]\gt ; 461 XII .
Ellipsoidal Harmonic Analysis . .
H. DARWIN , F.R.S. , Professor of rinity C , in the UnirersitrJ of Cambriclgc .
, \mdash ; Read May 2 , 1901 .
liolt . . . . .
TION O THE FUNCTIONS .
S 1 .
The inciples of ellipsoidal harmonic analysis . . . . .
463 S .
limits of so to represent all ellipsoick . . . . .
463 S 3 .
The differential equations . . . . . . . . .
466 S 4 .
The forms of the functions . . . . .
4 S 5 .
ration for determination of the functions . . . . .
47 S 6 .
Determination of the coefficients in the fumctions . . . . .
477 S 7 .
igorous determination of the functions of the second degroo . . . . .
488 S 8 .
Approximate form of the fnnctions . . . . .
489 S 9 .
Factors of transformatio1t between the forms of -function and of C-or ction 492 S10 .
The functions of the second kiod . . . . . . . . . .
497 PAltT II .
LLLIISOIDL Yblb . .
The of rmouic d } of ellipsoid . . . . .
) S .
The of homogeneous solid ellipsoid . . . . .
ration for the of the of surface rmonic over the . . .
the . . . . . . . . . . . . . .
S15 .
Integration in the case of . . . . . . . . . . . .
S16 , Portion of the integration in the case of . . . . .
S17 .
Portion of the tion in the of . . .
, .
) S18 .
Prcl)tion for the tegrations . . . . .
uation of the integrals . . . . .
533 S20 .
ction of preceding ; disappearance of logarithmic terms . . . . .
S21 .
Integrals of the squares of harmonics when and . . . . .
S22 .
of of squares of harmonics . . . . .
17 III . . . . . .
ondix .
of functiunb ( 298 ) . . .
7.12 .
1901 PROFESSOR G. H. ) ON } HARMONIC ANALYSIS .
INTRODUCTION .
unctions o ellipsoidal hal.monics have been successfully used in many tions , but ) have been presented has ) been such to ical cculation so difficult as to be impossible .
The object of investigation is to this fection in the method .
1 believe tha I now reduced hese functions to such a form that numerical results ) ] of the CitSO the ithmetic will llly } ] ellipsoidal rmonics I have enjoyed the immellse of uent deussionS with .
E. W. He me eely roJll g of knowledge , I did , in almost complete of the subject , I could ought l attempt to a successful issue without his advice .
In help derived from hinl has been of innmense value , even it is not possible to indicate a specific point as due to him .
In other cases he ; put me in the way of giving succinct proofs of propositions which I had only proved by clumsy and tedious methods , or where I mel.ely felt sure of the of a esult without igorous P. In particular , I should have been unable to out the investigation of , unless he had shown me how the , were to be determined .
My original object in attacking this problem was the hope of being thereby enabled to obtain exact numerical results with respect to M. POINCARE'S pear-shaped figure of of a mass of liquid in But I soon found that a partial inyes- tigation with one particular poin view acticable , and I was led on little by little to covel the whole tield , in it was to do so for the of plication .
This then to such considerable it seemed best it should stand by itself , and that the discussion of the specific should be of this kind hardly vell by the thenatician , unless happens be subject .
It to be useful to present nnlaly , which endei .
it possible the mathematical reader to uIlderstand the lature of the thod iesults , ving ) to pick it out from a coulplex of analysis .
Such a is given in III . .
THL 1 .
The lciples of of this method of analysis is ) ouuded in ious works the subject .
with a tement of results in * A paper giving the required rebult will ) ) lesented to the Society in the autumn .
, 1901 .
] PROFESSOR G. H. DARWIN ON LLIPSOIDAL H If denote the three roots of the cubic it and , may ) written down by nges .
If for breyity write , equation becolnes The solution is where functions of ?
ectively , and satisfy and two other equations with suffixes 2 and 3 , the , a constant , and the same a positive integer .
If are in nscending order of magnitude we may suppose to between and between , and .
and If denote the ) arcs formed the intersections of ) three orthogonal quadrics , and two other equations found by cyclical ) of suffixes .
S2 .
; limits of so to all I now change the notation , let the three roots be defined thus : ' where rano e from to between , i ) between Let the axes of the fundamental ellipsoid of reference be ' The ellipsoid definerl ) its three axes given by .
This mode of definin the axis is such as to indicate the relationship to the prolate ellipsoid .
But another hypothesis may be made which will bring the axes into relationship with those of the oblate ellipsoid ; for if we take a new , numerically equal to the old one but imaginary , and replace by we have .
If be made to range from to , all possible ellipsoids are comprised in either of these types .
It will , however , now be shown that , by a proper choice of type , all ellipsoids may be included with the range of from to Let us suppose the axes to be expressed in three forms , as follows:\mdash ; ( 1 .
) ( 2 .
) ( 3 .
) , .
Then we have Therefore And G. H. ON Now let increase from to As passes from to is appropriate .
As passes from to 1 , decreases from to , so that form ( 2 ) is appropriate .
Lastly , as passes from 1 to increases from to , so that form ( 3 ) is .appropriate .
But we might well have Wl.itten forms ( 1 ) and ( 3 ) so as to involve , and form ( 2 ) so as to involve , and it that all possible ellipsoids comprised in the range of from to , proyided that the ) be } ) riately chosen .
The developments in this ) aper are made in powers of .
It will , therefore , be well to show that there is a class of ellipsoids , analogous to ellipsoids of revolution , which might form the basis of developments similar to those carried out below .
Ellipsoids of revolution are defined by the condition In the class to which I refer .
Ellipsoids of this kind are ) riven b : .
in , Tlley also given by , \mdash ; then Hence if we only allow to range to corresponds with soids of revolution , to which spheroidal harmonic analysis is applicable ; and corresponds with this new class for which the corresponding analysis has not yet been worked out .
We shall see below that the solid harmonic for this case where will be of the , where and satisfy the I am not clear whether or not it would be advisable to proceed ab these equations , but at any rate I shall show hereafter the and -functions may be determined from the vsis of the present paper with an .
of accnracy desirable .
If it were proposed to use the functions corresponding to as a basis for the development of general ellipsoidal harmonics , we ] have to assume VOL. CXCVII.\mdash ; .1 .
3 PROFESSOR .
H. DARWIN LLIPSOIDAL HARMONIC ANALYSIS .
or eIse , The developments would then proceed by powers of In order to discover what is the greatest value of which must be used so as to comprise all ellipsoids , when we proceed from both bases of development , a comparison be made between this assumption the previous one .
Suppose in fact that \mdash ; \mdash ; 1 ; .
Then and therefore , or When and are both equally great , they must each equal the positive root of .
This root is or 236 .
Thus the greatest values will be In this case very nearly , whereas when .
Thus if the developments were to stop with we should double the accuracy of the result .
However , I do not at present propose to carry out the process suggested .
S 3 .
The Differential We now put ; ; ' and find from the formulae of S1 , \mdash ; 1 \mdash ; 1 , ( 1 ) .
PROFESSOR G. H. DARWIN ON IPSOIDAL HIONIC ANALYSIS .
It will be observed that is independent of , and that it has the same form as in spheroidal harmonic analysis when vanishes .
Since is less than 1 and greater than and real .
In all the ulier 1tion of this papel I always write thd 1 as to maintain perfect synunetry with respect to and now have , \mdash ; 1 , ; Let us denote the differential operators involved in our equations , thus : Then ) ) ' ( 2 ) .
Hence our differential equations , a similar equation with suffix 2 , and Let us replace by another constant such that so that \mdash ; G. H. DARWIN ON ELLIPSOIDAL HARMONIC ANALYSIS .
formula.s is a constant a con stant to be ed. Our equations are now similar for ( 3 ) .
And equation is ( 4 ) .
is equal to differential operatol in ( 4 ) , divided by It is well known that in spheroidal harmonic analysis there are two kinds of functions of and which satisfy the differential equation , and they are usually denoted .
The -functions of the variable have no significance , so that virtually there are and -functions of , but only -functions of .
The like is true in the present case , however , with the additional complication that each of the functions assume one of two alternative forms .
I adopt a parallel notation and write for and either , or , as the case may be .
Since and nter in the first two equations in exactly the same way , we need only consider one of them , and we may usually rite simply ( for example ) where the full notation yould be ?
or .
In the part of the investigation I shall only refer to the -functions , and the -functions will be considered later .
In spheroidal harmonic analysis the third function is cosine or sine of .
So here also we find functions of two kinds associated with cosines and sines , which shall denote , the variable being understood .
Throughout the greater part of this paper the functions will be of denoted by , and it seems useless to print the subscript hundreds of times .
I shall accordingly drop the subscript except where it shall be necessary or advisable to retain it ; for example , will be the abridged notation for .
The operators involved in the differential equations ( 3 ) will occur so frequently that an abridged notation seenoe justifiable .
I therefore write where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; where \mdash ; .
PROFESSOR G. H. ON ELLIPSOIDAL The equations are ( 5 ) .
or or It is well known that the function is a linear function of of degree made up in one of the following ways : .
When is even , linear function of ?
of degree 2 , 3 , 4 .
When is odd , linear of of gree , ltiplied by , or , or , 7 .
When is even , linear function ) of , ultiplied by , or , or 8 .
Whell is odd , a linear function of of degree , multiplied ) These eight classes may be conveniently spe.cified by initials O , AB , ABC , but it is better to them according as they are associated with the evenness or oddness of and , and with the cosine or sine functions .
This llew grouping may be defined by a shorthand notation involving the initials and or , which shall denote successively the evenness or oddness of and , clnd cosine or sine .
We shall see this rangement is as follows :\mdash ; or EEC ; even , even , cosine .
AB or even , even , sine .
A or OOC ; odd , odd , cosine .
or OOS ; odd , odd , sine .
, or OEC ; odd , even , cosine .
ABC or OES ; odd , even , sine .
CA or even , odd , cosinle .
or EOS ; odd , sine .
Since several functionls in they in the no tatio fnnctions of or , or of and Hence ?
and involve linear functions of of various degree .
multiplied by various factors ; and the same is true of the functions of In the case of the third root the linear function of powers of may be replaced by a series of cosines of even multiples of .
Further , in forming the @ , may as being , and since this only amounts to dropping constant factors which may be deemed to be included in the , as yet , undetermined coefficients of the several series .
I will now consider in detail the forms of the several -functions of ( those following by symmetry ) , and the same time indicate more precisely the nature of the notation adopted .
In the following series , indicated by 2 , the variable is supposed to proceed from the lower to the upper limit by 2 a time .
The reader will be able to perceive the manner of the formation of the functions when he bears in mind that \mdash ; Type EEC ; ?
AB or EES ; \mdash ; Type A OOC ; \mdash ; Type or OOS ; or OEC ; Type ABC or OES ; Type CA or EOC ; ) CB or ; Observe that is always associated with \mdash ; , and that , each form being epeated twice , there are two forms of function of each kind .
Moreover , a cosine a sine function are always associated with different kinds .
It is obvious that the -functions are expressible in terms of the ordinary -functions of spherical harmonic analysis , and that if we take out the factor -functions similarly expressible .
This factor will occur so frequently that I write ' and as cummonly put to denote .
PROFESSOR .
H. DARWIN ON ]LLIPSOII ) HA IONIC 471 .
We assume then the following forms for the functiolfs:\mdash ; For EEC , OEC , OOS , EOS For EES , OES , OOC , ( 6 ) .
In these series proceeds by intervals of one at a time , beginning from a lowel limit of unity .
In both forms the upper limit of the first is or according as is even or odd ; and the upper limit of the second is or ) according as and agree or do not agree in evenness or in oddness .
The factor contains in the denominator , but does not become infinite when , because when is not zero is divisible by and we shall see that is zero.ae When is zero there is no function of the type .
It may be noted that the limits of the series are such that neither nor can ever have a negative suffix .
We shall ultimately make and equal to unity , and this will be justifiable because there must be one arbitrary constant .
We have now to consider the forms of the cosine and sine functions .
They may be derived al once from the ceding results , for we have only to read as where is even ; as as , and as The factor will occur frequently , and I write and as elsewhere I commonly write to denote .
The following are the results : \mdash ; Type or ; Type AB or EES ; is clear that we may equally well regard the lower linlit in the latter as zero .
Type , or OOC ; each term is of type or Hence *This also follows from the fact that the series begin with ) in the case of EES , and with ] in the case of OES .
Thus in the former case there is no term in the latter no term PROFESSOR G. H. DARWIN ) ) tNALYSIS .
Type , or OOS ; since we now have ; rType ABC , or OES ; each is of ) , whicI ] .
Hence It is clear that we may equally well regard the lower limmit as unity .
rType CA , or ; each term is of type .
Hence each type ) .
Hence When ao to enness oddness have the forms independent of they differ in this respect the factor occurs .
Therefore ( in alternative form ) for EEC , , OOC , OOS for , OES , , EOS ( 7 ) .
In these series proceeds by intervals of one at a time , beginning with unity .
In both forms the upper of the first is or according as is even or odd .
In the first form the upper limit of the second is , and in the second it is .
We shall ultimately put and , which may be regarded as constants equal to unity . .
Preparation for of tl , In order to detel.mine the coefficients and , we have to substitute these assumed forms in the differential equations .
Where the functions .
involve and as factors , already given for the PROFESSOR G. H. DARWIN ELLIPSOIDAL HARMONIC YSIS .
differential equations are perhaps the most convenient , but in the other cases a reduction seems desirable .
By considering the forms of and in ( 3 ) it is easy to show that ( 8 ) , ( 9 ) .
By making vanish we reduce these operators to the forms appropriate to spheroidal harmonic analysis .
By making inite we obtain the differential equations specified in S2 as appropriate to ellipsoids of the class .
It is now ecessary to perform the operation on typical terlns P and , and on typical terms and To find .
The form ( 8 ) for is ] ) It is clear that \mdash ; 1 1 because is the solution of the differential equation f by the term from each side .
Again we have from the same differential equation It may be noted in passing that this is equally when ) of ation is , the function of the other form .
Therefore Hence V0L .
XCVIT .
PROFESSOR G. H. ) ] ON ELLIPSOIDAL HARMONIC ANALYSIS .
We have to eliminate It is and P. The differential equation satisfied by involves in the form HenCe can only from by a constant factor .
In order to find that factor suppose to be infinitely large ; then and Also the factor is ' .
It will be convenient to pause here and obtain the corresponding formulae for the -fimctions .
Various writers have adopted various conventions to the factors involved in these functions .
I write , Q. As the of we may change the sign of , if we introduce a constant fictor , be found by making infinitely reat .
Tn that case it is easy to show that By performing and ) on it follows that the constant factor is the same as before , and that the alternative for are exactly the same as for Hence the transformations which follow for the -functions are equally the -functions .
PROFESSOR .
H. DARWIN ON ELLIPSOIDAL ANALYSIS .
If we differentiate in its two forms we find And \mdash ; 1 \mdash ; 1 1 now write \mdash ; 1 .
It is clear that , and , 1 .
Now since } , by taking the sum and difference of the two forms of , we have ( 10 ) .
It is easy to verify , by means of the relationship ' that these ' equations are true when is negative .
They are also true when , although the second equation then becomes nugatory .
Multiply the first of ( 10 ) by and the second by , and apply them a second time .
Then since \mdash ; \mdash ; , ; .
These equations are always true although for and they become nugatory . .
Then -fence .
( 12 ) find It is to use in the , where is defined by ( 2 ) .
The latter terms of contribute Therefore But , and we find on reduction that stit uting f and their values , have .
( 13 ) .
To find .
In this case the most convenient form for is that in ( 9 ) , and we ea , sily.find PROFESSOR G. H. DARWIN ELLIPSOIDAL ONIC A ( 14 ) .
S. find 1 now use the form as defined in ( 5 ) , here l , is iven in ( 2 ) , so that We have and The latter terms of } ibute , erefore ( 15 ) .
S In this section I use successively the four results ( 12 ) ( 13 ) ( 14 ) ( 15 ) obtained in the last section under the ( ) , S .
The limits of the first are 1 to ) , and ) the second 1 to or .
Applying ) operation to and equating - ) to zero , have G. H. DARWIN ON LLIPSOIDAL HARMONIC ANALYSIS .
The coefficients of ) 's must vanish separately .
This gives from the coefficients of and the following :\mdash ; These equations may be written in the form ' ( 16 ) .
Whence by continued application , the continued fractions .
- We must now consider what I may call the middle of the series , which corresponds with ? .
In this case each of the 's contributes one term and the term gives .
The result is or .
Since and are expressible as continued fractions , we have an equation for , if the continued fractions terminate .
PROFESSOR .
H. ) RWlN O ELLIPSOIDAL tNALYSIS .
We consider those ) tions .
First , supl ) is even , The continued ) ends only on the ) tisfied is Since , \mdash ; 1 , we have , by to zero the coefficients of and , results which may be written Hence the 's disappear from the first continued fraction , which terminates with In this last term the ) which prevails elsewhel.e is eplaced by Observe that when 2 the fraction is replaced by fraction , so that the equation for Secondly , suppose that is odd , to the types OOS , EOS .
The condition to be satisfied is now ( , - ) O. Now , and equate ero the coefficients of and esults written Thus the again disappear , and the continued fraction ends with .
Observe that when , the continued fraction reduces to a simple fraction , and tlJe equation for becomes The case of must be considered separately .
We have next to consider the termination of the second fraction , which depends only on the second First , when and are either both even both odd , the types are EEC and OOS , and the limits are to 1 .
The condition to be satisfied is Now is zero , and equating the coefficients of and to zero we results which may be written Hence this continued fraction ends with Secondly , when i to evenness or oddness , the types OEC and EOS , and the limits are to 1 .
The same ation applies again is changed into Hence the continued fraction ends with be by themselyes . .
H. OA7 ELLIPSOIDAL HARMONIC ANALySIS .
When , the types are EEC .
The " " middle\ldquo ; of the series is now cqlso an end , and the condition is \mdash ; . . .
, 3 Writing for and the coefficients of and to zero , we have Therefore ) ) ending with } for EEC , and with for OEC .
Next when the types are OOS , EOS ; the " " middle\ldquo ; is again an end , and the condition is \mdash ; . . .
Writing for , 1 and equating to zero ths coefficients of and , we have 4 .
1 .
, 5 , 4 Therefore ending with for OOS , and with for We have next to consider the other form of -function for types EES , OES , OOC , EOC , namely , , where VOL. CXCVII.\mdash ; A. 3 1ROFESSOR G. H. ) ON IONIC ANALYSIS .
Let us write .
The 's are not now the actual coefficients of any -fimction , but we shall see that they are determinable by almost the same relationships as those already found , and therefore the notation is convenient .
We now have ' Applying the operation to and equating to zero , we have This is the same equation as before , if we replace by .
As we may equate coefficients of to zero ( instead of coefficients of ) , we obtain the same equations for the 's as before .
A certain change must , however , be noted with respect to the beginning of the first series , which determines the end of the first continued fraction .
We previously for and for , 1 But the esponding terms will now be and , 1 , and these are equal to and .
Hence it follows that when is even The term has disappeared from the latter of these , and thus the continued action is independent of .
This is correct , since whatever value ( short of infinity ) may have , being equal to , vanishes .
Hence the continued fraction is docked of one term and ends with It is important to note the deficiency of one term in the fraction , it indicates that when the first continued fraction entirely disappears .
When there is no function of the form , so the question of interpretation does not arise .
PROFESSOR G. H. DARWIN ON ELLIPSOIDAL HARMONIC YSJb .
When is odd ( OOC , EOC ) the only change is that enters with the opposite sign , so that the first fraction ends with .
When , we have equal to the same fraction as before .
When the 's are determined we have .
But it is desired that in the case should be unity , and that in the case should be unity .
This condition will be satisfied in the present case if we determine all the , put equal to unity , and finally take Thus in both and we put equal to unity , and in determine the by the above equation .
We now have to consider the cosine and sine functions .
For EEC , EES , OOC , OOS The first 2 has limits or to 1 , the second to 1 .
Apply the operation and equate \mdash ; to zero ; then .3 G. H. DARWIN ON ELLIPSOIDAL HARMONIC ANALySIS .
If we equate to zero the coefficients of , we find These as before , lead to continued fractions , and by elimination of the 's to an equation for .
The equation will agree with our former result , for it can of course make no difference from which equation we determine It follows then by comparison with the previous result ( 16 ) that Hence when the are found , the 's follow at once .
S. For OEC , OES , EOC , where The limits of the first are or to 1 , of the second to 1 .
Proceeding exactly as before we find .
By comparison with ( 16 ) we see that ' Therefore when the 's are found , the 's follow at cnce .
* I have of course verifled this is so .
PROFESSOR G. H. DARWIN ON ELLIPSOIDAL HARISIONIC ANAL We may now summarise our results , as follows:\mdash ; In the general case where is neither nor 1 , is the root which nearly vanishes of the equation 4 .
2 \mdash ; The continued fractions terminate va ] iously for the various types of function , The end of the first continued fraction is as follows : \mdash ; For EEC and when this is the whole fraction .
For EES and when the fraction disappears .
For OOC , : and when this is the whole fraction .
For OOS ; and when this is the whole fraction .
For OEC ; and when this is the whole fraction .
For OES ; and when the fraction } ) ears .
For EOC ; and when this is the whole fraction .
For EOS ; and when this is the whole fraction .
For the first four of these types , , OOC , OOS , the second continued fraction ends with ; and when this is the whole fraction , but with the sign changed .
For the last four , viz. , OEC , OES , EOC , , it ends with ; and when this is the whole fraction , but with the sign changed .
When .
, the equation becomes ending when is even with ; when is odd ( OEC ) with When the equation has two forms , which may , however , be written together .
If the upper sign refers to cosines ( OOC , EOC ) and the lower to sines ( OOS , EOS ) , the equations ending when is even ( EOC , EOS ) with ; and when odd ( OOC , OOS ) with It might appear at first sight that a difficulty will arise in the interpretation of these results when is small , for the numbers in the denominators of the fractions increase , and yet it is possible that the number at the end should be smaller than that at the beginning ; thus apparently the fraction ends before it begins .
But this difficulty does not really arise , because in such cases the numerator will always be found to vanish , and thus the whole fraction disappears .
For example , in the last case specified , if the denominators , according to the formula , begin with and end with ; but the fraction has for numerator { 2 , 2 } { 2 , 3 } , which vanishes .
When has been determined we find the 's by the formulae\mdash ; 4 The terminations of the continued fractions are as specified above in the equation for By forming continued products of ratios of successive , we can find all the 's as nultiples of , and In the cases , OEC , OOS , EOS , these are the required coefficients for PROFESSOR G. H. DARWIN ON ELLIPSOIDAL HARMONIC ANALYSIS .
In the cases EES , OES , OOC , EOS we put thus find the coefficients for The coefficients for in EEC , EES , OOC , OOS are determined by .
' The coefficients for in OEC , OES , EOC , EOS are determined by It follows that if we put and When is equal to that which would be given by the general formula for when we put in it .
Hence it follows that the 's for double the values giyen by the formula .
If we change the sign of , the two continued fractions in the equation for are simply interchanged .
Hence is unchanged when changes sign .
Also , since is equal to is unchanged when is writtell consideration of the forms of the 's shows that is equal to , and therefore ; .67 .
Rigorous of the Functions of the second If numerical value be ibuted to it is obviously possible to obtain the orous expressions for the several functions .
Thus , if were we could determine the harmonics of the ellipsoids of the class .
But I do not think it is possible to obtain rigorous results in algebraic form when is greater than 3 .
In order , however , to show how our formulae lead to the required result I will determine the five functions corresponding to , but I will not work out the case of although it is eas to do so .
When Therefore , or writing for brevity , .
Then putting , and remembering that the value of is twice that given by the general formula , Therefore ( 17 ) , where .
The coefficient of the cosine function is given by Therefore \mdash ; ( 18 ) .
, cosines ; EOC type .
The continued fraction has , 2 in the numerator , and vanishes because .
Therefore , where herefore But the coefficient is independent of , for , and Therefore \mdash ; ( 19 ) .
arly . . . . . . . .
( 20 ) . .
H. DARWIN ON ELLIPSOIDAL ANALYSIS .
, sine ; EOS type .
The continued fraction vanishes and , but it is not needed to express the functions .
Putting \mdash ; 1 . . . . .
, . . .
( 22 ) .
, cosine ; EEC type .
The second continued fraction ) islles because it , 3 in the numerator .
The equation is Therefore .
Then Therefore ? .
( 23 ) , where .
, and .
, sine ; EES Both fractions and anishes , but is not needed for functions .
Noting that , and , . . . . .
functions of by nd the products of the three functions give rigorously the solid tions of of the second degree .
As I have bove , seven hrmonics of the third degree may be obtained rigorously by a parallel process .
S 8 .
of the Functions .
It is clear that the first approximation to is zer ] ccond approximation , in the general case , is V0L .
XCVII . .
If this expression were inserted in we should obtain correct to .
But since the next approximation would only introduce it follows that would be correct to inclusive .
Now enters in the functions with a factor , and therefore this approximation would give results correct to inclusive .
Since the similar operation could be applied with equal ease in all the cases in which the continued fractions assume special forms , it follows that this degree of accuracy is very easily attainable .
However , the forms of the coefficients would be rather complicated , and it would render the subsequent algebra so tedious that I do not propose at present to carry the approximation beyond It now suffices to put in the denominators of all the continued fractions , whereby the coefficients are determined , except in the cases of , where we put .
In the general case we have ( 27 ) .
' ' When , we double the results given by the general formula and find There ( 28 ) PROFESSOR G. H DARWIN ON ELLIPSOIDAL HARMONIC ANALySIS .
When , with upper sign for cosines ( EOC , OOC ) and lower for sines ( OOS , EOS ) .
for all cases .
But for OOS , EOS we use the ?
form , for , OOC the form ; and for the latter Therefore for OOS , EOS and for EOC , OOC ( cosines ) ( 29 ) .
For OOC , OOS , with upper sign for cosine and lower sign for sine , \mdash ; , 2 , 4 ( 29 ) For EOC , EOS , with upper sign for cosine and lower sign for sine , , 5 ( 29 ) .
When the coefficients may be derived from the general formula .
When the upper sign applying to cosines the lower to sines ( OOS , FOS ) ; But for OOS , EOS the $ form applies , and for OOC , the pplies .
Also with , Therefore for OOS , LOS , 2 , 3 , 2 , 3 , 2 , 3 , 2 , 3 , 2 , 3 , 2 , 3 , 2 , 3 , 2 , 3 , 2 , 3 , 2 , 3 , 2 , 3 , 2 , 3 , 2 , 3 , 2 , 3 , 2 , 3 , 2 , 3 , 2 , 3 , 2 , 3 , 2 , 3 , 2 , 3 , 2 , 3 , 2 , 3 , 2 , 3 with ) lower for sine , [ :3 ] with ) lower for sine , [ :3 ] with lower for sine , [ :3 ] .
OOC , ( i30 ) .
with upper sign for cosine and lower for sine , , 5 , 7 / It will say much trouble to note that if we were to admit negative suffixes to the general formula would give us the term , where 128 .
2 .
1 this term is .
, 3 } .
But this is exactly that part of the term in ( 30 ) which arises , but which is not included in the general formula .
Similarly the general formula gives for those parts of the terms al.ising from which are not included in the general formula .
It follows that in much of the subsequent work we need not devote special consideration to the case of S9 .
xctors of mation bthe two forms of -function C- or S-fn The igorous eessions and always differ from one another , but approximately they are the same up to a certain power of , provided that is greater than a cert quantity .
Since , it is legitimate to develop in powers of 1 up to a certain power , say , provided that it is to be multiplied by a function involving at least as a factor ; for this condition insures that there shall be no infinite tevms when .
At present , I limit the development to so that PROFESSOR G. DARWIN ON ELLIPSOIDAL HARMONIC ANALYSIS .
Therefore It is obvious on inspection that we cannot rely on this development if is less If then is equal to , or greater than 4 , this value of , when properly deyeloped , to the adopted order of ) nation can only differ from by a constant factor , or shortly ; so that . . . . .
( 31 ) , and we have to determine the constant We develop the above expression for completely and compare it with , but this is unnecessary since the cp.ison of a single term suffices .
I now write ( 32 ) , or shortly 2 .
This notation is introduced because this function occurs very fi.equently hereafter .
We have seen in ( 11 ) ( slightly modified ) that We may write this where , Then ) .
Therefore the coefficient of is , or I luce a further abridgement and write ( 32 ) , or shortly T. Then , after reduction , I find \mdash ; . .
Accordingly the coefficient of in is But , and the last term in the above expression will be found to be equal to .
Thus the coefficient of in the development of is ; \mdash ; \mdash ; but the same coefficient in is unity .
Therefore ( 33 ) .
The squares of this constant and of its reciprocal are given because they will be needed at a later stage .
We next consider the cosine and sine functions .
As far as \mdash ; .
OFESSOR G. H. ON LLIPSOIDAL HONIC A Therefore \mdash ; cos2 , \mdash ; This expression , when developed , must lead to or multiplied by a constant factor .
Let . . . . .
( 34 ) .
Then or may be found by considering only the coefficient of .
Hence But ' .
Thel.efore , .
The reciprocals may clearly be written down at once .
There are no factol .
S by which be convel.ted into ; but this is not true of the cosine and sine functions .
In the case of , it will be found that the general holds ) the factor whereby are convertible into When Then . . . . . .
But , , 4 , , 6 , and , where .
Therefore the factors are , ( 36 ) .
It is easy to verify that the other coefficients of .
and @ are in fact reproduced .
The notation adopted herc and below for distinguishing the cosine and sine factors is perhaps rather clumsy , but I have not thought it worth while to take distinctive symbols for the factors in these cases , because they will not be of frequent occurrence .
When . .
This must be equal to Now , with upper for cosine and lower for sine , \mdash ; , 3 , 3 , 5 for its values , we find with the upper sign And with the lower .
\mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; \mdash ; It follows that ( 37 ) . .
H. FI , LIPSOIDAL We cannot in the present case use as an notation , it is infinite as involving in the denominator .
It is easy to verify that the other coefficients of and are , in fact , reproduced in the transformation .
Lastly when , we have only cosine functions .
As before This must be equal to , and therefore Hence ] -l .
( 38 ) .
Since in this case .
Thus the general formula again holds good .
It is easy to verify that the other coefficients of are in fact reproduced .
The principal use of the transforming tors , determined in this section , is that it will enable us to avoid some tedious analysis hereafter .
S 10 .
metions of the Kind .
The second continued fraction of S 6 terminates because when or , since one of the two factors then assumes the form Hence it that the equation for determining is the same as before ; but we cannot on that account assume that the coefficients vanish when their suffixes are greater than In considering the -functions it was immaterial whether or not we regarded them as vanishing , because vanishes if is greater than .
But the -functions do not vanish in this case , and therefore we must postulate the existence of 's with suffix greater than VOL. CXCVII.\mdash ; A. 3 In fact , whilst we have as before , when and are both odd or both even , we also have similarly a fraction for , and so forth .
It follows therefore that while the 's with suffixes less than or equal to depend on finite continued fractions , those with suffixes greater than depend on infinite continued fractions .
It thus appears that } yhile the first series in the expression for or for has limits 1 to or , as before , the limits of the second series are 1 to Thus we have found an expansion for this class of functions in powers of In the limited case in which the coefficients have been actually evaluated , namely , where the development is only carried as far as the squares of , we have These coefficients do not vanish when or are greater than , and this confirms the conclusion already arrived at .
In spherical harmonic analysis there is no occasion to consider the value of when is greater than , and the values are therefore not familiar .
I will therefore now determine them .
It is know Therefore differentiating .
And BRYAN , ' Camb .
Phil. Soc. Proc vol. 6 , 1888 , p. PROFESSOR G. H. tfiWIN ON ELLIPSOII)AL MONIC ANALYSIS .
therefore ) therefore ) therefore ) therefore ) therefore ) therefore ) therefore ) therefore ) therefore ) therefore ) therefore ) therefore ) therefore ) therefore ) therefore ) therefore ) therefore ) therefore ) therefore ) therefore ) therefore ) But ( 39 ) .
These are all the functions which can be needed for the expression as far as of or of when is less or equal to .
If is equal to , we shall have terms or , and these are the furthest .
But it is well known that there is another expression these functions of the second kind .
The differential equation is where may be interpreted as meaning also Let us that is a solution , where may be interpreted as meaning also Then since is a solution of the rential equatiolJ , have \mdash ; 1 PROFES OR G. H. DARWIN ELLIPSOIDAL lONIC ANALySIS This is ence V ' is a constant .
Hence ( 40 ) .
The general -ution of the differential equation must be and we have already found both and .
Hence the two must be different expressions for the same thing , for the form of as a series negatives the hypothesis that it involves in the form Having then two forms of or of , it to evaluate the coefficients , which are involved in the equations ( 40 ) .
In order to do this it will suffice to consider the case where is very great , so that As far as concerns the first term in the series It will be observed that if is equal to or the terms in in and disappear ; and if is equal to or that in disappears .
This agrees , as it should do , with the vanishing of and when the order is greater than the degree .
PROFESSOR .
H. DARIVIN OIDAL HALYSIS .
It ws write firsb of our becomes , when is very large , Therefore oes , and since the in the case of the only differ from these in the accenting of the 's we have . .
the same with accented Effecting the multiplication of the series the same wit , accented If we substitute for the values , the coefficient of inside in the expression for is In the expression for the first pair of these terms are multiplied by , and the second pair by coefficient of in the ession f is In the expression for the first pair of these terms are multiplied by ; the second pair by ; the first of the third pair by , and the second by ; and the last pair by The reduction of terms such as these will occur frequently hereafter , and I will therefore say a word on the most convenient way of carrying it out .
It is obvious that the coefficient of may be arranged in the form The coefficient A is equal to the coefficient of in the original expression , and if we put we have , and with .
Hence may be easily determined .
the coefficient of may be arranged in the form This may be written It is easy to pick out the coefficients of , and we thus obtain , C. Then putting successively equal to and \mdash ; 1 we have and In order to express the results succinctly I use as before the notation ; and I usually omit the superscript and subscript and PROFESSOR G. H. DARWIN ON ELLIPSOIDAL HARMONIC ANALYSIS .
Proceeding in this way I find ( 41 ) .
These results lray be verified , for if we multiply oes by , as given in ( 33 ) , we ought to find ; and this is so .
The formulae apparently fail when , 1 , 2 , 3 ; but when they still hold good because , as remarked above , the general formula for gives correct results when properly interpreted .
Thus it only remains to consider , 1 , 2 .
When the coefficients of remain as in ( 41 ) .
In coefficients of In the expression for the coefficient of inside the bracket is .
( 42 ) .
Effecting the reduction and writing for , we find ( 43 ) .
The coefficient of for may be got from ( 42 ) thus :Multiply the first and second terms by , erase the third , fifth , and sixth ter1ns , and multiply the fourth term by 4 .
Effecting the reduction we find ( 44 ) .
Observe that there is no factor by which can be converted into , so that this case cannot be verified like the general one . .
H. ON ILLIP , IC When ] The terms in .
and now contribute to the terms in For the in inside the bracket is The in of which the first portion is carried over from the term in , is - \mdash ; 1 \mdash ; 1 This is equal to \mdash ; \mdash ; As we cannot now use abridged notation with , which is infinite , I write .
Thus ( 45 ) .
For the coefficient of is three times as great as ) efore , and the coefficient of is On effecting the reduction I find ( 46 ) .
When have only to determine .
Here The term in is ] .
That in .
is .
PROFESSOR G. H. DARWIN ON LLIPSOIDAL HARhIONIC ANALYSIS .
hcrefore ( 47 ) .
since Collecting results from ( 41 ) , ( 43 ) , ( 44 ) , ( 45 ) , 46 ) , and .
( 48 ) .
\ldquo ; PART II .
APPLICATION 0F ELLIPSOIDAL HARMONIC ANALYSIS .
S 11 .
The Potential of harmonic of an Ellipsoid .
A solid harmonic , 01 solution of equation , is the product of two -functions of and of respectively , and of a cosine or sine function of surface harmonic is a -function of multiplied by a cosine or sine function of We found , ; and a similar formula held .
Hitherto we have ) to have exactly thesanle form as .
But since is less than unity this introduces an ary factor when is odd , and VOL. CXCVJI.\mdash ; A. 3 .
H. DARWIN ON oIDAL H IONIC ANAL llakes the successionl of 's alternately positive and negative when is even .
As this is practically inconvenient I now define and then reffiining the former meaning for coefficients , we give the following definition\mdash ; ; with a similar formula for .
Thus we need only remal.k that in the fulJctions of the 's corresponding to odd owers of enter with the opposite sign from that which holds in the functions of and the whole of our preceding results are true with this definition of .
If defines the ellipsoid the face harmonic applies , we require the expression for the o11 the tangent plane at , and that for an element of area of the of the ellipsoid at the same point .
By the usual formula ( 49 ) .
Let be the elements of the orthogonal arcs corresponding to tions of respectively .
by the formula at the end of S 1 , ' ( 50 ) .
Therefore ) ( 50 ) , G. H DARWIN ON LLLIPSOIDAL HARMONIC ANALYSIS .
and ( 50 ) .
Two functions , written in alternative form , , are solutions of LAPLACE'S equation , and ether form a function V continuous at the surface of the ellipsoid .
ReadinE the upper line we have a function always finite inside the ellipsoid , and the lower line one always finite outside .
Hence V is the of a layer of surface density on the ellipsoid and by POISSON'S equation that density is equal to ( outside ) \mdash ; inside ) Let the surface density , which it is our object to find , .
a surface harmonic multiplied by the perpendicular on to the tangent plane and by a quantity Then since , But Differentiating this logarithmically we find , a constant .
Hence surface density .
, where is constant , gives rise to potential the } \mdash ; ( 51 ) .
The same investigation holds good with , or with in place of the corresponding letters above .
Imagine that the surface of a homogeneous ellipsoid of density , defined by receives a normal displacement Sn , such that .
Then the equivalent surface density is , and we can at once write down the expressions for the internal and external potentials by means of ( 51 ) .
If be the co-ordinates of a point on the surface , it is clear that the co-ordinates of the corresponding point on the deformed surface are .
Hence the equation to the deformed surface is ( 52 ) , or since ) , it may be written .
If we substitute for value from ( 49 ) , this may be written in the form ( 52 ) This is the equation in elliptic co-ordinates to the deformed surface , but in actual computation the form involving rectangular co-ordinates might perhaps be more convenient .
S12 .
The Potential of a homogeneous solid Ellipsoid .
It is well known that the potential of a solid ellipsoid externally is equal to that of a " " focaloid\ldquo ; shell of the same mass coincident with its external surface .
If be the density of the shell defined by and , we have PROFESSOR G. H. DARWIN ON ELLIPSOIDAL HARMONIC ANALySIS .
Therefore or .
( 53 ) .
If Sn be the thickness of the shell at the point where is the perpendicular on the tangent plane , .
If we multiply both sides by , we see that the surface density of the focaloid shell is If therefore we can express in the form of surface harmonics , it will be easy to write down the external potential of the ellipsoid by means of the formula ( 51 ) .
Before doing this I will , however , take one other step .
It is easy to see that ( 54 ) , where for brevity Now on referring to , ( 17 ) and ( 23 ) , we see that , where \mdash ; .
If then we put , ?
it is clear that ' ; and OFFSSOlt C. H. DARWIN ON FLLIPSOIDAL HARMONIC ANALYSIS .
It is obvious then that our expression ( 54 ) is equal .
Then since , we have the surface density of the focaloid given by . .
where But since ) ( 55 ) , we have .
( 56 ) .
With the object of writing this function in surface harmonics , and besides to enable us to a rotation potential in similar form , we have to reduce in the required maIlnel. .
I now drop the suffix zero , since we are not concerned with any particular ellipsoid .
Referring again to S7 , ( 18 ) and ( 24 ) , we have If then we put we may write Let us assume , if possible , ; ; or From which it follows that PROFESSOR G. H. DARWIN ON ELLIPSOIDAL HARMONIC ANALYSIS .
These equations give and the condition Now ' Since these values satisfy the condition tJst the coefficients , the assumed ] for is justifiable .
I find then Whence .
( 57 ) .
This is the required expression for .
in surface harmonics .
Next assume If we put ; we have ; ; and Whence have the same forms as before , and the condition to be satisfied by the coefficients is will be found that the condition is satisfied , and that Hence ( 58 ) .
It follows from ( 55 ) that .
( 59 ) .
Whence \mdash ; 1 .
( 60 ) .
This is needed to express the rotation potential .
If we add to this we have ( 61 ) .
This expression will be needed hereafter .
Returning now to the formation of the expression for , I find On considering the forms of the functions , ?
, it is found that this result be written thus : Therefore .
writing for unity , the surface density of the focaloid shell , for which , is .
( 62 ) IVIN ON IC of ( 51 ) , we at once write down the potential of the It is V .
In this expression denotes the of ellipsoid , fficients determined tely in 10 .
ordel ' to find the ) ) , let ; suggested by the fornl of ( 61 ) , let I. .
Then a ofhout the of the ellipsoid , the , where , it is equal to x- ) consider the function ( r- ) ( 64 ) .
whole of it , .
the in is a solution of inside the ellipsoid , the tel in ives snrface , ) rees with .
Heltce potential of the ellipsoid internally .
The potentia } internal point lend itself ) ) -ordinates , but it be giyen ( which is ) ) ) resel l notation well-known is ' integrals may be ) essed in of the ) divisors o and for ) .
- .
CXCV 3 ' OIDAL I In this may substitute the expressions found above .
lt be worth noting that the f that if the of the , and if denotes nction ional to our , is ction of in therefore S 1 eparation fthe of the harmon ?
the Ellipsovd .
If it is intended to lction i harmonics , it is necessal.y to know the tegrals o the surface of the ellipsoid of the squares of harmonics multiplied by the on the tallgent plane .
urface hmonic h one of the eight ; and the -functions are expressible in of the 's where I in this portion of the investigation frequently write , and shall omit the or or P- , C- , .
Also I the subscript If denotes the element of of the ellipsoid , volulne ellipsoid , have , of S 11 , Then } , whele the limits of are to \mdash ; , and of are to G. H. DARWIN ] ) ONIC A 10- It will be legitimate to develop in powel .
S of ) up to any given provided involves as a factor such a power of that the whole function be integrated does not become infinite at the poles where I shall at present limit the developments to the square of We know that is of the same form as , but with the additional factor Suppose then that ; nnd let Then we put , Now pose that , a function independent of , denotes one of the fonr .
Then in cases involving -fnnctions and -fnnctions respectively , ] in dternative form\mdash ; for If it be supposed that the development in of is justifiable ' And a form , save that eplaces It ' clear that unless is divisible and by and will have infinite elements at the , and the development is not legitimate .
Since , it follows that the lower of by which is diyisible increases as increases .
SSOR t. H. DARWIN ON LLIPSOII ) HARMONTC LYSI , Let us consider the case of Then ( the same with accented 's for the other case ) .
Fron this it is clear that is divisible by and by , and the method of nent is legitimate when , but it is not so when and The investigation then into the general case , and the ] .
14 .
Integration in the ) .
We have and Also has the same form with accented , so that it be necessary to accent the to obtain the second case .
We have then , since . .
And has the same form , but with accented , and with replacing , and replacing .
H. ON IPSOII ) It is llow necessary to evaluate the sevel'a ] definite integrals involved in this expression .
It is well known that It easy to see that it is possible to express in the form .
, where , C. . .
do not involve alue of A may be found by considering the highest power of on ench side of the identity .
Now . . .
: . . .
Therefore Thell , since the integral of the product of two 's of different orders anishes , have We will next consider , where is The differential equation gives Multiply the first of these by and the second by , and we have G. IONTC , between limits Again since by ( 1.1 ) it follows that , unless or 1 .
It remains to find the integrals of , and We have freen in ( 11 ) ( ansformed to accord with our present definition of ) Hence The first of these involves integra]s already determined ; on introducing them on the right and reducing we find the result to be The first and last terms of the second integral vanish , and the integral is The second and third terms of the third integral vanish , and the whole is clearly * I owe this method of finding these last two tegrals to .
H. ON ONIC A Collecting results have / Then by means of ( 66 ) I I Therefore \mdash ; 1 . .
( 68 ) .
bSOR G. H. 1 ARWIN ON ]ONIC A Also has a with ccented q for \mdash ; When we substitute for its value / I the in \mdash ; ( 68 ) .
Also the last tern ] in becomes .
( 68 ) .
But it will ) pear later that we only need the parts of these ternls which involve developed as .
as the power of ; hence in both cases we write the latter term inside simply as Our general formulae the coefficients apply all values of down to , although the esnlt f needs } ) oper i } ) retation .
Jfence the result ) plies down to , inclusive .
1 -defined 2 , and I emind the if in ( 68 ) we oduce f the find that the coefHcient of the term is \mdash ; \mdash ; \mdash ; I ) .
The coeflicient of the is If be reduced by it process to that in S 10 , we find SSOR ( .
H. DARWIN ON .
( 69 ) .
We know that is derivable from by multiplication by , and we have found in ( 33 ) , S9 , Hence multiplying ( 69 ) by we hay .
I have also this estllt by direct elopnnent .
It ulay be surprising that the last term is now same in formulae , the clifference ] ) earlier , but if reader will go through the analysis be will see how this ] ) been brought about .
and ( 70 ) also hold true when ] ( as I yerified ) , notwithstanding the f that is to be from ' by a factor .
The next is the with ] to We have Therefore Also ) have the } forms with accented 1 ) , with unnaccented ) ' , we ) ) ) ON IPSOIDAL HARMONIC ANALYSIS .
] .
and yitl)ccented p 's we multiply by and we shall then have the functions denoted above by function has to ) multiplied by a function of the integrnted from to .
It follows that the only terms in which will not vanish are those independent of and those in ; moreover , the latter terms far as the first power of No , \mdash ; \mdash ; .
omitting which will vanish on integration However , the latter formula is not needed except ification , because it will he derivable from the former by multiplication if we substitute for the ) ' their values as given in ( 27 ) , , we find ( multiplying by or 1 \mdash ; ) , or developing directly or These represent ) K- of on 1 integraIs .
TheYl .
H. IC .
( 71 ) .
The ving 1estllts n be obtained eithel by development , multiplication by either or both the nnd into , the tter o @ into S. . . . . . . . .
( 71 ) .
15 .
Integrc ; tion in the case of Although the development in powers of is still l in this , yet the nulae f in the last section fail because contains in the natol .
Mol.eovel .
since is not convertible by uusl ) considel.edsepal.ately .
We now have therefore ' is same cconted s last We ( , associated with cosine function .
ssociated with a sine fullction .
It to note thebe by the general mula , because shows that much of eductions is still applicable . .
the eductions I find { This will ) associated with in ) osent c .
the same way This be yith S turning to the cosine and sine functions , we find that they must be eated , but the that in by the I.r the in ( 36 ) , S ( t ) .
We have for the sine function } \mdash ; .
lI .
lnaterial No , We these by ) ectively , the terms disefore .
of the do the factor excellent test of ectnesh the eductions throughout this part of ) Then ) } ( 7 .
) .
If multiply by \mdash ; , ) in result eplaces ; the only is that the last inside } .
If this by ) ) obtain the result when S the only change last .
This the rals , ninod I ) ) of ( .
in ) .
H. .
Portion of ) ' the ) of The eceding method may be used finding the foul integrals will then remain four of the type to evaluate .
The first of in clearly to be eated by putting , and then mining .
The condition the second pair only in the tccentuation ( the which anish , and in the use of vanishing of makes , .
of these ; and the econd the form is the same with accented Also , we ' ; is the with accented Ic is only to the and ) , since ) othel pair of may be determined by means of multiplication the factors , determined in S9 . .
H. ON HARMONIC ANALYSJS .
for ?
, associated with ; and for , associated Therefore In the present case we cannot nse as an abridgement , ce it infinite ; I therefore now write .
ffecting the reductions we have The former of these is associated with @ , the latter with In the cosine sine functions we , and \mdash ; ; far as material , we then have ; .
H. ON LIPSOII)JONIC ANALySIS .
' botb for sineH and , to the order necessary for our present purpos , 2 } .
Therefore , introducing for , ( 73 ) .
and not involve , and are of the first order in , ( material for the } portion of the work , .
( , to the order , , necessary Hence .
( 74 ) .
have only to replace the factor 1\mdash ; , and find ( 74 ) .
PROFESSOR G. H. DARWIN ON HARMONIC ANALYSIS .
Again , omitting intermediate steps , ( 74 ) . .
S 17 .
Portion of ?
of We are to find , leaving two integrals of the type to be determined subsequently .
It is only necessary to consider , since the other case is determine able from it by multiplication to as found in ( 38 ) of S9 .
Following the procedure of the case where , we have Then since Now and as far as material SOR ) SOIDA HARMONIC At present we only require this to the first power of , and since does not contain , the expression ( 75 ) as far as at present needed is simply unity .
Again , by ( 38 ) of S 9 , , therefore by multiplication ( 76 ) .
This is also unity to the order at present needed .
Hence . . .
( 77 ) .
S18 .
Preparation.for tloe Integrations when and We have now to evaluate the three integrals ( 78 ) , and from these to determine three others when replace have It is the second factor which alone involves , and as I shall now first integrate with respect to , the first factor may be dropped for the moment , and the second factor multiplied by the squares of the cosine or sine functions .
Since the integration is from to , those terms which vanish on integration may be dropped .
1ROFESSOR .
H. DARWIN ON ELLIPSOIDAL For brevity write . . .
, , Then we have seen in ( 73 ) and ( 75 ) that Therefore . . .
Therefore . . .
Therefore . . .
Therefore . . .
Therefore . . .
Therefore . . .
Therefore . . .
Therefore . . .
Therefore . . .
Therefore . . .
Therefore . . .
Therefore . . .
Therefore . . .
( 78 ) . . . .
Now pick out the parts of and of these integrals ( 78 ) which are independent of , and write , . . . .
/ Also write , ' so tnat ( 80 ) .
Lastly , in accordance with the usual notation for elliptic integrals , write ( 80 ) .
Then we have ( 81 ) .
The next step is to express the of the 's in of powers of When } , it is known that By putting we see that is the coefficient of in the expansion of .
By TAYLOR'S theorem this last is equal to sin .
[ terms involving powers of ] , ' \mdash ; when Also On putting , we see that the coefficient of in is PROFESSOR G. H. ON ELLIPSOIDAL 5.3.3 Hence we have* ( 82 ) .
Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Then clearly Therefore Now suppose ( 82 ) .
( 83 ) .
The evaluation of these integrals depends on two integrals only , namely , and , and these will be considered in the next section .
S 19 .
Evaluation of the Integrals and I will denote these integrals and respectively , and I propose to find their values in series proceeding by powers of The usual notation is adopted where is such a function that it is equal to accordingly when is a positive integer ! .
Since ' is less than unity ; and since Mr. HOBSON kindly gave me this proof lvhen had shown him the series which I belisved to hold true .
or , with the usual notation geonletric series , This series is of no service , since it proceeds by powers of , which in our case is nearly unity .
It is required then to transform the series into one proceeding by powers of It is known that , if If we apply this theorem with , the first becomes whose nth and all subsequent terms involve zero factors in the denominators .
Also the coefficient of the second involves , which has an infinite factor .
Hence the formula leads to an indeterminate result .
Let us therefore put , and proceed to the limit when We have then Limit Now , .
Therefore , when is very small , . . .
I to thank for giving me this , and for me the procedure whereb.v it can bo PROFESSOR .
H. DARWIN ON ) fONTC ANALYSIS .
Hence for the coefficient of the first series we have But , Therefore This is true from to 1 , but in the case of we have ' ( 0 ) , so that in that case Now consider the coefficient of the second series .
We have , and since Therefore the coefficient of the second series is , and The case of is an exception , for the coefficient of the the part inside replaced by * Proved by entiating the known formula .
putting PROFESSOR .
H. 1)ARWIN ON LLIPSOIDAL HARMONIC ANALYSIS .
It relnains to consider these two When When , put , and when Thus It follows that we may write the first term of as follows:\mdash ; ' The first of these terms becomes infinite when Turning to the second we have PROFESSOR G. H. ON ) Thus the second term of is The first of this becomes infinite when , but it is equal and opposite to the infinite term in the first part of , and ihihte one another .
Hence exalnining the case of we finrl that umla also ) races i ovided we interpret as mean i .
The coefficient in the la , st admits of some simplification , We thus conclude that I ) 0 to ( 84 ) .
The second integral may be found as follows:\mdash ; .
CX .\mdash ; A. G. H. ON LLTPSOIDAL HARMONIC this I find or This is applicable also to the case of provided tha is interpreted as zero .
In the particular case in hand I find , however , that it is shorter not to use this general formula , but to out transformation ( 85 ) in the particular cases where the result is needed .
S20 .
Red of preceding ppearance of logarithmic terms .
In the application of the integrals of the last section , we are to put ' and only to develop as far as to the proposed order , Also lt will now facilitate future developments to adopt an abridged notation .
I write observe that , and , .
Since is of the first order in , only the firffi series in the integral ( 84 ) enters when is greater than 2 .
In that case ( 86 ) .
This result may be obtained very shortly without reference to the general formula ; is greater than 2 .
G. H. integral of an odd powel of is it } be lound that the l.esult ( 86 ) is obtained .
It is , , clear if is not ) the development in of ; not legitimate .
When is not than ) mula ( of the section is , and find ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; .
( 87 ) .
We have to find the second integl .
A this be done more easily than by reference to the general of the last section .
We have lt will be obseryed even when is ) the general formula ( 86 ) gives .
as far a , s the of .
Hence in finding we use eneral formula except when , 1 .
Then since , whel 7 is . . . . . .
( 88 ) .
But when .
when / ( 90 ) .
G. 11 .
1)ARWIN ON ) HARMONIC ANAL Y I now wish to show that , in the use to hich these integrals to be put , the thmic t disappeal The followin is table of these ; collected , ( 89 ) , ) ) , in as ' involve Then formulae for , in so only as is at ])esent m , are On using the integl.als and only etaining squares of , we find .
But by definition of .
and in ( 80 ) of the 's in ( 82 ) , to the zero of small quantities , rithmic terms disa , henceforth ] ) Thus , nsfar , we have table of integrals PROFESSOR .
H. RWJN OilIS .
, ( 91 ) .
Before using these for the determination of , it is to obtain one othel result .
We have seen in ( 82 ) that But integral is .
; thel.efore Putting 1 and aring with the values of in ( 8 ) ) , we have ' , . .
of of monics w we Thel.efore , that , and using the PROFESSOR G. H. DARWIN ON ONIC ANALYbIS .
ubstituting f ( which the eadel .
must not confuse with the functional in use here ) its value ( 80 ) , the of ) is .
By ( 92 ) this is equal to The of the first .
in is , be reduced to the fornl is to .
The of the second order in is This may be reduced to the form which the term is Therefore Now .
It known that ( 93 ) helt since infinite factor ( 93 ) .
I to fur this is due to GALSb .
PROFESSOR G. H. ON LLIPSOlDAL HARMONIC Also Hence .
Introducing the value of as defined in ( 79 ) , we We in ( 74 ) obtained , and if it be added to onr last esult we see that the term which does not involve the factor is annihilated , nd ( 94 ) .
Now the square of the factor converting into .
Therefore .
( 94 ) .
These are two of the required integrals .
Next we have Noting as before that , and the integrals ( 91 ) , Substituting for its value , find the term of to be , or The term of the first order is OIDAL HARMONIC be reduced to the ; which by ( 92 ) becomes The term of the second order is This is reducible to ; which becomes Therefo ] Introducing for its value ( 79 ) , we find or But in ( 74 ) we have ) .
If this be added to the result.jnst found the terlJl which has not as factor is annihilated , and .
Now from ( S7 ) the square of the factor for ting into is .
Therefore ( 95 ) .
hese last two complete the solution ] .
ON Next we from ( 83 ) Proceeding as before , .
its value ( 80 ) , we find the .
order is , by ( 92 ) this is equal to of the first order is which be written in the form , clnd is to of the second order is which is equal to .
lIence ] , , .
But ON ONIC A the ) receding fnlula the sum of this last function had limits to 1 , but its lye see that it vanishes when , the uppe ] limit be changed to It follows that the terms of the second order in in this ) ession will be found to be - in will to .
Then since terms lLle togethel .
hole m be written .
last result follows from the fact that in accordance with ( 93 ) the of ypergeometric series an infinite factor in the denominator , and vanishes .
since ) ' the terms of the second to ) .
Hence , collecting terms , .
Substituting its , we G. H. DARWIN ON HARM[ONIC ANALySI But we have already found in ( 77 ) the value of , and on adding it to the last result the term independent of disappears , and we have .
The square of the ) whereby i , s converted )(38 ) , amely , .
These ore the last of the required integrals .
S22 .
of of In this section the results obtained in ( 71 ) , ( 72 ) , ( 94 ) , , ) are collected .
After having completed the evalnation of all these inbegrals , I fonnd that they may ) ated very shortly by means of the factors and of ( 48 ) , S10 .
I find that for all values of ( vriting the eight forms in a single formla ) , const .
part of I the eadcr to verify this is so .
tely I hitherto been mablo to pve the of this except the ) in the text .
I do not therefore whether the resnlt rolnains trnc for higher degrees of tion , :dthough I snspoct it does so .
If it shonld ) , it onld ) very ensy to the when higher powers of aro It may ) worth mentioning that the the into , .
Thns , denotcs one of the eight forms , The integrals present no difficulty , but with regard to the others we are met b.y the of expanding powers of for the lower ordors .
It would a stcp in tho ight d , if it could ) proved that / the which do not we the factor ncccssnrily ) nish .
It may be well to remind the reader that .
irst when Types Types Types Types Secondly , when .
Type EEC Type OEC Type EES Type OES Thirdly , when , 2 .
is infinite and we must use .
SSOR G. H. DARWIN ON ELLIPSOIDAL HARMONIC NALYSIS .
Type EOS Type OOC Type EOC Lastly , when .
There are only two types\mdash ; Type EEC Type OEC PART III .
SUMMARY .
The symmetrical form in which LAME presented the three functions whose produc is a solid ellipsoidal harmonic is such as to render purely analytical investigations both elegant and convenient .
But it seemed to me that facility for computation might be gained by the surrender of symmetry , and I have acted on this idea in the preceding paper .
Spheroidal analysis has been successfully employed where the ellipsoid is one of revolution , and it therefore seemed advisable to make that method the point of departure for the treatment of ellipsoids with three unequal axes .
In spheroidal armonics we start with a fundamental prolate ellipsoid of revolution , with imaginary semi-axes .
The position of a point is then defined by three co-ordinates ; the first of these , , is such that its reciprocal is the eccentricity of a meridional section of an ellipsoid confocal with the fundamental ellipsoid and passing through the point .
Since that eccentricity diminishes as we recede from the origin , plays the part of a reciprocal to the radius vector .
The second co-ordinate , , is the cosine of the auxiliary angle in the meridional ellipse measured from the axis of symmetry .
It therefore plays the part of sine of latitude .
The third co-ordinate is simply the longitude .
The three co-ordinates may then be described as the radial , and longitudinal co-ordinates .
The parameter defines the absolute scale on which the figure is drawn .
It is equally possible to start with a fundamental oblate ellipsoid with real axes .
We should then take the first co-ordinate , , as such that .
All PROFESSOR G. H. DARWIN ON LLIPSOIDAL HARMONIC ANALYSIS .
that follows would then be equally applicable ; but , in order not to complicate the statement by continual reference to alternative forms , I shall adhere to the first form as a standard .
In this paper a closely parallel notation is adopted for the ellipsoid of unequal axes .
] ) squares of semi-axes of the fundamental ellipsoid are taken to , and the three co-ordinates are still .
Although their eometrical meanings are now by no means so simple , they may still be described as radial , latitudinal , and longitudinal co-ordinates .
As before , we might equally well start with a fundamental ellipsoid whose squares of semi-axes are , and replace .
by where .
All possible ellipsoids are comprised in either of these types by making vary from zero to infinity .
But it is shown in S2 that , by proper choice of type , all possible ellipsoids are comprised in a range of from zero to one-third .
When is zero we have the spheroids for which harmonic analysis already exists ; and when the ellipsoid is such that the mean axis is the square root of mean square of the extreme axes .
The harmonic analysis for this class of ellipsoid has not been yet worked out , but the method of this paper would render it possible to do so .
We may then regard as essentially less than , and may .
l in powers of In spheroidal analysis , for space internal to an ellipsoid , two of the three functions are the same -functions that occur in spherical analysis ; one being a function of , the other of .
The third function is a cosine or sine of a multiple of the longitude .
In external space the -function of is replaced by a -function , being a solution of the differential equation of the second kind .
The like is true in ellipsoidal analysis , and we have and -functions of for internal and external space , a -function of , and a cosine- or sine-function of .
I will now for a time set aside the -functions and consider them later .
There are eight cases to consider ( S 4 ) ; these are determined by the eyenness or oddness of the degree and of the ordel of the harmonic , and by the alternative of ether they correspond with a cosine-or sine-function of .
I indicate these eight types by the initials , or S\mdash ; for example , EOS means the type in which is eyen , is odd , and that there is association with a sine-function .
It appears that the new -functions fall into two forms .
The first form , which I write , is found to be expressible in a finite series in terms of the where the 's are the ordinary functions of spherical analysis .
The terms in this series are arranged in powers of , so that the coefficient of has as part of its coefficient .
The second form , which I write , is such that or is expressible by a series of the same kind as that for ?
Amongst the eight types four involve -functions and four -functions ; and if for G. H. DARWIN ON ELLIPSOIDAL IONIC ANALYSIS .
iven s-function is associated with a cosine-function , the corresponding is associated with a sine-function , and vice verse Lastly , -function of is always associated with a -function of ; and the like is true of the Again , the cosine- and sin -functions fall into two .
In the itre either both even or both odd , the function , which I write or is expressed by of terms of a coeflicient multiplied by .
In the second form and difler as to eyenness and oddness , and function , writtell or , is expressed by similar series multiplied by The combination of the two iorms of -function with the of sine-function gives the eight types of solid harmonic .
Corresponding to the two forms of -function there two forms of Q-function , such that and are expansible in a series of ordinary -functions ; but whereas the series for and are tel.lninable , because vanishes when is ter , this is not the case with the -fun ctions .
In fact the series the -functions begins with or , and the order of the 's increases by two time up to when we have the principal central ; it then goes on incl.easing up to or on to infinity .
In spherical and spheroidal analysis the ential equation satisfied by involves the integer , whereby the order is specified .
So also the equations , satisfied by and by , or , involve a constant , but it is longer an integer .
It seemed conyenient to assume the this constant , where is the known integer specifying the order of harmonic , remains to be detel.mined from the differential equations .
When the assumed forms for the -function and for the cosine-and sine-functions substituted in the diflerential equations , it is found ( S 6 ) that , in order to the equations , must be equal to the difference between two finite-continued fractions , each of which involyes .
We thus have an equation the required root is that which vanishes when vanishes .
For the of , 1 , 3 , and for be found rigorously in ic fornl , but highel degrees the equation be solved tely , unless should definite ical value .
Whell , either rigorously or tely , uccessiv coefficients of the minable i such a that the of coefficient to the one is expressed by a continued fraction , which is , in tion of one of the two fractions involved in the equation Throughout the rest of the paper the greater part of the work is carried out with approximate , and , although it would be easy to attain to greater accuracy , have thought it sufficient , in the first instance , to stop : With this ' H. DARWIN ON ELLIPSOIDAL MONIC ANALYSIS .
tion the coefficients of the sel.ies assume simple forms ( S 8 ) , and we have thus definite , if approximate , expressions for all the functions which can occur in ellipsoidal analysis .
In expressions , and are essentially different from one another , but in approximate forms , when is greater than a certain integer dependent on the degree of approximation , the two are the same thing in different shapes , excel ) to a constant factor .
I have , therefore , in 59 determined up to squares of the whereby is convertible into , and or into or .
With the of approximation adopted there is no factor for converting the P'swlle , 2 , 1 .
Similarly , down to inclusive , the same factor serves for convel.ting into and into .
But , 1 , one form is needed for changing into , and another for changing into S. It may be well to note that there is no sine-function when is zero .
The use of these factors does much to facilitate the laborious reductions involved in the whole investigation .
It is well known that the -functions expressible in terms of the -functions by means of a definite integl.al .
Hence and must have a second form , which only differ from the other by a constant factol .
The factors connecting the two forms are determined in The second part of ] ) is deyoted to applications of the harmollic method .
In 11 the perpendicular the centre on to the tangent plane to an ellipsoid the of an element of surface of the ellipsoid , are found in terms of the co-ordinates , and the constant It is easy to form a function , continuous at .
the surface , which shall be a solid monic both for external and for internal space .
POISSON'S equation then enables us to determine the surface density of which this continuous function is the potential , and it is found to be a surface harmonic of multiplied by the perpendicular on to the tangent plane .
This application of POISSON'S equation involves the use of the -function in its form .
Accordingly , if the serial form for the -function is adopted as a standard , the expression for the potential of a layer of surface density involyes the use of the factor for conversion between the two forms of -function .
This esult may obyiously be employed to determine the potential of an harmonic deformation of a solid ellipsoid .
potential of the solid ellipsoid itself may be found by the consideration that it externally equal to that of a focaloid shell of the same mass .
It appears that in order to express the equivalent surface density in surface harmonics , it is only necessary to express the reciprocal of the square of the perpendicular on the tangent plane in that form .
This result is attained by expressing in surface harmonics .
When this done , an application of the preceding theorem enables us to write down the external potential of the solid ellipsoid at once .
In S 12 the external potential of the solid ellipsoid is expressed rigorously in terms of solid hal.monics of PROFESSOR .
H. DARWIN ON ELLIPSOIDAL ANALySIS .
Since have been found in surface harmonics , we can also write down a rotation-potential about any one of the three axes in the same form .
The internal potential of a solid ellipsoid does not lend itself well to elliptic co-ordinates , but expressions for it are given in S 12 .
If it be desired to express any arbitrary function of in surface harmonics , it is necessary to know the integr ls , over the surface of the ellipsoid , of the squares of the several surface harmonics , each multiplied by the perpendicular on to the tangent plane .
The rest of the paper is devoted to the evaluation of these integrals .
No attempt is made to carry the developments beyond , although the methods employed would render it possible to do so .
When is greater than unity , it appears that it is legitimate to develop the function to be integrated in powers ; and when this is done , the integration , although laborious , does not present any difficulty .
But when is either 1 or , the method of development breaks down , because it would give rise to infinite elements in the integrals at the poles where is unity .
However , portions of the integrals in these cases can still be found by the former method of development .
As to the residues which cannot be so treated , it appears that they depend on in , of the forms and where nearly equal to unity .
Development of the square-roots in powers of is useless on account of the slow convergence , and it is required to find series which proceed by powers of , where By a somewhat difficult investigation , in respect to which I owe special thanks to Mr. HOBSON , the needed series are found ( S 19 ) .
It appears that portions of the two integrals involve logarithms which become infinite when vanishes .
Since , in the application of these integrals , the vanishing of implies the vanishing of , we appear to be met by a difficulty .
It is known that in spheroidal analysis no such terms appear , and we Inay feel confident that they cannot really exist in ellipsoidal analysis .
In S20 it is proved that the logarithmic terms do as a fact disappear .
The residues of the integrals in the cases are thus found , and added to the previous portions to form the results .
The second part of the paper ends ( S22 ) with a list of the integrals of the S of the surface harmonics for all values of , as far as the of Finally , an appendix below contains a table of all the functionls as as .
It is probable that for the higher values of the results would only be applicable when is very small .
VOL. CXCVII .
\mdash ; A. APPENDIX .
Table of the P- and -Functions .
( OEC ) ( OOC ) ( OOS ) ) ( EEC ) ( EOC ) ( EEC ) ( EES ) ( OEC ) ( OOC ) ( OOS ) ( OES ) ( OOC ) ( OOS ) PROFESSOR G. H. DARWIN ON IONIC ANALYS ( EEC ) ( EOC ) ( OS ) ( EEC ) ( EES ) ( EOC ) ) ( EOS ) ( EEC ) ( EES ) ( OEC ) , i ( OOC ) ) ( OOS ) ( OEC ) ( OOC ) ( OOS ) ) ( OES ) ( OCC ) ( OOS ) } PROFESSOR G. H. DARWIN ON ELLIPSOIDAL HARMONIC ANALySIS .
Note that in this denotes , and is If the variable is , if accordingly the factol ) in is replaced by , the of all the terms which as coefficient must be changed .
still same neaning , must writteYl form Table of the Cosine Sine ( EEC ) ( OEC ) ( EEC ) .
( OEC ) ( EEC ) PROFESSOR ( G. H. ON ELLIPSOIDAL HARMONIC Note that in this table A table of and to is contained in ' paper Proc. Camb .
Phil. .
The functions there tabulated as and in the notation here adopted , would be factor ) and .
formula for , where reater than , is given above .

