]\gt ; [ 369 ] XV .
On the Problem of the In-and-Circumscribed Triangle .
A. CAYLEY , F.R.S. Received December 30 , 1870 , \mdash ; Read February 9 , 1871 .
THE problem of the In-and-Circumscribed Triangle is a particular case of that of the In-and-Circumscribed Polygon : the last-mentioned problem may be thus stated\mdash ; to find a polygon such that the angles are.situate in and the sides touch a given curve or curves .
And we may in the first instance inquire as to the number of such polygons .
In the case where the curves containing the angles and touched by the sides respectively are all of them distinct curves , the number of polygons is obtained very easily and has a simple explession : it is equal to twice the product of the orders of the curves containing the several angles respectively into the product of the classes of the curves touched by the several sides respectively ; or , say , it is equal to twice the product of the orders of the angle-curves into the product of the classes of the side-curves .
But when several of the curves become one and the same curve , and in particular when the angles are all of them situate in and the sides all touch one and the same curve , it is a much more difficult problem to find the number of polygons .
The solution of this problem when the polygon is a triangle , and for all the different relations of identity between the different curves , is the object of the present memoir .
which is accordingly entitled ' ' On the Problem of the In-and-Circumscribed Triangle the methods and principles , however , are applicable to the case of a polygon of any number of sides , the method chiefly made use of being that furnished by the theory of correspondence , as will be explained .
The results ( for the triangle ) are given in the following Table ; for the explanation of which I remark that the triangle is taken to be aBcDeF ; viz. are the angles , the sides ; that is , are the sides , ce , respectively , and are the angles , DF respectively .
And I use the same letters to denote the curves containing the angles and touched by the sides respectively ; viz. the angle is situate in the curve , the side touches the curve , and so for the other angles and sides respectively .
An equation such as or denotes that the curves or , as the case may be , the curves are one and the same curve : it is in general convenient to use a new letter for denoting these identical curves ; viz. I write , for instance , or , to denote that the curves or , as the case may be , the curves are one and the same curve ; the new letters thus introduced are , there being in regard to them no distinction of small letters and capitals .
The expression " " no identities\ldquo ; denotes that the curves are all distinct .
But I use also the letters , and quantitatively , to denote the orders and classes of the curves respectively ; thus , in the Table , for the casel ' ' no identities\ldquo ; the number of triangles is given as aceBDF , which agrees with tions times the order ; viz. the curve is determined by its order , class X , and ; and similarly for and Observe that , in the column " " Specification each line is to be read separately from the others , and , where the word " " or\ldquo ; occurs , the two parts of the line are to be read separately ; thus case 5 , the six forms are : the letter case accompanies the first of the given forms ; in the present instance , and it is to this first form that the number of triangles , here ceDF , applies .
I remark that what is primarily determined is the number of positions of a particular angle of the triangle , and that in some cases , on account of the symmetry of the figure , the number of triangles is a submultiple of this number ; viz. the number of positions of the angle is to be divided by 2 or 6 ; this is expressly shown , by means of a separate column , in the Table .
om co oo then toints ahave acertain correspondence , respondence withwhere toints aituate ourve , urve agviz.nstead oriangle aBcDeF wonsider tnclosed trilateral aBcDeFgThe foregoing results ahiefly obtained beans oheory oorrespondenceON ROBLEM OCIRCUMSCRIBED TBIANGLE .
each other ; and when are a " " united point\ldquo ; of the correspondence , the trilateral in question becomes an in-and-circumscribed triangle aBcDeF ; that is , the number of triangles is equal to that of the united points of the correspondence , subject however ( in many of the cases ) to a reduction on account of special solutions .
It may be remarked that by the theory of correspondence the number of the united points is , in several of the cases , but not in all of them , .
But in some instances I employ a functional method , by assuming that the identical curves are each of them the aggregate of the two curves : we here obtain for the number of the triangles belonging to the curve a functional equation given function ; viz. the expression on the right-hand side depends on the solution of the preceding cases , wherein the number of identities between the several curves is less than in the case under consider- ation ; and takin it to be known , the functional equation gives particular solution linear function of .
The particular solution is always easily obtainable , and the constants of the linear function can be determined by means of particular forms of the curve The Principle of Correspondence as applied to the present Problem.\mdash ; Article Nos. 1 to 6 .
1 .
Consider the unclosed trilateral aBcDeFg , where the points and are on one and the same curve , .
Starting from an arbitrary point on the curve , we have any one of the tangents from to the curve , touching this curve , say at the point and intersecting the curve in a point ; viz. is any one of the intersections of with the curve ; we have then similarly any one of the tangents Fig. 1 .
from to the curve , touching it , say at , and intersecting the curve in a point ; viz. the point is any one of the intersections in question ; and then in like manner we have any one of the tangents from to the curve , touching it , say at , and intersecting the curve in a point ; viz. is any one of the intersections in question .
Suppose that to a given position of there correspond positions of ; it is easy to find the value of ; viz. if ( as above tacitly supposed ) the curves are all of them distinct curves , then the number of the tangents ; there are on each of them points ; through each of these we have tangents ; on each of these points ; through each of these tangents ; and on each of these points ; that is , BcDeFa .
But if some of the curves become one and the same curve\mdash ; if , for instance , , the point aumber oemaining gents ieach tangent meetsthe curve ioint acounting ooint Bcounting twice , line is here a tangent from a point on the curve , we exclude the tangent at the or , we have such factor in its original form or diminished by 1 , 2 , or 3 , as the case may be .
Similarly starting from , considered as a given point on the curve , we find the number of the corresponding points ; thus in the case where the curves are all distinct curves , we have FeDcBa ( ; and so in other cases we find the value of .
The points have thus correspondence , where the values of are found as above .
2 .
There will be occasion to consider the case where in the triangle aBcDeF ( or say the triangle aBcDeFa ) the point is not subjected to any condition whatever , but is a free point .
There is in this case a " " locus of which is at once constructed as follows : .
starting with am arbitrary tangent of the curve , touching it at and intersecting the curve in a point ; through we draw to the curve the tangent touching it at and intersecting the curve in a point ; and finally from to the curve the tangent , touching it at and intelsecting the original arbitrary tangent in a point , which is a point on the locus in question .
We can , it is clear , at once determine how many points of the locus lie on an arbitrary tangent of the curve ( or of the curve F ) .
3 .
The general form of the equation of correspondence is ; viz. if on a curve for which twice the deficiency is we have a point corresponding to certain other points , . . .
in such wise that have an correspondence , correspondence , \amp ; c. ; and if ( a ) be the number of the united points ( b ) the number of the united points \amp ; c. ; and if moreover for a given position of on the curve the points are obtained as the intersections of the curve with a curve ( depending on the point P ) which meets the curve times at times at each of the points times at each of the points , \amp ; c. ; then the relation between the several quantities is as stated above : see my ' Second Memoir on the Curves which satisfy given conditions Philosophical Transactions , vol. 159 ( 1868 ) , pp. 145-172 .
I omit for the present purpose the term Supp treating it as included in the other terms .
4 .
In the present case we consider , as already mentioned , the unclosed trilateral aBcDeFg , where the angles are on one and the same curve ( the curve in the general theorem ) ; and the curve is the system of lines which by their intersec* To avoid confusion with the notation of the present memoir , I abstain in the text from the use of as denoting the deficiency , and there is a convenience in the use of a single symbol for twice the deficiency ; but writing for the moment to denote the deficiency , I remark , in passing , that perhaps the true theoretical form of the equation is ; viz. the point is here considered as having with itself correspondence , the number of the united points therein being THE IN-AITD-CIRCUMSCRIBED TRIANGLE 381 tion with the curve determine the points .
Considering these as the points of the general theorem we have : I change the notation , and instead of write ; viz. I take ( g ) for the number of the united points , and suppose that the points have correspondence .
The most simple case is when the curve is distinct from each of the curves ; here all the intersections of the line-system with the curve are points , that is we have only the correspondence ; and since the line-system does not pass through the point , we have simply 5 .
But suppose that the curves are one and the same curve , say that ; understanding by the point the point of contact of a line with the curve , then the intersections of the line-system with the curve are the points each once , the points each twice , and the points each as many times as there are lines through the point , say each times .
( In the present case , where the curves are identical , we have or according as the curve is or is not distinct from the curve ; in the cases afterwards referred to , the values may be or ; that is , we have always , as the case may be .
) We have to consider the several correspondences , , is as before ; and the form of the theorem is where the symbols denote as follows , viz. have correspondence and No. of united points so that the determination of here depends upon that of and 6 .
The curve might however have been identical with only one of the curves ; viz. if , but is a distinct curve , then the equation will contain the term but not the term ; and so if , but is a distinct curve , then the equation will not contain but will contain : it is to be noticed that in this last case we have or , according as the curve is not , or is , one and the same curve with F. The determination of ( g ) here depends upon that of or , as the case may be .
These subsidiary values ftIld are obtained by means of a more simple application of the principle of correspondence , as will appear in the sequel* , but for the moment I do not pursue the question .
Locus of a free angle .
Nos. 7 to 14 .
7 .
I consider the case where is distinct curve , and as was seen , the equation is simply *See post , Nos. 24 et seq. this line .
8 .
Suppose now that in the triangle aBcDeFa the point is a free point , we have , as above mentioned , a locus of , and the united points on the arbitrary line are the intersections of the line with this locus ; that is , the locus meets the arbitrary line in points ; or , what is the same thing , the order of the locus is 9 .
I stop for a moment to remark that in the particular case where the curve is a point , then in the construction of the locus of the arbitrary tangent is an arbitrary line through , and the construction gives on this lin positions of the point .
But drawing from a tangent to the curve , and thus constructing in order the points , the construction shows that is an -tuple point on the locus ; and ( by what precedes ) an arbitrary line through meets the locus in other points ; that is , in the particular case where the curve is a point , the order of the locus of is , which agrees with the foregoing result .
10 .
The construction for the locus of may be presented in the following form : viz. drawing to the curve a tangent , meeting the curves in the points respectively ; then if from any point we draw to the curve a tangent , and from any point to the curve a tangent , the tangents intersect in a point on the required locus .
Hence if in any particular case ( that is for any particular position of the tangent ) the lines become one and the same line , the point be an indeterminate point on this line ; that is , the line in question will be part of the locus of 11 .
The case cannot in general arise so long as the curves are distinct from each other ; but when these are one and the same Fig. 2 .
First-mode figure .
question of the tangent of , the points coincide with each other , and we have thus the coincident tangents and to the identical curves .
It is further to be remarked that the number of the points of intersection is ; from each of these there are tangents to the curve ( in THE IN-AND-CIRCUMSCRIBED TRIANGLE .
383 all tangents ) , and each of these counts once in respect of each of the tangents to the curve , that is , it counts times .
We have thus , as part of the locus of , ce .
lines each times , or , say , first-mode reduction 12 .
The second mode is that shown in the .
annexed ' ' second-mode figure The Fig. 3 .
Second-mode figure .
tangent from is here a common tangent of the curves , and .
This meets the curve in points , and the curve in points ; and attending to any pair of points , these give the tangents , coinciding with the common tangent in question , and forming part of the locus of .
The number of the common tangents is ; but each of these counts once in respect of each combination of the points , that is in all times .
And we have thus as part of the locus BD lines each times , or , say , second-mode reduction .
This is ( as it happens ) the same number as for the first mode ; but to distinguish the different origins I have written as above .
B. and respectively .
13 .
It is important to remark that each of the two modes arises whatever relations of identity subsist between the curves , and , but with considerable modification of form .
Thus if the curves are identical but distinct from , then in the first-mode figure may be a node or a cusp of the curve , or it may be a point of contact of a common tangent of the curves , and .
As regards the node , remark that if we consider a tangent of meeting the curve in the neighbourhood of the node , then of the two points of intersection each in succession may be taken for the point and the other of them will be the point ; so that the node counts twice .
It requires more consideration to perceive , but it will be readily accepted that the cusp counts three times .
Hence if for the curve the number of nodes be and that of cusps , the value of the first-mode reduction is , or , what is the same thing , it is As regards the second-mode figure , the only difference is that will be here any pair of intersections ( each pair twice ) of the tangent with the curve ; the value is thus It would be by no means uninteresting to enumerate the different cases , and indeed there might be a propriety in doing so here ; but I have ( instead of this ) considered the several cases when and as they arise in connexion with any of the cases of the in-andcircumscribed triangle .
14 .
Observe that the general result is , that in the case of the identity of the curves and , but not otherwise , the locus of includes as part of itself a system of lines ; or , say , that it is made up of these lines , and of a residual curve of the order Red .
, which is the proper locus .
of the order of the proper locus of 16 .
It is however to be noticed that if the curve , being as is assumed distinct from the curves , and , is identical with one or both of the remaining curves , the foregoing expression \mdash ; Red .
may include positions which are not true solutions of the problem , viz. the curve may pass through special points on the proper locus of giving intersections which are a new kind of heterotypic solutions* .
17 .
But this cannot happen if the curve is distinct also from the curves ; or , say , simply when is a distinct curve .
The conclusion is , that in the case where is a distinct curve we have \mdash ; Red .
, where the term " " Red vanishes except in the case of the identity of the curves ; and that when this identity subsists it is times the reduction in the order of the locus of considered as a free angle ; viz. this consists of a first-mode and a secondmode reduction as above explained .
Remarks in regard to the Solutions for the 52 Cases .
Art .
Nos. 18 to 23 .
18 .
Before going further I remark that the principle of correspondence applies to corresponding and united tangents in like manner as to corresponding and united points , and that all the investigations in regard to the in-and-circumscribed triangle might thus be presented in the reciprocal form , where , instead of points and lines , we have lines and points respectively .
But there is no occasion to employ any such reciprocal process ; the result to which it would lead is the reciprocal of a result given by the original process , and as such it can always be obtained by reciprocation of the original result , without any performance of the reciprocal process .
* More eneraHy , if the curve be a curve identical with any of the other curves , then if treating in first instance the angle as free we find in any manner the locus of , the required positions of the angle aro the intersections of this locus and of the curve ; but these intersections win in general include interseotiom which give heterotypic solutions .
The determination of these is a matter of some delicacy , and I have in gener treated the problems in such manner that question does not arise ; but as an example see post , Case 43 .
using one process only , it may be and in general is easier and more convenient to obtain directly one than the other of two reciprocal results ; for instance , to consider the case rather than , or vice ; and that it is sufficient to do this , and having obtained the one result , directly to deduce from it the other by reciprocity ; but that it may nevertheless be interesting to obtain each of the two results ectly .
20 .
It is moreover obvious that although the several forms of the same case , for instance Case 2 , , or , are absolutely equivalent to each other , yet that , when as above we select a vertex , and seek for the number of the united points , the process of obtaining the result will be altogether different according to the diffelent form which we employ .
For instance , in the case just referled to , if the form is taken to be or , then the equation is applicable to it ; but not so if the form is taken to be .
It would be by no means uninteresting in every case to consider the several forms successively and out the result each of them ; I shall not , however , do this , but only consider two or more forms of the same case when for comparison , illustration , verification , or otherwise it appears proper so to do .
The translation of a result , for instance , of a form or into that for the is so easy and obvious , that it is not even necessary formally to make it .
21 .
I do not at present further consider the general theory , but proceed to consider in order the 52 cases , interpolating in regard to the general theoly such further discussion or explanation as may appear necessary .
In the several instances in which the equation is applicable , it is sufficient to write down the values of , the mode of obtaining these being already explained .
The 52 cases for the in-and-circumscril ) triangles .
Case 1 .
No identities .
BcDeFa , , aceBDF .
Case 2 .
DeFx , , eBDF .
Second process , for form .
The equation of correspondence is here ; but the points being given as all the intersections of the curve by the line-system which does not pass through , we have ; so that ; and then , DcBx , giving the former lesul * Of course , the result is obtain in the form to the new form of spccification , viz. here it is cBDF ; and so in other instances ; but it is unnecessary to refcr to this ch BcXeFx , , XxceBF .
Observe this is what the result for Case 1 becomes on writing therein , viz. the opposite cul.ves may become one and the same curve without any alteration in the form of the result .
Case 5 .
, where ( X-2 ) ; ceDF .
Case 6 .
: perhaps most easily by reciprocation of Case 7 ; or Second process , functionally by taking the curve to be the aggregate curve .
The triangle aBcDeF is here in succession each of the eight triangles : ' where the two top triangles give and respectively ; the remaining triangles all belong to Case 2 , and those of the first column give each , and those of the second column each xBDF .
We have thus Hence obtaining a particular solution and adding the constants , we have ; THB IN-AND-CIRCUMSCRIBED IRIANGLE .
387 il is easy to see that are independent of the curves ; and taking each of these to be a point , and the curve to be a conic , then it is known that we have that is The case where the curve is a line gives , that is , ; but it is not easy to find another condition ; assuming however , we have , and thence or say : this is a good easy example of the functional process , the use of which begins to exhibit itself ; and I have therefore given it , notwithstanding the difficulty as to the complete determination of the constants .
Third process .
The equation of correspondence is but for the correspondence we have and for the correspondence we have whence ; and then ; that is Moreover ( if be the number of cusps of the curve ) , and the resulting value is ; that is where , however , the term is to be rejected .
I cannot quite explain this ; I should rather have expected a rejection , introducing the term .
For consider a tangent from the curve from a cusp of the curve : there are such tangents ; each gives in the neighbourhood of the cusp two points , say ; and from these we ; that The reductions of the two modes are as above , with only the valiation that in the present case is the same curve with the two curves .
That of the first mode is , and that of the second mode is , which is together they are , or subtracting , we have Case 8 .
DeFx , , Case 9 . .
By reciprocation of 8 .
No. Case 10 .
, Case 11 . .
By reciprocation of 10 .
No. Second process : form giving the former result .
Case 12 .
, yYBF .
Case 13 . .
By reciprocation of 12 .
, No. THB IN-AND-CIRCUMSCRIBED Case 14 .
Case 15 . .
By reciprocation of 14 .
No. Case 16 .
, Case 17 .
But we have here as an axis of symmetly , so that each triangle is counted twice .
or the number of distinct triangles is Case 18 .
XeFx , , Casc 19 .
YxDyXa , , Case 20 .
Case 21 .
xyXY Case 22 .
ZyXzYx , , xyzXYZ .
PBOEESSOR CAYLEY ON TllE PROBLElf OF Case 23 .
, xyzXYZ Case 24 .
yzYZ Case25 .
, But we have here as an axis of symmetry , so that each triangle is counted twice , or the number of distinct triangles is Case 26 .
, .
Case 27 . .
By reciprocation of 28 .
No. where each triangle is counted twice , so that the number is really one half of this .
Case 28 .
Here Red .
, .
The reductions are those of the first and second mode as explained above , with the variation that the curves and are here identical , , and that the curve is identical with the curves First-mode where refer curve , which is ; that is , the reduction is .
*Printed in the Table , erroneously , THE IN-AND-CIRCUIISCRIBED TRIANGLE .
And the second-mode reduction is where refer curve , which is ; that is , the reduction is Hence the two together are ; and from we have .
; but on account of the symmetry each triangle is reckoned twice , and the number of triangles is Case 29 .
, Case 29 .
Second process .
the ; here No. \mdash ; Red .
, and There is a first-mode reduction , viz. this is { which is ; and a second-mode reduction , .
Hence the two together are whence the result is which rees with that obtained above .
On account of the symmetry we must divide by 2 .
PROFESSOR CAYLEY ON THE PROBLEM OF Case 30 . .
By reciprocation of 29 .
No. On account of the symmetry we must divide by 2 .
Case 31 .
, Case 32 . .
By leciprocation of 31 .
No. Case 33 . .
By reciprocation of 34 .
No. Case 34 .
Table , erroneously , , Case 35 .
Case 36 . .
By reciprocation of 35 .
No. Casc 37 . .
By reciprocation of 38 .
No. xyXY - F. Case 38 .
, Case 39 .
Functional process ; the curve is assumed to be the aggre ate of curves , orming the enumeration THE IN-AND-CIRCUMSCRUBED TRIANGLE . .
x'X'x .
x'X'x .
Case .
39 \amp ; c. 10 ( where the second column is derived from the first by a mere interchange of the accented and unaccented letters ) , I annex to each line the number of the case to which it belongs ; thus XxDxF is , which is Case 10 , and so in the other instances .
that cases 10 and 14 occur each twice , we have thus multiplied into . .
( 10 ) . .
( 6 ) . .
( 14 ) . .
( 12 ) . .
( 8 ) where the 's refer to the like functions with the two sets of letters interchanged .
Developing and collecting , this is multiplied into and thence multiplied into where the constants have to be determined .
Now for a cubic curve the number of triangles vanishes ; that is , we have in each of the three cases , , IIDCCCLXXI .
3 I PBOFESSOR CAYIiBY ON TIfE PROBLEM OF and we thus obtain the three equatipns giving .
Whence , finally , Second process , by correspondence .
We have and thence , Moreover , and , as is easily obtained , but see also post , No. 29 ; hence multiplied into ( X \mdash ; 2 ) .
; but I reject the term as in fact giving a heterotypic solution ; I do not go into the explanation of this .
And then substituting for its value , we have multiplied into ( X\mdash ; 2 ) .
where the second factor is which is the foregoing result .
Case 40 . .
By reciprocation of 39 , No. THB IN-AND .
CIBCUMSCRIBED TRIANGLE .
395 .
Case 41 .
Case 42 .
Functionally , viz. the curve is supposed to be the aggregate of two curves , say The enumeration is Case .
Bx X X. X X. X X. X X. X X. X X. X X. X X. X X. X , ( 42 ) \amp ; c. ( 11 ) ( 11 ) ( 17 ) ( 10 ) ( 19 ) ( 21 ) ( 10 ) whence multiplied into . .
( 11 ) . .
( 17 ) . .
( 10 ) . .
( 19 ) .
( 21 ) where the 's refer to the like functions with the two sets of letters interchanged .
Developing and collecting , we multiplied into we have : and by the consideration that are given as intersections , a double intersection , of the curve with the first polar of the point , which first polar does not pass through whenoe and c\mdash ; 7 so that this is .
Also so that multiplied into , viz. this is ' THB IN-AND-CIRCUMSCRIBED TRIANGLB Third process : form \mdash ; Red .
, .
The first-mode reduction is here Fig. 4 .
; where the last term es from the tangents and , each with a cuspidal tangent , as shown in the figule .
The second-mode reduction is , so that the two leductions together are viz. this is ; or substituting for and .
the values and respectively , and reducing , it is Hence subtracting from written in the form the result is On account of the we must divide by 2 .
Case Suppose for a moment that the angle is a free point ; the locus of is a the order of which is obtained from Case 28 , by writing ; the locus .
in question meets a curve order in points ; wherefore the order of the locus is , and this locus meets the curve in a number of points , viz. this is the numbel of positions of the angle ; but several of these belong to special of the triangle aBcDeF , giving heterotypic solutions , which are to be rejected ; the required number is thus \mdash ; Red .
curve in points , and selecting any one of them for and any other for , we have from the last-mentioned point tangents to the curve ; and in respect of each of these a position of coincident with .
The reduction on this account is ; but since we may in the figure interchange and and , we have the same number belonging to the coincidence of the angles , or together the reduction is .
from the two points respectively drawing two other tangents to the curve , we have a triangle wherein the angles all coincid . .
The side may be a proper tangent to the curve , or it may TlfE IN AND-CIRCUMSCRIBED TRIANGLE .
399 pass through a node or a cusp of this curve , viz. it is either a common tangent of the curves and ( as in the figure , except that for greater distinctness the points and are there drawn nearly instead of actually coincident ) , or it may be a tangent to the curve from a node or a cusp of the curve ; we have thus the numbers Common tangent , Tangent from node , Tangent from cusp 2 ; but ( as we are counting intersections with the curve ) the second of these , as being at a node of this curve , is to be taken 2 times ; and the third , as being at a cusp , 3 times ; and the three together are thus , .
The reductions altogether are , which is ; and subtracting from the before-mentioned number the required number of positions of the angle is .
The number of triangles is on account of the symmetry equal to one-sixth of number .
Case 44 .
, : there is a division by 2 on account of the symmetry .
Case 45 .
, XYxy Case 46 . .
By reciprocation of 47 .
No. : thel.e is a division by 2 on account of the symmetry .
Case 47 .
functional process is exactly the same as for No. 39 , with only written instead of ; hence No. : there is a division by 2 on account of the symmetry .
Case 48 .
The functional process , writing , would be precisely the same as for Case 42 , with only the factor written instead of ; and we have thus the like result , No. which on account of the symmetry must be divided by 2 .
Case 49 .
, Case 50 .
Functional process ; by taking the curve as the aggregate of curves , say .
The cases are . .
X . .
X .
X'x'X'x .
X . .
X . .
X . .
X .
and we thus have multiplied intc .
2(40 ) .
2(32 ) ( 42 ) 2(33 ) .
2(38 ) 2(36 ) ( 28 ) ( 29 ) where as before the 's refer to the like functions with the two sets of letters interchanged .
Developing and collecting , this is multiplied into ' -22 ; whence multiplied into , where the constants have to be determined .
We should have for a cubic curve ; viz. ; or .
Writing first the equation is giving in the three cases respectively ; and we have then , so that the required number is As a verification , observe that for a conic , , this is Second process , by correspondence : form We have \mdash ; Red .
, into There is a first-mode reduction , which is where the term arises , as shown in the figure , and a second-mode reduction , which is Fig. 8 .
; and the two together are into ( X-4 ) ; that is , into / ; and subtracting this from the foregoing value of , which is into the result is as before .
and the functional equation then is ( 60 ) THE IN-AND-CIRCUMSCRIBED TRIANGLE .
' ' ' ' ' ' ' ' ' ( 51 ) . .
( 46 ) .
( 47 ) . .
( 48 ) .
( 49 ) . .
( 43 ) . .
( 44 ) . .
( 45 ) where as before the 's refer to the like functions with the two sets of letters interchanged .
Developing and collecting , this is found to be .
\amp ; c. I abstain from writing down the remaining terms , as they can at once be obtained backwards from the value of ; they were in fact found directly , and the integration of the functional equation then gives 1 where the constants have to be determined ; I have in the first instance written , instead of , thus introducing two constants only , since it is clear from the symmetry in regard to that we must have .
We must have , when the curve is a conic or cubic .
Writing , we have and then for the conic , Writing , we have and then for the three cases of the cubic ; and We have thus the four equations all satisfied by .
Whence , finally , on account of the symmetry the number of triangles is sixth of this expression .
THE IN-AND-CIRCUMSCRIBED TRIANGLE .
The Case 52 , as belonging to a different series of Problems .
Art .
Nos. 22 to 36 .
22 .
In the foregoing Case 52 , where all the curves are one and the same curve , we have the unclosed trilateral aBcDeFg , and we seek for the number of the united points .
But we may consider this as belonging to a series of questions , viz. we may seek for the number of the united points , , , , , ( the last four of these giving by reciprocity the numbers of the united points , , , finally number united points .
It is very instructive to consider this series of questions , and the more so that in those which precede there are only special solutions having reference to the singular points and tangents of the curve , and that the solutions thus explain themselves .
23 .
Thus the first case is that of the united points , viz. we have here a point on the curve , and from it we draw to the curve a tangent touching it at ; the points and are to coincide together .
Observe that from a point in general of the curve we have X\mdash ; 2 tangents ( X the class as heretofore ) , viz. we disregard altogether the tangent the point , counting as 2 of the X tangents from a point not on the curve , and attend exclusively to the X-2 tangents from the point .
Now if the point is an inflection , or if it is a cusp , there are only X\mdash ; 3 tangents , or , to speak more accurately , one of the X\mdash ; 2 tangents has come to coincide with the tangent the point ; such tangent is a tangent of three-pointic intersection , viz. we have the point and the point ( counting , as a point of contact , twice ) all three coincidin ; that is , we have a position of the united point ; and the number of these united points is 24 .
It is important to notice that neither a point of contact of a double tangent , nor a double point , is a united point .
In the case of the point of contact of a double tangent , one of the tangents from the point coincides with the double tangent ; but the point is here the other point of contact of this tangent , so that the points are not coincident .
In the case of a double point , regarding the assumed position of at the double point as belonging to one of the two branches , then of the X\mdash ; 2 tangents there are two .
each coinciding with the tangent to the other branch ; hence , attending to either of these , the point belongs to the other branch , and thus , though and are each of them at the double point , the two do not constitute a united point .
( In illustration remark that for a unicursal curve , the position of answers to a value , and that of to a value of the parameter , viz. are the two values of at the double point ; contrariwise in the foregoing case of a cusp , where there is a single value Hence the whole number of the united points is , and this is in fact value given ( as will presently appear ) by the theory of correspondence .
I recall that I use , to denote twice the deficiency of the curve , viz. that we have 25 .
The several cases are United points .
by reciprocity , by reciprocity , by reciprocity , and so on .
26 .
The mode of obtaining these equations appears ante , Nos. 5 and 6 , but for greater clearness I will explain it in regard to a pair of the equations , say those for , Regarding as given , we draw from the tangents , touching at and besides inter secting at ( viz. the number of tangents is , and the number of the points is ; from each of the positions of we draw to the curve the ( X\mdash ; 3 ) tangents touching at and intersecting at ; the whole number of these tangents is ; and this is also the number of the points , but the number of the points is .
Now this system of the ( X-2 ) tangents is the curve of the general theory viz. the curve ( which does not pass through a ) intersects the given curve in the three classes of points , the number of intersections at a point being at a point being , and at a point being .
And we have thus the equation where are the numbers of united points and the correspondences in the three cases respectively .
27 .
Observe that we cannot , starting from , obtain in this manner the equation for the number of the united points ; for we introduce per force the points , and thus obtain the foregoing equation for .
But starting from , the tangent at this point besides intersects the curve in points , each of which is a position of ; and from each of these drawing a tangent to the curve , we have the curve consist .
ing of these tangents , not passing through , but intersecting the given curve in the three classes of points , viz. the number of intersections at each point is , at each point it is , and each point it is ; and we have thus the equation where the numbers refer to the correspondences , and what same thing ) respectively .
28 .
Correspondence .
We have and thence It was in this manner that I originally applied the principle of correspondence to investigating the number of inflections of a curve , regalding , however , the term as a special solution ; it is better to put cusp and inflection on the same footing as above .
29 .
Correspondence .
Since , we have here and whence ; this is in fact , viz. we have ( X-3 ) and therefore above , viz. the united points are the points of contact of the double tangents , and the cusps each ( X\mdash ; 3 ) times in respect of the ( X\mdash ; 3 ) tangents from it to the curve .
This is the way in which I originally applied the principle to finding the number of double tangents of a curve .
30 .
Correspondence .
By reciprocation .
31 .
It may be remarked , as regards the cases which follow , that although the result in terms of when once known can be explained and velified easily enough , there is great risk of oversight if we endeavour to find it in the first instance ; while on the other hand the transformation from the form in terms of , as given by the principle of correspondence , to the required form in terms of is by no means easy .
I in fact first obtained the expression in , and then , knowing in some measure the form of the other expression , was able to find it by the actual transformation of the expression in .
32 .
Correspondence .
From the values of and we have and then , whence , which is .
And we then , by means of the equations ( X-4 ) , , ( X-3 ) , verify that 33 .
Correspondence .
From the values of have and then ; that is , which is ( 2 ) , and then , , ; and summing these values and comparing , figure .
34 .
Correspondence .
By leciprocation of .
, 35 .
Correspondence .
By means of the values of and , we have and then , , whence , which is This result includes proper solutions of the problem of finding the number of the triangles aBcDeF , which are such that the side touches the curve at ; and also heterotypic solutions having reference to the singular points of the curve : but I have not determined the number of solutions of each kind .
36 .
Correspondence : from the values of and ; we have and then ' wherefore , viz. this is 2 ) Comparing with the expression of , Case 52 , we have ( 3 ) which difference must be the number of heterotypic solutions having relation to the singularities of the curve ; but I have not further considered this .

