the parameters are taken to be contained in the equation of the curve homogeneously , their the parameters before made use of are in fact the ratios of these homogeneous parameters ; and using the term henceforward as referring to the homogeneous parameters , the numbers of the parameters will be =w+L .
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 betveen the parameters ; and the condition that the curve shall pass through given points , a I*fold lined relation between the parameters .
It follows that the number of the curves which satisfy a given Mfold condition , and besides pass through X-D given points , is equal to the order of the k-fold relation , or of the corresponding fold locus ; and thus if we define the order of the / fold condition to be 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+ 1+ &amp;c. )fold condition made up of a I-fold condition , a 1-fold condition , &amp;c. , is equal to the product of the orders of the component conditions ; and in particular if k+1 ?
&amp;c. then the order of the ev-fold 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 parametric point is then situate on a onefold locu .
It is to be noticed in reference hereto 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 one-fold locus of the parametric point ; this is , however , a case where the special locus is of the same manifoldness as the general locus ( ante , No. 8 ) , and is consequently separable ; throning 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 " 1 contact-locus of the given curve .
To each -point of the contact-locus there corresponds a curve having with the given curve a two-pointic 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 ( yr ich is a higher speciality ) in that the sought for curvre is or contains as part of itself o or more coincident curves ( ante , No. 3 ) .
To a point in general on the contact-locus there corresponds a curve having a proper contact with the given curve , save and except that to each point on any one of certain special loci on the contact-locus there corresponds a curve having some kind of special contact as above with the given curve .
To fix the ideas , it may be mentioned that for the curves of the order r which touch a given curve of the order rn and class i , the order of the contactlocus is zzn+ ( 2r4-.2)n If , then , the curve touch a given curve , the parametric point is situate on the contact-locus of that curve .
If it touch a second given curve , the parametric point is in like manner situate on the contact-locus of the second given curve , that is , it is situate on the twofold locus which is the intersection of the two contact-loci ; 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 , any higher order , or has two or more ordinary or other contacts with the same given curve , then if the total manifoldness be =k , the parametric point is situate on a k-fold locus , which is given as a singular locus of the proper kind on the onefold contact-locus ; so that the theory of the contact-locus 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 last-mentioned general case depends on the discussion of the singular loci which lie on the contact-locus .
And similarly , if the curve has any number of ordinary or other contacts with each of two or more given curves , we ' have here to consider the intersections of singular loci lying on the contact-loci which correspond to the several given curves respectively , or , what is the same thing , to the singular loci on the intersection of these contact-loci ; that is , the theory depends on that of the contact-loci which belong to the given curves respectively .
15 .
Suppose that the curve which has to satisfy given conditions is a line ; the equation is ax+by+cz=O , 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 line 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 ( a , 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 line is that it shall touch a given curve , the locus of the parametric point is a curve , the contact-locus ; ( this is in fact the ordinary theory of geometrical reciprocity , the locus in question being the reciprocal of the given curve ; ) and the case 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 contact-locus , 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 inflexiontangent , 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 , c , f , g , Ahx , y , Z)2 0 , we have here six parameters ( a , b , c , f , g , h ) , which are taken as the coordinates of a point in 5-dimensional space .
It may be remarked that in this 5-dimensional space we have the onefold cubic locus abef2-6g2 h2+ 2fgh=_ 0 , which is such that to any position of the parametric point upon it there corresponds not a proper conic but a line-pair ; this may be called the discriminant-locus .
We have also the threefold locus the relation of which is expressed by the six equations ( be-f2=O , ca-g2=O , ab-h2=O , gh-af=O , hf-bg=O , fyg-h=O ) , which is such that to any position of the parametric point thereon , there corresponds not a proper conic but a coincident line-pair .
I call this the Bipoint-locus * , 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 , l* a , b , c , of , Y , hi ) = 0 , ( *'3a , b , c , f , y , h ) =0 ; the equations of the locus are satisfied by a : 6 : C:f : g : hA=2 : 12 : r : fie :Yar : a ( where cc : j3 : y are arbitrary ) ; and substituting these values in the linear relations , we have two quadric equations in ( e , ( 3 , y ) , giving four values of the set of ratios ( o6 : ( 3 : Y ) ; that is , the order is =4 , or the Bipoint is a threefold quadric locus .
17 .
The discriminant-locus 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 line-pair ) ; thus for the conics which have a node and touch a given curve ( m , n ) , or , what is the same thing , for the line-pairs which touch a given curve ( mn , n ) , the parametric point is here situate on a twofold locus , the intersection of the discriminant-locus with the contact-locus .
It may be noticed that this twofold locus is of the order 3(n + 2rn ) , but that it breaks up into a twofold locus of the order 3n , which gives the proper solutions ; viz. the nodal conics 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 conic has 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 consideration of the Bipoint-locus 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 contact-locus ( a , b , c , f , g , h ) =O ( to fix the ideas , if the given curve is of the order qn and class n , then the order q of the contactlocus is =n ?
2r ) .
The contact-locus of any given curve whatever passes through the Bipoint-locus ; in fact to each point of the Bipoint-locus there corresponds a coincident line-pair , 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 contact at each of the points of intersection of the given curve with the coincident line-pair regarded as a single line , that is , in the case of a given curve of the qn-th order , mn special contacts , the Bipoint-locus is a multiple curve on the corresponding contact-locus .
*t In framning the epithet XBipoint , the coincident l ; ine-pair , is regarded as being really a point-pair : see Jost , N o. 30 .
If the conic has simply to touch a given curve of the order m , and class n , , then the order of the condition ( or number of the conics which satisfy the condition , and besides pass through four given points ) is equal to the order of the contact-locus , that is , it is =nl 2rnl If the conic has also to touch a second given curve of the order m2 and class n2 , then the order of the twofold condition ( or number of the conics 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 contact-loci ; and these being of the orders n , + 2m , and n2+ 2m2 respectively , the order of the intersection and therefore that of the twofold condition is =(n1+2ml)(n2+2m2 ) .
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 ( MI , n , ) , ( fi,2 , n2 ) , ( gna , n , ) respectively , we have here three contact-loci of the orders ( n , +2m , ) , n2 + 2m , n,3+2in3 respectively ; .these intersect in a threefold locus , but since each of the contact-loci passes through the threefold Bipoint-locus , this is part of the intersection of the three contact-loci ; and not only so , but inasmuch as they pass through the Bipoint-locus in , , in2 , in times respectively , the Bipoint-locus must be counted m , m2M .
times , and its order being =4 the intersection of the contact-locus is made up of the Bipoint reckoning as a threefold locus of the order 4rnlmnqn , and of a residual threefold locus of the order ( nl + 2m1)(n2+ 2mn2)(n3 + 2mf ) 4mim2 , r3 , =nln23+2(nln2nm+ &amp;c. ) + 4(nm2rm3+ &amp;c. ) + 4m1r2n3 , ; 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 conics touching four or five given curves , the same principles are applicable ; the contact-loci 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 conics 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 containing the six parameters ( a , b , e , f , g ) h ) ; but if the question as originally stated relates only to a class of conics 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 can be brought to the form in question , then in the latter case we may employ the equation in such reduced form , attending only to the remaining conditions ; andin either case we have the equation of a conic containing linearly 2 , 3 , 4 , or 5 parameters which parameters are taken as the coordinates of a point in 1- , 23- , or 4-dimensional space , and the discussion relates to loci in such dimensional space .
This is in fact what is done in Annex No. 2 MDcccLxvIu .
above referred to , where the conics considered being the conics which pass through three given points , the equation is taken to befyz+gzx+hixy=O , and we have only the three parameters ( f , q , h ) ; and also in Annex No. 3 , where the conics pass through two given points , and are represented by an equation containing the four parameters ( a , b , c , h ) : I give this Annex as a somewhat 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 , which 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 ==(r+1)(r+2 ) , and the space therefore at most I r(r+ 3)dimensional ; but we may in particular cases have W+I parameters , the coordinates of a point in c-dimensional space , where W is any number less than -1r(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 condition , and to show that , as a rule ( though as a rule subject to very frequent exceptions ) , the order of a compound condition is equal to 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 formule of DE JONQUIRES , post , No. 74 et seq. But I now proceed to a different part of the general subject .
PReproduction and Development of the Researches of CHASLES and ZEUTHEN .
Article Nos. 24 to 72 .
24 .
The leading points of CIIASLES'S theory are as follows : he considers the conics which satisfy four conditions ( 4X ) , and establishes the notion of the characteristics ( peh v ) of such a system , viz. p , =(4X ) , denotes the number of conics in the system which pass through a given ( arbitrary ) point , and v , =(4X/ ) , the number of conics in the system which touch a given ( arbitrary ) line .
We may say that pb is the parametric order , and v the parametric class of the system .
25 .
The conics ( :))( ) , ( 1 ) ( -111 ) , ( HH1 ) W which pass through four given points , or which pass through three given points and touch a given line , &amp;c. , ... or touch four given lines , have respectively the characteristice ( 1 , 2 ) , ( 2 , 4 ) , ( 4 , 4 ) , ( 4 , 2 ) , ( 2 , 1 ) .
26 .
A single condition ( X ) imposed upon a conic has two representative numbers , or simply representatives , ( ) , X3 ) ; viz. if ( 4Z ) be an arbitrary system of four conditions , and ( c , v ) the characteristics of ( 4Z ) , then the number of the conics which satisfy the five conditions ( X , 4Z ) isAaG+P7N As an instance of the use of the characteristics , if X , X ' , X " , , X " ' , X " " be any five independent conditions , and ( cc , j3 ) , ... ( cii " ' , j3ls ) the representatives of these conditions respectively , then the number of the conics which satisfy the five conditions ( X , X ' , X " , X " ' , X""1 ) is --=(15 2 , 4 , 4 , 2 , 1]tc ( oc , plll ( 3)(W " 131(6 ' 3)(oc " " , s P/ " " ) viz. this notation stands for. .
+1313ii313P/ '13"/ .
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 , in ) , whence the number of the conics which touch each of five given curves ( m , n ) , ... ( in " , nil ) is =(1 , 2 , 4 , 4 , 2 , l ; n , m)(n ' , m')( " , rn")(n " ' , m"')(n"ll , in " " ) .
29 .
A system of conics ( 4X ) having the characteristics ( p , v ) , contains 2P-pu line-pairs , that is , conics each of them a pair of lines ; and 2pi-P point-pairs , that is , conics each of them a pair of points ( coniques infiniinent aplaties ) .
30 .
I stop to further explain these notions of the line-pair and the point-pair ; and also the notion of the line-pair-point .
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 line-pair ( but the class is then= 0 ) : qua curve of the second class it may degenerate into a pair of points , or point-pair ( but the order is then= 0 ) .
The two lines of a line-pair may be coincident , and we have then a coincident line-pair ; such a line-pair ( it must I think be postulated ) ordinarily arises , not from a line-pair 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 line-pair as being in fact the point-pair formed by the two terminating points .
Similarly the two points of a point-pair may be coincident , and we have then a coincident point-pair ; such a point-pair ( it must in like manner be postulated ) ordinarily arises , not from a point-pair the two points of which become coincident , but from a proper conic sharpening itself to coincide with its asymptotes , and so becoming ultimately a pair of lines through the coincident point-pair ; and the coincident point-pair is regarded as being in fact the line-pair 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 point-pair as a line terminated by two , points on this line , and similarly to speak of a line-pair as a point terminated ( that is , the pencil of lines through the point is terminated ) by two lines through the point .
32 .
If in a point-pair thus considered as a line terminated by two points the two points become coincident ( the line continuing to exist as a definite line ) , or , what is the same thing , if in a line-pair thus considered as a point terminated by two lines , the two lines become coincident ( the point continuing to exist as a definite point ) , we have a " line-pair-point ; " viz. this is at once a coincident line-pair and a coincident point-pair ; it may also be regarded as the limit of a conic the axes 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 line-pair-point first presents itself in ZEUTHEN'S researches , as will presently appear ; but it may be noticed here that line-pair-points , and these the same line-pair-points , may present themselves among the 2v -d line-pairs , and among the 2 , -v point-pairs of the system of conics 4X .
33 .
Returning to the foregoing theory of characteristics , I remark that the fundamental notion may be taken to be , not the characteristics ( p , v ) of the conics which satisfy four conditions , but in every case the number of the conics which satisfy five conditions .
Thus for the conics not subjected to any condition , we may consider the symbols ( : )( : 1)GH*/ )S ( :/ // ) , ( / /)9(/ // /)5 denoting the number of the conics which pass through five given points , or which pass through four given points and touch a given line , &amp;c. or which touch five given lines ; these numbers are respectively =1 2 , 4 , 4 2 , 1 .
So for the conics which satisfy a given condition X , or two conditions 2X , ... , or five conditions 5X , we have respectively the numbers 3X , ( : ) , ( .7 ) , ( / 1 ) 4X , ( * ) , ( 7 ) 5X , where the X , 2X , &amp;c. belong to the symbols which follow : read ( X : : ) , ( X. ' .
/ ) , &amp;c. , or , as we may for shortness represent them , PiI II ( 1* Y , viz. the single condition X has the five characteristics ( , " , ... 7 " ' ) , ... ; the four conditions 4X , the characteristics ( ( , v ) as in the original theory ; and the five conditions 5X a single characteristic pi .
34 .
We thus see the origin of the notion of the representatives ( cG , 3 ) of a single condition X ; for considering the arbitrary four conditions 4Z , the characteristics whereof are ( p , v ) , and assuming that the single characteristic , or number of the conics ( X , 4Z ) , is -[-j}v , and taking for ( 4Z ) successively the conditions having respectively the characteristics ( 1 , 2 ) , ( 2 , 4 ) , ( 4 , 4 ) ( 4 , 2 ) ( 2 , 1 ) , we have Yfl/ =2a+4P , c"'=4ca+43 , - " 4a2-+ lf3 , that is , the characteristics ( LW " , Y " ' , gi " , o " ' , 7 " ) of a single condition X are not independent , but are representable as above by means of two independent quantities ( z , 3 ) ; or , what is the same thing , we have Pill 2p,4o , fi 1= 2 3(t + lc " ' ) , which being satisfied , the representatives ( a , / 3 ) are given by a=1 ( 2 , " A/ " ) , 3 ( 2 35 .
I find that a like property exists as to the characteristics ( p " " , 9 ' , &lt; , c " ) of the two conditions 2X , viz. these are not independent but are connected by a single linear relation , If_ 3y If+ 3jft-d ' O. This may be proved in the case where the conditions 2X are two separate conditions ( X , X ' ) ; viz. let the representatives of these be ( a , / 3 ) , ( ce ' , 3 ' ) respectively , then combining with them the three arbitrary conditions X " , X " ' , X " ' having respectively the representatives ( ao " , 3 " ' ) , ( at " , 13 " ' ) , ( oe ' " , / 3f " ' ) , we have the general equation ( X , X ' , X " , IX " ' , X"")-(1 , 2 , 4 , 4 , 2 , 13]oc , P)(dj ' P')(as ' ( 3"( f ' , f")(aflPfill ' ) ; taking herein successively , and observing that the representatives of ( * ) are ( 1 , 0 ) and those of ( / ) are ( 0 , 1 ) , we thus obtain for ( ai " , 9 ' , p " , d ' ) , characteristics of ( X , X ' ) , the values K-"(4 , 2 , 4lXe , 13)(Q ' , 13 ' ) , v-2,4 , 43(oc ) p3)(oc 1 ' ( ='(4I 4,1 21cc , p)(m % , 3 ( viz. Pi'=laz'+2Q43'+ ci)+4f3 , &amp;c. ) , and these values give identically 2p'"3yv " + 3g " 2a " =0 , which is the foregoing equation .
And I assume that the theorem extends to the case of two inseparable conditions 2X , but in this case I do not even know where the proof is to be sought for .
The characteristics ( pW , v ' , I ' ) of the three conditions 3X are in general independent .
36 .
It has been mentioned that if ( ce , 13 ) are the representatives of the condition X , and ( p , v ) the characteristics of the conditions 4Z , then ( X , 4Z)=apj+13v ; this is the most convenient form of the theorem , but as ( ex , 3 ) are known functions of the characteristics ( P " ' , vf ' , go " , " , a.11 ) of the condition X , the equation is in effect an expression for ( X , 4Z ) in terms of the characteristics of X and 4Z respectively .
There is , similarly , an expression for ( 2X , 3Z ) in terms of the characteristics ( p'W , v ' , I ' , a ' ) of 3Z ( satisfying the relation +---lv g'-a ' = 0 ) and the characteristics v , a ) of 2X , viz. we have ( 2X , 3Z ) , ( 4+ )+ ` 8MJ +1 6 &gt; +('( iWIJ This may be easily proved in the case where the conditions 2X are two separable conditions X , XI having the representatives ( ox , P ) , ( oa ' , 13 ' ) respectively , and the conditions 3Z three separable conditions Z " , Z " ' , Z " ' having the representatives ( do , PI3 ) , ( dx'1)((3 " ' ) , ( Wx " " , WY I ) respectively ; we have , in fact , ( IJ=(l , 2 , 4(ca , f3)(ex ' , P3 ' ) , pi ( 1 , 2 , 4 , 43](f , j3)(oJ"I , j3511)( eX11 , 1111 ) , Vt=(2 4 , 4 , , , v 2 4 , 4 , 2 , , , ) ?
I =(4 , 4 , 2 : , )9 g(4 , 4,2 , 1( , , , , , , d =(4 , 2 , 1 , , , , ) ; and with these values the function ( X , X ' , Z " , Z " ' , Z " " ) , =(1 , 2 , 4 , 4 , 2 , 12cc , / ( 3)(ex ' , ( 3')(0J ' , / 3")( " ' , 13"')(Q " ' 131111 ) is found to be expressible as above in terms of ( 1 , v , I ) , ( pi " , v , g ' , It ) ; 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 that is , ( K , a)(1 , 2 , 4 ) , ( 2 , 4 , 4 ) , ( 4 , 4 , 2 ) , ( 4 , 4 , 1 ) , we have in the first case ( 2X.- .
) --4'= -4 8 +8~ @-4 =~d + Q ' , ' --p 32 a't ) , Ad~ ' , and similarly in the other three cases , ( 2X / /=V ' ( 2X-/ /)=e5 ( 2X/ // )=dl38 .
Let Qo , v , i , a ) be the characteristics of 2Z , ( a-3+ 0 ) , and ( d , ' , , i ' , a ' ) the characteristics of 2X , ( I3f +3 '-o= 0 ) .
Then in the formula for ( 2X , 3Z ) , writing successively for 3X ( 2X ) , characteristics ( pt , Y , g ) , and ( 2X/ ) , , ( , i5 ) we obtain expressions for the characteristics ( 2X , 2Z .
) and ( 2X , 2Z/ ) of ( 2X , 2Z ) , viz. eliminating from the formulae , first the ( ao , o ' ) and secondly the ( p , py ' ) , each of these may be expressed in two different forms as follows : ( 2X , 2Z .
) ( 2X , 2Z/ ) = 17 I=1 8~ ~~~~~~~~~~-1 +{ +7V 4(V+ 4(+ +Zt ) 4g 2 d= aid -4V 4( +eb ) -2d+ ' ) | 4+ 2(va+v0 ) the two expressions of the same quantity being of course equivalent in virtue of the relations between ( ph , v , 6 , a ) and ( pI ' , ' ' , ?
' , a ' ) respectively .
The characteristics of ( X , Z ) , ( X , 2Z ) , ( X , 3Z ) are at once deducible from the beforementioned expression ap+py of ( X , 4Z ) .
39 .
ZEUTHEN'S investigations are based upon the before-mentioned theorem , that in a system of conics ( 4X ) , characteristics ( p , v ) , there are y point-pairs and 2v-p linepairs .
If in the given system the number of point-pairs is=X and the number of linepairs is = , then , conversely , the characteristics of the system are And by means of this formula he investigates the characteristics of the several systems of conics which satisfy four conditions ( 4X ) of contact with a given curve or curves , viz. these are the conics ( 2)(1)(1 ) , ( 2)(1 , 1 ) ( 2 , 1)(1 ) 5 ( 23 1 , 1 ) ( 2)(2 ) , ( 2,2 ) ( 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 8 nodes , x cusps , r double tangents , and * inflexions ( , n , , i ; &amp;c. as the case may be ) .
The symbols ( 1 ) &amp;c. might be referred to the corresponding curves by a suffix ; thus ( 1)m would denote that the contact is with a given curve of the order m ( class n , &amp;c. ) ; but this is in general unnecessary .
40 .
In a system of conics satisfying four conditions of contact , as above , it is comparatively easy to see what are the point-pairs and line-pairs in these several systems respectively ; but in order to find the values of X and w , each of these point-pairs and line-pairs 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 point-pairs and the reciprocal classes of line-pairs .
A point-pair may be described ( ante , No. 31 ) as a terminated line , and a line-pair as a terminated point ; and we have first the following point-pairs viz. A , line terminated each way in the intersection of two curves or of a curve with itself ( node ) .
B , tangent to a curve , terminated in a curve , and in the intersection of two curves or of a curve with itself .
C , 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 line-pairs , 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 .
C ' , intersection of two curves , or of a curve with itself ( node ) , terminated each way by the tangent to a curve .
VD , cusp of a curve terminated each way by the tangent to a curve : A= : mmj1 , 1 A'= rnn1 , B= ( n -4)m1 + inf(m-2 ) , 2 B'= 'r(m-4)n,1+rm , ( n-2 ) , i-+rnm1(n--2)(rn3 ) , +nnj(m-2)(n-3 ) , C= 4(m-4)mj1 +nn.-2(mrn-2)(m3 ) , 4 C'= S(n-4)n , +mm1J(n-2)(n3 ) , D= ( m-3)m , , 3 IY= z(n-3)n , .
1 )15 )l =(I , 1 , 1 , I)M .
As 3t(-1 ) , 1 A'= 2zf( 1 ) , B= B(n-4)(m-4 ) , 2 B'= r(mn-4)(n-4 ) , Cr.2(m-4)(m-5 ) , 3 C'= n 4)(n-5 ) , D= : .
( m-3)(m , -4 ) .
4 D'= z2(n-3)(n-4 ) .
43 .
Secondly , we have the point-pairs : E , tangent to curve from intersection of two curves or of a curve with itself ( node ) , and terminated at the point of contact and the last-mentioned point .
F , tangent to a curve at intersection with another curve or with itself , and terminated there and at a curve .
Go common tangent of two curves or double tangent of a curve , terminated at one of the points of contact and at a curve .
D , ut supra .
H , line joining cusp of a curve with intersection of two curves or of a curve with itself , and terminated at these points .
I , 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 curve and the corresponding line-pairs , 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 supra ' .
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 .
I ' , 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 , Thirdly , we have the point pairs : 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 line-pair-point .
0 , cuspidal tangent terminated each way at cusp , viz. this is a line-pair-point : and the corresponding line-pairs : K ' , intersection of two curves or of curve with itself ( node ) , and terminated by the two tangents .
L ' , intersection of inflexion tangent of a curve with a curve , and terminated by the inflexion tangent and the tangent at the intersection .
Ml , intersection of inflexion tangent of a curve with inflexion .
tangent of a curve , and terminated by the two inflexion tangents .
N ' , =30 , ine-pair-point as above .
O ' , =N , Iine-pair-point as above : which being so , we have ( 2)(2)~ =2 ) .
( 2)m , K -nn , , K ' =mrn1 , L =xnI+x1n , 3 L'-=m1+:rI , ( 2 , 2 ) , =(2 , 2 ) ... K =,9 K ' =L x(n3 ) , M= ; ( xI ) , M=1 M'=:(I - ) N,2 N'= , 0-= .
1 0'= 45 .
Fourthly , we have the point-pairs P , tangent of a curve at its intersection with another curve or itself , terminated each way at the point of contact-line-pair-point .
Q , common tangent of two curves or double tangent of a curve , terminated each way at one of the points of contact-line-pair-point .
J , Ut supra .
R , cuspidal tangent terminated at cusp and at a curve : and the corresponding line-pairs : to which may be joined ( Z)(4X ) a(4X -)+b(4X/ ) , where a , b are the representatives of the condition ( Z ) , and where ( 4X ) is to be considered as standing successively for ( 4 ) , ( 3 , 1 ) , ( 2 , 2 ) , ( 2,1 , 1 ) , and ( 1 , 1 , 1 , 1 ) , the values of ( 4X ) and ( 4X/ ) being in each case given by the foregoing Table .
52 .
The formulae are very convenient for the calculation of the numbers of the conics which satisfy five conditions of contact with two given curves ; thus if , for example , ( 3Z ) , =(3)m1 , denotes the condition of a contact of the third order with a given curve ( M , ) then writing for symmetry ( 2)m in place of ( 2 ) , we have ( 3 ) , l(2 )= Ia( 3/ = a(- ' 4mf1-4n + 3a51.)53 .
To obtain the foregoing expressions of ( 5 ) , ( 4 , 1 ) , ( 3 , 2 ) , ( 3 , 1 , 1 ) , ( 2 , 2 , 1 ) , ( 2 , 1 , 1 , 1 ) , and ( 1 , 1 , 1 , 1 , 1 ) , I assume that the given curve breaks up into two curves ( in , n , a ) and ( en ' , n ' , o ' ) , or , as we may for shortness express it , into two curves m and mn ' .
We have then ( 5)m+m ' ( 5)m + ( 5)m ' , viz , the conics which have contact of the 5th order with the aggregate curve m+ m ' are made up of the conics which have this contact with the curve m and the conics which have this contact with the curve in ' .
Writing this under the form ( 5)m+m ; ( 5)M-(5)M ' ?
0 and observing that ( 5)m is a function ; ( m , n , a ) , and that consequently this is a functional equation ( r+m ' , n-+n ' , +')-@(in , a ) o)-p(m ' , n ' , a')=O , the solution is p(mn , n , a)=ain+bn+cce , where a , b , c are arbitrary constants ; but as the solution should be symmetrical in regard to in , n , we have a-b , or the solution is 4(n , n , a(in+nj+c .
54 .
Similarly we have ( 4 , 1)m+m'(4 , 1)m ( 4 , 1)m'=(4)m(1)m , +(4mi)(1)m , viz. the conics which have with the aggregate curve m+m ' the contacts ( 4 , 1 ) are made up of the contacts which have the two contacts 4 and 1 with the one curve or with the other curve , or the contact 4 with the one curve and the contact 1 with the other curve .
The expression on the right-hand side is a known function of ( in , n , a ) , ( inm , n ' , ac ) ; hence the form of the functional equation is p(m+ no , i+n ' , n '-9( , as o)-O5(Xm ' , n ' )-F(m , na , on ' , 9 ' a ) and any particular solution of this equation being obtained , the general solution is found by adding to it the term ain+bn+cx .
Assuming that the particular solution is symmetrical in regard to ( in , n ) , then the term to be added is as before=a(mn+n)+cac .
And similarly for ( 3 , 2 ) , ( 3 , 1 , 1 ) , &amp;c. ; that is , in every case we have a solution containing two arbitrary constants a , c , which remain to be determined .
MDcCoL~xvJIJ .
Q Now in every case except ( 5)m the number of intersections of the conic with the curve is &gt; 6 ( viz. for ( 4 , 1)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 given curve m be a cubic , the number of conics satisfying the prescribed conditions is = 0 ; and since a cubic may be the general cubic or a nodal or a cuspidal cubic , we have the three cases ( in , n , a)=(3 , 6 , 18 ) , ( 3 , 4 , 12 ) , and ( 3 , 3 , 10 ) .
We have thus in each case three conditions for the determination of the constants a , c ; so that there is in each case a verification of the resulting formula .
56 .
In the omitted case ( 5)mw when the curve in is a cubic , the theory of the conics ( 5)m is a known one , viz. the points of contact of these conics , or the " 1 sextactic " points of the cubic , are the points of contact of the tangents from the points of inflexion ; the number of the conics ( 5)m is thus = ( n 3 ) : , viz. in the three cases respectively it is =27 , 3 , and 0 .
Hence for determining the constants we have the three equations 9a+ 18c=27 , 7a+12c= 3 , 6a+lOc= 0 , which are satisfied by a_ -15 , c=9 , and the resulting formula is ( 5)= -.-15m-15n+9a .
In the particular case of a curve without nodes or cusps , -this is ( 5)=12n-15mn , in(12m --27 ) , which agrees with the result obtained in my memoir " On the Sextactic Points of a Plane Curve , " Phil. Trans. vol. clv .
( 1865 ) pp. 545-578 .
57 .
The subsidiary results required for the remaining cases " ( 4 , 1 ) , &amp;c. are at once obtained from the foregoing formulae for ( 4Z)(1 ) , ( 3Z)(2 ) , &amp;c. ; for example , we have ( 4)m(1)m i n(-1 O1A 8n+ 6a ) +m'(8m.-lOn+6tz ) , with like expressions for ( 3 , 1)m(l)m , &amp;c. ) ( 3)m(2)mi =1 0C ( -8mn-8n+6c6 , ) , ( Mm(l : &gt; 1 )mI = ln/ 2 , n ) a x-n+36 Jr(m'd-Al-4t )(_ 8m , -8n+6a ) +(1q~2d2 -1m1)(-3m-4n+3a ) ; with like expressions for ( 2 , 1)m(2)m , ( 2 , 1)m(1 , 1)mi , &amp;c. &amp;c. 58 .
Calculation of ( 4,1 ) .
We have ( 4 ) l )m eml(4~ 1).-~(4~ l)wz=(4)m(1)m + ( 4)m , ( 1 )m --16nm ' -20(nn'+inn)-16nd ' +O6Q(an'+ cn ) + 6Q(zn'+ h 'm ) , the integral of which is ( 4 , 1 ?
l= n8ifn2-2Oinn8n~2+a(rn+n ) +os(6in+6n+c ) .
WHICh SATISFY GIVEN CONDITIONS .
101 The particular cases ( n , n , a)=(3 , 6,18 ) , ( 3 , 4,12 ) , ( 3 , 3 , 10 ) give respectively 0=252+9a+18c , 0= 64+7a+12c , 0= 36+6a+10c , satisfied by a= 104 , c=-66 .
59 .
Calculation of ( 3 , 2 ) .
We have ( 3 , 2)m+m , ( 3 , 2)m ( 3 , 2)m ?
( 3)m(2)m , +(3)m , ( 2)m -4(ma ' +m'mc ) 4(na ' +n % ) ? ?
6c ' ; the integral is ( 3 , 2)m=a(m+nn)+a(-4f-4n+c)+3c2 , and , as before , 0=324+9a+18c , 0= 96+7a+12c , 0= 60+6a+lOc , satisfied by a=120 , c_-78 .
60 .
For the calculation of ( 3 , 1 , 1 ) we have similarly ( 35 1 , 1)ffi+m , -(35 1 , 1)m-(35 l , 1 ) m =(3)m(1 , 1)M1+(3)M1(L 1)M + ( 3 , 1)m(1)mi+(3 , 1)mi()m .
The function on the right-hand side was of course calculated from the values of ( 3)m(15 1)mI &amp;c. ; but there is no use in this ( and the more complicated cases which follow ) in actually writing down the values of the function in question ; it can in each case be calculated backwards from the foregoing expressions of ( 3 , 1 , 1 ) &amp;c. , and the values so obtained be verified by actual substitution .
But assuming it to be known , the solution of the functional equation gives of course the foregoing expression for ( 3 , 1 , 1 ) , except that the terms in n+ n and a are therein a(mn+ n)+cmz ; and I shall in this and the subsequent cases give only the three equations which determine the constants .
In the present case these are 332+9a+18c=0 , *-454+7a+12c=0 , 306+6a+10c=0 , satisfied by a=-434 , c-291 .
61 .
Remaining cases ( 2 , 2 , 1 ) , ( 2 , 1 , 1 , 1 ) and ( 1 , 1 , 1 , 1 , 1 ) .
We have ( 2 25 l)tn+m -(25 25 1 ) , w=t ( 25 2)m(I)ml + ( 2 2)w(l)m + ( 2 , 1)m(2)M + ( 2 , l)m , ( 2)M ) and -1674+ 9a+18c=0 , 648+7a+12c=0 , 462+6a+lOc=O , satisfied by a-468 , c-=327 .
Again , ( % 15 1,1)m+ ?
n !
( 2 , 1 , I , )m -(21 1 1,1 )I ( 2,1,1 ) '(1)mr+(2j l , l)m , ( 1 ) +(2,1)m(1 , 1)mt+(2 , 1)mi(1)m ~~~~~~~~~~~+(2)m(Ij1 , 1 , )m , +(2)l(1 I 1 , )W and 5400+9a+18c-=O , 2280 + 7a+12c=0 , 1680 + 6a+10c=0 , satisfied by a 1320 , c 960 ; and finally , ( 1 , 1 , L1 , 1 , )M+Mr-(1j 1 , 1 , 1 , 1 - , I 1 , 1 , )mj= ( 1 , I 1 , 1 , )M(I)M + ( I 1 , 1 , I)7nt(I)m + ( 1 , 1 , l)M(l , l)M , +(l , 1 , l)Mf(il l)M , and -30618+90a+180c=0 , -14094+ 70a+120c= 0 , -10692+60a+120c=0 , satisfied by 10a=6318 , 10c=4860 , that is , a=3 1 59 , 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 ; but 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 ) , &amp;c. are sibireciprocal , the contact at a given point of the curve is the same thing 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 conics which satisfy such conditions as ( 1 ) , ( 2 ) , &amp;c. are obtainable in several ways .
63 .
( 10 ) When the total number of conditions is 4 , the question may be solved by ZEUTHEN'S method , viz. by determining the line-pairs and point-pairs of the system 4Z , with the proper numerical coefficients , and thence deducing the values of the characteristics ( 4Z ) and ( 4Z / ) .
A few cases are in fact thus solved in ZEUTHEN'S work .
64 .
( 20 ) 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 + 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 conics ( 3Z ) ( 1 , 1 ) : writing for shortness in place hereof ( 1 , 1 ) , or ( in order to mark the curve ( in ) to which the symbol has reference ) ( 1 , 1)m , let the curve ( i ) be the aggregate of the curves ( in ) and ( in ' ) .
Regarding the point 1 as a given point on the curve ( nO ) , that is , an arbitrary point in regard to the curve ( in ' ) , we have thus the equation where the right-hand side is known ; and so in general the form of the functional equation is always ( in+m')-.p(m)= given value , that is , p(m+m ' n+n ' , O+cd')-P(m , n , ca ) = given function of ( m , n , za on ' , n ' , da ) ; whence , as stated , the general solution is Particular Solution + Constant .
In the case in hand , taking successively ( 3Z)==( : .
) ( :/ ) , ( / / ) , and ( / // ) , we have in the first of these cases ( 1 , l~m , ,-1 , ).=nl+2m ' , whence ( 1 , 1)m=n+2mn+ const .
=(1 , 1)( : .
) ; and the value of the constant being in any way ascertained to be =-2 , we have ( 1 , 1 ) ( ..)=n+ 2i2 ; and the like for the other three cases .
65 .
( 30 ) The expressions for the number of conics which satisfy such conditions as ( 1 ) , ( 2 ) , &amp;c. are deducible with more or less facility from the corresponding expressions wherein ( 1 ) , ( 2 ) , &amp;c. are replaced by ( ) , ( : ) , &amp;c. ; thus from ( : : 1 ) *=n+ 2n we deduce ( .@1,1)(::/ )-2(.- .
2)=n+2mn-2 , viz. if one of the four arbitrary points of ( ::/ ) becomes a point on the curve , then the condition ( ::/ ) is satisfied specially by the conic ( ..2 ) which passes through the remaining three points and touches the curve at the point in question ; 2 of the conics ( ::/ ) coincide with the conic in question .
We have thus a reduction 2( : .
2 ) , =2 , and the number of the conics ( . .
1 , 1 ) is =n+2mn-2 .
Similarly , we have the system ( . .
1 , 1 )=n+2m-2 , ( : 1,1,1 )=-n+2i-4 , ( * 1 , 1 , 1 , 1)=n+2m-6 , ( 1 , 1 , 15 1 , 1)=n+2m-8 .
Again , two or even three of the given points on the curve may come together without any reduction being thereby caused , that is , we have ( :2 , 1 ) =n+2m-4 , ( 2,1,1 )( 3 , 1 )=n+2in-6 , ( 2 , 1 , 1 , 1)=( 3 , 1 , 1)=n+2mi-8 ; but if the four points on the curve coincide in pairs , or , what is the same thing , if in ( 2 , 1 , 1 , 1 ) the points 1 and 1 come to coincide , then there is a special reduction , and we have ( 2 , 2 , 1)=n+2m-8 [ -(m-2)]1n+n-6 , viz. here ( mn -2 ) of the conics come to coincide with the two points considered as a point-pair or infinitely thin conic .
If the points 2 and 2 come to coincide , that is , if the four given points on the curve all coincide , there is no further reduction , but we have ( 4 , 1)=m+n-6 .
V 66 .
The expressions involving a single ( 1 ) may in every case be reduced by the foregoing method to depend upon other expressions ; thus we have ( 3Z )(1 , 1 ) =(.i ) -2(2 ) ( 2Z ) ( ,2 ) =(-2 ) -3(3 ) , , ( ,1 , )-1 , 1 ) --2(2 , 1 ) ( Z ) ( 1 , 2 ) ( *1,2 ) -2(2 , 2 ) 3(1,3 ) , , , ( ,3 ) ( *3 ) 4(4 ( 1 , 4 ) =(*4 ) -5(5 ) &amp;c. , where , comparing for example the equations for ( Z)(i , 1 , 2 ) and ( 2Z)(i , 1 , 1 ) , it will be observed that in the first case the contacts 1 , 2 of the symbol ( 1 , 1 , 2 ) successively coalesce with the point 1 , giving respectively 2(2 , 2 ) and 3(1 , 3 ) , the exterior factor being in each case the barred number , whereas the second case , where the 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 ) +2(1 , 2 ) , =4(2 , 1 ) ) , but the identical symbol is taken only once , giving 2(2 , 1 ) .
Thus we have also ( 1 , 1 , 1 , 1)=( =1 , ( .e ) =2 , ( 3,1 ) 1 ) n+ 2n----6 , ( / / ) =2n1 6 ; ( 4)( ) 1 , ( I/ ) = 1 ; which are the several cases for the conies which satisfy not more than four conditions , and 69 .
For the conics satisfying 5 conditions , we have ( 5 ) _1 , ( 4,1 ) -m+n-6 , ( 3,2 ) ( 3 , 1 , 1 ) = w+2rmn+ In , n+27-U , ( 2,3 ) -4m-4n-6+3a , ( 2 , 2 , 1 ) = 6m+6n+54+c(Qn+n-15 ) , ( 2,1,1,1 ) -=*m3+m42fl+rnl2+*36 52 2S 3l fn+3 m+ 3n+os( -3n -q3 229 ) ( 1,1 4 ) -10m-8n-5+6 ( 1,11 3 ) Sin q2 _12 , jnn3n+60)n+57n+36+os(6mnT3n-45 ) , ( 1,2 , 2 ) =27m+24n+27-23ca+ ? ?
2 , ( 1 , 1.1 , 2 ) 45-nt2+30mn+ ?
2i2n2 mi 32 ln189 +oa(M2+2mn+In 2 -27qn33 n+ 34 5)3 ( 1 , 1 , 1 , 19 1)=_ 3 +2 3 5+ 124432 7n3 ' 12 1m2_ 5m2n O_49 n2+42 5nX+ 9I 3n+150 +a_ a2c(3 -nn 3n2+ 4n+-27n W ) +aOV2 .
70 .
The given point on the curve to which the symbols 1 , 2 , &amp;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 lzl ; similarly , a conic which touches the curve at the cusp , in fact there meets it in three points , and -I denote the condition by 2X 1 ; 1Xi , 21 are thus special forms of 1 , 2 , and the annexed I indicates the additional point of intersection arising ipso facto from the point 1 or 2 being a cusp .
Similarly , we should have the symbols 34 , 4X1 , 5X1 ; but it is to be observed that at a cusp of the curve there is no proper conic having a higher contact than 2z4 ; thus if the symbol contains 34 , or a fortiori , if it contain 44l or 5ac4 , the number of the conics is in every case -0 ; it is thus only the cases 14z and 2X1 which need to be considered .
71 .
The several modes of investigation which apply to the case of contact at a given ordinary point of the curve are applicable to the case of contact at a cusp : we may if we please employ the functional method ; we have here a functional equation of the foregoing form , p(m+rn')-p mgiven value ( that is , P(m+d ' , n+n ' , )( n , ao ) given function of ( in , n , cz , a ' , n ' , a ' ) , and the general solution is as before -Particular Solution+ Constant ; so that there is in each case a single arbitrary constant to be determined by special considerations .
The determination of the constant is in some instances conveniently effected by means of the case of the cuspidal cubic : see Annexes Nos. 4 and 5 .
The formation of the functional equation itself is similar to that in the corresponding case where the given point on the curve is an ordinary point .
For example , we have ( 2Z)(1 , 1 , 1)m+mi(1 , 1 , 1)M= ( 1 , 1)m(1)m , = nX(1 , 1* )m+i ( 1 1 / )m +(1)Mf(1 ) I)Mf 298 )( n+ ( mn 3a)(1 + ?
2(m2 - ' )(1 / /)m , and we may herein simply change l into 14 .
Writing successively 2Z=( : ) , ( / ) and ( / ) , we find ( / /)m+rn ' ( / )m='(2fl+ 4mg-.6 ) +m2(4n+ 4m3-m6 ) + ( n ' 2 2n')2 + ( m'n'_ 06-')4 + ( irn22TW ' ) 4 , which only differ from the corresponding expressions with 1 in that they contain " n+2m-3 , 2n+4nm-6 , 4n+4mn-6 , 4m+2n-3 in place of n+2m-2 , 2n+4m-4 , 4n+4m-4 , 4m+2n2 repectively , and they lead to the expressions for ( lxl , 1 , 1 : ) , &amp;c. , the arbitrary constant being in each case properly determined .
72 .
We have ( ../ ) =2 , ( :7/ ) =4 , ( 7/ // 7 ) 4 , ( / 7/ 7/ ) =2 ; The remainder of this table , being the part where the symbols ( ) and ( / ) do not occur , I present under a somewhat different form as follows : ( 5z1 ) zzO , ( 3x , 2 ) =0 , ( 3X,1 , 1 ) = 0 , ( 2,3 ) -(2X , 3 ) =0 , ( 2 2 , 1 ) -(2X1 , 2 , 1 ) =n -3 , ( 2,1,1,1 ) -(2Ul , 1 , 1 , 1 ) =2(n-3)(n-4 ) , ( 1,4 ) -(1X1 , 4 ) =1 , 1 , 1 , 3 ) -(1X1 , 1 , 3 ) _(2zl , 3 ) + ( n3)5 ( 1,2,2 ) -(1l , 2 , 2 ) =3(n-3)+x-1 , ( 1,1,1,2 ) -(1x4 , 1,1 , 2 ) =(2xl,1,2)+s(n-3)(n-4)+8+2n-3m-4 , ( 1,1 , ,1,1)(IX1 , 1,1 15 1 ) =(2xl , 1 , 151 ) .
These results relating to a cusp , are useful for the investigations contained in the Second Memoir .
It will be noticed that the symbols which contain 2x are not , like those which contain 2 , symmetrical in regard to ( in , n ) : the interchange of ( in , n ) would of course imply the change .
of a cusp into an inflexion , and would therefore give rise to a new symbol such as 2:1 ; but I have not thought it necessary to consider the formulae which contain this new symbol .
Investiqations in extension of those of DE JONQUIERES in relation to the contacts of a Curve of the order r with a given curve.-Nos .
74 to 93 .
74 .
DE JONQUIEREs has given a formula for the number of curves C ?
of the order r which have with a given curve Um of the mth order t contacts of the orders a , b , c , &amp;c. respectively , which besides pass through p points distributed at pleasure on the curve Urn ( this includes the case of contacts of any orders at given points of the curve Um ) , and satisfy any othr r~+3 ) which moreover satisfy any other ( a+ b+ c+ &amp;c. ) conditions ; viz. the number of the curves Cr is =M(a+l)(b+l)(c+l ) ... into ( [ rin-(a+b+c .
)p ]t I +[rin-(a-+b+c..)-p-1]t(a +b +c..)[D ] +[rin-(a+b+c..)-p-2]'-2(ab+ac+bc..)[D]2 +[rin-(a+b+c..)-p-t]0 ( abc ... where the curve Um is a curve without cusps , and having therefore a deficiency D=-1(m 1)(m-2 ) ; the numbers a , b , c,. .
are assumed to be all of them unequal , but if we have ca of them each=a , / 3 of them each=b , &amp;c. , then the foregoing expression is to be divided by [ c}]cf ] ... ; and p , denotes the number of the curves Cr 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 , &amp;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 curves Cr which satisfy the prescribed conditions , it is sufficient that the Lr(r+ 3)(a 6+ 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 formulk of the present section .
75 .
I will for convenience consider this formula under a somewhat less general form , viz. I will put p= 0 , and moreover assume that the 2r(r+3)(a+b+c. .
) conditions are the conditions of passing through this number of arbitrary points ; whence p=1 .
We have thus a curve C ' having with the given curve Urn contacts of the orders a , b , c. .
respectively , and besides passing through -ir(r+3)-(a+b+c. .
) arbitrary points ; and the number of such curves is by the formula=(a+1)(b+1)(c+1 ) , ... into 7 [ rm-(a+b+c. .
) ]t I +[rm-(a+b+c..)-']t1(a +b +c..)[D ] l +[rm-(a+b+c..)-2]t-2(ab+ac+bc.)[D ] It+[rgn-(a+b+c..)-t ] ?
( abed .
)[D]t where , as before , in the case of any equalities between the numbers a , b , c , ... , the expression is to be divided by [ ex}{f3 ... 76 .
I have succeeded in extending the formula to the case of a curve with cusps : instead of writing down the general formula , I will take successively the cases of a single contact a , two contacts a , b , three contacts a , b , c , &amp;c. ; and then denoting the numbers of the curves Cr by ( a ) , ( a , b ) , ( a , 0 , c ) , &amp;c. in these cases respectively , I say that we have ( a)(a+1)frm-a| t+aD f -a. .
( a , b)= ( a+l)(b+l)F [ rm a-6]2 21 +[rm-a-b-1]'(a+b)[D ] ' 1+ a6 [ D2 J2 r a(b+1)Jfrm-a-b-1 ] bDl + +b(a+1){[rm-a-6-1 ] ' Ill l+ ~~~aD| J+ a6 ... ..Ae ( a , b elG d , e)=(a+1)(b+1)(c+1)(d+1)(e+1 ) l rmM ]5 | [ rm -c_-1]4j [ D ] ' +[rm-c'-2]3 [ D]2 +[rm-c 3]2r[D]3 +[rnn-c'-4]'1 [ D]4 | 1+ s [ D]5J -[2e(a+1)(b+1)(c + 1)(d+1 ) " [ rmlj { +[rm-a 2 2]3]13 [ D ] I +[rm-_-3 ]2Y ' [ D]2 +[rm n-c 4]'f3 [ D]3 |+ Y ' [ D]4 + [ :de(a + )(b + 1(c +1)/ [ rmna -2]3 ][ ] + +[rntM 3 &gt; 1os [ D ] ' y"[D]2 V+ it " [ D ] F[Ecde(a+1)(b+1 ) [ rm-a-3]2 I ][z]3 + +[rm-o 4 ] s'al/ [ D ] ' + [ D + +[bede(a+l ) { [ rm t M-4]1 4"'D ] I+ fit/ [ D]i abode ... . .
[ X. 78 .
In all these formulae there is , as before , a numerical divisor in the case of any equalities among the numbers a , b , c , &amp;c. And D denotes , as before , the deficiency , viz. its value now is D= ( m--1)(n-2)-b-z ; or observing that the class n is= 2-m28-3x , we have D=ln-m+1+lz , or say D=1-n+2n+i , =-1+A if A= -m+ n+ .
79 .
It is to be observed with reference to the applicability of these formula within certain limits only , that the fQrmule are the only formulae 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 C " which have with a given curve Ua contact of the order a , and besides pass through the requisite number 1-r(r+ 3 ) -a of arbitrary points , is that given by the formula , viz. ( a ) ( a+ 1)(rma+aD ) ax .
Considering the curve Um and the order r of the curve Cr as given if a has successively the values 1 , 2 , ... up to a limiting value of a , the formula gives the number of the proper curves C'r which have with the given curve Urn a contact of the required order a : beyond this limiting value the formula no longer gives the number of the proper curves Cr which satisfy the required condition , and it thus ceases to be applicable ; but there is no algebraic function of a which would give the number of the proper curves Cr as well beyond as up to the foregoing limiting value of a. 80 .
The formulae are applicable provided only the conditions include the conditions of passing through a sufficient number of arbitrary points ; viz. when the number of arbitrary points is sufficiently great , it is not possible to satisfy the conditions specially by means of improper curves Cr , being or comprising a pair of coincident curves .
Thus to take a simple example , suppose it is required to find the number of the conics which touch a given curve t times and besides pass through 5t given points : if the number of the given points be 4 or 3 there is no coincident line-pair through the given points , and therefore no coincident line-pair satisfying the given conditions ; if the number of the given points is =2 , then the line joining these points gives a coincident line-pair having at each of its mn intersections with the given curve a special contact therewith , that is , having in " nz(Qn1)(Qn-2)ways three special contacts with the given curve ; if the number of the given points is 1 or 0 , then in the first case any line whatever through the given point , and in the second case any line whatever , regarded as a coincident line-pair , has m special contacts with the given curve ; and so in general there is a certainvalue for the number of given points , for which value the conditions of contact may be satisfied by a determinate number of improper curves Cr , and for values inferior to it the conditions may be satisfied by infinite series of improper curves Cr .
It is by such considerations as these that DE JONQUIERES has determined the minimum value T of the number of arbitrary points to which the conditions should relate in order that the formulae may be applicable : I refer for his investigation and results to paragraphs XVII and XVIII of his memoir .
I remark that in the case where the number of improper solutions is finite , the formula can be corrected so as to give the number of proper solutions by simply 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 in the first part of the present Memoir ; see in particular ante , Nos. 8 , 9 &amp; 10 .
81 .
The expressions for ( a ) , I(a , b ) , &amp;c. may be considered as functions of rm , 1+ A , and z , and they vanish upon writing therein rm= 0 , A=0 , z= 0 ; they are consequently of the form ( rm , A , z)'+(rm , A , z)2+ &amp;c. , and I represent by [ a ] , [ a , b ] , &amp;c. the several terms ( rrn , A , x)1 , which are the portions of ( a ) , ( a , b ) , &amp;c. respectively , linear in rm , A , and z. The terms in question are obtained with great facility ; thus .
to fix the ideas , considering the expressions for ( a , b , c , d),10 .
To obtain the term in rm , we may at once write D=1 , z=O , the expression is thus reduced to -[a , z b , G. d , e ] ( am 1)(b+1)(c+ 1)(d+ 1)(e+1 ) ( oa+ )(a+ 2)(a+ 3)(a+4 ) wn +(a+I)(b+I)(c+:I)(d+I)(e+I)| sc+)+2(+3a+4 A 13 ( xt+2)(oc+3)(oM ?
4 ( o4+3)Q(a+4 ) IV23 ( c%+4)1 + " Xe ( a+ ?
)(b+1)(c+1)(d+1)(e+1 ) ( oC+2)(Q+3)(v-+4 ) 'X + Id ( d+e+2)(a+l)(b+l)(c+l ) Qx+3)(a +4 ) -2:cde ( c+d+e+3)(a+1)(b+1 ) ( %+4 ) + 6 : bed ( b+c+d+e+4)(a+1 ) -24abcde where rl=a++b+c+d+e , f3- &amp;c. , s-abcde .
83 .
The complete functions ( a ) , ( a , b ) , ( a , b , c ) , &amp;c. may be expressed by means of the linear terms [ a ] , [ a , 1 ] , [ a , b , c ] , &amp;c. as follows , viz. we have ( a ) [ a ] ( a , b ) [ a][b ] + [ a , 13 ] , ( a , b , c ) .
[ a][b][c ] + [ a][b , c]+[b][a , c]+c[a , b ] + [ a , 1 , c ] , ( a , b , c , d)[a][b][c][d ] +:X[a ] [ b ] [ c , d ] +:X[a , b ] [ c , d ] +:X[a][b , c , d ] + [ a , b , c , d ] , and so on : this is easily verified for ( a , b ) , and without much difficulty for ( a , b , c ) , but in the succeeding cases the actual verification would be very laborious .
84 .
The theoretical foundation is as follows .
Writing for greater distinctness ( a)m in place of ( a ) , we have ( a)m to denote the number of the curves Cr which have with a given curve Urm a contact of the order a , and which besides pass through 2r(r+3)-a points .
Let the curve Um be the aggregate of two curves of the orders m , m ' respectively , or say Jet the curve Urm be the two curves m , in ' , then we have a functional equation , the solution of which is ( -)-N [ --a-mI where [ a]m is a linear function of n , mn , x , or , what is the same thing , of mn , A , z. I assume for the moment that when the coefficients are determined [ a]m would be found to have the value =[a ] .
Similarly , if ( a , b)m denote the number of the curves CU which have with the given curve Un contacts of the orders a and b respectively , and which besides pass through Wr(r+3)-a-b points then if the given curve break up into the curves in , in ' , then we have ( a , b)mfm , ( a , b)m-(a , b)mt= { ( a)M(b))m } + { ( a)mi(b)m1 } where I(a)m(b)m,4 is the number of the curves Cr which have with en a contact of the order a and with mn a contact of the order b , and which pass through the -1r(r + 3 ) a--b points ; and the like for I ( a)m , ( b)f4 .
Then , not universally , but for values of a and b which are not too great , the order of the aggregate condition is equal to the product of the orders of the component conditions ( ante ' , No. 12 ) , that is , we have { ( a)m ( b ) } =(a)m * ( b ) =[a]m [ b]mi { ( a ) , f(b ) .
}= ( a)mi .
( b)m [ a]mf[b]m , and thence the functional equation ( anb)Mmt-a)b)m.-(a , b)mr = [ a].[b]mfi+ [ a &amp; ]m[b]m_ But [ a ] .
&amp;c. being linear functions of m , A , z , we have [ a]m+m= [ a]m+ [ a ] , nl [ 6]m+mj [ bum + [ bigmi and thence a particular solution of the equation is at once seen to be [ a]m[b6 ] ; the general solution is therefore ( a , b ) ?
, =[a ] , n[b ] , , , -+-[a , b]m , where [ a , b]m is an arbitrary linear function of n , A , z. Hence , assuming for the present that if determined its value would be found to be =l[a , b ] , we have the required formula ( a , b)=[a][b ] +[a , 6 ] .
The investigation of the expression for ( a , b , c)m depends in like manner on the assumption that we have { ( a)m(b , c)mt } =(a)m CrRVES and comparing with the , equation for ( a , b ) , we 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 right-hand side of the equation in virtue of the remark as to the divisor of the functions ( a , b ) , ( a , 6 , e ) ... in the case of any equalities between the numbers ( a , b. .
) This is generally the case , and the foregoing expressions for [ a ) b6 ] [ a , 6 , c ] , - &amp;c. are thus to be regarded as true without modification even in the case of any equalities among the numbers a , b , c ... .
86 .
To complete according to the foregoing method the determination of the expressions for ( a ) , ( a , b ) , .
, we have to determine the linear functions [ a ] , [ a , 6 ] , &amp;c. , which are each of them of the form f +gA +ltz , where ( f g , h ) are functions of r and of a , b , &amp;c. .
; and I observe that the determination can be effected if we know the values of ( a ) , ( a , b ) , &amp;c. 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)=(a+1)(rm-a ) , Writing first A--1 Z0and secondly A=-1 , z=1 , we havy ( a6+1 )(rm a ) =fM a-g ( a+ 1)(rm-a)a=fin-g+h , w~hence ~ff(a +1l)r , get*(a+1l)a , h-=-a giving the-lforegoing value [ a ] =(a+ 1)rn + ( a+ 1)aA ax .
Similarly , for two contacts assume that we have in the two cases respectively ( a , b)=(a+1)(b+1)[rn-a-b- ] , ( a , b)=.(a+l)(b+l)[rm-a-b]2-{a(b+l)+b , ( a+l)}:rm-a-b-1 ] ' Starting here from the formula [ a , 6 ] ( a , b)[a][b ] =fm+gzA +liz , and writing successively A--1 , z=O , and A--1 , z1 , we have , ( a + )(b + ) [ rm !
]_{+1)(rn-a}{bl( b ) Feg ( a~)(bl)rsx66 0 ]_{a(b+l)+b(a+1)j[rm -a-b-I]I { ( ai+l)(rrn-a)-a } { ( b+l)(rm-b)-b } =fin-g+h ; a the first of which putting therein a-j 6=G , ab=43 , is at once reduced to whence ( f a -(a+1)(b ?
1)(t ; + 1)r , g= -(a+1)(+ 1)QM(I ?
1)-j3 ) .
And taking the difference of the two equations , we have { a6 +1f ) --(oM +1)}(rrn-a-6-1 ) +(b+ 1)(mn6 ) + 6(a + 1)(rrn-ad-at =h , that is ho=(a+1l)(6+1)(a+6)-a6 ; whence [ a , 6 ] has the value above assigned to it .
87 .
The actual calculation of [ a , 6 , c ] would be laborious , and that of the subsequent terms still more so ; but it is clear that the principle applies , and that the foregoing values , assuming them to be correct , would be obtained if only we know , for a unicursal curve without cusps that ( a , , c , .
[ )rn-(a~b++c , .
( t the number of contacts a , 6 , C ... ) , and for a unicursal curve with a single cusp , that ( a , b6 el..)= ( a+1)(b+1)(c+1 ) ... [ rm-(a+b+6c ... ) ]t -fab+l)(~l* * .
rm-(a+b+c .
)lbl viz. that the diminution of a , , b , ... ) occasioned by the single cusp is = [ an(a+D c,..)-I]'~ .
X { a(b+l)(c+1 ) ... } .
88 .
Consider a unicursal curve Un , and a curve Ct having therewith t contacts of the orders a , 6 , c , ... respectively .
The coordinates ( X , y , z ) of any point of the unicursal curve are given as functions of the order m of a variable parameter 0 ; and substituting these values in the equation of the curve Ct , we have an equation of the degree rm in 0 , but containing the coefficients of CU linearly ; this equation gives of course the values of O which correspond to the rm intersections of the two curves .
Hence in order that the curve Cr may have the prescribed contacts with Utm , the equation of the degree rm in 0 must have t systems of equal roots , viz. a system of a equal roots , another system of b equal roots , &amp;c. : this implies between the coefficients of the equation an ( a +c , ... )fold relation , which may be shown to be of the order ( a+1)(6+1)(c+1 ) ... [ rm-(a+b+c , ... )]t ; and since the coefficients in question are linear in regard to the coefficients in the equation of the curve Cr , the order of the relation between the lastmentioned coefficients has the same value ; that is , the number of the curves Cr which have the prescribed contacts with the unicursal curve Urn and besides pass through the requisite number of given points , is=-(a+1)(b+1)(c+l).X [ rm-(a+b+c,. .
89 .
The reduction in the case of a cusp appears to be caused as follows:-Consider on the curve Un a points indefinitely near to the cusp , and let the condition of the curve Cr having the contact of the a-th order be replaced by the condition of passing through the a points ; that is , consider the curves Cr which have with the curve UmN ( t-1 ) contacts of the orders b , c,. .
respectively , which pass through the a points on the curve Utm in the neighbourhood of the cusp , and which also pass through the requisite number of arbitrary points .
The number of these curves is=(b+l)(c+1l ) ... [ rm-a-(b+c+..)]t(the term rrn-a instead of rm , on account of the given a points on the curve : compare herewith DE JONQUIERES ' formula containing rm-_p ) .
Each of these curves , in that it passes through a points in the neighbourhood of the cusp , will ipsofacto 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 , &amp;c. ) , and be consequently , in an improper sense , a curve having a contact of the a-th order with the given curve Up .
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 In the case of the conic , ( 1 ) , ( 2 ) , &amp;c. are the expressions denoted in the former part of this Memoir by ( 1 : : ) , ( 2 : .
) , &amp;c. , the number of points being in each case such as to make in all five conditions ; calculating these functions by means of the formulae ( a ) =Ea ] , &amp;c. , the comparison of the resulting values with the values previously obtained will show aposteriori the limits within which the formule are applicable ; where they cease to be applicable I find the difference , and annex it as a correction to the formula value : I have in some cases given what seems to be the proper theoretical form of this difference .
VWTe have ( 1 : : ) _ 2m+n ; ( 2. .
) ( 3 : ) 4m-3n+3a , ( 4* ) =.-10m-8n+6oa ; ( 53 ~--1 8z-15n + 10a-[-3tn+a ] ) 2(1,1 ... ) ( 2m+n)2 -4m-n--3a ; ( 1 , 2 : ) ( 2m+nf ) +12rn+12n-14tz ( 1 3 .
) = ( 2m+n)(-4m--3n+3a ) +56m+49n-39mc ; ( 1 , 4 ) = ( 2m+n)(-10m-8n+6x ) +140m+122n84c -E[(m -3)( -1 2m -6n + 6 ) ] ( = -E[(I -3)(4i+ 2z)J)X 2(2 , 2 a)2 +54m+48n-40Oa ; ( 2 , 3 ) = ( -4m-3n+3ez ) +144m+126n90oe -[24m+6n+(n-12)cxl ( = -[67+(n --3)X ] ) ; 6(1 , 1 , 1 : ) = ( 2m+n)3 + 3(2m+n)( -4m-An3x ) 32m58n+ 78a [ 4mn(m 1)(m2 ) ] ; S which is =-(m-4 ) C 31(2+ 3)+ 110(2 + 3z )(27 +3 ) 1+ ( '-m3+1142m+3174 )( zi ) +(-14m-638n-1524 )(2 +3x ) +(-390m +llOn+4272)(2 r+3i ) I +(-210m2-m180n+2130m+990n-7110)z+135X2 J 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 ) , &amp;c. , or to their equivalents [ a ] , [ a][b]+ [ a , b ] , &amp;c. in order to obtain formulae for the cases beyond the limits within which the present formulae are applicable ; but this I am not in a position to enter upon .
If the extended formulae were obtained , it would of course be an interesting verification or application of them to deduce from them the complete series of expressions ( 1 : ) , ( 2..). .
( 1 , 1 , 1 , 1 , 1 ) for the number of the conics which satisfy given conditions of contact with a given curve , and besides pass through the requisite number of given points .
It will be recollected that throughout these last investigations , I have put DuE JONQUIERES'_P=O ; that is , I have not considered the case of the curves Cr which ( among the conditions satisfied by them ) have with the curve Um contacts of given orders at given points of the curve ; it is probable that the general formulae containing the number p admit of extensions and transformations analogous to the formulae in which p is put=0 , but this is a question which I have not considered .
93 .
The set of equations ( a ) [ a ] , ( a , b ) -[a][b]+ [ a , b ] , &amp;c. , considered irrespectively of the meaning of the symbols contained therein , gives rise to an analytical question which is considered in Annex No. 7 .
The question of the conics satisfying given conditions of contact is considered from a different point of view in my Second Memoir above referred to .
Annex No. 1 ( referred to in the notice of DE .
JONQUIERES ' memoir of 1861).-On the form of the equation of the curves of a series of given index .
To obtain the general form of the equation of the curves C " of a series of the index N , it is to be observed that the equation of any such curve is always included in an equation of the order n in the coordinates , containing linearly and homogeneously certain parameters a , b , c. ; this is universally ' the case , as we may , if we please , take the parameters ( a , b , c. .
) to be the coefficients of the general equation of the order n ; but it is convenient to make use of any linear relations between these coefficients so as to reduce as far as possible the number of the parameters .
Assume that the number of the parameters is =cZ j-- , then in order that the curves should form a series ( that is , satisfy jn(n+ 3)-1 conditions ) , we must have a ( x1)fold relation between the parameters , or , what is the same thing , taking the parameters to be the.-coordinates of a point in w-dimensional space , say the parametric point , the point in question must be situate on a ( W 1)fold locus .
Moreover , the condition that the curve shall pass through a given point establishes between the parameters a linear relation ( viz. that expressed by the original equation of the curve regarding the coordinates therein as belonging to the given point , and therefore , as constants ) ; that is , when the curve passes through a given point , the corresponding positions of the parametric point are given as the intersections of the ( w-l)fold locus by an omal onefold locus ; the number of the curves is therefore equal to the number of these intersections , that is , to the order of the ( X-1)fold locus ; or the index of the series being assumed to be -N , the order of the ( w 1)fold locus must be also _N .
That is , the general form of the equation of the curves Cn which form a series of the index N , is that of an equation of the order n containing linearly and homogeneously the x+1 coordinates of a certain ( co-l)fold locus of the order N. It is only in a particular case , viz. that in which the ( a-l)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 parameter 0 ; and consequently only in this same case that the equation of the curves Cn of the series of the index N can be expressed by an equation ( *(x , y , z)n= 0 , or ( *)2x , y , 1 ) =O , rational and integral of the degree N in regard to a variable parameter V. If in the general case we regard the coordinates of the parametric point as irrational functions of a variable parameter 0 , then rationalising in regard to 0 , 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 n , say qn .
Such an equation represents not a single curve but q distinct curves C " , 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 containing the coordinates of the parametric point , say an equation ( *Xx , y , 1)'involving the two coordinates ( x , 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 ( x , y ) , and also rational of the degree N in the differential coefficient do ; in fact since the number of curves through any given point ( x0 , ye ) is =N , the differential equation must give this number of directions of passage from the point die ( ; O , o ) to a consecutive point , that is , it must give this number of values of die , and must consequently be of the order N in this quantity .
die adreNi Conversely , if a given differential equation rational in x , y , , anof the deg the last-mentioned quantity lie , 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 series of the index N ; whence the general integral ( assumed to be algebraical ) is given by an equation of the above-mentioned form , viz. an equation rational of a certain order n in the coordinates , and containing linearly and homogeneously the w+1 coordinates of a variable parametric point situate on an ( a1)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 Cn , but a system of q such curves : the first-menfioned form is , it is clear , the one to be preferred .
Annex No. 2 ( referred to , No. 17).-On the line-pairs 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 lines ( x=0 , y-0 , zzO ) , we have to find ( f , g , h ) such that the conic ( 0 , 0 , 0 , f g , hjx , y , z)2=0 , or , what is the same thing , fyz+gzx+hxyz=O , shall reduce itself to a line-pair , and shall touch a given conic ( 1 , 1 , 1 , l , [ L , iny y , Z)%=0 .
The condition for a line-pair is that one of the quantitiesf g , h shall vanish , viz. it is fgh 0 ; the condition for the contact of the two conics is found in the usual manner by equating to zero the discriminant of the function 1 -(x+f )2(+Og)2-(+Oh)2+ 2(X +f)(p + 0g)(v + Oh ) ( a , 6 , c , dXo , 1)3 suppose ; the values of a , 6 , c , d being a= 2fgh , b=6f2+g2+h2-2Xgh 2pjhf2Yfg ) , C 3= ++ d_2 2+ 2xypv .
Hence considering ( fg , h ) as the coordinates of the parametric point , we have the discriminant locus a= 0 , and the contact-locus a2d2 + 4ac3 + 4Vd32c26abed= 0 , and at the intersection of the two loci , a= 0 , 62(4bd 3c2 ) = 0 , equations breaking up into the system ( a=0 , 6=0 ) twice , and the system a=0 , 4bd3c-0 ; the former of these is fgh= 0 , f2+g2+h2 2Xgh2hf2fg= 0 , which expresses that the intersection of the two lines of the line-pair intersect on the given conic ; in fact the system is satisfied byf=0 g2+h-2xgh= 0 , giving a line-pair x(hy+gz)=0 , the two lines whereof intersect on the conic ( 1 , 1 , 1 , X , P , , P:x , ,y , z)2=o ; and similarly , if g=O , then h2+f2p2thf=0 , or if 4=O , then f2+g2 2xfg-0 .
As noticed above this system occurs twice .
The second system is th-= 0 , ( f2 +g2 + h2 2xgh2phf2yfg)(1 d2 -P2y2+2Xpd ) +((P X)f+(pX-p')g+(p=0 , or , as the second equation may also be written , f2(1_2)(-Iv ) +g2( _V2)(1 X2 ) + h2(1 -)( 2 ) +-2gh(l ) k ) + 2hf(12)(vX- ) + 2fg(1 -2)(xz _)= which expresses that a line of the line-pair touches the conic ; in fact the system is satisfied by f= 0 , g(l 02 ) + h2(1-fly ) + 2gh(pi = 0 , viz. we have here the line-pair x(hy+gz)=0 , in which the line hy+gz 0 touches the conic ( 1,1 , 1 , X , d , v](x , y , z)2-=0 ; and the like if g= 0 , or if h= 0 .
This system it has been seen occurs only once .
Annex No. 3 ( referred to , No. 22).-On the conics which pass t1Irough two givenpoints and touch a given conic .
Consider the conics which pass through two given points and touch a given conic .
We may take Z=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 two given points respectively , the equations of the joining lines may be taken to be X=0 and Y=0 respectively .
This being so , we have for the given points ( X=0 , Z=0 ) and ( Y= 0 , Z_-0 ) respectively , and for the given conic aX2+by2+2hXY+cZ2=0 ; and since the required conic is to pass through the two given points its equation will be of the form WX2 +2xYZ+2yZX+2zXY=O , where ( x , y , z , w ) are variable parameters which must satisfy a single condition in order that the last-mentioned conic may touch the given conic .
The condition is at once seen to be that obtai:led by making the equation ( a+Xw)bc _ ( a+wX)(h+=z)2 + 2X2xy(h+ ?
\z ) =0 , considered as a cubic equation in X , have a pair of equal roots ; or if we write A= 3c(ab -h2 ) , B= ( ab-_h2)uW-2chz , C= -aeb y2 _CZ2 +2ht(y-zwo ) , C D= 3z(2xy-wz ) , then the required condition is A2D2+4AC3 + 4BI3D6ABCD3B2C2= 0 .
Hence the conic wX2+ 2xYZ+2yZX+2zXY= 0 satisfies the prescribed conditions , if only the parameters ( x , y , z , w ) satisfy the lastmentioned equation , that is , if ( x , y , z , 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 , C =0 ; this may be considered as the intersection of the quadric surface ACB2 0 and the cubic surface AD-BC= 0 ; and the cuspidal curve is consequently a sextic .
The surface has also a nodal curve made up of two conics ; to prove this I write for shortness k.h-V/ ab , k , =h+Vab ; the values of A , B , C , D then are A= 3ckk1 , B =-kw 2 chz , C _.g2b by2GzI+2h(xy -zw ) , D=3z(2xyzw ) .
And it is in the first place to be shown that the surface contains the conic x:y : z : W= V6b : OVa : 1 : k2+ where 0 is a variable parameter .
Substituting these values , we have A= 3ckk1 , B=k2Jkc 02_ c(3h+Vab ) , C =2kk,2 _j(3h-V/ ab ) , and hence AD-BC= =2k Qk12 +2c a)2 AC-B2 =-k2 ( kk02 ?
2+ c ab)2 BD-C2 ( kk O2+2c V ab ) 2 values which satisfy identically the equation of the surface written under the form ( AD -BC)24(AC-B2)(BD C2 ) O. Moreover , proceeding to form the derived equation , and to substitute therein the foregoing values of ( x , y , Z , w ) , we have bA : bB:6C : 3D=0-O k2 : 2k : 3 , and then the derived equation is -(AD-BC)( 3A-2kB-lk2C ) -2(AC-B 2 )( 3B 4k C+k2D ) -2(BD-C2 )(2kA212B that is , -k ( 3A-2klB-k2 C ) +k 2( 3B-4kC+kk2D ) + ( 2kA-.21F2B ) , or finally -/ k(A3Bk/ + 3k2Ck'D)=O , which is satisfied by the foregoing values of A , B , C , D ; hence the conic is a nodal curve on the sextic ; and by merely changing the sign of one of the radicals Va , s/ &amp; ( and therefore interchanging k , ke ) we obtain another conic which is also a nodal curve on , the surface , that is , we have as nodal curves the two conics ( x : Y : z : w=0v/ b : Oea:1 : kO2 ?
rand X : Y : Z : = b : ZOVa : I : k C02+ It is to be remarked that each of the nodal conies meets the cuspidal curve in two points , viz. writing for shortness 0= 21= V22$/ ab , for the intersections of the first conic we have X : y:z : W=Osa .
( E V/ D 1 : c and =a::-O v/ : 1 : Ac and for the intersections with the second conic x : y : z : w= 01va:1/ b : l and= OVa : 1jvb : 1 : The condition of passing through any arbitrary point establishes a linear relation between the parameters ( x , y , z , w ) .
Hence , if the conic in addition to the prescribed conditions passes through two other given points , the point ( x , y , z , w ) is given as the intersection of a line with the sextic surface ; the number of intersections is =6 .
If ( x , i , z , w ) is situate on the cuspidal curve , then the conic instead of simply touching the given conic will have with it a contact of the second order , and if we besides suppose that the conic passes through a given point , then the point ( x , y , z , w ) is given as the intersection of the cuspidal curve with a plane ; the number is =6 .
Similarly , if the conic has two contacts with the given conic , and besides passes through a given point , then the point ( x , y , z , w ) is given as the intersection of the nodal curve by a plane ; the number is 4 .
Finally ( observing that in the case in question of the contacts of a conic with a conic we cannot have three simple contacts , or a simple contact and one of the second order ) , a point of intersection of the nodal and cuspidal curves answers to a contact of the third order ; and the number is=4 .
That is , the theory of the sextic surface leads to the following values ( agreeing with those obtained from the formulae by writing therein m=n=2 , M=6 ) , viz. ( 1 : : ) -6 , 2m+n , ( 1 , 1.:.)=4 , 2mn2+2mn+ n2-2m in-cc , &lt; ( 2 .
)=6 , =t ( 3 : ) =4 , =-4m--3n+3oa .
I remark that the section by an arbitrary plane is a sextic curve having 6 cusps and 4 nodes ; it is therefore a unicursal sextic ; this suggests the theorem that the sextic surface is also unicursal , viz , that the coordinates are expressible rationally in terms of two parameters ; I have found that this is in fact the case .
In doing this there is no loss of generality in supposing that a=b=c=1 ; and assuming that this is so , and putting also -1+h=k , 1+h-k , , and therefore 2h=k+kI , we have A=3kkc , B=kklw-(k+kl)z , C= 2 y2_ 2+(k+kj(xy-zw ) , D= 3z(2xy-zW ) .
The equation of the sextic surface being , as before , A2D2 + 4AC3 + 4B3D 'B2C2 -6ABCD= , I say that this equation is satisfied on writing therein x+y =Vi(1-Al.a ) sin 7 ' x-Y= ( I ( 1+kk x ) cosp , z =1 , wr ( 2c k ) cos2p +(2Mz-k ) singp , where ( Oa , p ) are arbitrary .
In fact these values give 1A -kk , Cos 2p-kki sinc'P , B k(2xk1+ 1 ) cos20 -kl ( 2oak+1 ) sin2 ' , C=-k( 06k , +2)cos2e-k , ( ak+2)sin2'p , 1D_ -ka2 cos2 -ki ?
2 sin2 4 , whence w being arbitrary , we have ( A , B , C , D3w , 1)3 =-rk cosep ( kw+l)+kl sin2 p(kW+1)](W+oa)2 , viz , the equation ( A , B , C , D)(w , 1)'=0 considered as a cubic equation in , has the twofold root w= -O , that is , we have the above relation between ( A , B , C , D ) .
Whence also writing sin 0= 2X , cos '=l , the equation of the surface is satisfied by the values s+Z( : x-_y : z : X2 ) : 2/ ( I +koa ( I _,4 ) : ~~~~~(I +x42)2 2a-i ( I1X2)2 + ( 2a-k U2 , or the coordinates are expressed rationally in terms of are , X. Annex No. 4 ( referred to , Nos. 22 and 71).On the Conics which touch a cuspidal cubic .
In the cuspidal cubic , if s=0 be the equation of the tangent at the cusp , y-p=O that of the line joining the cusp with the inflexion , and z=0 that of the tangent at the cusp , then the equation of the curve is y3= 2Z ; the coordinates of a point on the cubic are given by x : y : z=1 : 0 : 0 ' , where 0 is a variable parameter ; and we have , at the cusp 0=00 , at the inflexion 0=0 .
In the cubic , mn=n=3 , ( =3n+x)=10 .
Considering now the conic ( a , be , c , q50f g , h3 yJ Z)2= 01 this meets the cubic in the 6 points the parameters of which are determined by the equation ( a , b , c , f g , hXl , 0 , 03)2=0 , or , what is the same thing , ( c , 0 , 2f , 2g , b6 2h , aXO , 1)6=0 .
The discriminant of this sextic function contains the factor c , hence equating the residual factor to zero , we obtain the equation of the contact-locus in the form ( caf , g , 6 , h , a)9=0 .
MIDCCCLXYIIIT U It follows that the -number of the conics ( 1 : : ) is = 9 , which agrees with the general value ( 1::)=2n+n .
If the conic pass through the cusp we have c=0 , and the equation in 0 is reduced to a quartic ; it is convenient to alter the letters in such wise that the quartic equation may be obtained in the standard form ( a , 4 , c , d , eat 1)(i viz. this will be the case if the equation of the conic is taken to be ( e , 6c , 0 -la , 2b , 2dIX , y z)2-0 and we then obtain the equation of the contact-locus in the formit ( ae4bd+ 3c2)3-27(ace+2beed-ad2 -4e-c3)2= 0 , which is a onefold locus of the order C. It follows that we have ( 1X1 , I. 6.)-C , agreeing with ( 1 , 1..)=n+ 2n 3 .
The condition in order that the conic may touch a given line is given by an equation of the form ( *Xa2 cb , b2 , 2ce3d2 , de-84d , ad 12bc)1=0 , which is a onefoid locus of the order 2 ; it at once follows that we have ( 1,1 , 1 , :/ )=12 , agreeing with , ( 1xl , 1 , :/ )=2n+4n-C6 .
It is a matter of some difficulty to show that we have ( 11 , l)*/ /)=18 , agreeing with ( 1 , z , 1* / /)=4n+49-C6 ; but I proceed to effect this , first remarking that I do not attempt to prove the remaining case ( T,4 , 1 / // )-15 , agreeing with ( 1Z1 , 1 / // )=4n+2m-3 .
Investigation of the value ( 11 ; A , I 7/ ) 18 : We have the sextic locus ( ac644d + 3c2r2 7(ac6 + 24cdad4be-c'8)2 = 0 , and combined therewith two quadric loci , ( *](a ' , ab , P2 2ce-3d2 , ae-84d , ad-124c)1-=0 , ( */ Xa2 &gt; ab , 42 , 2ce3d2 , ee-8d , ad-12bc)'-0 .
which intersect in a threefold locus of the order 24 ; it is to be shown that this contains as part of itself the quadric threefold locus ( a-O : 4 , 0 , 2ce3d2 0 ) taken three times , leaving a residual threefold locus of the order 24-C -18 We may imagine the coordinates a , 4 , c , d , e expressed as linear functions of any fonr coordinates , and so reduce the problem from a problem in 4-dimensional space to one in ordinary 3-dimensional space .
We have thus a sextic surface , and two quadric surfaces ; the sextic is a developable surface or torse , having for one of its generating lines the line a=0 , 40 , and for the tangent plane along this line the plane a=0 ; the two quadric surfaces meet in a quartic curve passing through the two points ( a=0 4=O , 3ce-2d2=O ) , which are points on the torse ; ; it is to be shown that each of these points count three times among the intersections of the torse with the.quartic curve , the number of the remaining intersections being therefore 24-6 ?
= 18 ; and in1 order thereto it is to be shown that each of the points in question ( a =0 , b=0 , ce -2d2=0 ) 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 torse in three consecutive points , or the two points of intersection each count three times .
The torse has the cuspidal line S=ae-4bd+3c2-=0 T-ace+2.bd-acd2 2e--3=0 , and the nodal line 6(ac-62 ) , 3(ad-be ) , ae+-2bd3c2 , 3(be -cd ) , 6 ( ce .
d2 a , b , c , d , e and the equations of the nodal line are satisfied by the values ( a=O , b=0 , 3ce-2d2=0 ) of the coordinates of the points in question .
To find the tangent planes at these points , starting from the equation S3 27T 2=0 of the torse , taking ( A , B , C , D , E ) as current coordinates , and writing a Aak+BMb+Cc +Dad+ Eke , then the equation of the tangent plane is in the first instance given in the form 2DS-i 81 TT=60 , which writing therein ( a=0 , b=O , 3ce-2d2=0 ) assumes , as it should do , the form 0=0 ; the left-hand side is in fact found to be 9c'(3ce-2d2))A .
Proceeding to the second derived equation , this is S262S+2S(bS)2-_STh2T-.18(bT)2=0or substituting the values of the several terms , the equation is 9c4(AEE-4BD + 3C2 ) + 3c2(eA-4dB+6CC)2 + 18C3 { e(AC-B2 ) + 2d(BC-AD ) +c(AE+ 2BD3C2 ) } 9 { ( ce-d2)A+2cdB-3c2C}2=0 ; the terms in BC , BD , C2 vanish identically , that in B2 is ( 48-36=)12c'd2 --l8c3e , -6c2(3ce-2d2)B2 , which also vanishes ; hence there remain only the ternms divisible by A , giving first the tangent plane A= 0 , and secondly the other tangent plane , A(6c2e2+l 8cd2e9d4 ) +B(6(O02de+36ed3 ) -i+C( 108c3e-54c2d2 ) +-D(36c3d ) +E .
27c4 =0 .
'taking the equations of the quadric surfaces to be ( Xt , En , Use a)la2 62 , ab A3ce -2d2 , ae-8bd , ad-126c)=- , u 2 the equations of the tangent planes are 6(3cE+3eC-4dD)+ao(eA8dB)+lr(dA-12cB)=0 , in all which equations we have 3ce-2d2=O ; and if to satisfy this equation we write c : d : e=2 : 3p : 3f32 , then the equations of the tangent planes become P3 ( A-3 8B)+ 8(3CP24DP+2E ) -O ?
o(3CB2 4D1+2E)+(ofP+ ?
r)(AI3-8B ) -0 'f( X , )+ ( c ' - ) , , )=-O .
or the three tangent planes intersect in the line Aj3 -8B O , 3Cj32 4D13)+2E= 0 which completes the proof .
Reverting to the sextic locus , ( ae + 4d 3c2)2 27(ace+2bed-ad2b2e-c3)2= 0 , considered as a locus in 4-dimensional space depending on the five coordinates ( a , b , c , d , e ) , this has upon it the twofold locus ae4bd+3c2=0 , ace+2bed-ad2-b2eC3=0 say the cuspidal locus , of the order 6 , and the twofold locus 6(ac b2 ) , 3(ad-be ) , ae+2bd-3c2 , 3(be-cd ) , 6(ce-d2 ) =05 a , 6 , c , d , e say the nodal locus , of the order 4 : there is also a threefold locus , a , b , c , d -0 6| , c , d , el say the supercuspidal locus , of the order 4 .
We thence at once infer ( 1x1 , 2 : )-6 , agreeing with ( hzi , 2:)=a4 , ( 1zl , 1 , 1 : )=4 , , , ( 1z1l51 , 1:)=2 n+ 2mn+-In_8m-2n1 ( 1hi , 3:)=4 , , , , ( hli , 3:)=-4m-n3-5+ &amp; 3 ; but I have not investigated the application to the symbols with / or HIf the conic , instead of simply passing through the cusp , touches the cuspidal tangent , then in the equation ( a , b , 0 , g , hXx , y Z)2 =0 of the conic we have f=0 , or , what is the same thing , in the equation ( e , 6c , 0 , -l-a , 2b , 2djx , y , Z)2==0 of the conic we have a=0 .
The equation in 0 is thus reduced to 4b63+66c02+4d0+e=O .
For the independent discussion of this case it is convenient to alter the coefficients so that the equation in D may be in the standard form ( a , b , c , dig , 1)3= 0 , viz. we assume the equation of the conic to be ( d , 3b , 0 , 0 , ia , -lca , y , z)2 0 .
The equation of the contactlocus then is a2d2 + 4ac ' + 4b3d6abCd 362c2=0 , viz. this is a developable surface , or torse , of the order 4 , and we at once infer ( 241 , 1 : )-=4 , agreeing waith ( 24,1 1 : )=2m-Fn-5 .
WHiCH SATISFY GIVEN CONDITIONS .
135 I will show also that we have ( 2z1 , 1 / )=6 , agreeing with ( 2z1 , 1 / )=2n+2n 6 , and ( 2x1 , 1 / /)=5 , , , , , ( 2z1 , I / /)= m+2n-4 .
The condition that the conic may touch an arbitrary line x+ ?
3y+-yz-0 , is in fact ( 0 , -4a2 , ( 4bd-3c0 ) , ac , -3tab , O ' , / 3 7)2=0 , which , considering therein ( a , b , c , d ) as coordinates , is the equation of a quadric surface passing through the conic a= 0 , 4bd3c0=0 ; the quartic torse also passes through this conic ; hence the quadric surface and the torse intersect in this conic , which is of the order 2 , and in a residual curve of the order 6 ; and the number of the conics ( 2z1 , 1 / ) is equal to the order of this residual curve , that is , it is = 6 .
If the conic touch a second arbitrary line dx+P'y+7fz=O , then we have in like manner the quadric surface ( 0 , -la23 ( 4bd3c2 ) , -ac , -tab , Old , at , rI)2=0 ; that is , we have the quartic torse and two quadric surfaces , each passing through the conic a=O , 4bd-3 200 and it is to be shown that the number of intersections not on this conic is 5 .
The two quadric surfaces intersect in the conic and in a second conic ; this second conic meets the torse in 8 points , but 2 of these coincide with the point a=0 , b= 0 , c =0 , which is one of the intersections of the two conics ( the point a-=0 , b=0 c=O is 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 points coincides with the other of the intersections of the two conics ; there remain therefore 8 2-1 _5 intersections , or we have ( 2z1 , 1/ / )5 .
Annex No. 5 ( referred to , Nos. 22 and 71).-On the Conics which have contact of the third order with a given cuspidal cubic , and two contacts ( double contact ) with a given conic .
Let the equation of the cuspidal cubic be X25-y3=0 ( X=0 tangent at cusp , z=O tangent at inflexion , y= 0 line joining cusp and inflexion ; equation satisfied by : y z-1 : d ' : 03 ) ; and let the equation of the given conic be U=(a , b , c , / ; g , h(X , y , Z)3 0 ; then writing te t ( a , b , cf , g , h1 , D 0 , 03)2 =cd6+ 2fd4+2g03+ b'+ 2hdH+c , the equation of a conic having with the given cubic at a given point ( 1 , 0 , 0 ' ) contact of the second order , and having double contact with the given conic , is 4/ U , x , ye z 0 , ( V ?
o 1 ) 0 63 where the form of the coefficients may be modified by means of the identical equations ( A , H , Ga , hI , g)=KI , ( HI B. F38 , , )=O , ( G , F , C( , ,)0 ' ( A , H , G:( , bf)=O , ( HiBI F3( , , )=K , ( G , F , CX , , )=O , ( A , H , G:(g , j ; c ) = O , ( HI , B , F( , , )=O , ( G. F. C3( , , )=Ki .
There is consequently a conic answering to each value of 0 given by this equation , or we have in all 12 conics .
In the case where the givei conic breaks up into a pair of lines , or say , ( a , b CJ f , hasX ye Z)2 -2(Xx+ Pie +PZ)(klX+Pty+P'Z then , writing for shortness piP -pi 'd , va AYE , Xp/ ?4'(k-Ad X , Y , ; , we have ( A , B , C , F , G , H ) = ( X2 Y2 Z2 YZ , ZX , XY ) .
Substituting these values , but retaining ( a , b , c , if , h ) as standing for their values a -22xx .
, &amp;c. , the equation in K is found to contain the cubic factor 2XdO3Y02+Z , where it is to be observed that this factor equated to zero determines the values of d 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 lines 2x+ IyI+z-0O and X'x+b'y+v'z=0 ; and omitting the cubic factor , the residual equation is found to be ( 2cX -12cY -8fX -20gX -iOhX -4OhX -20aX +15aY +5hZ +A 0 , 1)a -l12fY + 83 , Y SbY +17hY + 4hZ + 4c Z +7~qZ =0 , where the form of the coefficients may be modified by means of the identical equations aX+h1Y+gZ=0 liX+bY+-fZ=0 , gX+ fie +cZ= O. The equation is of the 9th order , and there are consequently 9 conies .
is even or odd ; thus in the expression for [ 11111 ] , the term ( 14)(13)(12)(11 ) has four factors , and is therefore + , the term ( 113)(12)(11 ) has three factors , and is therefore - .
The numerical coefficients are obtained -as follows .
There is a common factor derived from the expression in [ ] on the left-hand side of the equation ; viz. for [ 11111 ] , which contains five equal symbols , this factor is 1.2.3.4.5 , =120 ; for [ 1112 ] , which contains three equal symbols , it is 1.2.3 , -6 ; and so on ( for a symbol such as [ 11222 ] containing two equal symbols , and three equal symbols , the factor would be 1.2.1.2.3 , =12 , and so in other similar cases ) .
In any term on the right-hand side of the equation , we must for a factor such as ( 11 ) , which contains two equal symbols , multiply by 1 ; for a factor such as ( 111 ) , which contains three equal symbols , multiply by 4 , and so on .
And in the case where a term ( as , for example , the term ( 122)( ] 1 ) or ( 23)(12)(11 ) , vide suprA ) occurs more than once , the term is to be taken account of each time that it occurs ; or , what is the same thing , since the coefficient obtained as above is the same for each occurrence , the coefficient obtained as above is to be multiplied by the number of the occurrences of the term .
For example , taking in order the several terms of the expression for [ 1112 ] , the common factor is -6 , and the several coefficients are 6 , 6.4,6 6.4 , 6 .
1 X2 6.41 6.4 , 6.4 !
2.4x2 , 6 .
( ; and similarly in the expression for [ 11111 ] the common factor is 120 , and the coefficients taken in order are 120.4 120.4 , 120.4.4.41 &amp;c. , without there being in this case any coefficient with a factor arising from the plural occurrence of the term .
The foregoing result was established by induction , and I have not attempted a general proof .
I observe by way of a convenient numerical verification , that in each equation the sum of the coefficients ( taken with their proper signs ) is ( -)n-11 .
2. .
( n-1 ) ; if n be the number of parts in the [ ] ( n=5 for [ 11111 ] , =4 for [ 1112 ] &amp;c. ) , and moreover , that the sum of these sums each multiplied by the proper polynomial coefficient and the whole increased by unity is ' 0 ; viz. for [ 14 ] [ 23 ] [ 113 ] [ 122 ] [ 1112 ] E11111]1 the sums of the coefficients are -1 , -1 , +2 , +2 , -6 , +24 respectively , and we have 1+5(-1)+10(-1)+10(2)-+ ?
15(2)+10(-6)+1(24 ) , =75-75 , =0 .
If we have any five distinct things ( a , b , c , d , e ) , then the polynomial coefficients 5 , 10 , 10 15 , 10 , 1 denote respectively the number of ways in which these can be partitioned in the forms 14 , 23 , 113 , 122 , 1112 , 11111 respectively , and the last-mentioned theorem is thus a theorem in the Partition of Numbers .
On the Curves which satisfy given Conditions .
By Professor CAYLEY , F.1 ?
S. 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 else the curves are conics ; 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 theory of conics , the Memoir is based upon the researches of CHASLES and ZEUTHEN , as regards that of the curves of the order r , upon the researches of DE JONQUIERES : the notion of the quasi-geometrical representation of conditions by means of loci in hyper-space 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 , with remarks .
DE JONQUIERES : Theoremes generaux concernant les courbes ge6ometriques planes d'un ordre quelconque , Liouv .
t. vi .
( 1861 ) pp. 113-134 .
In this valuable memoir is established the notion of a series of curves of the index N ; viz. considering the curves of the order n which satisfy -n(n+3)-1 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 C. of a series of the index N can be analytically represented by an equation F(y , x)= 0 , which is rational and integral of the degree N in regard to a variable parameter X : this is not the case ; see Annex No. 1 .
CHASLES : Various papers in the Comptes Rendus , t. lviii .
et seq. 1864-67 .
The first of them ( Feb. 1864 ) , entitled " 'Determination du nombre des sections coniques qui.doivent toucher cinq courbes donnees d'ordre quelconque , ou satisfaire a diverses autres conditions , " establishes the notion of the two characteristics ( p. , v ) of a system of conics which satisfy four conditions ; viz. p is the number of the seconics which pass through a given arbitrary point , and v the number of them which touch a given arbitrary lines MDc'ccLxYIII .
N The Principle of Correspondence for points on a line is established in the paper of June-July 1864 .
Many of the lea.ding points of the theory are reproduced in the present Memoir .
The series of papers includes one on the conics in space which satisfy seven conditions ( Sept. 1865 ) , and another on the surfaces of the second order which satisfy eight conditions ( Feb. 1866 ) .
SALMON : On some Points in the Theory of Elimination , Quart .
Math. Journ. t. vii .
pp. 327-337 ( Feb. 1866 ) ; On the Number of Surfaces of the Second Degree which can be described to satisfy nine Conditions , Ibid. t. viii .
pp. 1-7 ( June 1866 ) , -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 hyper-space .
ZEUTHEN : Nyt Bidrag ... Contribution to the Theory of Systems of Conics which ' satisfy four conditions , 8 ? .
pp. 1-97 ( Copenhagen , Cohen , 1865 ) , translated , with an addition , in the Nouvelles Annals .
The method employed depends on the determination of the line-pairs 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 curees .
It is reproduced in detail , with the enumeration called " ZEUTllEN 's Capitals , " in the present Memoir .
GAYLEY : Sir les coniques determi'nees par cinq conditions d'intersection avec un courbe donnee.-Comptes Rendus , t. lxiii .
pp. 9-12 , July 1866 .
Results reproduced in the present Memoir .
DE JONQUIhRES : Two papers , Comptes iRendus , t. lxiii .
Sept. 1866 , reproduced and further developed in the ' Me'moire sir les contacts multiples d'ordre quelconque des courbes du degre r qui satisfont a des conditions donnees de contact avec un courbe fix du degre in ; suivi de quelques reflexions sir la solution d'un grand nombre de questions concernant les proprietes projectives des courbes et des surfaces algebriques , " Crelle , t. lxv.i ( 1866 ) , pp. 289-322 , -contain a general formula for the number of curves CCr having contacts of given orders a , 6 , c,. .
with a given curve Un &amp;c. , which formula is referred to and considered in the present Memoir .
DrE JONQUfi4RES : Recherches sir les series ou systemes de courbes et de surfaces algebriques d'ordre quelconque ; suivi d'une reponse &amp;c. 4 ? .
Paris , Gauthier Villars , 1866 * , On the quasi-geometrical re resentaion of Gonditions.-Article Nos. I to 23 .
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 contained in the equation of the curve .
The condition may be onefold , twofold , ... or , generally , k-fold , and the corresponding relation is onefold , twofold , ... or k-fold accordingly .
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 consider two or more conditions or relations , this case is in fact included in that of a k-fold condition or relation .
In dealing with such a condition or relation it is assumed that the number of parameters is at least =k ; for otherwise there would not in general be any subject satisfying the condition : when the number of parameters is =k , 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 I-fold condition or relation , and the same number k of parameters , the 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 maybe 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 two-pointic 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 w , of parameters may be considered as the coordinates of a point in w-dimensional space ; and if the parameters are connected by a onefold , twofold , ... or k-fold relation , then the point is situate on a onefold , twofold , ... or k-fold locus accordingly ; to the relation made up of two or more relations corresponds the locus which is the intersection or common locus of the loci corresponding to the several component relations respectively .
A locus is at most ca-fold , viz. it is in this case a pointsystem .
The relation made up of a k-fold relation , an i-fold relation , &amp;c. , is in general ( k+l+ &amp;c. )fold , and the corresponding locus is ( k+l+ &amp;c. )fold accordingly .
5 .
The order of a point-system 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 corresponding to any linear j-fold relation between the coordinates is said to be aj-fold omal locus ; and if to any given k-fold relation we join an arbitrary ( c.-k)fold linear relation , that is , intersect the k-fold locus by an arbitrary ( w-k)fold omal locus , so as to obtain a point-system , the order of the k-fold relation or locus is taken to be equal to the number of points of the point-system , that is , to the order of the point-system .
And this being so , if a k-fold relation , an i-fold relation , &amp;c. are completely independent , that is , if they are not satisfied by values which satisfy a less than ( k+1+ &amp;c. )fold relation , or , what is the same thing , if .
the k-fold locus , the i-fold locus , &amp;c. have no common less than ( k+l+ &amp;c. )fold locus , then the relations make up together a ( k+l+ &amp;c. )fold relation , and the loci intersect in a ( k+l+ &amp;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 k+l+ &amp;c. =a , then we have an w-fold relation , and corresponding thereto a point-system , the orders whereof are respectively equal to the product of the orders of the given relations or loci .
6 .
A k-fold relation , an i-fold relation , &amp;c. , if they were together equivalent to a less than ( k+l+ &amp;c. )fold relation , would not be independent ; but the relations , assumed to be independent , may yet contain a less than ( k-1+ &amp;c. )fold relation , that is , they may be satisfied by the values which satisfy a certain less than ( k+I+ &amp;c. )fold relation ( say the common relation ) , and exclusively of these , only by the values which satisfy a proper ( k-F 1+ &amp;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 k-fold locus , the i-fold locus , &amp;c. may have in common a less than ( k+1+ &amp;c. )fold locus , and besides intersect in a residual ( k+1+ &amp;c. )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 comnjplete independence .
) In particular if k+ l+ &amp;c. =w , the several loci may intersect , say in an ( x-j)fold locus , and besides in a residual x-fold locus , or point-system .
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 c ; arbitrary parameters : to fix the ideas , it may be assumed that these enter into the equation rationally so 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 w-dimensional space ; so long as the curve is not subjected to any condition , the point in question , say the parametric point , is an arbitrary point in the ce-dimensional space ; but if the curve be subjected to a onefold , twofold , ... or k-fold condition , then we have a onefold , twofold , ... or k-fold relation between the parameters , and the parametric point is situate on a onefold , twofold , ... or k-fold locus accordingly : to each position of the parametric point on the locus there corresponds a curve satisfying the condition , that is , a solution of the condition .
In the case where the condition is c-fold , the locus is a point-system , and corresponding to each point of the point-system we have a solution of the condition ; the number of solutions is equal to the number of points of the pointsystem .
8 .
Considering the general case where the condition , and therefore also the locus , is k-fold , it is to be observed that every solution whatever , and therefore each , special solution ( if any ) corresponds to some point on the k-fold locus ; we may therefore have on WiHICH SATISFY GIVEN CONDITIONS .
79 the k-fold 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 point-system , viz. there are in this case a determinate number of special solutions corresponding to the several points of this point-system .
We may consider the other extreme case of a special I-fold locus , viz. the k-fold locus of the parametric point may break up into two distinct loci , the special le-fold locus , and another k-fold 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 last-mentioned / -fold locus and regarding it as the proper locus of the parametric point .
But if the special locus be a more than k-fold locus , that is , if it be not a part of the k-fold 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 k-fold 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 therecorrespond special or improper solutions .
It is to be noticed that the special locus may be , but that it is not in every case , a singular locus on the k-foldlocus .
9 .
Suppose that the conditions to be satisfied by the curve are a k-fold condition , an 1-fold condition , &amp;c. of a total manifoldness=w .
If the conditions are completely independent ( that is , if the corresponding relations , ante , No. 5 , are completely independent ) , we have a k-fold locus , an l-f6ld locus , &amp;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 point-system , or to the order of the point-system , viz , it is equal to the product of the orders of the loci which correspond to the several conditions 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 k-fold locus , the 1-fold locus , &amp;c. intersect in a common ( a-j)fold locus , and besides in a residual point-system .
The several points of the ( a-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+ &amp;c. )fold , that is , a true c-fold condition ; the proper solutions are therefore comprised among the solutions given by the residual point-systenm , and the number of them is as before equal to the order of the point-system , 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 k-fold locus , the i-fold locus , &amp;c. , less a reduction depending on the nature of 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 b~e ao ; but if

