]\gt ; [ 397 ] XI .
A lllemoir on the of Elliptic Functions .
By Professor CAYLEY , F.R.S. Received November 14 , 1873 , \mdash ; Read January 8 , 1874 .
THE theory of Transformation in Elliptic Functions was established by JACOBI in the ' Fundamenta Nova ' ( 1829 ) ; and he has there developed , transcendentally , with an approach to completeness , the general case , an odd number , but algebraically only the cases and ; viz. in the general case the formulae are expressed in terms of the elliptic functions of the nth part of the complete integrals , but in the cases and they are expressed rationally in terms of and ( the fourth roots of the original and the transformed moduli respectively ) , these quantities being connected by an equation of the order 4 or 6 , the modular equation .
The extension of this algebraical theory to any value whatever of is a problem of great interest and difficulty : such theory should admit of being treated in a purely algebraical manner ; but the difficulties are so great that it was found necessary to discuss it by means of the formulae of the transcendental theory , in particular by means of the expressions involving the exponential of , or say by means of the -transcendents .
Several important contributions to the theory have since been made :SoHNKE , " " Equationes Modulares pro transformatione functionum Ellipticarum CRELLE , .
( 1836 ) , pp. 97-130 where modular equations found cases ; JOUBERT , " " Sir divers equations analogues aux equations modulaires dans la theory .
des fonctions elliptiques Comptes Rendus , .
xlvii .
( 1858 ) , pp. 337-345 ( relating among other things to the multiplier equatiou for the determination of JACOBI'S M ) ; and KoNIGSBERGER , " " Algebraische Untersuchungen aus der Theolie der elliptischen Functionen CRELLE , .
lxxii .
( 1870 ) , pp. 176-275 ; together with other papers by JOUBEIiT and by HERMITB in later volumes of the 'Comptes Rendus , ' which need not be more particularly referred to .
In the present Memoir I carry on the theory , algebraically , as far as I am able ; and I have , it appears to me , put the purely algebraical question in clearer light than has hitherto been done ; but I still find it necessary to resort to the transcendental theoly .
I remark that the case ( next succeeding those of the ' Fundamenta Nova on account of the peculiarly simple form of the modular equation , presents but little difficulty ; and I give the complete formulae for this case , obtaining them as well algebraically as transcendentally ; I also to a considerable extent discuss algebraically the case of the next succeeding prime value For the sake of completeness I reproduce SOHNKE'S modular equations , exhibiting them IIDCCCLXXIV .
3 for greater clearness in a square form , and adding to them those for the non-prime cases and ; also a valuable table given by him for the powers of ; and I give other tabular l.esults which are of assistance in the theory .
The General Problem.\mdash ; Article Nos. 1 to 6 .
1 .
Taking a given odd number , I write where are lational and integral functions of being each of them of the order , or , what is the same thing , being each of them of the order ; that is , Order ofP in is ; whence in the first case No. of coefficients of and is , and in the second case No. is , as before .
Taking .
, . .
, the formula is the number of coefficients being as just explained .
Starting herefrom I reproduce in a somewhat altered form the investigation in the ' Fundamenta Nova , ' as follows .
2 .
If the coefficients are such that the equation remains true when we therein change simultaneously into and into , then the variables will satisfy the differential equation constant , value which , will appear , given ; and the problem of transformation is thus to find the coefficients so that the equation may remain true on the above simultaneous change of the values of In fact , observing that the original equation and therefore the new equation are each satisfied on changing therein simultaneously into , it follows that the equation may be written in the four forms , the common denominator being , say , where are all of them rational and integral functions of ; and this being so , the differential equation will be satisfied .
ELLIPTIC BITNCTIONS .
3 .
To develop the condition , observe that the assumed equation gives suppose , where are functions each of them of the degree in .
( Hence , if with JACOBI denotes the value , we have , as mentioned .
) Suppose in general that being any integral function , we have ; viz. let be what becomes when is changed into and the whole multiplied by Let be the value of obtained by writing for ; then , observing that in the expression for the degree of the numerator exceeds by unity that of the denominator , we have whence ; and the functions may be such that this shall be a constant value , ; viz. this will be the case if which being so , the required condition is satisfied .
4 .
I shall ultimately , instead of introduce JACOBI'S but it is for the present convenient to retain , and instead of to introduce the quantity connected with it by the equation ; or say the value of is .
The modular equation in its standard form is an equation between , which , as will appear , gives rise to an equation of the same order between ; and writing herein the resulting equation contains only integer powers of , that is of , and we have an -form of the modular equation , or say an -modular equation , of the same order in as the standard form is in ; these -forms for , 5 , 7 , 11 will be given sently .
5 .
Suppose then , being a constant , that we have identically ; this implies ( In fact if then and the assumed equation ives ; that is , , . . .
and therefore ) From these equations , that is , as it should be ; so that signifying as above , the required condition will be satisfied if only ; or substituting for their values , if only , where each side is a function of of the order , or the number of terms is , the several coefficients being obviously homogeneous quadric functions of the coefficients of , Q. We have thus equations , each of the form , where are given quadric functions of the coefficients of , say of the coefficients \amp ; c. , and where is indeterminate .
6 .
We may from the equations eliminate the ratios , thus obtaining an equation in ( involving of course the parameter k ) which is the -modular equation above referred to ; and then denoting any root of this equation , the equations give a single value for the set of ratios , so that the ratio of the functions is determined , and consequently the value of as given by the equation , or The entire problem thus depends on the solution of the system of equations , .
The -Modular Equations , , 5 , 7 , ll.\mdash ; Article No. 7 .
7 .
For convenience of reference , and to fix the ideas , I give these results , calculated , as above explained , from the standard or uv-forms .
OF ELLIPTIF FUNCTIONS .
I ' we have , we have \mdash ; 1936 \mdash ; 7876 \mdash ; TRANSFORMATION : -30800 \mdash ; Equation-systems for the cases , 5 , 7 , 9 , ll.\mdash ; Article Nos. 8 to 10 .
8 .
, cubic mation .
( here and in the other cases ) . .
The condition heal is and the system of equations thus is , and similarly in the other cases ; for these it will be enough to write down the equationsystems .
, quintic transformation .
, .
OF ELLIPTIC FUNCTIONS .
, septic transformaliolL , , .
, enneadic transformation .
, , , .
, endecadic transformation .
, , , , , and so on .
9 .
It will be noticed that if the coefficients of .
taken in order are then in every case the and last equations are .
Putting in the first of these , the equation becomes each side is a perfect square ; and in extracting the square root we may without loss of generality take the roots positive , and write This speciality , although it renders it proper to employ ultimately in place of , producesreally no depression of order ( viz. the -form of the modular equation is found to be of the same order in that the standard or -form is in ) , and is in another point of view a disadvantage , as destroying the uniformity of the several equations : in the discussion of order I consequently retain .
Ultimately these are to be replaced by ; the change in the equation-systems is so easily made that it is not necessary here to write them down in the new form in 10 .
The case has to be considered in the discussion of order , but we have thus only solutions which are to be rejected ; in the proper solutions is not , and it may therefore for convenience be taken to be .
We have then .
The last equation becomes therefore ; or recollecting that / 3 is connected with the multiplier by the relation that is , and substituting for its value , the equation becomes ; that is , the first and last coefficients are 1 , and the second and penultimate coeffi ' cients are each expressed in terms of , M. The cases are so far peculiar , that the only coefficients are , or ; in the next case the only coefficients are , and we have in this case all the coefficients expressed as above .
The \mdash ; Order of the Systems.\mdash ; Article Nos. 11 to 22 .
11 .
In the general case , an odd number , we have and coefficients connected by a system of equations of the form where , . .
, . . .
are iven quadric functions of the coefficients .
omitting the , there remains a system of equations of the form . .
or say ' which determine the latios : : 7 . . .
of the coefficients ; and to each set of ratios there corresponds a single value of .
The order of the system , or number of sets of ratios , is .
; and this is consequently the number of values of , or older of the equation for the determination of ; viz. but for reduction the order in of the -modular equation would be .
In the case , this is , which is right , but for any larger value of the order is far too high ; in fact , assuming ( as the case is ) that the order is equal to the order in of the -form , the order should for a prime value of be , and for a composite value not con- taining any square factor be sum of the divisors of .
I do not attempt a general investigation , but confine myself to showing in what manner the reductions arise .
OF ELLIPTIC FUNCTIONS .
12 .
I will first consider the cubic transformation ; here , writing for convenience , the equations give , that is , and ; the equation in gives , and we have thence that is the modular equation ; and then , that is , or , which is , say we have ; consequently , and which completes the theory .
13 .
Reproducing for this case the general theory , it appears a priori that is determined by a quartic equation ; in fact from the original equations eliminating , we have an equation where are quartic functions of , ( 3 ; that is the latio has four values , and to each of these there corresponds a single value of ; viz. is determined by a quartic equation .
14 .
Considering next the case , the quintic transformation ; the elimination of gives the equations where \amp ; c. are all quadric functions of .
We have thence sets of values of ; viz. considering as coordinates in plano , the curves are quartic curves intersecting in16 points ; but among these are included the four points ( in fact the point four times ) , which are not points of the curve ; there remain therefore16\mdash ; 4 , intersections ( agreeing with the general value ) .
Hence is in the first instance determined by an equation of the order 12 ; but the proper order being there must be a factor of the order 6 to be rejected .
To explain this and determine the factor , observe that the equations in question are ; the first of these has at the point a double point , the second a triple point ; or there are at the point in question 6 intersections ; but 4 of these are the points which give the foregoing reduction ; we have thus the point , counting twice among the twelve points .
Writing in the two equations , the equations become , viz. these will be satisfied if , that is , the curves pass through each of the two points , and these values satisfy ( as in fact they should ) the third equation , ; it is moreover easily shown that the three curves have at each of the points in question a common tangent ; viz. taking as current coordinates , the tangent at the point of the second curve has for its equation and for this becomes , viz. this is the line from the point to the point .
And similarly for the other two curves we find the same equation for the tangent .
Hence among the 12 points are included the point twice , and the points .
each twice : we have thus a reduction 15 .
Writing in the equations , the first and third are satisfied identically , and the second becomes , that is the equations give ; writing , they become viz. putting herein , the equations again give ; hence the factor of the order 6 is , and the equation of the twelfth order for the determination of is where is the -modular equation above written down .
16 .
Reverting to the equation it is to be observed that for , that is , this becomes simply , which is the transformation of the orderl ; the corresponding value of the modulus is and the equation then gives , which is replaced by If in the same equation we write , that is , then ( without any use of the equation ) we have , the transformation of the order 1 ; but although this is so , the fundamental equation , which , putting therein , becomes ; that is is not satisfied by the single relation , but necessitates the further relation OF ELLIPTIC EUNCTIONS .
The thing to be observed is that the extraneous factor , equated to zero , gives for the value corresponding to the transformation of the order 1 .
17 .
Considering next , the septic transformation ; we have here between a fourfold relation of the form , V ' , where , as before , \amp ; c. are quadric functions , and the number of solutions is here 8 .
; to each of these corresponds a single value of , or is in the first instance determined by an equation of the order 32 .
But the order of the modular equation is ; or representing this by , the equation must be viz. there must be a special factor of the order 24 .
18 .
One way of satisfying the equations is to write therein ; the equations thus become ; or putting , which with instead are the very equations which belong to the cubic transformation ; hence a factor is Observe that for the values in question , if and therefore which is the formula for a cubic transformation .
19 .
The equations may also be satisfied by writing therein ; in fact substituting these values , they become ; the ffist and last of these are , which being satisfied the second and third equations are satisfied identically ; and these are the formulae for a cubic transformation ; that is , we again have the factor Observe that for the values in question , we have , ; so that , writing , we have for the value which is the formula for a cubic transformation .
20 .
It is important to notice that we cannot by writing or reduce the tran sformation to a quintic one ; in fact the equation shows that if either of these equations is satisfied the other is also satisfied ; and we have then the foregoing case , giving not a quintic but a cubic transformation .
And for the same reason we cannot by writing or reduce the transformation to the order 1 .
There is thus no factor 21 .
As regards the non-existence of the factor , I further verify this by writing in the equations ; they thus become , , which it is to be shown cannot be satisfied in general , but only for certain values of Reducing the last equation , this is , which , combined with the first , gives ; and if for convenience we assume , and write also ( that is ) , then the values of are ; which values , substituted in the second and third equations , give two equations in 7 , ; and from these , eliminating 7 , we obtain an equation for the determination of , that is of .
In fact the second equation gives ; or , dividing by 7 and reducing , , that is , or , as this may also be written , that is Moreover the third equation gives , that is , ; or dividing by , it is ; OF ELLIPTIC whence also Also wherefore or or that is or finally We have thus , that is or , or else , that is or ; viz. the equation in is these being in fact the values of given by the modular equation on putting therein The equation of the order 32 thus contains the factor at least twice , and it does not contain either the factor , or the factor belonging to the quintic transformation ; it may be conjectured that the factor presents itself six times , and that the form is ; but I am not able to verify this , and I do not pursue the discussion further .
22 .
The foregoing considerations show the grounds of the difficulty of the purely algebraical solution of the problem ; the required results , for instance the modular equation , are obtained not in the simple form , but accompanied with special factors of high order .
The transcendental theory affords the means of obtaining the results in the proper form without special factors ; and I proceed to develop the theory , reproducing the known results as to the modular and multiplier equations , and extending it to the determination of the transformation-coefficients The Modular Equation.\mdash ; Article Nos. 23 to 28 .
23 .
Writing , as usual , , we have , a given function of , viz. suppose ; and th , is being so , the several values of and of the other quantities in question are ] given in terms of The case chiefly considered is that of an odd prime ; and unless the contrary is stated it is assumed that this is so .
We have then transformations corresponding to the same number of values of ; these may be distinguished as ; viz. writing to denote an imaginary n-th root of unity , we have , , \amp ; c. , .
Observe , \mdash ; for The occurrence of the fractional exponent is , as will appear , a circumstance of great importance ; and it will be convenient to introduce the term " " octicity .
an expression of the form positive integer exceeding a rational function of q ) may be said to be of the octicity 24 .
The modular equation is of course ; say this is so that \amp ; c. In the development of these expressions , the terms having a fractional exponent denominator would disappear of themselves , as involving symmetrically the several n-th roots of unity , and each coefficient would be of the form , a rational and integral function of .
It is moreover easy to see that , for the several coefficients will denote the positive residue of , . . .
respectively .
Hence assuming , as the fact is , that these coefficients are severally rational and integral functions of , it follows that the form is having the foregoing values for the several coefficients respectively .
And it being known that the modular equation is as regards of the order , there is a known limit to the number of terms in the several coefficients respectively .
We have thus for each coefficient an identity of the form where A and being each of them given in terms of , the values of the numerical coefficients , . .
can be determined ; and we thus arrive at the modular equation .
25 .
It is in effect in this manner that the modular equations are calculated in SOHNXE'S Memoir .
Various relations of symmetry in regard to and other known properties of the modular equation are made use of in order to reduce number of the unknown coefficients to a minimum ; and ( what in practice is obviously an important simpliiication ) instead of the coefficients \amp ; c. , it is the sums of powers \amp ; c. which are compared with their expressions in terms of , in order to the determination of the unknown numerical coefficients . .
The process is a laborious one ( although less so than perhaps might beforehand have been imagined ) , involving very high numbers ; it requires the development up to high powers of , of the high powers of the before mentioned function ; and SOHNKE gives a valuable Table , which I reproduce here ; adding to it the three columns which relate to OF ELLIPTIC CTIONS .
PROFESSOR CAYLEY ON THE TRANSFORMATION OF ELLIPTIC FUNCTIONS .
26 .
I give from SOHNKE the series of modular equations , adding those for the composite cases and , as to which see the remarks which follow the Table . . . . . . . . . . . . . . . . . .
1 @ 414 PROFESSOR CAYYIiY ON 'HIE TRANSFORMATION 1 .
OF ELLIPTIC FUNCTIONS .
PROFESSOR CAYLEY ON @HH TBANNSFORMATIN OF ELLIPTIC FUNCTIONS .
Various remarks arise on the Tables .
Attending first to the cases a prime number ; the only terms of the order in or are , viz. or 5 Uae sign is - , but or 7 the sign i. And there is in every case a pair of terms and , having coefficients equal in absolute magmtude , but of opposite signs , or of the same sign , in the two cases respectively .
Each Table is symmetrical in regard to its two diagonals respectively , so that every non-diagonal coefficient occurs ( with or without reversal of sign ) times ; viz. in the case or 7 this is a perfect symmetry , without reversal of sign ; but in the case or 5 it is , as regards the lines parallel to either diagonal , and in regard to the other diagonal , alternately a perfect symmetry without reversal of sign and a skew symmetry with reversal .
Thus in the case , the lines parallel to the dexter diagonal are ( symmetrical ) , ( skew ) , ( symmetrical ) , and so on .
The same relation of symmetry is seen in the composite cases and , both belonging to or 7 , If , as before , is prime , then putting in the modular equation , the equation in the case or 7 becomes , but in the case or 5 it becomes 27 .
In the case a composite number not containing any square factor , then dividing in every possible way into two factors ( including the divisions and 1 .
) , and denoting by an imaginary b-th root of unity , a value of is ; so that the whole number of roots ( or order of the modular equation ) is , if be the sum of the divisors of .
Thus , the values are 1 3 5 l.oots ; and the order of the modular equation is .
The modular equation might thus be obtained as for a prime number ; but it is easier to decompose into its prime factors , and consider the transformation as compounded of transformations of these prime orders .
Thus , the transformation is compounded of a cubic and a quintic one .
If the of the cubic transformation be denoted by , then we have ; and to each of the four values of corresponds the six values of belonging to the quintic transformation given by The equation in is thus where are the roots of the equation in , viz. we have ; and it was in this way that the equation for the case was calculated .
Observe that writing , we have , or say .
The equation in thus becomes , that is 28 .
The case where has a square factor is alittle different ; thus , the values are , 1 , 3 9 , roots ; but here being an imaginary cube root of unity , the second term denotes the three values , , the first of which is , and is to be rejected ; there remain roots , or the equation is of the order 12 .
the equation as compounded of two cubic transformations , if the value of for the first of these be , then we have ; and to each of the four values of correspond the four values of given by the equation One of these values is however , since the -equation is satisfied on writing therein ; hence , writing , we have an equation containing the factor , and which , divested hereof , gives the required modular equation of the order 12 , which was in fact obtained in this manner .
Observe that writing we have , or say ; the modular equation then becomes that is The Multiplier Equation.\mdash ; Article No. 29 .
29 .
The theory is in many respects analogous to that of the modular equation .
To each value of there corresponds a single value of ; hence , or what is the same thing , is determined by an equation of the same order as , viz. being prime , the order is .
The last term of the equation is constant , and the other coefficients are rational and integral functions of , of a degree not exceeding ; and not only so , but they are , , rational and integral functions of , and , alternately of this form , and of the same form multiplied by the factor .
OF ELLIPTIC FUNCTIONS .
421 The values are in fact given as transcendental functions of ; viz. denoting by the values corresponding to , respectively , and writing then we have ' . .
( an imaginary root of unity ) Hence , the form of the equation being known , the values of the numerical coefficients may be calculated ; and it was in this way that JOUBERT obtained the following results .
I have in some cases changed the sign of JOUBERT'S multiplier , so that in every case the value corresponding to shall be The equations are:\mdash ; , this is , it is or 1 , this is , this is , it is this is , it is .
7.11.512 .
37.256 .
131072 OF ELLIFPfC IONS .
The Multiplier as a rational function of , v.\mdash ; Article Nos. 30 to 36 .
30 .
The multipher , as having a single value corresponding to each value of , is necessarily a rational function of ; and such an expression of can , as remarked by KoNIGSBERGER , be deduced from the multiplier equation by means of JACOBI'S theorem , ; viz. substituting for their values , and observing that if the modular equation be , then the value of is , this is ; and then in the multiplier equation separating the terms which contain the odd and even powers , and writing it in the form , this equation , substituting therein for its value , gives the value of rationally .
The rational expression of in terms of is of course indeterminate , since its form may be modified in any manner by means of the equation ; and in the expression obtained as above , the orders of the numerator and denominator are far too high .
A different form may be obtained as follows : for greater convenience I seek for the value not of but of 31 .
Denoting , as above , by , . . .
the valueswhich correspond to , . . .
respectively , and writing \amp ; c. , we have , \amp ; c. , all of them expressible as determinate functions of ; and we have moreover the theorem that each of these is a rational and integral function of : we have thus the series of equations , . . .
, where , . . .
are rational and integral functions of hese give linearly the different values of ; we have in fact where \amp ; c. denote the combinations formed with the roots , . . .
( these can be expressed in terms of the single root ) ; and we have also : the resulting equation is .
consequently , a determinate rational and integral function of ; but as the same formula exists for each root of the modular equation , we may herein write in place of ; and the formula thus is , viz. we thus obtain the required value of as a rational fraction , the denominator being the determinate function , and the numerator being , as is easy to see , a determinate function of the order as regards 32 .
The method is applicable when is only known by its expression in terms of ; but if we know for an expression in terms of , then the method transforms this into a standard form as above ; and by way of illustration I will consider the case where the modular equation is and where a known expression of is .
Here writing , , \amp ; c. to denote the sum of the powers - \amp ; c. of the roots of the equation , we have , as appears from the values presently given , ; and observing that being ultimately replaced by , we have that is we have , viz. this is .
But we have ; and the equation thus is ; OF ELLIPTIC FUNCTIONS .
to verify which observe that , substituting herein for its value , the equation becomes ; that is , , as it should do .
33 .
Any expression whatever of in terms of is in fact one of a system of four expressions ; viz. we may simultaneously change orintoor that isigns a Thus , starting from , we have ; and of course if from any two of these we eliminate , we have either an identity or the modular equation ; thus we have the modular equation under the six different : 34 .
Next .
Here , starting from , the changes give viz. the third and fourth forms agree with the first and second forms respectively ; that is , there are only two independent forms , and the elimination of from these gives which is a form of the modular equation .
35 .
In the case , starting from ( as to this see post , No. 43 ) , the forms are ( 2 ) .
, ( 3 ) ; .
( 4 ) so that here again the third and fourth forms are identical with the second and third forms respectively ; there are thus only two forms , and the elimination of gives which is a form of the modular equation .
36 .
If in the foregoing equation , , we make the change into , it becomes ; and combining these equations , we have ; or substituting herein the foregoing value this becomes or 5 , -for 1 or 7 , which must agree with the modular equation : thus in the last-mentioned case where we have or say and therefore ; the equation is which is right ; for JACOB.1 , p. 82 , the modular equation , gives .
OF ELLIPTIC FUNCTIONS .
Observe that the general equation no longer contains the functions , which enter into JACOBI'S expression of Theorem in connexion with the multiplication of Elliphc Functions .
Article Nos. 37 to 40 .
37 .
The theory of multiplication gives an important theorem in regard to transrmation .
Starting with the nthic transformation , we may form a like transformation , such that the combination of the two gives a multiplication , viz. for the relation between , deriving from as from , we have ; and instead of we have ; that is , we have and thence ; or writing , we have here , viz. it is for 3 or 7 , and or 5 .
Now in part effecting the substitution , we have where denotes its value in terms of And from the theory of elliptic functions , replacing sn , sn by their values we have an equation where are given functions each of the ordcl ; viz. the coefficients are given functions of , or , what is the same thing , of Comparing the two results , we see that in the nthic transformation the sought-fol function , of the order , is a factor of a given function of the order .
38 .
Considering the modular equation as known , then by what precedes we have ; that is , the given function has a factor , of which one ( the last ) coefficient is known , and we are hence able theoreticaJy to determine all the other coefficients rationally in terms of ; that is , the modular equation being known , we can theoretically complete the solution of the transformation problem .
I do not , however , see the way to obtaining a convenient solution in this manner .
39 .
The formula in question for is which , putting therein , and replacing by , may be written , where the signs indicate denominators which are obtained from the numerators by changing the signs of respectively .
The theorem in regard to , thus is , is a factor of ; viz. writing in the last-mentioned function , we ought to have that is which is in fact the modular equation .
40 .
And so for if , and for if , the formulae 1 1 ( ) ( ) 4 OF ELLIPTIC FUNCTIONS .
Term in has factor ; , term in is - Term in has factor ; , term in is The transformations , 6 , 7 , ll.\mdash ; Article Nos. 41 to 51 .
41 .
The cubic transformation , I reproduce the results already obtained ; since there are only two coefficients , ( 3 , these are also the last but one and last coefficient .
Hence , from the values of , ( 3 , , we have , ; the two values of are thus , giving the modular equation ; and we then have 42 .
The quintic transformation , Here there are the three coefficients , or are the last but one and last coefficients ; we have , Comparing the two values of , we have , and then so that only the modular equation remains to be determined .
The unused equation is , which , putting therein , may be written ; attending to the value of , this divides by ; in fact the equation may be written ; and then completing the substitution , and integralizing , this becomes , viz. this is ; and the term in being , the whole again divides by , and the equation thus becomes which is the modular equation .
43 .
The septic transformation , I do not propose to complete the solution directly from the fundamental equations for , but resort to the known modular equation , and to an expression of which I obtain by means thereof .
The modular equation is which may also be written as can be at once verified ; but it also follows from CAUCHY'S identity , ELLIPTIC We then have Moreover ; and similarly , whence Writing this under the form I find , as will appear , that the root must be taken with the sign - , and that we thus have , whence also 44 .
Recurring now to the fundamental equations for the septic transformation , the coefficients are , and we have , so that the coefficients are all given in terms of , M. The unused equations are , , which , substituting therein for the foregoing values , give two equations ; from these , eliminating , we should obtain the modular equation , and then in terms of Substituting in the first instance for their values , the equations are The first of these is viz. this is ; PROFESSOR CAYLEY ON TllE TRANSFORMATION or observing that in this equation the coefficient of is , the equation becomes 45 .
This should be satisfied identically by the foregoing value of ; .
it should be satisfied on writing therein ; that is , we should have where observe that the sign of the second term is the sign of the foregoing value of ; so that the identity being verified , it follows that the correct sign has been attributed to the value of 46 .
Multiplying by , the equation is viz. this is In the second column the coefficient of is , viz. this is , or it is Reducing also the other two columns by means of the modular equation , the equation thus becomes OF ELLIPTIC FUNCTIONS .
This is in fact an identity ; to show it , writing for convenience in place of , and observing that the terms al.e , the whole equation divides by ; or throwing out this factor , it is The first line is ; whence , throwing out the factor , the equation is that is ; or throwing out the factor the equation is which is an identity .
The other equation is ; that is which also be verified , but I have not done this .
47 .
The conclusion is where and of course ; but the resulting form may admit of simplification .
48 .
The endecadic transformation , I have not completed the solution , but the results , so far as I have obtained them .
, are interesting .
The coefficients are ; and we have , as in general , , The unused equations then are , , , ; but I attend only to the first and last , which , it will be observed , contain 7 , linearly .
If in the first instance we substitute only for their values , the equations become ; say , for a moment , these are giving 1 : 27 : Here where the terms containing disappear of themselves , viz. this is ( observe that the term in , equated to zero , gives the modular equation for the case .
It thus appears that 7 and are given as fractions , having in their denominator this function 49 .
To complete the calculation , we have ; viz. multiplying by 8 , and substituting for their values , this is 8(QB\mdash ; AS ) , OF ELLIPTIC FUNCTIONS . .
what is the same thing , - ( QB\mdash ; AS ) , viz. the left-hand side is ; or say we have - , where ; the value of is .
Similarly , writing we find ; in verification whereof observe that this being so , the first equation gives the identity 50 .
The result is that , writing for the moment , the values of the coefficients are and ; odular equation iownsion ferms o ' and to complete the solution we require only an expres51 .
We may herein illustrate the following theorem , viz. we may simultaneously change into Thus in the equation , making the change , we have ; that is , , which is light .
So in the equation , if for a moment ( ) , are what become , the equation is , that is , , or ; but obviously ; and the equation thus is , or say ; that is .
which is right .
The general theory by -transcendents.\mdash ; Articles Nos. 52 to 71 .
52 .
I recur to the formula and seek to express the ratios : in terms of .
Writing with JACOBI we have in general ( snc co am ; viz. snc 2 ) ) \amp ; c and the values of , . . .
which correspond to the moduli , . . .
, or say the values , , . . .
, are obtained by giving to the values ' , . . .
OF ELLIPTIC FUNCTIONS .
437 viz. the cases correspond to JACOBI'S first and second real transformations , and the others to the imaginary transformations .
I remark that gives for snc 2 an expression which is rational as regards but an expression involving , the real th root of ; the other values , , . . .
, give the like expressions , involving , . . .
( an inary n loot of unity ) , the imaginary roots of 53 .
I consider first the expression Here , writing 2 being used different sense ) , that is thence imaginary root unity ) , we have ( JACOBI , p. 86 ) ; that is where , for shortness , I write to denote the infinite product and similarly to denote the infinite product and the like for the terms in the notation , accompanied by its explanation , is quite intelligible , and it would be difficult to make one which would be at the same time complete and not cumbrous ; and then attributing to the values 1 , , and forming the symmetric functions of these expressions , we have the values of , \amp ; c. , or being put , say the values of ( 3 , , . . .
54 .
I stop to notice a verification afforded by the yalue of .
Putting ?
, that is , we and thence I[DCCCLXXIV .
3 we have ; and puttin ON THE TRANSFORMATION is } as above , the value of is ; whence a theorem relating to the imaginary roots of unity , an odd prime .
In particular , at once verified by , velified by ( viz. the theorem is also true for the real root ) ; in fact the term in is , that is ) ; which may be verified by means of ; and so on . .
I further remark that we have But JACOBI ( p. 86 ) , sn 2 ' where ( p. 89 ) , that is , Hence ; and giving to the values 1 , 2 , . . .
, and multiplying the several essions , we have the value of .
this is , where denotes the product of the several factors which contain 56 .
The of the denominator gives a factor , which destroys the factor .
We have then a factor , which is ELIJPTIC FUNCTIONS .
In fact , , this is viz. the numerator is , and the denominator is So , the , formula is , that is ; or viz. this is , which is right ; and so in other cases .
We thus have , which , on putting therein , that is , gives , as it should do , 57 .
As regards the expression of , observe that giving to its values , the factors and all the factors other than of , and so as to the other pairs of factors ; viz. we have viz. this is that is agreeing Nith a former result .
58 .
We have of course the identity ; that is , which , putting therein , is an identity before refel.red to ; form perhaps more convenient is obtained by dividing each side by 59 .
I notice further that we have snc snc ; the term in is where we have .
For example , , the term is ; , it is * ; , it is , S$ and so on .
The term in question thus is that is .
This has to be multiplied by , and we thus obtain , rreeing with a former result .
We have in what precedes a complete -transcendental solution for the transfonnatio ; viz. the modulus being given as a function of , then , as well the new modulus and the multiplier , as also the several functions which enter into the expression are all of them expressed as functions of 60 .
I consider in like manner the expression Here , writing 2 ( instead of JACOBI'S as before ) , that is and thence we hare ' OF ELLIPTIC FUNCTIONS .
where the notations are as follows : . .
is the infinite . .
, and .
the infinite product . .
; and the like as to the expressions with exponents containing\mdash ; And then attributing to the values 1 , 2 , . .
, and forming the symmetric functions of these expressions , we have the values of , . .
; or being put the values of , . . .
It is easy to see , and I do not stop to prove , that if instead of we have or , we simply multiply by an imaginary root of unity ; that is , replace the real nth root by an imaginary root of In the case , that is , we have , and thence the like the values , . . .
: the equation ives consequently for values each , agreeing with the multiplier equation .
61 .
We have for the formula and , as before , sn 2 .
; hence and we thence derive the value of ; viz. observing that we have in the denominator which destroys this factor in the expression of , this is Now , giving to its values , it is easy to see that we have where . .
denotes . .
, .
it is the same function of that .
is of also . . . .
where denotes . .
, it is the same function of that is of and the like as to the denominator factors : we thus have .
this is or we have , agleeing with a former result .
We have , that is a result which , substituting on the left-hand side the foregoing values of the several functions , must be identically true .
62 .
We have also snc snc where the term in is ; or observing that the sum of the exponents is , this is ; or , the last factor being , the expression is ; or , multiplying by , we have agreeing with a former result .
OF ELLIPTIC FUNCTIONS .
We have in what precedes the complete -transcendental solution for the transformatio secunda ; viz. the original modulus being given as a function of , then , as well the new modulus and the multiplier , as also the several functions which entel into the formula all expressed in terms of .
The expressions all contain , and by substituting for this an imaginary root of , we have the formulae belonging to the several imaginary transformations .
63 .
As an illustration of the formulae for the transformatio I write ; and putting for greater convenience , that is , then we have ' where ' ' ' rhele the numerator of A denotes . . . .
, and so in other cases , the difference of the exponents being always .
And we have , as mentioned , the identical equation The values of the several expressions up to are as follow.s .
Mr. J. W. L. GLAISIIER kindly performed for me the greater part of the calculation .
ind .
A Sum .
Multiplied by ofr . . .
C4 .
As already mentioned , the regoing expressions of the coefficients in terms of may be applied to the determination of the coefficients as rational functions of Representing by any one of the coefficients . .
, consider the sum OF ELLIPTIC FUNCTIONS .
a positive integer , and the summation extending as to the values of and corl.esponding values of .
This is a rational function of , and it is also integral .
As to this observe that the function , if not integral , must become infinite either for ( this would mean that the expression contained a term or terms ) or for some finite value of .
But the function can only become infinite by reason of some term or terms of becoming infinite ; viz. some must become infinite ; or attending to the equation snc it can only happen if , or if ; and from the modular equation it appears that if , then also : the expression in question can therefore only become infinite if , or if .
Now gives the ratios , . . .
, each of them a determinate function of , that is finite ; and gives also , so that the explession does ot become infinite for ; hence it does not become infinite either for or for any finite value of ; wherefore it is integral .
The like reasoning applies to the sum ; viz. this is a rational function of ; and it is quasi-integral , .
there are no terms having a denominator other than a power of , the highest denominator being ; viz. the expression contains negative and positive integer powers of , the lowest power ( highest negative power ) being 65 .
It is to be observed , further , that writing the expression in the form where refers values modulus ) , and considering the several quantities as expressed in terms of , then in the sum every term involving a fractional power acquires by the summation the coefficient , and therefore disappears ; there remains only the radicality occurring in the expressions of the ; and if , , or a positive integer less than 8 , then the form of the expression is into a rational function of .
Hence this , being a rational and integral function of , must be of the form 66 .
We have thus in general ; and in like manner DCCCLXXIV .
* We may in these expressions find alimit to the number of terms , by means of thebefore mentioned theorem that we may simultaneously interchange , 9 , into ; ( , . . .
( 3 , .
Starting from the expression of , let be the corresponding coefflcient to ; viz. in the series . .
, let be as removed from as is from ; then the equation becomes where the equation thus is .
; and by what precedes the series on the right-hand side can contain no negative power higher than ; that is , the series of coefficients , C. . .
goes on to a certain point only , the subsequent coefficients all of them vanishing .
In like manner from the equation for we have where the indices must be positive ; viz. the series of coefficients , . .
goes on to a certain point only , the subsequent coefficients all of them vanishing .
67 .
The like theory applies to the expression .
We have , putting as before , , and we find a limit to the number of terms by the consideration that we may simultaneously change into ; the equations thus become where , if , there must be on the right-hand side no negative power of ; but if , then the highest .
negative power must be , and , where on the l.ight-hand side there must be no negative power of OF ELLIPTIC FUNCTIONS .
68 .
It is to be remarked that being always given linearly in terms of , it is the same thing whether we seek in this manner for the values of or for that of ; but the latter course is practically more convenient .
Thus in the cases we require only the value of In the case , whele the coefficients are , it has been seen that 7 , are given as cubic functions of : seeking for them directly their values would ( if the process be practicable ) be obtained in a better form , viz. instead of the denominator there would be only the denominator .
69 .
I consider for the cases and 5 : , 1 , 2 , , then , 3 , 6 , 1 ; and we write down the equations giving ; viz. if we had in the first instance assumed . .
, this would have given . .
, whence and the succeeding coefficients all vanish ; and so in other cases .
have heal only the coefficients ; and these can be obtained without the aid of the formulae by the consideration that for the values of whence ; or we have the equations giving as before reducible by means of the modular equation to 70 . .
Corresponding to , 1 , 2 , 3 , 4 , 5 , we have , 5 , 2 , 7 , 4 , and we find PROFESSOR CAYLEY ON THE TRANSFORMATION giving , But for the corresponding values of are ; whence , or say the value of is .
The value of is found very easily by the -formulae , viz. neglecting higher powers of , we have ; hence ; that is , and the equations are ; whence , where , \amp ; c. are the coefficients of the equation viz. are , ; or the equation is OF ELLIPTIC FUNCTIONS .
or say where Hence also , reducing by the modular equation , the one of which forms is as convenient as the other .
71 .
Making the change into , we have - ; and comparing with the equation we obtain Writing for a moment , this is that is But we have Hence , replacing by their values , this is ; viz. writing , this is that is ; ; sented as an equation of the same order when prime ) between or between , or between , or between .
As to this , remark that in general an equation of the order gives rise to an equation of the ordel between ; viz. the equation is whele the left-hand side is a rational function of of the form ; or starting from a given equation , and transforming by the equations , the curve in is of the order ; in fact the intersections of the culve by the arbitrary line are given by the equations , and the number of them is thus .
Moreover , by the general theory of rational transformation , the new curve of the order has the same deficiency as the original curve of the order .
The transformed curve in ' may in particular cases reduce itself to a curve of the order twice repeated ; but it is important to observe that here , taking the single curve of the order as the transformed curve , this has no longer the same deficiency as the original curve ; and in particular the curves represented by the modular equation in its four several forms , writing therein successively , are not curves of the same deficiency .
73 .
The question may be looked at as follows : the quantities which enter rationally into the elliptic-function formulae are ; if a modular equation led to the transformed equation , then to a given value of would corre- spond 8 values of , therefore values of , giving the same number , , values of ; that is , the values of corresponding to a given value of would group themselves in cights correspondiLg to the 8 values of .
There is in fact no such grouping ; the equations are ; to a given value of correspond 8 values of and therefore values of , but these give in eights the same value of , so that the number of values of is 74 .
I consider the case : here , writing for , we have here the sextic curve I. ; and it is easy to see that the remaining forms wherein denote ; and respectively , are derived herefrom as follows ; viz. II .
, that is ; OF ELLIPTIC FUNCTIONS .
III .
, that is ; IV .
, that is where it may be noticed that the process is not again repeatable so as to obtain a sextic equation between standing for respectively .
The curve I. has a dp ( fleflecnode ) at the origin , viz. the branches are givcn by ; and it has 2 cusps at infinity , on the axes tively ; viz. the infinite branches are given by respectively .
These same singularities present themselves in the other curves .
The curve II .
has the four dps , that is .
Corresponding hereto we have in the culve III .
the 2 dps , and in the curve IV .
the dp The culve III .
has besides the 4 dps , that is , , , ; and corresponding hereto in the curve .
we have the 2 dps The curve .
has besides the 4 dps , or say .
Hence the 4 curves have respectively the dps and deficiency following:\mdash ; dps .
Def. ; viz. the curve .
representing the equation between and is a unicursal sextic .
It may be noticed that except fleflecnode at the origin , and the cusps at infinity , the dps in question are all acnodes ( conjugate points ) . .
The foregoing equations may be exhibited in the square diagrams:\mdash ; I. II .
These values give and the equation of the curve is thus verified .
76 .
Considering in like manner the modular equation for the quintic transformation , we derive the four forms as follows:\mdash ; I. II .
, that is III .
, that is : or expanding the two terms separately , this is OF ELLIPTIC FUNCTIONS .
77 .
The square diagrams I. II . .
PROFESSOR CAYLEY ON THB TBANSFORMATION III .
: IV .
where the subscript line , showing in each case what the equation becomes on writing therein , serves as a verification of the numerical values .
78 .
The curve I. has at the origin a dp in the nature of a fleflecnode , viz. the two branches are given by respectively ; and there are two singular points at infinity on the two axes respectively , viz. the infinite branches are given by respectively .
Writing the first of these in the form , we see that the point at infinity on the axis .
the point is ; and similarly writing for the other branch , the point at infinity on the axis point is Moreover , as remarked to me by Professor H. J. S. SMITH , the curve has 8 other ; *These results follow from the general formulae in the paper ' On the JJigher Singularities of Plane Curves C. \amp ; D. XI .
J. .
vu .
( 1865 ) pp. 212-222 ; but they are at once seen to be true from the consideration that the curve , which has only the singularity in question , is unicursal ; the singularity is thus OF ELLIPTIC FUNCTIONS .
viz. writing to denote an eighth root of - , then a is ) , .
To verify this observe that these values give or the derived functions each vanish .
Thus I. bas in In II .
we have in like manner l ; viz. instead of the 8 , we have the 4 dps what same thing , , , where .
But we have besides the 12 dps given by viz. we have in all In .
we thence have ; and , besides , the 12 dps given by in all And in .
we thence have ; and , besides , the 12 dps given by ( these curves intersect in 16 points , 4 of them at infinity , in pairs on the lines respectively ; and the intersections at infinity being excluded , there remain 16-4 , intersections ) ; there are thus in Or arranging the results in a tabular form and adding the values of the deficiency , we have I. ldps , II .
29 , 7 , III .
33 , 3 , IV .
35 , 1 , so that the curve .
is a curve of deficiency 1 , or bicursal curve .
It appears by JACOBI'S investigation for the quintic transformation ( Fund .
Nov. pp. 26-28 ) that we can in fact express , that is , rationally in terms of the parameters connected by the equation , which is that of a general cubic ( deficiency ) ; we in fact have 79 .
In the general case an odd prime , then in I. we have at the origin a dp ( in the nature of a fleflecnode ) and at infinity 2 singular points each .
I infer , from a result obtained by Professor SMITH , that there are besides ; but I have not investigated the nature of these .
And the Table of dps and deficiency then is I. II .
III .
, 1 IV .
viz. his values of the deficiencies being as in { must be as in the last but one column .

