node, for it is only in those
rihnfnode 61 '! r * "T" ^ ^ ^ ^ in ^ <
the thud node, a 6' 3 , may be and is taken to be in the biplane Z=Q, at A
6ach is a '
; rr. ^ "

O
to e _ There .
f
oot oT f (ex ? fc r the
X=0 being the ordinary binlane 7n tl , 7", ' (f 1 ' a bmod8 ^ or S "
^
X=0, Fo, those oth W blod /T T f ^ 8e OIld Mm)de
XII = 12ff.) the iraiplane is Tn TP , ' and that (exoe P t i
as represented b y to^^z ?%/?$* ^ ( /VH. 12
toing X = ^0, a ^ at , 2,f ^ + ^V0 Fol/ f & r
common biplane), and a C7 a at X. (therefore ^ = the
3 In the case, however, of asineleJ). JTTiQ n n. u ,
tfe^inia*,, thobipla^ are talcen to
412]
A MEMOIR ON CUBIC SURFACES.
363
13. Ifc will be convenient (anticipating the results of the investigations contained
in the present Memoir) to give at once the following Table of Singularities; the
several symbols have of course the significations explained in the former Memoir.
w
a'
5'
<J'
<j'
8" o ; '
<JT <j'
1
ii i . .
M
1 1
i, i "
n
^
ef
o ^
e? ft?
co T ef b u
4
B? ft? tT
^ ft? f
s &
3
""*
rH
I 1
III!
1
i i i
rT
CM
01 CM
CM CM <M CM
CM
<M CM (M
CQ
H
II
rl
U
1
11 II
II II
11 II II II
II
H II II
I
II II
II
II
tl
II
a
u
b ,
S Ei >
H E H
^!
M
R
M
H
H
fc>
p >
t> M M M
tl
M H M
t*i M H
M M
i
M
/
,,
n
H
!)
8
8
3
3
3
8
3
n
(1
l\
G
11
li
(i
G
G
C
(i
4
a
1
2
1
8
2
4
1
u
f)
t)
7
fl
7
(J
8
7
fi
8
y
3
1
b
k
t
a
u
2
3
f.
h
r
a
&
X.
I)
y
i
1
1
2
1
1
3
2
2
1
4
1
2
8
B
I
II
III
IV V
VI VII
VIII X XI XI
IX
XIII XIV XV
XVI XVIII XIX
XVII XX
XXI
XXII XXIII
r
10
(I
9
u
li
y
H
il
y
7
6
n
fi
fi
9
6
y
6
C
9
c
9
4
6
9
3
6
9
8
4
3
n'
a'
8'
' 2
' 21
', d
', ' 2
1C
(10
IB
IB
y
18
u
y
7
12
b
7
1
i)
8
t)
9
1
3
3
1
1
1
8
1
3'
G
1
k'
.
. , .
2
c'
' 2
' 18
' 8
18
90
24
10
72
42
12
88
17
10
24
24
G
G
y
12
32
2
5
o
2
2
h 1
' 1
*
12
12
10
10
9
8
1
fi
16
2
2
tf
K'
' fi
8(
18
18
8
1
P,
y
i'
'
'
1
'
1
. ,
1
3
B
364 A MEMOIR ON CUBIC SURFACES. [412
Article Nos. 14 to 19. Explanation in regard to the Determination of the Number of
certain Singularities.
14. In the several cases I to XXI, we 'have a cubic surface (n = 3), with singular
points G and but without singular Hues. The section by an arbitrary plane is thus
a curve, order ?i = 3, that is, a cubic curve, without nodes or cusps, and therefore of
the class tt' = G, having S'=0 double tangents and K '=Q inflexions. The tangent cone
with an arbitrary point as vertex is a cone of the order a =6, having in the caso
1 = 12, S = nodal lines and *6 cuspidal lines, but with (in the several other cases)
G nodal lines and B cuspidal lines (or rather singular lines tantamount to G doublo
lines and B cuspidal lines): the class of the cone, or order of the reciprocal surface
is thus '=6.52(0!<7}3(G + j5)=1223a.
15. In the general case 1=12, there are on the cubic surface 27 lines, lying by
3's in 45 planes; these 27 lines constitute the nodecouplo curve of the order
/>' = 27, and the nodecouple torse consists of the pencils of planes through these linos
respectively being thus of the class ?' = &' = 27; the 45 planes are triplo tangent
planes of the nodecouple torse, which has thus *' = 45 triple tangent pianos. But in
the other cases it is only certain of the 27 lines, say the "facultativo lines" (as will
be explained), which constitute the nodecouple curve of the order p' the pencils of
planes through these lines constitute the nodecouple torso of the class &' = // the t'
planes, each containing three facultative lines, are the triple tangent pianos 'of the
nodecouple torse. Or if (as is somewhat more convenient) we refer the numbers
r, ? Vi e ^i SUrfMfll then the lines ' *al8 of the fucultativo lines,
constrtute the noda curve of the order V; and the points *, each containing three
of these lines are the triple pomts of the nodal curve, Inasmuch as tho nodal curve
foZa W me V he T mber *. of its apparont doilble >^ * fcy *
fonnula 2fc =6J e^; and comparing with the formula a> = b'*b' 2// V fit'
we have g + 3 7 ' = 0, that is, ? '= (,' the class of the nodal curve), and also y.Q.
the OTbb > pn e CU1 ' Ve S tG C0m ' 3lot "^^ootion of
the cubic smface by the Hessmn surface of the order 4, and it is thus of the order
~ ; , ther CaS6S the m P lete mterseotion connate of the
, reoon connate o the
ogether w:th certain right lines not belonging to the curve, and tho
wiu be
curvo
are common tangent planes of the spinode tors and n
planes of the spincde torse; or we have%  L  T ^^ """"^
instead of the 27 lines we must lakemilv the\~ t V' r * GS ' howew '
is not a double or a single ^^ ^^M^^ ^ f hioh J9
of the surface at the points of contact are th, V ^ ** ta ^ nt P Unea
that is, the number of c ntacts give 7 O t ^ gte~
412] A MEMOIR ON CUBIC SURFACES. 365
18. There are not, except as above, any common tangent planes of the two torses,
that is, not only y'~Q as already mentioned, but also i' = Q. I do not at present
account d priori for the values ff = lQ, S, and 16, which present themselves in the
Table. The cubic surface cannot have a plane of conic contact, and we have thus in
every case G' = ; but the value of B' is not in every case =0.
19. In what precedes we see how a discussion of the equation of the cubic
surface should in the several cases respectively lead to the values b',t' t p', a', I3',j' } %, B',
and how in the reciprocal surface the nodal curve of the order b' is known by means
of the facultative lines of the original cubic surface. The cuspidal curve <?' might
alno bo obtained as the reciprocal of the spinodetorse ; but this would in general be
a laborious process, and it is the less necessary, inasmuch as the equation of the
reciprocal Hurface is in each ease obtained in a form putting in evidence the
cuspidal curve.
Article Nos, 20 to 28. The Lines and Planes of a Gubio Surface; Facultative Lines;
Explanation of Ditif/mms.
20. In liho general surface I = 12, we have 27 lines and 45 triple tangent planes,
or say simply, planes: through each line pass 5 planes, in each plane lie 3 lines. For
the mirftiooH II to XXI (the present considerations do not of course apply to the
Scrolls) Hovoral of the linos come to coincide with each other, and several of the
pianos also come to coincide with each other; but the number of the lines is always
reckoned HB 27, and that of the planes as 45. If we attend to the distinct lines
and Urn distinct pianos, each line has a multiplicity, and the sum of these is =27;
and HO oach plane has a multiplicity, and the sum of these is = 45. Again, attending
to a particular line in a particular plane, the line has a frequency 1, 2, or 3, that is,
it represents 1, 2, or 3 of the 3 lines in the plane (this is in fact the distinction
of a scrolar, toiml, or oscular lino); and similarly, the plane has a frequency 1, 2, 3, 4,
or 5, according to the number which it represents of the 5 planes through the line.
It requires only a little consideration to perceive that the multiplicity of the plane
into its frequency in rogard to the line is equal to the multiplicity of the line into
its frequency in regard to the plane. Observe, further, that if M be the multiplicity
of tho plane, then, considering it in regard to the lines contained therein, we get the
products (Jlf, M, M.}, (2Jf, M), or 3j, according as the three lines are or are not
distinct, but that tho sum of the products is always = 2M, and that in regard to all
the pianos the total sum is 3x45, =135. And so if Jf be the multiplicity of the
lino, then, considering it in regard to the planes which pass through it, we get the
products (M\ M', M', M', M'), (2Jlf, M', M', M'), ...(6Jf'), as the case may be but that
the Hum of the products is fiJf , and that in regard to all the lines the sum is
5 x 27, = 135, as before.
21 Tho mode of coincidence of the lines and planes, and the several distinct
linea and planes which are situate in or pass through the several distinct planes and
lines respectively, are shown in the annexed diagrams I to XXIC): the multrphcrty
i See the commencements of the several sections.
36G A MEMOIR ON CUBIC SURFACES. [412
of each line appears by the upper marginal line, and that of each piano by the
lefthand marginal column (thus in diagram I, 27 x 1 = 27 and 45 x 1 = 45, 1 is the
multiplicity of each line, and it is also the multiplicity of each plane); the frequencies
of a line and plane in regard to each other appear by the dots in the square
opposite to the line and plane in question, these being read, for the frequency of the
line vertically, and for the frequency of the plane horizontally ; thus " ! indicates that
the frequency of the line is =3, and the frequency of the plane is = 2. There
should be and are in every line of the diagram 3 dots, and in every column of tho
diagram o dots (a symbol ' '. being read as just explained, 2 dote in. the lino,
3 dots in the column).
22. For the surface 1=12, there is of course no distinction between tho linos,
but these form only a single class, and the like for the pianos; but for tho other
surfaces the lines and planes form separate classes, as shown in the diagrams by tho
lower marginal explanation of the lines, and the righthand marginal explanation of
the planes, I use here and elsewhere "ray" to denote a line passing through a
single node; "axis" to denote a line joining two nodes; "edge" (as above) to donofco
the edge of a binode; any other line is a "mere line." An axis is always torsal or
oscular; when it is torsal, the plane touching along the axis contains a third lino
which is the " transversal" of such axis; but a transversal may be a mere lino, a ray,
or an axis; in the case XVI = 124a 2 , each transversal is a transversal in regard to
two axes.
23. In the general case 1 = 12, each of the 27 lines is, as already mentioned,
part of the nodecouple curve; and the nodecouple curve is made up of tho 27 linen
and is thus a curve of the order 27. In fact each plane through a lino moots tho
cubic surface in this line, and in a conic; the line and conic meet in two points
and the plane (that is in any plane) through the line is thus a double tangent
plane touching the surface at the two points in question; the locus of tho points of
contact, that is the line itself, is thus part of the nodecouple curve. But in the
other cases II to XXI, certain of the lines do not belong to the iiodooouplo curve
(this will be examined in detail in the several cases respectively); but I wish to
show here how in a general way a line passing through a node, say a nodal ray, fe
not part of the nodecouple curve. To fix the ideas, consider the surface II =120,
there are here through G 2 six lines, or say rays: attending to any one of these, a
plane through the ray meets the surface in the ray itself and in a conic; the ray
and the come meet as before in two points, one of them being tho point C 3 : the
7 a 7 t6 tei> P int ' ** * d GS not M * face at 0..
node h Pen qUeSti n ' Wh6ther Ol ^ ht *> *t at a
n d 1
rh C ' *>* "ra7
through C a is not a tangent plane.) The plane through the ray is only a
r pl r ; and the * * ^ Ai
^  e" acoordin,
412 ] A MEMOIR ON CUBIC SURFACES.
Article Nos. 24 to 26. Axis ; the different lands thereof.
24 A line joining two nodes is an axis; such a line is always a line, and it
is a torsal or ocular line, of the surface. But _Bome furtherdistinctions are ^qu.sito
using the expressions in their strict sense, cmcnode = 0, bmode = * an ax, < a
COaxis joining two cnicnodes, or it is a OBaxis jommg a omcnocle and a bmode
or it is a%B axis joining two bmode.. A Offaxis is tonal, the transversal bomg a
14 L not a 4 through either of the encodes; a CBaxis is torsa^ the
EanS bg a ray of tL binode; a BB axis is oscular The *^
course carried through as regards the higher biplanar nodes ft, ft, ft, ^ ^
uniplanar nodes 17,. 17,, 17.: t!n,s (ft B) the edge of a bmode B not " s ^
nil but (ft = 20) the edge of a bmode ft is a OCaxis ; (B ? = .B+0) the edge ol a
all, ov>< OH axis (A = 30) the edge of a binode ft is a thrieetaken COax I3 ;
S?^. r of ^ ra^l regldod a 3 S a 0,axis ; (J7.H + .J, the ^b.e ray
regarded as a twicetaken CTaxis, and the single ray as a CCaxm; (U, = ^ + G)
the ray is regarded as a BBaxig + a twicetaken CBaxis,
25 It has been mentioned that the intersection of the surface with the tfarian
consists of the spinecle curve, together with certain right linos; these hues are m act
tlTaxesvis. the examination of the several cases shows that m the complete
^1 each 00axis presents itself twice, each OBia 8 fees, and each ^ax,
ftoes. We thus sec that a 00axis, or rather the tonnl plane along such ax.s, is
he pLhplane or singularity / = 1 ; the OBaxis, or rather the torsal pane a ong such
a s,7e closeplane of singularity *l; and the .BBaxis, or oscular^plane along s, eh
axis the bitrope or singularity B"  1 ; for a cubic surface TO tl. Bmgular hues Ihe
cxlsson *> being ta fact \> = 12  %" *  '. There are however aome casen
eS explanation f thus fer the case VIII = 12 ft, where the edge IB by what
Zedes a CTaxis, the complete intersection is made up of the edge 4 time and of
!n cctic curve; the consideration of the reciprocal surface shows, however, tha the
Sge aken once is really part of the spinode curve (vi, that tin. curve is modo up
of the edge taken once and of the oetic curve, lts order being thus , =B) and he
interpretation then of course is that the intersection u. made up of the edge taken
3 times (as for a OBaxis it ehould be) and of the spmode curve.
368 A MEMOIR ON CUBIC SURFACES. [412
Article Nos. 27 to 32, On the Detemination of the Reciprocal Equation.
27. Consider in general the cubic surface (*$X, Y, 2, W) S =Q, and in connexion
therewith the equation Xx + Yy + Zz + If w = 0, which regarding therein A r , F, #, TT
as current coordinates, and as, y, z, w as constants, is the equation of a piano. If
from the two equations we eliminate one of the coordinates, for instance W, wo obtain
($Zw, Yw, Zto, (X(s+Yy + Zz))* = Q t
which, (X, Y, Z) being current coordinates, is obviously the equation of the oono, vortex
(X = Q, F=0, #=0), which stands on the section of the cubic surface by the piano.
Equating to zero the discriminant of this function in regard to (X, Y, Z) t we express
that the cone has a nodal line 5 that is, that the section has a node, or, 'what in the
same thing, that the plane atf + y7+.e# + wTF=:0 is a tangent plane of tho cubic
surface; and we thus by the process in fact obtain the equation of tho cubic mirm.ce
in the reciprocal or plane coordinates (, y, z, w ). Consider in the same aquation
*, y> z, w as current coordinates, (X t Y, Z) as given parameters, the equation represents
a system of three planes, viz. these are the planes aX + yY+*Z + wW' = whoro W
has the three values given by the equation (^X, Y, Z, F') 3 = 0, ol . what / H tho samo
thing, X, l,Z,W are the coordinates of any one of the three points of interaction
of the cubic surface by the line  = f = ; (Z. F, *, W) belongs to a point on tho
surface, and
is the polar plane of this point in regard to a quadric surface Z +?> + #+ r0
'
the equation
(*jZw, 7w, Zw t
is thus the equation of a syatem of 8 planes, the polar planes of three point* of tho
cubic surface (which three points lie on an arbitrary line through tho J point i0
= = eqUatlUg t0 Z61 ' the disci ' imin ^ in regard to (X, Y t g)\ Q find tho
 ^ ..L.tv.iu j.,, my m 1.1 iiu I A . t io I \
envelope f the system of three plane,, m say O f a plane, the po ,,r pi
arbitrary point on the cubic surface,or wo have the equation of the
Z 6 ; mg ,' " ", Wni the Snme tWng " S the D[ l uatio " of Wo
wie reciprocal or plane coordinates (x it ?/A T 1
412] A MEMOIR ON CUBIC SURFACES. 369
29. But let the parameters, any 0, </>, be regarded as varying successively; if $
alone vary, wo have on the surface a curve , the equation whereof contains the
parameter 8, and when varies this curve sweeps over the surface. The envelope in
regard to c/> of the polar plane of a point of the surface is a torse, the reciprocal
of the curve , and the envelope of the torse is the reciprocal surface. In particular
the curve may "be the plane section by any plane through a fixed line, say, by
the plane P 0Q = 0; the section is a cubic curve, the reciprocal is a sextic cone
having its vertex in a fixed line (the reciprocal of the line P 0, Q = 0), and the
reciprocal surface is thus obtained as the envelope of this cone; assuming that the
equation of bho scxtic cone has been obtained, this is an equation of a certain order
in the parameter 6 ; or writing 8 = P : Q, we obtain the equation of the reciprocal
surface by equating to zero the discriminant of a Unary function of (P, Q).
80. With a variation, this process is a convenient one for obtaining the reciprocal
of a cubic surface; we take the fixed line to be one of the lines on the cubic
surface; the curve is then a conic, its reciprocal is a quadricone, and the envelope
of this qimdricone is the required reciprocal surface, This is really what Schlafli does
(but the process is not explained) in the several instances in which he obtains the
equation of the reciprocal surface by means of a binary function. I remark that it
would bo very instructive, for each case of surface, to take the variable plane
Hucoonaivoly through the several kinds of lines on the particular surface; the equation
of the reciprocal surface would thus be obtained under different forms, putting in
evidence the relation to the reciprocal surface of the fixed line made use of. But
thin is an investigation which I do not enter upon: I adopt in each case Schlafli's
process, without explanation, and merely write down the ternary or (as the case may be)
binary function by means of which the equation of the reciprocal surface is obtained.
H('l't)H
n
H2. Tim
clwrnimmuil, ul' a
itmti<!,
whit'h
r n /,, ,.
nMiil,inn li,.in w iJ
u, WO Hhlll
H of M
y ur a liinuiy !iinrli.,in'). vi., tlii<. i> <
., iv,n in ii )r . ).,
'
) yHiM,V. utnl wlii.li i, I!UM,,U ,nf ,,
jv.lnn.il h. llu pr..,,,! m.l,.. Th,, 1: fl t .
in nvid.^ii.n (in ni.pi.hl mn, .,) r),.
),;,,, (In* I..HM i.lti.mi.. ,, ,\'' /' n
'
y (N II. 7' .11,, ,.,, ,, ,1... ,, !M ,, V j,. ,/ /
'
'I tin ,v ..... (,,, .., (IM I)t ......, l4 ,,.. 4 l,!, ,.
HH ,nrvu m n ,:,(, ,,,,,,, , ,1,,. ,,,,,!,,,, ,,,...., ,;..,, i (ll , (l . Ml(i , t
'" v,,,,  .....  ' in, !!,,,, fl ,,< .1... ,,!, 1 ' l 
"I* u iJn'iiwti mil, liii'tur.
AH
, notation ,,,H
l In Homo onv MUM, for
v ft fllroot oll

mi,.. W 1 ,, XXIII ,' r.. I 
v
""" "' ''
A MKMOIU ON UIMIK! f
371
"I" lln i cipiiilinii M' tin nripiwiil mi Hunt iJirnn miiilylioul uxpniHwiim liuvn no
Hi'iil. !ipili'nli"ii, Ami mi in HHUIM ul' ,lm ut>xt liillnwin^ minimus, no applinil'iun IH
' '
liWili' nl ill' 1 utiillyl irill i'\p)'rs>iinll!i ol' l.hn lilh'hl lUlll
Ilk 1 I'ull I" wiutl llml il' a HUM lu< j;iwn ui Uin inUii'mMitiwi nl' Uin two plains
Uii'ii I In' i'K I'nni'iliimli'.'i nl' I he lint im
i, /i, r, f, if. It
AD' AD, Itlt' li'D t CD' H'D, IiC' fl'C, VA'U'A, All'" A' It,
ami llml. in l'iim "I it> n\\ cnimljtmli'i Uu< linn is i{i\c\\ JIM (Jut cuiniiion inrnr.siol.ioti
M' .)ii' luiir phtiir:i
( . A. //. H O.Y, i'. /, ii'i^o,
'A. , f, /
.*' ' / (I
',1. I, !. .
iiixl I lnil Uni'rM.'!iliiif; mi iisital in n'i;.ml In A' J I i' 1J I ^f ! IP  tl) (ho rnniiliiiiitrM of
llti* H 4 rijir..fu liii* HI. ( /', /, A. ii, ii, r} ; lliut IN, ihi'i in ,hu rnininmi inliU'Hiuiliinn nl' tlu 1
I '', ! /'. f O' 1 '. ,'/. . '") ' '0.
1 (', , H, *I i
! * I
A,  (i, . h s
/. //. A, .1
Jt, IN lit M'tint 1 rii^'^i ni>'i'< niiiV''iiifiit, in funkier 11 linn an ilcliniiinril us Um iiiLui*
NMI'tinn nl .,WM plilll *f Kllll'T I llUll ll.V HHUlll M' its HIX I'tml'lHuall'S j l.llllH, I'nl' illHllllKri',
In hH'iik ( I lif lin.' .Y , )*' niilirr tliuu f iln i liim (0,0,0. 1,0,0); iitnl in
wiiiit* nl* lit*' Mvlinn. 1 liuvr pi'r'uvit hfil. In jjivn tht i'Xpivw*iniiH nl' Uin nix (iimnlini 1 ' 1 "
n' tin* NiVrliil llKt'w,
A MI'IMllIII. ON rlMIH!
it in HO (iomplimLuil dlml> il. can Ininily IIP onwiili'ivil in at nil put I in;; in .'\i<l,n,v
Mio 1'iitaMimn of Mm MUCH ami plain's; Mml. "(' Dr Hurl (Sulnimi, " ' hi Mi Triplr
Tniitfont Plant)* of a Kiu'l'm'n n!' Mm Tliinl Onlir." MI ..... v.limi,, pp, ;!;,:! \\i\o) t
ilnimndiiig on an unuiigiiiM'iil of Uio l!7 Ihnvi iii'rni<liiii I" u iulu' { :i rni'ti \vay, i;(
a Hinuliu'l nh^iinl. IIIIK, anil will In* jni'innMy I'l'iirmiin'ftl,
.'Hi. JUil; Mm nui9l. (innvunii'iil HIM is Si'hliilli's, (iiHrlinif Innt u .Innl'Ir .is,, r ; vi.,.
wo can (mid lliaf. in fid dilVmviil. wnyn) [ii'lcnl. nut n(' the ;!V lin i iiv,. '.v.i'iirf ,w\\
of nix liii(*n, Hiitih Mmt no lwo linrn nl' Mm tiiinm Mynl'iu iiihi !'(, lnl ih.it .t.li Hn,.
dl' tlm ouu Hynliioii infiriw^ln all linl. dim HHTi;)iHinliuff linr ul" lit" .lli'i '.y^t. m ; nr,
Hay, if UKJ liniiH urn
I , li , It , i . .'i , II
r, *.!', ;i', i', .v, i;\
Mmn MiiiHii ltav<! tlm Miirl.y iiidi'i.iriiriMnn
Any two HIIIIH Himh IIH 1, il' !!' in u ilriiif vvhirli limy li<  ;i !!, d 1^',
linrH I', U lin in a plaiii wliirli nniy I'" rullr.l '^ ; ih,. lt . f WM piu
Him 12; nnil any Miivn limM mn^li \\a III, III, ,'iii m ! tit ..i)ii>, Ki
l2,U'l',5(i, \Vn liuvo UiiiH llitt ttiiiifit .'lydft'iu "" il^ %'} in,. > ..i.l i
ollUc.ti (ioniplobily oxpluiimil liy wlial. IIIIM ln'.n t.Ulnl, tntl wluuli i .^liil
tho iliagmin.
, f ,iinil,iilv I In*
n.'i KI* ( m =t
i.i,; in n \i\w
, [.l^tf^, * in
.i^'l in !<!! ut
;!7. Tlm. (liagmni oC Mm lines ami plum
412]
A MEMOItt ON OUBTO SURFACES.
373
Linos.
CJi hi", ill. C OJ OS tO W M K> M M M M M
OV CS t)l CTJ Ol hi* O til hi CW Q3 CJi K W IO O> til if. M K) H O) til rfx W U) M
1OI li
~ll 1
Il
!
X
~l I
Ifl'
.
10
Hi'
.
r
14'
.
z
]fi'
.
y.
1(1'
.
y,
81'
'
e
2H F
.
84'
.
f
ar'
.
Pi
an'
.
V
HI/
.
na*
.
i)
114'
.
B
fi'
.
1
nil'
.
'Ji
U'
.
r
w
.
y
ill'
.
h
, llfi'
> < *
m
1 'J' 1 '
.
m
K fit'
. > >
1'
r.a 1
.
y
r.:i'
IB x 1 ! i
.
i ;
04 '
.
i
fill'
.
i
ill'
. .
PI
IJM'
.
y
III)'
.
r
(14'
. '
n
06'
.
U
18.84.00
I i >
il!
18.8B.40
.
12.00.46
. .
18 . 84 . 60
.
18.85,40
.
1'
lit. 80. 46
.
' l7 l
14 , 28 , 60
* *
K
14 . 26 . 80
i '
h
14 . 2(1 . Hfi
> *
i
16 . 28 . Ill
1 '
<ii
16, 84. BO
.
riii
1.5. 20. 81
i > .
TI
10.28.46
* *
<i
10.8i.86
*
fii
10 . 86 . 84
46 46
.
.L
# jf t s o? * rjrrr** * * fi * ? ? Jf ^jLfjflf . JL.
M74 A MKMimt, ON (IUIIK1 HlMtKACIIM.
!)H. It IIIIH ImiMi mniitiininl t<hul< Mm imnilirr nf ionl'l<"M\<'ni \v't Mil,
us Pollown:
I , i!, ;t , 4, fi , li A:tmiui'il Mimiti\'< I
r *>' ;i' I,' r/ ii'
I i f r f i i i' i " ( f '
I, I', 2!l, 24, 2fi , 2ti liilti' iiiiaiii;''mi'iii i 1 ,'t
l >. '' l:l M. 15 Hi
*J ) M f I *' I * "  * ' * I ' "
1 , 2, II, .1(1, 411, 'l.l l.iUf iiwm.n.'iuriii
2:1. i;t, it:, 4, n, n
wlmro, if wi! I tilt u any itnliiiini j^ ut' IAVH lintvc, \\<> huvr llit< r.ntiil<>t.
pairH nf niiiiiut.ri'Hwtmj,; linn:i (rnrli linn nii ( nli 10 liiirn, iluir tM< fh.'h't'.ti" ; l ,V I III.
; " 1(1, whinll id ilui^ iinl, iiii'i<l, it nil I he iiiiiiiln<i' \' HUM iui>iur<:f iuu j,,,n , i , i  1(  fp
i. 27, Hi r. i 2 Id).
!l!l. Wn rim mil. nf llii> tn jiliim^ Hpti'.'l, uitil llnU in UNI w,i\n. ,i (nliHi.il p.tti 1 ,
thivt IN, l,wo [I'lttdd u!' tluinn t Midi lliui flie il!ini"( ..I din .in hiit'i, iisr.'i 'i".riitjj t)tti"
uf Mm olluu lriuil, KIVK !l "'' Mi*' V limvi. Aniil.vlii'iillv il ,V II. (' H, / ti fiml
//s.^0, I'V' 0, IT; (1 HIM tin* iNiiuli'tii'i nf tlh' ;>i\ jilaii'"', tlicii ihc i ^inuinii ! llir
liii'i! \H XY% \klIV\V il SCM in (M l)ii>t i^.f. ,\',., H,
Tlu> trilii'dnil plrmn jmiin mv :
12', 2H', :U'
N... i
,:iri,2i!. 2:. Ki.ru,
1 2 M x 1 5 (i
.'I I 2 5 I! 4
2 II I li ^ r,
li
14 23 JHi
Mf HI ;H
H IB immaterial how tltu Lwo compiiiuitt tritub ISJI mitl 4^ii an Arranif^l, wi bmin
fttways thu snmo Irihwlral jitiir.
A MMMdlll ON (UMIK! HlIJIKACIKS.
375
Hart, ai
^7 lini'n, rnliiriilly, l.
lhiv l.'i.ii'i'.'i il' ill. :mmi> iilpliuli.'t. ili'iinl.i' linns in lilin Kiutiti piano, if only bho lottos
V ih<' mum* "i lit.' !itilli\i"i Hut uiiiiii'i t'tiw yl,, A. it A a \'w in u piano ^d^i^a;
.. /v ''i I'*' " ' l I 1 '" 1 "' li'W 1.1'll.irn !' ilillVnnil, nlpliabnts tlonotn HnoH whioh moot
TMifliiiu l" th' Tiil'l' 1
V.. i
V, fl,
ttl(l , lv ,,, ,..., iu , t ,, ,,.),,, ,.!' Mi.' H,ii,m. .Ir.mhH a linn lying in bh KIHUO
uiMt It..* In,. .I.'iil".l i'.V Mir 1HI..H ..I* rurh v.r(.i(Mil pail in th<( Hiuno H.pm.u UUIH
,t lirri in Mi iilintr .1,'"^,. *'liM.. ^lt'*.7u ('i" 1 ' > l ' lln lH!l'mnii!iiUiH!(l two iihvnuH
H, / liu.'l liiui MII,' wy i> vvhii1. this may hn id.mtiliml with liho doublosixur
i*( l4i ivjn.^MiiI, Mir ulu'V.' urnt!iK''i"'nt. by
I. X, 13
li 1*1*1 Mli
t r i "* 
V. fi . 'I' 1 " 1
n, 24, fit
r, ui, o
,
nfltitjttid in (720xH(l) 25020 ways, viss.
a i fi '' lii
rh
VIK ukittg
!JJ J.J .!
.to/
.
,.,,Un, r II,, 4,1 pluu,, uru ..bUM^l i.> y Wr Int reforrocl to,
A M mom UN riniH MI ni ,\t i
/A' .) iiiY+n'/, I II'
A'O,
HI
\ ir ti,
'i. n'/,  H" ,
. ,,#.,, jr.
7* f'
/
;(/"") I
ir..
412] A MEMOIR ON CUBIC SURFACES. 377
X I  1 r + nZ + W = 0, [56' = I]
. 7 
m (p  a)
 ' [64'n]
Y + mY + + F = 0, [15  26 . 34 =
a)
nZ + F = 0, [16  24 . 35 = m]
IX +  1  K   7 2  ..... x ^ + TV 0, [14 . 25 . 36  n]
u ^ ?/i m (^J  a)
Z (?)  a) 7 1 z + F = [46' = inj
'
_0 [16.25,34 = 1!]
V/)
_A [15. 24. 36 = mi]
U, u
r , 1 y ? (^  ) 7 + IT t 14 ' 26 ' 35 =Sl ^
L p "" .* ' n.., '
W
" =0, [51' p]
= 0, [35' =q]
IX r  *  gm
a n
^ ,,_,_ 2
+ ^+(p
0. VT.
378
A MEMOIR ON CUBIC 8UEFAOES.
'If, 1 l\. , 2\ W
[I5.28.40q,]
, [or = p,j
The coordinates of the 27 lines are then found to bo UK follows:
7 TO  
* V m/ \ n
0)
1
o
w / i\/ i
m [ _
W \ n
=("i)('J)
*'
~ T M "I 
1
n
m)
4A, ?v
nk \ m j \
"^("~")(
n
i)
m
412]
A MEMOIR ON CUBIC SURFACES.
379
(/}
(?)
(A)
(12=0
(S'ft,)
( 1 ="i)

m
( 2 =,)
n
Z
(23 =* Jj)
 in
'
< 8 '*>
1
1
( 1 ~ a 3 )
n
m
1
1
( 3 = 6 a )
n
^
I
1
(13 = a,)
m
1
1
m
(34 =04)
n
n
1
;
(24 = 64)
1
I
(H = c 4 )
m
48
380
A MEMOIR ON CUBIC SURFACES.
[412
412]
A MEMOIR ON CUBIC SURFACES.
381
(/)
((/)
(h)
n
1
m
(56=a 6 )
1
n

(**,)
m
1
(*'=.)
o
.!*
(35 = a fl )
(25  6.)
m (p  a.)
?
(P + W
n (p a.)
(15 = OB)
2(p + j8)
?) + j8
(16  a,)
2 ( p + ^3)
w j (y) a)
(6=6,)
rtl / 1) 1 rt J
\/ /
,^
(5'^ 7 )
3J(pfl
f B}
(36 = a.)
^J a
(>.
p a
(26 = 6 B )
2*1 (7? /})
^j a
(,
(16 = c s )
2*(/>0)
(45 a)
2m (jfj /?)
pa
6~&^
_fj a,
2?i (^j )S)
jj a
( 6' = Cn
382 A MEMOIR ON CUBIC SURFACES.
44. We have ,Y = 0, Y=0, Z=Q, F=0 for the equations of the planes
(12.34.56 = 4 (42' = ?,), (14' = z\ (12' w);
and representing by f=^+ Y+Z+ W=0 the equation of any other piano
7/t ?i v I
the equation of the cubic surface may be presented in the several forms :
0= tf=
here .^ , ,, w)! , T=
= Fhh
+ kzxj',
TKniT, + ]ixj"/,,
TflJ f kyzx,
TTnijin + kzxy,
Fnjn + kxyz,
= Fqq,
= T7rr, +k
out of tlle
r.^ s r Ai if:^:r
therefore ^(= V ) = 54. d uble n S ent of * spinodo ourvo, a,,,!
. The ^ of the
,
may Me ^
A MJJSMOIB, ON OUEIC SUKPACES. 383
of a doublesixer is a doublesixer. Hence the 27 lilies of the reciprocal surface may
bo (and that in 3G different ways) represented by
1, 2, 3, 4, 5, G
1', 2', 3', 4', 5', G'
12, 18,.... 56,
whoro 12 is now the lino joining the points 12' and 1'2; and so for the other lines.
The lines 12, 34, 56 moot in a point 12.34.66; the 30 points 12', l'2...5fl', 5'G, and
the; fifteen points 12 . 34 . 56 make up the 45 points t'.
The above equation, S a  2'* = 0, shows bhat the cuspidal curve is a complete inter
section 0x4; o' = 24,
Section II = 12  G a .
Article NOB. 47 to 50. Equation W (a, b, c, f, r/, h^X, Y, Z}*+2hXYZ = Q.
47. It may bo remarked that the system of lines and planes is at once deduced
from that belonging to 112, by supposing that in the doublesixer the corresponding
linen 1 and 1', &o. severally coincide; the lino 12, instead of being given as the inter
section of liho planoH 12', 1'2, is given as the third line in the plane 12, which in
fact ropvoHontH tho coincident pianos 12' and 1'2.
384 A MEMOIR ON CUBIC SURFACES.
48. The diagram is
[412
Lines.
tt tt >t Oi W W K> to bS M
12
13
14
1;
l(i
24
25
ffi
o <)(j
1 S 34
35
36
45
46
56
12 . 34 . 56
12 . 35 . 4G
12.36.45
13.24.56
13 . 25 , 46
13 . 26 . 45
14. 23.S6
14.25,36
14 . 26 . 35
16.23.46
15.24.36
15.26.34
16.23.45
16.24.35
16.25.34
15x2 = 30
6xl =
Jiimilinl
oiicli pair of inyo,
onclt
Uirco nioro lines.
412] A MEMOIR ON CUBIC SURFACES. 385
49. Putting the equation of the surface in the form
T, Y, 2f
, , ,
where for shortness
a = mn I,
finl m,
<y = Im n,
5 = linn 1,
p = linn,
then taking X = as the equation of the piano [12], 7=0 as that of the plane
[34], 2 = as that of the plane [66], the equations of the 30 distinct planes are
found to be
Z0, [12]
7=0, [34]
Z = 0, [56]
m Z + Z 7 + = 0, [23]
m'Z + Z
m
= 0, [24]
l^Y+Z^Q, [18]
20, [45]
^0, [46]
^0, [35]
^O, [36]
Z=Q, [16]
# = 0, [15]
^O, [26]
^O, [25]
F = 0, [12.34.56]
Tf = 0, [12.36.45]
F0, [12.35.46]
TF = 0, [16.25.34]
r_ ^877=0, [15.26.34]
^r +a/3 F = 0, [14.23.56]
^_ 7 S Tf=0, [13.24.56]
mnZ + ttZ FH Zm ^ + a/3 7 S Tf  0, [16 . 23 . 45]
pX + n F + m ^ + ^878 Tf=0, [13.26.45]
wZ + jp Y+l J? + 75 ^=' [16.24.35]
mZ + Z 7 + ^ + ^ F0, [16.23.46]
Z + ZmF+to Zfat F=0, [15.24.36]
ZmZ+ F+m57S ^= [13.25,46]
ZZ + m F+ ^  /9S TF = 0, [14 . 26 . 35]
ZZ + m Y + n Zafa F = 0, [14.25.36]
49
386 A MEMOIR ON OUBIO SURFACES.
50. And the coordinates of the 21 distinct lines are
[41!
(a)
()
()
(/)
(?)
(A)
whence equations may bo taken to bo
I
 1
1
(!) X=Q, 7+ IZ = Q
T71
1
 1
(3) 7=0, #+Jr=0
n
1
1
(6) ^=0,Z+7=0
I 1
1
1
(2) Jf=0, 7+ ^^=0
m" 1
1
1
(i} 70 y i. )' r n
\tj t. u, ^ + ni A = u
'
 1
1
(6) ^ = 0, X+ jrU^O
1
n
m
Wy
n
() ^r + ^.^ +/9yI ^o
n
1
I
I
ya
n
ya
/I /1\ T" / ^ 1^~ A Tr
in
*
1
I
a/J
m
()^ + W + ^o^.,.^iro
\
1
n
*
m
1
naS
(46) z+ WUNW^O, ,rsrro
n
1
j
1
"^
(26) 7+ lV + nX^Q, 7/?Srr^O
1
I
1
i
yS
??iyS
(24) ^r + m ijr+ /r,, 0, ^  yftjr,, o
1
71
1
m
___
maS
aS
/*if)\ y v i ^ A
1
n
1
I
JL
1
(15) r,. W+B.JT.O, r^Hr.o
1
1
m
t
yS
(13) ^ f inX 4 ^ l J r ^ 0* % vS n^ 1 n
* ' " * J. * v/j ^/ *y(j j f ^j y
" .
^
1
1
1
1
1
/Ojn\  i ir
m
W/?y
nfiy
V^OJ Ji 1* flJ ^I^Nfl>i"^'f^*_fl T^ i /I lir ft
\ u, A f TO 1 I m /J = U, ft . /3y  r =:
1
1
1
n
*
fya
(25) 7 + l^ +n ^=Q, 7,yaIr,,0
1
1
1
1
1
la.fi
mya
1
^^ ~
i
1
 _
" 1 .
(34) 7 = 0, WsO
,
412] A MEMOIR ON CUBIC SURFACES. 387
51. The six nodal rays are not, the fifteen mere lines are, facultative. Hence
52. Resuming the equation W(a, b, o, f, y> K$X t Y, 2) z +ZkX7Z~Q> the equation
of the Hessian surface is found to be
6, o t f,g t h1X, 7,%)*
2kW{(a, b, c,f> g, Ji$X, 7, Z)*(FX + QY+HZ)3KX7Z]
7c a a*X* +
whore
(A B, 0, ^ (?, ff) = (6o/ f (W^,
K = a&c  ct/ a  fy a 
The Hessian and the cubic intersect in an indecomposable curve, which is the spinode
curve 3 that is, spinode curve is a complete intersection 3x4; o' = 12.
The equations of the spinode curve may be written in the simplified form
W(a, b, o,f, (/, liftX, 7, Z)* + 2kX7Z = Q,
8KXYZW
+ SkXYZ(afX + bg7 + ohZ)
 7c 2 (Z 4 + 6T 1 + &Z*  26c7 3 ^ 3  2caM 3  2a&Z a F 3 } = ;
and it appears hereby that the node O a is a sixfold point on the curve, the tangents
of fcho curve in fact coinciding with the six rays,
Eaoh of the 15 lines touches the spinode curve twice; in fact, for the line 12
we have X G t W0' t and substituting in the equations of the spinode curve, we have
(6I ra ~c^ a ) :==0 > tlmt is > wo have tlie two P oinfcs of contact ^' ^=0) 7*Jb = Z^c.
Hence ^'30.
Reciprocal Surface.
63. The equation is found by equating to zero the discriminant of the ternary
cubic function
(Xa+Yy + Zg)(a, 6, o, f, g, h^X, 7, Z)*2kwX7Z,
viz. the discriminant contains the factor w a which is to be thrown out, thus reducing
the order to n' = 10.
The ternary cubic, multiplying by 3 to avoid fractions, is
X*, 7* ,
3cJ, 3%, ,
49 2
388 A MEMOIR ON CUBIC SURFACES. [412
Write as before (A, B, C, F, ff, H) for the inverse coefficients (A=bc~f a , foe.), and
A" = ale  fl/ 3  Iff  ck + %fgh ; and moreover
<P=(A_0, 0, ^ G, HJa, y, ,)
P^Aa
Q = Hx
U = aft/2 + Igzeo + clmj,
V  2/%ir  flP?/2  6Q
= ~a Efz ~ b Fz* ~ a
 a Gye*  1 HzeP ~ G
+ (~ abc  a/ 2  bcf 
W=(A t , 0, F, G,
M=kwU+V,
JT = 2i'fl&C K1JZW ) W ;
Then the invariants of the ternary cubic ar
are
is
a  . 0)
and the required equation of the reciprocal surface
viz. this is
which, arranged in powers of kw, is as follows; viz.
Coeff.
( _ fl
412]
A MEMOIR ON CUBIC SURFACES.
389
= 2a6o mjz (12i#  8i 3 ) + W ( 3* + 1
+ U s ( 2$ + 4i a ) + 2 DT ( 4) + F a
 86060 fljy F18tfTP + 72a6o coyztU
16 a* 1
(Aw) 1 =
18QVW
1G7",
bub I have nob carried the ultimata reduction further than in Schliifli, viz. I give
only the terms in (IcwJ, (kwf, (Iswf, and (&w).
55. I present the result as follows; the coefficients doduciblo from those which
precede, by mere cyclical permutations of the letters a, 6, o mid /, g, h, are indicated
by ().
= (Aw) 7 .
I (Aw)" .
wfss
ct?bo I 1.
/i/tflf 1 A,
\.tf\f\ti J. J.
"
i)
a a %6
o%oA6
06 V  6
B io/ 
y/i+ 2
o^ I 2.
V i 42
abofgk+
5cW + 2
a<j/"/i a 
Ixsfgl  24
flj/'V* +
390
A MEMOIR ON CUBIC SURFACES.
[412
56, In explanation of the discussion of the reciprocal surface, it is convenient to
remark that we have
Node <7 3 , X = Q, 7=0, Z = Q,
Tangent cone is
(, 6, o, / ft AJJ, r,2)'0.
Nodal rays are sections of cone by planes
X=Q, 7=0, Z=0 respectively, viz. equa
tions of the rays are
2 +cZ* = 0,
7 = 0, cZ' +
Z =0, .J*
6'r
the lint
Reciprocal plane is w = 0.
Conic of contact is
(A t B, G, F, 0, /fjj> y, 2 y= o, w = 0,
Lines are tangents of this conic from points
(y=Q, = 0), (2 = 0, = 0), (,v=0, 7/ = 0)
respectively, viz. equations arc
w  0,
w  0,
y ^ plft
f * he Six
s to
, tftyyz
(reciFooal8 f tha
TO i
12 and
points 1
and we thus have
meet: and the
or, what is the same thing,
(equivalent to two equations) t
the second and third equations.
19
 
hllPS mr. '19 Vl rrt ^ni i*
' ltj "" fi(s  Wio Unoa
written
Jlf
aUrTe '
in the conic w = o i0 ~ w =. i
. " ", or say it is a ourve 4 x
. 0l which
_ 2 . O '_ 18
412]
A MEMOIR ON CUBIC SURFACES.
391
Soction III = 12jtf a .
Arl,i(!lu NOH, 00 to 72. Equation. 21f(Z Y+Z)(IX + mY + n%)\
(i(). Tho Hyntom of linos mid pianos is at ouco deduced from that belonging to
II 12 (/a, by mipposing thu tangent cono to reduce itself to the pair of biplanes;
It of thn pliimiH (n) of II = 12 O a thus coining to coincide with the one biplane, and
tlmio of thorn with fcho othov biplaiio,
(II, Thn diagram in
Linos.
Q) CM W IJ IO K> M M (J
OJ CJI rf*. O> 111 It OV Ol It
O til it" 0) IsO f 1
ill Mia
11
^1
Ol l
M
Oi
X
1W
II
CO
wit
4M1
. . .
Biplanes,
ll
*
Ifi
'
'
1(1
,
'
yi
ui
*
a
1 yfi
HO
'
'
BiraiUal plnnoa each con
taining a ray 1, '2, or 8
of the one btplano, and
a my 4, fi, or fl of tlio
othor biplftiio.
an
M , an . tto
14 . H(l . Jlfi
lfi.UH.H4
16 . U . ild
10 . aft . IM
flxltstt
i7 4fi
t
Plftnos oaoh oontainiiiR
tlu'co moro linos.
if?
s **
I!
392 A MBaroiE, ON CUBIC SURFACES. [4J.2
62. Taking X+Y+Z=0 for the biplane that contains the rays 1, 2, 3, and
lX + mY+nZ=Q for that which contains the rays 4, 5, 6, we may take X~Q, Y=0,
Z=0 for the equations of the planes [14], [25], [36] respectively; and thon writing
for shortness
m n, n ~ I, l m \, ft, v,
and assuming, as we may do, k**\pv, so that the equation of the surface is
W(X + Y+Z) (IX + mY+ nZ) + (m  n) ( 1) (I m) X YZ= 0,
the equations of the 17 distinct planes are
Z = 0, [14]
J'O, [25]
^=0, [36]
A r + 7+ ^=0, [123]
lX + mY+nZ=Q t [456]
IX
12=0, [25]
Q, [26]
Q, [35] .
IX + lY+nZ = Q, [36]
F = ' [14.25.36]
Ff (XZ^O, [14.26.35]
F + ^y=0, [10.26. 34]
TF+ n^^O, [15.24.36]
ImX i m7 + n?^ + F  0, [15 . 26 . 34]
mnZ W= 0, [16 . 24. 35]
412] A MEMOIR ON CUBIC SURFACES.
63, And the coordinates of the fifteen distinct lines are
393
()
(*>
()
(/)
(0}
(A)
whence equations may be written
1
1
(1) X = Q, Y + Z=Q
1
i
(2) 7=0, Z + X=Q
 1
1
(3) # = o, js:+r = o
~n
m
(4) X=Q t ?7>?fcZ=0
n
I
(5) r=o, # + ZJT=O
m
/
(G) ^=o, ;x +r=o
1
(U) JST = O, ir=o
1
(25) r = o, iro
1
(36) 2T=0, ir=0
I
n
n
n^v
 nlv
(16) lX+nY+n = Q, IP"+?iv^=0
I
in
HI
l)i?p
Imp
(16) UT + wr+m2' = 0, IT!w/i7=0
I
m
I
a
Imp
(26) /JT+mr+/^ = 0, ir + tt,JT=0
n
in
n
tnnv
 v
(24.) X+ mY + nZ^Q, IT+ nf ^=
m
m
n
mn/j.
Wl 8 /*
(34) wZ + wFf w^=0, ir+m/tr=0
I
I
n
nl\
2\
(35) ZX+ jri #=<), IF + /XZ =
Tho rays are
The equation
not, tbo mere lines are, facultative; hence b' = p'Q; t' = (j.
of the Hessian surface is
k (l*X*
64.
G5,
+ kXYZ {(I* + 81m + 32?z + nm) JT + (m fl + 8mn + 3ii + nl) 7+ (?i a + 32 + 3?im + toi) Z\ = 0.
The Hessian and cubic surfaces intersect in an indecomposable curve, which is the
spinode curve ; that is, spinode curve is a complete intersection 3x4; a' = 12.
The equations may be written in the simplified form
W(X + Y+ Z} (IX + mY + nZ) + IsXYZ = 0,
 4A r YZ [I (in + n) X + m (n + 1) Y + n (1 + m) Z\ = 0.
We may also obtain the equation
X a Y *&
>\ Y
= 0,
C. VI.
50
394 A MEMOIR ON CUBIC SURFACES. \
L.
which shows that there is at B 3 an eightfold point, the tangents boing givoti by
(JT+ Y+Z) (IX + m7 + nZ) 0,
(V, & v\ fw, ~v\ X/$7#, ZX t AM r )' J  0.
Each of the facultative lines is a double tangent of tho spinodo ourvo; wlumcn ft* i* IN.
Reciprocal Surface.
66. The equation may be deduced from that for 11= 12  (?, vk writing
(o. &> o, f, 9, APT, F, ^) a 2 (X + K+ Z) (IX + ?F+ ^f),
that is
(, &, c,/ g, h) = (2^, 2m, 2, m+n, wM,
we have
(4,5,0,^, ff, ^) = (XS ^ ^ ^ , ,x,
Writing also
X, /t, v = m~n, nl, lm as before,
I (m + n) 2/2 + w (ft + ;) m + w (; + Mt )
^y*+ mpz!s\ nv
(l + B)ffl+ (ft+i)y+ (i + W
we have
and then
i = fc'
so that the equation is
L eualbn i7 *""' S W fc throw out ft the whole
factor equabn i W trow out ft the whole equation tho
412] A MEMOIR ON CUBIC SURFACES. 395
or, what is the same thing, it is
(7c 2 w a  Zkwt + <r) 9 (/cV0 + kw
 36 (Afttf 3  2&wi + <r a ) (kwv + o
 32 (fewu + <rty ) 3
 108&w (4rfcw0  f )" = 0.
67. This is
(fay) 8 . 5
+ (/cw) 5 .  ^ ~ G<0 + u a
+ (kwy . <T H . 30 + <n/r . 2u + f a . Gi
+ (kw) s ,  2<ri/r B + o a (2u a 
. 37
288W  96u a ) + f 2 (Si 3 
(Jew} .  o^p + o 4 ( QtO + u a ) + o 3 f ( 8tv  1445)
o^ 3 .  72 + f * ,  108
which, reducing the last term, is
m) (lz  nes) (m>  ty) = 0.
08, I verify the last town in the particular case = as follows: the coefficient
of CT" is
(0, 2tt(Z + m)ffiy, 2(m + w)w + 2(ft+Z)
which is
{(I + 7)i) (Jus + /*P) 9 + [(^' l + w) + (n
{[(I + m) \ + (m f ) c] Xa* 1
+ [2 (Z + w) X/* + (m + ) /w + (n + i) i>X + 2?iy]
which, substituting for X, & v their values mn, nl, l^m, is
= ZtfwY .  2X/i (>  y) (m  ly) ;
or for ^ the coefficient of o 8 is
agreeing with the general value
396
A MEMOIR ON CUBIC SURFACES.
[412
69. In the discussion of the equation it is convenient to write down tho relations
of the two surfaces, thus :
Cubic surface.
B a , X = 0, 7=0, Z=0
Biplanes X + 7+ Z=0
intersecting in edge.
Rays in first biplane,
X = Q, 7+ = 0; 7=0, Z+X.
rays in second biplane,
X=0, mY+nZ=0 t 7=0, nZ
Reciprocal surface,
Plane w 0,
Points in 10 0, viz.
in line (m n)a + (n  1) y + (I  m) z~Q,
that is, \o) + py + vz 0, or <r = 0,
Lines in plane <u) = 0, and through first
point, viz.
i/z = 0, z~x = Q, /y7/ = 0;
lines through second point, v\x.
70. The equation puts in evidence the section by the piano w = 0, vk thin in
the hue <r = (reciprocal of the edge) three times, and the six linos (reciprocals of
the rays) each once. Observe that the edge is not a line on the cubic; but its
reciprocal is a line, and that an oscular line on the reciprocal surface; tho six linon
reciprocals of the rays) are mere scrolar lines on the reciprocal smfaco; they PBHB
three of fa, through the point a = ^ and the other three through the point
li/J v i : ; n; ff* 13 < th y ave six tangents of the pointpair (reciprocal of tho
pan ot biplanes) formed by these two points.
,  put in evidence the nodal CU1>VO on
at Pedes ^t oo t g g ^^ rf ^ mere ^
i, , d and 4, 6, 6 the hnes which pass through the points 0,
o: by
g
0*0 anil
72, The cuspidal curve is given by the equations
9 , 24 (kwv + o1/A _ 3G
Writing down the two equations,
= 0.
= 16.
_
J i:
s a partial intersection 4.84
412]
A MEMOIR ON CUBIC SURFACES.
397
Section IV =122(7,,.
Article Nos. 73 to 84 Equation WXZ + P (yZ + S W) + (a, I, c, d$X t 7) 3
73. The diagram of the lines is
l(i lO CM rf^ W W
Lines.
( M M U> tO OS
to a* e rf2 i
tf^WtOH^.COI.OI'
01
IV = 1
22G' a .
Ml Gi
M
M
X
X
X
bO
o
^
[0]
1x2= 2
Plane touching along
nxis,
11'
*
Planes through axis,
22'
* *
eitoh containing a
4x4=10
* *
ray of the one node
1)8'
*
and a ray of tlie
*
other node.
41'
12
111
l'I:
. 23
8
1 24
PH
84
Biradidl planes, each
containing two rays
t B
two rays of the other
12
.
node,
I'd'
*
I'd'
*
2'8'
2'4'
12.84
18.24
8x1= 3
i ,
.
Planes each contnin
ing three mere lines.
14 , 23
20 45
g
W
s
f"
p p
ht ,*s (b p
Sul^Sa
IB (D " m
lr
ransversa
tangent ;
the axis.
'
Qj
'I
Ci
S'e ^ 5
* B rt^ w
E 1
&
B * o P
pUft ^,
S B'
B 1
*~** p *^* p
3
o
fmt. ! .( . IH
H ^ H
hj ^4
<
<D T TO i fO
398
A MEMOIR ON CUBIC SURFACES.
= 0)
, And the 16 linos are
[0]
[irj
[22']
[33']
[44']
[12]
[13]
[14]
[23]
[24]
[34]
[1'3'J
[1'4'J
[2'3'J
[2'4']
[3'4/J
[13 . 24]
[14.23]
[412
<*)
(ft)
' ,
()
. _
(/)
(?)
W
whence equations may be written
1
"(oT^oTr^
y
d
(o) .A = 0, dY + yZ + 8W =:Q
*
o
n
rt n
V
1
f i
8
(\\ Y ~ f v n &v
fg
S
(2} *
V
f s
8
( )
V
ft
S
( 3 ) t,
1*
^
V / M j,
MI;M(IIII (N nnutf .SUHKAIIKS.
39!)
(") iM ' () ! (/)
t.v)
(M
''. '. V n n
v (i 1 ) Arf,r..o, y
ri.i' l ir^o
I'j I/ I" 11
n
v (0
I*
f, IV II II
n
/'I' V
y (' )
' * ' M
it
f, I, " it
v M'l
r \< i
* \ r t)
I*
A ., , 1 1 \ / 1 1 \
,,.'< I i  ( i \
V
' ( l ) ^ ' /' ^
1,1; M. i..J ' Mr, r/
Wi
(la.n'j*
<S M' ' ) : i '  /' l \
1,1; > ! '> ' J Mi',' < t )
V
" /I l\ 1 /I lv
Ci'Af, ''rj 1 rjAr/ 1 '^
(!.
fi ; M '  ' 1 i ^ ' \
M t " ( i, ^' ' v(f/J
V
i /i i\ i /i i\
f^Ar, iVVAr/fJ
(..i.^),,
.i ' ; '/,.!) ] *(!/;.)
f J
'l'(
fifAr/'J t^'Ar, '''];)
(a.r.r )n
' i M 1  ' J i ( ' . l \
M. ; vt, ',' MI, ij
r
f.f.(f B '"0 f.i'Xi;' 1 '!;)
< i ')
n fi (r'r) ' >(! '!)
v ,
iif/r/'i) f,Ui'"r)
(3.1 . 1'2')
uiu
ttvr
.!)'! *i ith liii.i 1^, :tT, ..hs.'i'vn Unit llu twu iiintiuim
Eh. P>
A*
lliul.
n
.
t ss 3,
es an tliw
mul thy wix iiiui'o liiii);
A .';' i i
/'(.V  y,,') 1
\'t' **'.(('.' i yii'}.,
4
loo
A Ml:\!nll ,, N ,  m , , ,
''"" H'mi, ,.!!,, , ,,...
7!l ' <! "'iiliiiiii.ir will, (I,,
1111 ..... ' ........... ^n
' J') 1 . whrjv ((,,. n^,,.,^
ir*
111 '.U
II
I',!'
,,
1 Ih
ir..!j' ;! I . y
HM. Tl
A*
 X
I A MM.MOIK OX iHMIC N
Sivtinn V \"}., /i',
H.V Tin iliiii;iain ..I I),,. MM, : , tlll  ,, IMI ,., ;.,
'/.  U,
V" J ^ II.
% \ / /i 0,
f iro.
I'tnin 'i ';!i'li lUi.'ii
H)t> llllll'Al'liiJll,
i
'
1 1 r *t^*
 1 I .MAT
U^' 4 I '
H , t
ON cmin< HUUKACKM. . i)v ,
A, U lt F, , _,
liquation i ./.'I /'".
A(A/t  A l ;i r  //,") Ml../, u,
whom UK, torm II nmy 1,,, oalrukto,! wi.J.nu,, lillin.lrv: ,). ,,.,. (l . nil ., ,,,, n
tho (lovoldpii
x,n, n
H nf } (' Av ... A \r i //n ,. I r
' } "' 7
^^^ ..... , ..... i..,
,,,,, i, ( ' ; ; ( " "i" l " 11 :*'" ( ',, (lll .i liM1J .
Uio plan,, c^0. ' V ' <! " 111 Nl "' ( ' A 'l '"""1'Hl.v H.r il,n ,. ,, v
HH. Thu tioilul iitirvn w iimitt* un ,<(' n,,. i; i ,
. I ..... " nn,l tin, ,,, u , N v al; ' ,''J' '"',', wl "" 1 ' ,'"'"
": ' ..... ' !. i ..... in,,, ""
!'" '" ..
""" T " t! l " fl " f "'"
..... ....... '". ........... ,,,,,,, ,,, ,;,,
A" .2.H/ii.. .fl/,tw i),
412] A MEMOIR ON CUBIC SURFACES. 405
The equations of the spinode curve may be presented in the form
XZ, flZ 8 + bZ*  Y 2 , aX 3 + IZ* = ;
X + Z, W , Y*
it is a curve 3.4 2, the partial intersection of a quartic and a cubic surface which
touch along a line.
The biuode is on the spinode curve a singular point; through it we have two
branches represented in the vicinity thereof 'by the equations
8 x_ mv^A
' w~~WJ (w) )
respectively.
90. The edge counted once is regarded as a double tangent of the spinode
curve (I do not understand this, there is apparently a higher tangency); each of the
four mere lines is a double tangent; the transversal is a single tangent; hence
'=2.2 + 2. 4 + 1, =13.
Reciprocal Surface,
01. The equation is found by equating to zero the discriminant of the binary
quartic
X<s + Za)XZ(X + Z
viz. multiplying by G to avoid fractions, and calling the function (*X, Z)\ the coeffi
cients are
Qw (a; + 2aj),
f } 'A (.7; + z] w 4 4 (a + 6)
and then writing
L =if + 4 (a; + z) w + 4 (a +
we find
and then the equation is
 54wW) a } 0,
viz. it is
9 = 0,
404
A MEMOIR ON CUBIC SURFACES.
[412
87. And the lines are
(i
&
c
/
*
A
equations may be written
1
(3) ^ = 0, ^=0
1
1
(4) X+^=0, TT=0
V&
1
(1) X=Q, 7^V6 =
V5
1
^ J / ' ~ > ~
1
V5
(V) ^=0, ZVa+r=0
1
V
(2') #=0, Z\/fti 7=0
\ / '
1
Vi
1
1
Va
2
2 ( Vrt  V&)
2
(11') but for bho otlior lines the
coordinate expressions are
1
1
1
tlio more convenient.
~7^
2
2 < V  Vi)
3
(12')
V o
2
2 ( V + Vi)
2
(21')
Vfl&
i
i
Vl
2
2 ( Vn + V&)
 2
(22')
88. The four mere lines and the transversal' are each facultative ; the edge is
also facultative, counting twice; p' = &' = 7, i' = 3.
That the edge is as stated a facultative line counting twice, I discovered, and
accept, d postwiori, from the circumstance that on the reciprocal surface the reciprocal
of the edge is (as will be shown) a tacnodal line, that is, a double line with
coincident tangent planes, counting twice as a nodal line. Reverting to the cubic
surface, I notice that the section "by .an arbitrary plane through the edge consists of
the edge and of a conic touching the edge at the biplanar point; by what precedes
it appears that the arbitrary plane is to be considered, and that twice, as a node
couple plane of the surface: I do not attempt to further explain this.
89. Hessian surface. The equation is
6;
Combining with the equation.
X2W+(X + Z)(7*  etX* !>&)'= 0,
and observing that from the two equations we deduce
it appears that the complete intersection of the Hessian and the surface is made up
of the line .X" = 0, Z~Q (the edge) twice (that is, the two surfaces touch along the
edge), and of a curve of the tenth order, which is the spinode curve ; c' = 10.
412]
and wo thus have
or, what is the same thing,
A MEMOIR ON CUBIC SURFACES,
407
LM+ wW =(),
L, 12M, fl/V
=
w 9 , L t M
for tho equation of the cuspidal curve Attending to tho second and third equations,
those aro tjuarbics having in common w* = 0, L = 0, that is, tho line y = 0, w = four
times; or tho cuspidal cnrvo is a partial intersection 1x44: e'~12,
Section VI ~12.B a 0,.
Articlo NOH. !)5 to 102. Eqiuilion WXZ+ Y*Z + (a, 6, c,
f)5. Tho diagram of tho linos and planes is
W >( rf K*. N> rf* W lS M O
Visi37i 8 a a .
(i ^
CO
w
K*
M
fj X
X
w
X
X
l ^l 1
II
II
CD
II
en
1x0=
Biplane touching nloiig axin, nntl
containing transversal ra.y.
00
Ixflss
, . .
Olhoi 1 biplnno.
22'
,
'
B
9
,
PlnnoH oaoli tlnouRh tho nxiB and
jjlj'
(J xfl = 18
.
,
containing a ray of tlio binodo
4 ' M '
ftiid a my of tho onionado,
1
* 12
'
*
Biinclinl pianos of tho binoclo,
18
8x8=
.
ctioh containing my of axial
biplnno tincl a ray of othor \)i
*
t
piano,
14
3'8'
2'4'
8x2=
t
' '
Diraclial planca of tho ouicnodo.
8'd'
n
* *
K
g
tr&
W
Et a"**!
p.
S
g
'I'i'
>d g
iL
K
Q
a BE
" S.
P
* G
3
'I
" ^
oa
*3j
a.
a is.
&
*
<o
S 1
B.
3"
g
t* P
o
406
A MEMOIR ON CUBIC SURFACES.
[412
92. This, completely developed, is
64?<J 8 . ab (a + 6) a {(a + 6) f  (te  s)}
+ 32W . 2t(6 f 3 (a + 6) [(a  26) as + ( 2a + 6) z] f
\+ (IB  zf [( Sa + 56) x + (5ft  36) z\
3a6(a a  7o6 + 6 s )2/ 1
[6 (9 a + 266  6 3 ) $  266 (ft 4 6) aw + ft ( "
Cffl  zf [6 f 12a + b) a? + ZZdbKZ + a (ft  126) z
V / L \ *
+ 8?w 3 r 36 [(2ft  6) w + ( a + 26) 0] # 4
26) a; 3 + Sabxz + a ( 2a + 96)
+ 6a%  (a + 6) V + cwa 3
2w
+ 4,'cV (w + z] f
96 s ) &} f 
where we see that the section by the plane w = (reciprocal of #,) is imulo up of
the line w 0, y = (reciprocal of the edge) four times, and of the lines w^O,
ttyaaPeOj w~0, 6?/ a ~s a = (reciprocals of the rays) each once.
93. The surface contains the line ?/~0, w Q (reciprocal of the cclgo) ; and if wo
attend only to the terms of the lowest order in y, w, viz.
2 a w 2 + 8
which terms equated to zero give
we see that the line in question (y = 0, w = 0) is a tacnodal line on tlio auvfacu, fcho
taenodal plane being w = 0, ft /a'etZ p/ans for all points of the line: it has already
been seen that this plane meets the surface in the line taken 4 times; every other
plane through the line meets the surface in the line taken twice. Wo havo in what
precedes the d posteriori proof that in the cubic surface the edge is a facultative line
to be counted twice.
94. Cuspidal curve. The equation of the surface may be written
 (LM + Dsylff) 3 = 0,
412] A MEMOIR ON CUBIC SURFACES, 4C
99. The equations of the spinodc curve mi\y be written in the simplified form
XZW+Y*Z + ( (t> b, c, cl^X, Y) s ^0,
4(a, 6, c, d$X, F)H3(4flo3& !l , d > H cd, d^X, Y)* = Q,
the line X  0, Y= here appearing as a triple lino ou the second surface ; tl
curve is a partial intersection, 3x43.
The node Q is a triple point on the curve, the tangents being tlio nodal raya
The node ,83 is a quintuple point, one tiuigonb being AT = 0, RdY+^Z^Q, and tl:
other tangents being given by Z = Q, (4ae  36 3 , ad, bd, ad, d^X, Y')***Q.
Each of the facultative lines is n double tangent to tho curve, or wo hivvo /3'~ G.
Reciprocal Surface.
100. Comparing the equation of the cubic surface with thai; for IV =02 2(7
it appears that tho equation of VI = 12 H s  O, in obtained by HubRtitiiling iu fch
equation the valnsa S = 0,7=!. But instead of making this substitution in tho ruin
formula, it is convenient to make it in tho binary qiwitic (#"5A r h K}', thus in fac
working out tho reciprocal surface by moana of tho funotion
a, >, o, f
the coefficients whereof (multiplying by 6 to avoid fractionn) are
Sony { IRbzw,
6w a .
We find
where
i 2/ a + 6 (a 1
M = Zdmj + 6
N =  4d B ffl a  Sd (3w  2tt?/ h 2d^) w  1 2 (36 1  4a o) w 9 .
The equation is
lOfte 5 ^ K /j2  122WJlf ) 3  ( /jB ~ 18*w/Jf  64s^JV)a  0,
viz. it is
i a (LN+M*)  ISzwLMN  IQtwM*  27^ 3 w'W a = 0,
where however ii\ 7 +Jlf u contains tho factor w, = wP suppose ; the
L*P  l&zLMN IG^Jlf 3  W&vN* = 0.
Write
A
G = 66tZw  4rfy 4 4^3 + 3 (36 s  4ao) w,
c. vi. 52
408 A MEMOIR ON CUBIC SURFACES. [412
96. Writing (a, I, c, dJ[X, 7) 3 = d(0 2 Z 7)(0 3 Z 7)(^Y 7), the planes are
Z0, [0]
Z = Q, [00]
2 Z7=0, [22']
^ 3 Z7 = 0, [33']
^7=0,
7 (03
[12]
[13]
[14]
[2'3'J
[2'4']
97, And the lines are
a
b
c
/
A
equations may bo written
.
1
(0) . JT0, 7=0
1
rf
(!) JtT = 0, dY *%~
1
o.
(2) o 2 A r  r = o, ^r o
1
8 3
(3) f? 3 Z7=0, ^r==0
1
04
(4) ' 0,X~ 7=0, ^T0
02 '
1
2 a
(2')  7=0, a "A'+ )r0
6 3
1
3 a
(3') 8 A r  7=0, O a *X+ ?r=.0
04
1
.
4 fl
{4') ^ 7=0, ^A'H ir0
de,
<ie t
d
d
1
1
(*.+*)
<,+**)
~Q$4
OA
d(e s e,~e z e 3 o,e,)
d(OAOAOA)
n 2 g' j/\
but for the remaining linos
(13 . 2'4') the coordinate oxpvoasiona
de,
d
1
<*,+*.)
0A
d(e,e a 6AoA)
nA o'o'\ aro mt >ro convoniont,
^1'i . A 1
The mere lines are each of them facultative ; 1)' = p' = 3 ; f = 0.
98. Hessian surface. The equation is
^+S(cZ + dP)}(Z^F+I^ + (a, 6, c, dJ2T, 7)'}
42 (a, 6, Cj dJZ, 7)
8(40036", ad, H od, d'JJT, 7)^0;
and it is thence easy to see that the complete intersection is made tip of the line
A0 1=0 (the axis) three times, and of a curve of the ninth order, which ia the
spmode curve; & Q.
412]
A JMUMOIR ON CUBIC SURFACES.
411
l/ = 0, Next as regards the surface LM 4 9zwN = ; the line ?/ = 0, w = is
a simple line on the surface, the terms of the lowest order being Qzw ( 4>clW) = ;
that is, we have ?y=0, and for a next approximation w = Aif, viz, L=y\ M^
:/V, and therefore  2r%' + 9.s. A 2/ 3 (4rfV)= 0, that is, A =\ , or
., 18rfnw
3 ; there is thus a threefold intersection with one sheet and a simple
intersection with the other sheet of the surface ISIZzwM = 0. The surfaces infcorseot,
as has been mentioned in the conic = 0, ?f + 4't,w = ; or we have the lino y = 0,
w0 four times, the conic once, and a residual cuspidal curve of the order
4.442, =10; that is, c'=10.
Section
Article Nos. 103 to 116. Equation WXZ+ YZ + YX* %* = 0.
103, The diagram of lines and planes(') is
Linos.
j_i M
O3 t^
05 W
ll
VII]
2B D .
Oil IO
X
to
X
X
X
H 1
Ot
On
IOI
M.
o
Ot
K*
O
01
1x15 = 16
4
Toi'flnl biplnno.
00
.
.
1x20=20
. .
Ordinary biplnuo.
si
a
03
*
fe
IB'
2x 5 = 10
Plfinos eaoh containing
EI more line.
T 45
I
a !
f!
td
f
en
* 3
P S"
B
V
s
i The marginal symbols in the preceding diagram* constitute a roal notation of the linos and
but hare, and still mow BO in somo of the following atagmme, thoy ai e anero marke of veforonoo,
which are the lines and planes to which the several equations respectively belong.
522
Mum wti have
A MKMiili! UN t II1H M Ki \. j ,.
A :
M '
A/. .....
101. Coimiiln HIM M'ttMni, l.il.M ,!, ,
tlur f,uv
Ill
ft
10m Ay, VIK. Wli
f ., ,
^ ; '' ' ( " 7''
' ' ' " lM '' It
^
r , v .,, ,.,, ,., t . 
. .
liu, ,(' ,, ,v,i,,n,.l ,,,,,ih.v.
lH ,,r rl,..,.....;.!.!! ..,
(hat i, A  '' ,; 1 "' / '""'' *"""* ""'' M "" S "" *"  m ^/*%'"0.
3SSS &
412
A MMMOUl ON (Until! HUHKAdKH. 4 \ ft
10!). Kach of tho linu.s (A' ~ IF' I), Y\X.0) , u .l (A'  W() t ,K #<)) is
iloublo liiuigonli of fchn Hpiuniln ontio; in lar.fc I'm l.lin Ih'Hl; nl' MmHn Ijm.'s \vn huv<'
Unit i,s,
HO I'Jmli bh lino UnudhuH uli (Jin l,wu poinU W iviii by ^^.hM); mu l
ittlior lino toimhoH ub lilio hvo pniuU ^ivnn by U0 J I : : 0.
in nbUin.'il l.
vk Ottlliii K tins (*^A r , AfV, Mm nioniKiiMtLH (tnnllij.lyinit l'.V ti) nn
(({<  M//WJ, y I 4,,'m, ([/HI, 2 )');
. ,
aw! tlitiii wnl
,A
,
we hiivn
\1
Tbo od^i A'O, ^r0 IIII
vix. Mm Uim(iionM A%, 0, ^ ,
nniHt bu riiokiiiioil only UH n tlntibhi l.iui^riiL ll.'iir, //',,,;. o .,. >> p , , (j,
mid fcluj i'iiia(:i.ni IH, an in lii
,A' J ( f,N  ,U y ) I Hw^A J/ A r  1 1 iv0/ J ... ^TwW ^ ;
but ^VHjV ,ut(l tlmrororn Uu whnh ..pmlim. ilivi.is by . ami v ll.n
.,, 07/M'A'',': o
ov, (Htinplotoly (Iuvcluii!il, thin IH
w' . (it
+ W.:)2 ( ,V.kc)
+ ?"'. l(i.i'( fly 1 4. Jis')
+ <" ( // 4 + at)// 1 ^ ( KiO//*.^ ^7^' >< lil..*)
4 7W' J . >kr ( 1 1, V ^ + 1 2yV ^. !l,y^  .k a ,.^)
h w . i'
412 A MEMOIll ON CUBIC! SUUKACfW,
104. The planes arc The linen uro
Z =0, [10] A r 0,
A r = o, [oo] yo,
0, [12'] A'O, yi
A r lF^O Y\ %::(}
JL It ""* \fj ,1 [ fj "' W>
105. The two mere lines are facultative, and tho od^i; in uho litmil.al.ivn ; /i';i //.'I
106. Hessian .surface. The equation is
X*&)~ 4A r 'r# + A 1 ' + \ft ; , ().
The complete intersection with the tuirfiico i,s thus givoii by t;ho (>iMiil.itiiiH
WXZ + Y*Z + Kf ~ jgs. o,  &*Y% I P I l^ 1 s 0,
which i B made up of the lino X**Q, %(} (tho I W ) loin ti m ,. i UI , a rnrvu ,,f Mi,*
eighth order. To see this, observe that tho luHtrimmUuiiuil mirliiiiCH li.uv in nu ......... i
the line A r = 0, ^ = 0, which is on the first miife,, u t,,, wi i j nt , ( l!lUl iti llM j,, vi,,;,,!^.
being ^A^.and on the second surfnco n triply linn (nqiiiLli.utN in vininit. lin,,
Z l yX* and X*=] 7 Z*}, But * ^A'. tonoto, Jgr.^V, un.l U... lu...
(2 + 2 =) 4 times,
107. I say that the complete in Demotion is th,, linn ( A' *..., ^^0) Mtivc linir,
ogether with a spinode curve made up of this mm lit,,, ,. and uf t lt , mnv ,,r
the eighth order; and that thus o' = 9,
The discussion of the reciprocal surface in f ll(!t show, M,u; ,; in .v.m.rnral ,,f (h,.
edge is a smgulav hne thoreef, counting onee us a nodal ,! ,,wi,, , H 1<
108. I find that the octic comnonont of Mi ^..;, ,,i
the equations of which may be written ' " ' '" ' irv " ' H u IUlblWlU m "' V1 ''
X ; F :
 . ;
the values of ^ at the binode B, arc tf tf i ,1 ,
bourhood thereof the two branches ' WU th " H " htllin iu thu >
412 ] A MEMOIR ON CJUBIG SURFACES. 415
that is, the surface has the nodal lines (* + y=sO, jyw = 0) and (zy = Q, #[ w = 0)
which are the reciprocals of the lines 12' and 13' respectively. The nodal curve is
made up of these two lines and of the line y=Q, w = Q (reciprocal of edge), as will
presently appear; so that we have b' = 3.
112. The equations of the cuspidal curve are
L*
LM+
0.
Attending to the two equations
48 w 1 = 0,
(8  72 =)  Oiaiio 3 + 18si^ = 0,
these surfaces are each of the order 4, and the order of their intersection is = 16.
But the two surfaces contain in common tho line (y Q t w = Q) 7 times; in foot on
the first surface this is a cuspnodal lino 4auw + ?/ + Ay* = ; and on the second
surface it is a nodal lino w (Amy + I8&o) ** Q \ the sheet w = is more accurately
4aMO + 2/ 9 HJ?^> 1 .. s 0; whence in the infcorsootioa wibh tho first sinface tho lino counts
5 times in respect of the first shoot and 2 times iiV respect of tho second shcot
together (5+2=) 7 times, and tho residual curve is of the order (167=) 9.
113. I say that the cuspidal curve is made up of this curve of tho 9th order,
and of the line j/ = 0,^ = (reciprocal of tho edge) once; so that c' 10. In foot]
considering the line in question ^ = 0, w = Q in relation to the surface, the equation
of the surface (attending only to the lowest terms in y t w) may bo written
giving in the vicinity of the lino
and then
_
that is, ^.= = 2 or 4>%w + f*= V2. *y* ; wherefore the line is a cuspnodal lino,
counting once as a nodal and once as a cuspidal line; and ao #ivin the forcgoine
results &'=3, c'=10.
114, I revert to the equation which exhibits the nodal lines (aiw^O, ,
10 = 0, yzfy for the purpose of showing that they have respectively no pinch
points; that is, that in regard to each of them we have j' = 0. In fact for the first
414 A MEMOIR ON CUBIC SURFACES, J'41.2
111. To transform the equation so us to put in ovuluiicn tlio nodul cum 1 ,
I collect the terms according to their degrees in (>j, z) and (;, w); vix. tint u<tmtlim
thus becomes
& a w fl 4
4
4 ^ 3 2/ a . w
+ f .  tu
+ z . w
and if fa momont we write .
^a momon we wte . + 7 , y. 7 mM , a)llBOtj u l (; . ilmit(1 , wji
, 7 by their values (* + ,,), ^y), the equation can bo oxpivHso.1 in l;hn ii.nn
4
( 4 y Y (tO + W ) a
(* + y Y (z  y) ( W + W ) (
 W (* 4 7/ ) ( Z  y)3 ( ftl
4 lw (2 y^(fl;^ w .
and observing that we have
f the 8coond wte h ' +
order m * y and W + M conjointly;
4I2 .1 A MJOMOIU ON CUBIC SUWACKS. 417
115. The cuspidal uinthio curva is a utiicursal curve, tlio equations of which can
be very readily obtained by considering it a* fcho reciprocal of tho spinorlo torso: wo
in fact have
(a : y : z : w = ZW+ 2 AT : VYZ + X* : WX K a 3^
or substitufcing for X, Y t Z, W their values (= 1(>^, 40 + 100', 1G#,  5  80'
and omitting a common factor 100 3 , wo find for the cuspidal curvo
A : y \ z : w80 + 240 s 160 : 240" + 820" : 4480* ; 160"
(values winch verify the equation A'+ >> + #* + Fw  0); tho spinoclo ourvo boiiiff
thus of tho ordor =!) as montionod.
For 0co we have the singular point (y 0, 00, w  0) (rociproeal of torsul
biplane), and in the vieimty thereof at : y i x : w=*l :  20~ 3 : JJ0" ;  0", tliereforo
For = we have the singular point .v0, ?/ = <), w = (reciprocal of tho other
biplane), and m tlio vicinity thereof > : y : g . w=*~ %0 :  fi0 a ; 1 : 40', therefore
110. Tho section of tho surface by fcho piano Q in an interesting ourvo.
Writing 2 = in the equation of tho mirfuco, I find Unit the rumilfciiie ommtion inuv
be written *
h ^"itf 9 + 27fl, J , T/ IJ  82rt;?y) a 0,
where observe tliat
HO that tho curve has tho four cusps w a +27^0, ^32ro/0; the piano
intcrsocts tho cuHpidal ninthic curvo in tho point (y~Q> js^Q, zy = 0) n,.Mni.{
and in tho lastmentioned four points; in liict, writing in the equal;!
curve 3 = 0, that is 1 + 120' = 0, \vo find w f ?/) w = 40, M0J,
1 + 120*) = 0, if  82ww 0,
41G
A MEMOIR ON CUBIC SURFACES. [412
of these lines, neglecting the terms which contain ce w, y\z conjointly in an order
above the second, the equation may be written
82w" (zy) ( + w) as (w  w) (z + y)
I iw (2y) 4 (ffiw) 9
+ f*(*y) (<e  w) (s  ?/)
,.. . ~ ^' (*#) ( 2 + y)^0,
this is ^
(^1, 5, 0$fl;w ( 2 + 7/) a =0,
vo, collecting the terms and reducing the values by mean* of tho c<matioiiH
= Q,z + y = Q, or ^y by writing w = W) y = e, we have
, .
Hence the condition 440flan rt f n i
points (if any) would be at the :^' U !'. 1 .7' 1S (f8)' = 0, 80 that tho
= 1,  2 A/2, 2 V2 1 But t. , ' ^ + * = ' ^  8w ' J = ' "' 8 y >'
ti S (y tho equals' I^s W ilf/r,/' " f> JV = 12 ' ]2 '  1C i l
"' '
J..WUVAVUI me obviously no f /in tl r " "" v> fu " 4 ' vo "J 1 
com l )0 '^ the cusp ^^^^ *% to on tho
The line y=0 w = * pmohpointe.
J>eing part of the cuspidal curvT no ^ t^ ^ * vefy poinfi a pinchpoint, but
M, in regard to this line also we' have i'o A i ^ rGgarclGcl ^ ft pinchpoint; that
? ==0  ' ? Aud thorefom for tho ontivo nodal curve
A iMUMOW ON (HrillO t
,H. Tukn iiin in,, IIH Um 1'ootn o' Mm equation (i
i I, Mien Mic jiluncN are
A'O,
wi 
4fH, HO tlmt
% K
o,
t J) ]
}'
% ! A r
0.
1. ll
';'
A'.f ir
r= 0,
[ .'!.]
r!
(JH,  1
v,
: S
r "
(/ a i
A;
' i/'i
r
O^i)
\
, H'J
[ 25)
r
(.!)
^
. 4(i
1 7y(t 1
LIU.
It
And tlic
ines un
,
h
I/"".!
(tquutiniiH miiy lu 1 \vi'i(lcii
1
(7) A' =0, l r
(I
i
t)
() ^ 0, I'
1
I)
n
(!') II' O t J': 11
1
I
1
(7) r1^.i.A o, ir n
1
1
1
t)
() r^A'i ir ii, ^ r tt
1
1
I
<!>) )'.i./r.i. ir o, A' 11
7/1, 1
I
1
(i) r <., Dr, C , (j i)^
1
.
o
I//., 1
1
1
till " I
(B) r:( M , a l).V (, r .)^f
1
I
HI,  1
1
!,. 1
() y (wj.i) ir .( W IJA*
I
1
w.,  1
t)
1
(i) 1' OH, 1) ir, (m.l)A'
1
1
1
(fl) r (m, i)jsr *( Wl i) r
__l_j
1
1
<rt, * v IJX ^ r
418
A MEMOIR ON CUBIC SURFACES.
412
Section VIII = 12 3G' a ,
Article Nos. 117 to 125. Equation P+ 7(A r + # + W)\
117. The diagram of the lines and pianos is
Lines.
O H td 031 )
I=!
23<7 2 .
M.
X
X
X
to
M
iK
tOl M
II
II
ql bj
w
LO
7
\"
__...
I'luiU'H cuuli LoimliiiiM
3x2= C
alotift an iixia nntl
(,
'
(ion tiilnin^ Uio oiii'i'ti
"
HlHUUllllH LriUlHVIH'Mlll,
12
34
8x2= 6
1 '
'
Bli'ftdml plnmiH onoh
t
(iontiuiiiiiif two nivH
G6
" '
.
of tlio Hiimu innlit.
18
10
26
6x4=24
(
1'lfinoH oiiuli {ie)iitniiiiii
(in axiH.nnd two myH
ItiroiiKh tho Utnnliial
4(i
nlii mniiooUvcly.
36
'
80
1x8= 8
, .
Piano Uii'oi^li Uio tlirco
ItXliK.
89
1x1= 1
H 46
.
I'lnno tliroiieli tho three
trnnHvoi'Balfi,
_
f
3
H
W
B* H
m
fi Culi
m CA
J"
  J~^
(3
S
=71
(6 S>
w 5"
fr
si
05
tfl B
E.
tf
*o S
S ff
0!)
a
Q. S
* i O
W C '
9
1
o
E, 09
 ~~ ~
<1 tD CO
~ _
~ ,
O
A MUAIOTU ON <!IMW! HIMIL'ACJUH.
...
UllH 1H
i <r.' J (('I 2 7
I 2. f((i 7 ' J
 (tv  i)' J (; ~ w) g (0 ~ VW)' J ^ 0.
W. Tilt) nodiil Miii'vo i^ mini.) up uf ,ht! liur (t/ ... t ,t }.* ), (y ,>=,,; , 3 , w ),
KwilH of Mm llnvn H
o Hhow thin I mi nark Lliul., writing
llio (!iniLtion uf Mm Hiirfiuui may bo wriUun
4y (//  j) (/y  *) ( (V  w )
+ . (,'/ J (12/8'fi' 7 ') . rt , , iH 7 'fi' . 278'*
I 2 ; V ( <i#'*S' ). 2/j VK H. !) 7 '')  h V + ^7^' J  W)
(.'.'  ?)' J (rtJ  WJ)' J (j H/)' ra 0,
whoiHfo ttlwui'vhitf Muit 7' JH oi 1 t,lin unl.'i I inul ^' !' .)iu onlor 2 in (. i/), (;.. y )
ronjtiinlily, i<m;h U'l'i.i tif Uio (]tliUinii in ul, Inmt uf the wrrun.l unlnr in (;'/). (j '//)
itnijointly; or w hnvti y,!^;, a iiu.lal linn; an.) Nintilnrly Mm nth,, Mvn ' MIMH urn
uoilal liinH,
121. Tho fim^oiii^ LrniiHluniidt iiiu,i.in is must iviwlily oliUiitotl by vnvnrtinjj in
tho cubio in 2 1 , W, vi. writiitK ;i ".*!/. ;*//, w iV( un.l LhcrHnro * + ),
J^'/Hr, wy*M, Mnj ttnlm; f'uncdi.ii duaLi>f tinmill J'm V\ylf) I
{ K+ ////) 1^  ( K / V
writing /3', 7 ', 6' m Jt + ,  , /!; h r. ;/, Uiu fut.'l1i<nciiiH unt (,'( (,t i 1),  // f '
HIM! tho uquation of tho Hiii'laou in thun ohtainml in tin; I'unu
27 <. + ])"'
which, arranging in powera of . ad ravoraing thi mgu. in tho fnroguinu
on
result.
420 A MEMOIR ON CUBIC SURFACES. [4.12
120. The three transversals are each facultative; p'~b'3\ t' Q.
121. Hessian surface. The equation is
The complete intersection with the cubic surface is made up of tho lintw (F0,
X = Q), (F=0, Z=0), (7 = 0, W0) (the axes) each twice, and of a aoxbic oiirvo which
is the spinode curve; o'=6.
The spinode curve is a complete intersection 2x3; tho equations may in iuoli bo
written
P + P (X + Z + IF) + 4&XZW = 0,
the nodes D, G, A are nodes (doable points) of the curve, tho tangnnts at cnoh nod,,
being the nodal rays,
Each of the transversals is a single tangent of the spinodo ourvo; in mt f..r
e transversal 7 + J + ZO. F=0, these equations of couL satisfy thi equation of
fcllt ^ ' the equation of tho Ho.snian, wo havo
= > ^0, F = is a point on the axi.s IF=() r
. i i . , ' J v " '" MU"ib uji me axiH r = () ) = n
not belongmg to the .prnode onrve ; wo have only tho point of contact .F + ,TU = o'
M' = 0, JC ^=0. Hence ^ = 3, '
Reciprocal Surface.
122, The equation is found by means of the binary cubic,
viz. writing for shortness
this is a binary cubic (jy f ff)", the coefficients whereof are
3(0+1), %j8, o^ + % .3^
and the equation is hence found to be
=
878}
^
A M15MOIU ON (MJIJIC) HUIIKAOEH,
127. Writing ( ( /,, b t o t tl$X, Kys^/.A'  Y)(f.X V) (./',, AT  V ), Mm
[ 7]
[ H]
, , ,.
and tho ILIUM art!
/, A' !'(), [2fi]
,/U' y0. ;i(i];
A'  0, y o, (())
/,A r  K(l, % 0, (I)
/ a A r y0, # 0, (2)
/aA'yo, ^ o, (:i)
/,A r  T.0, irt(>, 4.)
mu
I2H. Thovo in no lli<!nltulivn linn; p'
liif), H'dHMitui mirliini!. Tlio c(tiii,inu
X %]\r(oX h (/l r  HA" (j
H. bluit Uid JluHHinu liivukH up int.ii ill.
into a unMo
nf Mm
Tin! i:iniil(i ( n
Mm lino A r (), F() (Mm iixiw) lour tiimH;
Uiu HphmiUi c.iirvn; o'^.S.
In that c
and
tlm oiiiuliiHiH
(, 6, (t. (/'ox, rr
and
ulnu 9' = . 0.
0,
(axinl .i <.iinmnu l.m>luno)
witli thn o.Hblu sui'lmi.s is i.uulc np .if
J nl' u Hyntom nf tnur wmit.n, wliinli in
f" l,li
, ry o,
order wliiuli lnvakn
into four <smicHj fur wo <tuu fnnn thn two i!iimti<inH
(a, 6, 0 dJtX, y)*(oA r Kn'}'hHA r (o6.
that IH
(4aoafc, ad, bd, cd, tF&X t y )'(),
a system of four plnncn each inUswisntiupf fcliu cubic A / ^lK + ((/ J ft, a, tl$X t K)>0 in
tho axia and a conic; whence, a* above, Bpiimdti curvo is four conies.
It ia oaay to HOC that tho tangent planes along tiny conio on tho tmrfoco JIUHH
through a point, and form therefore a yuadrie cana; henco in particular tho wpmocto
torso is made np of tho quwlric conca which touch tho Hurfucu along tho four
rGBpootively,
422 A MEMOIR ON CUBIC SCTHVACI'H
125. The cuspidal curve is given by thn equations
S(ft + l), 2ay& ?/ + '?
or sa}' by the equations
that is
a (  3) f + 4y  3 (a.  1 ) 7 0,
and '
3 (a + 1} S H (2y H $) ((/*  iy) =3 Q I
consequently o' = 6. It is to bo addod tluit the
section, 2x3.
Section JX = 12
Article Nos. 12G to 136. Equation
126. The diagram of the lines and plimoH in
41 a
oumi ix a ,n,mIulo inln
'
Linos,
IX 1
22J? 3 ,
^ en
X
M
X
CD

II
"^1 CO
II
(O
1x0=
' 1
Ooinmou hljilnuv, OH.
oitlfiriiluiiKtluinxk
7
,
8
2x6=12
I
Otlior bii>luium of tint
tWO blllOtluH I't'lJllllot.
ri
ivol.y.
<8
(
25
3x9 = 27
* I
I'liuiCH onoli Hiionuh
llioaxlmuuloonlnlii.
30
' '
jii({ nyn of tho two
45
11?
!r
P si
Cl o
f
K&^
3t
w, a (0 H
o
?&
^ F
rt (B 7
~~ _L
" """I'.I.I.M
412]
A MEMOIR ON CUBIC SURFACES.
425
133. The cuspidal curve is a system of four comes; in fact from tho preceding
equations written in the forms
(ad bo,
eliminating zw, we obtain
t l) a =0,
2 ( ad*  Sicrf + 4c a ),
\CtfGCb ~T" u (L "" jD(s ff
6 (6c arf) 6,
= 0,
which shows that the cuspidal curve lies in four planes, and it honoo consists of four
conies; these are of course the reciprocals of tho quadric cones which touch the cubic
surface along the four conies which make up the spinode curve.
134. The equation of the surface, attending only to tho terms of the second order
in y, g, w, is 27dV,sw = 0; it thus appears that the point y = 0, &**Q, w0 (reciprocal
of the plane X = 0) (which is oscular along tho axis joining the two biuodcs, or
.SBaxis) is a binode on tho reciprocal surface, the biplanes being z = Q, w = 0, vis:, these
are the planes reciprocal to the binodes (A' = 0, Y=0, 1^=0) and (X = 0, I 7 = 0, #=0)
of the cubic surface ; we have thus B' = i ,
It is proper to remark that the binodo ?/ = 0, #=0, w = is not on the cuspidal
curve, as its being so would probably imply a higher singularity.
ISO. A simple case, presenting the same singularities aa the general one, is when
a = d, 6 = G = : to diminish the numerical coefficients assume a = d = fa, the cubic
surface is thus 12^^^! A' B + P= 0, mid tho equation of tho reciprocal surface,
multiplying it by 4, becomes
viz. this is the surface
(a; 3  f) = 0,
(o (a? + Szw)
(3ffi s + 0w) a =
considered in the Memoir " On the Theory of Reciprocal Surfaces."
is, as there shown, eomposed of the four conies y = 0, 3a; a + sw 
a 3  ssw ; and it is there shown that the tw o points (to = 0, y = 0, z 0)
each reckoned eight times, are to be considered as offpoints of the re
C. VI.
424
A MEMOIR ON CUBIC SUIIFACEH.
412
Reciprocal Surface.
130. The equation is obtained by means of the biimry wibio
XfaX + yVy + faufa, b, o> d$X, )') ;1 ,
viz. calling this (*jA r , F) 3 the coefficients are
The equation is found to be
432 (fftp  Galcd + 4c 3
1 216 [(ad*  36cd 4 2c 3 ) ^
+ 9 [3tfV  12<K&fy + (lOh? + Sc a ) icf  (lad I Me) w?/ 3 + (4(w  ft") i/'J j
 I/ 3 (c/ft? 3  Scflft/ + 36ffiJ/ 3  fl.7/ 3 ) =s 0.
The section by the plane w = (reciprocal of # s = .0) is tho
of edge) three times, and the lines w = 0, eta 8  Se<ify + B/M^"
biplanar rays). And similarly for the section by the planu c
The section by the plane ?/aO is made up of the linos
each once, and of two conies, ?/ = 0,
(), //
(ruri
imiiial
j . 0),
i>mi'iil
of thi
16 (a
86V)
4 rfW * 0,
181. There is not any nodal curve; &' = 0,
1S2. Cuspidal curve. The equations may he written
?
Forming the equations
(od  lo) . 144*V + (Sate"  2 Cffl y  6^) . ;i 2 2W a ^a ,
these are two quartic mrfiu* having in common tho HnoH (. 0 W 0) 0/0 0).
attending to the he (w = s^tn tin's k n M, \ H/  U J' ^^". * M ( U :
: 5 u ' * u * tms 1S on "w second Hiiifuci) mi oHoular line,
' on fche fil ' st "&ce it is a nodal line, tho ono tangmifc t ,lano boi, W
.^ 0 the other t,ngent plane being ,,0, but tho Huo being in
regard to this sheet an oscular lino, ^
42G
A MEMOIR ON CUBIC SURFACES.
[412
136. The like investigation applies to the general surface, and wo have fclnw
& = 16 ; the points in question are still the points (a = 0, y = 0, e = 0), (ss = 0, y = 0, w = 0) ;
viz, these are the points of intersection of the surface by the line (us = 0, y = 0), which
points are also the common points of intersection of the four conies which compose)
the cuspidal curve, that is, they are quadruple points on the cuspidal curvo ; it does
not appear that the points are on this account, viz. qucl quadruple points of the
cuspidal curve, offpoints of the surface, nor does this even show that the points ahotild
be reckoned each eight times. As already remarked, the singularity ioquiros a uioro
complete investigation.
Section X = 12^(7 a .
Article Nos. 137 to 143. Equation WXZ + (X + Z)(Y*~ X*) =0.
137. The diagram of the lines and planes is
M
FO M
Lines.
w
o
X=12
<]l M
X
1J
II
__
X
if
X
*>.
II
co
X
en
II
H"
X
CO
II
CO
1x12=1
*
Biplane touohtiiK along fixin,
and containing odgo,
3
1x12=12
1
Other biplano.
g 22'
1
1
2x 8=16
Planes oaoh through iho
axia and oontainiiig a hi
planar ray and a onieno
dal ray,
8'
lx 3= 3
* '
Plane touching along tho
edge and oontnining tho
more lino.
6 45
 _
Bimdial plane through tho
two cnionodal rays.
&
ID
CmcnotJal rays.
Biplanar rays in the non
axial biplane.
Edge of binode, being a
transversal.
O $>
l>
O
a
<
I
A MEMOIR ON CUBIC SURFACES. 427
138. The planes are and the lines arc
X = 0, [0] A r = 0, F=0, (0)
Z = 0, [3] A r = 0, #=0, (3)
x  r o, [irj x  F= o, z = o, (i >
A r + r  o, [22'j x + F= o, z = o, (2 >
Tf = 0, [3'] A r F = 0, W = Q, (!')
X+^0, [1'2'J A r + F = G, Tf = 0, (2')
A r + # = 0, W**Q, (12).
139. The facultative lines are the edge counting twice, nnd the mere lino ;
p = Z>' = 3; i' = l.
140. Hessian surface. The equation is
X (X + Z) (ZW + 3 X* ~ XZ} + F* (A'  ^) = 0.
The complete intersection with the surface consists of the line (A r = 0, F = 0), bhe
axis, four times; the Hue (A r = 0, #=<)), the edge, twice; and a sostic curve, which
is the spinode cui've ; c' = 6,
Writing the equations of the surface and the Hessian in the form
X (ZW + F fl )  Z* h # ( F u  A' a ) = 0,
we see that the equations of the spinode curve may bo written
viz. the curve is a complete intersection, 2x3,
Y f Z\1 y f 7\n
There is at B t a triple point ^^f^J . w == ~(~W) ' Mld afc a n double P" lb *
the tangents coinciding with the nodal rays W = Q t F" A' a = 0.
The edge and the mere line are each of them single tangents of the spinode
curve. But the edge counting twice in. the nodal curve, its contact with tho spinode
curve will also count twice, that is, we have $'=2.1 + 1, =3.
Reciprocal Surface.
141. The equation is obtained by means of the binary cubic
4wZ (X + ZJ + fa)Z (X + Z} (&>X + zZ) + tfXZ* ',
or calling this (*JX, ^) 3 , the coefficients are
542
48 A MEMOIR ON CUBIC
;uid thence the equation is found to be
ft'  5*) f2 (as ~ 2*) (a; 
where the section by the plane = (reciprocal of bimxlo) in
the line ^0^ = (reciprocal of the edge) four timon, and the
(reciprocals of the biplanar rays).
_ The section by the plane ,= is found to bo *  ti <.
>> v\ r rhi,
we have
thatis,
n .,/, ,,
two values, .
143. Cuspidal curve. The equations
are
pwon " complete
= onco,
A MKAKHlt ON <U!II(! HMItl'AOIX
Hwl.ion XI,:, 2 ff,.
Arl.idn NUM. ill m 111). Kpml.i.m \\' X Z *\ K 3 #H A'"
Tim itinerant i' rim liaoH run! plnui'H is
!! H
r',
M
ii 
II
!'i'
P ,p
p
1 .1
a
t' !'
11
F t
rf
Nj Nj
li ii
"3
II IF
O 'S>
. K.
O
XI li.
./V
0=1 U,
~<
>:
.
t.'
II
Ol
1'lmii'n nic
li h1
. i
II
i~
?ji
.V' (1 i]
] ;: Ifi 1 T.
Oxiiiilni' l>iilniii'.
.V II It
!:.. m
Ol'diliinj,' liiiilll( l .
a ifi
,,.{?
N
b w
r'
;;;
i
^ n
' flj
IIM oj 1 Ihn li
ni'N utiit 
llUH'S lill!
.nllUWJI j
i tlin margins <
f l.lui Uiu^v
gn in a lilt
nlliiUvii 1
till' (''Mill
liitf Ilin
"' L.iiitc.4; ih in
vilt npjinitr
145, Tim
l.lw ilimtiiHKion ni' lint )v(.!iivoi'iil nm'liin', Tlit^riHini (>' ft' . M ; i'^. ].
14(1. Hosriiuii mivtiuio. Tliin in
ii]i iiitti #(), Mil! iiHtiular biplniii', ami intu u cnhio Hiu'l'uco (itsnir n, Hiirlhai
XIil2 /i,,). Tlio mtiiijiltitii infcijiww.tion with tho itl)in HiirDw^ IN nuulti up ' (Jin
lino A'"0, ^r<() (tlio wl^o) nix linion, uiul of a iVHidiml HxU (^^ M (tonioH), wliiiih in
tfiu s[iiit(nlo (inrvij; o' (I.
'Hiu ctipiatiniiM uf Liu.) Hoxbiu uro in I'mit A r ^Hr j :^0, A'' 1 f J?* se 0, HII that Lhin !ti
uf tlirtsu ituiiifH, uiuili in a jilaiiu paHsing tln'ciu^li UK; (.td^u.
Thu otlgo timohoH enah of bliu fchrou nonicw tit thu puini A'^0, J?f
ii nuiHt bu ixickuijud jw a ningla tiuiguiit of tho Kpinodu uurvu, and
ihmo
430
A MEMOIR ON CUBIC SURFACES,
[412
Reciprocal Surface.
147. The equation is obtained by means of the binary cubic
. . L ,
VIZ. It IS
482w"
or, completely developed, it is
+ w"
,r
148. The nodal curve is the Hue = ft w 
th ^ & In to the fo 
^' Vlz  ther e are two aheets oaoulatinir alono
in thi S H Qe taUen three tin,, "
3 For the cuspidal curve we have
giving
= 0,
the
= 0,
.
' """""  ' 
* 
r
,, ,,
6*"
s= o
412]
A MEMOIR ON CUBIC SURFACES.
431
Section XII = 12 U s .
Article Nos. 150 to 156. Equation W(X +
150. The diagram of the lines and planes is
XYZ= 0,
Linos,
XII.12.tV
W to M W 10 M
OTl W
w
X
X
M
co
1
to
*
1x32 = 32
. . .
Uniplano.
a
o)
3x 4=12
Planes uaah touoliing
along a ray, and con
taining a mei'oliiiQ.
3
;
Ix 1= 1
5 45
Piano through the three
more linos.
i
I s
a>
ta
t !
p EJ'
tf
(D
161. The planes are
X +Y+Z=Q, [0]
A'=0, [1]
^=0, [2]
Z = 0, [3]
W=0, [1'2'3'J
The lines are
o, (i)
= 0, (2)
) F0 ^
= q,
0, (2')
z = a, w = o t (3').
162. The three mere lines are each facultative: ^ / = &' = g; t' = l,
153. Hessian surface. The equation is
 Jl " J
via the surface consists of the uniplane X+F+# = twice, and of a quadric cone
having its vertex at U 6l and touching eocli of the planes X=0, F= 0, #=0.
132 A MEMOIR ON CUBIC SURFACES.
The complete intersection with the cubic surface ia imulo up of (IH; my* omsh l,wi<
find of u residual sextic, which is the spiuode curve; o'=(>.
The equations of the spinode curve are
X* + F + # 27^ 2&L'  2XY = 0,
vix. the curve is a complete intersection, 2x3.
Each of the mere lines is a single tangent (as nt imoo appoint* by writing fur
instauce W=0, Z = 0, which gives (7^=0); that is, #' = &
Reciprocal Surface.
154. The equation is found by means of the binary cubits
 writu* for shortness 4 ( r  Z 0(^0)(^'D)
/3 = rc + 2/ + 3j
7 = y + zts + <sy t
_
then the cubic function is ~*
(12, W _4A Ay, 12SJ2',
aud the equation of the reciprocal surface is found to bo
432 S 2
I 64 y
(w  4/3) 3 S
+ 72 (w 4/3) 7 $
expanding, this is " ( w */?)V0;
w'.S

above.
12 J A MEMOIR ON CUBIC SURFACES.
156. For the cuspidal curve we have
12 , w4/3,
w  4/3, 4y ,
48y ( w
433
or say
whence the cuspidal curve is a complete intersection 2x3; c' = 6.
Section XIII = 12  B.,  20,.
Article Nos. 157 to 164. Equation WXZ+Y*(Y + X
157. The diagram of the lines and planes is
o
H>
to
( W
Linos
W M
o
O5 CII
Col M
M
M
to
X
to
II
fii
II
X
II
to
X
IT
M
10
1
2
2x = 1
'
Biplnnca.
OCR
1112 = 1
Piano through the three
axes.
C
s c
1
2x 0=12
*
Pianos each through an
axis joining the biuoclo
with n omonode.
M
Ix 4= 4
Piano through the ftxis
joining the two onia
nodos.
12
Ix 8= 8
' '
Planes through the hi
plnnnr rays.
x 2=i S
8 46
?lano
TransversaL
Cnicnodal rays, one
through each cnionode.
Biplauarrajs, one in each
biplane, and being each
a transversal
Axis joining the two cnie
nodes.
1
I
J Axes, each joining the bi
! node mth a, enicnode.

O. VI.
A MEMOIR ON CUBIC SURFACES. [412
158. The planes are The lines are
*0, [1] Z = 0,7=0, (5)
z => [2]
^ = 0, [056]
Z = 0, 7 = 0,
(S)
7 = 0, TF = 0,
(0)
X = Q, 7+^ = 0,
(1)
= o, r+^ao,
(2)
> [6]
7F=0, [34] W=Y=Z, (S)
X + Y+ZQ, [12] F=7=l r , ( 4 )
lf =  [] W = ; X + 7 H Z = 0, (012)
159. The transversal is facultative; />' = &'=!, (' = 0,
160. The Hessian surface is
The complete intersection with the surface is made up of the line 7=0 X
(fl*r three tnnes; the line 7=0, ^=0 (^axis) tlL times; li.o F=0,
(CCaxis) twice, and of a residual quartic, which is the spinode curve ; ff ' = 4,
161. Representing the two equations by U=Q, 7/=0, wo have
and (37 + Ar ^)^^^ a (^ a ^7Z+47^ + 4^X JfT suppose,
^^
whence  + 28^^+80^) 27P (,Y + ^
27 (X + ^) fr+ 9 ff + (_ 9X7
or, as this may also be written,
277'
_ ^
)}.0 J
and we thus obtain thp pmioiiA ~e >.i. , ,
equat.cn of the res.dual qu ftrt io, ov spinodo curve, in the form
4Z + ) =
'* l ^J A MEMOIlt, ON OUJUO HU.HFAOIW. 435
Thu Hpinodu uurvo is bhiiit a uomplotn intm'soolion, 2x2; HIM! ninao bho lirab surface
a comi having UH vni'tox on bho Hucond Hiirfiujo, wo HUD moreover that tlio npinodo
J'vo iH a undiil (iuitl)'i([(iiidna IiiHtoml of bho Itml; (iquabion \vu may wrlto inoro
omvti
0.
'lilt! iH[ual,ii)iiH nf him
iionH iif l,lu> HpiiHiilu
uro IK^ 0, A' I Y \ %=* 0, ami HiibHti luting in
vti \vn obtiiiii from ouoli ixjualion (A F #) 3 0, thai i,
l; of tho
1(12. Tim (tiiuulion of Uni oiibic m dorlvod IVoiu l;hab boloii^ing to VlralgJ^f/.,
hy writing thin'itn u**b  0, o^, (Z 1. Milking bhiM cluuigo in the Ibiinulio for b1u>
iHKiiiM'iJt'iil HurOitio if tin. cumi jimt roHiirtuI lo, uu huvn
JV ..a  4;H 1J ,
V . l(U 1J
ami HiihsliUitih^' in llir i<iiiaLion
= 0,
(<iiiiiLion divuioH hy c^; in Lliruwinjf IhiH oiib, thw
n
thw i
!l . Hi ;
+ w {
?* H
45 s )
552
436 A MEMOIR ON CUBIC SURFACES.
The section by the plane w = (reciprocal of B 3 ) is w = 0, y=0 (the edge) three
times; and w = Q, y~x = Q] wQ, y~s = Q (reciprocals of the CBaxes),
103. Nodal curve. This is the line y = x ~z\ wherefore &'=!, To put the line
in evidence, write for a moment x = y + ct, z = y + y, then the equation is readily con
verted into
= 0,
which, each term being at least of the second order in , 7 (aj_y ( z y\ exhibits the
nodal line in question.
164. Cuspidal curve, Multiplying by 27, the equation may be writtou
(1y3x3z~5w, y + Qw,  w% 2 + 162/zy12aw12swj + Uw\
 20/ + 2%aj + 24?/2  27^  8mo + IQw^ = 0,
where
and we have thus in evidence the cuspidal curve,
f + 1 Qyw  12 (a; + ) w + 16^ = 0,
 20/ I 24y (* + *) 27a  8j/w + IQ W * o,
which is a complete intersection, 2x2, or qadriquadrie curve; o' = 4.
412
A. J1KMOIU ON
437
Suution XI V~ 12 
Ai'ticlu NOH. 1(!fi to 171. ISipiatimi
Tho diagram of tlm HIU>H and plains in
^
N
i
II
ii
p
II
p
II
p
<n
*i
M
N
'I
In
(fl
II
11
o
o
P
ro
w
.0
M
o
ill H
M
M
M
\ I V ;.: 1'' .  //
( '
X
x
>:
>;
H
M
11
II
O
o
ii
II
li
M
II
1 11
Cl
O
O
I'lnntiH nro
/:..() I
lxlfiM
...
TinHill hi])liini!.
A':.:0 01
1K*.,
Onliimvy l)l]ilfuu',
r. :0 onit
1x10: 10
'
I'liinii lluoiiuli uxin niul
Din Lwo riiyH.
'ii i/i
O
ft)
M
1"
(3.
*^l'
P..
H
j p?
9.1
!"'
1
(T
'?'
"pi.U
'i'o l.lii 1 i'inaM(iim of tlm jtlitiiMH niul MUCH n! Hhitwn in l;hn
l(i(i. Tho <tlg{s IH a tiic.nltal.ivii lints HH will appear I'nua tlm diwiussiitn of t.lm
107. IlcHHiaii Hurfiuus. The equation is
m'# u + )'"^  8 A' 9 )'# I A' (I.
Tho (intnploin inturHtiOlion wifih the Hiirftum IH intultt up of Uio lino A r 0, K
(tlut nxw) live liuiow, the lino AT0, ^?^0 (tho eilgo) four tinuiH, and a nkaw nuliic,
tho orpmtionH nf wliinh may liu wriLlon
A', I', If 0.
, A", 5F
Tri fact; frarn th oipuitiorm CT0, jff0 we doihico HZUm X* (X* + *Y%) Q \ and
if in Z/0 we write Jf41'J8f ( it bocomcs ^ (AT IV + :)') 0; and thuu in nU~i) t
writing 5F a ^ A'TK, wo have
438 A MEMOIR ON CUBIC SURFACES. [412
1G8. I say that the spinode curve is made up of the edge X = 0, Z= once,
and of the cubic curve; and therefore a' = 4.
In fact in the reciprocal surface the cuspidal curve is made up of the skew cubic,
and of a line the reciprocal of the axis, being a cuspnodal line, and so counting
once as part of the cuspidal curve : the pencil of planes through the line is thus
part of the cuspidal torse; and reverting to the original cubic surface, wo have the
axis as part of the spinode curve : I assume that it counts once,
The edge is a single tangent of the spinode curve; /3'=1.
Reciprocal Surface.
169. The equation is obtained by means of the binary cubic
tw& (Xas + Zz} + X ( YZ  wX}\
or, as this may be written,
(3w>, 2yw, f + 4*ow, 12mJ_X, ZJ,
The equation is in the first instance obtained in the form
tys (f + 4&
* (f + 4,CTy)
but the last two terms being together  4w s a (y 3 + 4flw)", the whole divides by
and it then becomes
f 4
+ ffi(2/ 2
or, expanding, it is
w" .
The section by the plane w = (reciprocal of 5.) is w = 0, y0 (reciprocal of
edge) four times, together with w = 0, ^ = (reciprocal of biplanor ray).
The section by the plane = (reciprocal of ff.) is (f + 4flw) = 0, vi this is
*0, ^ + 4w = (reciprocal of nodal cone) twice, together with *0, w0 (reciprocal
of nodal ray). l
170. Nodal curve. This is the Jinej y = 0, y0, reciprocal of edge. The equation
in the vicinity is ^~^^/~~w^ showing that the line is a cuspnodal
line counting once in the nodal and once in the cuspidal curve; wherefore &'!,
412 ] A MEMOIR ON CUBIC SURFACES. 439
171. Cuspidal curve. The 'equation of the surface may be written
where 4aj . 3w  y* = 1 2a?w  if, This exhibits the cuspidal curve I%aw ~f = 0, $zw + 8a# = 0,
breaking up into the line ?o = 0, y = (reciprocal of edge) and a skew cubic; the line
is really part of the cuspidal curve, or c' = 4.
The equations of the cuspidal cubic may be written in the more complete form
8a;
Section
Article Nos. 172 to 176. Equation
172. The diagram of the lines and planes is
= 0.
tsj
H
o
b
B 1
o
o
o
(D
w
ta
^
XV = 12
r/ 7 .
Wl M
M
M
X
X
X
_A
o
ftnos aro
10
ll M
o
M
Ci
XssQ 21
.
.
1x40 = 40
*
Uuiplano.
#=0 28
Ix 6= 5
* '
Plnno touching along
the single ray.
2 IF
to
K
H
S?
B'
i
p
V
e.
P
1
4
where the equations of the lines and planes are shown in the margins,
173. The mere lino is facultative: p' = &' = !; f' = 0.
174. The Hessian surface is
viz, this is the uniplane Jt = twice, and a quadric cone having its vertex at U~ yt
440 A MEMOIR ON CUBIC SURFACES. [412
The complete intersection with the surface is made up of X = Q, 7=0 (torwil
ray) six times; A' = 0, = (single ray) twice; and of a residual qu'artic, which is
the spinode curve ; a' = 4.
The equations of the spinode curve are XZY* = Q, X W + 2# a = ; the firat
surface is a cone having its vertex on the second surface; and the curve 'is ttuw a
nodal quadriquadrie.
The mere line is a single tangent of the spinode curve; /3'=1,
Reciprocal Surface.
175. The equation is obtained by means of the binary cubic
viz. throwing out the factor y, the equation is
w 8 ( G4s; 3 ) + w ( 16aV + 72<e^r h 27?/ 1 ) + %V = 0.
The section by the plane w = Q (reciprocal of ff 7 ) is W = 0, ^0 (reciprocal of
tori ray) three trmes, and w = Q, y = (reciprocal of single ray) forioe. llMlplOClU
Nodal curve. This is the line .^ = 0, ^0, reciprocal of the mere lino: 6' = 1.
Cuspidal curve. The equation of the surface may be written
whera ( 64a > " 162 > Swja' + aflw, OT,*
4 . 64a;( 3w)  25G* 3 =  26fi t*
176. {The equation
resembles that of a quintic torse, vi z , the equation of a quintie torse i
s
y j zztsf =
which equation, writing 9y for y> 8* fo r , and ^ y fop W)
or, what is the same thing,
and developing, this is
only ia tl ,c
A MNMOIR ON (JUHIC HURFMMS,
441
Huotion XVIlii'l.a,,
Arli<iUi No. 177 to 1HO. Munition ll r (A r r XZ+ Y
177, Tho iliiigniiM ol' Uio UIIHH und plains in
! !' I' ' ' f if if 
"I N N , P P P P o
P P P ,'
M K N o
N 1 ^
II II II
O O o
=1
O O O O
if lU ,u
w ii to
XVI.. M'J..
I'huitm nut
lr: 8 .
l W
X
II
^1 W
X
tl
li
'/,\" IIV.'II IU
rl M r ;i) III
.
ritf no n
A'lll'^O Mil
llx'J.U'J
I'lfuii.'H cnoli tmiiiliiiiK nloi
itn nxiH, niul conlninii
n tvuuHvui'mil,
A*.IXMO ui
'
A'M'O ill
.v..o i
* >
\ :..0 U
.U.H,,,,
! ' .
I'lftllCH (Jllflll Ull'Ollull till'
ftXOH.
A'.. rixi iriit iviin
'ili
. .

I'luiin tlirou^h llio llu p
i i jr.
.' .' I
Li'niinv(ti'Hnln,
H

o B
Q *
L
iB'el
K M^
ffi is rt
s
!
a
E,
o f
I
whoro the oiuatioiiH of the linc nnd planes arc shown in the margins,
0. VI,
442 A MKMUlll ON ('inili: m
178. Tho tmiiHVOiwlH nro wmli tiuiiiltutivn: /' // ;i ; /' , ,
179, Hosaian miri'mso, Tim njimtioii I'M
*XYZW(X+y+%\. II')(,II'AT I \\'XfS i mV? ,
or, what is the siuno tiling,
,Y(1^ >l I'll' i ^11')
i^.v .(. ir.vi
The complete intonation with Mm oiihi,, H.trliin, i, .MU,],. (t( ,
twice, and thoro i. no H pino ( lo um,  ,'.,. ^, w ..... ,, i
i 180, Tho equation w inu.u.tmtuly uhiniii,.,! in , l( ,
,,!.. . . /';! /v 4'^ 1 \Vt,(t,
oi 1 ratioiwhfflng, it i.s
(^+^ + ^^. ,,,,.. .,,, ...... ,,.,,. ^,
so that this is ;,,  llct , S t oilU)1 .. H , 1111H . tin m . i( , t ,
Nodal
ourvo. Thin c IIK iH, a (lf u,,, n,,,.,,
,. = (); i ""' l '''".
'To pit any ono of tl mi ,, f m . ,', ,,,,,
wnto tho 0(1 ,,ati, m ,,f t l u , Hrtitt, in lln, f,,,',,, * '" """""' '" " vi>l ........ ' w "
There is no cugpidftl curve; '0.
412]
A, M18MOIU ON CUBIC! SUBFAOR8.
443
Hoobion XVn1227f fl 6 Y fl .
Avtiolo NOH. 181 to lfi. K<iuiiti<m WX%+ A
181. Tim diagram of tho limw and plauon i,s
'" = 0.
n n
ll' ll'
ll' r
O O
p p
P B
t *
ID
H H
53 *N
IV ">
1 !
if II
n' a
n it
o ro
11 li
O
rt. W
N l
o
xvu.ia y
lly'Cy.
1
X
X
X
OS
II
CT1
II
II
I'lnnoH (mi
^ll O>
IP
M
to
A'nO
ix Oe 11
. . .
(Jin ni mm liijdiuio, tln'ongli
tliii 'iixiH joining tho two
hinodoB,
tfi'iO lit
ir0 ill
ax iiwia
'
Itonuiinlng ItipliiniiH, ono
for cnoli liinodu.
r n o <ua
'
1
1'lnno Ihvoutth tlio tlivoo
*
HXOK.
A'MV;0 Oil I
lx .,: I)
I'lllllO tllVOIIRll lllO ftXJH
fi 'ifi
joining tho two liinodon.
ffi
"a*
6:
WK "
3'^,
O' _ p.
So'
ft H
I* 6 !.
i* H'
'5!
h
^ o
(D
P O
E d
rj
P CR
A
CO
p*tr
O
whovo fiho e([imliuiiM tif tho HnoH and pianos arc Hhowu in tho nmrgiiiH,
182, Thoro is no fncultfttivo Urns; b'**p' = Q t i'0.
183. Tho Hossinn Burfiwo is
X ( WXZ + 8 7ZW + X r j ) 0,
vis*, this breaks up into JT0 {tho common biplnno), and into ft cubit
14 A MEMOIR ON CUBIC SURFACES. [412
The complete intersection with the cubic surface is made up of Z=0 t Y~0
axiH) four times, of 7=0, Z=Q and 7=0, F=0 (OSaxes) each three times; and
of a residual conic, which is the spinode curve; a' =2. The equations of the apinorlo
curve t are F';JW = 0, 4^+37=0; viz. it lies in a plane passing through the
/Maxis; since there is no facultative line, /3' = 0.
Reciprocal Surface.
184. The equation is found to be
or say this is
16V + (%"  3<ky + 27tf) ^ + y 3 (y  a;)  0.
The section by plane W =0 (reciprocal of B^D) is w==0) y(w)0 vi this is
he^hne w . 0l 0 (reciprocal of edge) three tinL, and ihe line Li) y v  o
(reciprocal of ray) once ; and the like as to section by plane * = 0. ~
The action by plane .. (reciprocal of C^) is itf== o, (^ + 4^)^0 vi, this
is the come (reciprocal of nodal cone) twice. ' '
There is no nodal curve; i' = 0.
185. Cuspidal curve. The equation of the surface may bo written
i 1 ^ 5 aUd thei ' 6 is thus a G 8 P^ oonio pia,o
Attending only to the terms of the second m ^ ftl ,
fe"*0j that is, the point V = 2 o n y> *' W> tho e( l llfltio11 ^ocoinoa
Knodo of the surface; or there is'the"^ 1 ^ 1 f the C mm " bi l 31 ) ^
<l 12
A aiKiMOIH ON (JUIUU
445
Snc.lion XVIII I :!/*
AHiuio Nus, LSI! t,,> IN!). KqimUmi W X % \ P* (X \ Z)
Tim din^nun of Mm limm iunl pitman urn
01
=8
o
O 3
1*
II I?
P ii"
Muno
tltn (lin><!
o mul unn
titinhiK
I'ltuiu tflueliliiK ftloiiK uxiH
tlm l\vu cnii'iHiili.'H uitil coutiiiiiiiiK
line,
whuro tho tiijuatiioiitt ol' tho lino and plunou 111*0 uhown in this iiiurghtu.
187, Tho inoro Him in fiuuilUitivo; tlu; erlge in ttluo fiwultafcivo counting twice (this
will appear froim the dwcuwion of the reciprocnl aurfiice) : &'/>' 8, i'l.
446 A MHMOM ON CI'llli' .'iriM'.Vi f
188. Tlio UOHHWII Hiirfacii) in
(.V'./o ir.v/i i.v t\ \ .1
Tho flomploto inl(>rai'i!.i<m with lJn< ntltir Miilti., j, J n
(tlio 6^iwoH) ouch lour liinoi; I'.n, II' <i ,/W/,ui,i ?v,i,^'
odgo) twice. Tlusn; !H nu H]iin,>i!<> ntrv. ,' it; Hl h ,.,,,i.. ,,!.. ;
^ ..... r , ( ,. , . .
f nodal cmio) bwi(H , . mitt Hili(ry ( . ( , ^^.^ ; '  '., * I
n ' t * 1 J'i*n.!l3 t,J
Nocl.il cum,. Writing U,, t , (
and writing tho cinmtiuu j, t j,,. j;,,.
or Kiiy ' J ' * " * M * s ^ :J( * f/'.V/i'.
Tho o(iiutiou I'M thus
(//*'4H'.r  IfCij.. li.[,.> ... u
oi 1 in an mutioiml liinii
, ,. (li , ()
lour fcinioH. ' ','i> I* ;/ ...n n,,. ,1,1,...,  ,,,
12]
A, MEMOIR ON OUimJ
447
Suction XlX:i2yy 6V
Article No. !!>() to 1IJM. Kquatiuu WXZ+
UK), Thu diagram of l,lio linen and pliimiH in
ll'
.P
ii 1 e
N
N P
II
o
II ID
O
XIX1.
,,
lil M
M
X
Ul
t
1'llUlOH lll'O
II
^ll Ol
K

OHcnliu 1 biplfinc.
fc o
IK HO ^110
Oi'tltimry l)i)lniio,
'H ir'
W
fc,
;e of tbe biao
c. 1 ?'
S B
01'
K.
F
JT> O"
g'
ft
whcns fcho otjiuitiouH of Iho IhioH and pltuiun avo nliiiwii in tho
I!) I, Tin; uxm to a ftwjulttvtivo lino cmuUin^ Lhrots tiuicH (aH will appouv from tho
reciprocal Hnrfticn); f)'~~b'~~ 1\, i'l.
1!)2, Tin; Histwiim Hurfticu IH
vi/,. lihiH in tho iimulftv biplano ^>0 and a cubic Hiirfacu.
Tho complete intovHocitum with tho cubic mirfaco is nuulo uji of A'^0, ^0
(thu odgo) nix timuH, and A r 0, K0 (tho axin) six tunoH. Thoro is no Hpinodo curve,
cr'^0; whence nlno /3'wO.
Surface,
103. Tho uquntioH is at unco found to be
(f + 4ww) 0,
44B
A Mi;,MnMt (IN
Nnilnl curve. Tlit i><iiati>'H j;i\." (
Tlii'iv is iin rii',iii!ul
\\
llUlt' in
X H ii
UHI. TJ
1111(1 tlu ' i lll
JH i
in,
412]
A MEMOIR ON CUBIC SURFACES.
449
Reciprocal Surface.
197. The equation is at once found to be
27 s 4 4icw) a  G4fy = 0.
The section by the plane w = Q (reciprocal of the Unodo) is w = t * = (reciprocal
of ray) four times,
There is no nodal curve ; 6' = 0. But there is a cuspidal conic, y = 0, a 3 + 4wtv = 0.
The point 2/ = 0, s = Q, w = (reciprocal of the imiplaue A" = 0) is a point which
must be considered as uniting the singularities '=1, #' = 2.
I give in an Annex a further investigation in reference to this case of the cubic
surface,
Section
Article Nos, 198 to 201. Equation WX
198. The diagram of the lines and planes is
N *< *!} 11
o g
xxi=i2a 3 .
II II II &
o o o
cwj w
X
CD
Planes are
r=o o
#=o s
ir=o s
Oomi
tin
1x27 = 27
t .
3x6 =18
4 46
. . .
Honii
oat
o rt
a'
where the equations of the lines and planes are shown in tt
199. There is no facultative line; p'=ti / = 0, f' = 0.
200. The Hessian surface is XYZW=Q, the oomir
biplanes each once. The complete intersection with the s
each four times ; there is no spinode curve, o' ; whence
C. VI.
450
A MEJTOIH ON CUBIC SURFACES.
412
Reciprocal Surface.
201. This is 27^0^ = 0, viz, it is a cubic surface of the form XXI = 1.2 ~ &#,.
There is no nodal curve, &' = 0, and no cuspidal curve, c' = 0. Moreover .#'= 8.
Article No. 202. Synopsis' for the foregoing sections.
202. I annex the following synopsis, for the several cases, of the facultative lines
(or nodecouple curve) and of the spinode curve of the cubic surface: aim, of tho
nodal _ curve _ and the cuspidal curve of the reciprocal surface, It is to bo observed
curve ffj* 8 6 ? 1 * a curve ' for instance > 18 = 4x62, this moans that it is a
JZAnf 6 ', ' 1S) the I 31lrtial "^section of a quartic surface and a quinliic
fliuiaLt, out without any explanation of the nature of tho (Mint 01*1
causes the reduction, viz. without explaining whether this is a eonicor a^mir of Tinm
SLl 1 " e1 ' """' "" ^ be ^ b >' 1 '*" CO to P^loL of 2
Facultative lines.
Nodal curve.
Spinodo curve,
CuHpldal onrvu.
1=12
27
27
12 = IJ x 4
~" 2i~(\ i ""
111=12 IL
15
15
12 = 0x4
184xfila
IV12 W
9
9
12 = 3x4
104xfil
V12~ a
7
7
10 = 8x42
4
7 = 6 + edge twice
7 = 6 + I'ec.of edge twic
I'co. of edge taonodnl
10 = 8x42
12=4x44
~ 3 ~ C 2
3
n
y TT 10 n
"
s: 8 x 4 H
lOsat x4~d3
s
3 = 2f.edge
3=2Hieo. of edgo,
= (3ikoM '
roc. ofedgoisouspnodii
841)10 Uffft
10=reo. of cdpfl H
uniQiirxal (tUifc,
111=123(7,,
a
rec, of o(l(o ja oimpnodnl
IX = 122B
u
8
= 2x8
<!2x B
x=i2^;, a
none
3=l + edge twice
none
3 = 1 + reo, of edgo twice
veo. of edgo is tnonoda
8 = 4 oonlcs
= 2x8
8 = 4 coniofl
XII = 12 tf fl
3=edge 3 times
8= roo. of edge 3 times
'eo. of edge is osonodal
8 oonios
O.^l..
3
8
= 2x8
02
XIV12
1
1
4 = 2x2, noddl qua
1=2 xS qnndiiqundrio
l=edge
l = roo. of edge,
ce. of edge is ciispnodnl
^Z
4=84100. of odgo,
xvi=i 2 . 40a
1
1 = 2x2, nodal qua
drlqundvlo
4=2 x 2 ouBpulftl qnn
diiqiifuliio
XVII=l22B s  0a
o
none
3
nono
nono
SI
3=l+edge twice
none
1 + i'ee. of edge twice,
wo. of edge tnonodal
2=conio
none
2 = oonlo
nono
XX=i2t; fi
3= axis 3 times
=i i eo. of axis 8 times,
i'eo. of axis osonodal
none
noo
XXI=l2 3j5s
 ,,
none
none
nono
2=conio
2 ss. oonlo
I pass now to
~ .
the two fflooo n c
none
" _
310110
none
A MEMOIR ON CUBIC SUMTAdm 451
Article No. 203. Section XXI]: = (1, 1). Equation X a W+Y*JS = &
208. AH this is a Hcroll there is here no question of the 27 lines and 45 pianos;
thoro IH a nodal lino A r 0, Y = Q t (&=,!) and a single directrix: lino, # = 0, TK=0.
The Hessian Hiufaco is A r *7 a = 0; the comploto in tors oc bio n with the cubic surface
is mado up of X=0, r=() (the nodal liuo) eight times, and of the lines A r = 0, #=>0 }
and y=0, W = t ouch twice.
Tlio reciprocal siufuco is ;%^y = 0; via. this is a like scroll, XXII = 8(1, 1) ;
c'O, &'!. V "
Artiolu No. 204. Section XXIII = S (17 1). Equation A r (A r !K+ YZ}+ Y*  0.
204. ThiH ia alKO a scroll; there is a nodal lino X = Q t y = 0, and a single directrix
lino united therewith.
Tlio licaniau Hiufiioe i A' 4 0; the comploto intersection with the cubic surface is
A'0, Y~Q (the nodal line) twelve times.
The reciprocal nurfaco iw ^(aw + ^)^ = 0; via. this i,s a liko scroll, XXIII = (171) ;
o'0, i'l.
.w Gontuininff Additional lleiieawhes in rcyu.rd to the case XX = 12Z7 8 ; equation
Lull the Hiirfaco bo touched by tlio lino (a, b, o, f, g t A), that ia, the line the
equationH whereof arc
ft, 0, /, b
ff> ~f> 0, G
~ a,  b,  G,
Writing the equation in the form oW . cX + X (oZ? + &Y> = G t and substituting for
oW, c% bhoir valucn in termn of X, Y, wo Iwvo
that is
or say
vix, the condition of contact ia obtained by equating to zero
cubic function. Wo have thus
270 4 (*  cffY
f 4o 9 (2a6 H o/) 3
452 A MEMOIR ON CUBIC SURFACES. [412
viz. this is
+ 27 o j o a
36 ab a cc/
+ 18 b
+ 27 c
which is the condition in order that the line (a, b, c, / g, h) may touch tho aiufoco
X*W + XZ Z + 7 3 =0; and if we unite thereto the conditions that tho lino almll pass
through a given point (a, ft 7, S), we have in effect the equation of tho circumscribed
cone, vertex (a, /3, 7, S),
Writing (/ ffl h, a> b, c) in place of (a, b t c, / g, h\ wo obtain
+ 80/yAa
+ 18 <7 B /i
+ 27 A
as the condition that the line (a, 6, o, /, </, A) shall touch tho reciprocal surface
and if we consider a, 6, G, f, g, h & standing for
Wfa azyx, faay, Samo t $y/3w> Sz
412] A MEMOIR ON CUBIC SURFACES. 453
values which satisfy the relation
( 0, h,  ffi (i J, & 7, S) = 0,
A, 0, /, b
/* /\
a, &,  c,
then the equation in (a, b, c, / y, A) is that of the circumscribed cone, vertex
(, /3, % ^); the order being (as it should be) ft' =6.
The cuspidal conic is y = 0, 4aw + 2 a 0, and we at once obtain a a 4cgr = as
the condition that the line (,, &, c, / </, A) shall pass through the cuspidal cone.
Hence attributing to (, 6, c, / </, A) the foregoing values, we have
a*  4c0 =
for the equation of the cone, vertex (, (3, 7, S), which passes through the cuspidal
conic; this is of course a quaclric cone, o = 2. I proceed to determine the intersections
of the two cones.
Representing by <B)==0 the foregoing equation of the circumscribed cone, and putting
for shortness
X = 27/i a (/ 3  bk)  % a (2/0 + A),
= :
I find that we have identically
(/a _ M) X + ((/  4a/i a  BfgJt?) (
whence in virtue of the relation af + Iff + oh = 0, we see that the equations = 0,
(t a 4c</ = 0, arc equivalent to
or the twelve lines of intersection break up into the two systems 1
and
a/00,
To determine the lines in question, observe that we
( 0, A, g, a$>, /?
A, 0, /, 6
454 A MEMOIR ON CUBIC SURFACES. [412
and we can by the first three of these express a, 1), c linearly in terms of /, g, h ;
the equations / 3 &A0, a 2 % = 0, 27A 9 (/> Wt) 2&*(%fff + ah) = become thus
homogeneous equations in (/, g, A); the equations may in fact be written
= 0,
^ 0,
= 27A 8 (S/ a  A S I 7 A/) + 2# 9 (/3/t 2  7(7/1  28/0)  0,
viz. interpreting (/, (/, A) as coordinates in piano, the first equation represents a conic,
the second a pair of lines, find the third a quartic.
We have identically
4/3 a S (S/ 3  /t u + <yA/)
and it thus appears that the two conies touch at the points given by the equations
we have moreover
= 4/3S (S/ a
2S/ 7 A) [2/3S/ (f 4
hence at the lastmentiowed two points  J SA 3 + f y<//i + 2S/r/ is =0; aiid the quavtic Jt
thus passes through these two points.
The conic (a a c#) = and the quartic A' = intersect besides (as is evident) in
the point t/ = 0, A=0 reckoning as two points, since it is a node of the quartic; ami
they must consequently intersect in four more points: to obtain these in the most
simple manner, write for a moment
then we have identically
(jh + 2S/) 3 ,
+ 4aS) (? 3
AJ.CUUB wiiBii a' ^cg=i(), we have
412] A MEMOIR ON CUBIC SURFACES. 455
and substituting these values in the equation X = 0, it becomes
viz, multiplying by l(i/3 3 8, and omitting the factor O, this is
27/t a n + 16.<% 4 = 0,
or finally
16'/% 4  27 (7 3 + 4aS) (ftf + 27W = 0,
a pencil of four lines, each passing through the point y = 0, h = 0, and therefore inter
secting the conic
(7" + 4S) (f + /3 W ~ 2/%/t  4S/i/ = .
at that point and at one other point ; and we have thus four points of intersection,
which are the required four points,
Recapitulating, the conic a s 4c</ = meets the sextic (f 9 bh)X~0 in the two
points
2/9S/ (7* + 4S) g + jftyA =
each three times, in the point #=0, h~Q twice, and in the four points
27 (7* + 4aS) ^ + 27^ = 0,
=
each once. Or reverting to the proper significations of (a, &, c^/, (/, A), instead of
points wo have 2 lines each three times, a line twice, and 4 lines each once; the
line 00, A = 0, that is, f/ = 0, A0, a = 0, being, it will be observed, the line
#= drawn from (a, A 7, S) to the point y = 0, = 0, w = 0, which is the
reciprocal of the uniplane ,Y = 0: the twelve lines are the aV lines of intersection of
the circumscribed cone a! with the cuspidal cone c', viz. aV = [aV] + So' + %' ; [a'c']4
referring to the lastmentioned four lines; </ == 2 to the two lines; and %' = 2 to the
line </ = 0, A = 0, a = 0, which it thus appears must in the present case be rentf^
twice.
[413
413.
A MEMOIR ON ABSTBACT GEOMETBY.
[From the Philosophical Transactions of the Royal Society of London, vol. CLX. (for
the year 1870), pp. 5163. Received October 14, Eead December 10, I860.]
I SUBMIT to the Society the present exposition of some of tho elementary principles
of an Abstract mdimensional Geometry. The science presents itself in two ways, us a
legitimate extension of the ordinary two and threedimensional geometries ; and as a
need in these geometries and in analysis generally. In fact whenever wo ore concerned
with quantities connected together in any manner, and which are, or arc considered as
variable or determinate, then the nature of the relation between tho quantities m
frequently rendered more intelligible by regarding them (if only two or throe in number)
as the coordmates of a point in a plane or in space : for more than throe quantities
there M, from the greater complexity of the case, the greater need of such a repre
sentation; but this can only be obtained by means of the notion of a space of tho
proper dimensionality; and to use such representation, wo require the geometry of such
space. An important instance in plane geometry has actually presented itsolf in the
question of the determination of the number of the curves which satisfy given con
ditions: the conditions imply relations between the coefficients in tho equation of tho
curve; and or the better understanding of these relations it was expedient to consider
the coefficients as the coordinates of a point in a space of the proper dimensionality.
A fundamental notion in the general theory presents itself, slightly in plane geometry,
t ! !^ ~: V S0lid ^   >e here the 'difficulty as I
" CU1 ' Ve n SpaCG ' r < to 8 P eak mora wemrately) to the
U f the tW fold relation bot * coord nates of a
that of a *feU relation.as dis
elation8 > se S ^ ^e expression
a stern o f e ni "^ BU h ^ tio " * "* of
a system of equations (or onefold relations). Applying to the ease of solid geometry
f i d ^ , 7 atl nS ( r nefold relation8 > se S e expresson
* "^ BU h
413] A atJOMOIK ON AUSTItACT GEOMMT11Y. 457
my (joiiohiHion in tho gonoral thuory, it may bo mentioned that I regard tho twofold
rolabiou of a onrvo in HKJU iw being uompletnly and precisely oxprefwod by moans of
a HyHtom of (MiiatioiiH (7 J ^0, Q=*() t ... TQ), when no ono of tlio functions P, Q....T
in a linoiir f'umstion, witli oon.stant or variable integral oooffioioiits, of tlio obhora of
them, uiiil whim wwi/ mtrfitce whti tenor whioh passon through the curve hn its oquation
oxproHsiblo in bho fniin If ^l./ 1 I .A 1 ^ . . . + AT, with coiwbanb or variable) integral
(souttiuiimtH, A, Jt, ... 7i r , It in Imrdly nouonHJiry to rumark tliat all the fnnobionH and
urn ljukttii to bo rational fim<stionn of the coortlinataw, and that kho ' word
linn I'idttniimo 1m lihu
Artiiili N(m, '!. to .M(J, Gwioml .f&nplnnulions ; .Huhtiion, Loons, itc.
\, Any in (uantUirs may bo rqirowMitod by nwaim of m + 1 c[imntitioH nn the
m of m of thcno to tlin niinainin^ (in. \~ \ )th cinantity, and thus in places of the
al)Holul;o vjdnoH of linn in ojmntitioH wo may (ntiiHidor tlio ration of m I 1 quantities.
2, III in to lit! notkiod that wo urn Ihrowjhmil <!onenriu:d with tlio ratio.4 of bho
m I 'I. cinnntitii'M, not, with tint abnolnto valuta; thin biimtf undorHtenKl, any montion of
tho ration in in ^cncnil nniuntcHnnvy ; tlnin 1 whall Hpnak of a relation botwoon the
tn *\< I (iianlitiun t nuiuiiug fchrnsliy a rulatiiui as rogardn the 1 ratioH of tho iiuatititioH ;
at id HO in utlirr cascH. It may almt bo unbind [iliat in many itmtanctja a limiting or
(KXtviMiiu caHo in HoinoliiiK'H inolndml, .soinotiini'H not included, nndor a gonural uxproHHion ;
tho goiuTnl i!Xii'('HMiini in iiitiindrtl t,n inohidi? whattwor, having ro^ard to tlus md)jcot
imitttM' and oontoxt, can bo inoliuhul mulor it,
It. roMtulati 1 , W(s may <!nm>uivn butwiMin bhu m {I iniantition a relation^).
4. A rolatinn w oithoi' rttyitlur, that w, it him a dolinito manilbldiHws, or, Hay, it
IH n /.:lultl relation ; nr C()HD it w irt'ai/itlar, that in, onmpoaod of volatioim not all of
th<! Hiuno inanif'oldnt'KH. AH to lint word " (iomnoHod/ 1 HOO jmt, No. 14.
I'iH
A MI;M"IU UN
,
(Illlt lilt HI:. /.N.f. .,1 Wliri
..., ,., tltt ,, ..... , tl
limy ^I..IM IM i,, t <ln,.. ,h., j,,,,^^^,^ , 411 ,
H Ml.' ruMhlinHlv,. , ,, M , ,,,.,. I l( il( J!
. . ifjft< ,, is i ,. ! ..^
^.j , ( A ..... ( ^, (j ,
; ; ^^^ (!j f fa' H i
I .
i,
.toli.lN IM till'
J
hl; v IH ^,,^4 Hi
th, ,,,.,,, ta 1(t ,
t. fl in,'h rvintion,
*
413] A MEMOIB, ON ABSTRACT GEOMETKY. 459
14. Any two or more relations may be composed together, and they are then
factors of a single composite relation ; viz. the composite relation is a relation satisfied
if, and not satisfied unless, some one of the component relations be satisfied.
15. The foregoing notion of composition is, it will he noticed, altogether different
from that which would at first suggest itself. The definition is defective as not
explaining the composition of a relation any number of times with itself, or elevation
thereof to power; which however must be admitted as part of the notion of composition.
16. A /(jfold relation which is not satisfied by any other /ofold relation, and
which is not a power, is a prime relation. A relation which is not prime is composite,
viz. it is a relation composed of prime factors each taken once or any other number of
times; in particular, it may bo the power of a single prime factor. Any prime factor
is single or multiple according as it occurs once or a greater number of times.
17. A relation which is either prime, or else composed of prime factors each of
the same manifoldness, is a regular relation ; a &fold relation is e& vi termini regular.
An irregular relation is a composite relation the prime factors whereof are not all of
the same manifolduess.
18. A prime /dfold relation cannot be implied in any prime fefold relation different
from itself. But a prime fcfold relation may be implied in a prime morethanfefold
relation, or in a composite relation, regular or irregular, each factor whereof is more
than /(.fold ; and BO also a composite relation, regular or irregular, each factor whereof
is at most /cfold, may be implied iu a composite relation, regular or irregular, each
factor whereof is more than ft fold, In a somewhat different sense, each factor of a
composite relation implies the composite relation,
19. A composite relation is satisfied if any particular one of the component relations
is satisfied ; but in order to exclude this case we may speak of a composite relation as
being satisfied distvibutiwly ; viz. this will be the case if, in order to the satisfaction of
the composite relation, it is necessary to consider all the factors thereof, or, what is the
same thing, when the reduced relation obtained by the omission of any one factor what
ever is not always satisfied. And when the composite relation is satisfied distributi
the several factors thereof are satisfied alternatively ; viz. there is no o
throughout unsatisfied.
460
A MEMOIR ON ABSTRACT GEOMETRY.
41.;}
22. There is no meaning in aggregating a relation with itself; miuh
only occurs accidentally when two relations aggregated together become 0110 ami tho
same relation; and the aggregate of n relation with itself w nothing O!HO than Un
original relation.
23. A onefold relation is not an aggregate, but itj its own .solo ooiiHtitnnnt ; IL
more than onefold relation may always be considered as an aggregate) of two or mom
constituent relations, The constituent relations determine, they in font coiwtitutu, tin
aggregate relation ; but the aggregate relation does not in any wwo determine lilu/ 11011
stituent relations. Any relation implied in a given relation may bo (;o]ini<loii!<l iw a
constituent of such given relation.
24. The aggregate of a Mold and a /fold relation is in general ami at mo*!, a
(fc + J)fold relation; when it is a (& + J)foIcl relation, the mnntiDuonl; rolatioi.n an
independent, but otherwise, viz. if the aggregate relation ia, or has for factor, a I,*H
than (A+J)fold equation, the constituent relations are dependent or mtaominotnd.
25. Passing from relations to loci, we may say that the compomtion of rnlulinnn
corresponds to the congregation of loci, and the aggregation of relation, to (;hn ,',,,.
section of loci.
26. For first the locus (if any) corresponding to a given compoHitn rnlati.m is
the congregate of the loci corresponding to the sovoral prime faotom of t li,, K iv,,,
relation, the locus co 1T esponding to a single faetor being takou ,mo, an.l thn [
feetor bciug
28. It may be remarked that a Wold locus and a Mold loons
here the aggregate re.ation h ,no re than w fo,d) l.avo n ot
givon
not in involution is said to be
"' mluiwn ' ll
from the system any relatou
until we arrive at an
implied in the gil .
ens, and it 1 th, to * JJ
W m V oralt
'"" 8 ldatlora ' and 80 "aoeaHlTOly
* onofcW rolato
413] A MEMOIR ON ABSTRACT GEOMETRY, 461
be added on to and made part of the system. It may happen that, in the system
thus obtained, some one relation of the original system is implied in the remaining
relations of the new system; but if this is so the implied relation is to be rejected;
the new system will in this case contain only as many relations as the original system,
arid in any case the new system will be asyzygetic. Treating in the same manner
every other onefold relation implied in the given Mold relation, we ultimately arrive
at an asyzygetic system of onefold relations, such that every onefold relation implied
in the given Mold relation is implied in the asyzygetic system. The number of onefold
relations will be at least equal to k (for if this were not so we should have the given
fcfold relation as an aggregate of less than k onefold relations) ; but it may be greater
than fc, and it does not appear that there is any [assignable] superior limit to the
number of onefold relations of the asyzygetic system.
31. The system of onefold relations is a precise equivalent of the given Mold
relation. Every set of values of the coordinates which satisfies the given Mold relation
satisfies the system of onefold relations; and reciprocally every set of values which
satisfies the system of onefold relations satisfies the given Mold relation. But if we
omit any one or more of the onefold relations, then the reduced system so obtained is
not a precise equivalent of the given Mold relation; viz. there exist sets of values
satisfying the reduced system, but not satisfying the given /cfold relation.
32. In fact consider a Mold relation the aggregate of less than all of the onefold
relations of the asyzygetic system, and in connexion therewith an omitted onefold
relation j this omitted relation is not implied in the aggregate, and it constitutes with
the aggregate not a (k + l)fold, but only a Mold relation. This happens as follows,
viz. the omitted relation is a factor of a composite onefold relation distributively , implied
in the aggregate; hence the aggregate is composite, and it implies distributively a
composite onefold relation composed of the omitted relation and of an associated onefold
relation ; that is, the aggregate will be satisfied by values which satisfy the ^omitted
relation, and also by values which (not satisfying the omitted relation) satisfy the
associated relation just referred to.
33. Selecting at pleasure any k of the onefold relations of the asyzygetic system,
being such that the aggregate of the k relations is a Mold relation, we have a com
posite ifefold relation wherein each of the remaining onefold relations is alternatively
implied; viz. each remaining onefold relation is a factor of a composite onefold relation
implied distributively in the composite Mold relation. Hence considering the fc+1
onefold relations, viz. any fc+ 1 relations of the asyzygetic system, each one of these is
implied alternatively in the aggregate of the_ remaining k relations ; and we may say
that the jfc + l onefold relations are in convolution.
34. More generally any A + l or more, or all the relations of the asyzygetic system
are in oonwteto'on, that is, any relation of the system is alternatively implied in _the
aggregate of the remaining relations, or indeed in the aggregate of any k relations
(not being themselves in convolution) of the remaining relations of the asyzygetic
** G2 A MEMOIR ON ABSTRACT GEOMETRY. U I 3
system. It may bo added that, besides the relations of the system, there is not any
onefold relation alternatively implied iii the asyzygetic system.
35. The foregoing theory has been stated without any limitation as to tho valuo
ot t.aud it has I think a meaning even when k is >m; bnfc the ordinary case is km
Considering the theory as applying to this case, I remark that the last proposition, viz
nut no reduced system is a precise equivalent of the given /,fold relation, is aeuerallv
true only on the assumption of the existence or quasiexistcnce of sets of values
fold relation  For iet * be m *
more tha11 "*>* relation
th r "> reaton
the coordinates ; the number of relations in the system may be > w +l and if
s  "">=  '=
:rz
(In illustration of the foreeoine Nm 9Q t q T
"
But when the fac VB C g ftnch ni1 r ec l uafcions
i f /I of them constat '
. 8i
= is implied ; T'" I " e remam 'g equations Q = , .B = 0,...,
= ffi0 equations, and the relations (Q = 0, 7eLo ) ad
V U, ft = U, , . . I H,rP Pnillirolanl 4^ 1 i ^ " J U . i i I Uilill
0 equivalent to each other. But in the ease of a convolution
then the relation the equations Q = _R = im I
P = 0; or, what is the same thing, the relation ' '(0 = V T, >_!" \ ^ ^ A ~ r els
of the two relations (.4 = 0=0 ^ = i "U,..,) is a relation composod
any" 'of tf eXPreSS d ^ m re tha *"^Sl ( fp" i 00 p'S'^ {" ^
any t ot these equations which are not in c 1 t * "'). selecting
the remaining equations, we have a convolut^ fT' ^ Unifcin8> tllGrefco any one of
relation is precisely expressed by means of ^Vf e e . quafcions ' and lvh0 n a Afo]d
Q=0,...) ( then ev ery equation O = imnliprl T r more ec l uatio ^ (P=0,
same thing, the equation of any onefold 1 ^^ relati ni 01 *' what is th(S
the Afold relation is in involution with the ^T^ D gh tho locus & b 7
have identically O = ^ + ^ + ^+ f ^ P0, Q 0f ... ( that is, we
coordinates.} ">*. A 0,... being integral f unctions of fche
413] A MEMOIR ON ABSTRACT GEOMETRY. 463
Article Nos. 37 to 42, Omul Relation; Order.
37. A Affold relation may be linear or omal. If k = m, the corresponding locus
is a point; if k<in the locus is a fcfold, or (m ~ ^dimension al omaloid; the expression
omaloid used absolutely denotes the onefold or (m l)dimensional omaloid; the point
may be considered as a ?ftfold omaloid.
88. A mfold relation which is not linear or omal is of necessity composite, com
posed of a certain number M of mfold linear or omal relations; viz. the mfold locus
corresponding to the mfold relation is a pointsystem of M points, each of which may
be considered as given by a separate mfold linear or omal relation ; each which relation
is a factor of the original ?nfokl relation. The given ?n,fold relation, and the point
system corresponding thereto, are respectively said to be of the order M,
39. The order of a pointsystem of M points is thus =M, but it is of course
to bo borne in mind that the points may be single or multiple points; and that if
the system consists of a point taken a times, another point taken /9 times, c., then
the number of points and therefore the order M of the system is considered to be
<10. If to a given /cfold relation (k < m) we unite an absolutely arbitrary
(m~/e)fold linear relation, so as to obtain for the aggregate <a mfold relation, then
the order M of this mfold relation (or, what is the same thing, the number M of
points in the corresponding pointsystem) is said to be the order of the given fcfold
relation, The notion of order does not apply to a more than mfold relation,
41. Tho foregoing definition of order may be more compendiously expressed as
follows : viz,
Given between the m + 1 coordinates a relation which is at most mfold; then if
it is not mfold, join to it an arbitrary linear relation so as to render it mfold;
wo have a mfold relation giving a pointsystem ; and the order of the given relation
is equal to the number of points of the pointsystem.
42. Tho relation aggregated of two or more given relations, when the notion of
order applies to the aggregate relation, that is, when it is not more than mfold, is
of an order equal to the product of the orders of the constituent relations; or, say,
the orders of the given relations being >, /*',..., the order of the aggregate relation
464 A MEMOIR ON ABSTRACT GEOMETRY. [413
really as well as formally Mold; if they do satisfy certain relations in virtue whereof
the formally fcfold relation is really less than Wold, say, it is (k l)fo\d, then the relation
is in fact to be considered ab initio as a (k )fold relation : there is no question of
a relation being in general fcfold and becoming less than /ufold, or suffering any other
modification in its form ; and the notion of a more than mfold relation is in the
preceding theory meaningless.
44. But a relation between the coordinates (x, y, ...) may involve parameters, and
so long as these remain arbitrary it may be really as well as formally /ufold ; but when
the parameters satisfy certain conditions, it may become (kl)tol<\, or may suffer some
other modification in its form. And we have to consider the theory of a relation
between the coordinates (a, y ,...}, involving besides parameters which may satisfy certain
conditions, or, say simply, a relation involving variable parameters. If the number of
the parameters be m', then these parameters may be regarded as the ratios of m'
quantities to a remaining m' + 1 th quantity, and the relation may be considered as
involving homogeneously the m' + l parameters (to', ?/,...). And these may, if wo please,
be regarded as coordinates of a point in their own m'dhnensional space, or we have
to consider relations between the m+1 coordinates (a;, ?/,...) and the m' + l (parameters
or) coordinates i/', ..,). It is to be added that a relation may involve distinct sots
of parameters, say, we ^ have besides the original set of parameters, a set of m"\I
parameters /,...) involved homogeneously. But this is a. generalization tho
necessity for which has hardly arisen,
Article Nos. 45 to 55. Qualities, Notation, &c.
45. A homogeneous function of the coordinates (0, </,..,) i a represented by a
notation such as
(*$>, #,...)<>
(where (*) indicates the coefficients and ( . ) the degree), and it is said to be a
quantic; and in reference to the quantic the quantities or coordinates <, ...) are
also termed faw.nti. More generally a quantio involving two or more sets of coordi
nates, or iacients, is represented by the similar notation
t ' ite ' &c " accMdi ^ the number of
18 one two, three &c,; and with respect to any set of coordinates, it is binary
o es  ^
bee, foui, or + ; and it is linear, quadric, cubic, quartic ..... according as the
degree m regard to the coordinates in question is 1, 2, 3, 4, ... .
47 A quantic involving two or more sets of coordinates, and linear in regard to
1 if i taVLt it V antiparfcit6;   l^r t when there a. oTy t
sbe mear; may
413] A MEMOIR ON ABSTRACT GEOMETRY. 465
48. Instead of the general notation
(*<, </,... )<>(< 2/',...)' !) ..
"we may write
(a, ...)& */,... /(.< y', >,
where the coefficients are now indicated by (a,...), and the degrees are p, /i', ...
49. In the cases where the particular values of the coefficients have to be attended
to, we. write down the entire series of coefficients, or at least refer thereto by the
notation (a,...); and it is to be understood that the coefficients expressed or referred
to are each to be multiplied by the appropriate numerical coefficient, viz. for the term
ffl"P ... ftj^'... this numerical coefficient is
50. It is sometimes convenient not to introduce these numerical multipliers, and
we then use the notation
(a,. ..$0, y t ...Y(af, jA ...)".,
(a j ...i 1 y,...> 2/,.. >'...
In particular (a, 6, c]j>, y}\ (a, b, c, $%, yj &c. denote respectively
off" + 2% + cf,
&c. ;
but (a, &, ojffl, y) 3 , (a, &, c, djffi, i/) 3 , &c. denote
aa; 2 + 6ffy f cf,
a B + btfy + cwf + dy 3 ,
fee.,
and so (a, &. o, /, ffl ^ y. *> and (, &, c, /, ^ A]fe y. S ) fl denote respectively
oo 9 +
aa; 3 +
61. To show which are the coefficients that belong to the several terms respectively,
it is obviously proper that the quantio should be once written out at full length ;
thus, in speaking of a ternary cubic function, we say let U=(a,...$a> y, *)
(a, &, o,'f, g, A, , 3> k > 1 1> y> ^
= cw? + bf
+ 3 (ffz
and the like in other cases.
0. VI.
460
52, A nuufnlil n'talinii
oquabinn of (,ho form
A MKMUIH (IN' AIM'llAiT i.J"MMUY,
i in
'> m. nii'f .I it)l
nil. Tins oxprtiHHiim "mi
an equation ,,f ihn f,, nil j n ,,,,,(!,.. vi. it,., ,.,,,.,*
aom; th
ly Hft y
proHsiblu by an
rn!
itor, iviul to UK,, a
that Lho ti. K ln
I..
h,
//
t mrly ti. wril,
Wo may aim npnak uf M, ( , tl
ff=0, orfcho rm
H/i. A /,ful.l ivInMnn b,.i,w... il,,, rMhln,.,*
a HyHloin nf k ur in..r,, ,, M .;,], nl..!).,,,,, ,,,},
0, and UKI Mi.,l ,nla. '
>'Jtfh tfallv
*!nf iv .,,,
', ;;, , r  11
'b" * M*i<.ii
'
, ';"
, tho m>fta ,
,,
l '"' v 
h
' Uli ON AUSTHACT GKOMKTUY. 4G7
will mproHont u ] ](l)illt in tho ^npaeo, nay, tho oo^mon, point,
fl. A mfol.1 relation, or tho loeus, w pnintsy.stom tluuoby reprinted,
Z t 'TiV" 1 '. r ; ! ht \ lliliutM , r thu iwn! may b,
gonouilly ii Mold rolutiou (fc ;f> ,), or tho locus thoroby ropro.soutod may l.uvo n
or / pun,,; f r lob tliu ,oladon If U,s than JfohM .^oTfoW by
*> it n mmr iHh l '
l"ii.W *> it n mmr (iH^hl lvUt iou ^infi,,! by the, coordinate
h ; "tho ; ; ; Hin v7 iMri ^ ^ if t!i imiiit ^^ bi ^ ^w
I , r th V'fHyH ta ^^ Puntod, tho point i
11 ' Il< ] 10 ol * ilml "" Ii)ltl ! ' olrtbl<mi or of
00. A givnf, rrlati.m (^ ^ w ) bo.Am.n thn W1 +l owmliiurtoH. oi< the IOOIIH
Ui nby , l .., l .a w IUIH nut m ^.orul a ,ulnl point. But if thu Ration involvo thu
i lmmt. !W ( y ,,..}, thui,. If a uortuin nnofold rotniinn lo Ha^fiud btwooii tho
1 " "' ft ' ll
, I
t, rchttunt of 1,1ns ^i V( .i /:lbM rolation.
t. Tu l)h WIHO m .jnnstion, /,:(>, thu diwiimhuint relation i tho ronnltunt
o Uon nl a (*+ )lol,l y.hitmn whioli i.s tho )W O K ^O of bho givou ^fokl relation
with a iHufian. rclntiun culh.l tlu, .Aa^w/t .rf(/ (H , t or w ,on tho di^iuotion IH rommwd
fine t/acobuni rcltttwn in iogiml tu tho (a; t ?/,...).
(12. On,,Hi(l,r n /.(old nHatu.n (k > m, > m') hot^oon tho wn 1 eoonliiuittw (; v )
ami ho f /i +1 nwmljimh* ( ; ,; ,/,...). i, ; ^ b(!(m H , "
t i , s " ' nji.ii UIMIU uw u, ^ivuii ti\ii ui vcunoH oi
_ (<, y,...) or, Hiiy. (a a given point in the >aeo, thoro cornspun.]* a Jbfold ' IOOUH
in tho_ wwinncu. and that to a tfvon wot of valuon of tho (a,, y,...). or to a R ivcm
point m tlm ^Hpmto, thoro wirixHpraidH a M.ld lociu in tho m'Hpace, Tho /fold
Ueus m tlm w'Hpaco may haw. a nculal point; thiH will bo the OHM if thero in
witiflfind boh thu (/,', ?1 ...) a curtain onefold relation, the discriminant rotation of
Urn givon /.fold rohUmn in regard to tho (a', y' t ,}. This onofold relation repre^^
in tho wtHpaee a onefold looiw. tho envelope of the /.fold looi in tho m 
Hponding to tho Heveral points of tho w'Hpaoo. Tho property of the euv
to uiufh point thereof thoro coiTOHpondn in the wi'spnco a A: fold locus ho
point.
Article Noa. 0309. Consecutive Points; Tangent Omuls,
08. As tho notiona of proximity and remoteness have been thus far altogether
ignored, it seems necessary to make tho following
_ Postulate. Wo may conceive a point consecutive (or indefinitely near) to a given
592
4(18 A MEMOIR ON ABSTRACT GEOMETRY, [413
04. If the coordinates of the given point are (at, y, ...), those of bho consecutive
point may be assumed to be (x+Sae, # + &/,...)> where So;, By,... aro indefinitely small
in regard to (,, y, ...).
65. It may be remarked that, taking the coordinates to bo (,* + X, ?/4 Y, ..A thoro
is no obligation to have (X, Y,..,) indefinitely small; in fact whatever tho UKignitudoH
of these ([imntities are, if only X ; Y:...=x : #:..., then tho point (,v + X t y \ Y, ...)
will be the very same with the original point, and it is therefore clear that it con
secutive point may be represented in the same manner with iimgniUuloH, howovcir
large, of X, Y,... But we may assume them indefinitely small, bhnfc in, tho mtiiiH
.v+$,v : y +%,..., where Sx, %,.., are indefinitely small in regard to (as, ?/,.,,), will
represent any set of ratios indefinitely near to the ratios (m : y t .,.),
The foregoing quantities (&p, %,...) are termed tho increments.
_ 06. Consider a fcfold relation between the m + I coordinates (to, ,...) k > m tho
increments (&, fy, ...) are connected by a linear A fold relation. '
The linear Afold relation is satisfied if we assume tho increments proportional to tho
coordmatesthis is, in fact, assummg that the point remains unaltered. Wo may writ,
y " S1UC " in8UC an e the tios ''0 a to.
r ,. os ''0 aomor to. ut
F0portlmml to tho
"
as
by writing
,... t j ; c ^'^ r io f r' s r ses through th
Wold locus at the point fe ,, n f th<3 to .'/ (! *< of tho original
!<"' (, y, ...), which pomt 1S said to bo the po{ , a of ^^
^^S^Thir 06 ^/')"y( X ,y,...),and combine
elation (containing as paramete ' ^ ^ f"* J ate s <, 7,.) a M .fbld
the original Afold relaHm. ,^i , ^2/.0); these pammeberB satisfy
. but the point in qu^n 7sTn Z T Z? V ^  i d not
also follows that the tangentomal locus con d P m a 2 *" fold looiw
] cs. ng e ori ^ inal Afold relation, has for its envelope the
413] A MEMOIR ON ABSTRACT GEOMETRY. 469
69. We thus arrive at the notion of the double generation of a Afold locus, viz.
such locus is the locus of the points, or, say, of the ineuntpoints thereof; and it is
also the envelope of the tangentomals thereof. "We have thus a theory of duality;
I do not at present attempt to develope the theory, but it is necessary to refer to
it, in order to remark that this theory is essential to the systematic development of
a mdimensional geometry; the original classification of loci as onefold, twofold,...
(ml)fold is incomplete, and must be supplemented with the loci reciprocally connected
with these loci respectively, And moreover the theory of the singularities of a locus
can only be systematically established by means of the same theory of duality ; the
singularities in regard to the mountpoint must be treated of in connexion with the
singularities in regard to the tangentomal. These theories (that is, the classification
of loci, and the establishment and discussion of the singularities of each kind of locus),
vast as their extent is, should in the logical order precede that for which other reasons
it may be expedient next to consider, the theory of Transformation, as depending on
relations involving simultaneously the w + 1 coordinates (to, y, ...) and the m' + I
coordinates (of y 1 , ...).
41 !,
ON I'OLYXOMAI, fUMlVKS. uTIIKiaVls
s'/'W i )*,. ...
"S "< .*./ .V,..,rtV ../ A
HO. (,,,,,l ,i,, I.,,,,,,!,,,
, " 'K' . ...... . ,.n. v ,, ., ... ..
* r, , ,.,,,! i,, u, ......... ..... ,., .....  . .'/' "' ""'
curve W/W'.. (jr.. ,,ii ' ," ' ' ' ' "' " II.V/..IIH
I;,,
hav,, U !lm , ,,!. ,, '
*a If Un, n,,,,,!,,, ,  ....... ' >
. mu! o
u ,
" ' . mv
, ,, .
o, w.irthy ,,r (.<)i.i l l l , m iin it j, ; , , ....... '" ' " "'" '"" ' Hiir
on,,, t in,c, Bmmiiullt ,, w ,, , , '" " * ........ ,"""" ...... ' ">" .. i...l , ,, , ,,
& are o,,,, Ntall( , Tl)0 Ml^ 1 ' ^ ' ?' M "< "'" ' ....... ',*,.., ,.
<mrvl! i *u bm ..... 1M Uwr.n.r. U, H , ..... ,, ,,r I, '"" "'. """' '"
in i'og,ml to thu ^.omul mtw WlMt/.mj e ' '"" ""W" 1 " 1 '"
iu p"! "wwii a ...... ,, tH ;, r i,i,^;i, ll ;,;" 11 u * lii  hiiiiD *'''' .
onion , ftllll r _, ,,, loct . V( .  ' r *" '' l., ,,>, M,,,, .J..,, ,,r
* " ' ''' 
of
( .
to of th, ourvu bo , .
. p.
curto
41 4 ]
ON rOLYHOMAfc OUBVEH.
two trramml ouivoa, each oxpvo.sniblu by ntoaiiH of any thrno of the four fmiotmiiH
y, V, W, .'/'; for example, in the form */l'U+ Vw'F> Vyrr= 0. If, in ,hi theorem,
wo tolco _p = 0, then tho original curve M the tmomul V^ VwiKH vVl'K 0, .'/' m
any function (aZ/.bV>eF), whore, couriering J, m, us givon, a, li, o ,im
quantities Hiibjoot only to tho eomlition { + + .' 0, un.l wo havo ll.o theorem ,,f
11 I)
" tho variable HOI mil of a tribunal oui'vo," vix. tho (Hpuifciou of thn tmumal
yW+* / V'+<SnW**Q l may bo oxproHWid by nioaim ( ,f miy two of ,lu. throo functioiiN
(/, K, JK, and of a Aniotion 2' dotonninoil an ubovo, for txnmplo in I In tonn
Vm+vWl>>vV.7'0! whonco alHo it may bo OXIHOHHIM! in torniH of tl.m, now
functioiiH .7', clotonnincd an abovo. Thin thnownn, wlu.ih oeimpion a ]ioiniuonl, .u,iii im
in the wliolo tlioory, wna mi^.mtod to mo by Mr (Jiiwy'a tlicnruin, pnMuiitly ivfml
to, for tho csoimlii'iiotion of a bicirouliir quaHio as tho onvolopo of 11 vtirinblo injlc.
In tho i^onml curve Vi ( + //li) + & c . 0, if M bo a wmio, 0> a Itn.
tho annuls H + 7/1)^0, foe. arc TOIIKIH punsing thn.ugh tho HIUIIO l,wu pnintn M  0,
<P0, and fchoiu in nu real I,,HH of  om!m lity in baking lihesu to bo Llut (liroular pointn
ut iiihmty that in, iu taking tho OUIIJUH tu bo cimloH. .]),.h. K thin, and HHI]I K a H )Miiiil
notation *?["() for tho oration of a oirolo having i(, H cnntro ut a K ivcn ju.inl, J.
mid winilarly Sf0 for (ibo equation of an nvimowimt oirdu, nr nay .f Hi,, p,,i h f, ^1,
wo have Mm iMsomul curvo VAT^&e.O, and iho nioro Hpuciul form VU &.<).'
AH rogardn the lastmentioned unrvo, V^l + & (! . 0, the iioint yl to wlueb Mu
Pf0 boloi^H, in a oc.iw of thu curve, vi/. iu tho OHNO ^,'1, in  H an ..vdinary
and in tho UAHO ,$, it in a special land of f, muH , whioh, If tho tonn w.uv roi
miffhb be called a fotsofooim; (} M Memoir <!0)itaiiiH an i^planntion of tin
theory of tho foui of piano (jurvoH. Fur i.tt, tho equul^.n V/vf+ Vwi ( V H .<J
roully e(iiivalent to tlio apparently inoro general form ^I"H At^' 1 I "/(.*;' !). fit liwifc
thiH hiRt in in general a bioircmlar quartio, and, in regard to it, tho bi>lnni.iuniiiinii( H i
thoorom of thu variable mmml btiw.inoK fttr OVsey'H theorem, that " thn bi.^iroular quartie
(nnd, OH a particular 0,1*0 ehcniof. tho oimdar cubie) i H tho uivolopo nf a variable
endo ^ having j tH ounlim ()ll a ^ iv(!11 0()nic mul (m(iti , IK (lt ^^ tU} ^ H .
Clrclo< " Z lllH tjie(ll ' (!111 iH Jl wHUciont luiHi.H lor the cjoniiiloto thoory of tho
'
472 ON 1'OIVYXOiMAI, OUIIVMH. i ,. ,
Investigations; Part III., On tho Thtsory of J'Vi ; MIX! IW IV., (),, (J,,, TnVnnm! Mlll 
Tetrazoraal Curves whore tho /omuls urn mivlrs. Tlinv JM, ,,,uvvrr VMM,,. IM.V. ,,
intermixture of tho theories troutcd of, and ,h,. i,m.., K n,.n.l. will ',,,,. ..... n . :!.
detail from the heading* of thu Novoral mtii'lm Tit.. purmmi.!.* an ,n,,, ..... IVI  nm
tinuously through the Momoir. Thorn aro (! Anns.s. Hluring ,,, ,,,,;, .,,:,.,. .
seemed to mo moro convenient to (,ivat oi' UIIIH wpimih'ly,
It is right that I Hhould explain U, (! vrry K n,,i ,. X ,,,. ( w
position of the present Memoir, I an, in,!,!,!:,,! , A  r , Wv 
;'0n the Elation, and l^ wrtiw (,) , tllc . s^,,,,, ,, ri
in a plane; (2) of tho SyHtom of Sph.n, ron.hin, li.nr nphnH n
System of Circle touching tln,o ,i,, ( , on , H,,^.; { ,,, ' ,,,"
^nbed in a conic and touching tlnvo innniLnl ,(,, in
a. 2
,
s?=jssi; ::,K:";;;!,,;:: TV ""'v;;",;;:::
in hia lotto of Marcl, a IHOi ,, ' """""""'W ' Inn, U,,. ,,,,,l,l,,, ......  v ,,,
with ca^ Eqnati ,;; ^:: , ; /:: ll ," )if : liiv ..... '" ...... ........... 
a , , / , ,,
also a lw Annex No. IV o f ^ Z M,t ',! '''' ' :MI< :I " S .......
In connexion with this ilunr,,,,, i
'
s unr,,,,, i
,, , t ,, ';: ....... i"":; 1 " 1 ." ? ir ...... > ...... ' 
**> ., ,, (  ohl! U1 :,: V ':;,S"; ;';;: ';'*, " .
^, denote ^ mutlml , iHtoK ! ()S ,. ,; ;' 'V o ' ""' '"' '''
Mr Casey, in a lottor U, mo d.ilod MOII A ,", u '^ ""'"" '" " (1 ""'i'.
of viewing th,, , iuo8[irai ^ ; 1(1 ' A I'"L IWT. ,...,,, ........ hi ................
fc'.'./.ff.AjfeAtfo.wi ' 77 1 ,/""'; '"""" ........ ""  ......... ' "i
the equation (, 6, e , ^, *V' i/ n ? "' '" " l ' i " i '"" l " r ...... f ""' "' "'
" '' ..... " ..... " ' .....
o, , . ,.,
conic; denoting thin conic by /,' ftll , '''"" '.' , a ' ^ 1 "" "} I ..... i. i, ,!,,
'.I Foved that, if tt variabl c , , " """ '""" "' "' "''' ........ v 'V
orthogonally, it. onvcl wm ,,; ( '"' I ''i,.r ...... ,, ir ..... ,
down above;" aild J Mm , '" >l" l.' ..... li..,, i. ,,,,, wrf,.,,.,,
double foci of the quartic and II, ' ' " ..... """" """ ll "' '"'! "I' F HIV lh,
^ *, the ttcd'irv 1 ":: i: 1 "" 11 / r " ^ M >* :
^ Memou. on Bieiroular Q lmrt T, tl " '"' '''"'"'"' '""' '"' * I
Memon; as road before the ttoya ,,' A ""'" J ' , ' i " i " 1 "" 1 ' A " Alnlm,,. ,,r ,,,,
sit, r f""'^' "" H S h ;;;";;, """ " <iim ""^ ^ ..... '> ..... .1
30th Apnl, and 8omo othor thooron T i ....... '''""' " ll '" li "'""' i" Hi,. ,,. ,,r
d u t r wl i *  r^n ;'c;r w "^ u r ........ ^<
Memo;r U ,n m h , ^^ ^J^^y r ,,, ;
POLYZOMAL CURVES.
473
PART I. (Nos. 1 to 65).0 POLYZOMAL CURVES IN GENERAL.
Article Nos. 1 to 4. De/Vm fl rf Preliminary Remarks.
1 As already mentioned, ZT, 7, &c. denote rational and integral functions (*$, f/
all of the same degree r in the coordinates (, y, ,), and the equation
then belongs to a polyzomal curve, viz., if the number of the zomes *ffi t VF &c is
*, then we have a ,zomal curve. The radicals, or any of them, may contain 'rational
factors, or be of the form f*/Q. but in speaking of the curve as a ,zomd it is
assumed that any two terms, such as PVQ + P'Vg involving the same radical V0
are united into a single term, so that the number of distinct radicals is always *
m particular (r being even), it i s assumed that there is only one rational term P
But the ordinary case, and that which is almost exclusively attended to, is that in
winch the radicals </U, W t &c. are distinct irreducible radicals without rational factors.
2. The curves V=V=Q, &c. are_ said to he the zomal curves, or simply the
zomals of the poly.omnl curve ^+Vr+fra0j more strictly, the term zomal would
be_ applied to the functions U, V, fa, It is to be noticed, that although the form
Vtf+VK+&c.0 is equally general with the form Vjff + VmT+ &c. = (in fact, in
the former case the functions U t V, &c. are considered as implicitly containing the
constant factors l t m, &c., which are expressed in the latter case), yet it is frequently
convenient to express these factors, and thus write the equation in the form V7?+ v
'
c
1'or instance, in speaking of any given curves [7=0, 7=0, &c., we are apt, disregarding
the cons ant factors which they may involve, to consider U, Y, &c, as given functions;
but m this case the general equation of the polyzomal with the zomals. 170, 7
&c., is of course V#? + VmF+ &c. = 0.
3. Anticipating in regard to the cases = 1,^=2, the remark which will' be
presently made in regard to the pzomal, that V!? + VF+&c. = is the curve represented
by the rationalised form of this equation, the monozomal curve VF= is merely the
curve Z7=0, viz.,Jhis is any curve whatever U= of the order r; and similarly, tho
bizomal curve Vi7+VF = is merely the curve Z770, viz. this : ~
whatever Ii = 0, of the order r; the zomal curves U=b, V A  1 
not curves standing in any special relation to the curve in
may bo any curve whatever of the order r, and then 7=0 is a curve of the same
order r, m involution with the two ourves_fl = Q, 17=0; we may, in fact, write the
equation fl = under the bizomal form VZ7_+ VflTD 1 = 0. In the case r even, we
may, however, notice the bizomal curve P + Vtf=0 (P a rational function of the degree
r); the rational equation is here a=^P^=0, that is U=fl + P* t viz P is any
curve whatever of the order r, and 7 = is a curve of the order r, touching the
given curve il = at each of its r a intersections with the curve P = I further
0. VI.
474 ON POLYZOMAL COBVES.
remark that the order of the i/zomal curve ^F+&o. = is =2r; thin in right in
the case of the bizomal curve Vt/"+\/K=o, the order being =r, but it fails fur tho
monozoraal curve >JU=Q, the order being in this case r, instead of ), us given by
the formula. The two unimportant and somewhat exceptional cases v = I, i> =. 2, aro
thus disposed of, and in all that follows (except in so far as this is in fnet applicable
to the cases just referred to), v may be taken to be =3 at least.
4. It is to be throughout understood that by the curve </U+ VK+foc. =  H
meant the curve represented by the rationalised equation
Norm(VZ7+ VF+ &c.) = 0,
viz the Norm is obtained by attributing to all but one of tho KOHIOH V / J V &,.
each of the two signs +, , and multiplying together the several rolling values ,','['
the polyzome; m the case of a ,zomal curve, tho number of factors is thus =2".,.
(whence, as each factor is of the degree r, the order of the ourvo is 2 ..... Li,.
= 2 r. as mentioned above). I expressly mention that, as regards the poly,mal cnnV
we are not m any wise concerned with the signs of the radicals, which sifins , m , aad
nnuun essentially indeterminate; the equation VF+ VFh &e. = 0, is a mon) ' Hyml,l Ir
the rationalised equation, Norm (V(7 + VF+ C .) = 0.
Article Nos. 5 to 12. The Branches of a Polyzonal Ouna.
5. But we may in a difeent point of view attend to tho signs of ,;!, radicals:
f for all values of the coordinates we take the symbol ^ m . c(mHillm , ^ jy
&c. as signifying detennmately, say the P o^e valuos of W, ^V. fa,, then ,H ,h ol''
the several equations J U_ JV +ta ^ , fixing ftt ' . '
'
ofthe s x =
ft
, t
Cations of
Aginary, va, u , If for a? 
ot , ^ we
ON POLYZOMAL CURVES. 475
all one of the two Apposite values of V?+#, ca llmg it the positive value, and
representing it by ,/+#, then, for any particular values__of the coordinates, U being
= +#, thejalu^of Vff may be taken to be W^; and the like as regards
JV, &c. ,/ff, ^F &c. have thus each of them a determinate signification for any
values whatever, real _or imaginary, of the coordinates. The coordinates of a given
point 01^ the curve Vff+VK+&c. = 0. will in general satisfy only one of the equations
N/^s/F&o =0; that is, the point will belong to one (but in general only one) of
the 2" branches of the curve; the entire series of points the coordinates of which
satisfy any one of the 2' equations, will constitute the branch corresponding to that
equation. .
8. The signification to be attached to the expression J^Ji should agree with
that previously attached to the like symbol in the case of a positive or negative
real quantity; and it should, as far as possible, be subject to the condition of
continuity, ,jjH/9,' passes continuously to ' + /3'i, so V*T^ should pass con
tinuously to W+/3'*j but (as is known) it is not possible to satisfy universally this
condition of continuity; viz., if for facility of explanation we consider (a ft) as the
coordinates of a point in a plane, and imagine this point to describe a closed curve
surrounding the origin or point (0, 0), then it is not possible so to define
that this quantity, varying continuously as the point moves along the curve, shall,
when the point 1ms made a complete circuit, resume its original value. The' signi
fication to be attached to V + # is thus in some measure arbitrary, and it would
appear that the division of the curve into branches is affected by a corresponding
arbitrariness, but this arbitrariness relates only to the imaginary branches of tlie curve*
tbe notion of a real branch is perfectly definite.
9. It would seem that a branch may bo impossible for any series whatever of
points real or imaginary. Thus, in the bizomal curve Vtf+VTO, the branch
Vtf+Vp:=0 is impossible. In fact, forjmy _point whatever, real or imaginary, of the
curve, wejiave^= V, and therefore Vff V7; the point thus belongs to the otf
branch VETVr0, not to the branch VI7+V7=0; the only points belonging
the^ lastmentioned branch are the isolated points for wlr ]  ' "
47 G ON POLYZOMAL CURVES. [414
would have opposite signs), but there is no apparent reason, or at least no obviously
apparent reason, why this should be so for imaginary values of the coordinates, and
if the sign be in fact , then the point will belong to the branch VI? + V7> VF= 0.
11. But the branch in question is clearly impossible for any series of real points;
HO that, leaving it an open question whether the epithet "impossible" is to be under
stood to mean impossible for any series of real points (that is, as a mere synonym of
imaginary), or whether it is to mean impossible for any series of points, real or
imaginary, ^whatever, I say that in a yzomal curve some of the branches are or may
impossible, and that there is at least one impossible branch, viz., the branch
C. =0.
12. For the purpose of referring to any branch of a polyzomal curve it will bo
convenient to consider VF as signifying detenninately + VF, or else VFj and the
like as regards^ F^&c., but without any identity or relation between the signs pre
fixed to the VtT, V7, c., respectively; the equation VF+VF+ & .  0, so understood,
wiU denote determmately some one (that is, any one at pleasure) of the equations
^tfVF&c. = 0, and it will thus be the equation of some one (that is, any one at
pleasure) of the branches of the polyzomal curve  all risk of ambiguity which might
otherwise exist will be removed if we speak either of the citrw VF+V7, &c. = or
else of the branch Vtf + VFf&c. = 0. Observe that by the foregoing convention, whou
only one branch is considered, we avoid the necessity of any employment of the sign + ,
or ot the sign ; but when two or more branches are considered in connection ' with
each other, it u necessary to employ the ^ign  with one or more of tho radicals
u \ &G ' ; , US ^ the ,^ omal T ve ^+VF+V=0, we may have to consider
vt I ^ t, "t Tf=0 ' ^+^^0; vis., either of toe equations
the L,r ."I I 6 ? ^ ne bl ' anch at P leaSU1 ' e of thQ <. but when
h oZ T 88 , \ , ne eqUati U IS flX6d ' then th bl ' a ch Putod by
tne other equation is also fixed, J
Article No, 13 to 17. 2fc Points cownon to Two inches of a Polyzonal Curve.
13. I consider the points which are situate simultaneously on two branches of
thermal curve V^VF + & , = 0. The equations of the two branches may be taken
viz., fixing the significations of Vff VF VF frn i. m i ,1 ,
e i , , ' Y> Y w > fflc  in such wise that in the
414] ON POLYZOMAL CURVES. 477
V W t &c, to be those radicals which have the sign. , The foregoing equations break
up into the more simple equations
VF + c. = 0, VF+&c. = 0,
which are the equations of certain branches of the curves VtTf&o. =0, and VT7 + &C. = 0,
respectively, and conversely each of the intersections of these two curves is a point
situate simultaneously on some two branches of the original yzomal curve vU+ Vy+&c. = 0.
Hence, partitioning in any manner the yzome vl/+ VF+ &c. into an azome, Vt7=&c.
and a /3zome vW + &c, (a4/3 = v), and writing down the equations
of an zomal curve and a /3zomal curve respectively, each of the intersections of
these two curves is a point situate simultaneously on two branches of the yzomal
curve ; and the points situate simultaneously on two branches of the yzomal carve
are the points of intersection of the several pairs of an zomal curve and a /3zomal
curve, which can be formed by any bipartition of the yzome.
14<. There are two cases to be considered ; First, when the parts are 1, v 1 (v ~ I is > 1,
except in the case v = 2, which may be excluded from consideration), or say when the
yzome is partitioned into a zome and antizome. Secondly, when the parts a, /9, are
each >1 (this implies y=4 at least), or say when the i/zome is partitioned into a
pair of complementary parasomes.
IB. To fix the ideas, take the tetrazomal curve \/T7> VF+ Vlf+ V^^O, and
consider first a point for which V77=0, VFf VF + V2 T = 0. The Norm is the product
of (2 B ) 8 factors ; selecting hereout the factors
let the product of these
= U
be called F, and the product of the remaining six factors be called (?; the rationalised
equation of the curve is therefore FG = Q. The derived equation is GdF+FdG = Q;
at the point in question VT7= 0, VF+ VIF + VF=0; G and dG are each of them
finite (that is, they neither vanish nor become infinite), but we have
and the derived equation is thus QdU = Q, or simply dU=Q. It thus appears that
the point in question is au ordinary point on the tetrazomal curve; and, further, that
the tetrazomal curve is at this point touched by the zonial curve^Z7 = 0^ And similarly,
each of the points of intersection of the two curves V~Z7=0, VF+ VF + V^O, is an
ordinary point on the tetrazomal curve; and the tetrazomal curve is at each of these
points touched by the zomal curve U=0.
478
ON" POLTZOJTAL CURVES. [414
. y=0; to form the
m this case the two factors
let their product
" f ^ remainin six factors be called <? ' &* rationalised
atn * n '
uc^ r ^ th f deriV6(1 eqUati n 1S >+<W* I 0. At the point in
hi n 1, , ^ ^ eaCh f theM flnita < tlmt is ' &** neither vanish nor become
minute), but we have
becomes iclenticall)r 0=0; the
:B on ' 96e that ifc is in fect a node  OT *"7 doable
prat, on the tetrazoml ourw And similarly, e ach of the points of intention of
the two curves Vf7+^=0, VF + VF=0 i s a node on the tetrasonml curve.
v I! 1 ' P1 'f ^ " le foreg , hlg * W examplcs ras P^ly are quite general, and
T " "" Z0 ' ml UrVe ' emm te tlle res lte as f"o, viz., in a
T' , F mtS SltUatG sim "l to ^ o two laches are either the inter
, tt / 0mal , OU1 ' Ve aml it3 mtizomal curre  * "y are the intersection,
C .? m P lemental 7 P<1 curves. In the former case, the potato in
^ H T P , mt : n th0 " Z01m1 ' bU * t]le * are P inte f ntoot f >B
1. T J . ' my ^ addOCl ' tlmt the "te^^ons of the zomal and
In Z ' T tW10e ' ate "" the inta "ti f the omal and zomal.
L 'r,' P mtS '" ql ' eSti n am nodes of the " zoma li ;t >"V ^ ^ded,
Lhe ! has no, m ffeiKm l, any nodes other than the points which are thu s
" C mplementar y P""^ nd that it has not *, general
any
Article Nos. 18 to 21. Singularities of a vzomal dune.
18. ^ It has been already shown that the order of the ,zomal curve is =2",..
Considermg the ease where is =3 at least, the curve, as we have just seen, has
oontaot. with each of the ,o ma l curves, and it has also nodes. I proceed to determine
the number of these contacts and nodes respectively.
18. Consider first the zomal curve 17=0, and its antizomal VF+VF+&o=0
the lers
n w >
m S , points Hence the ,zomal touches the zomal in 2r' points, and reckoning
each of 'these twice, the number of intersections is =2r, viz., these are all th?
mtcrseotions of the ,zomal with the somal 7=0. The number of contacts of the
,zomal th the several zonmls [7 = 0, V= 0, &c., is of course =2"rt,
414] ON POLYZOMAL CURVES. 479
20. Considering next a pair of complementary parazomal curves, an azomal and
a /3zomal respectively (a + /3 = y), these are of the orders 2"~ s r and 2* 1 " 2 ?' respectively,
and they intersect therefore in 2" + ^~ 4 T* 2 k ~ 4 r a points, nodes of the yzomal. This
number is independent of the particular partition (a, /9), and the vzomal has thus
this same number, 2"~ 4 ? ta , of nodes in respect of each pair of complementary parazomals ;
hence the total number of nodes is = 2 1 '" 1 r' 2 into the number of pairs of complementary
parazomals. For the partition (a, /3) the number of pairs is [v] v = [] a [/3] p , or when
a.~/3 t which of course implies v even, it is onehalf of this; extending the summation
from = 2 to a = v 2, each pair is obtained twice, and the number of pairs is thus
= ^2{[y]"^[<t] a [/9] p } ; the sum extended from a = to a = v is (1 + 1)", =2", but we
thus include the terms 1, v, v t 1, which are together = 2^+2, hence the correct value
of the sum is = 2"  2i> 2, and the number of pairs is the half of this =2 ( "" 1 v 1.
Hence the number of nodes of the i<zomal curve is (2"" 1 v~l) 2 11 " 4 r fl .
21. The vzomal is thus a curve of the order 2" 9 r, with (2"" 1  v  l)2" 4 r 2
nodes, but without cusps ; the class is therefore
and the deficiency is
= 2"' r [(v + 1) r  6] + 1.
These are the general expressions, but even when the zomal curves E/=0, F=0, &c.,
are given, then writing the equation of the yzomal under the form */lU + \finV +&c. = Q,
the constants { : m : &o., may bo so determined as to give rise to nodes or cusps
which do not occur in the general case j the formulae will also undergo modification
in the particular cases next referred to,
Article Nos. 22 to 27. Special Gase where all the Zomals have a, Common Point or
Points,
22. Consider the case where the zomals U = 0, F=0 have all of them any
number, say k, of common intersectionsthese may be referred to simply as the common
points. Each common point is a S^tuple point on the yzomal curve; it is on each
zomal an ordinary point, and on each aiitizomal a 2" 3 tuple point, and on any azomal
parazomal a 2 a tuple point. Hence, considering first the intersections of any zomal
with its antizomal; the common point reckons aa 2 1 " 3 intersections, and the k common
points reckon as Z^Ic intersections; the number of the remaining intersections is
therefore = 2"~ 3 (r 8 &), and the zomal touches the vzomal in each of these points.
The intersections of the zomal with the yzomal are the /ccommon points, each of
them a 2"~ a tuple point on the yzomal, and therefore reckoning together as %"*k
intersections; and the 2"" (?' a  lc) points of contact, each reckoning twice, and therefore
together as 2" a (r a &) intersections (2" 2 ft + 2* a (r a ^) = 2 l  2 ?' 9 J =r.2"v); the total
number of contacts with the zomals C7"=0, F=0, &c., is' thus 2*" (r a  Is) v.
480 ON POLYZOMAL CURVES. [414
23. Secondly, considering any pair of complementary parazomals, an azomal and
a /3zomal, each of the common points, being a 2 a ~ a tuple point and a 2 /J  a tuplo point
on the two curves respectively, counts as 2 a+ ^~ J , = 2"~ J intersections, and the k common
points count as 2"~ 4 & intersections ; the number of the remaining intersections is there
fore = 2 1 '~ J (? k), each of which is a node on the czomal curve; and we have thus
in all 2" 4 (2" ) i>l)(?' 2 &) nodes.
24. There are, besides, the k common points, each of them a 2"~ B buplo point
on the vzomal, and therefore each reckoning as ^"""(S"" 3 1), == 2 ai '~" 2''~ a double
pointy or together as (2"~ 5  2" 3 ) k double points. Reserving the term nodo for the
abovementioned nodes or proper double points, and considering, therefore, the double
points (dps.) as made up of the nodes and of the 2^ tuple points, the total number
of dps. is thus
2" 4 (2'~ 1 i/l)(r 2 &) + (2 a " fl  2" 3 ) k,
= 2 1  4 (2" 1  v ~ 1) r" + {(v + 1) 2"'  2"} k ;
or finally this is
= 2* {(2'  v l)r+ (yl)l;
so that there is a gain =3r*(vl)k in the number of dps. arising from the k
common points. There is, of course, in the class a diminution equal to twice thia
number, or 2" 3 (i/ 1)&; and in the deficiency a diminution equal to this number, or
2">l),
25. The zomal curves ^7=0, 7 = 0, &c, may all of them pass through the same
v 1 points; we have then k = i* t and the expression for the number of dps is
s =(2'2")r, viz., this is =2"' (2l)r. But in this case the dps, are nothing
else than the ? common points, each of them a 2^tupIe point, the ,,zomal curve
m fact breaking up into a system of 2" curves of the order 7, each passing bhrouffh
the , common points. This is easily verified, for if = 0, * =, fl are somo two curvos
of the order r, then, m the present case, the zomal curves are curves in involution
with these curves; that is, they are curves of the form 1+ M> = 0, m8 + m'<I> &c
and the equation of the czomal curve is ' '*
+ Vm f m'$ + & c . = 0.
The rationalised equation is obviously an equation of the degree 2 in , *
therefore a constant value for the ratio 6 : ft; calling this q> or writing 0=
V^ + I' + Vmq + m' + &c. == 0,
ta 1 '? ^"fT,;: " eqU f n f the d8 ree 2M in ?. d ges there
loie 2 values of j. And the ,zomal clu , ve thn8 brealts into
8 6 9  =
eO!. O X ,
curves U0, $ = 0. The equation m g may have a multiple root or roots and the
ystem of eurves contain repetitions of the same curve or ourree an imtace of
tn (m rfat.on te the trizomal curve) will present itself in the sequel but Ido not
at present atop to consider the question.
414] ON POLYZOMAL CURVES. ' 481
26. A more important case is when the zoinal curves are each of them in
involution with the same two given curves, one of them of the order, r, the other of
an inferior order. Let = be a curve of the order r, # = a curve of an inferior
order r a; L = 0, Jf=0, &c., curves of the order s; then the case in question is
when the zomal curves are of the form .ZXl> = 0, @ + M*0, fee., the equation of
the vzom&l is
&o. = 0,
where l t in, &c, are constants. This is the most convenient form for the equation, and
by considering tho functions L, M, &o. as containing implicitly the factors l~ l , m~\ &o.
respectively, wo may take it to include the form */l + Z<T> + Vm + Jtf* + &o. = 0,
which last has the advantage of being immediately applicable to the case where any
one or more of the constants l t m, &c. may bo = 0,
27. In the case now under consideration wo have the r(r s) points of inter
section of tho curves = 0, <I> = as common points of . all tho zomals. Hence, putting
in tho foregoing formula }cr(rs), we have a yzomal curve of the order 2 11  3 ?',
having with each zomal % v ~*rs contacts, or with all the zomals % v ~ a rsi> contacts, having
a node at each of the 2'"~ 4 ?'S intersections (not being common points = 0, <D=0)
of ouch pair of complementary parazomals ; that ia, together 2''"' (2'"" 1 v 1) rs nodes,
and having, besidoH, at each of tho r(r s) common points, a 2"~ a tuple point, counting
aa 2 a "o_2rs djw., togoliher as (2 a "~ B 2") r (r s) dps. ; whence, taking account of the
nodes, tho total number of dps, is = 2''~ 4 r [(2' 1 2) r(vl)s],
Article Noa. 28 to 87, Depression of Order of the vtovnal Curve from the Ideal Factor
of a Branch or Branches,
28. In the caso of tho r(r s) common points as thus far considered, the order
of the yzonml curve has remained throughout = 2''~ a r, but the order admits of
depression, viz,, the constants I, m t &c., and those of tho functions L } M, &c,, may be
such that the Norm contains the factor <L> U ; tho pzomal curve then contains as part
of itself (<fc w = 0) the curve O ~ taken w times, and this being so, if we discard
the factor in question, and consider the residual curve as being tho yzomal, the order
of the pzomal will be = 2"~V w (r s).
482 (IN roLY/o.MA], ri'ltVKH. j il
or, wlwli is l;ho Htiinn Uiinjj, it w
which uxpiumion may duiihiiii lln' liutur <l>, or ti lii>{lhT ur, ,,\ I I>',. iMun.., h
wn Imvii V/ 1 tot^O, tlm pxpiiiiMinii wilt rtn ..... mlitiu llu> l',t.i. (1 l , mnl ii' u. j.,,,
liavo .A V/ [&<!, :: () (uliNorvi' iJiin i]iii!lr, t m :tl , u ,y r.,naii..n. ,n 1)1,1.. ,ti.' ir.\.u.fin
ln'niw in ,h,> wlioln Hnrirn ,i' 1'iiiu'lii.iiM / .!/, \v. ; ih.n., i! I,, M, A.. ,. ..,.1, .ViL,,,
' tl' l'"i'iii / J f./^ M./I!, willi (,hr> mini' vnl.h.t ,,r /'. I,J, /,', Ion v,,f!, .till. t,ui ,!.,,
< ( f tlui rnolliiiiKhlH , /, c, (Jinn it ittiili<>tt tin lliiv,. i ( imiiiit> K \ / ) ,v, , u, /. s ^ , ,v,. ..u,
V/ i &(..:.(); and KU in ,,, t .r ras,,.), if I HIV /,\7i,\, 1,.. lt ,,, u ,)...., , hl !
nxiBtHHinii will nmtmn Mm liichir K ,! ,. .. , ll,, ,.. ,.,, ,,, ^u.,,^^,,,, .,. l(1 ,,
tluil) Uiu uximiiHinn c.mliuns ,u. IUi..r n .vriuin jmw,,r ! i ! , ltll ..,, 1( . ,,), 1(  () ,M
l j,, J,i H .,. IUH   (1 , ri , fiUll M , ,,, .; ...,, lli .
, .
il 'l^hWti.ft (l tlin NDHII will rniihiiu l.lir liiri.n !".
30. U IHIH lh!un MM.ntin,,,,,! ,hnl,  l( . fi irill V /(H ,. /,,,, ( ,,.,,,,,1,.,,.., is , !
H H .M., U,nt i, ],.,,/... ,, ( . , iprlll v'M If i,, ,!. ,. imllll .., ^ t .
i "y Hiiiilt torni .li.r i.iHiuu,',, if i t , ,,,(,, ),, ^ l,\> , V )I:I H f
,, , ,. + fe . u,,. Nimi wi1 ,,, .
o Umt now .. A &,. ,,,, ......  ,, , ..... v ,,',' ',
 ...... ^
*,... vvii,
" wi
414] ON POLYZOMAL CURVES. 483
being so, the general theorem is, that if we have branches ideally containing the curves
0" = 0, <[:>0=o, &c. respectively, then the vzomal curve contains not ideally but actually
the factor <!>"> = (w ~ a + /3 + &c,), the oi'der of the i/zomal being thus reduced from
S"" 3 ? 1 to 2"~ 2 r <a (r s) ; and conversely, that any such reduction in the order of the
uKoinal arises from the factors <D = 0, <E>0 = 0, &c,, ideally contained in the several
branches of the ynomul.
32. It is worth while to explain the notion of tin ideal factor somewhat more
generally; an irrational function, taking the irrationalities thereof in a determinate
manner, may bo such that, as well the function itself as all its differential coefficients
up to the order l, vanish when a certain parameter <D contained in the function
is put =0; this is only saying, in other words, that the function expanded in ascending
powers of <D contains no power lower than <I> ; and, in this case, we say that the
irrational function contains ideally the factor <!>". The rationalised expression, or Norm,
in virtue of tho irrational function (taken determinate!/ as above) thus ideally con
taining <I> a , will actually contain the factor <D K ; and if any other values of the
irrational function contain respectively <l) 3 , &c., then tho Norm will contain the factor
<l)+fl+&0. >
88. A branch ideally containing <D = may for shortness be called integral or
fractional, according as tho index a is an integer or a fraction; by what precedes
the fractional branches present themselves in pairs, If for a moment we consider
integral branches* only, then if the jjfjomal contain <I> = 0, this can happen in one way
only, thore must bo some one branch ideally containing <E> = 0; but if tho ugomal
contain <I> Q =0, thon this may happen in two ways, either there is a single branch
ideally containing <D B  0, or else there arc two branches, each of them ideally con
taining <T?eO. And generally, if tho yswmal contain * M =0, thon forming any partition
to = a.  /9  &o. (tho parts being integral), this may arise from thore being branches
ideally containing <l>" = 0, <1>^ = 0, &o, respectively. Tho like remarks apply to the case
whero we attend also to fractional branches, thus, if the j/zomal contain <D = Q, this
may ariao (not only, as above mentioned, from a branch ideally containing <3> = 0, but
also) from a pair of branches, each ideally containing <t>* = 0. And so in general, if
tho //xonml contain <D0, the partition w = 4 ft + &o. is to bo made with the parts
integral or fractional (= J or integer H^ as above), but with the fractional terms in
pairs ; and then tho factor *" = may avise from branches ideally containing < n = 0,
484 ON POLYZOMAL CURVES.
is, through a single branch ideally containing = 0), than the i'zonml will havo Un
branches, each ideally containing = 0, and it will thus contain <I> 8 = 0, In fact, ii in
the zomal and antizomal, or in the complementary parozomala, tho branches whiiOi
ideally contain $ = are
+ &c. = 0, V^(0+ 17*) + &c, =
respectively (for a zomal, the f&c, should be omitted, and the first equation bo wril,l,i>n
= 0), then in the vzomal there will be tho two branches
rn or
'
&c.) = 0,
each ideally containing O = 0.
Conversely, if a ,zomal contain $ by reason that it ha. two bmnolios m.
ideally contannng $ = 0, then either a zomal and its antizomal will each of thorn
else a pan of complementary parazomals will each of them, inseparably contain * <).'
35, Reverting to the case of the yzomal curve
V((0 + i$~) + Vm(04TM) + &ci. = 0,
which does not contain 4> = 0, each of the r(r.) common points = 0, d>  u, n
of i e JZ n , tle t r ZOmal; h of * oounto thorefL for 2 i
 /r o 7 ia tr\ TT 0==
f1SeCtl nS f ourvo
pas.es 2 time t i ea h ^^^ ^^^ '
i h ; but
c
the ,zo^alpasses therefc e only 2 ' T ^ Th r 8iflllttl
each of the ,(,^) comnon , l ts if a f I^f *? 00min011 ^ J thftt !H
ve *. meets the ^ i "" '
r . r
the r(, 5 ) common points, each of them a ?2 u "i } P  ' Vif! ;' thfl8 inolml11
therefore counting together as tor* \, ~ ft))tu P 1 P oint n the wsomal, mirl
(r.) other intersections o/the ^^ StT
fche fact01 '
intersections of any zomal with it an^Vn T" 1 ^ 0== ' *" th0
the ,zomal, and the ^jj h ei Z cT ^ ' "^ f the
are nodes of the , 20 mal, it bif^Zd laTi ? ^ ' r*" 811 ^
contain a power of cD = such DOWP f <J 7 zoma1 ' ail(iiz oraal, or paraxomul
residual cmves attended io. The n,l! V ^ fc b ^^^ed, and only th
Articular case be investigated without diffl u 'f "^ f n dea V ^ ^'/
   ,, L
'1:14] ON rOLYZOMAL CURVES, 485
<i> w limy prismmt ibwslf, idoully in a mnglo branch, or in novoml branches, and lilio
<!01H(!(lUmli OniUlTOlKW 111 I,ll0 llllilX)! 1 (!IIH(! of jHnVOl'HOf t.D0 j a cOl'lilUll of tllO HOmills,
imhixonmlH, or jmruxmnuH Uio MWOH to bn uoiiHulomd would bo voiy numoimw, and
thorn IH mi I'oimim to luiliovn lihiif) fcho nwultH could bo prammljoil iu any inodomtoly
Cdiiijitio Corn i ; .1 lilnmifin'o iibsLain IVtini on boring on Uiu ([
Ucl(! NOH. JtH and Hi). Ow tho Trisomtd Ouruo and the Tetmsomtd Curve.
!tH. Tliu tri/i until wirvn
for iliH I'lvUoiiidiHi'il foriii nf
tf<+ K" + IP  2 K1K 2 IK/7  2Z7K ;
in 1 u.s thin may idnu bo wrilton,
(1, I, I, 1, I, l,}(f/, V, 117 = 0;
nut! wo imiy 1'nnn tltin mtioinil I'ljualJon vonfy tho xontM'til roHidtH ii])plioablo to tlio
OIVHH in hiind, vi/.., t.hai. l>ho U'ixonuil in u (survo of Llio ovdor 2j', and fchiit
// :(}, ill Uiush uf Us fl l<i itltoi'MIJtStioilH Wllill Y IT = 0,
nmH!Olivtily 1nunh ilio ii'i/.nninl, 'I'lmro urn not, in gonoral, any notion or (umpH, and
bho owloi 1 boiii(  2/', dltti cluHH in 2r (2? 1).
!H), 'L'lui tnlnvxninul i
48G ON POLYZOMAL CURVES. [4 I 'I
Article Nos. 40 aud 41. On the Intersection of two vZomah having the same
Zomal Curves,
40. Without going into any detail, I may notice the question of tho intorauctiim oK
two yzomals which have the same zomal curves say the two tmomals V(7+ VFh Vir () f
M+t/mV+t/nW^Q, or two similarly related tetrazomals. For the trissomals, writing
the equations under the form
VF + VK+ V IF = o, VJ VF + \lm \/ Ff Vw V"F = 0,
then, when these equations are considered as existing simultaneously, wo may, without
loss of generality, attribute to the radicals \/F, V?, \/W, tho sumo values in bho two
equations respectively; but doing so, we musf^ in_ tlio .second aquation suoeoHnivnlv
attribite to all but one of the radicals V Vm, V, each of its two opposite,
ior the intersections of the two curves we have thus
viz., this is one of a system of four equations, obtained from it by changes of Hi K n
say m the radicals V^ and V. Each of the four equations gives a not of 7" poinln '
we have thus the comnlete m w, .4,.. of the ^ flf intoraootion of fc , () ;
curves.
the form
41^ But take, in like manner, two tetrazomal curves;, writing their oquatio.iH h,
VF+ V^ VF + V;j Vf =
. , m ra " fl " 
,. ,r u 7 , eCOnd equatlon attribute successively, suv
hat' ea h f Ulei " ^ 01 " 3 Site U  ^ * "^oction, of tho ^
tnzoraal curve, of the order 2r rt. 7' . ^ ualloM raprosont each of thorn ft
if each of these , a point O ' f ,1 ""''T f 4 180 ' thBteforo in """ P oi ts . "
in all 8xV = 32,< intersections at, f "" tW teto20malfl . should havo
the order 4r, and they intersect tl" Lf tetra mala ara each of them a ourvo of
that not all tho 4^ P Lt s ta onf 2 ^f ^ n ' y 16 '" l )oi " te ' T1 ' explanation i.,,
In fact, to find M the interactions of the * t i , n ' eraection6 f the tahwnmnh
equations to attribute opposite sims tn 7 , tazomals  ll JL.neo6ssary in theii two
intersections from the equationa as tly 8 Z, 1 tM ^ S ^ W ' ^'' we obtai " 2 '"
two equations after we have in , , remalnm S 2r intersections from the
the second equation reversed the sign, sav of VP
4H] ON POLYJJOMAL CURVES. 487
Now, from tho two liquations as they utand wo can pass back bo the two tetraaomal
oqnationH, and tho IhHtmoutiunod 2r a points arc thus points of intersection of the
two totimomiil ourvon li'oni tho two equations after" such reversal of the sign of
v'J.\ wo ouniHit PIIHH buck to the two totrazomal equations, and the lastmentioned 2r a
pointH aro thiiH not points of iutomHition of tho two totrnxomal curves. The number
of intm'HOutionn of tho two curvuH IH thiiH 8x 2r'\ = lo'r a , as it should bo,
Artido NOH, 42 to 45. Tho Theorem, of the Decomposition of a Tetmzonud Curve.
42, I (soiiHidur ttio liotroxnmal ouvva
V/ if  Jin Y + */n'W t x/pT  0,
wlun'o tho HI mwl <mrvoH arc in involution, that in, whoro wo have an identical relation,
atfl l)Kf isFl d2' = 0;
and 1 jiroctMid to nhow that if t, m, n, p satisfy tho 1'olation
1 i m _i n a. ? J  n
I .(. ( r U,
a It o d
tho (jurvo hiciakn up into two tiiximialH, In fact, writing tho equation under the form
(V* It I Vwt V + Vn IF) 3  ^S 1 = 0,
and HiilmtiUitin^ for .7' HH viilun, in toniiH of //, 'I 7 , TK, tliiH is
(id ?m) ?/ + (MM! I ;b) K + (?id +^o) If
. 2 Vwrnl V K1K I a V/iiii V !Kf/  2 Vfowd V UV ;
or, (ioimHUtvin^ tho lofthand nidn as a qnadrie function of (Vtf, V K, Vlr), the condition
for itH breaking up into Ihutora in
j l(\~\pn,, d
d A/Hii , d ViHi,
that in ,11%
y/ J ((bed I 7/icdn. I dab +juabo) ~ 0,
or finally, tho condition JH
i + ?? !., L l + o.
a + b ' c + '
43. Multiplying by M+;MI, ftnd observing lihat in virtue of biio romtion wo
, , , , i i
(id h pa) (wwl +pb) fond" 
(hi + pa) (nd I yo ) ind" 
488 ON POLYZOMAL CUBVES.
the equation becomes
or as this is more .conveniently written
an equation breaking up into two equations, which may be represented by
where
where, in the expressions for Vi, &c., the signs of the radicals
may he taken determinate!)' in any way whatever at pleasure; tho only oHVct, >f
alteration of sign would income oases be to interchange the valuos of (VV^Vw,. \
with those of (V/ 3 , V^ 3l Vtt a ). The tetrazomal curve thus breaks up into two tm.mm
44. It is to be noticed that we have
, a m
c + bd
thati
s
and that similarly we have
414] ON POLYZOMAL OUEVES. 489
The meaning is, that, taking the tmomal curve */lJJ + VwijF"+ Vn 1 TT = 0, this regarded as a
totrazomoJ curve, V^ U+ Vm, 7 + VnTl? + VO!T= 0, satisfies the condition  + ^ + + ,==0;
t G Cl
and the like as to the trizomal curve \
45. The equation by which the decomposition was effected is, it is clear, one of
twelve equivalent equations; four of these are
,o
dV/'
bed
V ? 7iiJi V?l VI }(
a VHI ' ' '\
cda ?)i
, u , "vn + ,j ^ > ) 1 "
b v / \
c ??i
dab 11
n I i\ ./ ' i ^^ \ /
Vi , VTO , , \'p + V
cVp \
abc p
and tho others may bo deduced from these by a cyclical permutation of (U, V, W),
(a, b, o), (I, in, ri), leaving T, d, 2? unaltered.
490 ON POLYZOMAL CUllVJflS.  H I
viz., the trizomal curve JlU+ Vw7 + Vi? = 0, if a, b, c bo any (jiianhitiioH wmmutrd
by the equation
a b c '
(the ratios a, b, c thus involving a single arbitrary parameter); and if we tal<n 7' a
function such that aZ/>bF+cTK+d2 ? =0; that is, '/'=(), any ono of l.lus wunvs nf
curves af/"+bF+cF= 0, in involution with the given ourvon ?/=(), K0, IK" (),
has its equation expressible in the form
that is, we have the curve 2'=0 (the equation whereof contains a variable
as a zoraal of the given trizomal curve \/lU\ \^iiiV~\~^n \Y(); and \V(i liuvn lliu*
from the theorem of the decomposition of a tetrazomal dodmiud (.lio Uujurtiiii of t!i^
variable zomal of a trizomal. The analytical invc.sfiigation in .snmowhat, HJitipliiiiil \ty
assuming p ~ alt initio, and it may be as well to rupuut if, in UUH iorin.
47, Starting, then, with the trizomal curve
nW = 0,
mid writing
as the definition of T, the coefficients being connected by
a b c~ *
the equation gives
.' substituting in this equation for W its value in terms of U, V, T, wo Imvu
(aw + cl)U+ (bn + cm) F + 2c ^llmUV  dnT = 0,
hich by the given relation between a, b, c, is converted into
ac , r be
fttiu (F+ 2c vlftiU'V't dnT = '
Da *
lat is
n j i
z,, this is G
finally
414] ON POLYZOMM, OUBVES.
48, Tho result just obtained of course implies that when as above
491
the trizomal curve *JlU+ VmFt vnW Q can be expressed by means of any three of
the four siomals U, V, W, r l\ and we may at once write down the four forms
* _ A / A/ Wvif VT vw \
c" ' V b a ' VabcJr^ VK) VIT) ^
abo*
abc
nd
abc
Wl
abc'
abc'
the last of which is the original equation
that if the first equation be represented by
have
and therefore
= 0. It may be added
w,F+ V^T = 0,that is, if we
n
_. LJ. 0
i r r " i r i "" h ~ v >
bed bc\a b c
or if the second equation bo represented by
have
lf + Vp a 5P  0,that is, if w&
V4 =
and therefore
r
c d
or if the third equation be represented by
have
and therefore
nd
T~ "T T
b d
622
492 ON POLYZOMAL CURVES.
then the equation of the trizomal may also be expressed in the forms
ad '
ac
Ajjbe
1 ~V acf'
ab
ac

ab '
Mac
V bd
ab
be '
ab '
and
/
"V
jJ 3 ab
ed '
49, These equations may, however, bo expressed in a much more elegant form.
Write
a '= a b ' = ' c ' ~ d '
037*)'
where, for shortness, (^S) = (/9  7) ( 7  S) (S  /3), &c. ; (a, ^, 7 ) being arbitraiy quantitioH
or, what is the same thing,
Assume
: TO : re = pa'
: erb' (7  a) a : re' (a  /3) s ;
then the equation  + ^ +  = takes the form
a b c
414] <>N I'OLY/OMAL <:
anil tho lour forum <if ,ho o(uabiou ani found tu IMS
vix., bin wo urn l>ho tiquivalnnti forum of bho original tt<piubion awnumod to bo
(#  7) V,m'tf I (7  ) ^ II ' I' I' ( a  #) ^ Tl! ' 1r "= *
50. I rniiuii'k tiliuti tlin Mittoii'in nl' tlm variable minml may bn itbfciunwl us a
(jiuiwrnniHil.iini Llunrcin vix,, mMipimiitf tlu' tiuilii(m V///I Vml^l V/i.lF with tho
(!iiuiili()it \^/i"l ^mi/'\ V~ ! ; 0; l.hw liiHli boloiiKH t H tmi<i lumditHl by liht^ blmm linos
, T/r:i(), IT "''  Hill C(Ulll.inil 1)1' .}0 HlllllH *ini)i(l IUHKt, ili W lcMlV, lit) OXpl'(!HU)l ill
a Hiinilur I'nnn by ini'iiiw nl' any utlmr tlinm ttin^i'.nlH bliorctjl', Tmt tho oqimtiou of any
laugonl. "I' Mm c'unir IH iui I li/y I rc= '(), wluini a, b, art; any (niLul.itioH HiitwlyinK bho
romliUmi l '" !' ^O; whnniHi, wiil.iiiK miH ly/H^Hlw ^ 0, u ' (! llia }' infiioiltuio ?w =
(i h c
alontf with any Uvo iti 1 Mtit <r'^uml winiiiU /"(), ,v^0, s(>, or, instead of thorn, any
tihmi AuHiliiniiH 1 nl' Mm lunn w\ anil Mum Mm nmvn (thivn^i "'' "', ?A ^i ""' int_ 'A K ^. ^
givon bho Llii'iii'i'iii. Hub it, w IIM oany Mi iMiiiiliinb l;lm analyHW with (U, V, W, T) m
with (f, .i/, ~, w), unil. mi I'nniliic.l.nil, it in roaliy Lint Htinm iHialyHiH IIH bhab whoroby the
(ilworiMii in 1'ni.abliMhnil KH(, Nn. '17.
fil. Ib in worbli whiln in uxlubib bho iMiuutiinii of bhu *!iirvn
m n
, , . t .
in a liirm luMilainiiiK Mnvn now wniialH. OliMorvo ft hub th( oquaUon (  I  }) +  U is
by aty;x. bm^, ,,% if mily \~ $ I ^  ; or .say, if '(*",
(( ( ^ it rft". Thn ( M[iuibuni
X V(. f (it  '") / ^H ("' "") ('  ) " l ^H' ( ft " " (t ) < lt "  rt/) B 1V
+ ^V(/)70(ft/O^+X6'"'O(^'0t^ + CA''"'0^^^^
. p V ( ;  V o  ") it/+ (o 1  ") (o'  o)*'K+ (?" V) (0" V) If
IM ciniHwiuontty IIH omiabion involving thvoo /.onialw of tho pi.i ........... , 
inino \,n. v in Hiuliwim UK In identify thin with tho original wpuitiiMi VH7I
vi/., writing Hiiertssivuly //(), K0, ll r 0, wo find
(it 1  (O X I (V  b") p I (o'  c") i; = 0.
494
ON POLYZOMAL CURVES.
[414
equations which are, as they should be, equivalent to two equations only, and wlriuh
give
X : ft : v= I, I, 1 ; 1, 1, 1 ; 1 , 1 , I
b, b', b"
C, C , G
a , a , a '
c, c', c" a, a', a" b, b', b"
ami the equation, with these values of X, p, v substituted therein, is in fticb tho
equation of the trizomal curve >JlU + *JinV+ */nW = in terms of tliroo now xomuk
It is easy to return to the forms involving one new zomal and any two of tho
original three zoraals.
Article No, 52, Remark as to the Tetrazomal Curve,
52. I return for a moment to the case of the tetrazomal curve, in order to H!IDW
that there is not, in regard to it in general, any theorem such ay that of the variable
zomal. Considering the form vV+ Vmy  V* + */~pw = (the coordinates #, y, z, w nni
of course connected by a linear equation, but nothing turns upon this), tho curvo M
here a quartic touched twice by each of the lines to=>0 t 7/ = ( z = 0, w (via., oiicjh
of these is a double tangent of the curve), and having besides tho throe notion
(K=IJ z = w)> (x = 2, y^w], (01 = 10, y = z), But a quartic curve with throe nodos, m
trmodal quartic, has only four double tangents that is, hesidea the linoH a=0, ^=0,
s0, w=0, there is no line <w + fy + yz + &w = Q which ia a double tunwmt f bho
curve; and writing U, 7, W, T in place of ,, y> z } W> then if U, V, W, T um
connected hy a linear equation (and, d fortiori, if they are nob HO connected), bhoru \*
not any curve Z7 + /3F+ 7^+8^0 which i s related to tho curvo in the sanio way
with the hues ff0, 7=0, W = Q, 2'=0; or say there i s not (besides tho oum*
1 , J 31=0)l any ther zomal a^+jSV+V^ + Sa'^O, of bho liobrtusnnml
curve The proof does not show that for special forms of U t 7, W, T thoro nmy
not be Z omals, not of the above form ff+/97+ 7 F+ SZ'O, but bolonging to iv
separate system. An instance of this will be mentioned in the sequel,
Article No, 53 to 56. The Theorem of &, Variable Zonal of a Tribal (hn
53 I resume the foregoing theorem of the variable sonml of bho tfuomal ourvo
F.a The vanable zomal 2'=0 is the curve aZ7 + b7+oTr 0 wh.ro
a, b, c are connected by the equation i + * + ?_ ; that is, it belongs to a Bi
, the
of any one O f the om . vos
414" ON POLYZOMAL CURVES. 495
and of two of tho thrno givon emvow, in identical with the Jacobian of tho three given
flvu'voH. I (iall to mind that, by tho Jaeobian of the curves 7=0, V=Q, Tf = 0, is
mount tho enrvo
J(V t V,
^
' Y>
y , &
d a U,
tl a V,
d a W,
~ 0,
vis!., bhu imrvo obtained by equating to HOVO tho .Tuoobiun or functional determinant
of tho fmietioiiH U, V, W, Hume properties of Hie Jiusobian, which arc material as to
what follows, arn nuintioutid in tho Annex No, I,
lj\n tho (Mimiiloio statement of tho thooroin of the variable xomal, it would ho
miooHwuy liu intei']ii(!l, Hdoinntiicully the condition a H ','J +  => 0, tlioroby shewing how
tho Hinglo HoriiiH of tlnj vai'iiibhj wnnal i Holodtod out of tho double aoriim of tho
c.iirvoH nU + W+uW^Q i" involntion with tho givon ourvos. Huoh a geometrical
iiilitn'pi'otatiou of (,\w ooiulition may bo nought for OH followw, but it is only in a
liirti(iliH (!umi, HH aittirwnnlH umiitionod, that a w>iivoniont geomotrifial intcrpvetation in
th^itihy obUuned.
54. Uoimid.'i tho lixed lino 0=^1^1^ = 0, and let it be proposed to find
tho IOOIIH ol 1 tlio (r'l)' 1 iiuloH of tho lino 11 in rogaitl to tho norioa of curves
atf + l)V>Tr0, wltitnt '+7 + "0. Take (, i/, a) .w tho tsooHliimtoa of any one
a o (>
of tho noloH i iUOHtion, thon in order that (a, ?/, s) may belong to one of tho
(rl)' J poloR of tho line <}/' I wM'sO in regard to the curve ai/HbF
wo muHt havo
or, what in tho namo thing,
and filiOHo uquatioiiH give without difficulty
n : b : c,/(K, W, O) : J(W t U,
whonco, Hubntituting in tho equation ^ + ^ + 0"' wo Imvu
as tho locus of liho (*!) polOB in question. Each of tho Jaoobiana is & function
of tho order 2r2, and the ordor of tho IOCUB is thus =4r4. A a tho givon curves
y0 F0 TK0 bolong to tho single aeries of curves, it is clear that tho locus
puna' through tho 3(r~l)' points which are the (rl) pole, of tho fixed lins in
regard to tho curves Z70, 70, W 0 respectively.
49G ON POLYZONAL CURVES. [414
55. In the case where the given trizoinal is
= 0,
s = rl, that is, where the zomals + .M> = 0, + M> = 0, + lYcI> = are ouch of
them curves of the order r, passing through the r intersections of the lino <!' = ()
with the curve = 0, then, taking this line *() for the fixed lino fl0, wo havo
J(V, W, fl)=>J(& + M^, + m, <D) = <I>{j/; JV},
Hi for shortness, (tfjn '("# Q, *)+*/(# fl <!>), anfl tho liko M to ^ othci .
two Jacobians, so that, attaching the analogous significations to IN, L\ and \L W the,
equation of the locus is ' l Jj
where observe that each of the curves [M, N}^, {N> L} ^ ^ M] = Q ifi a flnrvo
of the order 2r3; the order of the locus is thus =4 r _ G and f aa bolbre) th[s
locus passe, through the 8 (rl)i pomts which are the (,!)* ob B of ^ Uno *0
m regard to the cnrves e+iiO, e + Jtf*0. 8 + JN&0 respectively.
56, In the case r=2, the trizomal is
where the zomals are the conies + i* = o, 6 + ^ = 0, 0 h TO = each naHsin f v
through the same two points = 0, * = 0; the loaus of tho pole of 1 o b ~
m regard to the variable zomal, i s the conic l '
polos
and second of tl e en V "' ^ ^ "^ ^ md tho (U '^
thereof, and it M y ", * ^ ^ jMobian c ic
case in hand. (See as to 1 Au LT' T** " f ^ thaOTom ^ tho
with the plan dopted i h L f lf 'n ^ "^ t0 ' } B '* if ' in acco ^ ce
point, = 0, = are til ,1 ^ memoir ' TO at nce " that the
. 
any three circles, then in the trizomal ^ the ec l uafcloris
a i
414] ON POLYJ50MAL OUBVIflS. 497
coefficients I, m, n are nob given in tho first instance, but arc regarded as arbitrary,
then tho hwlimontiouod conic in any conic whatever through tho three centres, and
there belongs to mioh conic and the norios of Komalw derived therefrom an above, a
Imomal urvo VWI" I Vm^}" 4. Vw(" = 0, Thin in obviously tho theorem, that if a
variable oireli! linn itn centre on a givon conic, and cuts at light angles a given cirolo,
tliou th(i envelope of thu variable) circle in a tmoinal curve V/?r + */inW { VV(,
whore VI" " 0, s iV'=0, ($"=s() are miy throo circlos, positions of the variable circle, and
I, lit, n an.) oniiHtaiiti inantitioH depending on thu soletttod tfu'co circles.
1,'AIIT II. (NoH, 57 to 104). iSUHHIDIAHY iNVEHTIQATIONft.
Article Nits, 57 and f>S, Preliminary Memitrlca.
57. Wo luivo JHHt henn led do consider tin; ooiiHiH wliifslt piws bhrough two given
s. Tlusro in no real IOSH of gunomlity in taking thenc to be tho circnlai 1 points
ut infinity, or my tho puinls /, ./" vix,, ovory tbiioreiu which iti anywine explicitly or
implicitly rolut.oH tu tht'Hc twn points, may, without tho noeuHsiliy of any change in the
Htateiuont tbiM'oof, bo undorstond IIH a thoorom rolating iiiHtoiul to any two points P, Q.
I cull to mind that a oircln in a (ionic passing through tbo two points /, J, ami
thai; Miuw at right angles to each otber aro lines harmonically related to tho pair of
linuH from thoir intnrsiiotion Lo tho points /, ./" ruHpocbivoly, so that when (/, ( /) are
ntplucud by any two given points whatever, tlie exproHHion a cirolo niUHt be understood
to mean u onnid pnHHing through tho two jyivou points; and in upoalting of UHOH at
right uugloH to (Mush other, it, ivmst be undtiwtood thub wo moan linen harmonically
nilutod to tiho pair of linen from their interdiction to the two given points respectively.
For iiiHtinico, the tluiorcm that tin.) Jac.oliiun of any throo circles ia their orthotomic
oirolo, will mean that the Jaoobian nf any throe conies which each of thorn passes
through the two given points IH tho orthotomio conic thnnigh tho name two points,
that w, thn conio Hiioh that nb catsh of its intowoctioim with any one of tho three
(WHICH, thu two tangnntH arc Imnmmiotilly rulated to tho pair of linos from this intor
Hootion to tho two given points rcHpeotivuly. Such cxtendod interpretation of any
theorem in applicable ovon to the thooromH which involve distances or angles viss., "~
torniH "distaiuin" iviul "angle" have a determinate nignification when interpreted
roferemio (not do bliu oircular points at inlinity, but hmteml thereof) to any two
pointH wluitovor (HKO as to thia my "Sixth Memoir on Qualities," NOH. 22"
Phil. !/ ! mwa. > v(l.cxT.TX.(lKfill),pp. (U~!)0; HOC p. HG; [1C8]. And thia boingao, the theorem
can, without change in tho Htatoment thereof, bo understood as referring to tho two
given pointH.
58. 1 nay than that any theorem (referring explicitly or implicitly) to tho circular
points at infinity J, J, may bo understood an a theorem referring instead to ^any two
givon pointH. Wo might of course givo tho theorems in tho first instance in terms
explicitly referring ta tho two givon points (viz., instead of a circle, apeak of a conic
through tho two given points, and so in other instances); but, aa just explained, this
is not really inoro general, and the theorems would bo given in a less concise and
0, VI. G3
498
ON POLYZOMAL CURVES,
4  .(
familiar form. It would not, on the face of the investigations, be apparent tlmt in
treating of the polyzomal curves
&c. = 0,
(0 = a conic, $ = a line, as above), tlmt we wore really treating of tho curve*
zomals whereof are circles, and therein of the theories of fool and focofnci us ill
to he explained, And for these reasons I shall consider tho two points <$ = (), <I>
to he the circular points at infinity 7, /, mid in the investigations, &c., mako MH
the terms circle, right angles, &c., which, in their ordinary .significations, havo hnp
reference to these two points,
The present Part does not explicitly relate to tho theory of polyzonml <jurvi%
contains a series of researches, partly analytical and partly geometrical, whioli will
made use of in the following Parts III. and IV. of the Memoir.
tint
Article Nos. 50 to 62. The Circular Points at Infinity; .Reotanyular and Circular
Coordinates.
80. The coordinates made use of (except in tho cases where tho gonoml (riliu^r
coordinates (,, y, ,), or any other coordinates, are explicitly referred to), will l,u ,i,,,,,
the ordmary rectangular coordinates K , y, or else, as wo may term thorn, tlu, nm.lur
in
for
. it boing uBdoto,Kl tlmt,
n ret i, '
ein
eitteftr
V d?f i Th s r? " ft Mlcl ' (=:l) ' olse
g> T/, .am 2^lj. ihe equation of the line infinifv i  n n , r r
by the equations (,, + i,/=0 , N !, t ^ T ! th P mtS J ' ^ '"'" 1V ""
by the . nations (f4 /= fa dtflTn' '""V! Wh ' ^ th MU " n
coordinates the Jfdinaies of^ ^'.T^TS ^ T m
the circular coordinates they are (1 00)11 m ' n ^ ', } r
"
the points /, J, are
*
respectively, or they are
ec t ua ^n S of the lines through A to
respectively These equations, if fo a ' } , (fl fl , flm . .
equations of any two lines through the JL. 7 r arbl ^ry, will, it is clear, bo
6 "ie pomes 7, ,/, respectively.
I t I (IN '(i,Y/,t)MAL (MJIIVKH.
lil, Wr liavc iruni I'ilhiT nl' iJh 1 1'iiiuit'ioiiK in (n\ //, 2)
flml i'i. iliu diMluiH'c I'miii riH'h nl.lii'i 1 n' any twn points (,', #, I), ami (<(, a', I) in a
lint* lhriiii'li / IT ./ in Ml. And in piu'liriiliu', il' 5 (I, llum nj 1 ( ;'/ J ; thub IH, t.bu
ili^.lain1' nl' tin 1 i"inl> (n, i/', I) fmiu / ur ./ in in ini'li CUHP.U).
li',1. ( IiMt'ihlri 1 i'T u fimnii'iil. miy llirrr {miiifs /', Q,A\ tlin miq)uu<!i(Miltt.r (lintiui(!(t
nl /' iVuhi (,?.! i' ! Iviinifji' I\)A ;> ilhtimiv (,).!; if (,J lin uny puinl, nn Uio Hun
tln.mi;li rt hi rifht'i 1 nl' ill" iuinM /, ,/, anil in wrUimliU' il' Q lin iiildim 1 ul" \;\w iioiiilH
/ t ./, llti'ii ilu< tiituujlt' I\)A i" liuilo. ttiil. tint (liM.nntt<! (,)<'! in ^:0; iluUi IH, llui
it'ipcuiliriilitr tli'iliuitc Ml' /' I'miii llif linn l.linnijfli ,!. In niUitT nl' Uin (niinlH /, ,/,
lliul iii. I'rnjii uny linr llimii[.;li i'illu1 1 nl' llicjio iminls, In . ; .xs , Hill., an jiuil, lal(!(l, (illr
hiiuii;!*' /'(,M i'i (in ih', ii' miy tin 1 iriiuitflfM f'(A, I 1 JA urn i>mtli linil.n ; vi/,., ihn
i'nniiliiMih<:i (riTtmi^uliii't "i /', ^l li"iiit; (*i\ tf, s  I ), (</, a', I) uv (riixuilar) (f, /, c I ),
\ ! t. '', I), llu 1 i'\Hi'^i"iri fur ith' ilnnlili'M ul' lliisn l,riiuij.;lt'M ivspiMitivnly urn
u, i'.
I lira i'i, tlit;v in 1 . (iv.'fMiinlur rHui'tli
(riniilur nti.i'ililiuh'Ml ^' fl., i/ ft',;.
It.'pi..riiliH^ llin i..til.l inriiM 1 1) 1 /'/.'I, KM, i
lit III.' pliilll'. .1. /', t'V VI, Wr IlllVi'
i'l ,(.*. us)' I {.V '"'
1,1,0
> i
ir, it, 1
;r  i(~ I '/{// u'c), w c/5 l <//  't'*). *"'
, mill tliu Hiuun;tt
500 ON POLY201TAL CURVES. [4L1
whence the equations of
AJ, A } J are f  as = Q, 7;  $z = 0,
BJ, B,J f&: = 0, 17^ = 0.
65. Considering any point P the coordinates of which arc , ?;, s ( 1), ,it
91, S3, SI,, 53, be its squared distances from the points .4, 7?, A l} .B, respectively; klmn
by what precedes
91 = fl'*
i ,
and thence
SI. 23 = 51,. 33,; .
that is, the product of the squared distances of a point P from any two points A, ft,
is equal to the product of the squared distances of the same point P from thu U,i
antipoints A lt E } . This theorem, which was, I believe, first given by mo in llu>
Educational Times (see reprint, vol. vi. 1866, p. 81), is an important ono in tho fclmury
of foci, It is to be further noticed that we have
if ' K t =(a a ')(/3/3'), be the squared distance of the points A, S, = Hnunrod distal*
of points A lt S lt
Article No. 66. Antipoints of a Circle.
66 A similar notion to that of two pairs of appoints is as follows, via, if
irom the centre of a circle perpendicular to its plane and in opposite HOUHOH, w,,
m asm, off two distances each < into the radius, the extremitioa of bhao
r cn ; , B is clear that the ****** f
hef ^ (ln the lailG f
ni Tt n
e7or L .f " ded ^^ 6aCh ******* * the cen ^ of
e on o he twn e ^ te crcle is * it
sect.on of the two cone spheres havmg their centres at the two antipomte reapootivoly.
Article No. 67. Antipoints in relation to a Pair of Orthotomic Circles.
*^ ^ ts c "' *
these two circles cut .U
if . i
,, Ah part cute fo , r n 6 Une AB V "ogh the
clebe a irde
tci ia
radii in the ratio 1 i and T * . bein ' lt la olear . concentric oirolos with thuir
^^r*^ r h f at > i
, 01 say they are concentric orfchotomic circles.
ON t'OUYXOMAL iJUttVKS. 501
Arl.idr NUM. ()8 to 71. Jfornitt of the Jfywtwn of u Circle,
liH, In i vrli tubular I'.nm'ilmnliss tho liquation of a oirclo, coordinate!* of centra
(it, if/, I) mid nuliiiH t*u", !H
VI"  O  (ts)" I (,y  it's? ~ rt'V  ;
niiil in nmmlur I'tiniiliiiuUiH, Uio itonriUniititm of tho cimbiii being (a, a', 1), and radius
II," us llrfni'i', (.lie. i!iMlll<Sini in
(ill. I nliMnmi in t )ll ' w '"K' (i ' ial ' tiM1J oi'iK" 1 boinjj at) tiho oontro and tho radius
lnMiig > ' I, tlii'ii wriUn^ iilnn ofil, Uu) (Miniition ol' Uio (JU'olo in ^==1, that ia tho
riiv.iihii' riKmlinul.i'H nl' any \\n\\ik n!' l,lm cirelo, xprHHt!(l by IIKUUIH of a vaviablo para
inilrr O t uriv (/', j, 1 J .
70, (I.niHiilt!! 1 u cumiut, (Hiiiih jf } , Lhii HoordinntDH ol' which (vcotangular) are
... /. (, ! 1), uiiil (nrcuinr) aro f, 7,, ^ (^ L), Hum blui Ibrogoing
ili>!in.i. il. in I'li'Hi, tin; Hiuiu'i* >!' tho tuiiKonMul iliHtuncu of bho point P from the
pivrli. y":0.
7 1. lint, Ilicrn in uimtlirr ml;t>r[inilntioii iC UUH Hanus function !?[", vin,, writing
Ihiivin ^ I, iunl thru
r^o) j i*(i/) il i('*v.
v H.M iluu. VI in l.lio H.Uivil iliHliiiitH! 'f P fmiu oitlior of tho antipoiiiliB of bho
.iivlM (iinintrt yin K , il. will l. nusnllwtl.Hl. out f tho plnnu of tho circle), and wo have
tln.M Mi. HI...HVIII llmt ill" K'l 1 ""' 11 (tf lll(> ^>K"nt,ial iliHtinico of any point 1 from tho
itl t..i Hir Miiiurn .if its iH^tiuiCrt from uithnv ant.pomt of tho cirolo.
t .,v,i n
Arl.iolu Niw. 7!i to 77. On a Myntom of Sixteen Points.
i (.1, /A r? . .^) llli y ''" lir < 1 " )1( y ulio point". ll11 ^ lot tll untipointfl of
502 ON POLYZOJIAL CURVES. [414
As regards the circle R, since its centre lies in BG, the circle passes through
(Bi, C',); and since the centre lies in AD, the circle passes through (A lt D,), that is,
the four points (A lt B lt G' n A) lie in the circle R. Similarly (A at B it G y , D 2 ) lie in
the circle 8, and (A 3) B 3 , 6' 3 A) in the circle T,
74 The points R, 8, T are conjugate points in relation to the circle ; that is,
8T, TR, RS are the polars of R, S, T respectively in regard to this circle ; aucl they
are, consequently, at right angles to the lines OR, OS, OT respectively; viz., liho four
centres 0, R, 8, T are such that the line joining any two of them cuts at right
angles the line joining the other two of them, and we see that the relation between
the four sets is in fact a symmetrical one; this is most easily seen by coii.sifloration
of the circular points at infinity 7, /, the four seta of points may be arranged thus :
A t A 5 , A,, A,,
A, S, B lt Bi,
C, C : , Q , G s ,
A, A, A, D ,
in such wise that any four of them in the same vertical line pass through /, mid
any four in the same horizontal line pass through J\ and this being so, starting for
instance with (A t , B s , 3) A) we have antipoints
of (A ,<U (4 a , A) are (A, C,), (A, A),
(0., 4 t ), (A, A) u (G lt AJ, (B lt A),
(^ A), (0,, A) (A, B), (G, LI
and similarly if we start from (A,, B }t G lt A) or (A, t B tt <7 a , A).
75. I return for a moment to the construction of (A lt B lt G lt A); those arc
points on the circle R, and (B t , 0,) are the autipoints of (7?, G); that ia, they are
the intersections of the circle R by the line at right angles to BO from its middle
point or, what is the same thing, by the perpendicular on W from 0. Similarly
(A, A) are the antipoints of (A, 7)); that is, they are the intersections of tho
circle J2 by the perpendicular on AD from 0. And the like as to (4,, B. t 6' 3> JJ a )
and (^ a , fi sj C 3) D s ) respectively.
7G. Hence, starting with the points A, B, 0, D on the circle 0, and consbiucting
AB CI AC* and C nStrUCting also fche P^'Pendiculars from on U
the peipendiculars on BO, AD meet circle R in (A, >)> (A t > A),
GA > BD S >t (0tt Aa)> (B> A))
" ^' ^ % >, (A a B) (G D)
system is " 1 '
414] ON POLYZ03IAL CURVES. 503
77. II' to fix the ideas (A, B, 0, D) are real points taken in order on the real
circle 0, then the points II, 8, T are each of them real ; but R and T lie outside,
$ inside the circle 0. The circles R and T are consequently real, but the circle >$'
imaginary, viz,, its radius is =i into a real quantity; the imaginary points (A 1 , B t , G l} Dj)
arc thns given as the intersections of a real circle by a pair of real lines, and the
like as to the imaginary points (A S} B 3 , G 3 , _D 3 ) ; but the imaginary points (A,,, B. 2l G' a , A)
lire only given as the intersections of an imaginary circle (centre real and radius a
pure imaginary) by a pair of real lines. The points (6' 3) A z ) quA autipoints of (G, A)
nro easily constructed as the intersections of a real circle by a real line, and the like
an to tho points (7J 9 , A) quA antipoints of (B, D), but the construction for the two
pairs of points cannot be effected by means of the same real circle.
Article Nos. 78 to 80. Property in regard to Four Oonfocal Conies.
78. All tho conies which pass through the four concyclic points A, B, 0, D, have
thoir axes in fixed directions ; but three such conies are the linepairs (BO, AD),
(CA, JW), and (AB, O.D), whence the directions of the axes are those of the bisectors
of the tingles formed by any one of these pairs of lines; hence, in particular, con
sidering either axis of a conic through the four points, the lines AB and GD are
equally inclined on opposite sides to this axis, and this leads to the theorem that
tho antipoints (A a> B a )(0 B , D 3 ) are in a conic confocal to the given conic through
(A, W, 0, J.))' } whence, also, considering any given conic whatever through (A, B, G, D),
liho points (/L 1( S lt 0,, A), (A,, B,, O tl D a ), (A 3 > B a> O a , A) lie severally in three conies,
each of thorn confocal with the given conic.
504 ON POLYZOMA.L CURVES. [414
ellipse we have the two points (G, D), then drawing the diameter OG conjugate to
CD, and through its extremity G a confocal hyperbola, the antipoints (0,, A) will
lie on the hyperbola. Suppose (A t B, G, D] are concyclic, then, as noticed, AB and
CD will be equally inclined on opposite sides to the transverse axis of the ellipse
the conjugate diameters OF, OG will therefore be equally inclined on opposite sides
of the transverse axis and the points F and G will therefore be situate symmetrically
on opposite sides of the transverse axis, that is, the points F and G will respectively
determine the same confocal hyperbola, and we have thus the required theorem, viz., if
(A, B, 0, D) are any four concyclic points on an ellipse, or say on a conic, and if
(A a , B s ) are the antipoints of (A t Jl), and (G 3 , A) the antipoints of (G, D), then
(A 3 , B 3 , 3> A) will lie on a conic confocal with the given conic.
Article Nos. 81 to 85. System of the Sixteen Pointe, ike Acoial Case.
81. The theorems hold good when the four points A, B, G, D are in a line; the
antipoints (#,, 6'i) of (B, G), &c,, are in this case situate symmetrically on opposite
.sides of the line, so that it is evident at sight that we have (A lt B lt (7,, A).
(Ag, #2, OB, A), C^BJ 83, G s , A), each set in a circle; and that the centres R, S, T
of these circles lie in the line, The construction for the general case becomes, however,
indeterminate, and must therefore be varied. If in the general case we take any circle
through {B, G), and any circle through (A, D), then the circle R cuts at right angles
these two circles, aud has, consequently, its centre R in the radical axis of the two
circles ; whence, when the four points are in a line, taking any circle through (B, 0},
or in particular the circle on BO as diameter, and any circle through (A, D}, or
in particular the circle on AD as diameter, the radical axis of these two circles
intersects the line in the required centre M, and the circle R is the circle with this
centre cutting at right angles the two circles respectively; the circles 8 and 2' are, of
course, obtained by the like construction in regard to the combinations (0, A ; B, D)
and (A, B \ 0, D), respectively. It may be added, that we have
*) (extremities^) IB.O,A,D,
S j centre and I of diameter S\ sibiconjugate points of involutions 40, A ; B t D,
T) ^of circles y j \A t B' t Q t D t
and that (as in the general case) the circles R, S, T intersect each pair of them at
right angles ; and they are evidently each intersected at right angles by the line
ABGD (or axis of the figure), which replaces the circle in the general case.
82. If the points A, B, G, D are taken in order on the line, then the points
R, S, T are all real, viz,, the point R is situate, on one side or the other, outside
AD, but the points S and T are each of them situate between B and 0; the circles
M and T are real, but the circle 8 has its radius a pure imaginary quantity,
83. If one of the four points, suppose A is at infinity on the line, then the
antipoints of (A, D), of (B, D), and of (G, D) are each of them the two points (/, J).
4 14] ON POLYZOMAL CURVES. 505
It would at first sight appear that the only conditions for the circles R, 8, T were
the conditions of passing through the antipoints of (B t G), of (G, A), and of (A, B)
reypectively, and that these circles thus became indeterminate ; but in fact the definition
of the circles is then as follows, viz,, 21 has its centre at A, and passes through the
antipoints of (B, 0): (whence squared radius = AB . AG). And similarly, S has its
centre at 23, and passes through antipoints of (G', A) t (squared radius =BA.BG)\ and
T has its centre at 6', and passes through antipoints of (A t B), (squared radius
= CA , GB} ; these throe circles cut each other at right angles. As before, A, B, G
boing in order on the line, the circles R, T are real, but the circle S has its radius a
pure imaginary quantity.
84', That the circles are as just mentioned appears as follows: taking the line
an axis of a, and a, b t G, d for the to coordinates of the four points respectively, then
the coordinates of A^ t A arc
& (a + d), $ (a  d) ;
whence, m being arbitrary, the general equation of a circle through A lt A is
a 3 h y~ fyn&iz + [ f m (a H d) ad] z z = 0,
writing heroin
m ~ d
this hecomoH
vix,, for rf co it w ^
which is a circle having A for its centre, and its radius an arbitrary quantity L If
the circle passes through the antipoints of B, G, the coordinates of these are
and wo find
85. Reverting to the general case of four points A, 3, G, D on a line, the
theorem a* to the eonfocal conic* holda good under the form that, drawing any come
whatever through (A lt B lt 6\, A), the points (4,, J5,, 0, A), and (A^ B a G A) lie
in eonfocal conies, these conies have their centre on the line and axes in the dueotaon
of and perpendicular to the line. When D is at infinity the confocal comes become
any three concentric circles through (A, <& (ft, A), and (A SI Bj respectively.
506
ON POLYZOMAL CURVES.
[414
where BCD, &c., are the triangles formed by the points (#, (7, D), &c, ; the analytical
expressions are
a : b : c : cl =
so that
6, tf, 1
c, c', 1
d, d', 1
c, c', 1
d, d', 1
a, a', I
a, a', I
a 4b 40 4d =0,
^fl L n/i i. r*f .4. n/7 M
tlLV Ut/ t^ ^J\J \ Ulu "^ \f
aa' 4 b6' 4 cc' 4 dd' = ;
this being so, it is clear that we have
a, a', 1
c, c', 1
z* [a (a?
a constant.
 a"') 4 b (& 3
2 4
87. I am not aware that in the general case there is any convenient expression
for this constant If; it is =0 when the four circles have the same orthotomio circle;
an fact, talcing as origin the centre of the orthotomio circle, and its radius to be =1,
we have
a a 4' 3 a" a = l, &c.,
whence
that is, if the circles A, B, G, D have the same orthotomie circle, then 51, 33, (", 55,
a, b, c, d, signifying as above, we have
and, in particular, if the circles reduce themselves to the points A, B, 0, D respectively,
then (writing as usual 2T, S3, (, 3) in place of [", S3 , <', 2)) if the four points
A> B, 0, D are on a circle, we have
88, This last theorem may be regarded as a particular case of the theorem
a9r4bS3 +oS + d5> =K#= K,
viz., the four circles reducing themselves to the points A, B,' G, D, we can find for
the constant K an expression which will of course vanish when the points are on a
circle. For this purpose, let the lines BO, AD meet in R, the lines GA, BD in #,
and the lines AS, CD in T\ we may, to fix the ideas, consider ASOD as forming
a convex quadrilateral, R and T will then be the exterior centres, 8 the interior
centre; a, b, c, d, may be taken equal to BOD, GDA, DAB, ABO, where the areas
BOD, &c. ( are each taken positively. The expression a2t 4 b23 4 c( 4 d2> has the same
value, whatever ia the position of the point P (to, y, g = 1), ; taking this point at R }
and writing for a moment
414]
then.
ON POLYZOMAL CURVES.
507
BGD = (ROD  RBD) = RD (RG  RB) sin R = (7  /3) S sin R,
with similar expressions for the other triangles; and we thus have
 = s 2 sin 72 (/3y  aS) (7  /3) (S  a),
that is, replacing a, & 7, 8, by their values, and writing also z = 1, we have
a[ + b33 + cCS: h d2>  ^ sin jfi . (KB , EG  RA . RD) BG . AD,
whore ^smH.BG.AD is in fact the area of the quadrilateral ABGD; we have thus
D
whore it is to ho observed that &4, 50 being measured in opposite directions from
S, must bo considered, one as positive, the other as negative, and the like as regards
SB, SD. This expression for the value of the constant is due to Mr Crofton. In the
particular case where A, B, C> .D, are on a circle, we have as before
89. If the four points A t B, G, D t are on a circle, then, taking as origin the
centre of this circle and its radius as unity, the circular coordinates of the four points
will be
the corresponding forms of 3(, &o., being
91 = (f  o)
the expressions for a, b, c, d, observing that we have
& 0', 1
1
1, & /3 3
S'vS
7> 7~ l 1
1, 7. 7 s
8, 8', 1
1$ 5a
, 6, 6 J
if (/9y8>, &c. denote (/3  7) (7  8 ) ( s  0). &0 " become
which are convenient formulEQ for the case in question.
508 ON POLYZOMAL CURVES. [414
00. If the points A, B, G t JD, are on a line, then taking this line for the axis of
, we may write 91 = (a  azf + f  'V, &c. It is to be remarked hero that we can,
without any relation whatever between the radii of the circles, satisfy the equation
in fact this will be the case if we have
a + b tc + d =0,
aa +b& +cc + dd =0,
a (a a  " 3 ) f b (6 a  6" a ) + c (c 3  c" 3 ) + d (cZ a  fZ" 3 ) = 0,
equations whicli determine the ratios a : b : c : d. In the case where the circles reduce
themselves to the points A, B, 0, D, these equations become
a +b + c + d =0,
a +b& +cc + Ad =0,
aft" + b& 3 + cc a + d# = 0,
giving
a : b : c ; d = (6cd) : ~(cda) : (dab) : (o&o) ;
if for shortness (bed), &c. stand for (6  c) (c  d) (d  6), &o. : and for these values, we
have
=0.
91. A very noticeable case is when the four circles are such that the foregoing
values of (a, b, c, d) also satisfy the equation
a2l
the condition for this is obviously
aa"
or, as it may also be written,
,
V s '"
"
Article No. 92. On a loous oomeated with the foregoing Properties,
92 If as above, A, B, G, D are any four points, and 2[, <B, K, 3) are the
>
cur vu
f u foui ' P inte ^spootively, then the
the foci of the comes wbch pass through the four points is the tetra Z omal
posits r + p e+ M +<tD n haa ' H has been aem  a (> t t TOI  * a "
pos,Uon s of the pomt P; tatog P to be the other foe U8 , its squared distanoes are
(kAf, &o., whence for the firstmentioned focus we have
14'
ON POLYZOMAL CURVES.
509
or rocollooting that a + b I o + d  0, it follows that wo have for the locus in question
a V?f  h Vi!3  o V( I d V'33 = ; this IOOUB will bo discussed in tho sequel. I remark
horn, that in bho ciuw whoro tho four points are on a circle, then (as mentioned above),
tho nxiw of liho Hovoru.1 conies am in tho Hamo iixcd directions ; there are thus two sets
of 1'ooi, thoKo on tho axln in ono direction, and those on tho axis in tho other direction;
it might thoroforo bo anticipated, and it will appear, that in this case the totrammal
broaltH up into two fcrissomul mirvoH.
Article NOH. 9J1 to !)H. Formula* <w to iho two Sets (A t B, C, D), and (A lt B lt O lt A),
cttoh of four Oonoyclio Points.
Oil Umwidoi' tho four points A, tt t C, J) on a circle, then taking, as before, their
uimilur ooordiuatoH to bo (, a', I), (/3, /3', 1), (7, /, 1), ($, ^ 1). ^ condition that
tint jiointH may bo on a oirolo is
1, a, a', aa'
1 t t'
1, 6 , 6 ,
may bo writton
vix,, thin o
or if, for shortnoHH, wo tako
ami fi
I'
thon tlio onuation i
A'/,
i .(. I t o S3 Oj tt' 1 &' + c' ~ 0,
a/: (47 : ofe ft/' : 6'ff' : c'V.
f)4. Lot a, b, c, d, donoto as boforo (a : b ; c : d~B01) :
thon wo havo
a : b : o ; cl
7i 7'* ^
8, 8', 1
8, 5 , 1
a, a', 1
a, ', 1
 ,
512 ON POLYZOMAL OUEVES. [414
Article Nos. 9D to 104, further Properties in relation to the same Sets
(A, B, G, D) and (A lt S lt G lt A).
90. It is to be shown that in virtue of these equations, and if moreover  + , H  h^ = 0,
1 abed
then it is possible to find /,, m lt n )t pi, such that we have identically
 m + w33 + g  p<$ + 1&  WjS,  !<, + p,3), = 0.
This equation will in fact be identically true if only
ffl+yg'm + hh'n , gh'^g'hn^ = 0,
cc'm + bb'n ffp . 4 cb'in^ + bc^ = 0,
gtfmhVn + ffh i(/&S ho'ih =0,
cg'm  bk'n , + ch'm, + bg'^ +ff$i = 0.
From the first and second equations eliminating m^ or ,, the other of these quantities
disappears of itself, and we thus obtain two equations which must be equivalent to
a single one, viz,, we have
bc'ffl + o'g'afm + 6 h a'f'n + y'hffp = 0,
b'cffl + cga'f'm + b'Vafn + gh'ffp = ;
which equations may also be written
and it thus appears that the equations are equivalent to each other, and to the
assumed relation
i+ + + =0 .
abed
100. Similarly from the third and fourth equations eliminating m or f the other
of these quantities disappears of itself, and we find
cg'ff'l,  cga'fm, + a/G'g'n,  o'gffp, = 0,
bh'ffl,  o/6'A'm, + 6/m/X  Vhff pl 0,
equations which may be written
/'/' W it'

414] ON POLYZOMAL CURVES. 513
where we see that the two equations are equivalent fco each other and to the
equation
+ + !!! +! = .
ftj b, c, di
It thus appears that the quantities l lt iti^, n l} p lt must satisfy this last equation. It
is to be observed that the first and second equations being, as we have seen, equivalent
to n single equation, either of the quantities m lt n lt may be assumed at pleasure, but
the other is then determined; the third and fourth equations then give / 1( ^; arid the
quantities l lt 7^, n l} p lt so obtained, satisfy identically the equation + T i + + y 0.
fti D! G! Q!
101. Now writing
ff\   g (c'm + b'mi) + h (b'n + C'HJ),
ffl\ =  G (r/'m  h'm^ + b (h'n  g'n,} t
and
ff'P = c ( G ' m + t ''
wo find
+ frnj) (A'?t  </'!) + Q/'m  /t'm,) (b'n + o'n,)],
thai) is
.^' (^i^i ~ ^) = rt
via,, this equation is satisfied identically by the values of J,, m,, n,, p : determined as
above.
102. Hence if wii^amn, we have also Ifa^lp, and we can determine TOI, !. so
that wiiTh shall m, vin., in the first or second of the four equations (these two being
equivalent to each other, as already mentioned), writing m* = On, and therefore n t = g in,
wo have
cc'm + Wn ffp + oft'n^ + Mm  Q = 0,
which are, in fact, the same quadric equation in Q, viz., we have
_ _
Wnffp cV bo' '
The final result is that there are two sets of values of l lt m^, n lt p it each satisfying
the identity
i  w,j + PII 0,
65
c. VI.
514 ON POLYZONAL CURVES. [414
and for each of which we have
li , mi , HI , pi A , ,

= wwi.
103. Consider, in particular, the case where p = ; the relation
l + ^ + 5+^o
a+b+c+d U)
here becomes
, __ cu/' a'h
i ~~lf m ^f n 
The equation in Q is
. _ (co'm + Wn) & + eb'nP + bo'm = 0,
viz., this is
gvng
tf* m  cw
o ' Ml . o ' ^"T
or else
Since in the present case ^, = 0, wo have either ^ = 0, or else ^ = 0, and as might
be anticipated, the two values of 6 correspond to these two cases respectively, viz,,
proceeding to Bnd the values of l lt p l} the completed systems are
e \ ' *.=^K~H' 7 ' ! '=f> ^=.
s n r .o '.. <=^, P,.
so that for the first system we have
and for the second system
n\ p\
f 5 6 fol ' egoing investi g ation would have assumed a more eimplo
AMD v o o
cncle AMD as ongm, and the radiu, of this circle been put =1; W e should then
have <*=, &c. ( and consequently

f * he Cil ' leS AB01) mA *AA A * have been
414] ON POLYZOMAL CURVES. 515
I will however give tho investigation in this simplified form, for the identity
(51 + m33 1 w(S = l$\ 4 in$ + n$ ; viz., in this case we have
I _ in (/3 7) (/3 S) (/3 7) (y S)
  J ft"~"x ' S) ~ y (
mid tho identity to bo satisfied is
writing foe, q^i*, wo find m,, and writing = as, y^z, we find w,, and it is
then oasy to obtain tho value of l t , viz., tho results are
mi =
'
a
and therefore m,, = m?i; it may bo added that we have
8 a S \ 7 S/ '
/ tf?) 7f
vk, this is tho form assumed by tho equation :r + tr + ~ = Oi
Hj MI \>\
PA.UT III. (Nos. 105 to 157). ON THE THEORY OP Foci.
Article Nos. 105 to 110. Explanation of the General Theory,
105 If from a focus of a oonic we draw two tangents to the curve, these pass
respectively through tho two circular points at infinity, and we have thence the
generalised definition of a focus as established by Hiicker, _ TIB, m any or "
focus ia a point 8 uoh that tho lines joining it with the two circular points at
are respectively tangents to tho curve; or, what is the same thing, i
tho circular points at infinity, say from the points /, J tangents
curve, tho intersections of each tangent from the one point with e
the other point aro tho foci of the curve. A curve of the class n ntuj , 
, nem ' fcd It ia to be added that, as in the conic the line joining the points
nlact of the two tangents from a focus is the directrix ~_* M  to*
in question,
516 ON POLYZOMAL CURVES. [414
106, A circular point at infinity / or /, may bo an ordinary or a singular
point on the curve, and the tangent at this point then counts, or, in tho case of
a multiple point, the tangents at this point count a certain number of times, say
q times, among the tangents which can be drawn to the curve from the point; tho
number of the remaining tangents is thus = n  q. In particular, if the circular point
at infinity be an ordinary point, then the tangent counts twice, or we have g=2; if
it be a node, each of the tangents counts twice, or q = 4 ; if it bo a cusp, tho tangent
counts three times, or </ = 3. Similarly, if the other circular point an inanity bo an
ordinary or a singular point on the curve, the tangent or tangents there count a certain
number of times, say q' times, among the tangents to the curve from this point ;
the number of the remaining tangents is thus =n~q'. And if as usual we disregard
the tangents at the two points I, J respectively, and attend only to the remaining
tangents, the number of the foci is = (  q) (n  q'),
107. Among the tangents from the point / or J there may be a tangent which,
either from its being a multiple tangent (that is, a tangent having ordinary contact
at ^ two or more distinct points), or from being an osculating tangent at one or more
points, counts a certain number of times, say r, among the tangents from the point
in question. Similarly, if among the tangents from the other point J or /, there is
a tangent which counts / times, then the foci are made up as follows, viz. \vo have
Intersections of the two singular tangents counting as , r ' r
Intersections of the first singular tangent with each of
the ordinary tangents from the other circular point at
infinity, as ... / , ,.
y ......... (n~q r')r
Do. for second singular tangent, ... f _ _ *. 1 /
' ' V* '1 ' J *
Intersections of the ordinary tangents .... (nqr)(n q' ~r')
Giving together the ........ frff) ("flO
fn " CaSG where ** tang^t" from
each of the points I, J include more than one singular tangent.
or r case to b6 considered; the line infinity may be an
he I 1 f g "^ b * fche : Burning that it counts / tinJs among
the tangents from either of the circular points at infinity, the numbers of the
ON 1'OLYXOJIAH CURVES. 517
IK). (loimithsr any two foci A, 11 not in lined with either of the points /, J,
Minn joining tlmHo with tho points /, ,/", anil taking ./In ./^ the iutorflootions of AT, HJ
and tif ^li/, ///' (yl u 7J, being tlmroi'oro by u foregoing definition the antipolnts of (A, J3)),
then A i, //, are, it !H clour, foni of the curve, Wo may out of tho ;i a foci soloct, and
Mutt in I.2../) ilill'm'tmli wiiyn, a system of p foci such that no two of them Ho
in linwl with either of Mm point* J, ./'; and this being HO, taking tho antipoints of
rai'h nf thn i/i(;)1) pati'H nub of tho p foci, wo havo, inclusively of tho p foci, in
all /II2. 4/>(/> 1), that is y/ 4 foci, tho ontiro Hyntom of foci.
Arl.itiln Nun. I'll tin II 7. (;. tho Foot of Oonioa.
Ill, A onnii! in u (iiirvn of thn (iliwa a, and th'o inunbor of foci w thus =4.
Taking MM loci any two piiiiilw A, H, tho roinainiug two foci will bo tho antipoints
J,, //,. In nvdiu thud a givtsii point ^l may bn a 1'ocun, tho conic miwt touch tho
lini'H ;l /, /I./; Hiniiliirly, in urdi'i' that a given point Jt may bo a fooua, tho conio
iiniHi, l.uirh tlui linoH ,/{/, 7U; Mm cipiation of a conio having tho given points A, .11
Inr t'ici c'oiilninH tlii'i'cfuro a ningln avhitrury paranmtnr.
111!. In Mm CIIHO, liuwnvov, of liliu ]mmhola tho curvo touolioH tho lino infinity;
Mmm in ruitKi!iUiiiiL)y from nmh nf tlio points /', J only a wnglo tangont to tho
I'.iiiv.', anil i:itiiHuiiuml.ly only cum fuouH: Mm juiraholtt having a givon point A for its
liitii* in a iuiii.'. Iniirliing Mio lino infinity and the liiuw AI t AJ, or aay tho throe
nidiit nl 1 Mm iriaugln AU] il.H liquation wmtimw Mioroforo two arhitrary pammoteiu
i. Hutuniing to Mm gonoml ctinic, thmi aio c.ortain tnx.omal 1'orniH of tlio focal
, mil, ul' any grnvb intnvHt, but which may bo mentioned. Using oirailar
lrH, unil lukiiitf (, *', I) and (/3, /H 1 , 1) for tho coordinates of tho givon foci
.I, n vrhpi'p.tivt'ly, tlm conin bmiclum Mm lines f  M 0, ?/  fit's = 0, f^0,
i) "tl's> (), Mm cqnatimi of a oonio Umching tho first three lines in
Vi (" ) 1 Vw (f W) + ^ ('/"a 1 ^) 0,
vvlnru /!, /, ui'o arbitrary, and it is any to obtain, in order Mmt tho conic may
hitmh tin; fourth lint* j #'*(), Mm eondiMou
[U. In ihul, rt luiving thiH vttluu, tho otiimtion gives
i (^  M) 4 M (f  /?*) + 2 N/MF^^TF^) Iv (m ~
uml tiiking uvor tho lunn
 , " ft , (?  08'  ') *, (J8  ) (  *.
# a
this givtw __
I (?  #*) + m (f  ) 1 2 Vim <)!*"  /3*)   IT?
518 ON POLYZOMAL OUHVES. [414
which puts in evidence the tangent y 13'z. It is easy to sec that the equation may
be written in any one of the four forms
^) +
(m  (*  '*) = 0,
( m _
viz., in forms containing any three of the four radicals Vf^T^ Vf^, V?/V^
vnfie. The conic is thus expressed as a tmomal curve, the Homals being ouch a
hue, viz., they are any three out of the four focal tangents; the order of the cuivo,
as deduced from the general expression 2V, is = 2 ; so that there is hero no dopraaHion
of order.
J\ 5 '. But the ordinai 7 form of ^e focal equation is a more interesting one ; vix.,
,$ being as usual the squared distances of the current point from the two mvon
loci respectively, say
then 2ft being an arbitrary parameter, the equation is
the equation is here that of a tmomal curve, tho .omals being curves of the
'' e s ( ^ o the ]in
rttd'^v;' e rf s T ( ^ o) the ]in ^ ^ *
4 but in f', 7 PeCtlV t f y: the 8^ Pum 2r gives therefore tho orclor
4, but m the present case there are two branches, viz., tho branches
tline .
the Ime mmt tw and onuttmg this factor the order IB =2, as it should bo.
i ^ ^' ting tho
ared distances of the current point from A lt J3,
414] ON POLYZOMA.L CURVES. 519
whovo k in tilio squared distance of the foci A, S, =4a a e 3 suppose: whence putting
tt a (1. o 3 ) 6 a , tho equation becomes
that is
which is the required now form, It is hardly necessary to remark that the equation
2((S + V91 I V35 = 0, putting therein z~\, and expressing SI, 23 in rectangular coordinates
measured along tho axen, is the ordinary focal equation 2a = V(  aef + f + V(aj + aej a + if.
117. I remark that tho aquation 2(t2+V2l+VS = gives rise to 4a 3
but hero 91  93 =  'kecflw, HO that the equation contains 2 = 0, and omitting this it
buoomoH (tte  ow) + VSt = 0, a bmmial form, being a curve of the order =2, as it should
bo; thiH iw in fiiot tho ordinary equation in regard to a focus and its directrix.
Artiolo Now. US lio 12JJ. Theorem of the Variable Zomal as applied to a Conic.
i:iH. The equation 2ib + V?T ! + ^W 1 = is in like manner that of a conic; iu
Cacti, thin would Ins a curve of tho order = 4, but there are as before the two branches
2/^ + VVl"'Var0, 2/^V9r + VffiT = 0, each ideally containing (s = 0) the line infinity,
and tho onluv IH thus retluced to bo =2. Each of the circles 51 = 0, 58 = is a
circle having double contact with tho conic (this of course implies that the centre of
the eivcio SH on an oxiw of tho conic). We may if we please start from the form
2& + VW+ Vaj'0, and than by means of the theorem of the variable zomal introduce
into tho oquHtinii one, two, or throe such circles.
110. It IH in thin point of view that I will consider the question, viz., adapting
the formula to tho cimo of tho ollipao, and starting from the form
2s + V(;'^ae0) a "T]/ a + V(ai+w) fl + f = 0,
the ufiutttion of the variable zomal or circle of double contact may be taken to be
4aV (<a  aas) B + f_ (<& + M*+ = Q
2" + ' 1ff 1 + 3
whovo q IB an arbitrary parameter; writing for greater simplicity ,1, and reducing,
the equation IH
120. If q< 1, then writing q=s\n9, we obtain the ellipse
as tho envelope of the variable circle
<>N* IMH.Y/tiM.VI. IT lili I .
vix., of ri. mrdit huvimr th fcntn mi iln> iihtj.>! ;i\\ M ., ,!,. , ,...,. ,/ ,
, i ., ... , , ' ;! * ' ''"'" Ulr
t.tMHt.1'0, lincl ll,s I'lllllllH JT/HIM//. (1 linllrn. Ill li.tm;. ll^ n, ); . <,>,.: m .i.[i t 
vi'ry (ioiivuniiml KTHihinil nimilrurtinii nl itm *lli.i..i h mm j, i.. lll( ,;., ,j ,(,,,,
(t : : \ HIU"' 1 !!, lllr f.'mlc. Il
iiH in rlic riivln nl 1 riirvuiui.' nf ..H,' ..i ,.Hi. i .. (
w ; m^,,IVH,n (I , ,j h ttl , , lp ,.,
IK Liu.
n
if,, illt(( ,,
I.V I'tir (I  flu . Ih.. ii ( ,'., (  ,,. , iihM , .,, S4 ,
l!il. Ill thi' I'HH. f/ >, Wl  IntlV M trill ,,.>,...
lhii li^un^ vi/.
llllVl!
"I 1 , whnl. In Ilir HUH,, U,i nW)
A'
1N ll , ( ,
minor HXIH of t! M , ,,liu,. ,i V ' V * %iirw1 ' !tl n "' Iil
M'hiH i. h, fun d,.;,;.' * "H" ' naH lh "

f, (JUHVES.
""'I,;' "" IVHllir ^' '> <' other extremity of the minor axis; from
' ""' ' /! ' "'" """riUnm, of tho consecutive circle, arc real, and give
,.' tu *flO", tho circles are .till real, but
' !I 'I': '""" '/' J:I,i,,'
!, K(!ll( , ml , illK (liralo wo interchange /, , o , 6, the
 h?/ = HOG" ^,
M I i t,iu>f t ti. )1 , tn ,ho Ilinnur oquation
1 ' 'to HIII ^) <J + y/ J 6 u cos fl,
^lJiMrrhol y iUCUMH ttf Ulii (((lialilOU
*"'' , iiud Mun'oliiro Hiud^i tan <L kmO
I !.!( i? r
i.t^tuiuy !uiii' ; r.tiuiutiun in tlu; thoory of Elliptic Functions.
','j (,i ;Jn /*%;* f;/* tho ('ircular ('nine and the JBiairoiiiar Quartio.
!iH' i (In dlnws in in gwioral 6, and tho number of the
hilly u<iinrNtiii^ (rtwu m thtib of a circular cubic, viz., a cubic
jli finului 1 jininlH at indnity. Horu, at each of the circular
^nt u? lliiH point, icokoiw twioo among tho tangents bo the
M iHunltpr ul' tlui voinaining bangonts ifi thus =4, and the
Ii., 1* IVnm any two pointa whatovor on tho curve tangents
i!in ihts lwi poucils of tangents are, and that in four
ititttt^K ,*>. hMUiH^Mi., h. .tfli othur, viu., if tho tangents of the firat pencil are
(1. a, II. **/nnl U<^.. ,,f ttti> fMc.iiitl in<:il, taken in ft propoi order, are (!', 2', 3', 4'),
tlint H,* I.iur U. '.*, :*. ** h.iuMl^ouH with ouch of tho avrangoments (!', 2', 3', 4')>
(2 1 T 4' tti in 1 4 T , I i'(, il*. H", 2', 1'). And iii each ease the intersections of the
IMUT *''U'j,,HihiiK i*.iirul li^' ', a conic piuwuig through tho two given points
(;*1 i 1 '^ t* * Mt.
...'i in .:iu, littt u
jt<tiii}* tlit'Oi'tlg Mi>'h
f.mu nf intiMit>. (hr
run. tim ih' j^"^
ti>n,.l:< i t ,i !h jj,i
r< lh' v
ON POLYZOJIAL CURVES. [414
125. Hence taking the points on the curve to be the circular points at infinity,
we have the sixteen foci lying in fours upon four different circles that is, we have
four tetrads of concyclic foci. Let any one of these tetrads be A, B, 0, D, then if
Antipointe of (B, G)(A, D) are (S lt G } ), (A lt D t ),
(G,A)(S, D} (0,,A 2 ), (B,, A),
(A ) B)(G,J)) (A 3> S s \ (<7 a ,D 3 ),
the four tetrads of eoncyclic foci are
A, B, G t D
^i, A, G lt A;
A> &, Cf,, A;
^3, J a , 0,, A
'^r^ '^ ^ ^' *' ^
"i ""^ 'T""' and take
fl ' ^ th r the rigina
h t tf  , 6 TTf ^ SySt6mS
is m tact that already discussed ante, No. 72 et seq.
if, as
^ipoints of each pair, the four
A ' *> G > D > of ointe on
vi, 'the
The hss s 
at the
h
f
altamtion
f the
the
at
f the
quartic the 'pencils of tants h
sixteen foci are related tf. IT
circular ou ic T
brea,
in a binodal
res P ecbivel y e homologous, the
f ^ f ^ of
Article No , m to 129 .
17
tangents in
the case of
and in the case of a
to H * convenient to
Qircular
Quartic.
is c i eal . ,,, >
hey 1 T e *
asymptotes.
^ ^
r tan from ^
he ly *
to the
^ m P totea  Z ., in
as y m P totea f ^e om've,
intersection of the two imaginary
414] ON POLYZOMAL CURVES, 523
129. In the case of a bicircular quartie, the two tangents at I and the two
tangents at J meet in four points, which (although not recognising them as foci) I
call the nodofoci; those lio in pairs on two lines, diagonals of tho quadrilateral formed
by the four tangents (the third diagonal is of course the line IJ), which diagonals
I call the "nodal axes;" and the point of intersection of the two nodal axes is the
"centre" of the curve. The nodofoci are four points, two of them real, the other two
imaginary, via., they arc two pairs of antipoints, the lines through the two pairs
respectively being, of course, the nodal axes; these are consequently real lines bisecting
each other at right angles in the centre (with the relation 1 : * between the distances).
The centre may also bo defined as the intersection of the harmonic of IJ in regard
to the tangents at /, and the harmonic of this same line in regard to the tangents
at J. Speaking of tho tangents as asymptotes, the nodofoci are the angles of the
rhombus formed by the two pairs of parallel asymptotes; the nodal axes arc the
diagonals of this rhombus, and tho centre is the point of intersection of the two
diagonals; as such it is also the intersection of tho two lines drawn parallel to and
midway between tho lines forming each pair of parallel asymptotes,
Article No. 180, Circular Cubic and Bicircular Quartic; the Axial or Symmetrical
Qase.
130. In a circular cubic or bicircular quartic, tho pencil of tho tangents from
/ and that of tho tangents through /, considered as corresponding to each other in
some one of tho four arrangements, may bo such that tho line IJ considered as
belonging to bho two pencils respectively shall correspond to itself, and when this is
so, tho four foci, A, 1$, G, D, which arc tho intersections of tho corresponding tangents
in question, will lio in a lino (viz,, tho conio which exists in the general case will
break up into a linopair consisting of tho lino IJ and another line). The line in
question may bo called tho focal axis ; it will presently be shown that in the case of
the circular cubic it passes through the centre, and that in tho case of the bicircnlar
quartic it not only passes through the centre, but coincides with one or other of the
nodal axes, vise,, with that passing through the real or tho imaginary nodofoci; that
is, the curve may have on the focal axis two real or else two imaginary nod'
The focal axis contains, aa has been mentioned, four foni t.ho vnmainine twelv
524 ON POLYZOMAL CURVES, [4 1 4
the points 7, J is a cusp, There remain then for the circular cubic and for the
bicircular quartic the cases where there is a node or a cusp at a real point of the
curve; and for the bicircular quartic the case where each of the points I, J is a
cusp in general the curve has no other node or cusp, but it may besides have a
node or cusp at a real point thereof.
132. I consider first the case of the bicircular quartio where each, of the points
J, J" is a cusp. The curve is in this case of necessity symmetrical ( J ) it is in fact
a Cartesian ; viz., the Cartesian may be taken by definition 'to be a quartic curve
having a cusp at each of the circular points at infinity. But in this case, as dis
tinguished from the general case of the bicircular quartic, there is an essential
degeneration of all the focal properties, and it is necessary to explain what these
become. The centre is evidently the intersection of the cuspidal tangents; the iiodo
foci (so far as they can be said to exist) coalesce with the centre, and they do not
in so coalescing determine any definite directions for the nodal axes; that is, there
are no nodal axes, and the only theorem in regard to the focal axis or axis of
symmetry is, that it passes through the centre. Of the four tangents through the
point /, one has come to coincide with the line IJ\ and similarly, of the four
tangents through the point J one has come to coincide with the line J7; there
remain only three tangents through / and three tangents through J, and these by
their intersections determine nine foci viz,, three foci A, B, G on the axis, and besides
(J5j, C,) the antipoints of (5, Q)>. (0,, A a ) the antipoints of (G, A) and (A a> B 3 ) the
antipoints of (A, B),
133. The remaining seven foci have disappeared, viz., we may consider that one
of them has gone off to infinity on the focal axis, and that three pairs of foci have
come to coincide with the points I, / respectively. The circle (as in the general
case of a symmetrical quartic) has become a line, the focal axis; the circles R t S, T
(contrary to what might at first sight appear) continue to be determinate circles, via.,
these have their centres at A> B, G respectively, and pass through the points (B lt G\),
(0i, A z ) t and (A 3> B a ) respectively, see ante, No. 83. But on each of these circles wo
have not more than two proper foci, and it is only on the axis as representing tho
circle that we have three proper foci, the axial foci A, 8> Oi in regard hereto it
is to be remarked that the equation of Jfoe curve can be expressed not only by
means of these three foci injhe _form Vj2l + VmS + V^ = 0; but by means of any
two of them in the form Vm + Vm53 + K = 0, where K is a constant, or, what is the
same thing (2 being mtroducedjbr homogeneity in the expressions of [ and S
respectively), in the form Vj8[ + VmSJ + K& = 0.
134. Using for tbe moment the expression "twisted" as opposed to symmetrical
Jit Witt appear ^f Ho, 161164, that if storting with three given points as the foci of a bioiroular
ciuarho, we impose he condition that the nodes at I, J shall be each of them a cusp, then either the
quartic wrtl be the onole through the three points taken twice, in which case the assumed focal propeL
~" or else ""
414] ON POLYZOMAL CURVES. 525
(viz,, the curve is twisted when there is not any axis of symmetry, but the foci lie
only on circles) then the classification, ia
Circular Cnbics, twisted,
symmetrical,
Bicircular Quartics, twisted,
("Ordinary,
symmetrical, !
[Bicuspidal = Cartesian,
and each of these kinds may. bo general, nodal, or cuspidal viz., for the two last
mentioned kinds there may be a node or a cusp at a real point of the curve.
135. In tlio case of a node, say the point lY; first if the curve (circular cubic
or bicircular qnavtic) be twisted then of the four foci A, B, C, D we have two,
suppose B and G, coinciding with N ; and the sixteen foci are us follows, viz,
B > 0, A , D are JV, N, A, D;
8 lt O lt A,, A N, iY, Antipoiuts of (A, D);
G' a , A a , S 3 , A . Antipoints of (N, A), Antipoinfcs of (iY, D)\
A at A, Oa> A Do. do.
viz., we have the points (A, D) each once, the node N four times, the antipoints of
(A, D) once, and the midpoints of (JT, A) and of (N, D), each pair twice. But
properly there arc only four foci, viz., the points A, D and their antipoints. The
circle subsists as in the general case, and so does the circle R(BG, AD}, viz., this
has for centre the intersection of the line A'D by the tangent at N to the circle 0,
and it passes through the point JV, of course cutting the circle at right angles :
the circles 8 and T each reduce themselves each to the point N considered as an
evanescent circle, or what is the same thing to the linepair NI, NJ,
136. The case is nearly the same if the curve be symmetrical, but in the case
of the bicircular quorbio excluding the Cartesian: viz,, we have on the axis the foci
]3, G coinciding at JV, and the other two foci A, D\ the sixteen foci are as above
and the circle R is determined by the proper construction as applied to the case in
hand, viz., the centre 11 is the intersection of the axis by the radical axis of the
point N (considered as an evanescent circle) and the circle on AD as diameter; that
is Sfficx&A.RD. And the circles S and T reduce themselves each to the poi"" * r
considered as an evanescent circle.
137. Next if wo have a cusp, say the point K; first if
or bicircular quartie) be twisted then of the four foci A
A, B, G, coincide with K ; and the sixteen foci arc as follows,
S , , A , D are K t K t if, D,
B lt 6',, A lt A &> K> Antipoints of
O tl A,, B,, A Do. do,
A, t B 3t G a> D a Do. do.
, , ,
ut tin* Lw i,,,, t . r,.H,,rLivHy. Mu r.pmlMm will ,
ON POM/MMAI, ruuvr i, '. .
vi/., wit havti tlio puinl. It HIHV, MM pninf i\ nim i;m> . ;uM Hi .s^r^^inf < ,, A" /i
Mint<! l.imi'H. Dud pinperly l.hc puinl /) in flu MH\ trii'i Tin iur, o P , r ,t u.mM
uppcur, (HJ// (tiniln Minnigh A', /', 1ml pn'iilily IH< pmhrulu * n !.. i\l
tillHpiiltll illll^'llli limy Itit M tnIIPl lTH'i".rlililMV, M flu. . n,'!, " ,!' Hi
Mil' ririilcfi A 1 , iS', '/' rciliici tlu'iiini'lvtt riti'li h. \\\< p.in! /i" , .,i s >i'i. p J
poinl.
INN. The liliit is ihr ni'u if MM rum li. i.\imniii 1 ',,!. hK ; .,
hi^iiTiilai (jiiiulir t>xi > linlin/<; Mm t ';uh",mi t ; ilir . ,!. u >, h.i f! ( . ,.
llU.il. Mli> rllHJiiilill ttili^rlil..
l.'ll), Km the (!iirli*iiiiui, it' MI.IV .i it n,.,!, .V, M<m ..( Hj. Mit.
l.wn, Hiippittiu /f mill ft, niiin'iili' uiili A'; tin' ninr !' i MI, ,1 . . fl , .,
furl nrc A" nine linn'.'i; Imi in i (l 'i tli>j< \,> t ,, t pi,,,,i ,., M .<
I'M). A eiruultir eultin euiuiul, liuv. tw. h<I. <i m,i. ., ,i l,,,.^,
rmiln; itiid Miinilarly u lijrjrriilur ipmrfje inn,., 1  til v. )V,M t ,. !, , K
nl Mm piuntH /, ,/J iinlt'i^t ii ImaK up iutt fv.. ((*].!, s!i ?!).>?
Ins (UtiiMiileri'il in Mie inrrpifl in ri'leifin'e In lh.< M.I)]..I M >( in, ii^.n . f
'  
r..r.r.Tt r.;at^A B :^ in
Arllolu N, w , , b, IH ,I,,W, /( ,W 7W, / w rt, r fm ,,,, ,,,.,,
' ..... ' ' " l ,., T,,
ON POLY2OMAL CURVES.
527
is fche same thing,
&l (P$ + W + <&) + z" (at; + b*} + eg) = 0,
"3. (, v), z) being any coordinates whatever, this is fche general equation of a cubic
passing through the points (f0, 2 = 0), (, = 0, = 0), and at these points touched by
the hues f0, *? = respectively. And if ( T,, 2=1) be circular coordinates, then we
have tho general equation of a circular cubic having the lines f = 0, t}=Q for its
asymptotes, or say tho point f=0, <? = for its centre; the equation of the remaining
asymptote is evidently p% + qy + ez = ; to make the curve real we must have (p, q)
and (, &) conjugate imaginaries, e and c real.
"143. Taking in any case the points /, J to be the points = 0, z = and ^ = 0,
2 = respectively, for the equation of a tangent from I write p%=Qss\ then we have
dtj (Oz + qri + ez) \ z
that is
2 a (ad + op) h ^* (0
and the line will be a tangent if only
(6 3 + e0 + ftp)*  4jtf (a^ + qp) = 0,
that is, the four tangents from / are the lines p%=Qz, where is any root of this
equation ; similarly tho four tangents from J are the lines gy = fa, where < is any
root of the equation
cq) = 0.
+ bpy + cpa) = 0,
6p) + rf . ^ = 0,
Writing the two equations under the forms
0,
e 9 + 2aj 
Seay 
the equations have the same invariants; viz., for the first equation the invariants am
easily found to bo
/= 3(e 3 4&j)4ag) a + 72(ce~2a&)p2,
J = _ ( e s _ 4Jjj 4 ft!? )3  36 (ce  2al)pq (e a  %  4a(?)  2 1 ft " a " 9 
and then by symmetry tho other equation has the same irm
invariant! I 8 * / 9 has therefore the same value in the two ei
equations are linearly transformable the one into the other, which is the before
mentioned theorem that the two pencils are homographic.
144. The two equations will be satisfied by # = </), if only ljp aq\ that is, if
p j , q j'j putting for convenience j in place of e, the equation of the curve is then
0.
528
ON POLYZOMAL CURVES,
[414
In this case the pencils of tangents are a% = kdz t li) = k&3, where 6 is determined by
a quartic equation, or taking the corresponding lines (which by their intersections
determine the foci A, S t G, D) to be (a% = k6tf, bi) = k9tf) t &c., these four points Ho in
the line albi) = Q, which is a line through the centre of the curve, or point 0,
T) = Q: the formula) just obtained belong therefore to the symmetrical case of the
circular cubic. Passing to rectangular coordinates, writing #=1, and taking y~0 for
the equation of the axis, it is easy to see that the equation may be written
or, changing the origin and constants,
tcy + (to  a) (a  6) (a  c) ~ 0.
Article Nos. 145 to 149. Analytical Theory fm the ^circular Quariic.
145. The equation for the bicircular quartic may be taken to be
k (f  V) (^  /9V)
k <& T,, ) being any coordinates whatever, this is the equation of a
r?, f
f I
h in f; T
th S am; li r g o th ' an
ot the same me 2 = m regard to the tangents at ( = (), *o) If
we o h r the
= for one pair, and the lines fl,
v' rr tho point *
we , h.ffl I' " dal . aXe& In OTder that tha
and 70 ,m . *>. aol n f to MffMri". *, 4 o real. The points (f
itdVlf^r The "i: 16 /;* J ' ^ " "7' the
of a^biousida, o ^ be meS & ^ ^ fol * h,cle 8
ourvo
of
of
n
O fni
real
0, 0)
^B a
tho case
we
thati
^uation of the curve (.*
.
and the condition of tangenoy is
ON 1'OLYtfOMAL CURVES,
529
via., the tangontH from 7 arc *, w heio is any root of this equation. Similarly,
if wo have
tho ttuiguntH from ,/ iiro ^c/,/30, wlioro 4, in any root of this equation.
147. Tho tiwo (!i[un(ii<niH may bo written
which () ([ iuifci(.im huvn lih (! HIUMU iiivarianfcHj iu iholi for tbo first equation tho invariants
iu'o fcmml tu l)o IIH Co I lows, vix., if for HlnutnoHH (7= ~tikW{3* 4Ao + fl a , thon
J ((t"a a + 6 8 /3 H )  21 OJt^fiS
and Uum by Hyimiudry bliu othor (!(ati(iu HUH tliu Hanio invtirianUi. Tho ftbaolufco
invni'iuiit 7 il ^,/ a IUIH thiiH thi) sinno vuluo in tho two oqiuitimiH, that i, tho oquafcions
iu0 linoarly traiiHfunnahln (Jin one into tlui oliliov, which in tho beforementioned
thoorwn that tlio iiMic.ilw urn
14S. Tim (iqiiatiiniH wilt ho Hutinfiud hy 00 if only a = ft/9, thnt i, if (t t &m& m;
or ^hy 0*z~~<l> if only rtaft^, that IH, if , ft^wi/9, ,; tho oquation of tho curve
is in tlumu two WIWJH
7/) ( o^' 1 0,
^ (f ~ a'^' J ) (T;' J ~ ^*s") + 85 1J ^ I ?HS" (/9f  mj) ( ca j  0.
1C to fix tho idww wo ntluiid In tho (imt OIIHU, thuii tho otiuation in ia
wo may take OH corresponding tangents through tho two noclew respectively %~
Qftz\ tho foci ^1, 7A 0, D t which aro tho intciwjotions of tho pairs of lines (k ss
0. VI. 67
530 ON POLYZOMAL CURVES. "414
y=d,(3z) l &c,, lie, it is cleav, in the line ftict'tj Q t which is 0110 of the nodal axes
of the curve. Similarly, in the second case, if & be determined by the foregoing
equation, we may take as corresponding tangents through the two nodes respectively
%=Q'i.z, i)~ Q$z\ the foci (A, B, G, D), which are the intersections of the pairs of
linos (^Bjftz, t) = d$z\ &c., lie in the line /3 + ?; = 0, which is the other of iho
nodal axes .of the curve. In either case the foci A, B, 0, D lie in a line, that is,
we have the curve symmetrical; and, as we have just seen, the focal axis, or axis of
symmetry, is one .or other of the nodal axes.
149. In the case of the Cartesian, or when a = 0, /3 = 0, via., the equation am = b{3
is satisfied identically, and this seems to show that the Cartesian is symmetrical ; ib
is to be observed, however, that for = 0, /3 = the foregoing formulas fail, and it i.s
proper to repeat the investigation for the special case in question. Writing a. = 0, /3 = 0,
the equation of the curve is
AV + ezfy + 8 s (of + by) + cs 1 = 0,
and then, taking f0fo for the equation of the tangent from /, we have
+ i)x.b (e6
+ 3
and the condition of tangency is
viz., we have here a cubic equation, Similarly, if we have i } = 6az for the
of a tangent from J, then
60 + c)  (4 + 1)" = 0.
Hence 6 being determined by the cubic equation as above, we may take A = 6, and
consequently the equations of the corresponding tangents will be g=*$bg y = 0az viz
the foci .A, B G will be given as the intersections of the pairs of lines (f^fil'
VtoM\ &c. The fod he therefore in the line of6,0; or the curve is symmetrical
the local axis, or axis of symmetry, passing through the centre.
Article Nos. loO to 158. On the Property that the Points of Contact of the Tangents
/rom a Pair of Qoncydic Foci lie m Gircle,
(A GA , four cc ses .
Ci. A, a.. A), (A B 2) C t , A), (A tl B a> C s> A), that is, A, B, G, D are in a e rclo
rzt
414] ON TOLYZOMAL CURVES. 531
151. Consider tho caso of tho bicirculnv quartic, and take as before (f ~ 0, = 0),
and (^ = 0, s=s()) lor tho coord i nates of tlic points J, J" respectively. Lot the two
tangents [Voin tho focus A bo oar = (), ?; '# (), say for shortnosa # 0, p' = 0,
thon tho equation of tho cuvvo is expressible in the form pp'V= V s ( l ), whore tT= 0,
V0 aro ouoh of tliom a oirolo, viz., V and V arc each of them a qtiadric function
containing tho torniH &, etj, z% t and ?j. Taking an in dc tormina to coefficient X, the
oqnation may bo writ ton
pp' ( 7 + 2X K + Tfypt ) = ( V + Xp/) 8 ,
and thon A, may bo HO determined that U"h SXVlX^y  0, shnl! bo a QcIrcle, or
pair uC IhuiH tlirougb / and ,/'. Ii w titwy to HUO that wu have thus for X a cubic
equation, that is, bhuro aro throo vahiOH of X, for each of which the function
f/H^xK + X^/ aHHUinoH tho form (%  &z) (y  p'e), = ttf Buppoao : taking any one of
thiiHO, and dianging tho value of V HO aa bhat wo may havo Fin place of V+\pp' t
tho (iquation in pp'qq' + V' J , whero 7=0 in an boforo a ctrclo, the equation whowB that
tho point* of (sontiict of tho tangents p = 0, ^>' = 0, f/ = 0, </ = () Ho in HUH circle V^= 0,
This (.urcnmHtanco (ihut X fa dotoniiiinid by a cubic oipiutioH would auggoHt that the
fwiUH v = 0, (/'^O IH 0110 of tho throo foci JB, C7, D ooucyclic with A\ but tins is
tlio vory tiling whioh wo winh to prove, and tho investigation, though HomowLmt long,
IH an intuvoHting oni!.
1fi2. Starliing from tho form pp'qy' V'\ then introducing aa boforo an arbitrary
onoftlciout \, thu oijuabion may IMS writton
and wn may dotonninn X HO that yr/ 1 2XK1X'V nlmU be a pair of linos.
Writing V 77 f 7;  ,V  7,'f* 1 Jlfa 8 , and Hubatitntiug for ;y/ and ryr/ their values
( j) (ij  's) and (?/3)(i;^), tho t;([uabion in quoHtion iu
(I  2X// 1 I V) ft;  (jS I 2X/, I X Q ) 7;^  (y3' + 2X// I Xtf ) f * h <^/9' 4 2XAT + XW) = 0,
and tho required condition in
(1 I 2X/r  X B ) 03)9' I 27Jlf + V/) (/9 + 2X7, + X 9 ) (^ + 2X77
or ntduoiug, this IH
+ X ((  0) (a'  /S') 4 4/f Jf
'  27/  2/V)  0,
vix., X is dotorminorl by a quadrio equation. Calling its roots X n and X,, tho for
o<iiation, Hulwtitubing thoroin fluccoHsivoly those values, bocomos (f  7*) (o?  /*)
8>) respectively, Bay H' = and **' = 0.
i ThU InvoiiUgatlon is similar to Hint in Salmon's Wglw Plane Qurwt, p. 100, in ngntd to tli
tixngontu o( n qimvllo ourvo.
G
532
ON POLYZONAL CURVES.
[414
153. WG have to show that the four foci (p = 0, #' = 0), (<? = 0, q' = Q), (r=0,
/ = (}), (s = 0, s' = 0) are a set of concyclic foci; that is, that the lines p~Q t <? = J
r=0, 5 = correspond homographically to the lines p'~Q t q' Q, r' 0, s' ~ ; or, what
is the same thing, that we have
1, a, a', act' =0,
1, A /3'>./3/3'
1. 7. y 1 . 77'
1, S, S', SS'
or, as it will be convenient to write this equation,
a$ <y^ _ aS j9
cc'^'yS'^a'S' jS'
154. We have
V '
8 =
The expressions of a&, &c., are severally fractions, the denominators of which disappear
from the equation ; the numerators are
for  8 , = a (1 i 2\,# + V)  03 + 2\ a
) _
for /9  %
for 7  S,
.and it hence easily appears that the equation to be verified is
, + a V) (1 +
G8 + 2^ + a\*) (1 +
+ X a 3 )
+ V),
fi
155. This is
if for shortness
B=_ a
the equation then is
, ' = 2 (a' '
414] ON POLYZOMAL CURVES. 533
156. Calculating AffA'B, GA'C'A, OD'C'D, BD'B'D, those are at once
soon to divide by {(a0'a'0) #!(</  /3')~ '(a0) ; we have, moreover,
 {('  00') 7^  L (a'  0') '( 0)} K  ') # + L ('  0') '(
viz., this also contains the same factor ; and omitting it, the equation is found to be
{(a  0) ('  0')  4 (01?  1) (&B  ') }
2 {(*' 00') # i(
+ { (  ft) (a 1 00 +
vix., substituting for \i + X a and \\ their values, this is
{(  0) (a' 004 ((3H  L)
 {(art 1  00') /f  L (a'  0')) {( 
+ {( 0) (a'  0') 4 4 (ff  i) (a'Jf  L 1 )} [M + Hffl  Lj3'  Z'0} == 0,
which should bo identically true. Multiplying by H> and writing in the form
+ {_ (a  0) (tf  0') + 4 (off  i) (a'J5f  L') } (m  LU + (0/f  L) ($B  L')) = 0,
wo ab onco see that this is so, and the theorem is thus proved, viz., that the equation
being pp'q(f= V, the foci (pO, p'O) and ( ? = 0, ^ = 0) are conoyclio.
157. By what precedes, \ being a root of the foregoing quadnc equation, we may
Wlt
tho focus r = 0, r' = 0_ is concyclic with the other two foci; but from the
.equation of tho curve V^Jffqq', that is we have
qq'
or, what is the same thing,
viz., this is a form of the equation of the ourve; substituting for p, p 1 , q, ?', r, r'
their values, writing also
'
*nd changing the constants X, K (*. X ^ : ZM : ^ : ^ the equation is
ws 4 Vwi = o,
534 ON POLYZOAIAL CURVES, [414
viz., we have the theorem that for a bioiroular quartio if (f  az  0, i)  a'z ~ 0),
(f fiz = Q, yft'z^Q, ( 72=0), t] f/z Q) be any three concyclic foci, then the
equation is as just mentioned; that is, the curve is a trizouml curve, the zomals
being the three given foci regarded as 0circles. The same theorem holds' in regard
to the circular cubic, and a similar demonstration would apply to this case,
158. It may be noticed that we might, without proving as above that the .two
foci (p = 0, p' = Q) t (y0, </=Q) were coucyclic, have passed at once from the form
Pp'<tf=V*, to the form \V^ + t/tg+K Vr/ = (or Vm Wm33 == vV( == 0), and then
by the application of the theorem of the variable zomal (thereby establishing the
existence of a fourth focus concyclic with the three) have shown that tho original
two foci were concyclic. But it seemed the more orderly course to effect the demon
stration without the aid furnished by the reduction of the equation to the trizomal
form.
PART IV. (Nos. 159 to 206). ON TRIZOMAL AND TETRAZOMAL CURVES WHERE THE ZOMALS
ARE ClKOLES.
Article Nos. 159 to 165. The Trizomal CurveThe Tangents at I, J t So.
159. I consider the trizomal
where A, 2, G being the centres of three given circles, S[, &c. denote as before, viz.,
in rectangular and in circular coordinates respectively, we have
51 = (  azj + (y  dzf  a'V, = (f  a z) (y  dz )  a' V,
S3 = (  Mf + (y  v s y*  y v, =   _ yv
By what precedes, the curve is of the order =4, touching each of the given circles
twice, and having a double point, 01 node, at each of the points i, /; that is, it is
a bicircular quartic: but if for any determinate values of the radicals V Vm, >Jn,
Wif
ii ii . , , " + + =0,
then there is a branch
containing (, = 0) the line infinity; and the order is here =S; ., the curve here
t,r is an"" 7, f t ?. P t ta 7l J mCl """^ an0ther P int * infini; (lala
theie is an asymptote), and is thus a circular cubic. ' *
points 6 / 1 T Z m t M 1 ^ inVeSt ; gaU "? the e " atioM f *1 odal tangent, at the
poms 7 , / respeotwely; using for this, purpose the circular coordinates (fe ,, ,i)
0 O^eTaf, TV" "f ed l'n * ^ding the tangents a
(f0, ,0} we have only to attend to the terms of the second order in (f ,) and
414] ON POLYXOMAL OUBVKS. 535
mmilavly lor finding tho tangents at (?> ~ 0, z Q) wo have only to attend to the terms
of t)ho Hooowl order in (?;, z), But it in easy to see that any term involving a", I",
or o" will bn ol' l;lm third ordor at loa.st in (, 2), and mmilarly of tho third ordor at
leant in (?/, z}\ honeo for finding tho tangents wo may reject tlio torniH in question,
or, what in tho Hiunn thing, wo may write <!,", h", G" each = 0, thus reducing the three
oir<;loH In thuir mnpootlvo centres. Tho equation thnn becomes
z) (y  a's) I ^n (f ~ Js) (7, ~^9'0) ! V^T(f^)Oy*)  0. ,
Fur linding tho tuiigtnitH at (^O, e  0) wo liave in tho rationalised equation to
ubkoiitl only to thn tornm of Lhu wmoml order in (^ f 0); and it i easy to BOO that
any twin involving ', #', 7' will Iw of tho third order at least in (, 0), that is,
wn may rcduoo a', $', 7' (ih to x.ovo; tho irrational equation thon booomcs divisible
by V?;, and throwing out thin factor, it in
V; (f  ff )  Vw (f >) + VB "(f  7*) = 0,
VIK., thiH (iuation whicli ovidently lielongK to a iwiir of linon passing through tho point
(gcsO, ()) givotn tho tungontH at the point in question; and similarly tho tangents
ut tho point (j) .= 0, s <=. 0) aro given by tho tHpiatiou
^(ij ^)l ^1(7; ^) I Vw'ft " V*) O
101. To complotu Mm nolutinn, aUending to tho tangents at ( = (), ^ = 0), and
putting for HhnvtiiuwM
\ sa I in  n,
/t K3  i "1 'HI ~ it,
Jt S3 J '/?t 'I' 
A =~: /' J h wt u I ;J
tlu^ rationaliHod nqiiutiiHi in oiwily found lio be
 ;
iiurt it IH to bo notinod tlmt in tlio uuwi of tho circular cubic or when
thon A0, w> that Iho (!qunliiini contaiiiH tho factor 3, and throwing this out, tho
munition givo* a Hmtflu lino, which in in fact tho tangent of tho circular cubic.
162. Hutuming to tho bicircular quartie, wo may Book for tho condition in
that tho mulo may bo n oiwp: the required condition w obviously
h wvy) a 0,
or obHoi'ving that
ft A 
A + ^i; a  2iX, &c.
tbia is
536 ON POLYZOJIAL CURVES.
or substituting for X, /*, v, their values, it is
or, as it is more simply written,
I m n n
j_ _L n
t T~ \J,
[414
163. If the node at (?j = 0, 2 = 0) be also a cusp, then we have in like manner
I I I i
Now observing that
a, a', 1
A ', 1
7. 7'. 1
= 03 7) (7 ')(/?' 7) (7 4
= fi suppose : the two equations give
I : m, : = fl (/9  7) (/3'  7') ; fl (7  a) (7'  a') : il(a
or if fi is not = 0, then

164. If
or, what is the same thing, if
A #
7. y.
, =0,
6, &', 1
c, c', 1
= 0,
the centres A, B t G are in a line; taking it as the axis of ta t we have a = '==,
/3 = ' = &, 7 = 7 ' = c; an d the conditions for the cusps at /, / respectively reduce
themselves to the single condition
1
so that this condition being satisfied, the curve
is a Cartesian; viz, given any three circles with their centres on a line, there are
a singly infinite genes of Cartesians, each touched by the three circles respectively;
414]
ON POLYZOMAL CURVES.
537
the line of centres is the axis of the curve, but the centres A, B t G are not the foci,
except in the case a" = 0, &" = 0,o" 0, where the circles vanish. The condition for
I, m, 11 is satisfied if I : m : n~(b~c) 5 : (c ~ ft) 2 : (a 6) a j these values, writing
VZ" : Vm. : */n ~ b ~ c : c : a b, give not only VT+ Vm + Viz = 0, but also
a yT+ 6 Vm + c v'n = ; these are the conditions ' for a branch containing: (z* = 0) the
line infinity twice ; the equation
(b  G
(c 
f (tt  6)
is thus that of a conic, and if a" = 0, &" = 0, c"~0, then the curve reduces itself to
2/ 9 =0, the axis twice.
166. If O is not =0, then we have
I : m : n = (/3  7 ) (/3'  7') : (/) (7 ') = (  /3) (a'  #),
viz., Z, ??i, ?i ai'e as the squared distances 7^C' a , ffjl 8 , AB* t sa}' as / 2 : g* : k s ; or when
the centres of the given circles A, li, G are not in a Hue, then f, g, h being the
distances BO, CA, AH of these centres from each other, wo have, touching each of
the given circles twice, the single Cartesian
which, in the particular case whore the radii a,", V, c" are each = 0, becomes
viz,, this is the circle through the pointy A, B, G, say the circle ABO, twice,
Article Nos, 1G6 to 169. Investigation of the Foci of a Gonic represented by
cm Equation in Areal Coordinates.
166. I premise as follows: Let A, J3 } be any given points, and in regard to
the Jiriangle ABC let the areal coordinates of a current point P be , v, iv\ that is,
writing PEG, &c,, for the areas of these triangles, take the coordinates to be
uiv. w = PBQ ; PGA : PAS,
or, what is the same thing in the rectangular coordinates (a 1 , y, 2=sl) ( if
(a, a', 1), (6, &', 1), (G, o' ( Ij,
be the coordinates of A, S, respectively, take
. y> z
:
j y . z
fa,
I, b', 1
G, C' , 1
a,
c, c', 1
, a, 1
b,
C. VI.
538
ON POLYZOMAL CURVES.
or iu the circular coordinates (, y, 2 = 1), if (, a ', i), (fr p> t 1), ( % y, ].) bo tho
coordinates of the three points respectively, then
: ^ MO =
& ?}> z
7) 2
Si '/i *
, fl? >
ft ', 1
7. 7. 1
a, a', 1
7. 7. 1
a, ', 1
ft /9'. 1
167. For the point / we have ( ,, ^) = (0, 1, 0), and honco if its aroal
coordinates be (, v a , w 6 ), we have
U :
and hence also, (, y, w ) referring to the current point P, we find
Vrfw  w a y = ( 7  a ) [(a'  /3') (f  u)  ( 
whence
and in precisely the same manner, if <, <, w ; refer to the
point
168. Consider the conic
i 11 ta ^XV^T
which ar ( X ' w j
as is well knoivn
(4, B, a, j, G,
consequently for the conic
w, v t w) a = 0,
;: hate : er; and
then f01 '
ten 8 flnta
poiat the ooordimtten of
this P^* *o tho eonic i a
and that of the pair O f tangents from / i s
W. i, 0. J, ff,
these two linepai,, in t eraeotin& rf
= 0)
rf
414] ON POLYZOMAL CURVES. 539
169. In particular, if the conic is a conic passing through the points A, B, G t
then taking its equation to be
the inverse coefficients are as (I 3 ,, m 3 , n*,  2mn, toil,  2Zni), and wo have for the
equations of the two linepairs
" 0,
= 0.
Article No. 1.70. The Theorem of the Variable Zomal
170. Consider the four circles
5[" = 0, 33 = 0, <T = 0, ) = (91  (m  fl*) + (y  aty*  a'V, &c.),
which have a common orthotomio circle ; so that as before
a2! + b5i + cS + d3) = 0,
where
a : b ; c : d = CW :  OD.d : DylU : yIJ?C'.
I consider the first three circles as given, and the fourth circle as a variable circle
cutting at right angles the orthotomic circle of the three given circles; this being
so, attending only to the ratios a : b ; c, wo may write
a : b : c =DBG : DGA : DAB,
that is, (a, b, c) are proportional to the nrool coordinates of the centre of the variable
circle in regard to the triangle AHG,
171. Suppose that the centre of the variable circle is situate on a given conic,
then expressing the equation of this conic in areiil coordinates in regard to the
triangle ABO, we have between (a, b, c) the equation obtained by substituting these
values for the coordinates in the equation of the conic; that is, the equation of the
variable circle is
aSr + b33 + ctr ==0,
540 ON POLYZOMAL CURVES.
then the equation of the envelope is
(l\ w a , ', mn, nl t Imfy 31, S3 , (S) 3 = 0;
that is, it is
or, what is the same thing, it is
[414
172. It has been seen that the equations of the nodal tangents at fcho points
/, J respectively are respectively
0,
a*) + Vm ()&) + V*T(f
cte) + Vm (7;  0'*) + V (^
and that these are the equations of the tangents to the conic Ivw + mwii + nnv =
from the points /, / respectively. We have thus Casey's theorem for the generation
of the bicncular quartic as follows :The envelope of a variable circle which cutfl at
right angles the orthotomic circle of three given circles 91 = 0, 33 = 0, S = 0, and 1ms
its centre on the conic tm + mwu + miv = which passes through the centres of the
three given circles is the biciraular quartic, or trizomal
m8 + VjiS = 0,
which has its nodofoci coincident with the foci of the conic.
178. To complete the analytical theory, it is proper to express the equation of
the orthotmmc circle by means of the areal coordinates ( U> v, w ). Writing for shortness
a a a * =
a a,* = a\ & Cl) and therefore
then if as before
&o.,
> y, z
:
at, y, z
:
x, y, z
6, 6', 1
G, C', 1
a, a\ 1
G, C', 1
ft, of, 1
6, 6', 1
and therefore
: y *
the equation of the orthotomic circle is
ffla*, ya! 2) aa + afytf*
fflte, yVe t Iw + VyVe
acz, y G 'z> ca +c'y~ oV
viz., throwing out the factor *, this is
0,
414] ON FOLYZOHAL CURVES. 541
or, what is the same thing, it is
(au + bv + ovf) 10 + (a'u + b'v + c'w) y  (a\i + $v + c\v) z = 0,
viz., it is
(aw + bv + ow) 9 l (a'tt + b'v + c'w) a  (a v zt + b"v + c'w) ( + y + w)  0,
that is, substituting for .\ 6\ C N their values, it is
tt'V + 6" 2 y a 4 c" a (o a
4 (6" 3 + c" 2  (6  c) a  (&'  c') a ) wo
+ (c" a + a" 2  (c  ) B  (c'  a') )wtt
+ (a" 2 4 i" 2  (a  6) 2  ('  t') a ) uv = 0,
aiirl it may be observed that using for a moment a, & 7 to denote the angles at
which the three circles taken in pairs respectively intersect, then we have 2&"c" cos a
= &" a + c" fl (6c) a (&'c') 3 f &c,, and the equation of the orthotomio circle thus is
(1, 1, 1, cos a, cos/3, cosyJa'X V'v, c"w) s = 0.
174. We have in the foregoing enunciation of the theorem made use of the
three given circles A, B, C, but it is clear that these are in fact any three circles
in the series of the variable circle, and that the theorem may be otherwise stated
thus :
The envelope of a variable circle which has its centre in a given conic, and cuts
at right angles a given circle, is a bicircular quartic, such that its nodofoci are the
foci of the conic.
Article Nos. 175 to 177. Properties depending on the relation between the Oomo and Circle.
175 I refer to the conic of the theorem simply as the conic, and to the fixed
circle simply as the circle, or when any ambiguity might otherwise arise , then as the
flrthotomio circle This being so, I consider the effect in regard to the tnzomal curve,
of the various special relations which may exist between the circle and the come.
If the conic touch the circle, the curve has a node at the point of contact.
If the conic has with the circle a contact of the second order, the curve has a
cusp at the point of contact.
If the centre of the eirele lio ou an a=ds of the eonio, then the four intersections
He in pat syletrioally in regard to this axis, or the curve has tins ax, S a, an
axis of symmetry.
eirc.es intersecting in the, t point,
542
ON POLYZOMAL CURVES.
the circular points infi ' 7 ! " , thfl general case there are at each of
of the one pan ^ [ h "2 T "' "' ^P '* of the *
sections, the four foci of the lie H ^ ""' there are thu3 fo r *
of circle., the two tangent T U 1^' T *" "" COT JS
the two foci on tll f axis " " ^ "T^ * . h
are each of them the intersect O f ,' 7 , 7 f C '' r a " t; P oi nte of these,
the other circle. mtosM " of a tangent of the one circle by a tangent of
f th
i. a circn.ar cubic having the
of the conic), then the c at ^"1 ""T""^ "* "* ""^'y f a " axi
circles touching eaoh other Id TL^ "c He", ^' ' ?' bre!>lB " **>
centres at the two fo ci situate on t,,at a $ * "*
I7n Tf jt
*ur int^secti^
 *~ of the P ara.o,a
H the eirde touch the paraoda, the e, has a node at the ^int'of contact.
a cusp at^l^f coif * * ^ <* one, order, the curve ha,
TJ* 41
intersectioL Te'itlfi^te^^*:; "" ^ / the P a ^ol a , then the four
1 this axis for an axis rf s X '" '^ * " lis 1>xis ' md tho
the
contact, , the curve
, and the c,c,e
f the P oi t3
vertex), then the curve has a ta e T f ''I'" 1 ' ' ^ >uahll the P arab la *
oh other and the parabola at h 've to tt T^ * ?" ^ ir le
parabola at its vertex, and the ch' e h ^1 , "! M i8 the tan S ent to the
for its centre, and pa s in throu.hl , haVmg * he feous of the P*ol
>i of the seLa^s ^ ^ * fe  'hing, haLg its
", bicirc 1 qnartio such that its
&& irc1 ^  " curve is a
!' ""* ^ ** * he ^<*
llrteraect 'o of the conic with the
', ^ f " n  ax!al fooi of the
* he ial fooi ' ^he third axial
Tf +1,
four nodofoci oeindae
Cartesian having the ce of
of the cuspidal tangent, of the Cassia
other droJa, or say with the orthoZ
Cartesian; vk, the antipoint ^ heT . ,
focus i, the centre of the'o tLoli
414] ON POLYZOMAL CURVES. 543
Article No, 1*78. Gase of Double Contact, Casey's Equation in the Problem of factions.
178. In the case where the conic has double contact with the orthotomic circle,
then (as we have seen) the envelope of the variable circle is a pair of circles, each
touching the variable circle ; or, if we start with three given circles and a conic
through their centres, then the envelope is a pair of circles, each of them touching
each of the three given circles ; that is, we have a solution of the problem of
tactions. Multiplying by 2, the equation found ante, No. 173, for the variable circle,
and then for the moment representing it by (a, b, c, f, g, h$w, v, w) a = 0; then
attributing any signs at pleasure to the radicals Va, Vb, Vc, the equation of a conic
through the centres of the given circles, and having double contact with the ortho
tomic circle, will bo
(a, b, c, f, g, h}j>, v, ?y) a (Va + ?jVb + wVc) a = 0,
viz., representing this equation as before by
hw + mwu + nuv = 0,
wo have _ _, _
I : in : ufVbe : gVca : h  Vab,
that is, substituting for a, b, c, f, g, h their values, and taking, for instance, a, b, c
= fi"V2, fc"\/2, o"V2, we find
' 67,
that is I, on, n are as the squares of the tangential distances (direct) of the three
circles taken iu pairs, and this beingjo, the_equation of a pair of circles touching
each of the three given circles is Vm + VmS + VS' = 0. It is clear that, instead
of taking the three direct tangential distances, we may take one direct tangential
distance and two inverse tangential distances, viz,, the tangential distances corresponding
to any three centres of similitude which lie in a line; we have thus in all the
equations of four pairs of circles, viz, of the eight circles which touch the three
given circles. This is Casey's theorem in the problem of tactions.
Article No. 179. The Intersections of the Come and Orthotomic Circle are a set, of four
Qoncyclio Foci.
179 The conic of centres intersects the orthotomic circle in four pc
each of 'these the radius of the variable circle is =0, that is, the points
are a set of four coneyclic foci (A, B, C, J3) of the curve. Regardmg the
the circle which contains them is of course the orthotomic circle; and
singly inHnite series of curves, viz., these correspond to the singly mfii
conies which can be drawn through the given foci. As for a given cur
544 ON POLYZOMAL CURVES. [414
four sets of concyclic foci, there are four different constructions for the curve, viz., the
orthotomic circle may be any one of the four circles 0, R, 8, T, which contain the 'four
sets of concyclic foci respectively; and the conic of centres is a conic through the
corresponding set of four concyclic foci. We have thus four conies, but the foci of each
of them coincide with the nodofooi of the curve, that is, the conies are confocal; that
such confocal conies exist has been shown, ante, NOR 78 to 80.
Article NOB. 180 and 181. Remark as to the Construction of the Symmetrical Curve.
180. It is to be observed that in applying as above the theorem of the variable
zomal to the construction of a symmetrical curve, the orthotomic circle made use of
was one of the circles R, 3, T, not the circle 0, which is in this case the axis in
fact, we should then have the conic and the orthotomic circle each of them coinciding
with the axis. And the variable circle, qua circle having its centre on the axis, cut*
the axis at right angles whatever the radius may be; that is, the variable circle is
no longer sufficiently determined by the theorem. The curve may nevertheless be
constructed as the envelope of a variable circle having its centre on the axis viz
writing 51 =(tf_^) 2+ ^ a 'V, & c> , and starting with the form
then recurring to the demonstration of the theorem (ante, No. 47), the equation of
thejanable circle BSr + bS' + oC'O, where a, b, c are any quantities satisfying
iT" ' r ' What is the same thin g, Baking q an arbitrary parameter, and writing
S
  1 J. m 1 n
a H<r, ^iff, 
n i
= 4 the equation of the variable circle is
Compare MOJL 118188 for the like mode of construction of a conic; but it is proper
to consider this in a somewhat different form.
181. Assume that the equation of the variable circle is
3> = 0^) 3 + y a cZ% 9 = 0;
we have therefore identically
ar + W + or + dfc'O,
viz., this gives
a +b +c =d ,
aa +bfi +cc=d<2,
a (a?  a"*) + b (6 a  V*) + (p _ c "*) = _ d ^ __ d ^
and from these equations we obtain a, b, c equal respectively to given multiples of d;
substituting these values in the equation ^^ = 0, d divides out, and we have an
UJN i J OLYZOMAL CURVES. 545
equation involving the parameters of the given circles, and also d, d", the parameters of
the variable circle; viz., an equation determining d", the radius of the variable circle,
in terms of d, the coordinate of its centre. I consider in particular the case where
the given circles are points ; that is, where the given equation is
The equations here are
a 4b +o d,
a + b& 4 cc = d",
act 3 I b6 a + cc 9 =  d (rf a  d"),
and from these we obtain
U? 1?
so that the equation  + +  = becomes
1
I (a  1)) (ao) m(b jO(& Z^L + " ( c ~ a ) ( c  & ) =
^b) (d  c)  d" a + (d  t>T(d  )  &** "*" (d  a) (d  6)  d" 8 '
or, as this is more conveniently written,
I 1 ___, * + J!i * y/ ^ Q t
T^"o (d  6) (d ' c)  d" 8 c  a (d  c) (d  a)  d" 2 a  6 (d  a) (d  b)  d" 2
vk, considering d, d" as the abscissa and ordinate of a point on a curve, and repre
senting them by w, ?/ respectively, the equation of this curve is
I 1 _m 1 , _^ i  o,
which is a certain quartic curve ; and we have the original curve
as the envelope of a variable circle having for its diameter the double ordinate of
this qnai'tio curve.
I m n T M A
Write for shortness ^ , ^^, ^Tfi* 1 " '
of the quartic curve may be written
2i [( )'(*  fiXfflo)^^ ) ( 2i
viz., this is .
%L[co(a>ct)(e>'V)(>o)
 (a + b + c) cc + (06 + o + i
0. VI.
546 .ON POLYZOMAL CURVES.
or what is the sanle thing, the equation is '
 (La + Mb + No) (x  a) (a  b) (a  c)
+ f {(La + Mb H jYc) + Lbo + Jfoa + #aft) = 0.
In the particular case where L + M+N = Q t thai is, where
4.
r
7 ' ' /  vi, .
0' C C (i ft (J
the quartic curve becomes a cubic, viz., putting for shortness
S =
the equation of the cubic is
aiS"
viz,, this is a cubic curve having three real asymptotes, and a (linmotor ab ri
angles to one of the asymptotes, and at the inclinations + 45, '15" to fcliu other
two asymptotes respectively say that it is a "rectangular" cubic. Tho rohition
+ a^6 =5 impli08 that tho ourvc ^+Vwii6 + Ve'0 is a OavliOMuin, and
we have thus the theorem that the envelope of a variable circle having for dtamotoi
the double ordinate of a rectangular cubic is a Cartesian.
I remark that using a particular origin, and writing the equation of tho ruofciuiguliu
cubic in the form j sa ^2ma! + + ^ > the equation of tho variable circle in
ft
1
d
that is
where d is the variable parameter, Forming tho derived equation in regard to d, wo
n
have
^A
and thence
nn P 
known form of the equation of a Cartesian.
= ( which is a
414] ON POLYZOMAL CURVES, 547
Article, ,Noa. 182 and 183. Focal tforimikQ for the General Curve.
182. Considering any three circles centres A, B, G t and taking 91, &c., to. denote
as usual, let tho equation of the curve be
then considering a fourth circle, centre D, a position of the variable circle, and having
therefore tho same orthotomic circle with the given circles, so that as before
bho formula No. 47 (changing only U t V t If, T into 91, 2T, (T, 3>) are at mice
applicable to express the equation of tho curve in terms of any three of the four circles
A, B, G, D. ' ,  :
In particular, the circles may reduce themselves to the four points A, B, G, D, a
act of concyclic foci, and here, the equation being originally given in the form
the sarao formulas are applicable to express the equation in terms of any three of
the four foci.
183. It is bo bo observed that . in this case if the positions of the four foci are
given by moans of the circular coordinates (a, , lj, &o., which refer to the centre of
tho circlo A BOD as origin, and with the radius of this circle taken as unity, then
tho values of a, b, e, d (ante, No. 90), are given in the form adapted to the formal*
of No. 49, vis;,, we have
a : b : c ; <U(j8yS) : /3( 7 Sa) : ?(S0) : S(#y).
, / if*t fi , 
where W) = ( ~ 7) <7  S) (S  /5), &o. The relation  + ^+ = 0, putting therein
I : m : n p(/3 7)" : r/3(7) : T 7 (/3) a , (or, what is the same thing, taking the
equation of the curve to be given in the form 03
vw. thie equation, considering p, a, r, a, ft, 7 " given, determines the position of the
fourth focus D, or when A t B, 0, D are given, it is the relation which must exist
between p, <r> T; and the four forms of the equation are
vi,, the curve is represented by means of any one of these four equati
each of them three out of the four given foci A, B, U, V.
548 ON POLYZOMAL CURVES, [414
Article Nos. 184 and 185, Case of the Circular Cubic,
184. In the case of a circular cubic, we must have
Va/3 OS  7) + V/3~<r (7  ) + V<yr (a  /3) = 0,
which, when the foci A, B, 6, D are given, determine the values of p \ a : r in order
that the curve may be a circular cubic, We see at once that there are two sobs of
values, and consequently two circular cubics having each of them the given points
A, B, Q, D for a set of concyclic foci. The two systems may be written
viz., it being understood that V^S means Va.VS, &c,, then, according as VS has one
or other of its two opposite values, we have one or other of the two systems of
values of p : ff : T. To verify this, observe that writing the equation under the form
Vap : Vjfo : V^=aVg Vo^y" ; ySVgVa^ : ryV'SVo/gy,
the second equation is verified ; and that writing them under the form
where
M =
the second equation is also verified.
185. If we assume for a moment = cos + i n = fl fa & c>( viz< if ffl 6 c rf
t^nchnat 10 ns to any fixed line of the radii through A t S. C, iJ res^eedvely then
and thence
or else : cos  (c + rf _ fi _
sn
: sin i (6 + d  o  a) sin J ( a)
Putting in these formula, : sin ^ + d "~ " " 6 ) sil1 4 ( a ~ & )
^ ^ then we have B  = (6 _ )
(7  J. = ^ ( c _ a ) f
41<L] ON POLYZOMA.L CURVES.
and for nitlusr sob ol' values tho verification of tho relation
549
V (/3  7 ) + V/3^ (7  a) H VTT (a  /3) = 0,
will dupond on liho two identical equations
tiin .4 Hiu (;j  0} I Bin 7^ mi (G  A} + sin (7 sin (A  5)  0,
COM A Hin (./i  0) 4 OOH Ji sin (0  A) + cos a .sin (AB)^Q:
although tho forogoing Holution for tho caso of a circular cubic is the most elegant
ono, I will prcwjutly rotuin to the quostion and give the solution in a different form,
Artiolo No. 180. *W Formalee for the Symmetrical Curve.
180. In this nynunobrical cfwo, whoro the foci A, B. G, D are on a line, then if,
ns UHiuil, a, 6, o, d .lonoto the dirttancoa from a fixed point, we have the expressions
i)l' (a, 1), (!, d) in a form adapted to tho formulas of No. 49, viz.,
&)M^
tlmli, assuming
: m :
thu
 r i r
a 1) o
,vud tho .^nation of tho curve may bo presented under any one of theourforms
 V?&d V(o&) (VSt.^.V
Article No. 187. tttsfl of the Symmetrical Circular Cubic.
187. Kor a circular cubic we must have
*/ '
Theno oquatiunfl givo Vp
equations), or G!HO
Vrl ' 1 : 1
. VT 1 . 1 .
which obviously satisfy the two
580 (IN I'lU.v/MMAi. iTuvr/i,
In (noli, Miomt valucH ulivimifily .''iiti'ly (In 1 <rMn>i t.puh.m. ,1,^1 p,, f , , ,, ,t . f
HiitiHly tlui Ih'Ht, (iignuMon, we liavi< niily I" wnh< Hn in HU<M ijj.t (..,,
/>:<;: T' /]/l(/i !'')(" I </): .V Mr * H in'. * <?* If .(,,.*/.,,.> ;.
' ' ? * ' P * f f
I/'^(<H /).(<' .. ( /y. The lir:.[, ; i... jivr. ir !.. MIK.
(//r\'VI  (r ,n\\*\ i in /*IM .M,
ll t'.lllH (Illlll.llillS lln> lilir ,? I) i,,[. y, ), U f r ,.(,,, jr ti ,
in liwirn; tl il\ j.rnu..' >'!, vvilli lli> ),. i v ) /; / / ,
" ' " '' *
(A(j)('/.l(//i)\ / yi I (... r)</. I ./ r ,n\'W * ,.j
C(iaii(lll ill' Wllittll i,'l. Hi' .'Mill;,.., r\.,.iM.r ,, >;,,!, ,,j (f.
Artirlc
INN.
in A', anil
>"" N """ '"' "'"" <.",,,,, ,:,,<,.,;;,;,; : i ;;...'ir 1 :.:';;:; 1
anil
tlutf, I.,,
micl Oho Hko HH rogftixln
ON 1'OLY/OMAL UUUVJ58.
.551
INI). ThoHO valuuH of V? : Vwi ; VK givo tho oquationR of the two circular cnbica
with' tho foiii (A, 1i t <!, ;j), the equation' of oaoh of tlitjin undor a fourfold form,
vix, ( wo Imvu
mid
c'i it, ,
], (I, ~ CJj
,, 6,  !
<2, I &, ,  6, l d
(6 (t j (Jj (t
(first OUI'VO),
/>,
=
(HOCOUtl OUllVO),
HJO. Similurly 6VI and 7l/> niunt in 8, "mid if wo dcmoto by a u , & 3 , j, c?j tho
irs fnun iS T nf l,ln; I'onr points roHpoclitvoly, HO that <y&j M^ ~ nwl, B S' (obnorvo
bhub if UN UHual /I, ,/J, (7, .1) avo talccm in owlor on tlio citclt) 0, thon A, MO on
opptmHid HidoH of W, rind mmilurly Jl, D aio on oppiiaito Hidon ul' V> HO that leaking
"JP ftu lmitivo <?.j, tl t will Ins iid^itbivcOj w liavo
a ; b : o : d (ft,,  rf.,) : rf a (6 a (( a ) :  , (&s ~ tl^ : ^(y a f(,),
and Uion thu (ii
of vnbniH
and
I , :  : 0, Vi + Vm I V;i  0, uro mitinfiotl by tbo two sotn
It I) I*
?1 /. .,.__/. / J. /i
 (/,J * (jy , [(,j {(jj . IJj "~ (/.J.J
and wo havo tlui mpiutioiiH of tho wain** two cubic. iwi'WH, ouch utpiattou undor a
fourfold lorn i j vix,, UUHO am
:Q
and
6,  ff a ) (V?f, VS,
h il tl
H t'a .
(uoooud
101. And again ^17i and C'D moot in T, and donating by ^, 6 a , c dl rf 3 the
tlistanooB from T of the four points roapootivoly, ao that ajb tt = c a rf a = md, *%\ wo havo
u : b : c ; d = & a (c a  <i) : ft,(c,(Q : f4(s^) : o a (fi Z^);
552
ON POLYZOMAL CURVES.
[414
ami the equations  + ^ +  = 0, VJ + Vm + V = 0, then give for */i t
1 a b c
of values, viz,, these are
two sots
and
and we again obtain the equations of the two cubics, each equation under a fourfold
form, viz., these are
and
_ /* J~ ft mm ft JLm /I
I "3 T l<3 t( 3 f 1/3 ,
"3 ~ &3 I fy ~ MS > Bg 3 ,
63
 a.
= 0.
192. The three systems have been obtained independently, but they may of
course be derived each from any other of them: to show how this is, recollecting that
we have
MA, MB, HO,
then to compare
the similar triangles
and the similar triangles
j.^ , j.jj =a a> 3> eg,
(jive Ji G
RAG give 0,Cs : Oj a,,
RED =b 
using these equations to determine the ratios of ,, 6 a , Ca , ^ a we have
_
~'
that is
414] ON POLYZOMAL CURVES. 553
and henco
& a ( &iC! + fl! 3 + chdj.  di 2 ) + c a ( hdi + Oidi + ^d  CjriO = 0,
that is
Z> a (d a &) + c a (OiO,  61 c4) = 0,
but
O 1 c l t 1 d l (e 1 d,) l
or the equation gives & a +jC 3 =0, or say 6 a : c a = &, : h, and this with ^r^ = ^ = ^'
gives 'all tho ratios, or we have
a, : & a : c a : d a &, (ffl!  d.) : 61(610.) : ^(a.dO : di(6 1 c l ).
We have then for example
& a c a : c 3 ff 3 : a a 6a6,Ci : GJO, : dft,; &c.,
showing the identity of the forms in (d, &i, c,, d,) and (a a , 6j, c a> cZ a ).
Article No. 193. 2Vons/ormo*ion to a #ew Set o/ Conoydio Foci.
193, Consider the equation
which refers to the foci 4. 5, 0, and taking D the
i^w/^i l \?ii:
tliafc we can find i,, w u such that identically
~Wt +m95 + = Zi
and that m lttl == mn. The equation of the curve gives
$
wo have therefore
_
that ia>
vi,, this is the equation of the curve expressed in terms of the concyclic foci
Article No. 194 2fe Itonuomol OHT^ JfcwmpawW or
194. I consider the tetrazomal curve
where the zomals are circles described about any gi
a vi.
554 ON POLYZOMAL CURVES. [414
There is not, in general, any identical equation a9P + b33 + c( H d3> = 0, but
when such relation exists, and when we have also ~+b~ + Q+3 ==0 > tne11 tne ourvo
breaks up into two trizomals. When the conditions in question do not subsist, tho
curve is indecomposable. But there may exist between I, m, n, p relations in virtue
of which a branch or branches ideally contain (^=0) the line infinity a certain number
of times, and which thus cause a depression in tho order. of the curve. Tho several
cases are as follows :
Article No. 195. Gases of the Indecomposable Gurve.
195. I. The general case ; I, m, n, p not subjected to any condition. The curve
is here of the order =8; it has a quadruple point at each of the points /, J (and
there is consequently no other point at infinity); it is touched four times by each of
the circles A, S, G, D; and it has six nodes, viz., these are the intersections of tho
pairs of circles __
v?w + Vil 3 = o, VM + Va>" = o,
= o, ?r + = o ;
the number of dps. is 6 + 2 . 6, = 18, and there are no cusps, hence the class ia = 20,
and the deficiency is = 3.
II. "We may have
there is in this case a single branch ideally containing (#=0) tho line infinity; tho
order is =7. Each of the points /, J is a triple point, there is consequently one other
point at infinity ; viz,, this is a real point, or the curve has a real asymptote, Them
sire 6 nodes as before ; dps. are 6 + 2,3, = 12 ; class = 18, deficiency = 3.
III, We may have
m = 0, */n + Vp = ;
there are then two branches each ideally containing (a=0) the line infinity; the order
is =6. Each of the points J, J is a double point, and there are therefore two moro
points at infinity. These may be real or imaginary; viz., the curve may Imvo
{besides Jibe asymptotes^at /, J) two real or imaginary asymptotes, Tho circles
V/[ + Vm23 = 0, V*iS + Vio==0, each contain (*0) the line infinity, or they reduce
themselves to two lines, . so that in place of two nodes we have a single node ab the
intersection of these lines; number of nodes is =5, Hence dps. are 5 + 2, 1, =7. Olaaa
16, deficiency =3.
IV, We may have
Vr : Vm : V ; V^ = a : b ; c : d
411 I ON I'O&YKOMAL CURVES. 555
l,h*'Ui in hcin it hiu^li) brannli containing (^ J = 0) bho lino infinity twice; the order is
. . (i. Kuril nl' llic points /, ./' IH a double point, and there are therefore two more
pninlM ul. inlmity, t.linl. in (bimidtsH tlm asymptotes at /, J), there are two (real or
imaginary) miyiiiplntni. Tim mnnbui 1 of iioduH, as in the general case, is 6, Hence
. I 11.1. : >N; I'liiHH in "K; dullciuuuy =2.
lun
I tunic" Mm imiliidud partimdai' WHO whuvo the circles reduce themselves to their
rn; vi/., wr liavu hern llw tnivvn
wl.i.li (M.T ti/i/ii N. Id) in in IUl tho (iiirvo whioh in tho locus of the foci of the
ri(M l ( . H whirl, ,HIHH ilmMiKli thn lour 1M nfcH ,L, ., (J. .. It in at present assumed that
,,, ,.,. .,,,., lin , ,,ul, a rirnlts tliw tuuso will bo ooiwidorad jrort No. 199. 11 we
huvn /', . I/' m,.li IlK in /,:; (,M. 7W iu ^ W d ^ (W> in 2 f then those pomt.
If. N. 7 1 iav lluv.' "f I'bn nix nudcK. In i'aot, writing down the equations of the two
im( , n(hi . rvi M . whpl t , l(1 IJlireoilt point in taken at ll t we have 23 : 
"< = ^^ ; ^(ABOf.VB
I 1 f .I I I \' ' ffr'llM"!iL<" i ll,llllil MI* 11 i*iv j i   ' .
tt !'?" Vul ,th nl Urn tw mroh* im bhnmgh tho point * or tins pomt is a
,,Mrlr. Mimilurly, UM ),niulH N and '/' aro uwh of thorn a node.
V. II 1
(M. ( . in.' IMMV Lluvt' bmiu'lui
P L ! f U'.t >ili ill Mil 1 KM IlLH i. i </ *" ' v * "" ~nf j. _ b ,
"" U "' ' < .,""" < .";. in Ul val, or ono col and two imaginary; that
,,, 1,,,,,1,H MI n,l....ly """ !"'";' J, \ lu ,, , wympt otos, all real, or one real and
it. ,,!.!,, i 1 1m iitVIllltllMfN ai I, >' ) uu'ii* i"> i i .,../.. ,! it.
! . Kmli i.f Li"* "ir'JH V9HVSU0. &o., contains tho^hno ^^^ Q 1S
llniH n'lniT'i i' ' li'
rlitHH !  !'l'", di'iiiit'iiry  .
to
556 ON POLYZOJIAL CURVES. [414
IV. VT+Vm + Vtt+N/p^O, a Vr+fiVm + cVttf d Vp0; corresponds to IV. supra,
viz., there is a branch ideally containing (z* = 0) the line infinity twice. But, observe
that whereas in IV. supra, in order that this might be so, it was necessary to impose
on l t fii, n, p three conditions giving the definite systems of values VT : Vm : */n : */p
= a : b : c ; d, in the present case only two conditions are imposed, so that a single
arbitrary parameter is left.
V. 'Ji~*Jm = >Jn = *jp\ corresponds to V. supra.
VI. V^H Vm = 0, V?i + Vp = 0, a*Jl + &Vra4 c V + d Vp  0, or what is the
same thing, *Jl : </m : V?i : Vp = c rf ; d  c : b  a : a  b ; the equation is thus
(c  fZ)(Vf  V3T)  (a  6) (VST ~ VgT) = 0. There is here one branch ideally containing
(s 3 = 0) the line infinity twice, and another branch ideally containing (2 = 0) the line
infinity once ; order is = 6. Each of the points I, J is an ordinary point on the
curve, the remaining points at infinity are a node (21 = 53, ( = 3)), as presently
mentioned, counting as three points, viz., one branch has for its tangent the lino
infinity, and the other branch has for its tangent a line perpendicular to the axis;
or what is the same thing, there is a hyperbolic branch having an asymptote perpen
dicular to the axis, and a parabolic branch ultimately perpendicular to the axis. The
number of nodes is =5, viz., there is the node 31 = 33, S = !D just referred to; and
the two jKrire of nodes ((c  d) VSP  (a  b) V(T = 0,  (c ~ d) V$T 4 (a  6) V^ = o) and
( c rf)V2[ + (a6)V3) = 0, (c^V^ + ^^VrO), each pair symmetrically situate
in regard to the axis. Hence also dps. = 5; class =10; deficiency = 1.
And there is apparently a seventh case, which, however, I exclude from tho preaonb
investigation, viz., this would be if we had
a , b , c , d ,
a 3 , & , G\ d* ,
a"\ b" s , c" 3 , d"*,
that is, a, b, c, d denoting as before, if we had
VZ : Vi : V : >jp = a : b : c ; d, and act" 3 + b&" a 4 co" 3 + drf" a = 0.
For observe that in this case we have
dSyO, and + +  + == Q
abed '
that is, the supposition in question belongs to the decomposable case.
Article No. 197. The Decomposable Gurm,
197. We have next to consider the decomposable case, viz,, when we have
414] ON POLYZOMAL CURVES. 557
ftco ewifo, Nos, 87 et seq.ii there appears that (unless the centres A, B, 0, D are in
a lino) tho condition signifies that the four circles have a common orthotomic circle;
and whan we have also
d
formulas for  the decomposition are given ante, Noa. 42 et seq. Writing therein
, 38, (T, 35 in place of ^> K W > T respectively, it thereby appears that the tetra
our Jo V&T + VmW WW + V* 3 = 0, breaks up into the two trizomal curves
= o, vpp + v^ + vis 5 = o,
whovo
A=^ + , ^=vr + 5,
c
111111 whmo w
, J . 2i . + .
"b + o ' a b
Artiolo Wos. 198 to 203. Oases of 0. Dec m po S Me Ou,;e, Centres not a K.
198 I assume, in tho fivst instanco, that the oenta. of the cu,les ave not m a
llno! wo have tho following ca.cs: ^ ^ Mnud ^ =g .
I. No furthc. relation between J ,  , P , ^ '^ j.^^ qim , ic .
tho ordov of oaoh of tho tmonuda as =4, that
 / / , ^n. rtip order of the tetrazomal is =7, that ot one
II. \fl + Vi + V + Vj) = ; me omu
tho trizomals must bo = 3.
To verify this, observe that we have
  i
_4..
VI VI V bod
for Vl+ VS+V5 the valne __
H^^ 1 '
,Ti
n the siga
condition, , b y .
558 ON POLYZOMAL CURVES. [414
and hence one or other of the two functions
V + Vw, + Vj, V + Vm^ + V?^ is = ;
that is, one of the trizomal curves is a cubic.
III. Vr+ Vp = 0, Vm + Vn = ; order of the tetrazomal is = 6 ; and hence order
of each of the trizomals is = 3, To verify this, observe that here
which since a + b+c + d=0, gives ?S so that, properly fixing the sign of the
1 1 if Lid
/aH
radical, we may write Vi + jJ =r Vm = 0. Wo have then
A = ^TT V V^ + V = A /~~ (b + c) Vm ;
a V abc
which last equation, using A/r~ to denote as above, but properly selecting the signi
fication of +, may be written
,/ . ,/ . b + c /ad /
Vni! H vi = + T A/ r vm.
Hence
V + (VS + V) = i {(a + d) VI + (b + o) A M
w ( V DC
afd f /r
viz., V^TCVflh+Vw,) with a properly selected signification of the sign + is =0; and
similarly VJ> + (Vw^ + V 3 ) with a properly selected signification of the sign T is =0;
that is, each of the trizomals is a cubic,
199. IV. V^ : Vm : Vw : ^ = a : b : o : d (values which, be it observed, satisfy
of themselves the above assumed equation l  + + * + = Q] the. order of the tetra
f* U u Cl /
zonml is_ = 6; and the order of each of the trizomals is here again =3. We in fact
have Vt = a + d, V^ + Vn^b + c, and therefore V + V^ + V^ = 0; and similarly
+VH 3 = 0; that is, each of the trizomals is a cubic,
I attend, in particular, to the case where the four circles reduce themselves to
the pomts 4, B, 0, jD; these four points are then in a circle; and the curve under
consideration is
414] ON POLYZOMAL CURVES. 559
in tihe gonoral case where the points A, B, G, D are not on a circle, this is, us has
boon soon, a aextic curve, the locus of the foci of the conies which pass through
tho four given points; in the case where the points are in a circle then the sextic
breaks up into two eubics (viz., observing that the curve under consideration is
V^T + Vm& + V?i + V = 0, where VF : Vm : Vn : Vp = a : b : o : d, these values do
of lihomsolvos satisfy the condition of decomposability  + ^ + J; + d ==0 ' ) ' that 1H ' the
looua of the foci of the conies which pass through four points on a circle is composed
of two circular eubics, each of them having the four points for a set of concychc
foci Tt is easy to .sec why the sextic, thus denned as a locus of foci, must break
up 'into two eubics; in fact, as we have seen, the conies which pass through the lour
Jmoyolio points A, B, G, D have their axes in two fixed directions; there is con
H0 ( ,ucntly a locus of the foci situate on the axes which are m one of the hxed
tUroutioi*. and a Repavate locns of the foci situate on the axes which He in the. othu
of tho flxod directions; viz, each of these loci is a circular cubic.
200. Adopting the notation of No. 188, or writing
.(and thovoforo 6,0, = o,d,) we have
a . b : e : d *ft
'Movcovor
and wo have
ugroeing with the formute No. 188.
The tetrazomal curve
560 ON POLYZOMAL CUBVES,
is thus decomposed into the two trizomals
(6  o,) VI + (c,  V33 + (a,  60 Vg = 0,
(6j  cO VI + (c, + VS  (a, + 60 VI = 0.
201. Observe that the tetrazomal equation is a consequence of either of the
trizomal equations; taking for instance the first trizomal equation, this gives the
tetrazomal equation, and consequently any combination of the trizoraal equation and
the tetrazomal equation is satisfied if only the trizomal equation ia satisfied. Multiply
the trizomal equation by  Oj f d\ and add it to the tetrazomal equation ; the resulting
equation contains the factor c^, and omitting this, it is
where observe that ^d is the distance BO, and 0,1 dt the distance AD. But in
like manner multiplying the second trizoraal equation by ct^+di, and adding it to
the original tetrazomal equation the resulting equation, omitting the factor a it is
(6,  cO ( E + 2>)  (a,  d,) (S8  S) = ;
viz., it is in fact the same tetrazomal equation as was obtained by means of tho first
trizomal equation,
The new tetrazomal equation, say
(6,  cO ( VIT+ V) + (a,  d,) (VS  VS) = 0,
is thus equivalent to the original tetrazomal equation; observe that it is an equation
of the form VJ2I + VwS + VM 4 Vp5) = 0, where
and where consequently Vi + Vp=0 J Vm + V?i=0, that is an equation of tho form
(198) III., decomposable, as it should be, into the equations of two circular cubics.
Writin
where 8 is an arbitrary parameter, the curve is obtained as the locus of tho inter
sections of two similar conies having respectively the foci (A, D) and tho foci (, G)
(see Salmon, Higher Plane Ourves t p. 174): whence we have the theorem, that if
A, B, G, D are any four points on a circle, the two circular cubics which are the
locus of the foci of the conies which pass through the four points ^1, B, (7, J), are
also the locus of the intersections of. the similar conies, which have for their foci
(A, D) and (B, G) respectively; and of the similar conies with the foci (, D) and
(G, A) respectively; and of the similar conies with the foci (a, D) and (A, B) respectively,
202, V, Vr=Vm = V7i = Vp. The order of the tetrazomal is =*5 t whence those
of the trizomals should be =3 and 2 respectively. To verify this observe that the
ON 1?OT,YHOMAL CURVES. 561
..mutiim ' ' + '" \ M I 1 '. () irivtiH L I ; l + l + \ = 0, and combining with a + b + <Hd = 0,
1 u I) c d n a b (s d
lu"HM iivc <nlv nitUfitii'd liy ono itf tint HyHtoran (a + b = 0, ctd = 0), (a + e 0, b + d==0),
la , .1 ,0, It I (^ 0), SulniiUnK U> lix the ideas the first of these, or writing
(a, b, o, d) = (a, a, c, c),
>tn thiil wi>
,.,. n.iuiitinn whirh HiKhilicH lhat thu radical axis of the circles A, B is also the
i
m.i,nl \i< I' ill" iiivnltH (h J>\ tlum, writing Jus wo may do,
Ul' lltU'M
1  
V, I 11, 2, V;ll, 0.
,., AiVw,.^,:.!). whuh iv !H ono of bho tmomals a cubic, viz., this is the
T ....... 1,,r ,n, ....... I B  ..... Wr U, U.
UH,
;;:;: ;:: ,   
. // iiu"l "I 1 llm iiirlrH C 1 , A
tlte
,*r ..... u Hit*'; Hi" , . .
. , , " A" d" Wo establish as before the relation  c d
i.\vr lr thn mUu it , , o
lUV HM ininnv". . , .
of toti'azomal = o, 01 wia
I. "
* unit 4,
i 17. nt tvi7omals =4 and 3; same
J j. V i II j order of totvazomal = 7 ; oi tuzoraaia .
" tr'^W^o, V.HVS.O., o^ or .,  B  s ;
nro iw III. /. 71
C. VI.
562
ON POLYZOMAL CURVES.
[414
204, IV. vT+ Vm + Va + Vp = 0, a V/ f b Vm + c Vra + cl Vp ~ ; order of tetrazomal
= G ; this is a remarkable case, the orders of the trizomals are either 3, 3 or else 4>, 2.
To explain how this is, it is to be noticed that in the absence of any special
relation between the radii, the above conditions combined with  +  r I  h T  give
abed
vT : Vm : V?i : Vjj = a : b : c ; d( l ); when I, m, n, p have these values, the case is
the same as IV. supra, and the orders of the trizomals are 3, 3, But if the radii
of the circles satisfy the condition
= 0,
1 >
1 ,
1 ,
1
ft ,
& ,
C )
d
a',
&' ,
c 3 ,
rf a
ft" 2 ,
6",
o" 1 ,
rf" a
then the two conditions satisfy of themselves the remaining condition  + ~j  uii = o,
_._ abed
and the ratios Vf ; **/m : Vw :
arbitrary parameter.
"We have
instead of being determinate as above, depend on an
and between I, m> n, p only the relations
Vr+ Vw + Vn + Vjj = 0, a^l + b V?n, + c Vw + d*/p~ 0,
We find first
^ = Vi +
4.VPJ5
fa
(b v "
o?, i/ a , a 3 , tn 9 In place of \/f,
, \/a, \^, wo have to find *, y, a, w ftow the conditions
X + IJ+ & + Ws=0,
+rfiu=0,
where tlio oonatanta are connected by the relation
It readily appean that tha lino represented by tho first two equations towhe s _ tho quadrio aurfaco in tho point
x:y:xi w=>& : b : o : d, so that the B e are in general the only values of *Jl; Vw ; VJ: v In the oaso next
referred to m the text the line lies in the surface, and the values are not determined.
41 4" ON POLYZOMAL CUKVES. 5G3
and thou _
(cl  a) Vi = (6  d) Vm + (c  d) Vn,
(d a) Vp = (a  6) Vm + (a  c) V,
wlu.m(!0
d \T a VjJ = j (b V?t  c Vm),
Cti Cl
mid wo hiwo thus
(ol,,rv that itL the ca^not__ under consideratioa bVSo^0, ami thevefove
\//, I VwiTl Vwi O, V+ Vms + Vw^O).
In liho proHont ctiao wo have
thouoo ad_(6c)
, mly 0110 of th
wo havo V/l
"Wo havo then
hub wo find
and thence
P = : Hence
in virtue ot A/ b c d; triaomal ls a qua*
Bonding Vernal is , co^ but the othe.
564 ON POLYZOMAL CURVES. [414
205. V. \ll = *Jm= Vtt = Vp; order of tetrazomal is =5; orders of fcrizonmls =3, 2;
same as V. supra.
VI. ^+p = 0, m + m=sO, a\+&w + oVw + d = 0; order of liotriizomal
= 5 ; orders of trizoraals are 3, 2.
We have here
Vm, = Vm + A/ r^j b Vw,
Vjij = Vm + A /
V
r~i
bed
or wilting the values of Vm,, V^ in the form
./ j~ , /ad c
Vn, = ~ Vm + A/f T
V be d
ad
g cnau as oeiore t =
equations are
V/j = ._ ,\/; anc j guniiaviy
then observing that as before l=~m, if to fix the ideas wo assume v7= A V^ fcho
DC V be; '
,
a
whence
We have moreover
and thence
so that
the
414]
ON POLYZOMAL CUEVES.
565
VII. If we have 1,1,1,1 = 0, and ( 1 , 1 , 1 , 1 ) (VF, 'Jm, \ ! n, Vp) = 0,
a t I , o , d ft , & , c , d
a? , & 8 , c a , d 3 a a , ft 2 , c a , tP
n "a /."a "s ,7"a r,"a 7,"a n "3 ,J"s
(0 , U , C , tt II j O , (/ , to
tho totraaomal has a branch ideally containing (s 3 == 0) the line infinity 3 times ; order
is = 5 ; orders of the trizomals are 3, 2. We have here
Vr : Vm : */n : Vw = a ; b : c : d,
and thonoo
= a + d
= a + d
. bed / .
, = b   , Vnifl  b
bed
which give
Moroovei 1
bcd ]
a
bed
V^ + V?H t + VTIJ = 0, V^ a + V?ft a f V?l. a = 0.
a V/i l 6 VM&! + c V?ij = ft (a + d) + &b h cc
0o)
bed
bed
and similarly
whence in virtue of
one of the two expressions is
^
bo "(5 a)" 1
; and the trizomals are thus a conic and a cubic.
Article No. 206. The Decomposable Oww f Transformation to a differen
Conmjclio foci,
206. Consider the decomposable case of
Vm + Vm23 + VftS + VpJD = ;
vi,, the points A, B, G, J) Ho here in a circle, and we hav (
Taking (A, A) the antipoints of (4, D); W. ft) the antipoinf
566 ON POLYZOMAL CURVES. [414
VdDj = 9135, 93^ = 186: (No. 65) and referring to the formula, ante, Nos, 100. et scq.,
it appears that we can find Z,, in l} n ls p t such that identically
and moreover that ^j = ^j 1] mn
The equation of the curve gives
which may consequently be written
2 = ;
viz., this is
= ;
that is, the two tmomals expressed by the original tebrazomal equation hivolviiyr Un
set of concychc foci (A, B, C, J)) are thus expressed by a now totra.omal oqukfciou
involving the Afferent set of concyclic foci (A lt A, 4, A); and wo might of coinso
m like manner express the equation in terms of the other two set* of oonqyolio foci
a, #*> 9 A) and (A 3 B 3 O a> $ 3 ) respectively. It might have boon anticipated that
^ Uld
naralv 7  ^ Uld " "^^ ^ f b ' le ^P^"" ^HlH
epazately pass fro.n the original set to a different set of concyclio foci, and the two
s M wouicl {i raiht be J d to

into a smgle tetrazomal equation; bnt the direct transformation of the
equation is not on this account less interesting,
ANNEX I. On the Theory of the Jacobian.
Consider any three curves ZT , 7=0, F=0, of the samo order r, then writing
J(U, 7, F)=
*, z , ,
we have the Jacobian curve J(U, V t F) = 0, of the order 3r3.
A fundamental property is that if the curves U~Q F0 W n t
common point, this is a nnint *U T i.. f ' U> ^~ lmv o any
double pit, that L P I 1 i, ':"' "^ "? nl ^ S ' but ifc k,
 0,<V=o ? qUeSt W have /= . ^ have also
oo , e f roY 0+z<i r m0+
. i = 0, Jf.fl, > = each f V' = ' " l = ' ' l = each of th
each of the order ,) which have in common the
414] ON I'OLYXOMAL OURVK3, 567
(r ft') (r ~ H) points of intersection of tho curves = 0, = 0, each of these points is
a nodo on tho Jaoobian, and houco that the Jacobian must bo of the form
<I' 3 = 0,
whom obviously tho degrees of A, B t must bo i + 2s'8, r h s h s'  3,
respectively. In tho particular ease whore '=0, that is whore I, m, n arc constants,
wti have A. ; tho Jaoobian curve then contains as a f'uotor ( ( I> = 0), and throwing
out, thu (iiirvo in ./^)  G'<D ~ 0, vix., thin is a ourvo of tho order 2r + s 3
through tiiuih of r(r ) pointu of intomxition of tho curves W = 0, = 0.
Li partionlar, if r = ii l ,v 1, that in, if the cui'voH are tho conicH B + M ) = 0,
M  JI/0) o ( () H jVtj) ~ 0, pasHing through the' two points of intersection of tho conic
HsaQ by tho liim <J>~0, thon tho Jonobiau is a conic passing through tlicso sumo
two points, vin., ltn liquation IH of tho form $> H il<I> = 0. This intorsects any one of
tho givon (ionioH, Hay HIM' O in thu points (), <!' = (), and in two other points
( H H iitp 0, ii .=(>; at mo/* of the lastmentioned points, the tangents to the two
oiirvoH, and tho linos drawn to tho two pointn H = 0, <1> ~ 0, form a harmonic pencil.
Although \,\\w in, in faot, tho known thoorom that tho Jooobian of three circles
is thoir ortliotnmie oirtilo, yet it is, I think, worth whilo to give a demonstration of
tho theorem as ahovo stated in reference to tho conies through two givon points.
Taking (s 0, //; ~ 0), 0?  0, y  0) for tho two given points <"> = 0, = 0, tho
gouoral ittpiation of a c.onie through the two points is a qundrio equation containing
terms in 2 1J , SM, zy, tvy ; taking any two Much conies
os' J H 2e  2i/^ h 2/(,^ == 0,
thoso intniwwt in tho two points (a> 0, 50), (y0, ^ = 0) and in two other points;
hit (n;, y, s) bn tho coordinates of either of tho lastmentioned points, and tako (X, Y, %)
as (jurnmt (soordinateH, thu is(uutioim of tho linos to tho lixod points and of the two
tangents are
h A'*#w0, \ r eZy =0,
(hy I gz ) (Xz  %v) H (A +/s ) ( F*  %)  0,
(//// 1 Cte) (.A^  &;) + (//* + Fz) (Yz  Zy)  0,
whonco tho condition for tho harmonic rotation is
(hy I yz) (Hat I Ity f (Jus +f*) (Hy + Gz)  0,
0,
but from the otiuatioim of tho two conios multiplying by &//, Jft and adding, wo have
 (o/jf h 7t(7) z* + (W+fH) yz
vix., tho condition in thus reduced to
568 ON POLYZOMAL CURVES, [414
so that this condition being satisfied for one of the points in question, it will bo
satisfied for the other of them. Now for the three conies
C2 a + 2/ yz + 2g zee + 2/t an/ = 0,
cV + 2/' yz + tym + Way = 0,
cV + 2/ V + V^ + 2A"y = 0,
forming the Jacobian, and throwing out the factor 0, we may write tho equation in
the form
Of + 2Fyjs + 2 to 4 %Hmj = 0,
where the values are
G = g (fo" /V) 4 ^ (/" C ft' ) + ry" (/c' /''),
tf  (A'/"  h"f) + </' (/."/  A/') + cf (hf  A'/),
2F = A (fo" /V) + A' (/' C fo" ) + A" (/' / ),
2ff = A (<ff  cV) + 4' (fy ~ off" ) + h" (eg'  o'fl ) ;
and we thence obtain
off + AC = 
viz., the condition is satisfied in regard to the Jacobian and tho first of tho throo
conies; and it is therefore also satisfied in regard to the Jacobian and tho other two
conies respectively.
I do not know any general theorem in regard to the Jacobian which given tho
foregoing theorem of the orthotomic circle. It may be remarked that tho UHO in this
Memoir of the theorem of the orthotomic circle is nob so groat as would ali first
sight appear: it fixes the ideas to speak of the orthotomic circle of throo givon oivoloH
rather than of their Jacobiau, but we are concerned with the orthotomic circle loss aH
the circle which cuts at right angles the given circles than as a circle standing in
a known relation to the given circles.
ANNEX II. On CASEY'S Theorem for the Circle which touches three gwen.Circles,
The following two problems are identical :
1. To find a circle touching three given circles.
t<n,eVen p" sVI^of ^ ^ "" "^ f ^ '* = 0) *>*
bp Jjf, f ' ^ v 6 fll ' St r b ! em if We use * to de te a given constant (which may
be 0) then takmg a , * and {(*,") for the coordinates of the centre and for the
radms of one of the givon circles ; and similarly fc, 6', J(f _^. c> $_
414]
ON POLYZOMAL CURVES.
569
other two given circles; and 8 t S', i(f") for the required circle; the equations of
the given circles will bo
? + (* a") 2 = 0,
and that of the required circle will be
( fl! _S) a + (yfl l ) !l + (*O = 0.
In order that this may touch the given circles, the distances of its centre from the
centres of tho given circles must be i(S"a"), i(S"b"), i(S"~c") respectively; the
conditions of contact then are
(8ay + (8'ay + (8"a! f y > = V,
(S  6) 3 + (&'' 67 + (S"&'7 = 0>
(ir we have from these equations to determine S t S', 8". But taking (a a', a"),
(b, V, I"), (c, c', c") for the coordinates of throe given points in space, and (8, 8 , b )
(or tho coordinates of the centre of the conesphere through these points, we have the
very aiuno equations for the determination of (S, S', S"), and the identity of the two
problems thua appears.
I will presently give the direct analytical solution of this system of equations.
But to obtain a solution in the form required, I remark that the Aquation of the
conosphoro in question is nothing else than the relation that exists between the
coordinates of any four points on a conesphere; to find this, consider any five points m
Bpacc, 1, 2, 8, 4, 5; and lot 12, &c. denote the distances between the points 1 and 2, &c.^
then 'wo have between tho distances of the five points the relation
= 0;
0,
1,
1,
1,
1,
1
..a
o
2
2
1,
0,
12",
13",
14,
16
g
9
2
1,
2f,
0,
23,
24,
25
2
2
3
1,
31,
32,
o,
34,
35
^_ o
a
M 2
2
1,
41",
42,
43,
0,
45
1, 5l", 52", 53", 5'
whence taking 6 to be__the centre of the conesi
we havo 15 = 25 = 35 = 46 = 0; and the equation b.
o, Ti a , i?, I
21, 0, 23, 2
ii a , 32 a , o, s
2 a 2
41, 42, 43,
C. VI.
570 ON POLYZOMAL CURVES. [414
which is the relation between the distances of any four points on a conesphere; this
equation may be written under the irrational form
28.14 + 31.24 + 12.84 = 0.
Taking (a, a', a"), (b, b\ b")., (c t c', c"), (#, 2/, *) for the coordinates of the four points
respectively, we have
23 = V(6  c)' I (&'  G J + (ftTT^, ii = Vfo  a) a + (T/  ')' + (*  '?.
31 = V(c  a) 9 + (c'  a') 2 + (c"  a") 5 , 24 = vfy &) + (# &')+(*.&''),
12 = Vfa^)' + (a 1  &7 + (of"  ft' T )T 34 = V(fl C )* + (y c '> + ( C ' ; ) a ,
or the symbols having these significations, we have
23.14 + 31. 24 + 12. 34
for the equation of the conesphere through the three points ; or rather (since the
rational equation is of the order 4 in the coordinates (, y, z )\ this is the equation
of the pair of conespheres through the three given points; and similarly it is in
the first problem the equation of a pair of circles each touching the three riven circles
respectively.
In the first problem the radii of the given circles were {(*<&")> *(&") *'C*o")
respectively; denoting these radii by a, & 7 , or taking the equations of the given
circles to be
a') a a 3 =0,
(&)H(2/&') 3 ~/3 3 0 J
Oo} a + (z/c') a y = (
the symbols then are
12 = (o^. o/ + ('  6') a  (  jS)', ^^"c
and the equation of the pair of circles is as before
23.14 + 31.24 + 12,34 = 0;
where it is to he noticed that 23, 3l, 12 are the tangential distances of the circles
2 i and S, 3 and 11 and 2 respectively; viz., if , ft, 7 are the radii taken positively,
then these are the direct tangential distances. By taking the radii positively o
Z^T'f ^ P , lmUr ' btain in ^ fo  equationsthe tangential dices big
all direct as above, or else any one is direct, and the other two are Inverse: we have
thus the four pairs of tangent circles.
The conespheres whioh pass through a given circle are the two spheres which
th" he'for U ^ ^ antiP intS f ^ given Cil ' cle ; d ^ ise ^ to
that the foregoing mvestigatzon gl ves the following (imaginary) construction of the
414] ON POLYZOMAL CURVES.
571
tangent circles; viz,, given any three circles A, B t G in the same plane, to draw the
tangent circles, Taking the antipoints of the three circles, then selecting any three
antipoints (one for each circle) so as to form a triad, we have in all four complementary
pairs of triads. Through a triad, and through the complementary triad draw two
circles, these are situate symmetrically on opposite sides of the plane; and combining
each antipoint of the first circle with the symmetrically situated antipoint of the second
circle, wo have two pairs of points, the points of each pair being symmetrically situate
in regard to the plane, and having therefore an anticircle in this plane; .these two
antioirclos are a pair of tangent circles; and the four pairs of complementary triads
give in this manner the four pairs of tangent circles.
I return to the equations " '
(a  gy + (y 8J + (z  8J = 0,
(aS)* + (a'S') 9 + (a"~S") = 0,
(b  Sy + (&'  S')' + (6"  8"? = 0, ;
(c  S) 2 + (c'  SJ + (o"  S'J = ;
by eliminating (8 t 3', 8") from these equations we shall obtain the equation of
the pair of conespheres through the points (a, a', o"), (&, If, &"), (c, c', c"). Write
to 8, y8', z~8" = X> Y, Z, then we have Z" + Y* + Z* = 0, and, putting for shortness
then, by means of the equation jusfc obtained, the other three equations become
SI I 2 [(o  a) X + (a'  y) Y+(a" z}Z} = 0,
Those last otiuations give
X : 7
where
X = b'o"  6V + (c'  V ) e  (G"  6") y t
p = O 'a"  c"a' + (a'  G' )  (a"  c" ) y,
v = a'6"  a"6' + (&'  ft' ) z  (b"  a") y,
V6"o 60" +(o"6'>(o 6 )*,
/*' = o"ct  ca" + (of 1  c") so  (a  c ) 2,
y ' = a "b  al>" +(!>"  a") as  (b  a ) *,
X"=6c' fc'c +.(o 6 )i/(c' 6'),
fj,"=ca' o'a +(a c )y~(a! o' ),
y"=:a6' a'6 H(6 ft }y(b' a!}\
722
572 ON POLYZOMAL CURVES. [414
and the result of the elimination then is
(Xffl + + & ) + (X'Sl + P'% + !/($)' + (X"9l + ^"33 + i/'<)' = 0.
But aubstituting for 21, 8, 6 their values, and writing, for shortness,
 i = VG"  b"c' + cV  c V + a'&"  o"&,
j = &"c 6o"le"a ca" + a"& a 6",
 & = 5c' &'c + c ft'  C'B H a 6'  a' 6,
A = a (6'c"  6"c) + a' (b"c  bo") + ft" (bo 1  6'c),
^ = (6' C "  &V) (f* 2 + a' 2 + ft" a ) + (c'a"  cV) (& 2 + 6' a + &" a ) + ('&"  a"b") (c + c" J + c" 3 ),
 q = (i"c  6c" ) (ft 2 + ft' J + ft" a ) + (c"ft  cci" ) (6 a + i' 2 + 5" 2 ) + (a"6  alt" ) (o a + o' a + c" 3 ),
 r = (6o /  b'o ) (a 2 + a' s + a" a ) + (eft'  o'a, ) (6 9 + 6 /a + i" 3 ) 4 (a6'  ab' ) (c 3 + o' a + o" 3 ),
Z =(c 6 )(a a fft /3 + fl" 2 ) + (a c ) (& 3 4 b'^ + b" 3 ) + ( 6 ft ) (c a + c' a I c" a ),
7rt = (c' 6 1 )(ft a + a /a + a" 2 ) + (ft'c' ) (6 2 + 6' 2 + &" 3 ) + ( &' ft' ) (o^ 1 + c' a 4 o" u ),
 n  ( c"  b" ) (a 2 + a' 3 + ft ") + ( ft "  c " ) (^ + ji + &" 3 ) + ( &"  ft " ) (c' 1 + c' 3 I c" 3 ),
we find
4 z*
+ 2i (a? + f + tf)  %x (KB +j _
with similar expressions for X'Sl + /*'33 + 1/(, V2E + /53 + y'% and the result) I
ny  msp}*
Iz ~q } a
viz., this is
(ffi 2 + 2/
+ ) (4A (it, +j
+ 4 A 2  2 (ip +jq + fo.) + (^ a + m" + n n )J
(te + my + s) 9 + 4 (ins + jy + Jb) (pa + qy + rz]
no) + r ( maj
i n of fche Pk of ""wphoim The
norm of " be 8iWe t0 a numei ' ical
^
. I /i I ON VOLYXOMAL CURVES. 573
t.hn numuricul l'iu',t,(H' ol' tins (^pimmm in question is in fact =4, that is, the
norm in
K=  I. (/ J  f + 2 1J ) 3 (# Vf I /.; 3 ) I &c. ;
MO l.liiit aU.nndhig only Lit tin; higlumb powciH in (i, y, z) wo ought to have
(/>o)' J l</'' ..... ( 0'M(//WriV( ( ;< t W
.....< laM'jMn
1 1, l'i itisy Ui Him t.hai, Uui norm is in fuot conipoHcd of the torms
2(//o')M (fto) a (tfrt) a ('tW.
I 3 (</  ')' J 1 (b  c) a I (o  <0 a ~ ( ft  W>
.,. o ( (i  _ h'Y ( (6  o) 3  (o  (i)' J I (<t  &)"],
llll( ) ul' ,hn mnuluv U.rnw (, A, o), (i^ //', o"), and in K f, o') (", ^ Oi 'ho above
writlcn lijriim urn := !< inio
(//7(ft/0('to)l(''OMfto)(6tt) + (tt l 6')'Cofl)(o&).
,,,,,
w
 I,!,, vnln. ..I 1 Uu. nnnn is tluu. KC+J' + ftO. i' al "> 111
ANNKX HI. ,, rto W.m ,,/'(/, ...)V(,(o
Tho iinnii <( V(/.VK.Vir in
, ,i , i Jn i /P.I, \l\f\ V \* Vlfl 1^ J
vhinnt tliiili ol V a I (/ ! v i i i i " '
a 'I 'I 'I __ 'I
., L, 1, li " L >
wluMu tho laHt tonn i ^2 into
is obviously composed
anil tho norm
in n similar manner,
574 ON POLYZOMAL CURVES. [414
Now, applying the formula to obtain the norm of
(b  c) */(7+0 + a + (c a) */+8^& I (ft
the expression contains six terms, two of which are at once seen to vanish ; and
writing for shortness () in place of (1, 1, 1, 1, 1, 1} tho remaining terms ftio
+ 2 () {(6  c) 3 a, (c  ) fr (a 
+ 2^ Q ((6  C ) , (o _ a). A ( fl _ ft
+ 2^ () ((& _ C ) a ( c  of 1>\ (a  6) c ^(&  c ) 3 , (c  ) 9 , (a  Z>)' ) ;
the first of these terms requires no reduction; the second, omitting tho factor 2,
(6c) a a [ (6c)a'(oa)>&'( ft 6)v;]
+ (o  ) 2 /3 [ (6  c) 3 a 3 + (o  a) 9 6 3  (a  6)2 *j
+ (i)'7h(6o) a a(oo)fi + ( a 6) atf .i.
which is ' J '
n
Similarly the third term, omitting the factor 20,
s
which i
Bonce, restoring tho omitted factors, and oollootmg, we find
ON" I'OIVY/OMAI, OIMIVKH. 57f)
oiiuo, fh'Hf, writing w, h~~m, tim in plow of </,, /, c ; (Jinn //* lor
ifc" a ,  ft" 3 , o"' J ) I'm (, #, y); uml linally hitrndiioinj.,' s lor hniiingnmiity, wu Fl
Norm ((ft  o) vV' ~. ,,4'  j/' ~ "" ^ I (o  ,/,) ^ . (^ ~. A) v* j  s' J into
S' J ((6  ())' tl"' ] (I! )//'* I (fi 6V (/'*
 2 (d  ((.y (  /0 1J /'" a o'" j  a (i/  fty (/.  *i)' j <;" y ( f," u  2 <ft  (t ) ij ( (J  (
'*//' (6  d) (;  ./,} (if, ft) [ (A  n ) u" .  (d .  .) },"> \. ( lt , ... ft) (I "J
 ') (ft  o) (o  ./,) ((  l>)  (ft  f i) <(," ;l (z'> hn  w (h i c ) l (*;')
O t iind
Flint
HU na l,li) cqnuti.it. (t> V) \W\ (o* ) \\>\' (<i,. !>}\ f (*".(
' J :!(!) l,hn linr inliniLy twii:n, unil llir <iur\'o in UHIM a i<
Until tihn tix>n^Hii)ii ul' thn nonn !
vix,, wlimi Mm rlinr niittli'K Jntvn nniOi of Miriu rJn ttituui nuliim //', flin tiurvn IN Uin
iuir nl' iurallrl limi //^ /:" a" ;.! j atid in mrlii:iiliir wlirn /*;" 1 0, nr Mm oimli'H ntilmto
ihiMHMiKvM uiMili lit a inini., Lhuu Um mimi in // J :M(, Mm uxis t.wioo.
ANNKX IV. o. //IH Tristmwt. ttttrw* Jltf*\>\ ( itiV\ v''. ,. 0, WV/i Aiuw ft ('/iw/j,
Tilt! l,iixoinnl citrvn V///4 V/l / '+ ^ulK^O, him nnl in ^ninjrul nny nnctnn v (!impn;
in ilui paHimiliu 1 imnc whom l.hd ^mnul CIUTOM unj firrl<% \vn liavn hownvor wuin how
Urn mlititt I : m \ tt, nmy hi) iltitisnninwil MU Mint thn HIUVU Mltull uctpiim a notlo, (i\vo
nixli'H, or a miHjt; vix,,, n^itrdin^ ti, h, r. IIH iwmmt uivul fiinnliiiut^H, wu havn lutrn n
 'I,'!' M ^0, Uui lortiH ul' i.lu.i (H'liUvH tif tlu; viiriahli! ninih), anil Lho Holution
it ti (.*
on (.'HtublwhiiiR a rnlation hoiwrun UHH <:ituio luid tlm nrLhuLoiiiui oinilis or Jnnobian
of Liu) limits givmi oiniluH. X 1m vo in my pitpur " InvtHtixutiuiw in (uimiuaUnn with
Ciwwy'H K.mution," Quirt. Mulh. Jnur. vol. vnr. (1HOT), pp. HIM IM.2, [:)()] ^ivon, uftor
, a xtilution tf tho xni.'ml (piittitiuu in Htitl Lho nuiubur of Lho curvtw
O, whitih liavu a cimp, or \\hiith Imvn twct itodnM, and I will horn
ioprocluoo thu loiulinx jiuinUi ttf thn mvun%niiim. I ivumrh, that allliouyh ouo of tho
loci involved in it is tho Hivinu iw Umfc oceumiijf in Lhu wwo of tho throo oiroluu
<vix,, wo Imvo in wutli ctiaa this Jnoohiiui of tho givon ciirvw), fchu other two looi
S and A, which prutMiil thomaolvoa, nuoin to Imvo no voliUiuii to fclio ctmio <ti" c
which ia nmdo uso of in tho pavtioulw caso.
,T T *
f>7() ON I'nl
\Vn havo Urn curves V I), I' 0, II' I), niih .. I)M' ,,,,. ,.,,(,., r , U1(r j
Hidwiiitf a pninl. tlm i nlimilrx wlmrml 1 mv (/, m, ). ur r,i;anl . ( ,, i, ',ii..ntlin L r i,,
lliiK poinli Urn niirvtt V//AI ^m I'  v'/< II" (I, ),,iy l..r nti.. m. , *. ih, t . mv ., (i  lrilu ,' ,,.,
uliovo a liiirvi! u' Urn nn ( >r :!r, linvin ? { ; ; ruuiu.h, mil, ...,), ,, ,., nvi (l ,u'ivi' .'
any iiodn, and in nrdcr Dial, thi't mm' miiy lm\.< n nu.tr, u i ., , ,, ; ., MV ,!(.,, ( j
to Htm, lii mi Uic mirvi' ./, (hi Jan.Iiiiui <.)'  ( > i n ,.,,
i/ and A will t'uin!Miniid In ciuili MIT puiu! tu i (l ,i,ii
puinti wlialii'Vt'i mi llir niivi 1 A, (In dim' ,ti will Imv
of ,/; and rnnvi'i'ii'ly, in nidrr tJial, UK. n
Tim (itirvtt A IIIIH, Imwi'ViT, iiMih<ri and riij.n, ; ( n  t i,..,, ,,3 \
nl' ,/, vi'/., llir (/, JK. ) al, a iindt. i.C ;\. i,,, ,nn, If p, tt I.hn.i.a
1 1
, .,!.
"
.,,, M im ., ,.
A. HUH ' n ,i Wllir ( . n . llllim ll , ) . 11Hli , ll ,  m , , (( . n ^ MMft
A inlu rnmM'x '
nl. IN ihafi lur Um ,'imv ih \
" ' '""i'"'!
,,., , , , m<1H  ,
418
415.
rnllltKCTKiNS AND ADDITIONS TO TIIK MKMOIH ON THK
TIIKtillY OK UKl'IIMKX'Ah SUHKAOKS (Phil. .'Avow. vol. cux,
IHttll, ll I ).
n /*AtVt,v*yt/nVfi/ Vniii.v*fi7iW i J M/* //n 1 MJ/<I/ NfictWi/ i;/' London, vol. tn.xil, (fur tho
ynir iNYli). tp, MM HV. Knvi\vrl July i!i!,Ht'inl Nnvtunlinr Hi, 1H7I,
I, ! AM iiuMtMl hi l)r Xt'tilliKii Ilir ill' 1 iviuurli: I,luil. ull.lmuj,(I) ilin " tilVpoinlH 11
iitiil "nil' jtlitMf'V '' fVliiiil iii tile iiit'iiM'ii', am rt'itl hin^iihiriUt'w, Ihoy urn nul, Iliu
Midgiilniiiit''! In uliiili Ilif if', /' lit 1 tin 1 Itn'imilm ri'lnr, Tin) IIIOHI, couvtuiitMili \vny of
i'liiTfcfiuj,; (Kin i 1 * I" Hiiiin nil lln< titnniihi* with /?, /)' UH tlny uliuitl, hul< l,u wri
in, w* IMV (In* nniuhiT tii "Mll'i'"iiil!i" iui"l "nlVplniH'H 11 niHjn'rtivnly; vix, u  t> tlnm IH
ninl
M', (iiri!iini',
^', uiixplniiii'il nii
ihc litriiniltM UN thry Nhiml, tnkinx ntrtiiint !' flu; iiiirxplitiiiitd Hiii^uluntiDM itml /"
hul, nut; tnktnx' ">" iicntmil nl iill !' tin* ullHiiulH untl ufl'plaucH o>, r./. Tlio
in whii'h iht'M' n lukiii ini" wciunt niv:
i? (n  3) Her h *$ + 7 I' (? H w,
CD  2) (  M) a (S f ."  :iw) I M (HO  Mff  *  :f&>) + ii
if, ( g) (  ;i) 4JL 4' (6  2p  j) + '1
c (  3) (  MJ OA + (o 3tr  x " ilw ) *!" 2
which roploco fialnntn'ii nrixinnl fomuilw (A) und (B).
C. VI,
H AND AIUUTHINM Til Till; MKMOIK
2. In Mm
'''"' lh " " ......... ...... ...... 'ml
=
" '"".V
 .. ( . .,.,,,). ..,. ,, 7 ,
('*l>~M('fl)(H..:\).. ( R..<!.. i ..... , Ht . MA . ,..,,
inn! Hiili.HriMiriu K th.Mfn in n 1 : <ii,i . n /. M, i
"' "<" H J ('// I U!, , I/,' , H/ ( i M,, ,  V
H/ ..SA I 1N/7 ! ; v , .j,'. (i,
"^.'.'i/f...M;
mi icnii iit r, fl
....... '' .....
 ' .,, .,,,,, 7 .. a,
i)'"' HA ...'IhV..,^ ; ,, 7 ,
^ 
aiS! (*//, f , 4 * ..//
.
415]' ON THE TH130BY Off 2U50IPKOOAL SUBFAOMS. 579
Tho reciprocal of tho first of those ia "
a' =  n + '  2/  3^'  26"  4/f  3a>' ;
vk writing a = u(n 1)  26 3o, ami 3(^ 2) 06 80, tins in
<T'  'in (  2)  86  llo  2y"  S^'  26" , Mi'  8<a' ;
and ifc thun appears that tho order a' of the apiiiodo OULVO w roduood by IJ for enuh
offpiano a)',
4. AH fco tho otliov two o<uationH, writing for. p, a lihoiv valuer, bluwo bocomo
JK10+3Z + 5/3467 = ^(2^ 4) 23,
2^ + 3<o h 4i + J H/9  7 = ( j (fin  1 2)  I 8 0,
o(iiuiti(m wluoh admit of u gtiomolritsal inturpi'oliatioii. In fiiot, whou tlunx3 is only n
nodal cnrvo, tho fiwt oqnabi(H) in
j + .= ft (2 4)2</,
which wo may verify when tho nodal ourvo IK a oomploto inliuiHcotion, Jf'^0, Q0;
tor if fclio otiuution ol 1 tho mirAioo is (A, ]j t 0$2> t Q)' J = 0, whc;ru Uio do^rooH of
A, Ji, 0, 1\ Q (mi w2/ nffj, n$f/,f 9 g rcwpeotivoly, then tlui piiiwhpoiutH iiru
givon by tho equatioiw P0, ^ = 0, y!.6 v ;i j = 0, IUK! blio nuinbor ,/ of piiiohpttiutH
i/  2) ;
but for tho cnrvo ./ J =0, Q wo havo 0, and i(,H ordor mid O!OHH aro
tf"/^/ (/!//" 2), ov tho formula JH thiiH vuriiiod.
Hiniilivi'ly, whou bhoro i only a ouHpidtil ourvo, tho Koenml oquabinn IH
2 X H !l w o (r, H ~ 12)  (h I 30,
wluoh may bo vorifiod wlion tho oiiHpidul ourvo IH a nomplobo intornon(iion, J J 0,
tho oquation of fcho Hurfnco w lioro (vl, J?, OJP, (3J 9 = 0, whoro /lf?..// 9 Mj ,
and tho pointH ^> "' givon IIH tho iiitoraocfciouH of tho curvo with bho aurfaou
(A, X,
Now AO& vanishing for P 0, Q wo mufifc Imvo ^1 A 3  yl/, 7
A^ + G', whoro ^.', .//, 6 V vanih for P = 0, Q = 0; and thouoo M = Ajlf I Jlf
' + JT, whoro jlf, JV" vanish for P0, Q0, Tho uquabiou
writing thuroin P = t Q = 0, bhua booomoH A 3 (jY'  jlf fly ** Q ; and its itiboracotionH with
the ourvo P0, Q0 aro tho points P0 > Q=.0 > A0 oaoh bhroo thnos, and bho
points ^ = 0, (3 = 0, JTajfa0 oaoh twice; viz. bhoy are bho pointH 2x + 3w.
But if bho dogroo of A is X, bhon bho degrees of JV", j)/' s a a , a/3, ^ aro 2?i  3/ 2r?  X,
2n2/8^X, n 2/X, n~fg\ t n %g~\ whouoo bho dogroo of A a (N'a
is ^SwG/'G^ aiid tho number of points ia =*/? (6?i  (i/ G.</), vix. this ia
or it is =fl(5)t12)Gr; so that being 0, the equation is verified.
7S 2
580 CORRECTIONS AND ADDITIONS TO THE MEMOIR [415
5. It was also pointed out to me by Dr Zeuthen that in the value of 24tf given
in No. 10 the term involving % should be 6^; instead of + 6%, and that in consequence
the coefficients of % are erroneous in several others of the formulas. Correcting those,
and at the same time introducing the terms in , and writing down also the
in 6 as they stand, we have
4i =... 2# + 30 So),
m =... Q x + 90 9w,
2o = ...  e~ to,
8/j = ...+ GX 90 + Qai,
c' = ...% + 100 20u.
The equations of No. 11, used afterwards, No. 53, should thus be
** + 6r = ( on  12) c  18/3  5 7  2# + 30  3o>,
 24 89 + 18r = ( 8w + 16) & + (lfi  36) o 84/3 + 9y + 4j ~ 6^+ 90 9
and from these I deduce
6. In No. 32 we have (without alteration) = 16; but in the application (Nos. 40
and 41) to the surface J?P* + G&Q* = Q we have = 0, and there are ^~f n offpoiute,
F=0, P = 0, Q = 0, and x = gpq closepoints, Q = Q t P = Q } Q^Q, The new oquatimin
involving w are thus satisfied.
7. I have ascertained that the value of {3' obtained, Nos. ,51 to 64 of the memoir
is inconsistent with that obtained in the "Addition" by consideration of the duficusimy',
and that it is in fact incorrect. The reason is that, although, as stated No 53 the'
values of two of the coefficients D, E may be assumed at pleasure, they cannot in
conjunction with a given system of values of A t B, 0, be thus assumed at plotvmra:
viz. A, B being =110, 272, 44 respectively, the values of D t E are really deter
minate. I have no direct investigation, but by working back from the formula in fcho
Addition I find that we must have D = ^ = 315; the values of the remaining
coefficients then are "
.,  , .
or the formula is
 hfff  g'B'  M  \y _ M y _ v < & _y v .
butj^have not as yet any means of determining the coefficients /, /' of the torma
415] ON THE THEORY OF RECIPROCAL SUBFA.CES, 581
From the several cases of a cubic surface we obtain as in the memoir; "but
applying to the same surfaces the reciprocal equation for jS, instead of the results of
the memoir, we find
h 1  4,
t/' + lQv =198,
</' + 2/A = 45,
g + (/' = 18,
X =5
(so that now X + X' =  2, as is also given by the cubic scroll), And combining the
two seta of results, we have
h
 24,
X
= 5,
/*
= ^~
V
= 3f 
h'
= 4j
9'
= 18
X'
___ 7
hut the coeffioionts (/, ft', *', /, /' are still undetermined. To make the result agree
with that of the Addition, I assume = 8G, a/=l, ,9 = 428; whence we have
 24(7
4 40'
and if we substitute herein the foregoing value of 44</ 4 ^ >', we obtain
4 ( 93 4 252) o
4158^4937 + 66;
24(7 285 ^27j33x +'
+ 4(7 + 103' + ;'+ 7/+ 8 X '
which, except as to the terms in to, en', the coefficients
agrees with the value given in the Addition,
Dr Zeuthen considers that in general i' = i ; 1 1 pref
verified it,
416.
ON THE THEORY OF RECIPROCAL SURFACES.
rt as an Addition by Prof. OAYLBT in Dr SALMON'S Treatise on the Analytic
becmelnj of lime Dimmsiom, 4th Ed. (8vo. Dublin, 1S82), pp. 592601.]
620. IN further developing the theory of reciprocal surfaces it has been found
necessary o take account of other singularities, some of which are as yet only imperfectly
to g!re * ***** ** "i <
n, order of the surface.
' a, order of the tangent cone drawn from any point to the surface.
S, number of nodal edges of the cone.
, number of its cuspidal edges.
p, class of nodal torse.
o, class of cuspidal torse.
b, order of nodal curve.
Je t number of its apparent double points.
/, number of its actual double points.
*, number of its triple points,
j, number of its pinchpoints,
g, its class.
c, order of cuspidal curve.
h, number of its apparent double points.
416] ON THE THEORY OF RECIPROCAL SURFACES. 583
6, number of its points of an unexplained singularity,
X, number of its closepoints.
a, number of its offpoints,
r, its class.
/3, number of intersections of nodal and cuspidal curves, stationary points on cuspidal
curve.
7, number of intersections, stationary points on nodal curve.
i, number of intersections, not stationary points on either curve.
G t number of cnicnodes of surface.
B, number of binodes.
And corresponding reciprocally to these :
n't class of surface.
ft', class of section by arbitrary plane.
8', number of double tangents of section.
', number of its inflexions.
p', order of nodecouple curve.
a', order of spinode curve.
b', class of nodecouple torse,
k', number of its apparent double planes.
/', number of its actual double planes.
if, number of its triple planes.
/, number of its pinchplanes.
q', its order.
0', class of spinode torse,
h', number of its apparent double planes.
$', number of its planes of a certain unexplained singularity,
^ t number of its closeplanes.
to', number of its offplanes.
r', its order.
/3', number of common planes of nodecouple and spinode torse, stationary planes of
spinode torse.
7', number of common planes, stationary planes of nodecouple tor
i', number of common planes, not stationary planea of either torse
O' t number of cnictropes of surface.
B' } number of its bitropes,
In all 46 quantities,
584 03S T THE THEORY OF RECIPROCAL SURFACES. [416
621. In part explanation, observe that the definitions of p and <r agree with thow
given, Art. 609: the nodal torse is the torse enveloped by the tangent pianos along
the nodal curve ; if the nodal curve meets the curve of contact a, then a tangout plant!
of the nodal torse passes through the arbitrary point, that is, p will bo fcho uumbur
of these planes which pass through the arbitrary point, viz. the class of tho torse. So
also the cuspidal torse is the torse enveloped by the tangent planes along fcho cuspidal
curve ; and a will be the number of these tangent planes which pass through tho
arbitrary point, viz. it will be the class of the torse. Again, as regards p' and a' : tho
nodecouple torse is the envelope of the bitangent planes of tho surface, and l;ho node
couple curve is the locus of the points of contact of theso planes; similarly, tho
spinode torse is the envelope of the parabolic planes of the surface, and tho Hpinodo
curve is the locus of the points of contact of these planes; vi 1 /,. it is tho ourvo UIl
of intersection of the surface and its Hessian ; the two curves aro tho rociiprooalu of
the nodal and cuspidal torses respectively, and the definition!* of p', a' correspond to
those of p and o%
622. In regard to the nodal curve 6, we consider k tho number of its apparent
double points (excluding actual double points); / the number of its actual double
points (each of these is a point of contact of two sheets of the surface, and fchoro in
thus at the point a single tangent plane, viz. this is a plane /', and wo thus linvo
/'=/); * the number of its triple points; and j the number of its pinchpoints
these last are not singular points of the nodal curve per se, but aro singular in regard
to the curve as nodal curve of the surface; viz. a pinchpoint is a point at which
the ^ two tangent planes are coincident. The curve is considered as noli having any
stationary points other than the points 7, which lie also on tho cuspidal ourvo; mid
the expression for the class consequently is q = tyl 2/c 2/ $y Qt.
G23. In regard to the cuspidal curve c we consider A tho number of its 
double points; and upon the curve, not singular points in regard to tho ourvo per so,
but only in regard to it as cuspidal curve of the surface, certain points in number
0, x, <o respectively. The curve is considered as not. having any actual doublo or other
multiple points, and as not having any stationary points except tho points (3, wliich
he also on the nodal curve; and thus the expression for tho class is r = c 3  c  2/t  rt/3.
624. The points 7 are points where the cuspidal curve with tho two ahootH (or
say rather halfsheets) belonging to it are intersected by another shoot of tho surface;
the curve of intersection with such other sheet belonging to tho nodal oiirve of bho
surface has evidently a stationary (cuspidal) point at the point of intersection.
As to the points ft to facilitate the conception, imagine tho cuspidal ourvo to bo
a semicubical parabola, and the nodal curve a right line (not in tho piano of tho
curve) passing through the cusp; then intersecting the two ourves by a series of
parallel planes any plane ^ which is, say, above the cusp, meets the parabola in two
rea points and the line in one real point, and tho section of the surface is a ourvo
with two real cusps and a real node; as the piano approaches the ousp, these
4lfi" ON' TIIM TUK011Y OV K1SOI l.'UOOAFj HTJIU'ACJISS, 585
approach tn^oUior, and, \vhim tin 1 , piano pawoM through the nun]t, unite into n Hingular
poinl, in the imknro of. 1 a triple point; (node  two <SUHJ)H) ; anil when tho plnno pannes
below llio (!iwp, fclui t.wo (IUHJIH of tho Hookian boeomo imaginary, and (ilia nodal lino
on iVoni minodal to mmodul.
(i2i"i. At it  mink i bho initial oiiivo OI'OHMOH tlin miHpidul onrvo, boing on fcho nido
awuy from Llio two liiUfHlinolH of klus mii'laco inniodnl, iviul on tliti Hides of kho two
liiiH'Hhi.!Ol,n nvuiiodal, vix. lihn two liallHluiotn iutoi'Hiiiik oatili oklior iiloag bliin portion of
tlm nodal oiii'vu. Thorn in ub Uio point a mngln tun^ont platiOj whioh IH a plains i' ';
and we tlnw have ijt'.
(iiit). AH uliTiuly inoiitinnrd, a cnionodo (J in a point wlinro, iimtciul of a
lilium, wo have u tiinj^mli iimdriroii(9 ; mid itl; a binodti H lilio ipiiulricono (liig
inko a pair of plunnH, A. enii;kro])i! (' in a pliinn toiuihin^ tho HurJiino alcm^ a oonio;
in klm oivmi of u bitropo l t tlus donio dogonnmU^ into a flab omiic or }iir of points,
(Iii7. In llii! original Ibrmuhu for tt, (n~ a), ft(?*2), o (  2), wo have hi wiito * It
iimU'iul nf , and tlm fonnnlin aru furthor ivwulillutl by VOIIHOM of (.ho Hin^iiliiritioH
and w. Ho in dim original lorninlm for (/ .(n 2)(n JJ), />(< 2) (it 8), ( 2) ( '1),
wo have iiiHltnwl o(' R to wrikn fi  (V  MM ; uwl to milwliilnito now (ixjutJHHionH for
[((/], [f/fi], [lit)], v\'/.. lihemt aru
ltib'\ ~nb %p~j,
[ltd] s= rfo Jicr ^ ,
Tho whole HorirH of ((iiat.ioim thiiH in
(]) a'^M/.
(3) /'/
(!i) '*.
(4) (* u (tt 1)~ 26 Mo,
(5) '.( li) lift Ho.
(It) S'  J?i (n  2) (H U  i)  (n a  3t  0) (26 I lie) + 26 (6  1) I (\bo I I!G (o  1).
(7) tt( 2) ,// H ^ H 2<r + Hra.
(H) &(w2) p + SS/3 + S7'l^
(0) o(M2) 2(r h 4/9 h v + ^lw.
(10) (t (n 2) (  II) 2 (B   w) I Jl (c  .V  %  flw) I 2 (al>  2/>  j ).
(11) 6(H2)(H~n)*U h (62p'/ ) + a(6o8^27t>
(12) c (n  2) (  ) = /t I (c  3ff  # " 3ffl ) I 2 (60  30  2?  0
74.
C. VT, ' A
586 ON THE THEORY OF RECIPROCAL SURFACES. [410
Also, reciprocal to these
(15) '= ?>'('!) 26' So'.
(16) * = 3'(tt'2)6&'8o'.
(17) S = ' ('  2) (''  9)  ('  ri  6) (26' + 3o') + 26' (6'  1) + M'o' + jjo' (o'  1).
(18) a'('2)~*'.B' + / 3' + 2ff' + 8iu'.
(19) &'(' 2)= p' + 2/9' + 37'+3i'.
(20) c' ('  2) = 2<r 7 + 4/3' + 7 ' + 5' + w '.
(21) ffl ' ('  2) ('  3J = 2 (8'  f?'  SM') + 3 (aV  3d'  #'  Sw') + 2 ('&'  2/a'  j').
(22) 6' ( B '  2) ('  3) = ^ + (a'6'  2^' /) + 3 (6V  3/3'  2y  ")
(23) c'(n'2)('3)= 6A' + (aV3tf' x /  3') + 2 (6V ~ 3/9' 87' O
(24) r/ = 6'^  6'  2tf  2/'  3y  Gi'.
(25) j' = c' a c' 2^3^9',
together with one other independent relation, in all 26 relations between tho 46 quantities.
G28. The new relation may be presented under several difforout forms, oqmvalonl;
to each other in virtue of the foregoing 25 relations; these are
(26) 2(l)(n2)
(27) 26^I2c4(7
in each of which two equations S is used to denote the same function of tho accontod
letters that the lefthand side is of the unaccented letters.
( 2S ) /3' + 0'= 2(ft2)(lln24)
+ ( 66tt + 184) &
 24(7  285  27j  33^
r 11 + 4 ^' + 10^'+ r? v + 8v'
Or, reciprocally, v T
+ (98w' + 262)o /
+ 22 (2^ + sy + 30
+ 27(4)8'+ y + ^
'  285'  27 
ON TIIK THKOUY OK HKOHMlOOAf, HUHFAOKB.
587
inj.;' i!(iuitinii (S>(i) iu (mil, oxpi'OHHtm blind (Jin sui'limo ami iln rooiprminl Imvn
Hiinio tlnl'minncy ; vi/. l;lm oxproHHimi for trho dulioiuiwy in
(ItO) ;i Mlcimiuy  \ (,,  I )(  2)(w  I))  ( ~ H)(6 I o) I ~i (f/ I r) + 21 H 5/9 I 57 I 1  fttf,
(^7
) in a mttil.ioti
for ilrH vahiti n(n
(i
I 2
'2
. Tilts n(uiil.inii (2S) (diitr do Prof, Oiyltsy) in Uus nmTiMil; Itirin of mt
HiMi; obliiincd liy him (widh wnnti isrnirn in dh tiuiuorutal oonil
, linl; whirli in btuil. obluimsd by niciutM of l;ho injuiilJtHi
ilwill 1 in tint invtiMligaUon. lu liml; t cuimidnrin^
i;to ( \vd lutvu from (Jui liivdi 2fi ttipiiitioim
_.. v
i; K.  1 ~
ninl ntiill'IplyiiiK Utcw* i!(iutliiiiiH Uy l\w ituinbnm wil niwiLo (in Uioin ii
tuliHii 1 , wt liinl
IVoin
(2tt) ;
rt UH
, unct
itn
1lioi'i'tn (2(1) wn Imvc !l* ('iiuil.iiiii (2V); iuid from thin (2H). or by a liko
, (2H). w olthiiiicil witliuuli inudi ililliimlty. AH Ln Llin H SJtjquivUuiiH or HyiiiiimbrioH,
' Uiul. (lie lir.M., ihin], loiirtJi, and litlh iu' in I'mit iimlurliMl tunoiix llin original
(lii itti t'XnvHMJnii wliioli vuniHlirs in in llust il); wn htivn I'nun Minin
Ko^M' ..... ', and llii'imi' !t;t (J  < Ma' ', wlvich in , or wn hnvi*
lliiiM lilt! Mimind nimtiim; but, tlio Mixl.lt, wvuiitli, and uiglitli ttquuLiniiH Iiavn yt to
IIP cibtiiiiuid.
liW), 'IMm nnn.ion (!">). ("'). (17) tfivo
ftwm (7). (), ()
(a ft
(a  2ft
2) (K fl)  ('* a  "  ) (28 I H) H 28 (S r '!) + 8* 1 U (*  1) J
liuvu
)(B) Jtt fi/3 4y.8iflH.aM.
w  2) (  M) 2 (fi  0')  Hi'  1HA ~ flfto + 1/9 H I2y i Gt  Ow,
742
588 ON THE THEORY OF RECIPROCAL SURFACES, [41(5
and substituting these values for K and S, and for a its value =n(n ~ 1) 26 3c
we obtain the values of ri, c', V ; viz, the value of ri is
ri = n (n  I) 3  n (76 + 12c) + 46 a + S& f 9c a + 1 oc
 8k  ISA, + 180 + 12? + 12t  9rf
223530.
Observe that the effect of a cnicnode G is to reduce the class by 2, and that of n
biuode B to reduce it by 3.
631. We have
(n  2) (n ~ 3) = H S  n + ( 4n + G) = a + 26 + 3o + ( 4?i + 6),
and making this substitution in the equations (10), (11), (12), which contain (2) (ji8)
these become
6 ( 4n + 6) = 4i  26 2  9/9  67  Si ~ 2 p j,
c(fe + 6) = 6/i3c a 6^4 7 2t~3 t r~ % ~Sa) )
(the foregoing equations (0) Salmon p. 536); and adding to each equation four time*
the corresponding equation w:th the factor ( B 2), these become
Writing in the first of .
means of the values of "* ^ ^ Othm '
3r + c
which give at once the last three of the 8 Sequationa.
The reciprocal of the first of these is
'o + * / 2/8 x '20'4B'3',
or writing herein a = ^i)_ 26 _ 3c aud ^(nqUBo, thia is
2)  86  llo  2/  3^ _ \ < . 451
ON TIM'l TIIIOOIIY OK KI'lOUMUtUAL HlWtfAdM. 58D
(IN2, IiiHlniul (if nbtainhitf the, mwnml and third ripmtintiN an above, wo nmy to
ihn vn.hn) nf /> ( <\<)i. I (i) add Uvi(! thn vulini of 6( 2); und to liwiuo bhu vtiluo of
o ( >ln. ( (i) mid tlirno l.hncH tho vahtn of o (u 2), tlnin obtaining <!([iiuU(niH frets from
{) and <T ri'Hpnsf.ivnly ; UIUHU uptatioiiH an;
ft ( 2u I a) :* 4/,:  2// 1  fiyj/  III I W ,7,
( fin I (J)  m  ({' J  fi 7  <li  % I M w,
wliitth, iiiti'oduning lihornin l;ho valiutH of (/ and r, may also bu wribtuu
(i (f/  J 1 2) I Hfl ^ (ir h 1 H/3 I 5 7 h 'h' I 2^ I Sw.
iii^ IIH /(ivnn, llm nrdrr of Mm mirlmn! ; Mio iiuilal mirvn with Jin mnjj
h, A',/, /,; l.lio I'.iinpidid cui'vn mid its niiif{ulanUi's <\ h\ uuil iJit! unauliilidH /if, y, ?! wliiuh
ri'latn in tin! inl.mwd.iiniH of thr nodid tuid tsuspidal C.III'VUH; thn lirnl, of lihi! t.wo
rijiial.iunti I^'IYIM /', I, ln i iniiiihin 1 nf pinchpoints, Imin^ Hingnlnrilitis ol' lihn nodal curve
tpinad Uin mirlHcn ; and Ihn Hrnnnd ciiualinii I'Hhililwlmn a viiltttion ImUviuni 0, ^ ( w, thn
nf Miujjiiliu 1 pDinlii nf tint c,nnpidal c.urvd tpioiul thn HUi'liKU).
In thn ram 1 of a nodal r.urvn only, if thin tin a mmpldiM intiTHi'iiiioii ./'(), (^ 0,
(Jin i'lpiatii'ii n' l,lu' mirl'an w (A, tt, f'O/', Q} 1 <(), and l.lin lirnl. npiaiion iw
in 1 , iiH'tiiining I i), nay j $(n ... l)/i 2// 4 "W 1 , wliich may Iw vrrilird ; ami HO in Iho
rn.'M' nf a riifipiilal tMirvr niily, wlit'ii t.liifi in a rnniplrtu itiiiuwt'tion V :(}, f j *.: 0, l\w
r.pmlinn nf UK Hinfaw in (A, li.HQr, <$*>(), wlinm A<! H^.tiU* ] NQ\ and Li to
^rrond ripial.inii in
,1 (. . fin I (1) r. = 1 2A  liu' J  2% I :U?  Hw,
nr. miy
^1  Mot ; '.(r.H. })(i.G g lJ2yi .IJW,
whitli may II!MH 1m vnrilii'tl.
liMII. \\*<! may in thn 11 ml innlanir mil, of Ihn Hi tpmnl.itiw cimmdoi 1 UH givon
ihn !! tintnli[it'ft
; /', A,/, t\ o, A, /', x ; ft 7. ' : : '' ^i
ilim nf tin! ail ntlatiiuiH, 17 iH.Tinini! tlio 17 fpmiitities
u, fi, /c, p, ff\ j, 7 ; r, w 1
';*', S', *' J A'./ i a' : '.
and thciv rrniitin thir {) u'pHttinim
(IK), (I!)), (20), (SI), (22), (US), (24), (25), (2H),
eliH ihti la ipiiiutitinrt
^ cr'j Jf, t', j', v'; /', '. x. '. '*'; j8'. 7'; ^ ^.
590 ON THE THEOHY OP RECIPROCAL SURFACES. [416
Taking then further as given the 5 quantities j', x ', a', G f , B',
equations (18) and (21) give p', <?',
equation (19) gi ves 2/9' + 3 7 ' + &',
( 2 ) 4/9'+ 7'+ ff t
( 2S ) /9' + i0',
sojhat taking also t' as given, these last three equations determine /3' ; y, 0'; and
equation (22) gi vea ',
(23) /,',
" (2*) </,
(25) M r ',
viz. taking as given iu all 20 quantities, the remaining 26 will be determined.
o = (!),
S = J w ( W  l) ( ?i  2) (  3),
a=
' = 3 (n ~ 2),
6' = fa ( n _ i) ( w _ g) (,  n 8 +  12),
^ = ^ (M _ 2)(? p^ G , B + 16? ^^^ 164w ^ miB + 647; ^ io ^
t ^ (7!  2) (r _
 2) (IGn*  G4m 8 + SOtf  lQ8n + 156),
2) (ft  3) ( 
the remainiag quantities vanishing.
I Hi I ON TIIH TIIKOUV OK HK(?I IMKJOA li HUlirAOKH. 5!M
fittn. The (ii(>sl,ioii of mn^nlaritie,4 lian bnon oinmiilernd inulor 11. nmns ^onoval point,
nf view liy /million, in Mio memoir "Kiwhnrdlit! den HititjuliivHiJs ipii null mpjioi'li i\ mu!
druHo nndliple d'uim mirlime," Mtilh. Amittlcn, (,. iv. ]>p. I 20, 'IH7I, Vfe nUiibnUw l,u
A ninnbm 1 of Hiiigulur pnini,n, vix, points nl. any oun of whidi Uin iiui^nnlH form
a rcHin nf tin 1 (H'ddi /t, inn) nliiHH M, wil.li // ] 1) douhlii liimn, nl' \vhi(;li ;/ iLrn hint((?nls
In tinuinlinH ill' Miii nndal rui'vc Lliroiitfli Mio puinl,, and ^ !," Hlialiuiinry lines, wlidntnl'
^ MIX tinij^i'iilH lit Itriinolicn ill' UHI i^mpidal diirvo lhroii^h Llm puini, and wiMi . ilnnUIn
plani'H and '/t Mlaliiuiniry pliincs; innntovitr, Mnw poinl.H liiivo unly (Jut prnptti'i'.inn wliioli
ui'i' .lni IMOH(. KViirml in Ilin CIINII nf a Hin'l'urr n^'tirdifd nn u Inimn tH' puinis; and ^
iliwiilt'ti H fiiun I'xl.i'iiditijif i.o all mid i iiiiul.H. Thn Inn^iiiiff jfi'iinnil didiniliimi indndivH llit
i'lliMiiiidiiH (/, i l' ; ; tl, )/ < ; i) :;;'. t,*, !//.: 71 = = ()), iviill 'alnn, bill, nod prtijiul'ly] Hut ItinndiiH (ft, "'%,
i; ' I, '','/ "fen. : ()), [id innlinlis ulsn (,iu> {^tVpniuis (/t ^ = c !J, j  n I , //. sij:i (^ 0>',j
And, liirUinr, ti innnhiT nl' nin^ulfi.1' jilanrH, vix. jilaiuH any mm of \\liinli
a riirvn ol' the cliiHM ^ uiul nrdnr i/, wil.li //*!?/ dnnlilcs (.itii^tmtH, ul' whinh //'
ure ^I'lii'riil.in^ HIICM of .ln< iiiiiltM'unplii lomi!, ,o'l f' ntul.innary tan^t'iil.M, of wluiili c' arc
!joui'Vn!.iii,' liui'M u' iho Mpinuilit forno, u' duulitn poinln und )/ eunpn ; ii, \n t inorenvrr,
uiipjui^i'd l.liiil, llu'si' plnin'M Inivn only ( li (1 jinipriiit!^ wluiih are 1. 1 in IUOH). Lfunenil in
1 lln riiHf nf a HiirfiKin rr.;jinlcil UM nit eiivi'lnpu u' JM liiiij^cul, planen ; nml ^' iltMinl'prt
a. Kiilit Msileiidiliff in all Hiirh plannn. ('IMm ilH'mUiuii innlutlcM lln^ oiuolFropPH (//   v i!,
// ; i/ ' ; g' ; ! (,'' ; it'.. ii' 0), and julHM, hu(, not, propi'vly'l l,hn ItiirnpnH (;t' :  
i' 1 i/'^fer.. . (I), fit. inrltidi'N ulno MM; (itVplniK's (fi. 1 t>' =.'t, ;;' ; ^/^"l, >/' /ri T
ll.'lti. ThiH Inill^ HO, uinl wriditix
flic dpiiLUniiM (7), (H), {!)}, (10), (11), (liJ), cnutaiii in ivnpur.1, ol' ilm nt;w
uilditiiKiul liirniH, vix. Uiisr urn
/*( a) :.... i s [//(/ ay).
2)(n
aiitl lUartt avc nf wmwc Mm reciprocal tcnnn in Llm reripronil eipiutioiiH (1H), (I!)), (20),
(21). t^iJ), (:0. Thenu loviiiuliu art! tfivm willitml. ilitinniiHirutiitn in Um nioiium' jnnl
rofiarrwcl Ut: Un principal nltjiict of tlui mnmmr, tw Hhown liy iln l:iLl(.', i tlw roiiHuler
uliuil mil nf Klieh Kingilliir puhitH untl ]tliilHjA, hill "if l.lle, umlti]ilu i\H\\\, HHOH of n
Murfiicc; nml tu \v^iw\ tit thuwf, bhu muinnir Mlnmld bn
NOTES AND REFEBENCES.
384, THE conclusion arrived at Nos. 27 30 that the transformed curve of the
order D + l depends upon 4D G parameters is at variance with Kiemann's theorem
according to which the number of parameters is 3p3, (p Kiemann D Cayley), = 3133,
and this last is the correct value, My erroneous conclusion is referred to 111 the
preface to Clebsch and Gordan's Theorie der Abel'schen Jfimctwnen (Leipzig, 1806),
"Unter den von Riemaim behandelteu Theilen der Theorie haben wir die Frago imch
der Anzahl der Moduln einer Klasse von Abel'soheii Funcbiouen ausschliessen m mltsson
geglaubt. Diese Frage ist durch die achaifsinnigen Betraclitungon des Herni Cayley
Gegenstand der Controverse geworden : sie ist iiberhaupb wohl zmiftohst iwr durch tiofe
algebraische Untersuchungen endgilltig zu entscheidon, fllr deren Sohwioiigkeifcen die gogcn
waitig bekannten Methoden nicht mehr ansaureiohen scheinon," In the case D(or 2>) 3
my value is 10, Riemaun's is 9 : that the latter is correct was shown by a direct
proof in the paper Brill, "Note beziiglich der Zahl der Moduln einer Ktossu von
algebraischen Gleichungen," Math. Ann., t. i. (1860), pp. 401406 : the explanation of
my error is given in the paper, Cayley, "Note on the Theory of Invariants," Math.
Ann., t. in. (1871), pp. 268271.
400, The question here considered, viz,, the expression of a binary sexfcic f in
the form u s  w 3 , and a cubic and a qimdric respectively, forms the basis of
the very interesting investigations contained in the Memoir, Clebsch "Zar Theorie
der binSren Formen seohster Ordnung und zur Dreitheiluug der hyperelliptisohoii
Funotionen," Gdtt. Alii,, t. xiv, (1869), pp. 1 6&. Considering / as a given sextic it is
remarked that the number of solutions, or wliat is the same thing the numb
the functions u or y, although at first sight = 45, is really = 40 ; supposing tha*
is a given solution it, v, or that the sextic function is in tho first instance gi
the form ^w", then if any other solution is u', v', we have v 3 w 8 = v' 3  it' 3 , wiiere
v f , u' are functions to he determined : there are in all 39 solutions, a set of 27 and a set
of 12 solutions; via. writing the equation in bhe form (v+if)(v~i/)=(uu!)(uev!)(u eV),
e an imaginary cube root of unity, then either the v + v' and the vif contain each
of them as a factor one of the quadric functions uuf, u eu', weV (which gives
the set of 27 solutions) or else the v + v r and the vV are each of them the product
of three linear factors of the quadric functions respectively (which gives the set of 12
a vi. 75
594 ' NOTES AND REFERENCES.
solutions). It may be added that the 27 solutions form 9 groups of 3 each and that
these 9 groups depend upon Hesse's equation of the order 9 for the determination of
the inflexions of a cubic curve; and that the 12 solutions are determined by an
equation of the order 12 which is the known resolvent of this order arising from
Hesse's equation and is solved by means of a quarfcic equation with a quadrin variant
= 0. As appears by the title of the memoir, the question 13 connected with that of
the trisection of the hyperelliptic functions,
401, 403, On the subject of Pascal's theorem, see Veronese, " Nuove teoremi sail'
hexagrainmum mysticum," It. Accad, dei Linoei (187u' 77), pp. 761 ; Miss Christine
Ladd (Mrs Franklin), " The Pascal Hexagram," .Amen Math. Jour,, t. II. (1879), pp. 1 12,
and Veronese, "Interpretations ge'ome'triques cle la thdorie des substitutions de n lettres,
particulierement pour n = 3, 4, 5, en relation avec les groupes de 1'Hexagramme Mysti
que," Ann. di Matem., t. XI, 188283, pp. 93 23G. See also Richmond, "A Sym
metrical System of Equations of the Lines on a Cubic Surface which has a Conical
Point," Quart. Math, Jour., t. xxn. (1889), pp. 170179, where the author discusses a
perfectly symmetrical system of the lines on the cubic surface and deduces from them
equations of the lines relating to a Pascal's hexagon : there are of course through the
conical point 6 lines lying on a quadric cone and these by their intersections with the
plane give the sis points of the hexagon : the interest of the paper consists as well
in the connexion established "between the two theories as in the perfectly symmetrical
form given to the equations,
406, 407. A correction was made by Halphen to the fundamental theorem of
Chasles that the number of the conies (X, kZ} is = a/A + j3v, he finds that a diminution
is in some casea required, and thus that the general . form is, Number of conies
(X t 4Z) = a^t + /3y  F : see Halphen's two Notes, Gomptea Rendus, 4 Sep. and 13 Nov.,
1876, t. LXXXIII. pp. 537 and .886, and his papers "Sur la the'orio des caractdristiques
pour les coniques," Proo. Lond. Math. 8oo. t t. ix, (18771878), pp. 149170, and " Sur
les nombres des coniques qui dans un plan satisfont . i\ cinq conditions pvojectives et
inde'pendantes entre elles," Proo. Lond. Math. Sac., t. x. (1878 79), pp, 76 87 : also
Zeuthen's paper "Sur la revision, de la" theVie des caracte'riatiques de M. Study,"
Math. Ann., t, xxxvu. (1890), pp. 461464, where the point is brought out very clearly
and tersely. . ,
The correction rests upon a more complete development of the notion of the
linepairpoint, viz, this degenerate form of conic seems at first sight to depend upon
three parameters only, the two parameters which determine the position of the coincident
lines, and a third parameter which determines the position therein of the coincident
points: but there is really a fourth parameter. (Compare herewith the pointpair, or
indefinitely thin conic, which working with pointcoordinates presents itself in the first
instance as a coincident linepair depending on two parameters only, but which really
depends also on the two; parameters which determine the position therein of the vertices.}
As to the fourth parameter of the linepairpoint the most simple definition is a
metrical one; taking the semiaxes of the degenerate conic to be a and b (a = 0, & = 0)
then we have two positive integers p and q prime to each other such that the ratio
NOTES AND REFERENCES. 595
a? : bv is finite ; and this being so the fractional or it may be integer number p : q
is the fourth parameter iu question. But it is preferable to adopt Halphen's purely
descriptive definition, viz. we consider a conic 1 in reference to three given points
y t z, t on a given line, and take is, ot for the intersections of the conic with the
lino : we take a = (y, z, t, K) (y, 2, t, #') for the difference of the corresponding anhar
monic ratios of the three points with the points #, #' respectively; and 2 we consider
the conic in reference to three given lines Y, Z, T through a given point and take X ) X'
for the tangents from the given point to the conic; we take l(Y, Z, T, X) (Y, Z. T, X')
for the difference of the corresponding anharmonic ratios of the three lines with the
. / Q> jy' 11 t
lines X, X' respectively  observe that these values are a =  . ~, , , and
1 J \ zm.etf z y,tt t
X ~X' Y ~T \
b v v~v v> *" ~r7 \7~n m } Here when the conic is a line pairpoint, to ~ &f and
/j A . Zi A A X , & 1 /
X = X', where a = and & = 0, but we have as before the integers p and q such that
a v : 6'' is finite, and we have thus the fourth parameter p : q.
Halphon's correction is now as follows, starting from the formula number of conies
(X, 4)Z) ttfj,+ @v, we may have among the a/it/3y conies Imepairpoints any one of
which if we disregard altogether the fourth parameter is a conic satisfying the five
conditions, but which unless the fourth parameter thereof has its proper value is an
improper solution of the problem and as such it has to be rejected: if the number
of such solutions is = P, then there is this number to be subtracted, and the formula
becomes, Number of conies (X, 4#) = /* + /9? P.
It may be asked in what way the fourth parameter comes into the question at
all: as an illustration suppose that a, l> denoting the semiaxes of a conic, or else the
above mentioned descriptively denned quantities, then p, q, Is denoting given quantities
(p and q positive integers prime to each other) the condition X may be that the
conic shall be such that a 1 ' h fr* = & ; this implies a" : finite, and hence clearly if the
system of conies (X, 4<#)< contains linepairpoints, no such linepairpoint can be a
proper solution unless this relation OP *& k is satisfied. .
NOTES AND REFERENCES.
and assuming the correctness of Zeuthen's values it would seem to follow that tho
four forms of surface have
12, 6, 12, 1
actual double planes respectively.
413. In the equation No. 36, tl = AP + BQ + CUZ + .. = 0, it is implicitly assumed
that the number of terms P, Q,R,.. is finite, viz. the implied theorem is that any
given fcfold relation whatever (k of course .a. finite number) there is always a finite
number of functions P, Q, X,'... such that every onefold relation included in tho fcfold
relation is of the form in question fl, = AP + BQ+ GE + ..., 0; this seems self
evident enough, but I never succeeded in finding a proof: a proof of the theorem has
however been obtained by Hilbert, see his papers "Zur Theorie dor algobraisoheii
Gebilden (Erste Note)," Oott, Nadir. No, 16, (1888), pp. 450457.
411, 415, 416. The first and second of these papers precede in date Zoufchon's
Memoir of 1871 referred to in 416, but I ought in that paper to have referred also
to his later Memoir, "Revision et extension des formulos nume'riques do la blidorio
des surfaces reciproques," Math. Ann, t, x. (1870), pp. 446546. I compare tho
notations as follows, viz. for the unaccented letters we have
Cayley,
n, a, , K, p, a
&. q> k *. 7
G, r, h, j8, 0, a
G,B
23 letters in all,
Zeuthen.
n t a, S, K, p, a
b, q, k, t, 7 ; a
G, r, h, /3 ; in
J>
B, U,0
f, i, d, y, e
27 letters in all
Here for Zeuthen's k, h, I have written k, k, viz, these numbers represent the
Plitckerian equivalents of the number of double points for the nodal and cuspidal curves
respectively. Zeuthen considers also the general node, say ( (^, v, y\v),0 + & u, ti),
see 416, this includes the cnicnode and offpoint w, and accordingly lie includes
under it and takes no special notice of these singularities, but it does not properly
include, and he takes special notice of, the binode B\ it does not extend to tho
case where the tangent cone breaks up into cones each or any of them more than
once repeated, and accordingly not to the case of a unode U where tho tangent
cone is a pair of coincident planes, He introduces this singularity, and also the
singularity of the osculating point which is understood rather more easily by means
of the reciprocal singularity of the osculating plane 0', this is a tangent plane
meeting the surface in a curve having the point of contact for a triple point ; and lie
disregards my unexplained singularity $. The letters s, m do not denote singularities ;
s is the class of the envelope of the osculating planes of the nodal curve, m the
NOTES AND REFERENCES. . 597
class of the envelope of the osculating planes of the cuspidal curve. Finally d denotes
the number of stationary points (cusps) of the nodal curve, exclusive of the points 7
which lie on the cuspidal curve; and g and e denote, g the number of ordinary
actual double points of the cuspidal curve, e the number of stationary points (cusps)
of the same curve, exclusive of the points ft which lie on the nodal curve.
Moreover with Zeuthen, the nodal curve has
30'+2' double points
/+30' + 2', if k denotes, as with me, the number of apparent double points
of the curve), and it has
7 + d + 2' stationary points.
The cuspidal curve has
g + QX' + l ^' + U' + 40' + 2 + 2' double points
', if A. denotes, as with me, the number of
apparent double points of the curve), and it has
ft 4 e + 20' stationary points
and the nodal and cuspidal curves intersect in
2 + :' points;
where I have written 2 and 2' to denote sums (different in the different equations)
determined by Zoutheu, and depending on the singularities ( and (' respectively,
For comparison of my formula? with Zeuthen's it is thus proper in my formula to
write = 0, a> = 0, Q = (but in the first instance I retain 0) and in his formula to
write *7=0, = 0, d = 0, # = 0, e = 0, 2 = 0, S'^0. Doing this the last mentioned
formulro give as with me 3+/ double points and 7 stationary points for the nodal
curve, but they give for the cuspidal curve 6^' + 12y9' (instead of 0) double points and
ft stationary points; and the two curves intersect (as with me) in 3/3 + 2y + i points.
There is a real discrepancy in the number 6^' + 12y9' of double points on the cuspidal
curve.
5 gg NOTES AND RKFKHKNCKS.
I compare his (6 + 26 + 1=) 33 relations:
(1) a = a', d = d'.
k = h'.
(6) (l)a + 26 + 3c,
(7) ft (((!) = + 28'* 3/c'.
(8) cW = 3(4
(9) b(bl)
(10) [3 (b  q) ~ 7 + d  s + 2', dotonninofi sj.
(11) c (c  1) = r + 2A + 3/3 + GO' ! 3d,
(12) [3(or)Hm + 20' + S', dotoiminos
+ 3i H 00' +
(15) o(2)
(16) a(
(IV) K
(18) c(fl
with the liko ieoiprooal oqutitioiw (0) to (IK) ;
(19) <r+mr/9 4/  3^  14(7' + ^
NOTES AND KEFEllENCES. 599
and my (3 +22 + 1 =) 26 relations as follows
(1) rt = it'..
(2) //'. , ,.
(3) i=f.
(']<) a = (nl)2&~3c.
(5) tf' = U?i(?i~ 2) (J6 8c.
(0) ' in ( 2) (' !))&&
(A.) (!!!)
(11) (14)
(0) (7) a (n  2)  k  Ji + /> + 2a +
(B) (8) fc(;t2)p + 2 + 87 + 3f.
(IS) (!)) o(n2)2ff
(F) (10) (tOi2)(ji8)2(803w) + 8(ao3ff x 3o
(0) (11) &(<>i2)(3)= ' 4& + (wfr2/3j ) + 3(6c
(H) (12) o(n2)(8)= flA + (ao8o %3)4 2(6o
with the liko rooiprocol equations (4) to
(I) (20) 2(?i l)(?i 2)(n 3)120i3)(&
= 2(n'  !)('  2)(?i'  3)  12 ('  3)(&' +c') + Gg' + 6)'' + 24t' + 42/9' + 30 7 '
A + <y + 6 + 1 2;r + D 7 + 40' + S + S'.
600 NOTES AND BEPERBNCES.
Substituting for k t h their values we have instead of (A), (5), (C), (D) the
equations
(A 1 ) ttb
(JS'J c 2 c
Writing as before 0=0, &> = 0; tf = 0, = 0, rf = 0, r/ = 0, e = 0, and neglecting the
terms in 2, 2', the two equations (E) become
Zeuthen c (  2) = 2a + 4/3 + 7 + 8^' + 167:2',
Cayley c (,  2) = So + 4/3 + 7 + 0,
which can be made to agree by writing = 8^' + 165'. But we have
Zeuthen (5') c 9  c = r + 2fc + 3/3 + 12^' +
Cayley (5) C 2 e =
values which differ by the terms 12tf + 24ff, or if 5 has the value just written down,
the term f 0.
I refrain from a comparison of the two equations (I.), and of the expressions for
the deficiency given by these two equations respectively but I notice hove tho
expression for the deficiency obtained by Zeuthen in the last section (XIV.) of his
Memoir, viz. this is
85+ 245' + 18tf + 6J7' + GO'
The problem is a very difficult one, and it cannot be held that as yet a complete
solutum has been obtained, Take in plane geometry the question of reciprocal curves 
here, using throughout pointcoordinates, we start with a curve represented by the general
equation ( ffl) y, *)> , such a curve has only isolated singularities, viz. the linesingularities
of the inflexion and the double tangent, we know the expression in pointcoordinates of
any such singularity (inflexion or double tangent as the case may be), viz, we can at
once write down the equation of a curve of the order n having a given stationary
tangent and point of contact therewith, or a given double tangent and two points of
on act herew^h Returning to the general curve fc y, ,)._<>, we know that the
recip ocal curve has other isolated singularities, viz, the pointsingularities which corre
spond to these, the double point (or node) and the stationary point (or cusp), and we
know the expression of any such singularity (node or cusp as the case may be) vi,
we can at on ce write down the equation of a curve of the order n having at a given
Zdi^Sr &?**&"* OT S P with given tangent. And then starting
afiesh with a curve of the order n having a node or a cusp we obtain the effect
NOTES AND REFERENCES. 601
thereof as regards the linesingularities of the inflexion and the double tangent. We
are thus led to consider as ordinary singularities in the theory the abovementioned
four singularities of the inflexion, the double tangent, the node and the cusp; and we
know farther that any other singularity whatever of a plane curve is compounded iu
a definite 'manner of a certain number of some or all of these singularities,
But in the theory of surfaces, starting in like manner with the general equation
(OB, y, z> w) n == 0, such a surface has torse singularities, the nodecouple torse, and the
spin ode torse ; each of these is in general an indecomposable torse of a certain kind
(hut there is the new cause of complication that it may break into two or more
separate torses), but we do not know the analytical expression of these singularities,
nor consequently the analytical expression of the curvesingularities which correspond
to them, the nodal curve and the cuspidal curve. Thus if we attempt to start with
a surface (ft 1 , y, e t w) n ~Q having a nodal curve, we can indeed write down the equation
in its most general form, viz. if the nodal curve has for its complete expression the /
equations P = 0, Q = 0, 11 = 0, &c. (viz. if the curve is such that every surface whatever
through the curve is of tlie form fl, =AP + J3Q+GR + ..., = 0) then the most general
equation of the surface having this curve for a nodal curve is (A, B, G, ...]P, Q, R, ...) a = 0,
but this form is far too complicated to be worked with ; and if for simplicity we take
the nodal curve to be a complete intersection P = 0, Q = 0, and consequently the
equation of the surface to be (A, B, CJP, Q) a = 0, then it is by no means clear that
wo do not in this way introduce limitations extraneous to the general theory. The
same difficulty applies of course, and with yet greater force, to the cuspidal curve;
and oven if we could deal separately with the cases of a surface having a given
nodal curve, and a given cuspidal curve, this would in no wise solve the problem ^ for
the more general case of a surface having a given nodal curve and a given cuspidal
curve. It is to be added that the general surface of the order n has no plane or
pointsingularities, and thus that such singularities (which correspond most nearly to
the singularities considered in the theory of reciprocal_ curves) present themselves in
the theory of reciprocal surfaces as extraordinary singularities.
END OF VOL. VI,
CAMBRIDGE :
PRINTED BY C. J. CLAY, M.A. AND SONS,
AT THE UNIVERSITY PRESS.
SKLKCJT SCIENTIFIC PUBLICATIONS OF
THE OAMBlfflXlE UNIVEESITY PRESS,
The
I
I
Collected Mathematical Papers of Arthur Cayley, Sc.B., F.R.S., Sadlorian
iT iT\J iw r ft ' ml 'i' 1H S " th(1 ' T '>ivi'Hity nf Oan.bri.lgo. ])cmy dto. 10 vote. Vols. I. II.
II. 1\. V. and VI. lir,*. onoli. [Vol. VII. / tAa 2>r<ua,
The Scientific Papers of the late Prof. J. Clerk Maxwell. Edited by W D
Mvi'lN, i\I.A,, ftinnorly I'Wlmv uf Trinity (!i>Hotfo. In 2 voln. lloyul .Jto. 3. 3s. (not).
Mathematical ana Physical Papers, by Lord KKI.VIN (Sir W. THOMSON), LL.D., D.O.L.,
Klt.K, I nilwtiii uf Natural IMiiloHitphy in Urn Univm'Hity of WIWKOW. Oollootoil from tlifi'orant Scientific
I iiniHlinilii Inini Muy 1HI1, l,(i tho proHonb Unit). .Dumy 8vo. Vol. I, IBs. Vol. II. IDs. Vol. III. L8s,
Mathematical and Physical Papers, by Sir G. 0. WTOK.KM, So.D,, LL.D., F.K.S., Lucaauwi
I njli'iimir nl Alal.hniiiiU.KiH in Mm Univornlliy of (Imnl.ridgo, Misprinted from tho Ori^im! Journals
mid Imn.'imitiniiH, with Aililitiomil Noti'H by tho Author. Dimiy 8vo. Vol. I. Ifis. Vol II 15s
[Vol. III. Jn the Press.
A Treatise on the Mathematical Theory of Elasticity by A. E. H. LOVE, M.A.,
I'Vllnw of St, .lolin'n Collt'jft', Ciunln'idi,'!!, in Two Voluuwa Vol. I. Doiny fivo, 12. Vol, II. 12s.
A History of the Theory of Elasticity and of the Strength of Materials, from
liahloi .i i Uio \\ww\\\, tinii>. Vol, 1. (liililnl to HaiiitiVwmut, 1(130 IHflO. lly tho lato I. TOUIIUSTKH,
N(i.l)., K.H.S., i!(iti'<t iinil t:ciinih)[,i!il Ity I'roltiHHtir KAIU I'KAHHON, M.A, Dainy Hvo. 26a,
Vol. II, Suitit'VKimnl to l,onl KKLVIN (Sir \Vn,i,iAM '.riniMHON). In 2 pjirtH, lly tho aurao Editor. 30a,
A Treatise on Natural Philosophy, .lly Lunl KHNVIN (Sir W. THOMSON), LL.D., D.C.L.,
K.U.M., and I'. (!. TAIT, M.A. Part I, Doniy Hvo. HI*. Part II, Demy Hvo. 18s,
Elements of Natural Philosophy. .Hy Lord KELVIN (Sir W. THOMSON) mid Professor
!'. U. TAIT. Domy Hvn. 11,*.
A Treatise on Analytical Statics. lly K J. ROUTII, Ho.D., 1'YR.S., Follow of the
Uni\WMil,y of Lninlon, Honorary l''ollow of L'litorliouHn. (Jainhridgij. 3)oiny 8vo. Vol. I. ids.
Vol. II. UM.
An Elementary Treatise on Quaternions. By I'.d.TArr.iM'.A. \\wlKditiwi. Domy Hvo. 18s.
Theory of DifTorontial Equations. J'mt 1. J^ximl; Ktiimtions and Pfaff'a Problem. By
A. II, KIIHHYTH, Stt.l)., I'Mt.H,, Knlhtw nf Trinity (!oll%'t, ('Hinlniil^i. l)my 8vo. 12s.
A Treatise on tho Theory of Functions of a Complex Variable, by A. It. IfonSYTii,
Hi!.n., I'MI.S, Hoynl Hvo. A Nut.
A Tretitiao on Plane Trigonometry, Hy 10. W. ILoiisoN', So.D., Follow and Loctursr
i if I'hrinl.'N ('iilli^i, (liiitiliriit^it. Uitiny Hvo. lt!.
A Treatise on the Theory of Determinants and their ftpplioations in Analysis nntl
(liHinii)lry, hy It. F. Sin ITT, M'.A., Follow of Ht Johii'M (lollngo. Domy Kvn. 12,?.
Hydrodynamics, u 'IWtimi on tlio Mtithoiimtifltvl Theory of tho Motion of Fluids, by
H. LAMII, M.A. Demy Hvo. 12*
Klomontai'y Thermoclynamlofl. Hy J. PAHKKH, M.A., Follow of St John's College,
I 'aiiiluiilgo. l.Vnwn Hvo. I).*.
A Treatise on aeometrical Optics. Jly It. S. H)5A.TJ[, M.A., ProfoHSor of Mathematics
in MiiMnn Huionua l.'olli^i 1 ! ItirininfjliHiii. Dciuy Hvo, 12s. (it/.
An Elementary Treatise on aeometrical Optics. By It, S. HEATH, M.A. Or, Svo. 6s.
Catalogue of Soientiflo Papers compiled by fcho Koyal Sooioty of London : Vols. 16
for Uio vi'iuH mH)l(lI. Dumy 4U. clitlh (Vol. 1 in half morocco) M (not); half morocco 5. tfs,
(not). Vul. 7 H for tho yeiiw IHttl 1878, cloth ,Bl. Us, Oil, (not); half morocco Z. to. (not),
Singlo vohinu'H ninth 2()jt. or halfmorocco Sto. (not). Vol. 0. No\v Son w for tho years 1874
1883. Cloth 2,1s, (not), half I.IOOTJ, 32i. (nut). [Vol. X. In tto Press.
Pontoon: C. J. CLAY AND SONS,
CAMBRIDGE UNIVERSITY PllESS WAKEHOTJSE, AVB MAEIA LANE.
CAMBRIDGE UNIVERSITY PRESS,
A Treatise on Elementary Dynamics. By S. L. LONEY, M.A,, late Fellow of Sidney
SU.T.S College. Xew ami Enlarged Edition. Crown Svo. 7s. Gd.
Solutions to the Examples in a Treatise on Elementary Dynamics. By tho
.Mint. 1 Author. Crown Svo. 7,*. Gd,
Plane Trigonometry. By the same Author. Crown Svo, 7s. Gd. Part I. up to anil
incliniiiig the Solution of Triangles is published separately. Os.
A History of the Study of Mathematics at Cambridge. By W. W. Rousifl BALI,,
M.A., Fellow and Lecturer on Mathematics iu Trinity College, Cambridge. Crown, Svo. (is.
The Electrical Researches of the Hon. H. Cavendish, F.K.S. Written between
1771 and 1781. Edited from the original MSS. in the possession of the Duko of Devonshire, K.O,.
ly tiic Lite J. CLERK MAXWELL, F.B.S. Demy Svo. 18s.
A Catalogue of the Portsmouth Collection of Books and Papers written by or
i.'L>lfjiihig to Sir ISAAC NEWTON. Demy Svo. 5s.
The Analytical Theory of Heat, by JOSEPH FOUIUEB. Translated, with Notes by
A. JHEEMAS, M.A., formerly Fellow of St John's College, Cambridge. Demy Svo. 12s. '
A Short History of Creek Mathematics. By J, Gow, Litt.D., Fellow of Trinity ColWu
JJiriiiy Svo. 10s, 6d, J '
Diopnantos of Alexandria ; a Study in the History of Greek Algebra. By T L HEATH
M.A., hollow of Trinity College, Cambridge. Demy Svo. 7s. erf. '
A T'S ta ? rP n ? e C; 6 " 6 ^! Principles of Chemistry. By M. M, PATTISON Mum, M.A.,
I.!!mv of Gonville and Cams College. Demy Svo. New Edition. 16s.
j
"v?T.9 U ^ matl 7 e An ^ sis  Concise aod Explanatory. By H. J H FBNTON
.A, F.I.U, Demonstrator of Chemistry in the University o/Cambricfgo. A 4to. ffwWtfa
MATHEMATICAL SERIES.
' SMITH ' 11A " Maste of Sid ^ Sussex College, Cambridg
Sus c, ^ Lo^v, M.A., , a te Folio, of
' ' '
Part IT p " ^AWJ. jwua leap. 8vo, 4s. Gd t
s  Tj  LosEr, M.A. is. Qd. e Element s of Statics and Dynamics. By
S":c B srr , T " s L . ^ 
*Yt .. *h_ . ^, _
! G. J. CLAY
MATHEMATICAL PAPEBS
ffiotrtxm: 0. J. CLAY AND SONS,
CAMI3HIBGE UNIVEBSITY PBESS WABEHOUSE,
AVE MAMA LANE.
! DEIOHTON, BELL AND co.
j p. A. BBOOKHAUS.
Jforfti MAOMILLAK AND 00.
THE COLLECTED
MATHEMATICAL PAPERS
ARTHUR CAYLEY, So.D., F.E.S.,
BADLKUIAN PHOPHSBOK OP PURE MATHEMATICS IN THE UNIVERSITY OP CAMBRIDGE.
VOL. VI.
CAMBRIDGE :
AT THE UNIVERSITY PRESS.
1893
[All Mights reserved.]
ADVEETISEMENT.
THE present volume contains 33 papers numbered 384 to 41G published
for the most part in the years "1865 to 1872; the last paper 416,
of the year 1882, is inserted in the present volume on account of its
immediate connexion with the papers 411 and 415 on Reciprocal Sur
faces.
The Table for the six volumes is
Vol. I. Numbers 1 to 100.
II. 101 158.
III. 159 222.
IV. 223 299.
V. 300 383.
VI 384 416.
CONTENTS.
384. On the Transformation of Plane Curves ..... 1
Proo. London Math. Society, t, i, (186566), No. in. pp. 1 11
385. On the Correspondence of Two Points on a Curve ... 9
Proc. London Math. Society, t. i. (1865 66), No, vn, pp. 1 7
386. On the Logarithms of Imaginary Quantities .... 14
Proc. London Math. Society, fc. n, (186669), pp. 6054
387. Notices of Communications to the London Mathematical Society . 19
Proo. London Math. Society, t. u. (186669), pp. 67, 2020,
29, 6163, 10310']., 123125
388. Note on the Composition of Infinitesimal Rotations . . 24
Quart. Math. Jour. t. vm. (1867), pp. 710
389. On a Locus derived from T^VQ Conies ...... 27
Quart Math. Jour. t. vm. (1867), pp, 7784
390. Theorem relating to the four Conies tuhich touch the, same
two lines and pass through the same four points
Quart. Matli. Jour. t. vm. (1867), pp. 162167
391. Solution of a Problem of Elimination ....
Quart. Math. Jour. t. vm. (1867), pp. 183185
392. On the Conies which pass through two given Points an
tivo given Lines ......
Quart. Math. Jour, t. vm. (1867), pp. 211 219
v l CONTENTS.
393. 0/i the Conies which touch three given Lines and pass through
a given Point ....... , .
Quart, Math. Jour. t. vm. (1867), pp. 220222
304. On a Locus in relation to the Triangle . . .
L/ * * "
Quart Math. Jour. t. vm. (1867), pp. 264277
305. Investigations in connexion with Casey's Equation . ,
Quart. Math. Jour, t, vui. (1867), pp. 33d 841
3DC. On a certain Envelope depending on a Triamjk imcribcd in
a Circle
........ .
Quart. Math. Join. t, ix. (1868), pp. 3141 imd 17fi 170
397. Specimen Table M , <M (Mod. N) for any prime or composite.
Modulus .
.........
Quart, Math, Jour. t. ix. (1868), pp. 9690 ami pl,vte
398. On a Certain Sextic Developable and Sextic Surface conned
t/ieremth .
' ' ' *
Quail Math. Jour. t. ix. (1868), pp. I20.ua and 373370
890. On the Cubical Divergent Parabolas .....
Q^rt. Math. Jour. t. ix. (1868), pp. 18G189
W. On tlu, Cubic Ct. m inscribed in a given Pencil of flfe .^,,
Quart. Math. Jour. t. ix. (1868), pp. 210221
401. A Notation tf tjle foints md ^ ^ pasmts
Qa.t Math. J 011 , t Ix . (1868) , )M68 __ 274
2. On a Singularity of Surfaces
Quart. Mail, Jom . t . , x . (1868) pp jM!uas8 ' ' ' ' ia.1
3. On Pascal's Tlieorem
Quart. Maft . Jou , , I18 . . . , lgl
404. Reproduction of Euhr's Memfr of ,,,. rt
Solid Sody. . J 758 on the Rotation of a
th. Jou , t . IX .  ias
otvi , fe 1 p> HJ ^ . 147
CONTENTS. IX
FAQE
406. On the Curves which satisfy given Conditions .... 191
Phil. Trans, t, OLVIII. (for 1863), pp. 75143
407. Second Memoir on the, Curves which satisfy given Conditions ;
the Principle of Correspondence . . . . . 263
Phil. Trans, t. CLVIII. (for 1868), pp. 145172
408. Addition to Memoir on the Resultant of a System of two
Equations .......... 292
Phil, Trend, t. OLVIII. (for 1868), pp. 173180
409. On the Conditions for the existence of three equal Moots or
of two pairs of equal Roots of a Binary Quartic or
Quintic . . . . . . . . . . 300
Phil, Trans, t. OLVIII. (for 1868), pp. 577588
410. A Third Memoir on Skew Surfaces, otherwise Scrolls . . 312
Phil. Trans, t. CLIX. (for I860), pp. 111126
411. A Memoir on the Theory of Reciprocal Surfaces . . , 329
Phil. Trans, t. CLIX. (for 1869), pp. 201229
412. A Memoir on Cubic Surfaces ....... 359
Phil. Trans, t. OLIX. (for 1869), pp. 231326'
413. A Memoir on Abstract Gfeometry ...... 456
Phil, Trails, t. CLX. (for 1870), pp. 5163
414. On Polyzomal Curves, otherwise the Curves 77+Vl 7 t&c. = . 470
Trans, R. Soo, Edinburgh, t. xxv, (for 1868), pp. 1110
415. Corrections and Additions to the Memoir on the Theory of Reci
procal Surfaces 577
Phil. Trans, t. OLXII. (for 1872), pp. 8387
416. On the Theory of Reciprocal Surfaces ..... 582
Addition to Salmon's Analytic Geometry of Throe Dimensions, 4th ed.
(1882), pp. 593604
Notes and References ......... 593
Plates to face pp. 52, 122, 190
Portrait ......... to face Title.
CLASSIFICATION.
GEOMETRY :
Abstract Geometry, 413
Curves, 114
BeciprocHl Smfaces, 411, 416, 416
Cubic Surfaces, 412
Skew Surfaces, 410; Developable, 398
Singularity of Surfaces, 402
= Curve., ! pencil of aix K MS ,
ANALYSIS :
Hotatio, ls , 388
M cmoh . f
of
eight,, M enioi ' ' > 8, 40D
384.
ON THE TRANSFORMATION OF PLANE CURVES.
[JTi'om the Proceedwcjn of the London Mathematical Society, vol. I. (1865 1866), No. in..
pp. 111. Eead Oct. 16, 1865.]
1. THE expression a "double point," or, as I shall for shortness call It, a "dp,"
is to be throughout understood to include a cusp : thus, if a curve has S nodes (or
cloTuble points in the restricted sense of the expression) and K cusps, it is here regarded
as having 8 4 K dps.
2, It was remarked by Cramer, hi his "Theorie des Lignes Courbes" (1750),
that; ft curve of the order n has at moat (?&l)(ji 2), =s J (n 9 3n) + 1, dps.
3, For several years past it has further been known that a curve such that the
coordinates (fa : y \ z) of any point thereof are as rational and integral functions of
tlie order n of a variable parameter 6, is a curve of the order n having this maximum
mi rnboi 1 (?i !)( 2) of dps.
4 The converse theorem is also true, viz. : in a curve of the order n, with
(ra l)(?i 2) dps, the coordinates (so : y : z} of any point are as rational and integral
functions of the order n of a variable parameter Q or, somewhat lesa preciselv. him
coordinates are expressible rationally in terms of a parameter Q.
OS THE TJUHSFOKlttTION OF PLANE CUBVBS. fgg.
series of curves of the order n1, given by an equation V+9V = (\ eontaininrr ,,,
rbitrary parameter 6; any such curve intersects the given curvo in tl.o ,1,,, ,,
>.mt,, lg as two points in the S.S points, and in o e other point; hone,, w ' tl,,,,
' only one vamble pomt of intersection, the coordinates of HUH point, vii! ., ,,,', ,,,
:.tes of an arbitrary pomt on the given curve, are expressible mtio mllv in t, r
"*"
* = the foregoing theorem is tha
o expressible rationally in terms of a parameter 9 S
fet po ints ^succeed each other in ^^ AR ^" ^ T" V " ',""
tamed by g lv i llg to the parameter it s different real Un fron , " ,'"'
rve may be termed a , curve. "^ '"' ' lllui
inciation it , s necessary to refer to naee l<f! , Tr PP ' , " <mi ' lll!to th "
; Abelschen Functioned Or*, t J "( 8 7, 1 'TT '""' ^P' " Thll '' rio
I lei L1V  ^ a7 >: PP' Ho],,,,, Vlai> th(J ( ,, mil(!i(l . ilm
of any order
form (1, f)(i, , ) = . P aiamete " (6 ';), connected by nu (!(I , llUi , m
>0  rationally in terms of the p a mn rn IK \
a certain form, viz.: ' te ' S (fi ') O ' mn *<l by an o,,,,atioii of
fl = l, the equation is (1, t\t n ~fn , ,
j^'^erootof;^;,^: oT "' tho """"
"4 the equation is (1, py> a x,_ n v
i. the square root of i 6 'exU 'fltbn O f P" to " 10 * mo
" > 2, viz. ; '
5 odd, = 2/1 3, the equation is a tw, ,
that treating (f ^ , r,^' f r(1> '" =. and is besides such
represented Lyi 2 )f d J rteSmn CM '*^s, the curvo thoT'
that treating (f^' 1?' P" (1 ' '' ) " = . a "d is besides sue
" ^ &
of this, f , . , f
coordmates (. s , ; , ^ , f. place of (f, ,),
Possible rationaUy and hooge neously i
n teni
384] ON THE TBANSFOKSfATION OF PLANE CURVES. 3
of (f. % D connected by an equation of the form ( )* ( ^ = 0. Such an aquation,
treating therein (, 1;, f) as coordinates, belongs to a curve of the order 2/t, with a
/ituplc point at (f = 0, =0), a /*tupIe point at (77 = 0, = 0), and which has besides
(M2) 3 or (/* !)(/* 8) dps, according as D = 2/i3, or 2/*2. The coordinates (x.y.s)
of a point of the given curve are expressible rationally in terms of the coordinates
($''>}' ) of a point on the now curve ; and we may say that the original curve is
by means of the equations which give (to : y \ z} in terms of ( ; 7; : ) transformed
into the new curve.
11. A curve of the order 2/a may have ^(2/41) (2/t 2), 2^ s 3/t+"l dps;
hence in tlie new curve, observing that tho /t tuple points each count for Kffft
we have
In the caso jD = 2/i ~ 8,
Deficiency = 2/t fl
In the case D = 2/j, ~ 2,
Deficiency = 2/i 3 8/t + 1
/*"+
9,i 9 n
*""* fffM ' ' >Wj " J^/
Moreover for 7J = 0, tho transformed curve is a conic, with dps, and therefore wibh
deficiency =0; in' the case J)~l, it is a quartic with 2 dps, and therefore deficiency
= 2; in the caso D = 2 it is a quintic with a triple point =3, and a double point
= 1, together 4 dps, and therefore deficiency = 2. Hence in every case the new curve
has the wamo deficiency as the original curve.
12. The theorem thus is that tho given curvo of the order n, with deficiency D,
may bo rationally transformed into a curvo of an order depending only on the
deficiency, and having the same deficiency witli the given curve, viz.: D = 0, the new
curve is of tho order 2(=D + 2); = 1, it is of the order 4 (=. + 8); JD2, ifc is
of the order 5 (= + 3); and D>2, it is for D odd, of the order D + 3;
even, of the order .0 + 2. It will presently appear that these are not the lo
which it is possible to give to tho order of the now curve, Riemanu' r
ON THE TRANSFORMATION OJT PLANE CUIWES.
 of ( " i * ): H is con to
iew  and
[384
11
the s
toV* with deflcie 7 A which is by the
'^z^^^Htf eacl i
i / r, 1 w> 7 ?> 6; ~u. ine transformed
<leficiency D ^ ^ ^^ ^^
equations f fP  n . /? f i ^ ' J * , t0 each of fch
i ^ / ' " 5 ' * : ^ theie C01 'responds a ainde set of
of tlie
nf /fc . M
'
IT. If, however, the curves Po = ffn

curve is equal to this number.
I assume that we have k < n :
P though t , le same ii(*8)l pia i"t," ' ' f "* " ler
sections common, the equations bein/f fee onms have all their inter
the tan P + ^ + Jlo To mX t ***, "T" ' ^ " Manti<lal relati
pebble, TO ^ tike the c , v f P * (P^'o' f T*"? m ' Ve " ' W a
as many poiuta ns possiblej and it ""'3^ from T]^ m89 ' .r "" on > ff =
groateat possible number i s =*/,% + 31^2 i,f? , , ""J"* , J " St " lade ' that
the number of points m the eSvt ^0 rflTff'' T ^T'' ""^
8  3, K'^ 3^) 2, respectively. " """" ^ to
7 e ; 00 f de S ^e curve ff = with deficiency D, or with i(tf lh,)
i Ai = ?t 1, we mav asmmn flinf IK i. r ^^ '
of the order !, ^ pas f *** ^ '"fa* Curves P0 ( 0, J20
through the * (m  3n)  .D + 1 dps,
and through 2 + JD  4 other poiiitB,
together j^j^ T^ points of thfl
384] ON THE TRANSFORMATION OF PLANE CUEVES.
This being so, each of the three curves will meet the curve U=0
in the dps, counting as ?i a 3?i 2.D + 2 points,
in the 2)i + D~4> points,
and in D + 2 other points,
together ?t a )i points ;
whence the order of the transformed curve is D I 2,
20. In precisely the same manner, secondly, if k = n~ 2, then we may assi
that the transforming curves P = 0, Q = 0, R  0, of the order n  2, each pass
through the 4( a 3w)D + l dps,
and through n + D4 other points,
together ^(?i a ~?i) 3 points of the curve 7=0;
and this being so, each of the three curves will meet the curve U =
in the dps counting as n*  3n W + 2 points,
in the n+ D 4 points,
and in 7) + 2 other points,
together a  2??, points ;
whence the order of the transformed curve is also in this case = D + 2.
21. I was under the impression that the order of the transformed curve a
not bo reduced below D + 2, but it was remarked to me by Dr Clebsch, that in
case JD > 2, the order might bo reduced to D + 1. In fact, considering, thirdly,
ease & = w3, we see that the transforming curves P~0, (2=0, .# = of the o
, 8 may bo made to pass
through the (w 9  3)  + 1 dps,
and through JD  8 other points,
together (n a  8w)  2 points of the curve U = ;
and this being so, each of the three curves meets the curve U= 0,
in the dps counting as (?t a  3n) 2.0 + 2 points,
in the 8 points,
and in > + ! other points,
together n*  3 points ;
whence the order of the transformed curve is in this case =/)+!,
22. The general theorem thus is that a curve of the order n with deficiency
can be, by a transformation of the order nI or n2, transformed into a curv
6 ON THE TRANSFORMATION OF PLANE CUEVttS. [8g
the order + 2; and if D>2, then the given curve can bo by a transformation of
the order 3 transformed to a curve of the order .Z) + l: tho trans formwl curvo
having iu each case the same deficiency D as the original curve.
23. In particular, if jD1, a curve of the order n with deficiency 1, or with
^ (n s  3) dps, can be transformed into a cubic curve with tho aamo doluuoiioy, thud
is with Odps; or the given curve can be transformed into a cubic, Thin WIBO is
discussed by Olebsch in the Memoir "Ueber diqjenigou Curvcn doron G'ooi'diiuiton
elliptisehe Functioiien eines Parameters sind," Grelle, t. r.xiv., pp. 210271. And lu<
has there given in relation to it a theorem which I establish as follows:
21). Using the transformation of the order n 1, if besides tho 2H + .H 4(=: g w ;j)
points on the given curve U~0, we consider another point on tins curve, thon wn
may, through the (w 3 3) dps, the 2w3 points and tho point 0, draw a wrios of
curves of the order nl, viz., if P 0t Q 0t R , are what tho function* P, Q, M, become
on substituting therein for (ts, y, z\ the coordinates fa, y 0i *) of tin; givuu point
then the equation of any such curve will be aP + bQ + olt^Q, with tho relation
aP + bQo + cll between the parameters a, b, c; or (what ia tho same thing) olimimitiii"'
o, the equation will be a^R* 2\R) + b(QR, Q ^) = 0, which contain* tho Hind?,
arbitrary parameter a : b. In the cubic which is the transformation of tho givuu ourvi
we have a point O' corresponding to and if (f a , %] ,) bo the Gi.ordinatoH of this
point, then coire spoil cling to the series of curves of the order nl, wu havu a HOHCB
of lines through the point 0' of the cubic, viz,, the linos f + * + < with thu
relation a D + 6^ + cft = between the parameters; or, wluit is tho samo tliinif, w,
have the series o lines <& ffi) + 6 ^o) = 0, containing tho Hmnn inglo
parameter  I By determining this parameter, tho CUVVOB of tho ordor n~l, will
vlicMoiich .h * rder ^7 h t dp3 ' th 2?1  8 P 01 ^ ^ * 1*** ^
hich touch the pven curve ^=0; and the lines will bo tho tangent* to tho onbu
nm 1 ' tane " tS *"
'
theorem, viz.: p0m; mt 18 ' ATO have the following
* cency V 1 wo
curve, four curves of the order Ul  vifTl ^ draw S M to touch tho
where the ratio ft : 6 is determined bv .' T^ 9 ^ by ftn equation
~, ttmi ULUVBS oi uiue order n i' viV tl vuu\,n unu
the ratio a : 6 is determined by a certai^ 9 ^7" 7 ^ eqUati n ^ + ^'"0,
( wn the absolute invariant J*.J O f tho '1,^ f 8 4 ( ! ua1a ? 1 (*!* & ) 4l=0 i then
01 the quartic funchon, is independent of the
384] ON THE TRANSFORMATION OF PLANE CURVES. 7
positions of the "2 2 points on the curve U 0, and it is consequently a function
of only the coefficients of the curve U = 0, being, as is obvious, an absolute invariant
of the curve U= 0.
26. And, moreover, if the curve U= is by a transformation of the order n l,
by means of 2 3 . points on the curve as above, transformed into a cubic, then the
absolute invariant I s s J of the quartic equation which determines the tangents to
the cubic from any point 0' on the cubic (or, what is the same thing, the absolute
invariant S 3 = 2 lJ of the cubic, taken with a proper numerical multiplier) is independent
of the positions of the 2 3 points on the curve U=(), being in fact equal to the
abovementioned absolute invariant of the curve U=Q. The like results apply to the
transformation of the order 11 ~ 2.
27. Suppose now that we have D>2, and consider a curve of the order n with
the deficiency D, that is with (n a 3n) D + 1 dps, transformed by a transformation
of the order n 3 into a curve of the order Dl with deficiency D', then, assuming
the truth of the subsidiary theorem to be presently mentioned, it may be shown by
very similar reasoning to that above employed, that the absolute invariants of the
transformed curvo of the order JD + 1 (the number of which is =47) G), will be
independent of the positions of the D 3 points used in the transformation, and will
be equal to absolute invariants^) of the given curve 7 = 0,
28. The .subsidiary theorem is as follows: consider a curve of the order D + l,
with deficiency J), that is, with ^J) (D  1) D = $(& 3D) dps; the number of
tangents to the curve from any point ()' on the curve is (JD + l)D(7> 3D)  2,
=3 4kb 2, (this assumes however, that the dps are proper dps, not cusps,) the pencil of
tangents has 4>D  5 absolute invariants, and of these all but one, that is, 4>D 6,
absolute invariants of the pencil are independent of the position of the point 0' on
the curve, and are respectively equal to absolute invariants of the curve.
29. To establish it, I observe that a curve of the order D + 1 with deficiency D,
or with (J9 3 3Z)) dps, contains ^(D + l)(D + ^)^^(D^W) 1 ~4D+2 arbitrary
constants, and it may therefore bo made to satisfy 4D+ 2 conditions. Now imagine a
given pencil of 4Z*  2 lines, and let a curve of the form in question be determined
so as to pass through the centre of the pencil, and touch each of the 4.D2 linos;
the curve thus satisfies ID  1 conditions, and its equation will contain 4D + 2  (4D  1), 
arbitrary constants. But if we have any particular curve satisfying the 4.D  1 oonditu
then by transforming the whole figure homologously, taking the centre of the pencil
as pole and any arbitrary line as axis of homology, ao as to leave the pencil of linos
unaltered (analytically if at the centre of the pencil 0, y = Q, then by writing
<uu + (3y \ yz in place of z) the transformed curve still satisfies the 4>Dl conditions,
and wo have by the homologous transformation introduced into its equation 3 arbitrary
constants, that is, we have obtained the most general curve which satisfies the conditions
in question. The absolute invariants of the general curve are independent of the
1 It is right to notice that the absolute invariants spoken of hew, and in what follows, are not in gaucral
rational ones.
OE P LANE .
3 arbitrary constants introducer] 1 *], i
consequently functions of on y the cn.ffi 'T 8 '7 te sfo >*tion ; and they are
"'being eo.it is obvious that Lr^'t 8 ^ f" ^ f * D ~* "
of the pencil of _ 2 ,. at ^ " be respectively equal to absolute invariant,
curve of the order D + l i, =J(/+ nTfl ' f, he j* 50 '"' 6 ln * of the general
each of the dps, honee i,, the prLt oL *n, , ~ i^n '^ ^ " MdaoHon =1 ' for
406i and there are th, 4^7 a h !,7 ' * ( ^ + !) ( ^ + ^H*SI))8,
oql to absolute invariant the pet l^TT* * %' ' ^ f ">">
of the pencil, there a re 4D6, each o he ', t ^~ 3 abs Ulte """*
cmve, and consequently indepe, <l e f O f H ?" '" " absolllte "variant of the
d,ffe out planes, thail joilli ^ ^ e ^ d ransformed curve as situate in
point on the transformed curve we hnvf g " Wl m ' Ve with the oorrespondi lw
*): if the t,vo curves : %r;.; ol , s S 7 S , of ". Arming a scroll"
see ,on by the plane of the oriri lm l "! vn ; ' " res P eo "ly, then the eon pleto
d of ' Derating En.. J'ti , y l^! ,"? f ^ of the order
transformed curve i s made of ^^ &l* sect.on by the plane of the
inc.. Conversely, given a sc ^,, of ' *e order ', and of genei ^
tic od .
, c o
be.ng ln ,,. Olu , ves rf the tic od _ + sMtona ^
e _
of the other; but for t !lo ^ scroll of ,1 ' , "^ ta 8f otion e the one
. breaking up as above th <ndar + ' ; ' i> not possible to find
UN TIIK ('uKUKHl'nNUKXi'K OK T\Vi) I'nlNTrt ON A (IIMtVK.
A.ii .l/Mf/mdM/t'tif ,S'.,,!V/y. tul, 1. {iMli
V, II* i<l A'iil Itl, iNiiii }
VII.
fiiul
lllf '!
111 Mtiriii
M ti J*
I h .'I.'
t** 6?.*'"!i r
hit'.*' im l'6Utl"l (I s ', I) 4 '* II ul lll' 'lrl)M' I * 1
at I ',
I thcit tf
10 OX THE CORRESPONDENCE OF TWO POINTS ON A OUHVK [{J8. r J
HirvL' () (the equation of the curve @ will of course contain the coordinntGH of J*
us maim:ters, for otherwise the position of P would not depend upon tlmt of ./>),
I iiml that if the curve has with the given curve k intersections ut the point /*,
hon in the system of (P, P) t tlie number of united points is
'.vhenc. in particular, if the curve does not pass through tho point 1\ thon
"ibvr oi united points is = + ', as in a unicursal curve.
* The foregoing theorem is easily proved in tho particular liasu whom th X

y, ') m,d of the order in u . 'Tf fu oto " t tho ordur iu
' rf the,,, vanish, id.nBc.ny. o, b^* 1P '7 T BU h "^ "'"^ <lu '"
y.) = fc ,. 4 ' "* " =
,
of ft .8
o the n ooo,,,
. have ff=o, an el similarly if' rv / J" 1 *' ' } ' m tho
!"'. J wo have ff' = . The eo Z n ( !' ^ *? a '' " 10 "'''i^ of tho
wnluiatcs of the given point P ^7 ' COH8ld( g therein (,, ,, s ) , t ,, n
/, as L ters "
=
have ft,, a the case above SUHM8 .j"^^ a havmg a /{4l , 1)lo iui
1
"f h ,
but i T ? 0+ii the total "umber of 7,,, 7 ' W ' UOh m '
of
un e ;"v;d,tha t upon writing ( ,
are ~ iv ' "
ni
vJii Is, *'. 'i. i !<
*,( < ( s oj I * /. , f
MtlVr. / " It, Julcl ()(
tmtf
I "1
.Mn]si):i'i . S'i tie [!. tfi"it "I n rune il ill
ini,i m" lli mid i i n.'H M! llii" IMIV v, ifh <h
!!l" nmd'l J"mf i S'i !hl','
(D (.,i I t/,,
/'
(Sti .. d
in
* w i'i"i, 
12 ON THE CORRESPONDENCE OF TWO POINTS ON A CURVE. [385
tangents from F to the curve, that is, ct  n  2. Hence the number of inflexions is
(2)+4J} I =m + n4 + 2(2ro+2), =3(wm), which is right.
8. For the purpose of the next example it is necessary to present the fundamental
equation a = a + a'+2/ i ;D under a more general form, The curve may intersect the
given curve in a system of points F each p times, a system of points Q 1 each
7 times, &c, in such manner that the points (P, P'), the points (P, Q'), & c ., are pairs
ot points corresponding to each other according to distinct laws; and wo shall then
have the numbers (a, a, a'), (b, ft 0'), &c, belonging to these pairs respectively j via.
(1,1) are points having an (a, a') correspondence, and the number of united points
is =a; similarly (P, Q') are points having a (ft ff) correspondence, and the number
ot united points is = b ; and so on. The theorem then is
9. Investigation of the number ef double tangents :Take P', an intersection with
the curve of a tangent drawn from P to the curve (or what is the same thing P P'
"* "" Cm ' Ve); th Unikd P^ ta hero the "o nts "of
c ero the po nts of
ontact of the several double tangents of the curve; or if T be the number of double
"
r r 
s^wKvTsv 1
^ , v;> ana ^T, /y, ) to the points (P, P'),
2(a~a') + 2T/3 / 3'=
Moreover, n,m the last example the value of .._, is ,, D , ftnd fchfl
2r^^2(n0)JDj
but from above it appears that we have fiff (n  2)(m _ 3)) whence
ON Til 10 milWWI'ONI>l'JN(!K Oh 1 TWO PO1NTM ON A (HWVIO.
henre, il' (/i, ;>) he tlm iOmm<!lerinl,i<!M n' Mm nyHtein of (lOiiinn (4#), Mm number of tlm
nonien through ./' in '/ A l l!IU ' n ' these I'"" 1 with Mm tfiveu <mrvn I iiitnrriiutlJnn ut
/', mid (KinmtipmnMy k.fi, Mnrnover, eaeh ni' 1,1m (umiiH bi^idus nmntH Mm e.urve in
(2m 1) pointN, and (uiimnmmnMy a t .; a.' /L (2w,  I), jlnncii Mm rnnniila ^iviiH l'.ho
ininilici 1 of nnili'il
or, HM tliis may bn
': IUI I l'l I ill (tyi I'}.
llulp Mm HyHlinu !' rnnicM (l^) rniidiiim (iljt  1/) i(inlriiurH (t'Ofw/Hfls wfnmnuiit
1'iinli M' whir.h, vi'Kimlril us a ptiir nl 1 r.iiiii(!)(lonf> lini'M, nmnln tlm givnn r.nrvn in
nuiuniilt'iil, pninUM; thai, SH, UIM ]Hmil,]tuir in l.n bit niiimidiiniil IIH IL i
hlit! tfivi'ii c.iirvn in HI puiulH; mid Mmm in on thin lumouiit n t'nliu^
() in tlm number nl 1 Mm unil.id pniuln ; whounti, Ilimlly, Mm niinibm 1 f Mm
H (I./) (!) in M/WC inm. It in liardly nrnwary l,u remark Mini. il> IH iwmnimtl
Mm nmdiMiniH {$%} \\\\\ ntntUl>mim having im Hjmdiul ri'lalimi do M
II. An H linal example, HiipptiHi that Mm puinl, 7' uu a ^ivnu curvo of Mm inilnr
nt. and t.lm point. (,? nu a K' v "" nill ' Vl ' "' )lln nnll<r m '' ' iavn au t. ') nim'iiHpinHlounn,
ami ln, il, In ivipiinid hi linil tlit fla^t ul' Mn> mirvi' cnvolnpcd by Mm linn i'Q,
Take un iirbitravy pitinl O, jnin fUJ, and Ittl. Miw meet llm uni've m in /", then
(/', /") itit* puintN nu Mm eurvn m, Imvin^ a (Vet, ma') nurn i Hpniitlitin!o; in limt., l.n a
^iven pciHitiinu ul' /' I here (tuiTi'Mpuml ' piwilimiM "!' Q, and t,o eiL^h id' Lhese. ?
puKiMuiiH til 1 /*', lhat in, I,M 1'iieh pusitinn ul' /* there mnvNpund waf pnMiMntin nl' /";
and Himiltivly In each piiHiMnn nl' /'' there mm'spttiiit m' pnsiiiunH "I" I". The dill've
H in the nyHleni ul' Mm lines drawn IVmu itwh of llm ' puMitioiiH nl' Q to Mm
pnint U, henen Mm IMIIVC H ilnei imt JIUHH Mtniit^b 7 1 , mid we have /,:'(). Uniirm the
uiLinht'i' nl 1 tlm united puinls (/', ]"). Mind in, Mitt mnnher !' the liimw ./'^ whicih
thrungh tht pninl f>, in i 7itV I ?it', r thin ! the rliiHM til' tlm niii've envnlnped by
1JJ, 1 1. may bit remarked, I hut ii' Urn two enrve.s are tiurviH in Mpatie (plann ni nl'
dimhln rnrvatiirc). tlmn the liki? retmoning H(IWM that Mm numhor if tlr '" ..... '"'
whinli nittub a given lino in ja'HHt'o, tliitt. in, Mm tirdnr ol' Mm HCIM
by Mm Un /'(, in MHI' + ?'.
ON THE LOGARITHMS OK IMA<;,\.\  !V
10* O.f tlm ./.<
,"
[From the PIWQBC
1'P ^0 H Knit 
y ol' (ihu low
but wo may ah) fl p ca i f
Writing tluis
and similarly
wo have of course
We have
_ _ P KJ j' fl rt
moreover bo
^
a,. HO that wo Jmvd a^t^^y
i( > ON THE LOGARITHMS 01? IMAGINARY QUANTITIES.
mid similarly
log P = log r' + iff )
and
logp=:logp + f
Hence
log P  log P = log + i (0 ff ,/,),
so that, by what precedes, log P log P', if the chord 1"P, coiMuloral H drawn I,,
P' to P, cuts the negative part of the axis of . upward.,, fc = l (lg .,. 2lV . if ^
chord out, the negative part of the axis of , downwa,*, iO is = log 2i7r , a,,<l
in every other case it is = IOET
*
Consider the integral
path of the variable , ; in order fco ? , Iimi J *~ P > <' <V it nay dopcmd on bho
* 1 p rt vAvi\j4. uu yivti IjiiO llrtrnrii\ r. & ^
therefore fix the path of the variable , and I r T^ H ^ mlicaliion ' must
the right line P'P w f ' j 7 y tlUt " lff tho l )ath fco bo
gw nne j P. Wnte Mw ^ p/> ^ du
PI " ilUIVU , ! fflrtss / J (Vllr,M, ... . 'I . 1,
^ w
******>> ** *
is a, on g t , le right lino , , P (fclwt is from uio
coordinates whereof are 1 *o M H
*> yu, to the pomt
We have thus
'*"* ^! U
the path in each case beW a ririlt , fl
n , p g UG ^ ab Ve ' The indefinite integral ^^lo,^
and as u pa8ses f rom j t P . ^ / w Io tt '
P" there 18 no diseontinuity in the value of lo gw; tho
386] ON THE LOGARITHMS OF III AGINARY QUANTITIES, 15
It is to be observed that 6 has always a determinate unique value, except in
the single case y 0, us negative, where we have indeterminately Q ir,
It is further to be remarked that, taking A for the origin of coordinates, we have
6 angle asAP, considered as positive or as negative according as P lies above or
below the axis of is.
Starting from the equation.
P = re ie ,
we have similarly
P'rV ',
and
P ' u
~ ft 1 ?
P' r '
p
where </> is derived from  in the same way as Q from P t or & from P'.
Consequently
e i<00'<) l f
and therefore 6' <f> a multiple of 2??, say
06'$ = 2flwr,
and in this equation the value of in is determined by the limiting conditions above
imposed on the values of 6, &, <$. To see how this is, suppose in the first instance
that the finite lino or chord P'P, considered as drawn from P' to P, cuts the negative
part of the axis of as upwards ; P is then above, P' below, the axis of as ; that is,
6, 0' are each positive ; and drawing the figure, it at once appears that the sum
& + ( Q'}> that is Q 6', is a positive quantity greater than IT. And in this case the
angle </> will be equal to 2?r (Q &') taken negatively, that is, $ ~  {2?r (0 0')},
or ~ & iJ3 = 2vr. Bub, in like manner, if P'P out the negative part of the axis of
so downwards, P will be below, P' above, the axis of CD ; 6 and &' are here each
positive, and the figure shows that the sum + 6' is greater than ?r; and in this
case the angle $ is = 2?r ( + 6'} ; that is, we have & & <f> =  27r. In m
other case, (that is, if the ' chord P'P cither does not meet the axis of .u <w
meets the positive part of the axis of to,) 6 Q' and </> are each in absolut'
less than IT, and we have #' (= 0, So that w '' '  1? 
16 ON THE LOGARITHMS OP IMAGINARY QUANTITIES.
and similarly
, lg P' = log r' f i&\
and
ll> 15J llf>M<f.
J^ o A' 7^
Hence
.0 that, by what p.eede, logP^P', if the choi , d
* to P, cuts the negative pa,t of the axis of Hpwards , is .^ if
chord cut. the ^ w part of the ax , rf m
m every other case it is = loo L
& p '
or from telow to above ti o f PM P C m & m ab Ve * Wow
the logarithm chan g e S from +1 to ' "1 o f ". rf ^ " le imf P art '"'
Consider the integral
path of the variable ,; iu order to fL , ' ^ ^ '* may de P end on tlu!
' ta
Wrfh now ^ p ,, ( _ we lmye _
p; a,d it is easy to see tllat , the path of
the righfc Une x fo P
coorfinatea whereof are l, v . 0> to the point ) .
We have thus
'J* Ji '
the path in each case being a right line as abov e. The
and aa * passes from 1 to ~
386] ON THE LOGARITHMS OF IMAGINARY QUANTITIES. W
value of the righthand side is thus =log~. As regards the lefthand side, the
indefinite integral is in like manner = \ogz; hut here, if the chord P'P cuts the
negative part of the axis of to, there is a discontinuity in the value of log 2, viz.,
if the chord P'P, considered as drawn from P' to P, cuts the negative part of the
axis of as upwards, there is an abrupt change in the value of log z from  iir to + iV ;
and similarly, if the chord cut the negative part of the axis of at downwards, there
is an abrupt change from + iir to ITT; in the former case, by taking the definite
integral to be log P log P', we take its value too large by 2zV, in the latter case
we take ib too small by 2wr; that is, the true value of the definite integral is in the
former ctiso = logP  logP'2wr, in the latter case it is = log P  log P' + 2m, But
if the chord PP' does not cut the negative part of the axis of ?, then there is not
any discontinuity, and the true value of the definite integral is log P  log P', We.
have tints in the three oases respectively
P
jp
which agrees wilih the previous results.
It may bo remarked, that it is merely in consequence of the particular definition
adopted that there is in the value of log P a discontinuity at the passage over the
negative part of the axis of to ; with a different definition, of the logarithm, there
would bo a discontinuity at the passage over some other line from the origin ; but a
discontinuity somewhere there must be. For if, as above, the chord P'P meet ^the
negative part of the axis of IB, then forming a closed quadrilateral by joining by right
p p
MUCH the points 1 to P, P to P, P' to p r and p to 1 ; the only side meeting the
fdz
negative part of the axis of a> is the side P'P ; the integral J , taken through the
closed circuit in question, or say the integral
+v\ dz
h
has, by what precedes, a value in consequence of the discontinuity in passing from
P' 'to P; viz., this is =2t7T or == 2wr, according as the chord P'P, considered as
drawn from P' to P, cuts the negative part of the axis of at upwards or downwards ;
but) this value  2wr or + 2wr must be altogether independent of the definition of
the logarithm ; whereas if, by any alteration in the definition, the discontinuity could
bo avoided, the value of the integral, instead of being as above, would be = 0. The
foregoing value  2wr or +2tV is in fact that of the integral taken along (in the one
0. VI. 3
18 ON THE LOGARITHMS OF IMAGINARY QUANTITIES. [386
or the other direction) any closed curve surrounding the point 2 = for which the
function  under the integral sign becomes infinite : but in obtaining the value as
z
above, no use is made of the principles relating to the integration of functions which
thus become infinite.
The equation
log P = log r + id
gives
PIU ~ gm log P J.TO gi'iiiO
or say
(ea 4 iy} = r m e !me ,
where, m being any real quantity whatever, r m denotes the positive real value of r m ,
We have thus a definition of the value of (0 + fy)* and the value so defined may
be called the selected value. And similarly, for an imaginary exponent m^p+qi, we
have
r p
which is the selected value of (
It may be remarked, in illustration of the advantage (or rather the necessity) of
having a selected value, that in an integral fa, taken between given limits along
a given path it is necessary that we know, for the real or imaginary value of z
correapondmg to each point of the .path, the value of the function Z, and consequently!
Z is a function involving log, or.*, the indeterminate,! ess which present itsol
m these symbols (considered as belonging to a single value of ,) i S) P SO to
indefinitely multiplied, and fa is really an unmeaning combination of symbols,
by selecting as above or otherwise, a unique value of log* or & we rondo,,
function to be integrated a determinate function of the variable. '
387.
NOTICES OF COMMUNICATIONS TO THE LONDON MATHE
MATICAL SOCIETY.
[From the Proceedings of the London Mathematical Society, vol. n. (1866 1869),
pp. 67, 2526, 29, 6163, 103104, 123125.]
December 13, 1S66. pp. 67.
PROF, CAYLEY exhibited arid explained some geometrical drawings, Thinking that
the information might be . convenient for persons wishing to make similar drawings, he
noticed that the paper used was a tinted drawing paper, made in continuous lengths
up to 24 yards, and of the breadth of about 56 inches 0); the halfbreadth being
therefore sufficient for ordinary figures, and the paper being of a good quality and
taking colour very readily. Among the drawings was one of the conies through four
points forming a convex quadrangle. The plane is here divided into regions by the
lines joining each of the six pairs of points, and by the two parabolas through the
four points; and the regions being distinguished by different colours, the general form
of the conies of the system is very clearly seen. (Prof. Cayley remarked that it would
be interesting to make the figures of other systems of conies satisfying four conditions;
and iii particular for the remaining elementary systems of conies, where the conies pass
through a number 3, 2, 1 or of points and tor ' n "* "" A " f K """
0 XOTICES OF COMMUNICATIONS TO THE [387
algebraical sum of the distances of a point thereof from three given foci is (this
V,;H ^lecUfl for facility of construction, by the intersections of circles and confocal
<'uni<s). The ijiuirtic consists of two equal and symmetrically situated pearshaped
curves, exterior to each other, and including the one of them two of the three given
f'R:i, the other of them the third given focus, and a fourth focus lying in a circle
with the given foci: by inversion in regard to a circle having its centre at a focus
the two pearshaped curves became respectively the exterior and the interior ovals of
:i Cartesian. There was also a figure of the two circular cubics, having for foci four
iriveu points on a circle; and a figure (coloured in regions) in preparation for the
'"iistniction of the analogous sextic curve derived from four given points not in a circle,
March 28, 1867. pp. 2526,
_ Professor Cayley mentioned a theorem included in Prof. Sylvester's theory of
Imvation of the points of a cubic curve. Writing down the series of numbers
1 2, 4, o 7, 8, 10 11, 13, 14, 16, 17, &c,, viz., all the numbers not divisible by 3
then (repetitions of the same number being permissible) taking any two numbers of
he H*IC.,, we have m the series a third number, which is the sum or else the
ifiom.ce of the two number (for example, 2, 2 give their sum 4, but 2, 7 give their
ftore.ce o) and we have thus a series of triads, in each of wnich one number
elLruct I ,e!t tf loirT ***** *' **"* * ****** m & CuWc
rhCH ^ ^l?^ 1 ?^^ ^^l^"
'
.
H >. the third point g of f anf 7 th tc'ond nf ?? **!, *""' ^ fil ' 8t f them '
Wo have here L theo rem J't! t W o'l ^ *"", " " "f ****** of 4.
"f i Similarly, 10, 11, 13 ,, each J Ten /I L m t  7 ,", a ' S
It. 10, 17, 19, each of them by t l u , e e nslSnl } i ". tamaaA ^ two
tractions increasing by nnity L , ZB of ' 1 n; &e numhor of
that the.0 constactions, 2, I ^^ SLJ?* ""^ ^ ^ theorem is 
o point. Prof. Cayley mentioneTthat on , 7' b ' 5 al " OM md *e
p. 29.
O to
on a straight line. ' ^ f * heae P mte >=eing flxed , whie
387] LONDON MATHEMATICAL SOCIETY. 21
March 26, 1868. pp. 61 G3.
Prof. Cayley made some remarks on a mode of generation of a sibireciprocal
surface, that is, a surface the reciprocal of which is of the same order and has the
same singularities as the original surface.
If a surface "be considered as the envelope of a plane varying according to given
conditions, this is a mode of generation which is essentially not sibireciprocal ; the
reciprocal surface is given as the locus of a point varying according to the reciprocal
conditions. But if a surface be considered as the envelope of a qitadrio surface
varying according to given conditions, then the reciprocal surface is given as the
envelope of a quadric surface varying according to the reciprocal conditions; and if
the conditions be sibireciprocal, it follows that the surface is a sibireciprocal surface.
For instance, considering the surface which is the envelope of a quadric surface
touching each of 8 given lines; the reciprocal surface is here the envelope of a quadric
surface touching each of 8 given lines; that is, the surface is sibireciprocal. So
again, when a quadric surface is subjected to the condition that 4 given points shall
be in regard thereto a conjugate system, this is equivalent to the condition that 4
given planes shall be in regard thereto a conjugate system or the condition is sibi
reciprocal; analytically the quadric surface oaf + by* + cz* + dw = is a quadric surface
subjected to a sibireciprocal system of six conditions. Impose on the quadric surface
two more sibireciprocal conditions, for instance, that it shall pass through a given
point and touch a given plane, the envelope of the quadric will be a sibireciprocal
surface. It was noticed that in this case the envelope was a surface of the order
(= class) 12, and having (besides other singularities) the singularities of a conical point
with a tangent cone of the class 3, and of a curve of plane contact of the order 3.
In the foregoing instances the number of conditions imposed upon the quadric surface
is 8; but it may be 7, or even a smaller number,. An instance was given of the
case of 7 . conditions, via., the quadric surface is taken to be o# a } 6?/ a + c# 3 _+ dw* =
(G conditions) with a relation of the form
Abo + Boa, + Gab + ffacl + Gbd I Red =
between the coefficients (1 condition) ; this last condition is at once seen to be sibi
reciprocal; and the envelope is consequently a sibireciprocal surface viz., it is a
surface of the order (= class) 4, with 16 conical points and 16 conies of plane contact.
It is the surface called by Prof. Cayley the " tetrahedroid," (see his paper "Sur la
surface des ondes," IAQU/V. torn. xi. (1846), pp. 291 296 [47]), being in fact a homo
graphic transformation of Fresnel's Wave Surface.
[Prof. Cayley adds an observation which has since occurred to him. If the quadric
surface am? I faf + ex* + dm* = 0, be subjected to touch a given line, this imposes on
the coefficients a, I, G, d, a relation of the above form, viz., the relation is
A z lo + B*ca, + &al) + Pad + G*bd + H*od = ;
where A, B, G, F, G, H are the "six coordinates" of the given line, and satisfy
therefore the relation AJF + BG+ <7I/= 0, It is easy to see that there are 8 lines for
which the squared coordinates have the same values A*, B*, 0\ F\ G 2 , f/ a ; these
8 lines are symmetrically situate in regard to the tetrahedron of coordinates, and
22 NOTKWS Ol' 1 rOMMinVK'ATIONH TO TIM; I
moreover they lio in a hyporbolmd. Tim iuiulrin mirliuv, instnid ..!' Imin^ ilrliinul
above, may, it in cloav, bo dnfinod by thn i!iiiiviilniil. u lil.iuu>i n' fniii'liimf ( . m ')i ,,f , i"
8 given linos: that is, wo havo Um oiivrlnpi> nf a ipmdrin Miirliuv hnuliiiiir ,,(,[, ii(
8 given lines; thosu HHUH nob bring urliit.niry MUCH, bill liciii^ a K.Vdfrm il' u \vi'v liiMri'il
form. By what procedoH, tho nnvnL.pn i>t n .jiiai'Uc Hiiriinv. li. 'IIIHIU;I, linwovcr I'l'^r
in virtue of tho rohition AF + JH}* (!Jf ?>(), Uim \ i,u tnu^.r M piupn'^uuili,' nmfi,!!^
but that it rosolvon itsoU' into MK abdvciiioiil.i d liyp.'rbi.lni.l mKru i.win ' TrY
is, icatoi'ing tho oiiginal A, it, &,!., in p llt v n f / '/.' J . ,<T M ll ivrln.. ' !' i')",
Xic I Jka + (Jab + 2i\id \Qbtl \ Jfal  0, (which in in ijPi'irnil u h'tiahrdmitl), in wi,".*"
^1, A (7, /'', ff, // am tin) Hi]imnl (!oonliiml,oH uf u Im, (nr, wltiif i^ il,r ^i,,,, u^)"."
passing through tho givon lino and tlirnngh the HyimMulnVMlly''^!., 'u,' L'v'n nllMrii!',!!' J' 1
la, IHli.S. pp. ion, nil,
Frofiw Onyloy g,tvo an uooouni, h, i lr M.tin,/ ul' ,i M,,,,,,,. liv ,.. ((i , H(( ,.
Comus rmnnhohor ( on,ploxo ndo. Vonill^nHMimiMK ,,., Ktilrr',,!,,,!, H,,.,,J v u %
^ ....... ''"
) 
number of spaces, un,l .orUiu Huppl.n.nnlnry mnu.riH,, A' " J
1
.
H tho bottom bo also We,,,, away, ,,,, H i . t ' ' ^ ' , 7 ,"T"i " '' "" '' ........ "
supplomonhuy (1 nni.lilii!H ,., i,, ,,, ,,,,, ,T, .',, ' .' "'" , ..... ' ..... "' " f "'
The ohiof diniculty ,! i ..... ,, r ,,, ''., "'.' """ lv "' ' *'l(./..*")..
8l Wlomo,,tavy ...Jatta ^ J ,?, ..... '' '"" '" ' ........ "" ...... ""' ...... '' ......
December 10. 1808, ini l'
1P l
OlMorratioiw by I'rofoas
, 
I 1. 2, 3, au,l any ,, tl lr(!u i jT '," , '"' ""* >' liui " ..... >
io C T OS ' caoh ^ * Hi \ ;. T , ,; v ;r"' r *
*. o, 8, wo havu a tmnfo, vt io n ,, . ,, . i " y ' ,:' ''' " "' ,! hl ' "'"'I' 1 " !
system " '' ' *"A, . 1, ; ^, l,, u , illK U u t , lllV( , nM ,
Si o >
points 4, s, 0, rat i n >0 8implo ^ r "' 2 , "f," ' ^'" Ol " ro 'l 1 ""' 1 ^ having
387]
LONDON MATHEMATICAL SOCIETY.
23
Analytically, Cremona's transformation is obtained by assuming the reciprocals of
"ai 2/a 2* fa ke proportional to linear functions of the reciprocals of as lt y lt z l (of course,
this being so, the reciprocals of *' lp ft, ?i will be proportional to linear functions of
tho reciprocals of flj a , 7/ 3) z.j). Solving this under the theory as above explained, write
M
a , /> , c \
i i_
r T
a' 3
u6i i/i Z\
1
} =
. L J, /_
. =
1
: + + i
if
Honco n
QJB^O, &c., arc qnavtioH, or generally aQ.12, + ^R l P l + yP, Qt = is a quartic, having
throo double points (y^O, ^ = 0), (^ = 0, ^ = 0), (^ = 0, ^ = 0), and having beaides the
throo points which are tho remaining points of intersection of the comes (&0, ^0),
( Ji, = 0, P, = 0), (Pj = 0, Q! = 0) respectively ; viz., these last are the points
"1 'I 'I o
:L . . :  = ei hf : fy  id : dh (/e> &c. &o.
Tho double and simple points are fixed points (that is, independent of a, /3, 7). and the
fonnulm coino under Cremona 1 * theory. It is, however, necessary to show that if the
point* < 5', 0' are in a line, the points 1', 2', 3' are also in a line. This may be
rtono as follows :
Lot thoro bo throo pianos A, B, 0, and let the points of the first two correspond
hv ordinary triangular inversion in respect of the triangle , on the_plane A, and _/3,
oil the piano S. Lot also the pianos S, C correspond by ordinary triangular inversion
in respoot of the triangle ft on tho plane B, and 7 , on the plane lh " "^
Hpondonoo between A and is tho one considered, the points 188 to ming the
r,,.^',:;". r ^^ r^js, i~
A^nlnrmim nrnnP.vtioa musl) apparently belong to uiemonuB uw
interesting part of the theory.
i's "TJeber
t
ip" Grelle, torn.
but in the present Paper wwnwty proportional to
388.
NOTE ON THE COMPOSITION OF INFINITESIMAL ROTATIONS.
[From the Q mrtef ly J omd of Pw . e and
^ vm .
pp. 710,]
n the last Smith ,
^ COS
ion to Denote M infinitesinml
*X rigidly ^ in the point
388] NOTE ON THE COMPOSITION OP INFINITESIMAL ROTATIONS. 25
where I, m, n, p, q> r are constants depending on the infinitesimal motion of the solid
body.
Hence, first, for a system of rotations
&>! about the line (a lt b lt c,, f lt </,, h^,
0)3 ,, (tt a , y 3 , C a , J 2l (/ 2 , fly),
the displacements of the point (#, y, z}, are
Bo) = . ?/Sca> s2ow 4
Sy 2cw , + zZaw 4
and when the rotations are in equilibrium, the displacements (80, Sy, 8,2) of any point
(a), ?/, z) whatever must each of them vanish ; that is, we must have
2wa = 0, S<u& = 0, SMC = 0, Sa>/ = 0, "Za>(f = 0, Sw/i 0,
which are therefore the conditions for the equilibrium of the rotations w lf a) 2 , &c.
Secondly, for a system of forces
P l along the line (a,, &,, c,, / 1( (/j, /M,
* 3 )i Jl V*3) ^3) C a , _/ a , (/ 2 , ft fl ),
&c.
the condition of equilibrium as given by the principle of virtual velocities is
SP (al + Imi + on+fy + gq + kr) == ;
or, what is the same thing, we must have
which are therefore the conditions for the equilibrium of the forces P u P 3 , &c.
Comparing the two results we see that the conditions for the equilibrium of the
rotations WD w a , &c. are the same as those for the equilibrium of the forces P l} P a , &e. ;
and since, for rotations and forces respectively, we pass at once from the theory of
equilibrium to that of composition; the rules of composition are the same in each case.
Demonstration of Lemma, 1,
Assuming for a moment that the axis of rotation passes through the origin, then
for the point P, coordinates (as, y, z], the square of the perpendicular distance from
the axis is
= ( , y cos y + z cos /3) 9
+ ( x cos 7 . z cos ) a
+ ( to cos /S + y cos a . ) a ,
C. VI. 4
2ti .VOTE OX THE C03IPOSITION OF INMNIT.ESLVIA1, IIO'I'A'J'IONH.
aid tho expressions which enter into this formula donoto (is fnllmvH ; via if l.lih Hll r]
^ lmt * at right angles to the plane through P ftnd t h axm nf mlul,i, m
Mil ""f n* : Iie rt Ular diStanCG f '
< Mirmtw oi Q refonwl to P as origin are
a cos 7 , _ g cos ffj
 A'cos/9+7/cosa
oach
' .V*, *) in place of ( an H
coordinate., (, J, c , f . n : '' ^' allt
" 4l <
cto .
of l emma 2.
gts ~
,"' "'" '''"' ........ ' ""'i''
to t ], ..inilind.],,,^ ( ,, u ,
n,l, ...... . l,v ,
I G f M .
Motion of the
389.
ON A LOCUS DERIVED FROM TWO CONIOS.
[Prom tho Quarterly Journal of Pure and Applied Mathematics, vol. vm. (1867),
pp. 778*.]
REQUIRED tho locus of a point which is such that the pencil formed by the
tangents through it to two given conies has a given auharmonic ratio,
Suppose, for a moment, that the equation of tho tangents to the first conic is
(HJ ay) (as by) = 0, and that of tho tangents to the second conic is (co cy) (at dy) 0,
and write
G (a
so that
write also
/,* / 
**~ A* lv *~A'
then tho anharmonic ratio of the pencil will have a given vali
(&/<;,)(& 70 = 0;
that is, if
or, what is the same thing, if
that is, if
 (B0}* =0,
^ 8 ON A LOCUS DERIVED PROM TWO CONICS. [389
where
are each of them symmetrical in regard to a, b, and in regard to G, d, respectively.
Let the equations of the two conies be
U =(a, b, c,f, (J , A>, y, 2) a = 0,
V (*',&', ef i f,g',kJa ll y l z^ = Q i
and let fa, ft 7 ) be the coordinates of the variable point. Putting as usual
(A, B> G, F, G, tf) = (i c / 2 , caf, i/ t , g} t af, kf~bg t fgch),
K = abo  af*  hf  c /i
the equation of the tangents to the first conic is
(A,B, 0, F, G,
where
X^Wfy, r^
and therefore
5l ,b sti t u ti, lg for 5 the vall>e _i (tf
tangents, an equation of the form aZ + 2hZ7+h7*n w l,i n i i ^ ,
to b e (Z7)(X~ & 7) = 0; that is, we ht e ^ " efa ' ^ taken
1 : ft+& : 06= a ; 2h : b;
an<l, in like , Mme , if the accented letters refer to the second code
1 : c f d : cd = a' :  2h' : b'.
Substituting for a, h, b their values, and for a' h' b' th
. ana 101 a, h, b the corresponding values, wo find
'
' a + b : ab
i c
0/aa
We then have
...Ja, A 7 ),
389] ON A LOCUS DERIVED PROM TWO CON1GS. 29
and similarly
( d)' = 4W. ..., /3, 7)'
We have, moreover,
(a + 6) (c + d)  2 (06 + cd)
= 4 (//7 3  ^7  Gf3>y
 2 (7V
 2 #
and substituting the foregoing values, we find
or putting for shortness
= (EG 1 + '0  2J!f, ., ., Gil 1 + (?'//  ^J"  ^'^, ., . Ja, ^3, 7 ) a .
the equation of the locus is
where (a, /3, 7) arc current coordinates. The locus ia thus a quartio curve havi
quadruple contact with each of the conies Z7=0, U' Q; viz. it touches them at th
points of intersection with the conic = 0, which is the locus of the point such ti
the. four tangents form a harmonic pencil.
The equation may bo written somewhat more elegantly under the form
30 ON A LOCUS DERIVED PROM TWO (.IONICS,
so that
A  2k ,.F=* mr
(BO ^ ,..) = (w)' jiff, H/J /,
and substituting these values the oqimtion in
which, if A, B, denote
a, p, 7
. /?, 7
i
fl, j0, fy
a > ft> 7
I , m, 11
f , m', n'
i , W, ?l
Xi tf'i ''
P, 9. r
p'. g', ?'
f, wi', n'
^. 2, '
}' J ,
i P> 7
a
a /3 > 7
a_{
i /3, 7
a, ft 7
_.
i, m, ?i
/' / /
f , m, n
^ fli, w
J?>> f/. ''
P. ff '
X. ?', *'
r, W', ft'
^'. '/, ''
, 7
respectively, (A + B + 0=0} is, in fact, tho
(2fi + l)M 9
or, what is the same thing,
that i
s
', m', *r'
,
7
jugate axes, the
we have K^abo, K'^
O.
ON A LOCUS DERIVED FROM TWO CONICS.
I suppose in particular that the two conies are
a? 4 inif 1 = 0,
fcho equation of the quurtic is
1 ma? +  1 _ a m ?  m  1 =
,, N ,, . .
or putting A, = jf/o/nrv5 tnis 1S
njiS _L 1
To fix the ideas, suppose that m is positive and > 1, so that each of the conies
in nu (illipso, tho major semiaxis being =1, and the minor semiaxis being = ,, ; .
3?or any roul valuo of Js the coefficient X. is positive, and it may accordingly be assumed
thai) X in poHitivo,
Wo have ..^..."t.. 1 ,> < 1, or the radius of the circle is intermediate between
m (m h I) in
liho aomiaxoH of tho ellipses, hence the points of contact on each ellipse are real points.
Writing for shortness
^  >
??i a + m
tho equation in
(a? + iwf  1) (ma? + f  1)  X (& + if a) 9 = 0.
For thu points on the axis of eo t we have
(fl)(flMJ a l)M< 1 a) B = 0,
that i
( m  \) & + { (1 + m) + 2X] a; 3 + (1  Xa fl ) == 0,
mid bhoucu
or, substituting for a its value, this ia
Eemarking tlmt tho values
are
32 ON A LOCUS DERIVED FROM TWO CONICS. [389
1 (m + l) a
and considering successive values of X; first the value \ = ,, =  ~, wo have
+ )
m/
m
1
m
or observing that
this is
^ = 0, or
\ w m
The next critical value is Xm. The curve here is
(a? + mf  1) (met? + fi\ m (& + ,*  a )a = Q
that is ' '
m (tf + # 4 ) + (l + m s ) a,' 2 ?/ 3  (m + 1) (a 3 + f) + 1
that is v ,/ y *
f m  I) 3 .ty + ( 2)w  m  1) (^ + /) + 1 _ , ?ia a = o,
or, substituting for a its value,
_ >
lmtfl&??li a8 _( m l)'(m
rt ,. . m( a
the equation is
.
m(m+iyt'
or, as this may also be written,
1
m H
in
or, what is the same thing,
in
389'
ON A LOCUS DERIVED FROM TWO OONICS.
33
which has a pair of imaginary asymptotes parallel to the axis of so, and a like pair
parallel to the axis of y, or what is the same thing, the curve has two isolated points
at infinity, ono on each axis,
linos
TKo next critical value in X = J(mtl) 3 ; the curve hero reduces itself to the four
,
(m
'
m
0;
and it in to be observed that when X exceeds this value, or say \ >(* + 1) 8 , the
cui'vo has no von! point on either axis; but when X=oa, the curve reduces itself to
(? I y 3 )"<), i.e. bo liho eirclo ro 3 + f  a. = twice repeated, having in this special
ciwo real pointw OH the two axes,
Ik i iww eauy to trace the curve for the different values of X, The curve lies
in ovory eano within the unshaded regions of the figure (except in the limiting coses
aftormuutiomsd); and it also touches the two ellipses and the four linos at the eight
points /i), at which points it also cuts the circle ; but it does not cut or touch the
four HnoH, tho two ellipses, or tho circle, except at the points k Considering X as
varying by HUCCUBnive stops from to oo;
A, = 0, tho curve is tho two ellipses,
\,
^^ M.!?JljA Vj the curve consists of two ovals, an exterior sinuous oval lying in the
\ W
four vegionH a and tho four regions 6 ; and an interior oval lying in the region e.
C. VI. 5
q *
ON A LOCUS DERIVED FROM TWO CONICS. Tggc,
1) 3
 there ,s still a s i oua oval M above>
aove> ^ .^^
\ w .
dwindled to a conjugate point at the centre.
; X "" ;
TO
; there is no interior oval, but only a
J
fc
, the onive becomes the four lines.
k>i(m+l) 2 , the curve lies wholly in tho f,,.
eonasting thereof of f olll . detaehed siuL ov a l s A TT  ^ ^ *" ^
!( + !). e^h oval approaches mm nea r' tnf ' ^ Y , lateS ' eSS from the
and inflate l ine . porti l w hieh temd " ' "^ W * ini1 f ed ^ the eide
And a X depart/ f fom the 1 tit T + , Z"" * V WhMl th OTal be
approaches more nearly to the circuil^ h PP '' 0a heS * "' each si
cont ^ 6eparates the 4
e crcu h
contains the dnnon. oval. ^ 6eparates the 4 ^giona 4 e , w hich
Finally, x = 0, the curve is t ], e ci role Wee repeated.
390.
THEOREM DELATING TO THE FOUR CONICS WHICH TOUCH
THE SAME TWO LINES AND PASS THROUGH THE SAME
FOUll POINTS.
[From the Quarterly Journal of Pure and Applied Mathematics, vol. vm. (1867),
pp. 162167.]
THE aides of the triangle formed by the given points moot; one of the given linos
in tliroo points, nay P t Q, 11; and on this same line we have four points of contact,
Bay A lt A tl A a , Ai\ auy two paii^i say A lt A t \ A B , A 4t form with a properly selected
pair, sn,y Q, R, out of the abovementioned three points, an involution j and we have
thua the three involutions
(A lt A; A aj Ai\ Q, .fi),
(A,, A s \ A it A, } H, P),
(A lt A,; A 2> AS\ P, Q).
To prove thia, lot ffi = 0, i/^O ^ G tno equations of the given lines, and bake for
the equations of the sides of the triangle formed Toy tho given points
b & + a y ~ab =0,
If oi + ' y  a' &' = 0,
the equation of any one of the four conies may bo written
Lab LW . L"of'b"
\j  ab b'os + a'y  a'V b"ai + a!'y 
and if this touches the axis of ic, say at the point co= a, then we must have
_^L JL' .. V/ = /f^") a ,
tvacKaiv a" () )(; a') (a;  a") '
52
Uti THEOREM RELATING TO THE FOUft CONICS WHICH TOUCH THIS
>r, aiming as we may do, #=('<)(") ('), this gives
L a =( ft op (a 1 a"),
MUW
manne, if the conic touch the axis of y, 8il y at the poi.,;
ifi =(b ^(fi'fi", '
and thence
p, ming 
P = a & (a'  ft ") (y j/^
f =a'b' (a,"  a )(b"~l)),
wo have ^V(a0( ~V),
and thenco
; * ;
wo have in like mamel .
 _
"' '
p , Putting
equation is '" ' ^
390] SAME TWO LINES AND PASS THROUGH THE SAMS TOUH POINTS. 37
and "by attributing the signs + and  to the radicals, we have, corresponding to the
four conies, the equations
(a  ,) V (X) + (a 1  ffl ) V (A") + (ft"  a.) V (X"} = 0,
 (a  ,) V (X) + (a'  a a ) V (A") + (a"  a 3 ) V (X") = 0,
(a  .) V TO  (a'  a,) V (T) + ( a "  ,) V ( JT") = 0,
(a  Oi) V (X) + <'  ,) V (*')  ("  <0 ^ (A) = 0,
where a n 3( a 3) a., aro the values of a for the four conies respectively,
Eliminating a" we obtain the system of three equations
(2a  ,  a,) V (X) + (a,  ,) V (A") + (a,  ,) V (A r ") 0,
foi  ) V (A') + (2a'  ,  ,) V (A r/ ) + (a 3  ,) ^ (.Y")  0,
(a, + a  a,  a.,) ^ (A r ) + (a, } a 3  a,  a,,} V (A r/ ) + (a, + 4  a,  a 3 ) V (^") = 0,
and then oliminating the radicals we have
, a 
a, + 4 ~ a 
==0,
which is in fact
1, . +'j act,'  0,
as may bo verified by actual expansion; the transformation of the determinant is a
peculiar one.
The foregoing result was originally obtained as follows, viz, writing for a moment
a V (X) + (if V (*"') + " V (X") = ,
tho four equations aro
 a, <I> = 0,
 <E> = 2
these give
(a,
2 (a  B
38 THEOREM RELATING TO THE I'OITR CONICS WHICH TOUCH THE
From the last equation we have
(a,  ,) fc = 2 (0  a V (X)  a' V (Z')J  2a 4 {>  V (X)
= 2(i^)*2(a
that is
or substituting for J(X), *J(X') their values in terms of <]>, we find
which way be written
that is
T
or
ft Of.,
that is
or finally
which is a known form of the relation
I ~1 _ ft .
V~ aa >
o + ft', acs' = 0,
gives the involution of the quantities a, a'; ,, 4
We have in like manner
1 t . ,, . .
0,
and
''olutions of the systems < a "j a . a
[390
, w
and < a; 1( 3 ; . a4
39
390] SAME TWO LINES AND PASS THROUGH THE SAME FOUR POINTS.
It may be remarked that the equation of the conic passing through the three
points and touching the axis of m in the point ; = ia
(a  a)* (of  a") 6 , (ft'  ? (a"  a) V (a"  ) 3 (a  o') b" _ Q
" + ' ''' ^ ' '"'
and when this meets the axis of y we have
>0.
yb " r yV ff6
Hence, if this touches the axis of y in the point y = {3, the lefthand side must be
 (a ~ a) 2 (ft'  a") + , (ft'  a) 3 (a"  a) + j (a"   ) 3 (a  a') (y  /3) s
, ft a )
and equating the coefficients of  a , we have
i/
( a _ a ). (a'  a ") + ^ (a'  ) 3 C^"  a) + j/ ("  a ) a ( a ~ a ')
P (a _ ) (a'  O + ^ (a'  )* ("  a) + ^ (a"  ) (  a j
ft
_a
or what is the same thing,
fa  ,f (rf  a") + (  ) ("  a) +
s ft t( '
"
= 2/3 (ft  a) 3 ('  a") + ~ ('  ) 9 ("  )
which gives /3 in terms of a, that is ft, ft, A, A in terms of ,, a a , ,, 4 respectively.
Cambridge, 30 November, 1863.
391.
SOLUTION OF A PBOBLEM OF ELIMINATION.
[From the Quarterly Journal of Pure and Applied Mathematics, vol. viu. (1867),
pp. 183185.]
IT is required to eliminate en, y from the equations
a > b , c , d , e
a' , b' , G , d', e'
a", b", G " , d", e"
This system may be written
= 0.
if for shortness
Or putting
we have
2/ 4 = SXe;
 ft + X a j &c.
 A
"* fV
391] SOLUTION 01? A PROBLEM OF ELIMINATION. 41
or, what is the same thing,
X (6 + Ao) + X' (6' + &0 + A." (&" + Aw") = 3
X (o + to*) + V (c' + fctf) + V (c" 4 Ad")  3
X (d 4 Are) + V (d f + fre 7 ) + X" (<?" + to") = ;
and representing the columns
a b> a' b f , a" I",
b c, b' c', b" c",
c d, c' d f , c" d",
d e, d 1 e', d" e",
ty
1, 2, 3, 4, 5, 6,
each equation is of the type
X(l + fc2) + V (3 + fc4) + V (6 + fc6) = 0.
Multiplying the several equations by the minors of 135, each with its proper sign,
and adding, the terms independent of k disappear, the equation divides by k t and
we find
X 2136 4 V 4135 + X" 6135  ;
operating in a similar manner with the minors of 246, the terms in k disappear, and
we find
X 1246 + V 3246 4 X" 5246 = ;
again, operating with the minors of (146 + 236 + 245 + &24C), we find
X {1236 + 1245 + k (2146 + 1246)}
+ X' {3146 4 3245 4 k (4,236 4 324G)}
+ X" {5146 4 5236 4 k (6245 4 5246)} = 0,
where the terms in Is disappear, and this is
X (1236 4 1245) 4 V (3146 4 3245) 4 X" (5146 4 5236) = 0.
We have thus three linear equations, which written in a slightly different foi
X 1235 4X' 3451 4 ^" '
X (1236 4 1245) 4 X' (3452 4 3461)
X 1246 4V 3462
and thence eliminating X, X', X", we have
1235, 1236 + 1245,
3451, 8462 + 3481,
5613, 5614 + 6623,
C. VI.
42 .SOLUTION OP A PROBLEM OP ELIMINATION.
which is the required result. It may be remarked that the second and third
are obtained from the first by operating on it with A, A 3 , if A ~ 2^ + 4<8 S f (
say the result is
1235
(1, A,
3451
5613
= 0.
In like manner for the system
if the columns are
then the result is
a? , afy, ay, ohj\ ay*, f
a , b , o t d , e , f
ft' > &' , c' , d' , e' , f
a", 6", c" , d", e" , /"
ft'", &'", c'", d"', e"' , f"
a b, a'V, a"b", a,'" V",
b G, V c' } 6"o", b'"o'",
c tZ, c' d', c" d", c'" d'",
d e, d'e', d"e", d'" e'",
*/> e'f, e"f"> e'" f",
= 1,2, 3,4, 5, G, 7,8;
(1, A, JA
where
12357
34571
56713
78135
=0,
43
392.
ON THE CONIOS WHICH PASS THROUGH TWO GIVEN POINTS
AND TOUCH TWO GIVEN LINES.
[From the Quarterly Journal of Pure and Applied Mathematics, vol. vm. (1867),
L pp. 211219.]
LET ro = v0 be the equations of the given lines; s = the equation of the
line ioining the given points. We may, to nx the ideas, imagine the implicit constants
so determined that + tfM0 shall bo the equation of the line infinity.
Take ffmyO.flnyO as the equations of the lines which _ by their inter
section with = determine the given points. The equation of the come is
y (m) + \i (n)J V (wj) = a + y V () + 7*
or, what is the same thing,
(co  my) ((o  ny} + 2 (w + y V (*)) y* + y V = >
so that there are two distinct series of comes according as ^ (vm) is taken with the
positive or the negative sign.
The equation of the chord of contact is
& + y V (w) + y* 0,
which meets ,0 in the point { ^y^nn)^ , = 0} that is in one of the ^centres of
the involution formed by the lines (, = 0, y0). (flW0. ^0). It to be
observed that the conic is only real when im i* positive, that is (the lines ^ and points
being each real) the two points must be situate in the same region or in opposite
redon3 of the four regions formed by the two lines: there are however other real
cases e. K . if the lines a0, y0 are real, but the quantities m 7, are conjugate
imageries; included in this we have the circles which touch two real lines.
b ^
ON THE CONICS WHICH PASS THUOUGH TWO GIVEN POINTS [3Q2
f V (m) + V ()} V W = + y V (mn)  7 (a + y),
which gives two coincident points, that is the conic is a parabola, if
that is (1 " 7) {A/ (wm) * ?)  * (V W + V ()},
or 7 9  7 {1 + V 0)) == i {V (m)  ^ (n))
7 = HI + V (m) V {(1 + m) (1 + n )J],
where it is to be noticed that
y == i [1 + V (m) + V {(1 + m) (1 + 71)}]
is a positive quantity greater than V(Mi), say 7 =^
7 = HI + V (mn)  V {(1 + m) (1 + )}]
is a negative quantity, say T = g, ff being positive.
The order of the lines is as shown m fig. i, 8ee plafce facing p _
7 =  ro to Y =  3 , curve is ellipse; 7as _ fff parabo]a
7  ff to p, curve is hyperbola; 7 ^ pambola Pj)
7=p to 7 = OT, ellipse.
Resuming the equation
the coefficients are ~ ' *+ TV 0,
(. & ' c. / ^ /O  [1, wi, 7 9 , 7V(m), 7,
and thence the inverse coefficients are
(4, 5, a, j, t?, jj) =
or, omitting a factor, the inverse coefficients are
W. a, a, j ffl a,. [ 0) a,
Considering the line
the coordinates of the pole of this line ar
are
392] AND TOUCH TWO GIVEN LINES. 45
or (what is the same thing) introducing the arbitrary coefficient k, we have
kx + 7/i  v v* (inn) 0,
ky + 7\ v = 0,
kz~\*/(iwi)  /* g" y(m)  V ()1 3 v =
the first two equations give
k ; 7 : li^A W (*)) : v \V V(w) ffl ) :*#/*&
that is
/t\V(m)l ~"iy V (*") 0)
,
or, substituting this value of 7 in the third equation,
lAW(m)l* , w (ffw)} + ^W ^ 0>
Xw/*y l ^ (Gy*j(mi
that is
(m)  V ()}' I ( ~ 2/ V (
+ * {  ^ V (*)! y {/i  X V (wwi)} = 0,
which is the equation of the curve, the locus of the polo of the line X0 + /*y+M*=0
in regard to the conic
(<a  my) (a ~ny) + 2{ + y*/ (ma,)} i* + 7^ a = 
In particular, if A, = /i = y = 1, then for the coordinates of the centre of the conic,
we have
to . y : 2 = 7 + V(<) = 7 + 1 = V (> + * + jj^ W ( m ) ~ V ()) 8 i
and for the locus of the centre,
(ffl^i{V(m)V())' + (y>l^
so that the locus is a conic, and it is obvious that this conic is a hyperbola. Putting
for greater simplicity
{oy =A,
co  y V (ron) = Y>
z Z,
the equation of the curve of centres is
Z B . HV (m)  V ()]' + XY i 1 + V (*)] + YZ {1  V (n)} = 0,
or, writing this under the form
Y[X {1+ V (mft)} + J? {1  V (>))] + i IV (*)  V (n)) a X* = 0,
46 OX THE COMICS WHICH PASS THROUGH TWO GIVEN POINTS [Q
the equation is
YQ + X* = Q,
where
 V (
these values give
{1  V (m)] * = [\l (m)  V ()} Q + 2 (1 + V (wi)) ^,
or, what is the same thing,
(1  V (mi)} w=  V ( mn ) X + Y,
[I ~ V (mn)} z = 2 (1 + V (mn)J Z + { ^ ( w )  V ()}'
whence also '
{1 + V (')} JT + 27+ (^(m)  V (
or the equation of the Hue infinity is
{1 + V(0) Z+ 2F+ {V(m)  V()}' Q = 0,
bo apKed to
e asymptotes
7Q + X = 0.
In fact we have identically
asymptotes of the oouio\B=oH, n w ( . Ue n " ty> PUte iu "vidonoo the
fa' a Wntnl S *' /* 7^ in the pl ao o of ^
, and . y. that is , ' * *
(u
392] AND TOUCH TWO GIVEN LINES.
that is
 4/3y&' (a s
47
 (a 3 + 4/37/c') ( 2/Vc'a  a?;*) 2  40y F 1 (OHO 4
or, what is the same thing,
_ 43# (> + W) (&V + 2/2) = {20tf oj + 2/3 a /,/y  (2/37/0' + a 3 ) ^} a
 (a 2 + WO (2^'<w  ^) 3  *^^ C
which, when tw + /5i/ + 7^ = is the equation of the HUG infinity, puts in evidence the
asymptotes of the conic #" + # 0.
Now writing X, 7, Q in the place of en, y, z\ tt = l, and == (1 + V (")). = 2 >
y={V(ft) V()l a we navo
 16 [{1 + V (w)) a + 8 (V (0  V ()i a ] (7Q + ^)
= [4 (i+ v ()) ^ + sy  (4 iv w  v ()) + (i + v (flw
 [(1 + V (m)" + 8 [V ()  V ()] s ] [4^  (1 + V <*)} Q] a
 16 [{1 + V <ro)l ^ + 27 + [V (0  V ()} 3 Q] 3 .
and the asymptotes are
= + V(l + V (WMI)}" + 8 IV C" 1 )  V ()} a [4A r  {1 1 V (mn)} Q}.
At the centre
4 {1 + V ()} A r + 8 F  [4 (V (w) ~ V ()} 2 + {1 + V <)} 9 ] Q = 0,
4Z{l + V(m)}Q ! =0.
but the first equation is
{1 + V (wn)) [4X  Q {I + V (*n){] + 8 F  4 (V (0  V (n)) 9 Q = 0,
so that we have
4A'  {1 + V (flw)J Q, 2 7 = {V (0  V ())' Q.
the first of these is
2 {V (TO)  V ())' (  ff)  {1 + V (ift)l* (*  y)  (1  ) * = >
and the two together give
2X {V (m)  V ()!'  {1 + V (W3i)] F = 0,
so that we have
2 {V (m)  V (*)!" (  ^ )  [i + V ()$ {^  ff V ())  o,
to determine the coordinates of the centre.
48 OX THE CONICS WHICH PASS THROUGH TWO GIVEN POINTS
The equation of the chord of contact is
} I 70 = 0,
which for 7=1 is parallel to y = and for y = V(0 is parallel to <y=0. Jlut
coordinates of the centre are
a : y : z =  7 + V (mn) ;  7 + 1 : ^ ( wm ) + l + _. (^ ( m ) _ ^ ( n )}^
which for 7 = 1 give
V(w) H 1 + i{V (*)  V())'i 2 + 2 V * : 2im
and for 7 = V (ni) give
a = 0,
if : *lV(w) : V
y (tttu
the cent,, !, J
+ y V (m) + 7'^ = 0,
where, writing for *, y , , the coordinates of the ^^ w(j
+ 2 v () + 4 y (m) + 1 + {V (m) _ ^ (B)} ,J = Oj
(mn)
7(1 +
that is
or, what is the same thing,
77 =
and consequently 7' = 7 only for 7==0 ,
TJ.
the fixed point Lx /^ \ ft con ' es P on{lll ig positions of the chnvrl nf o rtll *. rt . n
tL _ ^u pome (a + i/V(m) = 0, *=,( i * ^  cl ot con(; acli thmuwh
see fig. 2 in the
392] AND TOUCH TWO GIVEN LINES, 49
Chord of Contact. Centre,
01\. P a , at infinity on hyperbola,
OL(z = Q). L, (z = Q, a,'i/ = 0).
OH. 0, the lino joining this with being always
behind 00.
OP i. P., at infinity on hyperbola,
X {' + y V (mi') = 0] .  ' X (m = 0, y = 0).
OG (parallel to y = 0). G (on line # = 0).
077 (purallol to ^ = 0) and so back to H (on line ffi=0) and so on to
I have treated separately the case */(mn) = I.
Consider tho conies which touch the lines y a 1 = O f yM = and pass through
the points
Tho aquation is of tho form
f  a? + Je(t *? = $>
and to determine ft, AVG havo
c a
1 o" 1 + /c (1 )' J = 0, and therefore k ^  r a .
(.* ~ a )
Tho (Muixtion thiiB bocinnes
(1  ) 3 (f  ffiS ) + c a (  ) = >
thiiti in
(1  a) 3 y a + (c a  (1  a) 3 } ffi 9  2c 3 aa; + c a a 9 = 0,
or IIH this may bo written
{r&fi 1 a /i3/yS C\ _ \a
'^^f^o
Honoe tho nature of the conic depends on the sign of c G (la) a , viz. if this b&
poflilivo, or between the limits l + o, lc, the curve is an ellipse,
(Bcoordinate of centre ;
c (i a)
which is positive,
o (1  a)
a'semiaxis =
ysemiaxis
Tho coordinate of centre for a = 1 + c is = + oo (the curve being in this case a pai
P!> and for 1  o it is also = + oo (the curve being in this case a parabola P 3 ).
coordinate has a minimum value corresponding to a=V (1 o)> viz  fehis is ~i I 1 +
C. VI.
50 ON THE CONICS WHICH PASS THROUGH TWO GIVEN POINTS, &0.  ,'JU'J
Hence as () passes from 1 + c to V(l~ c3 ). the coordiimfce of tho eontn; pus^n
from x to its minimum value {1 +*/(! c a )j; iu the passage wo hnvo a= rrivJiiff
the coordinate =1, the conic being in this case a pair of coincident linos (; 1)' J =. ; O.
And as (2) passes from the foregoing value Vflc 2 ) to 1  o, tho (inordinate of (In 1
centre passes from the minimum value ^{1 + ^(1 c a )} to GO.
The curve is a hyperbola if a lies without the limits 1 + c, I  c,
.'(.'coordinate of centre  ZJ^. (
which has the sign of a,
0semiaxis = ~~ ..L...T /
ysemiaxis =
semiaperture of asymptotes =
' [ " A/  (i );
^l^r 1C ^.= 0. (Parabola), but inereases as l illoroiu(oa
coordinate of centre is =
lc, coordinate of centre
s
u ' " >, 0,
tho hyperbola being in this case the pair of lines * *,'
a re
.. . c')}, and then as a !>/ ,7 a mnximum l"itiv viiliu.
, i. . "'ii'H us a passes irom //i _ ^ <. ..
centre diminishes from HlV(l<fll n T ~ C I cooiflinatp i>l
negative, the Hnes y ^^ = n ^f.^,1/,.,, , i8 to bo Barker! that a 1
51'
393.
ON THE CONICS WHICH TOUCH THREE GIVEN LINES AND
PASS THROUGH A GIVEN POINT.
[From the Quarterly Journal of Pure and Applied Mathematics, vol. VIIL (1807),
pp. 220222.]
CONSIDER the triangles which touch three given lines; the three lines form a
triangle, and the lines joining the angles of the triangle with the points of contact of
the opposite sides respectively meet in a point S: conversely given the three lines and
the point S t then joining this point with the angles of the triangle the joining lines
meet the opposite sides respectively in throe points which are the points of contact
with the throe given lines respectively of a conic ; snch conic is determinate and unique.
Suppose now that the conic passes through a given point; the point 8 is no longer
arbitrary, but it must lie on a certain curve ; and this curve being known, then taking
upon it any point whatever for the point S, and constructing as before the conic
which corresponds to such point, the conic in question will pass through the given
point, and will thus be a conic touching the three given lines and passing through
the given point. And the series of such conks corresponds of course to the
points on the curve.
52 OX THE CONICS WHICH TOUCH THEEE GIVEN LINES, &O. [""SO 3
the condition in order that the conic may pass through the given point is a J & + G == O>
and we thus find for the curve, which is the locus of the point 8, the equation
{(*)
or, what is the same tiling,
the rationalised form of which is
= 0.
This is a quartic curve with three cusps, viz. each angle of the triangle in a
and by considering for example the cusp (?/ = 0, # = 0) and writing the equation uiitlor
the form
o? (y  zf 2a (yz* + y z z) + if& = 0,
we see that the tangent at the cusp in question is the lino s/,3=0; that is, fclio
tangents ^at the three cusps are the Hues joining these points respectively with tins
given point (1, 1, 1). Each cuspidal tangent meets the curve in the cusp counting 1 IXH
three points and in a fourth point of intersection, the coordinates whereof in tho Oiuscs
of the ^ tangent y* = 0, are at once found to be a : y : z^ 1 : 4 : 4, or say thia IH
the point (1, 4, 4); the point on the tangent xto = Q is of course (4, 1, 4), and thai*
on the tangent *y0 is (4, 4, 1). To find the tangents at these points reHpoctivoly,
I remark that the general equation of the tangent is
that is
X . Y . Z
or for the point (1, 4, 4) the equation of the tangent is 8Z+r+^0, or
8 + y + ^0; that is the tangent passes through the point , = 0, or+y+^O, boi
the pomt of mtersection of the line , with the line . + + ;. ^hioh IB t
e ne , wt the line . + y+ . 0 hioh IB tli
a 4 Sr/fT^'JV' 1} " r ^' d fc the U ^ *
o tl haimoni ' it ' ^ 1' 'n } T^^ ^ to)I h th ^ ints
ot the harmonic lino a + y + g = Q mt h the three given lines respectively,
of
* " "'
r
without much difficulty. ' w Sles P 10 *^ may be effected
Plate I.
$5 face p. 52.
394.
ON A. LOCUS IN RELATION TO THE TRIANGLE.
[From the Quarterly Journal of Pure and Applied Mathematics, vol. viir, (1867),
pp. 204277.]
IF from any point of a circle circumscribed about a triangle perpendiculars are
kit Tall upon the aides, tho feet of tho perpendiculars lie in a line; or, what is the
Htimo thing, tho IOCUH of a point, such that the perpendiculars let fall therefrom upon
tho Hides of a given trianglo havo their feet in a. line, is the circle circumscribed
about iiho triangle.
In this well known theorem we may of course replace the circular points at
infinity by any two points whatever; or the Absolute being a pointpair, and the
terms perpendicular and circle being understood accordingly, we have the more general
theorem expressed in tho same words,
But it is loss easy to sec what the corresponding theorem is, when instead of
being a pointpair, the Absolute is a proper conic; and the discussion of the question
affords some interesting results.
Take (so = 0, / = 0, s = 0) for the equations of the sides of the triangle, and let
tho equation of tho Absolute be
(a, &, o, /, g, hfco, y, *0 2 = 0,
then any two linos which are harmonics in regard to this conic (or, what is the
same thing, which are such that the one of them passes through the pole of the
other) are said to be perpendicular to each other, and the question is:
Find the looiiB of a point, such that the perpendiculars let fall therefrom on the
sides of the trianglo have their feet in a line.
Suppoain* as usual, that the inverse coefficients are (A, B, 0,1, &, 5),
K IB the discriminant, the coordinates of the poles sides respee
54
ON A LOCUS IN RELATION TO THE TBIANGLK
[394
(A, H, G), (H, B, F), (G, F, 0}. Hence considering a point P, the coordinates of which
are (x, y t z), and taking (X t Y, Z) for current coordinates, the equation of the perpen
dicular from P on the side X = is
X, 7, Z = 0,
A, H, G
and writing in this equation A'=0, wo find
( , AyL
for the coordinates of the foot of the' perpendicular. For the other perpendicular
respectively, the coordinates are
and
(BxHy, , BzFy},
(CxGs, OyFz , o ),
and hence the condition in order that the three feet may lie in a lino is
or, what is the same thing,
, AyEss, Az~
CsaGz, CyFz,
= 0;
that is
( ^  OH) & + 5
f
the,ofore a cubic.
and /f' for the discriminant
or as th l8 may also be written
, the equation i s
^
394] ON A LOCUS IN RELATION TO THE TRIANGLE. 55
that is
r 2 f A nn nTT\ A B C'l /A B G \fa y z\
vffrftffi (ABO  FGII)  rv a  M ~ rrT a *F+ ( Jf lit + 7v ^ + W ^) ( B? + O? + /// )
and the cubic will therefore break up into a Hue and conic if only
2 ABO
TFG'H.' ( ~~ ' ~ j^a ~ (yi ~ p3 ~ '
and it is easy to see that conversely this is the necessary and sufficient condition in
order that the cubic may so break up.
The condition is '
n = zFQ'ii 1 (ABO Fan)  ACPIP  BHW  ow a = o,
we have
A A' + ]&' + GO' = 3vl56'  AF*  5G a  C// a , = 7f + 2 (ABO
and thence
a j'<?7/' (^^' + J5J3'
that is
tt = A G'H' (Q'ff  A'F)
so that the condition il is satisfied if If = 0, that is if the equation
(A,B, 0,F, ff.SJfc fl.0^0,
which is tho lineequation of the Absolute breaks up into factors; that is,, if the
Absolute be a pointpair.
In the case in question we may write
(A, B, G, ff, G,
that is  . ,
(A, B, 0, F, G, J/) = (2aa', 2/3/3', 277', 0</ + 0<y>
whence also, putting for shortness,
(/  /3V> 7'  y' a > a $' " a '^^ = ( x> ^ v ^'
we have
(A 1 , B', 0', F', G', #') = (^> /*"' ^
and also
JTf = 0, 2 (X^a  FGH) = AA' + BB'+ CO', =  2
50 ON A LOCUS IN RELATION TO THE TRIANGLK
The original cubic equation is
J? + 77V) ay& + eufayz (py + vz} + pfipza (\cv \ vg) 4 77'^ (Xft. 1 +
mid this in fact is
(aofayz + pfi'ttxte + w'vwy) (X# + p,y I MS) = 0,
The equation \x\ py +vz = ft is that of the lino through tho two pointn which
constitute the Absolute; the other factor gives
s
= 0,
which is the equation of a conic through the angles of the trianglo (to = 0, y = (), z =s ()),
and which also passes through the two points of the Absolute; in fact, writing' (, /3, 7)
for (a;, y, z) the equation becomes a/3y ('X + ftp + y v ) = 0, and so alw> writing (', /3', y')
for (.B, y t 2) it becomes a'/9V (oX + ^ + 7^) = 0, which relations nro identically maiw fusel
by the values of (X, /*, v). Hence we see that the Absolute being a pointpair, tlm
locus is the conic passing through the angles of the triangle, and tho two points of
the Absolute; that is, it is the circle passing through the angles of tho triimglo.
But assuming that 1C is not =0, or that the Absolute in a proper conic, thu
equation fi = will be satisfied if
AFG'ff + BQH'F + CHFG' + PG'H'  0,
we have F, ff,H' = Kf,Kg, Kh respectively, or omitting the factor K\ the equation
becomes
AF<th + BGkf+GHfg + Kfgh = Q,
which is
/VA 3  bctfh*  coA 8 / 8 4 a6/ a (/ a + 2oio/^A = 0,
or, as it may also be written,
tho l r e "^^i anbstitutiug also for ^, ^
the values AT/ % /a, the equation of the cubic curve is
2 (a&c /jA) X yz + Ay, (hy +
and the transformed form is
we have
so that the foregoing condition
_
a/*
394]
ON A LOCUS IN" RELATION TO THE TRIANGLE.
0* *?/ P
being satisfied, the cubic breaks up into the line 7.+  + 7 = 0, and the conic
/ 9 h
'
A
G
It is to be remarked that in general a triangle and bhe reciprocal triangle are
in perspective; that is, the lines joining corresponding angles meet in a point, and
the points of intersections of opposite aides lie in a line ; this is the case therefore
with bhe triangle (# = 0, y = Q, 2 = 0), and the reciprocal triangle
(uai i hy f gz = 0, ha> + 6y +fz = 0, gio +fy + cz = 0) ;
and it is easy to see that the line through the points of intersection of corresponding
sides is in fact the above mentioned line ?4^ + ?0. It is to be noticed also that
f y h
the coordinates of the point of intersection of the lines joining the corresponding
angles are {F, G, H). The conic
9
is of course a conic passing through the angles of the triangle {& = (), y 0, 2 = 0);
it is not, what it might have boon expected to be, a conic having double contact with
the Absolute (a, b, o, /, y, Ajw, y, zf.
I return to the condition
=0
'
abo c(/' fl by 1 * c/t 3 fyli
this can be shown to be the condition in order that the sides of the triangle
( 0, y = 0, z = 0), and the sides of the reciprocal triangle (eta; + % + 0*  0, /we + fy +/* = 0,
r/ft> +/?/ H c# = 0) touch one and the same, conic; in fact, using line coordinates, the
coordinates of the first three sides are (1, 0, 0), (0, 1, 0), (0, 0, 1) respectively, and
those of the second three sides are (a, h, g\ (h, b, /), ({/, f> o) respectively ; the equation
of a conic touching the first three lines is
^ + + 0.
and hence making the conic touch the second three sides, we have three linear
equations from which eliminating L, M, N, we find
which is the equation in question.
0. VI.
58
ON A LOCUS IN RELATION TO THE TRIANGLE.
= 0,
We know that if the sides of two triangles touch one and the same conic fclmir
angles must lie in and on the same conic. The coordinates of the angles are (1 ' O ())
(0, 1, 0), (0, 0, ]) and (A, ff, G), (H, B, F], (ff, ff t G) respectively, and the anglon 'will
be situate in a conic if only
111
A' IP
111
H' B' F
1 1 1
G' F' G
an equation which must be equivalent to the last preceding one; this is easily voriliiil
In fact, writing for shortness
V =
we have
1
1
1
1
I
I
a'
A'
ff
'
A'
H'
G
I
1
1
I
I
I
h>
b'
7
H'
B'
F
1
1
i
I
1
I
ff*
/'
G'
ff'
G
S.
and the second factor is
But
 aGH (AF + Iff) + A.FKBG + AFgOff,
= AF(a6ffI
so that the second factor is
which is
" ^ (A* 
+
~  L _ JL 1 , 2
o . "
394] ON A LOCUS IN RELATION TO THE TRIANGLE.
so that we have identically
 ABCffWH* Q = Kalcf^h V ,
and the conditions V =0, D=0 are consequently equivalent.
The condition
L L_JL_1 + A =0 "
abc a/ 3 If/" ch* fgh '
is the condition in order that the function
may break up into linear factors; the function in question is
/ bo ca ab
(<^. c '7 7 j
whichiH
i A B , 0\
so that the condition is, that the conic
(a, o, c, /, </, /iji. ft; 1/> z) "T \^j:J g h /
(which is a certain conic passing through the intonation, of the Absolute
A n Is ~* i it t . 
(, fc o t /, g t Kt*. y, *0, and of the locus conic ^++^0) 1 be a
pair of lines. Writing the equation ef the eenic in question under the form
6c
the inverse eoefficients A' t B', G', F, ff, ff of this conic, are
Ale Boa Gab _^jr  a G
""
i, B"  r<  trF ' G : S. Hence, if in regard w i.uia u  
so that wo have M .(r "' ^ = ), and join the corresponding
form the reciprocal ef the tmngte (.Oj 0,j J, J ^ ^ ^
angles of ^e t,o t *. ^^^^ fa recipvocal in reg ,d
It is to be noticed that the conic
60
N A LOCQS IN RELATION TO THE TRIANGLE. [394
contains the angles of the reciprocal triangle, and is thus in fact the conic in winch
are situate the angles of the two triangles. For the coordinates of one of the aiielon
of the reciprocal triangle are (A, H, (?); we should therefore have
which is
or attending only to the second factor and writing
GH=Kf+AW )
the condition is
Kfgh + APgh + SGhf+ GSfg = 0,
or substituting for If, A, B, G t F, a, H their values and reducing, this is
~__4.__
aio a/" Ziy" c/i' fgh~
t t if the trian f b i suoh that the re >
tuaugle he m a oomo (or, what j. the same thing, if the sides touch a
conic) then the cubic locus breaks up into the line  + 8+^0, which is the linn
inTtt otic* "' 3 f intei ' Se ti0n f the ^% des of the two triangles, a,ul
S

which is the conie through the angles of the two triangles.
its alrie^rrfTh MiaeS ; g! T , a C0ni (the Abs lute ) to constraot
' S
in a cic
in re Md to the conic, Ho
394] ON A LOCUS IN RELATION TO THE TRIANGLE. Gl
I suppose that two of the angles of the triangle are given, and I enquire into
the locus of the remaining angle. To fix the ideas, let A, B, G he the angles of the
triangle, A' t B', C' those of the reciprocal triangle; and let the angles A and E be
given, We have to find the locus of the point G': I observe however, that the lines
AA', BB', GG' meet in a point 0, and I conduct the investigation in such manner
as to obtain simultaneously the loci of the two points and 0. The lines C'B', G'A'
are the polars of A, B respectively, let their equations he 3 = 0, and i/=0, and let the
equation of the line AB be 2 = 0; this being so, the equation of the given conic will
be of the form
(a, b, c, 0, 0, fc
I take (a, (3, 7) for the coordinates of and (a?, y, 2) for those of G'; the
coordinates of either of these points being of course deducible from those of the other.
Observing that the inverse coefficients are
(6c, co, ab h 2 , 0, 0, oA),
we find
coordinates of A are ( b,  h, 0),
B (h, a, 0).
The points A' and B' are then given as the intersections of AO with 6"^'(y=0) and
of BO with C'S'Cw^ )') we find
coordinates of A' are (Aa + && , fcy),
5' ( , act + Aft A?),
Moreover, coordinates of 0' are (0, 0, 1),
C^ { w it %\
<J W !/ z )'
The six looints A, B, G, A', ff, G' are to lie in a conic; the equations of the
MA M AT* AY 4 670 aX + hY=Q, Z=Q, and hence the equation of a
lines G.O., dx*, Aj> ^ ie /tArw _ '
conic passing through the ,points G', A, B is
L M ft^Q
ST+Af + WC + 6F ^
Hence, making the oonio pass through the remaining points A', * 0. we find
62
ON A LOCUS IN RELATION TO THE TRIANGLE.
[304
and eliminating the L, M, N, we find
1
a
1
h '
A + 6/9
1
1
A
b '
ct 4 A/9
1
1
Ay
aa + h
y* hie + by '
z
= 0,
or developing and reducing, this is
h cwB
A Aa + 6y
1 a + A/9 _ 1 Aa + & _
As? + 6 ~ '
We have still to find the relation between (a, ft 7 ) and (a;, y, *); this is obtained
by the consideration that the line A'ff, through the two points A', 7/ tho coordinntOH
of which are known in terms of (a, ft 7 ), is the polar of the point 0, tho coordinates
of which are fa y, z). The equation of A'B' is thus obtained in the two forms
and
+ (Aft! +
and comparing these, we have
: y : z = a.
or what is the same thing
chy
(ate + %) (Jig + by)
eliminate
where
_
/m& ^ A a
or, completing the elimination,
wtioh is a quartic curve having a node at each of the oi
points
394] ON A LOCUS IN RELATION TO THE TRIANGLE. 63
that is, at each of the points B t A, 6'. The righthand side of the foregoing
equation, is
=  (ab  7i 3 ) (ha, aft,
so that the equation may also be written
(am + hyf (hx + %) a + eAV f oo; a + i/ + f xy) = 0,
\ "'I
Secondly, to eliminate the (as, y, z) t we have
obh*Y 2 1 w
_
h ah<t + l3 b C '
where
7_ _ chf
or, completing the elimination,
(a&  A 3 ) <% 3 = (A6,  06, Aa$aa + A/3, Aa +
 (06  /i a ) A f ao? + 6^8" + ^ *P
\ ii
that is
fa, 6, o,0, 0,^/3,7)^0.
\ ft A /
Writing (a 1 , y, *) In place of (a, /9, 7), the locus of the point is the conic
Ub,o t 0, 0, f]Uy, *Wo,
\ /t A /
which is a conic intersecting the Absolute
(a, 6, o t 0, 0, AJ, T/, ^) 3 = 0,
ali its intersections with the lines ffi=0, i/ = 0, that is the lines C'B' and (TA 1 .
In regard to this new conic, the coordinates of the pole of C'B'(v = Q) are at
once found to be (~h, a, 0), that is, the pole of G'B' is B ; and similarly the coordi
nates of the pole of Cf'A'fo0) are (b, A, 0), that is, the pole of ^ IB A. Wo
may consequently construct the conic the locus of 0, via given the Absolute and the
points A and B, we have O'A' the polar of , meeting the Absolute m two points
L, th\ and O'ff the polar of 4 meeting the Absolute m the pomts (b, and b,) the
lines O'A' and 0'*' meet in G'. This being so, the required come puses through the
points ou .. fti. &, the tangents at these points being Ah. **,& Bb> JP^^
eight conditions, five of which would bo sufficient to determine the conic It * to ta
remarked that the lines 0'S t O'A! (which in regard to the Absolute are h epotai of
A, B respectively) are in regard to the required come the polars of B, A respectively.
The conie the locus of being known, the point may be taken at any point
of thie conic, and then we have A' as the intersection of O'A w.th AO B as h,
interaeetion of O'S? with BQ> and finally, G as the pole of the line AB in legaid
64 ON A LOCUS IN RELATION TO THE TBI ANGLE. [[394
to the Absolute, the point so obtained being a point on the line G'O, To each position
of on the conic locus, there corresponds of course a position of G\ the locus of
is, as has been shown, a quartic curve having a node at each of the points G' t A, 7?.
The foregoing conclusions apply of course to spherical figures ; we see therefore that
on the sphere the locus of a point such that the perpendiculars let fall on the aides of a
given spherical triangle have their feet in a line (great circle), is a spherical cubic. If,
however, the spherical triangle is such that the angles thereof and the poles of the sides
(or, what is the same thing, the angles of the polar triangle) lie on a spherical conic ;
then the cubic locus breaks up into a line (great circle), which is in fact the circle
having for its pole the point of intersection of the perpendiculars from the angles of tho
triangle on the opposite sides respectively, and into the beforementioned spherical conk.
Assuming that the angles A and B are given, the abovementioned construction, by
means of the point 0, is applicable to the determination of tho locus of the remaining
angle G, in order that the spherical triangle ABO may bo such that the angles and
the poles of the sides lie on the same spherical conic, but this requires somo further
developments. The lines Q'E>, G'A' which are the polars of the given angles ^ J1
respectively, are the cyclic arcs of the conic the locus of 0, or say for shortness tho
come 0 and moreover these same lines O'ff, G'A' are in regard to the conio O,
the polars of the angles S, A respectively. If instead of the conic we consider tho
polar conic it follows that A, B are the foci, and G'A', O'ff the oompondimr
duectrices of tiie come 0'. The distance of the directrix G'A' from the centre of tho
conic measuring such distance along the transverse axis is clearly  00"  distance of
' '
hn
boft'tfn T Vei ;r T 1S =9 ; that i8 ' the C0nic & is ft co described
0* l! ' , " aUSV r ^ (01> SUm r dUteMM of *" fci diBtomooH)
90. Considering any tangent whatever of this conic, the pole of the tangent is n
^ ^ ^ f interSeCti011 " *' P^onS lo
ta tiofl  ' P^on
from the angles of the spherical triangle on the opposite sides; hence to
(viz.
i , ~ " "" v jjuino. luiiujyitj ne in a snhor
mly to construct as before the conic 0' with th f A H "
SArrjs t j^^^i'itrr*!* ^
f the conic 0'. It i s moreovpr nW ^\ opposite sides respectively, is a taragoiit
 10 uiuieuvei Cieai, tnat ffivon n fnionnln ^ nn i ,1
i question, if with the foci AS d niangie^^^a having the property
7 in like manner with the foci A *P nn^r^ 88 aX1S ==9 W desci ' ibe a conic, ctnd
?, 0' and the same transverse axis\ve 1 h T tr ^ SVOrse axis ' aud with th foci
onics will have a common tan^nt the Tol! wL P Tu^ f fln that th threb
action of the perpendiculars fmm +i, i * wneieoi will be the point of infcer
B spectively, P IpentllCUlaiS flom the "^ of the triangle AEG on the opposite siclo*
395.
INVESTIGATIONS IN CONNEXION WITH CASEY'S EQUATION.
[From fcho Quarterly Journal of Pure and Applied Mathematics, vol. vm. (1867),
pp. 384341.]
IN 1,1 paper road April 9, I860, and recently published in the Proceedings of the
Royal Irish Aoad&niy, Mr Casey has given in a very elegant form the equation of a
pair of oh'oloH touching each of throe given circles, viz. if U = Q, F = 0, TP = be
bho equations of the three given circles respoctively, and if considering the common
(iangonlM of (F = 0, lf=0), of (1F=0, U=Q), and of (ZJ^O, 7=0) respectively, these
oommoH tangents being Huoh that tho centres of similitude through which they
roHpuotivoly posH Ho iu a lino (vis;, the tangents are a.11 three direct, or one is direct
and the other two arc inverse), then if /, g, h are the lengths of tlio tangents in
intention, bho equation
belongs to ti pair of circles, each of thorn touching the three given circles, (There
mil, it IH olenr, four combinations of tangents, and the theorem gives ther