]\gt ; On a Rotating Mass of Fluid .
On Jacobi 's Figure of Equilibrium for a Rotatin Mass of Fluid By G. H. DARWIN , M.A. , LL. D. , F.R.S. , Fellow of Trinity College and Plumian Professor in the sity of Cambridge .
Received October 12 , 1886 .
I am not aware that any numerical values have ever been determined for the axes of the ellipsoids , which are figures of equilibrium of a rotating mass of fluid .
* In the following paper the problem is treated from the point of view necessary for reducing the formulae to a condition for computation , and a table of numerical results is added .
Let be the semi-axes of a homogeneous ellipsoid of unit density ; let the origin be at the centre and the axes of be in the directions Then if we put\mdash ; , and ' ( 1 ) it is known that the potential of the ellipsoid at an internal point is given by\mdash ; . .
2 Now let us introduce a new notation , and let * The following list of papers bearing on this subject is principally taken from a report to the British Association , 1882 , by W. M. Hicks:\mdash ; Jacobi , ' Acad. des Sciences , ' 1834 ; Liouville , ' Journ. cole Polytech vol. p. 289 ; Ivory , ' Phil. Trans 1838 , Pt. I , p. 57 ; Pontecoulant , ' Syst. du Monde , ' vol. ii .
The preceding proofs of the theorem , and in more detail we have:\mdash ; C. O. Meyer , 'Crelle , ' vol. xxiv , p. 44 ; Liouville , 'Liouville 's Journ vol. p. 241 ; a remarkable paper by Dirichlet and Dedekind , 'Borchardt 's Journ vol.lviu , p. 181 .
and 217 ; Riemann , ' Abh .
K. Ges .
Wiss .
Gottingen , ' vol. , 1860 , p. 3 ; Brioschi , 'Borchardt 's Journ vol. lix , p. 63 ; Padova , 'Ann .
della Sc. Norm. .
Pisa , ' 1868-9 ( being Dirlchlet and Riemann 's work with additions ) ; Greenhill , 'Proc .
Camb .
Phil. Soc vol. , p. 233 and vol. iv , p. 4 ; Lipschitz , ' Borch .
Journ vol. lxxviu , p. 245 ; Hagen , ' Schlomilch Zeitsch .
Math vol. xxiv , p. 104 ; Betti , 'Ann .
di Matem .
' vol. , p. 173(1881 ) ; Thomson and Tait 's ' Not .
Phil. ' , Part very important by-Poincare , ' Acta Mathem 7 , 3 and 4 ( 1885 ) .
Thomson and Tait 's 'Nat .
Phil. ' ( 1883 ) S494 , .
The form in which the given is slightly different from that in ( 8 ) , ( 11 ) , ( 15 ) of SS494 , 1886 .
] EquiliMu for a Rotating Mass of Fluid .
In the present case .
Thus ( 8 ) and enable us to complete the required transformation to elliptic in .
tegrals of then in the expression oos , we have V- : Now snppose the ellipsoid to be rotating the axis of with an angular velocity , and let us choose the axes , a , a and the angular velocity , so that the surface be a SUl.face of equilibrium .
For this purpose constant , must be identical with Now in 10 we have in the form , . . . . .
( 11 ) whence ) Hence or ( 12 ) There are two kinds of solutions of these equations ( 12 ) .
First , since ( 13 ) it is obvious that vanishes when is satisfied by That is to say there is a solution of the problem which makes Thus there is solution which gives us an ellipsoid of revolution .
When , we have also , and , .
Therefore .
( 14 ) and the eccentricity of bhe ellipsoid of revolution is To find the other solution when is not zero , we have by comparison between ( 10 ) and ( 11 ) , .
( 15 ) Hence the first of ( 12 ) gives *Comp with Thomson and Tait 's .
Not .
Phil S771 , ( 8 ) or amy other work which gives a solution of the problem .
8@6 .
] Equihbrium for a Rotating Mass of FluiA or . . . .
In order to adapt this for computation , we may introduce the auxiliary angles defined by , . . . .
( 17 ) and the equation becomes . . .
( 18 ) The second of ( 12 ) gives , whence ' ( 19 ) Some of the subsequent computations were , however , actually made from a formula deduced from the third of ( 12 ) , which leads to . . .
( 20 ) By subtracting from ( 19 ) we can deduce ( 16 ) ; hence it follows that ( 19 ) and ( 20 ) lead to identical results .
Most of the subsequent lesults were computed from both ( 19 ) and 20 ) , thus verifying the solution of ( 18 ) .
The formuIae ( 18 ) and ( 19 ) are suitable for finding the soIution , exceI ) when is small or nearly , when the elliptic integrals become awkward to use .
I have , therefore , found approximate formulae for these cases , but as the algebraic process necessary to establish them is somewhat tedious , the details are given in a note .
* *Approximate Solntions of the Problem .
From ( 7 ) we have . . . . . .
( ) It is easier to develop the equations when written in the forms ( b ) and ( c ) han when we work from the elliptic integrals .
Write for brevity .
Firsf , when is small\mdash ; The followin definite integrak are lired : .
( ) . . . . . . .
( ) From ( d ) and from , and . . . .
( ) Again , , and Equilibrium for a Rotating Mass of Fluid .
325 Approximate Solutions of the Problem\mdash ; continued . . . . . .
( ) The equatiou ( b ) is , whence , equating ( g ) and ( h ) , If be we have This reduces to , . . . . .
of which the solution is as stated in the text .
Now and with the above value of is equal to Ako , with this value of the coefflcient of is ; so that or , which is the equation ( 21 ) of the text .
Again Therefore Now Hence from ( c ) and ( j ) which is the equation ( 22 ) of the text .
Secondly , let be nearly unity and Then we require the following deiinite integraIs:\mdash ; . . . .
( ) . . . . . . . . . .
( ) Now . .
By ( k ) and Now therefore therefore . . . .
( ) 1886 .
] quilibrium for a Rotating Mass of Fluid .
pproximate Solutians of the Problem\mdash ; continued .
Again , . .
( n ) The equation ( b ) for determining the axes is Hence equating ( m ) and ( n ) and dividing by we have . . .
which is the equation ( 23 ) in the text .
The equation ( c ) for may be written Now VOL. which is the equation ( 9 ) , S778 ' , of that work .
The ellipsoid of tion of which the icity is belongs to the revolutional series of fig.ures of equilibrium , and is the starting point of the Jacobian series of .
As shown by Sir William Thomson , it is the flattest revolutional figure which is dynamically stable .
The Jacobian figures of equilibrium are initiaJly stable , and as stated by M. Poincare , *there is for this value of crossing point of the two series , and an exchange of stabilities .
If be small , it appears that is given by . . .
( 21 ) and by ( 22 ) Approximate Solutions of the .
Thelcfore is the equation ( 24 ) in the text .
" " Sir l'Equilibre d'une Mass de Fluide , \amp ; c ' ta Mathematica , ' 7 , 1886 .
* 1886 .
] Equilibrium for a Rotating Mass of These formulae are , it must be admitted , of but little use , since it vould be necessary to take in higher powers of to obtain results for a variation of of more than Tf be near , so that is small , the oximate equation between and is - . . . .
( 23 ) And given by ( 24 ) We shall return later to a modification of ) will be applicable to very long ] ipsoids of equilibrium .
Besides the angular velocity and the axes of the ellipsoid , the other important functions are the mome1lt of momentum , the kinetic energy of rotation , and the intrinsic energy of the mass .
In order to express tl1ese numel.ically we must adopt a unit of , and it be concnient to take , where Let be the density of the fluid which has hitherto been treated its ullity , and let be the moment of momentum and kinctic energy , then The function ( 25 ) is the quantity which will be tabnlctted .
so ( 26 ) integrated throughout the ellipsoid .
This will be denoted by , or , so that ' will be positive .
Now , and if we denote by , the principal moments of inertia of the ellipsoid , we have and similarly , Also .
, Hence Bnt if we take the values of given in ( 15 ) , and note that it easily follows that Hence Therefore For a sphere becomes infinitely small , and becomes equaI to so that .
Thus .
Therefore the exhaustion of 1886 .
] Equitibrium for a Rotating Mass of Fluid .
nergy of a sphere of radius a is ; which is the known result .
For an ellipsoid of revolution , and , and ; so that The function ( 27 ) is the quantity tabulated below .
It seemed preerable to tabulate a positive quantity , and it is on this account that the intrinsic energy corresponding to the infinitely long ellipsoid is entered as unity .
Having now obtained all the necessary formulae , we may proceed to consider the solution of the problem .
We have to solve , ( 28 ) where and The axes of the ellipsoid are If are the eccentricities of the sections through , ab respectively , we have .
( 30 ) Having obtained the solution , we have to compute ( 31 ) we next compute and and from the formulae ( 25 ) , ( 26 ) , .
The functions and are tabulated in Table IX of the second volume of Legendre 's ' Traite des Fonctions Elliptiques , ' in a table of double entry for and for each degree .
The solution of ( 28 ) by trial and error was laborious , as it was necessary to work with all the accuracy attainable with logarithms of seven figures .
The method adopted was to choose an arbitrary value of , and * As stated above , some of the computations were actually made from the formula ( 20 ) .
Prof. G. H. Darwin .
On 's Figure of [ No 1886 .
] Equihbrium for a Rotating Mass of Fluid .
then by trial and error to find two values of one degree apart , one of which made the left-hand side of ( 28 ) positive , and the other negative .
The smallest value of is but after that value integral degrees for were always chosen .
The solntions for and could not be found very exactly from the elliptic integrals with logarithms of only seven figures , the solutions were confirmed by the approximate ] ( 21 ) and ( 22 ) .
The solution for was confirmed by the approximate formulae and ( 24 ) , and that for was only computed therefrom , since when the approximate formula gave nearly identical results with the exact one .
The solution obtained is embodied in the table opposite .
The first three columns give the auxiliary angles , from which the remaining results are computed .
As a graphical resuIt is much more intelligible than a numerical one , I have given two figures , showing the three sections in two cases , namely , where , and .
For these figures a is taken as 2 cm .
, so that the volume of fluid is cubic cm .
334 Prof. G. H. Darwin .
On It will be noticed that the longer It is interesting to observe that while diminishes , the moment of momenbu long llipsoids are very nearly ellipsoi perpendicular to that of rotation .
Th and is not much flattened .
The mosi remarkable point is that ] energy when is about 3 , or when about five times its diameter .
mum of kinetic energy , the total : length of the ellipsoid .
The kinetic energy is the product always increases , and the other of which obvious that it must have a maximu quite unforeseen , and it seems worth for the case of the long ellipsoids .
Th ] parameter , or the length of the elli ] From the table we see that in the 1886 .
] Equilibrium for a Rotating Mass of Fluid .
equal to , and that becomes very nearly equal to .
Hence we may put , and Thus , approximately , : and The axes of the ellipsoid are Now if in formula we only retain the higher powers of we have But Therefore writing , so that , we have If we put , this formula gives .
The full value in the preceding tables was ; thus even with so short an ellipsoid as this , the resuIts agree within 4 per cent. With rougher approximation we have of which the limit , when is large , is zero .
For the moment of momentum we have Since we have If we like may express these several results in erms of the minor and major axes of the ellipsoid , for , and therefore .
If we like may express these several results in erms of the minor and major axes of the ellipsoid , for , and therefore .
If we like may express these several results in erms of the minor and major axes of the ellipsoid , for , and therefore .
If we like may express these several results in erms of the minor and major axes of the ellipsoid , for , and therefore .
If we like may express these several results in erms of the minor and major axes of the ellipsoid , for , and therefore .
If we like may express these several results in erms of the minor and major axes of the ellipsoid , for , and therefore .

