]\gt ; [ 379 ] XIII .
On Figures of Equihbrium of Rotating Masses of Elluid .
G. H. DARWIN , M.A. , LLD .
, F.R.S. , Fellow of Trinity College , and Plumian Professor in the University of Cambridge .
Received April \mdash ; Read June 16 , 1887 .
[ PLATES 22 , 23 .
] IN a previous paper*I remarked that there might be reason to suppose that the earliest form of a satellite might not be annular .
Whether or not the present investigation does actually help us to understand the working of the nebular hypothesis , the idea there alluded to was the existence of a dumb-bell shaped figure of equilibrium , such as is shown in the figures at the end of this paper .
These figures were already drawn when a paper by M. POINCARfi appeared , in which , amongst other things , a similar concIusion was arrived at .
My paper was accordingly kept back in order that an attempt might be made to apply the important principles enounced by him to this mode of treatment of the problem .
The results of that attempt are , for reasons explained below , given in the Appendix .
The subject of figures of equilibrium of rotating masses of fluid is here considered ffom a point of view so wholly different from that of M. POINCARE that , standing his priority and the greater completeness of his work , it stiU appears worth while to present this paper .
The method of treatment here employed is simple of conception ; but it is unfortunate that , to carry out the idea , a very formidable array of analysis is necessary .
In the last section a summary will be found of the principal conclusions , in which analysis is avoided .
S1 .
Formutoe of Sphericed Harmonic Let there be two sets of rectangular as shown fig. 1 ; and let be sured from to , whilst is measured from ; let ; and let Then . . .
( 1 ) Let , denote the solid zonal hannonics of degree of the coordinates , reSpectively .
Now we shall require to express the solid zonal and certain tesseral harmonics of * ' Phil. Trans Part II .
, 1881 , p. 534 . .
PROA'ESSOII G. H. DARWIN ON FIGURES OF no , ative rees with respect to the origin as solid .
zonal and tesseral harmonioe positive frees with respect to the origin , and vlce versed ; moreover , the will have to be applied to a sphere of radius with centre , and to a sphere radius with centre .
This last clause is introduced in order to explain the duction of the symbols , in this place .
The formulae required will be called " " transference formulae because they are to used in the origin from one point to the other .
The obvious symmetry of our axes is such that every transference formula to has its exact counterpart for transference from to ; thus a second symmetrical formula with capital and small letters interchanged will generally be unwritten .
When necessary , , will be written for co-latitude and longitude with regard to ; and , for the same with respect to Then , since we have the usual expansion in zonal harmonics ( 2 ) The usual formula for the derivation of the zonal harmonic of negative degree from is ( 3 ) Hence , on differentiating ( 2 ) times with respect to , or , which is the same with respect , we have , from ( 3 ) , But Hence ILIBRIUM OF ROTATING MASSES OF FLUID .
this formula , it will be noted that , if is less than , the term hence the summation runs from to ; it is therefore better to for and we thus obtain .
( 4 ) is the first transference formula by which the solid zonal harmonic of degree with respect to is expressed as a series of solid harmonics of positive ith respect to .
The formula ( 4 ) includes ( 2 ) as the particular case where The right-hand side of ( 4 ) is convergent for less than .
A similar formula , overgent for greater than , is easily obtainable , but with this we shall not ocern ourselves .
It remains to find the transference formula for certain tesseral harmoni If we put , ( 5 ) general expression for the zonal harmonic is , ( 6 ) the summation extends from to or .
From ( 6 ) we have .
( 7 ) since , we have .
( 8 ) .
( 9 ) Subtracting ( 9 ) from , and simplifying the difference , we have ; ( 10 ) the last transformation being derived from ( 7 ) with in place of in place of Differentiate ( 10 ) with respect to , and notice that , and we have Then , with in place of Now by ( 10 ) : Differentiating again , by ( 1 !
) , or : : ( 12 But since , it follows that , in operatin , g ' a involving and only in the form , and AIso so that Now let us put ( 13 ) Then and therefore ( 12 ) may .ritten .
: ; ( 14 ) EQUILIBRIUM OF ROTATING MASSES Ull ) .
These expressions in ( 14 ) are obviously solid tesseral harmonics .
The transference formula required is for By the formula ( 4 ) we have operating on both sides by , and appIying ( , we have .
( 15 ) Now the general formula ( 6 ) for the zonal harmonic shows us that is zero when , 1 , 2 , 3 , and hence vanishes for the same of .
Thus the summation in ( 15 ) is from to , or , if we write for , from to 2 .
Hence ( 15 ) gives . . .
( 16 ) This is the second transference formula required .
We observe that the transference of a negative zonal harmonic gives .
us positive zonals , and that tesseral harmonics of the type give us harmonics of the type S2 .
The Mutual Influence of two Spheres without Rotation .
Imagine two approximately spherical masses of fluid of unit density , with their centres at the origins and respectively , and with mean radii and respectively .
We shall find that.each exercises on the other certain forces , one part of which has a solid zonal harmonic of the first degree as potential .
This part of the fo rce must remain essentially unbalanced in the supposed system , but we shall see hereafter that it is balanced by the rotation to be afterwards imposed on the system .
Meanwhile it will be supposed that it is annulled in some way , and we shall content ourselves with finding the mutual influence of the spheroids , and the outstanding term of the first degree of harmonics .
Let us assume that the equations , referred to our two origins , of the surfaces of the two spheroids , when they mutually perturb one another , are ( 17 ) The 's and 's are unknown coefficients , to be determined .
We have now to find the potential at any point in space .
The mass of the spheroid is , and its potential is The potential due to the departure from sphericity , represented by the term in the of ( 17 ) is This is written in a form convenient for passing to the case of .
It may be written in the form when it is in a suitable form for application of the transference formula ( 4 ) .
We shall now introduce two new symbols , namely , Then ( 19 ) may be written ; and , of course , the similar potential with the other origin is . .
The whole potential at any point of space consists of the potentials of the spheres and of the inequalities on each .
The potential of the inequalities of the may be written in the form ( 18 ) , and of sphere in the form ( 21 ) .
Thus the whole potential is The first line of ( 22 ) refers to origin , the second to origin , and to this half the transference formula ( 4 ) must be applied .
Now apply ( 2 ) to the first term of the second line , and ( 4 ) to one term of in the second term , and we have and EQUILIBRIUM OF ROTATING MASSES OF FLUID .
the second line of ( 22 ) when transferred is Then and together constitute the potential now entirely referred to origin We want to choose the 's and 's so that each spheroid may be a level surface , save as to the outstanding term of the first degree .
In order that ( 17 ) may be a level surface , when we substitute for its value ( 17 ) in ( 22 ) , the whole potential must be constant .
In effecting this substitution , we may put in the small terms , but in we must give it the full value ( 17 ) .
The constancy of the potential is secured by making the coefficient of each harmonic term vanish separately\mdash ; excepting the first harmonic , which remains outstanding by supposition .
We may consider each harmonic term by itself As far as concerns the term involving we have , from and , as the value of the potential , and the value of which must make this constant is but in the small terms inside we may put simply .
Make , therefore , the substitution , and equate the coefficient of to zero .
On dividing that coefficient by , we find Therefore , ( 23 ) and , by symmetry , .
( 24 ) Multiplying both sides of ( 24 ) by the coefficient of in ( 23 ) , we have .
\mdash ; A. Performing on both sides , and substituting from ( 23 ) , .
( 25 ) Now let And ( 25 ) may be written .
( 27 ) By imparting to all integral values from 2 upwards , we get a system of linear equations for the determination of the , and it will appear below that as many of them may be found numerically as may be desired .
We now have to consider the series ( 26 ) .
Let and denote the operations or by Consider the function .
Now Therefore Thus .
( 28 ) Next consider the function .
As before , and EQUILIBRIUM OF ROTATING lfASSES OF FLUID .
We must now develop the symbolical sums of the series in ( 28 ) and ( 29 ) .
The following theorems are obvious:\mdash ; Then , by their aid , we have from LEIBNITZ'S theorem\mdash ; , in which we interpret as .
Thus , ( 30 ) with .
Again Hence .
( 31 ) .
In ( 30 ) and ( 31 ) the infinite series are replaced by finite series .
From the form of the series it is obvious that the result must be symmetrical with respect to and , so that , but this is not obvious on the face of the formula ( 31 ) .
We shall find , therefore , the symmetrical form of ( 31 ) for the first few terms .
If , we obviously have First .
The second term arises from , and .
The two corresponding values of ( 31 ) will be found to add together , and we get Second The third term arises from , and we find\mdash ; Third A symmetrical form for further terms may be obtained by writing ( 31 ) first with before and then with before , and taking half the sum of the two results .
In computing these coefficients it is a useful check to compute from both unsymmetrical forms , when the identity of results verifies the computation .
The following Tables have been computed from ( 30 ) and ( 31 ) .
The numbers are the coefficients of the quantities at the heads of the columns for the values of and written in the first column .
The series is terminable with , and the series is terminable with In the coefficients have only been computed as far as , so that the which is given completely is 2 , 4 , ; however , with such values of as we require , the series are carried far enough to give numerical results with sufficient accuracy .
TABLE of .
EQUILIBRIUM OF ROTATING MASSES OF FLUID .
must now go back and determine the value of the outstanding potential of the gree of harmonics , which will be annulled when rotation is imposed on the The potential is given in and contributes nothing , and gives us , for if we call the outstanding potential of the first degree , when referred to two origins respectively , we have ( 32 ) .
S3 .
The Potential due to Rotation .
Intermediate between the two origins and take a third , and take the axes of and parallel to those of and , and that of identical with that of .
Let Then suppose that the system of the two spheroids is in uniform rotation about the xes of with an angular velocity The potential of the centrifugal forces is given by .
( 33 ) ( 34 ) hen , remembering that nd if we put have .
( 35 ) Similarly the rotation potential , when developed with reference to the other , is .
( 36 ) The last terms of ( S5 ) and ( 36 ) are constants , and the term in , and that in are symmetrical about each origin , and so the corresponding forces can produce departure from sphericity in either mass ; thus these terms may be dropped .
we have in ( 35 ) and ( 36 ) the outstanding potentials and will be annulled by other similar terms , and so need not be considered now .
left , therefore , with the terms in and , or in and .
The is a harmonic , the a zonal , and it will be convenient to treat them separately .
shall begin with the zonal term .
S4 .
due to the Zonal Harmonic Rotational Term .
The potential whose effects we are to consider is or , according to origin which we are considering .
If an isolated spheroid of fluid of unit density be rotating with angular velocity the ellipticity of the spheroid is ; therefore we put Let us assume , for the equations to the two spheroids , where , are unknown coefficients which are to be determined .
We now have to determine the potentials at any point of the inequalities ( 38 ) on the two spheroids .
The potential of the inequality in the of ( 38 ) is The similar inequality in the second of ( 38 ) gives us The term in in the first of ( 38 ) gives us , as in S2 , EQUILIBRIUM OF ROTATING MASSES OF FLUID .
term in in the second of ( 38 ) gives us , as in S2 , .
( 42 ) The potential due to rotation is or , being the second term of ( 35 ) or 36 ) ; this term we find it convenient to write ( 43 ) The sums of the several terms ( 39 ) , ( 40 ) , ( 41 ) , ( 42 ) , and ( 43 ) are to be regarded as potential of perturbing forces by which the spheroid or the spheroid is disurbed , and the arbitrary constants and are to be so chosen that each may be a figure of equilibrium .
We may consider the spheroid by itself , and the solution for it will afford us the olution for the spheroid by symmetry .
In order to find the disturbance , the formulae 40 ) and ( 42 ) must be transferred .
Now by ( 4 ) , with ( 40 ) nd again , by ( 4 ) , .
( 42 ) Then ( 39 ) , ( 40 ) , the sum of ( 41 ) from to , the sum of ( 42 ) from to , and ( 43 ) together constitute the disturbing potential , all now referred to the rigin In order to find the disturbance of the spheroid , we add the perturbing potential , give its value ( 38 ) in this term , put in the perturbing potential , and make the whole potential constant by equating to zero the coefficient of each hannonic term .
We will begin by putting , and considering only the perturbing otentials ( 89 ) and ( 43 ) .
We have then , for the coefficient of Now , with the value of in ( 37 ) , and - -L the coefficient of vanishes , and the term in in ( 38 ) has been properly to satisfy the perturbing potentials ( 39 ) and ( 43 ) .
Following the similar process with the remaining terms of ( 38 ) , and equating zero the coefficient of , we have from ( 40 ) , ( 41 ) , and ( 42 ) , ; whence By symmetry , the condition that the spheroid may be a level surface is Multiply both sides of ( 45 ) by ' and perform on the whole substitute from ( 44 ) ; and we have Introducing the notation ( 26 ) for the series involved in ( 46 ) , we have Each value of gives a similar equation , and there is a similar series of with small and large letters interchanged .
Now put and ( 47 ) becomes EQUILIBRIUM OF ROTATING MASSES OF FLUID .
We attribute to in ( 49 ) all values from to 2 , and thus find a series of equations the .
A similar series of equations holds for the We must now find the outstanding potential of the first degree of harmonics .
No term exists in ( 39 ) , ( 41 ) , ( 43 ) , but it al.ises entirely out of ( 40 ) and ( 42 ) .
If we write for the outstanding potential , we have clearly ; , ( 50 ) by symmetry , .
( 51 ) S5 .
due to the Harmonic Rotational Term .
In ( 35 ) and ( 36 ) we have found this term to be \mdash ; or \mdash ; We have already observed that , if the operation or be performed on result vanishes when , 2 , 3 .
Now , by ( 6 ) , , and , since , it follows that , and . . . .
( 52 ) Hence the sectorial rotational term is\mdash ; or \mdash ; ; this potential is of the second order of sectorial harmonics .
Now , with as defined in ( 37 ) , let us assume as the equations to the two surfaces , ( 53 ) We have now to determine the potentials of the inequalities on the two spheroids expressed by ( 53 ) .
The potential of the inequality \mdash ; in the first of ( 53 ) is The potential of the similar inequality in the second of ( 53 ) is The term in in the first of ( 53 ) gives us The term in in the second of ( 53 ) giyes us Lastly , the sectorial term itself is The sums of the several terms ( 54 ) , ( 55 ) , ( 56 ) , ( 57 ) , and ( 58 ) are to be regarded as the potential of perturbing forces by which the spheroid , or the spheroid , is disturbed , and the arbitrary constants , are to be so chosen that they may each be figures of equilibrium .
We may consider the spheroid by itself , the solution for it will afford the solution for the spheroid by symmetry .
In order to find the disturbance , the formulae ( 55 ) and ( 67 ) must be transferred .
For this purpose we require the second transference formlae .
By ( 16 ) , with , we have for ( 55 ) ( 55 ) And by ( 16 ) we have for ( 57 ) ( 57 ) Then ( 54 ) , ( 55 ) , the sum of ( 56 ) from to , the sum of ( 57 ) from to , and ( 68 ) together constitute the disturbing potential , all now referred to the origin EQUILIBRIUM OF ROTATING MASSES OF FLUID .
In order to find the disturbance of the spheroid , we add the perturbing potential , give its value ( 53 ) in this tenn , put in the perturbing potential , make the whole potential constant by equating to zero the coefficients of each harmonic term .
We will begin by putting , and considering only the perturbing otentials ( 54 ) and ( 58 ) .
We have then , for the coefficient of Now , with the value of in ( 37 ) , , and Hence the coefficient of vanishes , and the term in ( 53 ) has been properly chosen to satisfy the perturbing potentials ( 54 ) and ( 58 ) .
Following the similar process with the remaining terms of ( 53 ) , and equating to zero the coefficient of , we have , from , ( 56 ) , , or .
( 59 ) By symmetry the condition that the spheroid may be a level surface is .
( 60 ) Multiply both sides of ( 60 ) by , and on the whole , and substitute from ( 59 ) , and we have .
( 61 ) Now let us write so that ( 61 ) may be written .
( 63 ) Next put and ( 63 ) becomes We attribute to in ( 65 ) all values from to 2 , and thus find a series of for the .
A similar series of equations holds for the We now have to sum the series ( 62 ) .
Consider the function Hence Next Hence The differential in ( 67 ) must now be evaluated .
We have , by LEIBNITZ'S theorem , . .
Substituting in ( 67 ) , we have EQUILIBRIUM OF ROTATING MASSES OF FLUID .
6 ) and ( 68 ) , give the values of and S6 .
Determination of the Velocity of the System .
The angular velocity of the system must now be determined in such a way as to annul the outstanding potential of the first degree of harmonics .
Referring to origin , we have from ( 35 ) directly from the rotation potential ; the remaining terms are , since the sectorial harmonic term does not contribute anything .
Thus , taking from ( 32 ) , and from ( 50 ) , we get for the potential Equatin this to zero , .
( 69 ) And , by symmetry , Add ( 70 ) to ( 69 ) , note that , and , and solve for we have Now let denote the factor by which is multiplied in ( 71 ) , Then , if the two masses were particles , would be zero , and ( 71 ) would simply the usual formula connecting masses , mean motion , and mean distance in a orbit .
Hence is an augmenting factor by which the value of the square the angular velocity must be multiplied if it be derived from the law of the time of two particles revolving about one another .
, in fact , gives the correctioq KEPLER'S law for the non-sphericity of the masses .
This completes the solution of the problem , for we have determined the velocity in such a way as to justify the neglect of the harmonic terms of the degree in SS2 and 4 .
S7 .
Solution of the Problem .
We may now collect from the preceding paragraphs the complete solution of problem .
In ( 38 ) and we have found that there are terms in as Now , and ; hence and these terms are therefore equal to .
We note that , and that is the ratio written in works on the figure of the Earth .
Then , from ( 17 ) , ( 38 ) , ( 53 ) , equations to the two surfaces are EQUILIBRIUM OF ROTATING MASSES OF FLUID .
( 72 ) From ( 27 ) , ( 49 ) , ( 65 ) , we see that . . . .
are to be found by solving the equations resulting from all values of from 2 to infinity in the foUowing:\mdash ; ( 7S ) and symmetrical systems of equations for obtaining the , and With the values found by the solution of t.hese equations we then evaluate by formula ( 71 ) ; and we have .
( 74 ) We are now enabled to find the 's and 's by the formulae ( 48 ) and ( 64 ) , viz. , ( 75 ) and the symmetrical forms give us the 's and Having thus evaluated all the auxiliary constants , ( 72 ) gives the solution of the problem .
It is well known that is the ellipticity of a single homogeneous mass of fluid rotating with angular velocity .
Hence the first terms of ( 72 ) simply denote the ellipticity due to rotation in each of the masses , as if the other did not exist .
Now the rigorous solution for the form of equilibrium of a rotating mass of fluid is an ellipsoid of revolution with eccentricity , the value of being given by the solution of ( 76 ) $ See , for example , THONSON and TAIT'S ' Natural Philosophy ' , S771 , with Hence it will undoubtedly be more correct to construct the surface , of which the equation is ( 72 ) , by regarding the part of under the symbo12 as the correction to the radius-vector of an ellipsoid of revolution with eccentricity determined by ( 76 ) , where is found from ( 74 ) .
S 8 .
of the Solution .
The principal object of the preceding investigation is to trace the forms of the two masses when they approach to close proximity ; we shall thus be able to determine the forms when they are on the point of coalescing into a single mass , and shall finally obtain at least an approximate figure of the single mass .
For this purpose we require to push the approximation by spherical harmonic analysis as far as it will bear .
We shall below endeavour to estimate the degree of departure from correctness involved by the use of this analysis .
The results will , therefore , be worked out numerically for such values of as bring the two masses close together , and it will appear that the largest value of assumed for numerical solution is such that the surfaces cross ; in this case the reality will be a single mass of a shape which it will be sible to draw with tolerable accuracy .
The computations are facilitated if , instead of assuming to be an exact multiple of , we take a multiple of ; that is to say , we shall take as an integer , and therefore also an integer .
We shall in the first instance suppose the two masses to be equal .
In the following examples , then , we have , and the two masses assume the same shape .
The computations will be carried through in detail in two cases , viz. , when and when .
The results will also be given for When , and when .
Thus the distances of the centres apart are and of the mean radius respectively .
The numerical details of the two computations may be stated pari , and the numbers applying to will be distinguished by being printed in small type .
In the case of , we have ; but only the final result is given , without intermediate details .
The first step is to compute the values of the several series by means of the Tables in SS 2 and 5 .
The numerical ] are as follows .
EQUILIBRIUM OF ROTATING MASSES OF FLUID .
With these values for the series , we have to compute the coefficients of the systems of simultaneous equations ( 73 ) .
The equations lend themselves more readily to solution if we consider , as the unknowns ead of The results are given the following equations .
CCLXXXYII .
The upper coefficients correspond to the case of ; the lower ones , prin ed small type , to the case of EQUILIBRIUM OF ROTATING MASSES OF FLUID . .
With these values of the 's and , I find the summations , of course , stopping with Applying these in ( 71 ) , we have , when ; or , when whence Thus the angular velocity of the system has been found .
Next we have Introducing this into ( 48 ) and ( 64 ) with the previously found values of the 's and By taking the differences of , we may conclude that and this sixth nonic term will now be included .
It appears from the values of the 's that the harmonics of the type are negligible , excepting the term , and that in that we may neglect the part depending on Now , if denotes the radius-vector due to the rotation , and Sr the increase of radius-vector due to the mutual influence of the two masses , we have ' . .
( 77 ) We next have to consider the values of , the radius-vector of the ellipsoid , rotation .
We might compute from the spherical harmonic formula .
The results so computed will be compared with the others computed as shown The following Table of the angular velocity and corresponding eccentricity equilibrium ellipsoid of revolution is extracted from THOMSON and TAIT'S Philosophy , ' :\mdash ; From this we find by interpolation that , when ; when These , then , are the eccentricities of the ellipsoids whose radius-vector is in two cases The equations to the generating ellipses are for , and for The following are the computed values of for each of , the latitude , small figures written below appertaining to the case of : , 953 , 906 , 875 , 865 Computing from the spherical harmonic formula , I find : , 9692 9230 , 8892 , 8768 EQUILIBRIUM OF RaTATING MASSES OF FLUID .
The greatest discrepancy occurs when and , and the difference between the two results is of either .
It follows , therefore , that in drawing the figures it is not of much importance which results we take .
But , as above remarked , the radius-vectors computed from the true ellipsoidal figure are the more correct .
The formula ( 77 ) for Sr consists of a series of zonal harmonics .
The pole of symmetry of these harmonics lies in the equator of the ellipsoid of revolution defined by , and is that point of each mass which lies nearest to the other .
Then , denoting by co-latitude estimated from this pole , I find that the numerical values of Sr for each of are as follows:\mdash ; These have to be combined with , so as to give the radius-vectors of the mass of fluid along two sections , one perpendicular to the axis of rotation ( which may be called the equatorial section ) , the other through the axis and the two centres ( which may be called the section through the prime meridian ) .
Taking the case of , we add the successive values of Sr to the equatorial value of , viz. , , and thus find the equally-spaced radius-vectors along the equatorial section .
Next we add the successive values of to the corresponding values of , and thus find the equally- spaced radius-vectors along the prime meridian .
The results are as follows:\mdash ; These results apply to the case of ; those for are found in the same way , and are given in the figures referred to below .
also worked othough numerical details aivenWhen istance between tentres iiven b here .
In figs. 2 , 3 , 4 , and 5 ( Plates 22 and 23 ) are exhibited the figures which result from some of computations .
Figs. 2 and 3 refer to , 4 and 5 to that of , and the numerical , given above , make it easy to draw a figure for Since in these cases the masses are equal , the t.wo halves of the figure are images of one another .
The numerical value of each radius-vector is entered on plates ; and other numerical data and explanations are given .
Figs. 2 and 3 correspond to , and here the figures as computed cross another .
The reality must , therefore , be two bulbs joined by a stalk , like a bell .
The dotted lines have been filled in conjecturally , and must show pretty what that single figure , formed by the coalescence of the two masses , must be .
Figs. 4 and 5 show in a similar manner the case of , and here the two are separate , although nearly in contact .
When the shapes presrent characters , but are wider apart .
S9 .
On the Use of Spherical rmonic Analysis Method of Approximation .
Spherical harmonic analysis gives less accuracy as the bodies considered more and more from spheres .
How far , then , do our results present an approach accuracy ?
To answer this question , we have to find how nearly the potentials at surfaces of these figures may be computed from the spherical harmonic formulae .
It would be laborious to make an accurate computation of the potential , and fortunately appears to be unnecessary to do so , since a sufficient answer may obtained in another way .
The potential of an ellipsoid of revolution may be computed either rigorously or harmonic analysis .
With a certain degree of eccentricity the approximate result agree badly with the rigorous one .
If the ellipsoid consists of a fluid of unit density , there is a certain angular which makes ib a level surface .
If be that aIlgular velocity , then we know the spherical harmonic solution would give as the ratio of the minor the major axis .
If then , are the rigorous values of the minor and major semithe harmonic approximation is good if does not differ much from If we denote by the factor by which the approximate value of the ratio the axes is to be multiplied in order to obtain the ligorous value , we have and may be regarded as a measure of inaccuracy .
A table of the values of , corresponding to , is computed from the transcendental equation in TAIT'S ' Natural Philosophy , ' S772 .
From these I compute as EQUILIBRIUM OF ROTATING MASSES OF FLUID .
The measures of inaccuracy corresponding to he values of in the first column , or the values of in the second , are the reciprocals of the numbers in the last column .
We thus see that there is still a considerable degree of approximation when or when the ratio of the axes is 3 to 5 , for the measure of inaccuracy is ; but for the approximation is bad .
Now the shapes of certain egg-like bodies have been computed by the spherical harmonic method , and it seems safe to assume that the approximation has given about the same degree of accuracy as would hold in the case of an ellipsoid of revolution whose minor axis bears to its major axis the same ratio as the shorter axis of the egg to the longer .
Turning now to our computation , and considering only the more elongated or meridional sections , we see that , when , the longer axis is and the shorter 2 ; and the ratio is 6 , which correlponds to the measure of inaccuracy .
It might , however , be more legitimate to adopt two different measures , and at the pointed end of the egg to take the ratio 773 : , which will correspond to a measure of inaccuracy about ; and at the blunt end to take the ratio 773 : , which would correspond to a measure of inaccuracy or In the case of the two masses cross one another , and the result has been used to give an approximate picture of the dumb-bell figure of equilibrium .
We now see that even in this case there is a suffioient degree of approximation to give a very good idea of the accurate result .
the case of the meridional section , where , we have for the ratio of axes at the pointed end of the egg , and measure of inaccuracy about ; and at the blunt end , and measure of inaccuracy In the case of the similar figures are , for the pointed end , and measure of inaccuracy about ; and for the blunt end , and measure of inaccuracy perhaps about It thus appears that as the bodies recede the accuracy increases with great rapidity , and in the two cases considered last it is hardly necessary , from a physical point of view , to consider greater accuracy than that attained .
It must be remarked , however , that this way of estimating the degree of inaccuracy must necessarily give much too unfayourable a view .
If we have a single mass of fluid departing considerably from the spherical form , it is clear that the potential computed on the hypothesis of a layer of surface density on the true spbere will come to depart largely from the potential at the surface of the fluid .
If , however , we compute the potential of such a mass at points a little remote from the surface , the approximation will be much closer .
Now , where there are two masses , as in our problem , the potential at the surface of either mass consists of two parts , one due to the mass itself , the other due to the other mass .
As regards the first of these two parts , the above criterion is applicable , but as regards the second part it gives too unfavourable a view .
Now in the case of the single mass the deformative forces due to centrifugal force are considerably vitiated by computation at the spherical surface instead of the surface , whilst in the case of the two masses the tide-generating forces are computed with greater accuracy than is shown by the criterion .
Under these circumstances it has appeared worth while to give another figure below , which , judged by the criterion , would be no approximation at all .
The reasons for giving this figure will be stated when we come to it .
S 10 .
To find the Moment of Momentum of the System .
Rotating figures of equilibrium are classified according to the amount of moment of momentum with which they are endued .
It is , therefore , interesting to determine the moment of momentum of the systems now under consideration .
We must begin by finding the moments of inertia of the two masses .
Let SI , Si , denote the moments of inertia of the shells of zero mass lying on the mean spheres of radii Then where , and where the integral is taken tbroughout angular space .
Now , and is the sum of a series of .
Then , in consequence of the properties of harmonic functions , we need only consider the harmonics of the second degree in , and EQUILIBRIUM OF ROTATING MASSES OF FLUID .
and the moment of inertia of the mean sphere is ; hence , if we write the moments of inertia , and , are given by .
already have in ( 71 ) .
Hence the sum of the rotational momenta of the two masses is The whole system revolves orbitally about the centre of inertia with an angular velocity : hence the orbital momentum is But Hence the orbital momentum is and this is equal to It will be convenient to refer the mass to the radius of a sphere of the same mass as the sum of the two .
Let be the radius of such a sphere ; then Thus the whole moment of momentum is We shall therefore compute the coefficient of Computing from this formula , I find the following values of the moment momentum in the case where the masses are equal , when Now I find by a numerical that , if we imagine a mass of fluid to in the form of a Jacobian ellipsoid of three unequal axes , then , the momentum is , the axes of the ellipsoid are and , when the momentum is , the axes are It seems probable , then , that the Jacobian ellipsoid of mass becomes least as soon as when the moment of momentum is somewhere about It may be worth mentioning that the greatest moment of momentum for which ellipsoid ( of mass ) is stable , when it is a figure of revolution , is S11 .
the Conditions under which the Masses may be close to one If at any point on the surface of either mass the sum of the tide-generating centrifugal forces is greater than gravity , it is obvious that equilibrium cannot It is also clear that , if this condition is to be found anywhere , it will be at that of the smaller mass which lies nearest to the larger mass .
Hence , in order that system may be a possible one , we must satisfy ourselves that at that point gravity the itself exceeds the sum of the tide-generating and centrifugal forces .
To determine the limitations of size and proximity of the smaller of the masses to a high degree of approximation would be very laborious , and we therefore , content ourselves with a rough investigation , to be explained below .
We shall now find approximations for the shapes of the two masses and for potentials .
The radius-vector of either mass and the potential may be expanded in and , and a term involving in the denominator will be referred to as of the order .
Now the telm of the highest order which can be included without the of great complication is the 7th , and we shall content ourselve with that term .
The expressions for the various parts of the potential have been developed but it may be observed that the terms involving the first order of harmonics * ' Roy .
Soo .
Proc vol. 41 , 1887 , p. 319 .
EQ OF ROTATING MASSES OF FLUID .
omitted , since they are subsequently annulled by a proper choice of the angular velocity .
From we have The last term in the development to the th order is that .
Then it is dear that we require correct to the 4th order , to the 3rd , and so on .
But ( 25 ) shows us that the 's are equal to unity to the 4th order inclusive .
Hence , in the above , all the 's may be treated as unity .
Again when written in reference to the origin affords other terms , in which all those included under are of the 8th and higher orders , and negligible ; and the rest ( with omission of the first harmonic term ) gives Thus this first part of the potential is , to the 7th order inclusive , Next , from the expression for in S3 , we have a term in the poterItial due to rotation .
The remaining terms due to rotation will be taken up later .
From ( 71 ) we see that , to the 7th order inclusive , Hence and are of the 3rd order ; from ( 48 ) and ( 64 ) it follows that the factors by which the 's are derived from the 's and 's are of the 5th order .
And , since the 's and 's only differ from unity in terms of the 5th order , it follows that the 's and 's are of the 5th order .
Then ( 41 ) and show us that all the terms in and are negligible .
The first set of terms due to rotation and to the cotTesponding deformation given in and ( 43 ) , and together contribute The second set of terms due to rotation , and to the corresponding deformation , are given in ( 54 ) and ( 58 ) , and together contribute Hence , to the 7th order inclusive , we have .
Now the expression ( 72 ) for the radius-vector of the mass to the same approximation gives us ' and a similar expression for To determine the inward force at the pole of the mass , where it is nearest to the mass , we must evaluate , and in the first term substitute the expression for , and in the remaining terms put ; also at this pole and Then , differentiating ( 78 ) , ( 79 ) But at the pole Substituting this for the first term of ( 79 ) , we have But ; hence EQUILIBRIUM OF ROTATING MASSES OF FLUID .
Thus the criterion of the possibility of equilibrium is that ( 80 ) should be positive .
But the radius-vectors of the poles are and , similarly , Therefore .
Now the interval between the two masses is ; hence , if the two masses are just in contact , .
( 81 ) In order , then , to test whether equilibrium is still possible when the two masses are just in contact , it is necessary to determine from ( 81 ) ; and then , substituting in ( 80 ) , find whether is positive or not .
The solution of an equation and the determination of can only be performed by trial and error .
Now suppose that the solution is 8 where is small ; and that Then it is obvious that . .
and and It is not hard to find an approximate solution by trial and error , and the correc results may then be found thus .
Consider the case where the two bodies are equal to one another , and put The equations then become By trial and error we find From this we conclude that equilibrium ] subsists when the two masses are in contact .
When , we have , and Our figures showed that when the two masses were nearly in contact , and when they crossed .
This result is , therefore , in accordance with the figures .
Next pass to bhe case of an infinitesimal satellite , and suppose irfinitely small compared with and , and that .
The equations are The solution of the ffist equation is , and this value uf makes Hence we conclude that an infinitesimal fluid satellite cannot revolve with its surfaoe in contact with its planet .
vanishes when .
Hence it appears that the nearest approach of the infinitesimal satellite to the planet is mean radii of the planet .
The nature of the approximation adopted is , however , such that in reality the satellite 1rmst lie further from the planet than this , perhaps at two radu distance .
* The satellite and planet of which we here speak are , of course , supposed to revolve as parts of a rigid OCHE , ' Montpellier , Acad. .
Mem vol. 1 , lS47-1860 , p. ( added Oct. 5 , 1887 EQUILIBRIUM OF ROTATING MASSES OF FLUID .
Now , if for equal masses equilibrium still subsists when the two masses are in whilst for infinitesimal mass of one equilibrium is impossible with massed contact , it follows that for some ratio of masses equilibrium can just bsist when they are in contact .
The question , therefore , remains to determine this ting ratio of masses .
I find , then , that when , we have the numbers in being the logarithms of the coefficients .
The solution of this is , which makes Again , when , we have the tion of which is , which makes .
Since , and , it follows that the ratio of the masses in the first case is 1 : , and in the second 1 : From this it appears that when the ratio of the masses is about 1 to 38 equilibrium still just possible when the two masses touch .
It must be borne in mind , however , that the nature of the approximations adopted in this investigation is such that the results in this limiting case are only given very roughly , and it is certain that actually the limiting size of the smaller of the two masses must be greater than as thus computed .
We can only conclude that the limiting case occurs when the ratio of the masses is about 1 to 30 , or the ratio of the radii about 1 There is one other case which it is interesting to consider , namely , to find the limiting proximity of the Moon to the Earth , both bodies being treated as homogeneous fluids of the same density , revolving as a rigid body .
The case of Moon and Earth is well represented by ; for this gives 1 to as the ratio of the masses .
With these values I find and Now will be found to make vanish , and , with this of If be 4000 miles , miles , and miles Thus , as far as this investigation goes , it appears that when the flnid Moon is on the point o.f breaking up from stress of tidal and centrifugal forces the distance between centres of Moon and Earth is 6500 miles , and the shortest distance between the cwo surfaces is 380 miles .
This result must , however , from the nature of the approximation , be an underestimate of the distances .
The whole of the present section has been suggested by a pamphlet by Mr. NOLAN* in which he criticises some of my preyious papers .
I have commented elsewhere on his criticisms .
S12 .
On the Case the two Masses are unequal .
The results of the previous section point to a very remarkable limitation to the possibility of approach of two masses of unequal size .
It has , therefore , seemed worth while to consider this case numerically , and a case is therefore chosen which shall approach near to that which we know is the limit of possibility .
I choose , therefore , , which makes the ratio of the masses 1 to 27 , and , which brings the protuberances into close proximity .
The numerical details are omitted , but figs. 5 and 6 ( Plate 23 ) give the results , the numerical values of the radius-vectors being , as before , entered on the figure .
The elongation of the smaller mass is so extreme that it is obvious that , rigorously speaking , the spherical harmonic approximation must be considered to break down .
Nevertheless , I conceive that these curious figures may be held to indicate the general nature of the true result .
It is remarkable that the smaller mass exhibits a marked furrowing round the middle .
This seems to indica , te that such a system tends to break up by the separation of the smaller mass into two parts .
S13 .
The intention of this paper is , first , to investigate the forIns which two masses of fluid assume when they in close proximity about one another , without relative motion of their parts ; and , secondly , to obtain a representation of the single form of equilibrium which must exist when the two masses approach so near to one another as just to coalesce into a single mass .
When the two masses far apart the solution of the problem is simply that of the equilibrium theory of the tides .
Each mass may , as far as its action on the other is * ' DARWIN'S Theory of the Genesis of the Moon .
' ROBERTSON , Melbourne , Sydney , Adelaide , and isbane , 1885 .
'Nature , ' February 18 and July 29 , 1886 .
EQUILIBRIUM OF ROTATING MASSES OF FLUID .
onoerned , be treated as spherical , and the tide e. potential is given accuracy by a single term of the second order of harmonics .
As the masses are brought nearer to one another , this approximation ceases to be sufficient , terms of higher orders of harmonics become necessary to represent the potential adequately , and the departure from sphericity of each mass begin to exercise a sensible deforming influence on the other .
When the departure from sphericity of one body produces a sensible deformation in the other , that deformaticn in its turn reacts on the first , and thus the actual figure assumed by either mass may be regarded as a deformntion due to the primitive influence of the other mass , on which is supelposed the sum of an infinite series of reflected deformations .
But each mass is deformed , not only by the tidal action of the other , but also by its own rotation about an axis perpendicular to its orbit .
The departure from sphericity of either body due to rotation also exercises an influence on the other , and thus there arises another infinite series of reflected deformations .
It is shown in this paper how the summations of these two kinds of reflections are to be made by means of the solution of three sets of linear equations for the determination of three sets of coefficients .
The first set of coefficients are augmenting factors , by which the tides of each order of harmonics are to be raised above the value which they would have if the perturbing mass were spherical .
It appears that , the higher the order of harmonics , the more do these factors exceed unity .
The second set of coefficients correspond to one part of the rotational effects .
They appertain to terms of exactly the same form as the tidal terms , and in the final result the terms to which they apply become fused with the tidal terms .
These terms are the zonal harmonics of the several orders with respect to the axis joining the centres of the two masses .
The third set of coefficients correspond to the remainder of the rotational effect , and they appertain to a different kind of deformation .
These deformations represented by sectorial harmonics involving , where is azimuth measured from the plane passing through the axis of rotation of the system and the centres of the two masses .
That term of this set which is of the second order of harmonics , and which represent the icity of either mass augmented by mutual influence , is the only term which is considerable , even when the two masses are very close together ; but the existence of the other harmonic deformations of this class interesting .
We may say , then , that all the tides of either mass are augmented above the values which they would have if the other mass were spherical ; that the ellipticity corresponding to rotation is augmented ; and that the deformation due to rotation is no longer exactly ellipticspheroidal .
The angular velocity of the system is found by the consideration that the repulsion due to centrifugal force between the two exactly balance the resultant 4 ] 8 PROFESSOR G. H. DARWIN ON OF ttion between them .
If the masses were spherical , the square of the orbital Jgular velocity , multiplied by the cube of the distance between the centres , would be equal to the sum of the masses .
When the sses are deformed , however , this law is no longer true , and the ular velocity has to be augmented by a factor a little greater than unity , which depends on amounts of the deformations .
The theory here sketched is ap plied above numerically to several cases , and the results will be found in the preceding paragraphs .
We shall first consider the cases where the two masses are equal to one another .
In the first example solved numerically , the distance between the centres of the two masses is times the mean radius of either of them .
The two bodies are found to be elongated until they approach near to one another ; but , as the character the distortion is better illustrated in a subs.equent case , the result is not given graphically .
All the data , however , are found which will enable the reader to draw bhe figure if he should wish to dc so .
In the next example , with the masses still equal , the distance between the centres is reduced to of the mean radius of either .
The result of the solution is illustrated by two figures .
In fig. 4 , Plate 22 , the section of the masses by a plane endicular to the axis of rotation is shown , and in fig. 5 , Plate 23 , we have the section by a plane the axis and the centres of the two masses .
On both figures are inscribed the values of the radii for each of latitude in terms of the mean radius as unity , and the mean sphere , from which the distortion is computed , is marked by a short line on each radius .
The elongation of the masses is , of course , greater in the section through the axis than in the other section .
Each mass is shaped somewhat like an egg , and the small ends face one another and come very nearly into contact .
In the to the figures , amongst other numerical data , are given the square of the angular velocity and the angular momentum of the system .
The density of the flnid being unity , the angular velocity is given by the value of ; this is the function of angular velocity which is usually given when reference is made to figures of equilibrium of rotating fluid , such as the ] utional or Jacobian ellipsoids of equilibrium .
The moment of momentum of the system is given by reference to the angular velocity of a sphere , of the same mass as the sum of our two masses , rotating so as to have the same momentum .
If , in fact , be such a length that a sphere of fluid of that radius has the same mass as our system ( so that ) , then the moment of momentum is given by a number in the expression .
By this notation the angular velocity and moment of momentum are made comparable with the results given in a previous paper on the Jacobian ellipsoid of From that paper the following table of the axes , angular velocity , and moment of momentum of several solutions of JACOBI'S problem is extracted .
* Ro.y .
Soc. Proc vol. 41 , 1887 , p. 319 .
h'QUILlBRIUM OF ROTATING MASSES OF FLUID .
In figs. 4 and 5 , Plates 22 , 23 , is , and the momentum is 472 .
On comparison with the Table of JACOBI'S ellipsoids , we see that this corresponds with a considerably slower rotation than the 6th solution , and nearly the same moment of momentum .
In the next case the two masses are still closer , the distance between the centres being only times either mean radius .
The result is illustrated in figs. 2 and 3 ; the explanation of figs. 4 and 5 serves , mutandis , for these two figures also .
This case is interesting because the masses approached so near to one another that they partially overlap .
Two portions of matter cannot , of course , occupy the same space , and the continuity of figures of equilibrium leads us to believe that the reality must consist of a single mass of fluid .
In figs. 2 and 8 conjectural dotted lines are drawn to show how it is probable that the overlapping of the two masses is replaced by a neck of fluid joining them .
The figures as thus amended serve to give a good representation of the single dumb-bell shaped figure of equilibrium .
The angular velocity is here given by , and the moment of momentum by 482 .
In the sixth entry of the Table of Solutions of JACOBI'S problem we find , and the moment of momentum .
This ellipsoid has , then , the same moment of momentum , and only about 4 per cent. more angular velocity , than our dumb-bell .
It has seemed , therefore , worth while to mark ( in chain-dot ) on figs. 2 and 3 the outline of this Jacobian ellipsoid of the same mass as the dumb- bell .
The actual vertex of the ellipsoid just falls outside the limits to which it was to extend the figure .
In the above referred to it is shown how the energy of the Jacobian ellipsoid is to be computed .
If we denote the kinetic energy by , and the intrinslc energy by then it appears that in the case of the ellipsoid in these figures , and the total energy * The intrinsic energy being negative , it is more convenient to tabulate a positive quantity .
PROFESSOR G. H. DARWIN ON GURES OF Now in the case of our dumb-bell figure it appears , from calculations reforred to in the Appendix , that , and .
Hence in the dnmb-bell the kinetic energy less , but the intrinsic energy is so much greater that the total energy is a half per ) .
greater .
These numbers are , of course , computed from the approximate formulae , and must not be taken as rigorously correct for the dumb-bell figure of equilibrium .
With reference to a figure of transition from the Jacobian ellipsoid , Sir THOMSON has remarked:\mdash ; * " " We have a most interesting between the unstable Jacobian ellipsoid , when too slender for stability , and the case of smallest moment of consistent with stability in two equal head portions .
The consid.eration of bow to fill up this gap with intermediate figures is a most attractive question , towards which we at present offer no contributio1 Figs. 2 and 3 are intended to form such a contribution , but it is certain that tter is far from being probed to the bottom .
M. POINCARE ' has made an admirable investigation of the forms of equilibrium of single rotating of fluid , and has especially considered the stability of ellipsoid .
He has shown , by a difficult analytical process , when the ellipsoid elongated ( he has not arrived at a numerical result ) instability sets in by furrowing of the ellipsoid along a line which lies in a plane perpendicular to longest axis .
It is , however , extremely remarkable that the furrow is not symmetrical with respect to the two ends , and thus there appears to be a tendency to form dumb-bell unequal bulbs .
If M. POINCARE\ldquo ; S result shall appear to be not only true , but to contain the truth concerning the mode in which instability of the ellipsoid supervenes , then must be some other transitional form between the unsymmetrically fUrrowed *THOMSON and TAIT'S ' Natural Philosophy , ' 1883 , S .
In he remarks that " " a deviation from the ellipsoidal figure in the way of thinning it the middle and thickening it towards the end wouId , with the same moment of momentum , give energy I conceive that the energy referred to throughout this paragraph is kinetic only , and we seen that the kinetic energy is less for the dumb-bell than for the ellipsoid .
[ If we write for a quantity proportional to the excess of kinetic above intrinsic energy , so , then figures of equilibrium are to be determined by making stationary for of the parameters involved in it .
This course is actually pursued in the Appendix below , the ( viii .
) being , in fact , this ; and the variations of it , being made stationary , ffiord a controlling of the problem of this paper .
The similar method may easily be applied to the case of JACOBI'S soids .
From this point of view the interesting functiou to tabulate iS , and we observe in the case of the Jacobian ellipsoid referred to on the last it is 6052 , and for the dumb-bell it 6156 .
Is not the energy referred to by Sir W. TROMSON this function ( Addihon to dated Ocfober 10 , 1887 ' Acta Mathematica , ' 7 , 3 and 4 , 1885 .
EQUILIBRIUM OF ROTATING MASSES OF FLUID .
enipsoid and bhe dumb-bell ; except , perhaps , in the case where the two bulbs pass on to two masses of a definite ratio .
M. 's work seemed so important that this paper was kept back for a year , whilst I endeavoured to apply the principles , which he has pointed out , to the discussion of the stability of the two masses .
The attempt , which is in the Appendix , is apparently abortive , on account of the imperfections of spherical harmonic analysis when applied to bodies which depart considerably from the spherical shape .
We must , therefore , leave this complex question in , and merely point to the Ap.pendix as an example of the method which must almost certainly be this problem is to yield it , answer to analysis .
Allusion has just been made to the imperfection of spheyical harmonic } alysis , and this brings us naturally to face the question whether that ] ysis may not have been pushed altogether too far in the of the figures of equilibrium under discussion .
This question is considered in S 9 , and a rough criterion of the limits of applicability of this analysis is there found .
From this it appears that even in the cases of figs. 2 and 3 the result must present a fair approximation to correctness .
The criterion , indeed , appears to be such as necessarily to give too unfavourable a view of the correctness of the result .
The rigorous method of discussing the stability of the system having failed , certain considerations are adduced in S 11 which bear on the conditions under which there is a form of equilibrium consisting of two fluid masses in close promixity .
It appears that there cannot be such a form with the two masses just in contact , unless the IlaIler of the two masses exceeds in mass about one-thirtieth of the larger .
If we take into consideration the that the criterion of the applicability of harmonic analysis is too severe , it appears to be worth while to find to what results the analysis leads when two masses , one 27 times as great as the other , are brought dese together .
The numerical work of the calculation is omitted , since the numbers can only represent the true conclusion very roughly ; but the result is illustrabed graphically in figs. 6 and 7 , Plate 23 .
These figures can only serve to give a general idea of the truth , but the form into which the smaller nlass is thrown is so remarkable as to be worthy of attention .
The deep furrow round the smaller mass , lying in a plane parallel to the axis of rotation , cannot be due merely to the imperfection the solution ; and it appears to point to the conclusion that there is a tendency for the smaller body to separate into two , just as we have seen the Jacobian ellipsoid become dumb-bell shaped and separate into two parts .
In this paper , indeed , we have sought to trace the process in the opposite direction , and to observe the coalescence of two masses into one .
The investigation oomplementary to , but far less perfect than , that of M. POINCARE ' , who describes the series of changes which he has been tracing in the following words:\mdash ; " " Considerons un mass fluide homogene animee origin airement d'un mouvement de rotation ; imaginons qne cette mass se contract en se refroidissant lentement , mais de facon a rester toujours homogene .
Supposons que le refroidissement assez lent et le frottement interieur du fluide asbez fort pour que le mouvement tation rest le meme dans les diverses portions du fluide .
Dans ces conditions fluide tendra a prendre un figure d'equilibre seculairement stable .
moment de la quantity de mouvement restera d'ailleurs constant .
" " Au debut , la density etant tres faible , la figure de la mass est un ellipsoid6 revolution tres peu different d'une sphere .
Le refroidissement aura d'abord pour d'augmenter l'nplatissement de l'ellipsoide , qui restera cependant de revolution .
Quand sera devenu a peu pres egal a , l'ellipsoide cessera d'etre revolution et deviendra un ellipsoide de JACOBI .
Le refroidissement continuant , mass cessera d'etre ellipsoidale ; elle deviendra dissymetrique par rapport au des , et elle affectera la form representee dans la figure , p. " " Comme nous l'avons fait observer a.propos de cette figure , l'ellipscide semble creuser legerement dans sa party moyenne , mais plus pres de l'un des deux du grand axe ; la plus grande party de la tend a se rapprocher de la spherique , pendant que la plus petite party sort de l'ellipsoide par un des du grand axe , comme si elle cherchait a se detacher de la mass principa]e .
" " Il est difficile d'annoncer avec certitude ce qui arrivera ensuite si le meant continue , mais il est pertnis de } ) oser que la mass ira en se creusant de en plus , puis en s'etranglant dans la party moyenne , et par se partager en corps isoles .
" " On pourrait re tent de chercher dans ces considerations un confirmation ou refutation de l'hypothise de LAPLACE , mais on no doit pas oublier que les sont ici tres ditferentes , car notre mass est homogene , tandis qne la nebuleuse LAPLACE devait etre tres fortement condensee verse le centre It was in the hope that the investigation might throw some light on the hypothesis of LAPLACE and KANT that I first undertook the work .
It must admitted , however , that we do not obtain much help from the results .
It is remarked by M. that the conditions fcr the separation of a satellite from nebula differ from those of his problem in the great concentration of density in central body .
But both his investigation and the considerations adduced here to show that , when a portion of the central body becomes detached through angular velocity , the portion should bear a far larger ratio to the remainder than observed in the satellites of our system as compared with their planets ; and it is probable that the heterogeneity of the central body can make so great a difference the result as would be necessary if we are to make an application of these idea.s .
appears then at present necessary to suppose that after the birth of a satellite , if takes place at all in this way , a series of changes occur which are quite unknown .
* The furrowed ellipsoid of COBf .
, ' Acta Mathemat 7 , 1885 , p. 379 .
EQUILIBRIUM OF ROTATING OF FLUID .
On the Energy and Stability of the System .
N. PorNcAls6 has shown in his admirable memoir , referred to in the Summary , how the dynamicnl abiliby of a rotating fluid system in relative equilibrium depends on the energy .
Certain factors in the expression for the energy , which he calls coeffioients of stability , are there proved to afford the required Cl.iterion .
It will now be shown how in this case coefficients of stability are determinable , at least as far as spherical harmonic analysis permits .
The results will also cast interesting light on the methods by which the equations to the two masses have been ined .
The task before us is to determine the " " exhaustion of potential energy\ldquo ; of the two masses in presence of one another as due to the deformation of each from the spherical by yielding to gravitation and to centrifugal force .
The work will bc rendered simpler by the introduction of a new notation .
Leb us write , then , as the aeqnations .
to two shapes , which are nob necessarily togethor a figure of equilibrium:\mdash ; be observed that these equations have the same form as ( 72 ) , that the constants introduced are different from the , which were determined , so that the figures might be figures of equilibrium .
At present we do not assume that do represent figures of equilibrium .
The energy iost may be divided into several , the energy lost by the mass yielding from its sphorical figure to the gravitabion of the mean sphere , the exhaustion of mutual energy of that layer of matter on the mass which constitutes ils departure from sphericity .
, the Ioss of energy due to the deformation of the mass in presenoe of the mean sphero , the similar quantities for the mass , the loss of energy of the two layers in presence of one another .
, the loss of energy due to the deformation of the mass in the presence of ifugal force duo to rotation , the similar loss for 1st .
is equal and opposite to the work required to raise each clement of the layer on through half its own height against the gravity due to the mean sphere .
This gravity is .
Co-Iatitude and ongitude being denoted by ) , let , an element of soJid angle .
In effecting the inte- grations , the properties of spherical harmonic functions are used without comment , .
:\mdash ; Then , taking only a typical term of the first of , ' whence , with all the terms , and remembering that The formula for may be written by symmetry .
2nd .
, the exhaustion of mutual energy of the layer on itself , is balf the potenbial of the layer any element , multiplied by the mass of the element , and integrated over the whole sphere .
The potential of the layer is .
Then , at an element of the layer , and taking a typical term only , we have whence The formula may be written down by symmetry .
The addition of to , and of to , simplifies these expressions by cutting out the factor immediately following the in either , and replacing it by unity .
3rd .
is the loss of energy due to raising the layer on in presence of the an sphere .
We multiply the potential of the sphere by the mass of the element on , and rate throughout angular space .
The potential of the sphere , wheu transferred to the igin o , is Then , at an element of the layer and taking a typical term , whence . . .
( iv .
) The expression for may be written down by symmetry .
On collecting results from , , and , we have and a similar expression for 4th .
is the loss of energy of one layer in the presence of the other .
We take the pote , ntial of the layel .
on , multiply it by the mass of an element on , and The potential of the layer on when transferred to the , as in , is cing this into the , only taking a typical term , and neglecting those terms in the integral which must vanish , we get EQUILIBRIUM OF ROTATING MASSES OF FLUID .
Then , effecting the integrations , and putting in the , we get This involves the two figures symmetrically .
5th .
is the loss of energy in the yielding of the figure to centrifugal force .
To find it , we multiply the potential due to rotation by the mass of eacb element of the layer , and integrate .
By ( 35 ) and we know that the rotation potential is As this only involves harmonics of the second order , we may iu ths layer all terms except those of the second order .
Thus we get .
( vii .
) The expression for may be written down by symmetry .
Collecting results from , , and ( vii .
) , we get , for the whole exhaustion of energy , .
( viii .
) The expression is found without any assumption that the two masses are bounded by level surfaces , and therefore in equilibrium .
But the condition for equilibrium is that the differential coefficients of with respect to any one and all of the parameters , shall vanish .
If we equate to zero , we get If , however , , thpre is on the left-hand side an additional term The equation of to zero gives a similar equation .
we equate to zero , we get PROFESSOR G. H. DARWIN ON FIGURES OF If , howevor , , there is on the lefkhand side an additional term The equation of to zero gives a similar equation .
Now , if we put for except when , and then pnt , and similarly introduce the 's and ; and if we put except when , and then pul , and ] introduce the , it is easy to see that the eq11ntions to two faces become the same as ( 72 ) , and the eqnations of condition between and , and between and , become exactly those which we found by a different method above in ( 23 ) , 44 ) , and ( 59 ) The only difforence is that the equations for and are fused together .
This , therefore , forms a valuable confirmation of the correctness of the long analysis employed for the determination of the forms of equilibrium .
The formula ( viii .
) also enables us to obtain the intrinsic energy of the system , that is to say , the exhaustion of onergy of the concentration of the matter from a state of infinite dispersion to its actua ) shape , with changed .
The last line of ( viii .
) depends on the yielding of the fluid to centrifugal force , and musr be omitte8 from the exhaustion of energy .
Besides the rest of ( viu .
) , we have in th ' exhaustion of energy of the system , bxhaustion of two spheres and their mutual exhaustion .
It is clear , then , that the intrinsic energy is where the , have their values determined in accordance with the condition that the surfaces level surfaces .
In evaluating ) intrinsic energy from this formula , it is convenient to refer the to that of 1 sphere of such radius , , that its mass is equal to the whole mass of the systein .
Then b$ and we may take the intrinsic energy as .
Thus be a numerical quantity which is positive .
I find from ( ix .
) , with and , that With Tegard to the kinetic energy of the system , we have seen in S10 that the moment of is , where is a numerical quantity , and we find in the course of the determinatioik funcbion .
Then , since the energy is half the moment of momentum multiplied by the volocity , it is clear that the kinetic energy is The kinetic energy , as represented by , is oomparable with the intrinsio energy represented by In the case of , and , I find the kinetic energy Thns the total energy *[ Seefoot.note added to the Summary above .
October 10 , 1887 .
] EQUILIBRIUM OF ROTATING MASSES OF FLUID .
If the energy of a system be expressed as the sum of a number of coefficients , each multiplied by the wquare of a parameter , it has been shown by M. that the stability of the system depends on { he signs of these coefficients , which he calls " " the coefficients of stability But , if the expression for the energy involves the products as well as the squares of the parameters , the coefficients of stability are the roots of a determinantal equation involving the second differentials of the with respecb to the parameters .
Let , a linear quadratic function of , \amp ; c. , be the energy of a system in equilibrium ; then tlJe determinantal equation is . . . . . .
The solution of this equation in involves the determination of the several fundamental modes of vibration of the system ; and the roots are the coefficients of stability .
Now that involves a constant , then , in causing that constant to vary continuously , we have a series of systems of eqnilibrium of the same kiud ; and the coefficients of stability vary continuously at the same time .
If the system be initially iu stable equilibrium , the stability ceases when a coefficient of equilibrinm vanishes .
The system at the moment of instability is in a condition of " " bifurcation that is to say , there is another series of shapes of a different kiud , of which this shape a member .
In making the constant vary past the critical value , we find this second series of shapes stable , whilst the first is unsbable .
If the system be in uniform rotation , so that instead of absolute equilibrium there is equilibrium relatively to uniformly rotating axes , the same theorems hold true , provided that only oue root of the determinantal equatiou vanishes at a time .
This last is the case which we are considering , and the constant , which we suppose to vary continuously , is , the distance between the two centres of the mean spheres of radii and When the two masses are far apart the equilibrium is stable , but wheu they are brought cIoser a time may come when one of the coefficients of stability vanishes .
The condition for the vanishing of a coefficient of stability is determined by the determinant with To find the determinant , we have to evalnate the second differentials of with respect to the parameters If we form the determinant corresponding to with , it is obvious that two infinite squares of entries which are diagonally opposite to one another , and which meet at a corner , are to be filled in with differentials involving , in the denominators .
All these entries are zero , and hence the infinite determinant splits into two independent infinite determinants , one only involving the differentials with respect to , and the other only those with respect to .
The determinant may be called " " the tidal determinant the one " " the rotational determinant for the origin of the tel.ms in each is obvious .
By considering only the tidal determinant , we see how the other may be treated very shortly .
For the sake of brevity , write Then stability.vanishes , as far as regards the tidal forces , when It will be observed that , by the notation ( xi .
) , and the appropriate division of the columns and rows of the tidal determinant , it has been converted into a determinant in which the diagonal consists of ones .
If we drop certain factors common to the whole , we have , e , by differentiating ( viii .
) , ; . . .
xiii .
) With these values ( xiu .
) we easily find , by substitution in ( xi .
) , that [ This expression gives the valne of each of the entries in the infinite determinant .
Now it is possible , by a certain laborious investigation which I do not here reproduce , to develope this infinite " " tidal\ldquo ; determinant in the form of a series , and then to show that , however close the two masses may be to one another , the series arising from the tidal determinant can never vanish .
It may also be proved that the other infinite determinant , which resuIts from " " rotational\ldquo ; terms , is necessarily greater than the tidal determinant , and a fortiori can never vanish.]* Thus it might be held that stabihty must subsist in the figures of equilibrium until the two maases come into contact .
But , as appears from S 11 , it is certain that , if one of the masses be smaller than other , this cannot be the case .
In , the investigation must break down on account of the imperfecMl arising from the use of spherical harmonic analysis .
We have ssen that the infinite determinant , which gives the coefficients of stability , splits into tma parts when we rely on spherical harmonic analysis ; but when instability ensues it must be about by the joint action of the tidal and rotational forces .
It appears certain then that , if a analysis were used , this separation would not take place .
Although the present investigation proves thus to be abortive , I have thought it worth while to skeIph the process , because it almost indicates the line that must be pursued whenever a more rigorous analysis shall be applied to this difficult problem .

