]\gt ; At the end of cvions p it WttS stated that the bility of when pcar-shapod could not bc definitely without furthel approximation .
After some correspondencc with during the course of luy vork ou that paper , I lttompted to out the second ) foumd my self foiled ccrtaiIl stage of the work .
hile he had ttlrned his tteIltio to the subject , and he has shown how the difficnlty hich stopped me may be overcome .
He not , howcvor , uIsued the rdnous t of onverting his into ltlmll ) , so hc the question of stability unI .
so kind to mc to his manuscript on its to the Sociely for a felv , thus assisted I able to resume my attempt under favourable conditions , and this paper is the resnlt .
The snbstance of the dysis of this paper is , of course , essentially the same as his , but the two present but little superficial resemblance .
It is well that the two investigations of so complicated a subjcct ) nearly independent of ons another .
If a of liquid be rotating like a rigid body with uniform angular , the ation of the ( of cqnilibrinm may ) * .
R. Acad. , , pp. 17\mdash ; 30 .
'Phil .
, vol. 198 , pp. 301\mdash ; 331 .
' Phil. rans .
, vol. 198 , pp. 333\mdash ; 373 .
of of a of treated a statical problem , if the mass ) subjected to rotntio ] ] potential .
The lost in the conccntration of system from a condition of infinite dispcrsion consists of .
The first of these , say , is ths lost energy of the system at rest ; the second is equal to the etic energy , say , of the systcm in motion .
The whole lost , is equitl to the condition for a figure of librium is th( shall ) st subject to velocity .
It might appear at first sight the condition for stability is that shall be a .
But I. that this condition is insufficicut , th it is for that the whole energy , hich is equad to be for all tions , subject to the conditio1l of of monleutnn .
He , however , tion , hich enables to determine the stability from the tions of without direct consideration of the functiou U. He shown , iu fact , if for giveIl angular momentum htly less that of the cobi ellipsoid , from which the figures , there is only one possible figurc , namely , the Jacobian ; if for slightly angular there two fignres , namely , the and the pear , then exchange of bility b the two series mnst occur at the bifurcation .
If , on the other , the smaller momentnm corresponds with the two figures itnd the larger with only one , one ot two namely , mnst bc stal ) , and the othel ' ( namely , the pear ) unstable .
The question is then completely by the of the of the pear ; if it is th thnt of the critical Jacobian , the pear is if less , mstable .
It suffices then to determine the pear from the variations of velocity , fterwards to the angular momentnm .
In the first approximation the pcar-shaped figure is esonted b the third zonal harmonic inequality with reference to the gcst axis of the critical cobian ellipsoid .
In proceeding to the higher approximation I suppose its mplitudc is measured ) a meter , which to be regarded a quantity of the first order .
must now also suppose the ellipsoid to be deformed by every harmonic , but with amplitudes of order In the first approximation was proportional to , but it now becomes to as lts the .
A ohange in the sign of means that the figurc is rotated in azimuth through .
As this rotation cnnnot aflect the cnergy , the odd powers of must be absent from the exprcssion for We have further to find the moment of , as far as the * For the salre of , I spcak of one ] of ) ( liffcring by ] 180 Prof. G. H. Darwin .
of the .-shaped [ Jume 19 , of order , and thence to find the kinetic energy T. The function is then to In order to attain the isite dogree of accuracy it is convenicnt to regard the pear as ] ) eing 1uilt up in an artificial manner .
I construct an ellipsoid similar to and concentric with the critical therefore itself possessing the same character .
The size of the ellipsoid , which I call , is undefined ; is subject only to the condition that it shall Iarge enough to enclose the whole .
The between the pear being called I supposc bhe pear to consist of posive density throughout and negative density throughout R. The lost cnergy of the consists of that of with itself , say 21JJ ; of with which is filled with negave density , say of itseIf , say .
This last ibntio ( which had )affled me ) mnst be broken several parts .
If we gin J to be intersected by a tmily of orthogonal curves , and if we suppose for the moment that the region is filled lvith positive matter , we may further imagine the matter lying inside any orthogonal tube to be transl ) orted along the tube , and deposited on the surface of in the form of a concentration of positive surface In the actual system is filled negative density , and we may clearly add to this two cmd opposite surface densities and on J. The matter lying in the region It may then be regarded as consisting of negative surface density , together a double system , namely negative volums density , conjoined with e and opposite surface density .
This double system , say , is therefore The lost cnergy may now ) considered as consisting of three parts , first , the energy of with itself , say .l , ; secondly , that of with itseIf , say DD ; and thirdly of with D. This third item is obviously equal to , and therefore is equal to .
Thus , the gravitational lost energy of the , may ) written symbolically\mdash ; In this discussion no attention has as yet been paid to the rotation , but fortunately it ppens t the introduction of this coYlsideration actually simplifies the problem , if we suppose , JJ and JR to mean the lost energies of with itself and with on the supposibion that the mass is rotating with the angular velocity of the critical Jacobian , the formuIae become much more tractable than otherlvise have been the case .
The incIusion of part of the tngular velocity in this part of tho function only leaves outstanding the exccss of the kinetic energ .
of 1902 .
] a of .
181 the bove the kinetic energy which it would have if it rotated with the angular velocity of the critical Jacobian .
If de1lotes the latter angular velocity , and the actual angular velocity of the pear ; if Aj , .
denote the moments of ertia of of considered as filled with positive density , we The -ordinates of points are determined by reference to the ellipsoid which enveIopes the whole pear .
The size of is indeterminate : therefore the formulae must involve ftrbitrary coustant expressive of the size of J. But the final result for cannot in any depend on the sizs of the ellipsoid which is chosen the basis measurement , and therefore the arbitrary constant mnst nltimately disappear .
Hence it is justifiable to it as zero from the beginning , and we may use the formula for the internal vity t the investigation .
Although the constant pressive of the size of is put equal to zero\mdash ; which means that the pear is really partly protuborant beyond the ellipsoid\mdash ; yet there is a considerable amount of convenience in continuing to discuss the subject as though the ellipsoid completely enveloped the pear .
When an ellipsoid is deformed by an harmonic inequality , the volume of the deformed body is only equal to that of the ellipsoid , to the first order of small quantities .
In the case of the , all the inequalities , excepting the third zonal one , are of the second order , and as far as concerns them the volumes of and of the pear are the same .
But it is otherwise as regards the third zonal harmonic term , and the first task is to find the volume of such an inequality as far as When this is done , we can express the volume of in terms of that of the pear , which is of course a constant .
By aid of ellipsoidal harmonic analysis we may now express the first four terms of in terms of the mass of the pear and of certain iite integrals which depend on the shape of critical Jacobian ellipsoid .
The energy DD presents much more difficulty , and it is especially in this that M. Poincare 's insight skill have been shown .
The system consists of a layer of tive volume density coated on its outer surface with a layer of surface density of equal opposite mass .
His procedure virtually amolmts to arding this system consisting of an infinite number of magnetic layers , whose cnergy evaluated and summed .
The rednction of this part of the onergy to calculable forms is not very simplc .
* A term depending on the shift of ) of inertia proves to be negligible .
Compare with M. Poincal. 's treatment of samo point , ' Phil. rans , ' , p. 352 .
18 Prof. H. .
Stabiliiy of Pear-shaped [ June 19 , The moment of inertia of the pear presents but little difficulty , since it only invoIves those harmonic inequalities of which are expressible by harmo1lics of the second degree .
On multiplying the moment of inertia by obtain the last contribution to ession for E. The portion of indepcndent of ) cannot involve , since vallishing of the coefficient of that term is the condition whence the Jacobiau ellipsoid was detcrmined .
If denotes the coefficient of any harmonic inequality other than the third zonal one , this portion of is found to contain terms in .
The coefficient of consists of a constant term terms in , where thess 's denote the coefficients of the second zonal and harmonics .
If refers to any har1nonic of odd degree , the coefficient of the corresponding term in vanishes .
If , then , we make stationarJ7 for variations of the coefficient of any odd harmonic , that coefficient is seen to vanish .
Hence it follows that the expression for the peal cannot involve any odd harmonic other than the tbird zonal one .
ditions of symmctry also negative the existence of even harmonics of the type , and of even harmonics of the cosine type but of odd rank .
On ting to zero the variations of Fi for the remaimng excepting , we at once obtain their in terms of Equating zero the variations for three tions , hich g as multiples of It seems mnecessitry to explain here methods adopted for lncing the analytical resuIts to numbers ; it suffice to say that the task lvas very The harmonic terms included in the computation were those of degree 2 , and ranks , 2 ; of degree 4 , and ranks , 2 , 4 ; and of degree 6 , and ranks , 2 , 4 .
The sixth sectorial harmonic would certainly have proved negligible .
The expression for was found in the form of a fraction , of which the denominator is determinate , and the numerator is the sum of an series .
terms of this series were compnted , namely , constant term and the contributions of the eight harmonics ave cnumerated .
The result shows that the square of the angular velocity of the pear is less than that of the Jacobian in about the proportion to 1 .
On the other hand , the angular momentum is greater in about the portion of to 1 .
If this last result were based on rigorous summation of the infinite series , it would absolutely prove the stability of the pear .
The inclusion of the uncomputed residue of the series undoubtedly tend in the direction of reducing the coefficient given above in round numbers as , and if it were to reduce it to a negative quantity we should that the pear is unstable ftcr all .
1902 .
] Figure of Equilibrium of a Rotating JTass of Liquid .
; The apparently rapid convergence of the series seemed to render such a reversal of the result almost incredible .
In order , however , to feel yet more sure , I made a rough estimate of the contribution of the eighth zonal harmonic , and found that it would only amount to part of that critical total which would just show the peal ' to be unstable .
Since the convergency of the series is obviously rapid , I it as proved , but by something short of absolute aIgebraic proof , that the pear is stal ) The numbers obtained in the course of the vork afford the means of giving a second approximation to the form of the pear , and the result is shown in figures , drawn the largest value of hich seemed consistent with a fair degree of approximation .
I originally called the figure " " pear-shaped\ldquo ; because Poincar 's conjectural sketch in the 'Acta Iathematica ' was very like a pear .
In the first approximation , in my former paper , the to a pear was not strikin it needs some imagination to see the pear-shape in the new figures ; ) a distinctive name is so convenient that we may as well continue to call it by that name .
The effects of the new are almost entir.ely concentrated at the ends .
They tend to angment the otuberance of the stalk end .
and to diminish the depression at the end so much as nearly to fill it up .
Over the greater part of the figul.e the depressions protubelances are less conspicnous ) they were .
I think it is hardly too much to say that in a well-developed ' ' \ldquo ; the Jacobian ellipsoid has nearly regained its primitive figure , but thnt it has shot forth a protubsrance at one end .
A consideration of the figures and of a conjectural extension of them almost reminds one of some such phenomenon as the protrusion of a filament of protoplasm from a mass of living matter .
Notwithstanding the warning of M. Poincare as to the dangel of applying these results to heterogeneous- masses and thence to cosmogony , I cannot restrain myself $rom.ioining him in seeing in this almost life-like cotulterto at least one form of the birth of donble stars , planets , an satcllites .

