]\gt ; [ 187 ] VI .
The Waves on Rotating Liquid Spheroid of Finite Ellipticity .
G. H. BRYAN , Communicctted Professor G. H. DARWIN , F.R. Received November 6 , \mdash ; Read , 1888 .
1 .
THE hydrodynamical problem of finding the waves or oscillations on a gravitating mass of liquid which , when ) disturbed , is rotating as if with finite angular velocity , in the form of an ellipsoid or spheroid , was first successfully attacked by M. POINCARE in 1885 .
In his ) ortant memoir , " " Sir Mass Fluide animee d'un Mouvement de Rotation POINCARE has ( S13 ) obtained the differential equations for the tions of rotating liquid , and shown that , by a transformation of projection , the determination of the oscillations of any particular period is reducible to finding a suitable solution of LAPLACE'S equation .
He then applies LAME'S functions to the case of the ellipsoid , showing that the differential equations are satisfied by a series of LAME'S functions referred to a certain auxiliary ellipsoid , the boundary-conditions , however , involving ellipsoidal harmonics , referred to both the auxiliary and actual fluid ellipsoid .
At the same time , POINCARE'S analysis does not appear to admit of any definite conclusions being formed as to the nacure and frequencies of the various periodic free waves .
The present paper contains an application of POINCARE'S methods to the simpler case when the fluid ellipsoid is one of revolution ( MACLAURIN'S spheroid ) .
The solution is effected by the use of the ordinary tesseral or zonal harmonics applicable to the fluid spheroid and to the auxiliary spheroid required in solving the differential equation .
The problem is thus freed from the difficulties attending the use of LAME'S functions , and is further simplified by the fact that each independent solution contains harmonics of only one particular degree and rank .
By substituting in the conditions to be satisfied at the surfnce of the spheroid we arrive at a single boundary-equation .
If we are treating the forced tides due to a known periodic disturbing force , this equation determines their amplitude and , hence , the elevation of the tide above the mean surface of the spheroid at any point at any time .
If thele be no disturbing force , it determines the frequencies of the various free waves determined by harmonics of given order and rank .
Denoting by the ratio of the frequency of the free waves to twice tlJe frequency of rotation of the Acta Mathematica , ' vol. 7 .
liquid ) its lxis , the values of the roots of a rational algebraic equation , and depend only on the eccentricity of the spheroid , as well as the degree and rank of the nnlonic , while the number of rent free waves depends on the degree of the in .
At instant the of the disturbance at any point of the surface is proportional to the corresponding surface harmonic the spheroid , ]nultilied by the central perpendicular on the tangent plane , and is of the same form for waves determined by harmonics of any given degree and rank , whatever be their frequency ; but the motions of the fluid particles in the interior will differ in nature in every case .
the case of zonal harmonics of the degree , we find that , according as is even or odd , there will be or different periodic motions of the liquid .
These are essentially oscillatory in character and symmetrical about the axis of the spheroid .
In all but one of these the value of is essentially less than unity , that is , the period is greater than the time of a semi-revolution of the liquid .
Taking next the tesseral harmonics of degree and rank , we find that they determine periodic small motions .
These are essentially tida ] waves rotating with various angular velocities about the axis of the spheroid , the angular velocities of those rotating in opposite directions being in general different .
All but two of the values of are numerically less than unity , the periods of the corresponding tides a point fixed relatively to the liquid being greater than the time of a semirevolution of the mass .
The mean angular velocity of these waves is less than that of rotation of the mass by of the latter .
In the two waves determined by any sectorial harmonic , the relative motion of the iquid particles is irrotational .
The harmonics of degree 2 and rank 1 give rise to a kind of precession , of which there are two .
I have calculated the relative frequencies of several of the principal waves on a spheroid whose eccentricity is The question of stSbility is next dealt with , it being shown that in the present problem , in which the liquid forming the spheroid is supposed perfect , the criteria are entirely different from the conditions of secular stability obtained by POINCARE for the case when the liquid esses any amount of viscosity , and latter depend on the energy being a minimum .
In fact , for a disturbance initially determined by any harmonic ( proyided that it is symmetrical with respect to the equatorial plane , since for unsymmetrical displacements the spheroid cannot be unstable ) , the limits of eccentricity consistent with stability are wider for a perfect liquid spheroid than for one possessing any viscosity .
If we assume that the disturbed surface initially becomes ellipsoidal , the conditions of stability found by the methods of this paper agree with those of RIEMANN .
The case when the ellipticity and , therefore , the angular velocity are very small is next discussed , it being shown that all but two of the waves , or all but one of the oscillations for any particular armonic , become unimportant , their periods increasing LIQUID SPHE ROID OF FINITE LIPTICITY .
indefinitely .
In the case of those whose periods remain for a non-rotating spherical mass , the effect of a small angular velocity of the liquid is to cause them to turn round the axis with a velocity less than that of the liquid by Finally , the methods of treating forced ; are further discussed .
The general cases of a semi-diurnal\ldquo ; forced tide , or of permanent deformations due to constant disturbing forces , are mentioned in connection with some peculiarities they present ; and these are followed by examples of the determination of the forced tides due to the presence of an attracting mass , first , when the latter moves in any orbit about the spheroid , secondly , when it rotates uniformlv about the spheroid in its equatorial plane .
The effects of such a body in destroying the equilibrium of the spheroid where the forced tide coincides with one of the free tides form the condusion of this paper .
POINCAB for or of 2 .
Suppose a mass of gravitating liquid is iu relative when ting as if rigid about a fixed axis with angular velocity , and that it is required to determine the waves or small oscillations due to a slight disturbance of the mass .
Let the motion be referred to a se of orthogonal moving axes , of which the axis of is the fixed axis of rotation , while the axes of rotate about , with angular velocity .
In the steady or undisturbed motion the positions of the fluid particles relative to these axes will remain fixed .
In the , let be the small component yelocities of the fluid at the point to the axes .
The actual component velocities referred to axes fixed in space and coincidinoo with our axes of , at the time considered , will be , and the equations of hydrodynamics may be written* , , being the potential due to the attraction of the liquid and forces which may act on it , t , he pressure , and the density .
For small disturbances we may neglect squares and products of the relative velocities ( as is usual in wave problems ) , and , therefore , the tbove equations reduce to *BASSET , Hydro ( fynamics , ' vol. 1 , S23 ; or ] ENHILL , 'Encyclopaedia Britannica , ' article " " Hydromechanics . . . . .
( 1 ) , where . . . . .
( 2 ) .
We have also the equation of continuity , . . . . .
3 .
Eliminating from equations ( 1 ) , ( 3 ) , we obtain the differential equation . . .
( 4 ) , where , as usual , stands for LAPLACE'S operator 3 .
Let us now consider separately the ] harmonic oscillations of one particular period .
Assume that , and all vary as , so that the ratio of the period of oscillation to the time of a complete revolution of the liquid mass about its axis is equations ( 1 ) , ( 4 ) reduce to , ( 5 ) , ( 6 ) .
Put ( 7 ) , and ( 8 ) .
Equation ( 6 ) now becomes ( 9 ) .
LIQUID SPHEROID OF FINITE ELLIPTICITY .
If be greater than unity , and , therefore , also will be real .
We may t , ake to be the coordinates of a point corresponding to the point of the liquid .
We thus obtain a new region of points derivable from the original region by homogeneous strain parallel to or by projection .
This region may be called the auxiliary region , and the surface formed by points corresponding to points on the fluid surface , the auxiliary surface , Our problem thus reduces to that of finding a suitable value of satisfying equation ( 9 ) within the space bounded by the auxiliary surface .
But we must revert to the original system in order to satisfy the boundaryconditions , which must hold at the actual surface of the liquid , at the auxiliary surface .
If the surface of the liquid be free , .
be constant oyer it , and , therefore , the condition to be satisfied all oyer the surface of the liquid is const . . . . . .
( 10 ) .
In forming the expression for we must remember that the gravitation potential is due to the disturbed configuration of the liquid mass .
If be less than unity , will be imaginary , and , erefore , the auxiliary surface will also be imaginary .
But the results arrived at by this method in the where is real wil ] still hold good even if be imaginary , provided that the expression obtained for is a real function of the coordinates .
The method breaks down if , when vanishes ; this must be treated as a limiting case .
Solution for the Spheroid by Spheroidal 4 .
Let the liquid be in the form of a MACLAURIN'S spheroid the equation of whose surface is . . . . .
( 11 ) , so that . . . . .
( 12 ) , and is the eccentricity of the spheroid , being the radius of its focal circle .
The locus of the corresponding point is the auxiliary quadric ( 13 ) .
This quadric will be a prolate spheroid if lies between zero and , that is , if lies between unity and .
If is greater than , or greater than , the spheroid will be obla , te .
If be tive , or less than unity , equation ( 13 ) represents a hyperboloid of one sheet , but the part corresponding to the liquid surface the imaginary portion for which is less than cos is negative ; this is the imaginary auxiliary spheroid .
We shall take as our standard case that in which equation ( 13 ) represents a prolate auxiliary spheroid .
Let it be written in the form ( 14 ) , so that Solving , we find ( 15 ) , ( 16 ) .
solution of the differential equation ( 9 ) must be effected by means of spheroidal rmonics applicable to the auxiliary spheroid ( 14 ) , whilst the expressions for the gravitation potential of the liquid mass and the boundary-conditions will involve spheroidal hal.monics referred to the actual liquid spheroid ( 11 ) .
We must , therefore , use two different sets of orthogonal elliptic coordinates for the auxiliary and the actual systems .
Let these coordinates be denoted by respectively , and let them be colmected with the rectangular coordinates in the two systems by the relations and , therefore , ( 17 ) .
The surfaces of the spheroids will be given by the equations . . . . . . . .
( 18 ) , or . . . . .
; moreover , all over these surfaces , at corresponding . . . . .
( 19 ) .
LIQUID SPHEROID OF FINITE ELLIPTICITY .
The angular coordinate is the same in both systems ; but , except over the surfaces , will not be equal to , nor will any other two of the surfaces constant and constant coincide .
Put .
On transforming to , equation ( 9 ) becomes ( 20 ) , of which a solution , finite and continuous at all points within the spheroid ( 14 ) , is . . . . .
( 21 ) , where is any constant , and ( 22 ) , ( 23 ) , denoting the zonal harmonic of degree In our standard case is real and greater than unity , and in every case lies between the limits and \mdash ; 1 , and is real .
I have adopted the above notation ( according to which the functions differ in form by the constant factor in order to avoid introducing imaginary coefficients unnecessaril It is easy to see that the solution ( 21 ) is applicable in every case .
For , if be greater than , both and are purely irnaginary ; whilst , if be negative , we may show that will be imaginary , but will be real and less than unity .
In any case will be either real or purely imaginary , so that can be always made a real function of the coordinates .
Moreover , the right-hand side of ( 21 ) is finite , single valued , and continuous throughout the liquid spheroid , and satisfies the rential equation ( 6 ) .
It therefore only remains to investigate the boundary-conditions which must be satisfied by at the surface of the liquid .
5 .
The spheroidal harmonics referred to the liquid spheroid , required for these boundary-conditions , will be formed as follows:\mdash ; Let . . .
( 24 ) , ( 25 ) , tesseral harmonics may be replaced by the associated functions of the first kind of HEINE without any change in the formulae , the constant coefficients being supposed included in and let ( 26 ) , ( 27 ) .
Then the expressions ( 28 ) , . . . . .
( 29 ) , are soIutions of LAPLACE'S equation which are finite and continuous , the former throughout the interior of the spheroid , the latter throughout all space outside the spheroid , and vanishing at infinity ; while at the surface both expressions become equal to . . .
( 30 ) .
The Elevation of the Waves on the Surface .
6 .
Let be the normal displacement at any point of the liquid surface , i.e. , the height of the wave above the level of the undisturbed spheroid .
Let be the central perpendicular on the tangent plane , and an element of the outward drawn normal to the surface of the liquid spheroid ( 11 ) .
We readily find ' ( 31 ) , and at the surface , since the element is a tangent to the curve const .
, const .
; therefore , ; whence , . . . . .
( 32 ) , the differentials in ( 32 ) being total AJso , so that . . . . .
( 33 ) .
LIQUID SPHEROID OF FINITE ELLIPTICITY .
From equations ( 5 ) we obtain ( 34 ) .
Multiplying by , and adding , we find at the surface .
But is the normal velocity of the liquid relative to the moving axes , and is therefore equal to or to .
We have , therefore , ( 35 ) .
Now , taking as given by ( 21 ) , we have . . .
( 36 ) ; moreover , since by ( 23 ) where the symbol stands for differentiation with respect to , therefore ( 37 ) .
Also at the surface is equal to Hence , we find by ( 15 ) , ( 16 ) .
Whence where ( 39 ) ; moreover , the equation of the disturbed surface of the liquid is ( 40 ) , where .
( 41 ) .
The Boundary-Conditions .
7 .
Let be the potential of a mass of the liquid filling the spheroid . . . . .
( 18 ) , and let be the potential of a distribution of the liquid of thickness everywhere equal to over the surface of the spheroid .
The combination of these two distributions is equivalent to the liquid mass as disturbed by the waves , so that . . . . .
( 42 ) .
For the free waves , the boundary-equation ( 10 ) requires that const , . . . . .
( 43 ) , all over the surface ( 40 ) .
Now , S , are small quantities of the first order .
Hence , expanding by TAYLOR'S theorem , we have to first order LIQUID ROID OF FINITE ELLIPTICITY .
const .
( 44 ) , where the square brackets indicate that is to be put equal to In the case of forced tides due partly to small disturbing forces whose potential at any instant is , and partly to periodic variations of pressure over the surface of the liquid , the condition at the surface becomes const .
Equating to zero the non-periodic terms , we obtain the well-known condition for steady motion const . . . .
( 45 ) .
Here const . . . . . .
( 46 ) , where ( 47 ) , being the constant of gravitation , and being put equal to unity if the density is expressed in astronomical units .
From ( 45 ) we have , in the usual manner , ( 48 ) , whence . . . . .
( 49 ) , which can also be put in the form . . . . . . . . . . .
From , ( 45 ) the boundary-condition for the is ( 51 ) .
Now , by ( 48 ) .
( 52 ) .
8 .
To find v.\mdash ; Suppose that the values of this potential inside and outside the spheroid respectively are given by the formulae . . . . .
( 28 ) , . . . . . . .
( 29 ) .
Since are due to a surface distribntion of surface density , therefore , and , therefore , at the surface , by ( 30 ) . . . .
( 53 ) .
9 .
Lastly , in the forced oscillations , whatever be the variable conservative bodily forces or sulface tractions producing them , we know that it is always possible to expand the value of over the surface of the spheroid and at all times in a series of the form . . . .
( 54 ) , where is a constant , and the summations may extend to all possible values of but only to integral values of and .
The effect of each term may be considered separately .
To do this , let us take the case when there is a single term only , i. e. , take . . . . .
( 55 ) .
LIQUID SPHEROID OF FINITE ELLIPTICITY .
In the waves produced the values of will be the same .
Substituting from ( 21 ) , ( 52 ) , ( 53 ) , ( 55 ) in , we obtain ( 56 ) .
This equation , combine ed with ( 39 ) , suffices to determine the unknown constants in terms of the known coefficient , and thus the amplitude of the forced oscillation is determined in terms of that of the disturbing force .
10 .
The most interesting point is to determine , in order to find the height of the corrugations on the surface .
This plan has , moreover , the advantage that , in considering the effect of several disturbing forces of different periods , we may add together the elevations ( h ) due to the separate forces , whereas , in determining the value of , the terms having different periods are referred to rent auxiliary systems .
Substituting for in terms of from ( 39 ) in ( 56 ) , and writing , for brevity , . . . . .
( 57 ) , mass of spheroid . . .
( 58 ) , we find , after several reductions , the required equation for ( 59 ) , in which it must be remembered that ( 15 ) .
The Period-Equations for Free Waves .
11 .
If the oscillations of the liquid be free , we must put equal to zero in ( 59 ) , and we thelefore obtain ( 60 ) , , together with ( 15 ) , determines the admissible values of and .
In reducing ( 60 ) to a rational algebraic equation for we must distinguish three cases .
I. Let , and let be even .
Then we know that contain only even and odd powers of espectively , and , therefore , that is divisible by .
Multiplying ( 60 ) by , we find iting K for ( 61 ) .
MR. G. H. BRyAN ON THE WAVES ON A ROTATING Expanding and in powers of , substituting for by means of ( 15 ) , and ying the resulting equation throughout by , we find ( 62 ) .
This is rational algebraic equation in of the degree , involving only even powers of .
It is , therefore , satisfied by values of occurring in pairs corresponding to values of II .
Let , but leb ?
be odd .
Then is not divisible by .
Hence , we must multiply the equation ( 60 ) throughout by and obtain \mdash ; 4 ( 63 ) .
If this be developed in the same manner as in the preceding case , we shall obtain sin2 ( 64 ) .
This is satisfied by values of , but , as before , the positive and negative roots are numerically equal , so that there will only be different values of III .
Let be diflerent from zero .
Multiplying by the expression , we find a ( 65 ) , which reduces to the following equation in - .
- ( 66 ) .
2 .
LIQUID SPHEROID OF FINITE ELLIPTICITY .
This equation is of the degree , and involves both odd and even powers of .
It therefore has roots , but in the present case these roots do not occur in pairs of equal and opposite values .
Equations ( 62 ) , ( 64 ) , ( 66 ) are the period-equations of the various free harmonic waves or oscillations of the liquid spheroid .
Their roots depend on the value of or the eccentricity alone .
The periods of the waves are the corresponding values of and depend also on of the Oscillations and 12 .
The periodic movements determined by zonal harmonics and those determined by tesseral harmonics differ in character considerably .
The former are symmetrical about the axis .
Taking the solution another solution got by changiug the sign of is given by Compounding these , we get the real motions of the liquid determined by . . . . .
( 67 ) , being any constant .
These are of the liquid about the spheroidal form .
By what has already been shown , there are either or such free oscillations , according as is even or odd .
In all of these oscillations the expression for is of the same form , that is , the corrugations produced on the surface are similar in each .
But this will not be the case with the values of , because the auxiliary systems of spheroidal coordinates to which they are referred are different for each different value of Thus , the motions of the fluid particles in the interior of the mass are different for each of the oscillations .
13 .
Taking next the case when is different from zero , let us change the sign of everywhere that it occurs in our investigations .
The results will still hold good when this is done .
Hence , for every root of ( 66 ) we get two solutions of the equations of oscillation , giving respectively MR. G. H. BRYAN ON THE ways ON A ROTATING , tund also which )bine to give the real ( 68 ) These epresent a system of waves travelling round the axis of the spheroid with ] angular velocity .
But it must be remembered that the coordinate axes to which we have referred the way motion are themselves rotating with angula1 velocity .
Hence , the angular velocity of waves in space is .
According to our convention , positive values of give waves rotating more slowly thaIl the liquid , and vice versed .
There are such waves determined by harmonics of degree and rank and , since the values of are not equal and opposite in pairs , these waves do not combine into oscillations fixed relatively to the moving axes .
As in the symmetrical oscillations , the form of the corrugations is the same for all the waves , but the motion of the fluid pal.ticles different in each .
14 .
If . .
, be the roots of ( 66 ) , it is obvious that . . . . . . . .
( 69 ) .
Hence , the relative angular velocity of all the different harmonic waves of degree nd rank is ' in direction opposite to that of rotation of the liquid , whilst mean actual angular velocity in space of the Pe , riod-Equations .
15 .
From POTNCARE\ldquo ; S investigations it appears that the spheroid will be seculal .
lie stable , even if the liquid be viscous , provided that the coefficients which are here denoted by or LIQUID ROID OF FINITE ELLIPTICITY .
are positive for all values of greater than unity .
From our equations ( 49 ) , ( 50 ) we have , so that is essentially negative and positive .
In accordance with we shall now show that , if and be both positive , the roots of the period-equation for harmonic waves of degree and rank are all real , and we shall find their situations .
In the first place , let us suppose is different from zero .
The period-equation ( 66 ) , as it stands , may be written where NF , if we write for brevity We know that the roots of the equation . . . . .
( 70 ) are all real , and lie between and ; also they are separated by those of Let , . .
be the values of ( taken in descending order of magnitude ) corresponding to the roots of ( 70 ) .
These values of all lie between and , also decreases as decreases .
Moreover , if is put in turn equnl to , . .
, the corresponding values of are alternately positive and negative .
We are now in a position to trace the changes in as decreases from to When is greater than is imaginary .
But when written in the form of the left-hand side of ( 66 ) is obviously real ; also when the sign of is that of the coefficient of .
It is therefore positive .
When passes , hrough the value becomes infinite and then becomes real , but does not in general change sign .
When is positive , and is negative .
When is positive .
When is negative .
When is positive .
and so on ; thus , when has the same * We shall in future leave out the suffixes in and , using to denote the surface values , as these surface values alone occur in the remaindel of our investigations .
In general does not change when \mdash ; cosec , but when has the same as ) Hence , the equation must have one real root between each of the following values:\mdash ; , 1 , , . .
Thus , if is positive , all the roots of ( 66 ) are real .
Let us now examine what happens when vanishes and becomes negative .
POINCARE proves* that , if is divisibIe by , the equation has no real root ; we must , there , have even , so that is not divisible by When the equation reduces to of the roots , 1 , , . . .
, Since is ever ) , the equation has not zero for one of its roots .
Thus , the roots of the period-equation are all real and different .
Therefore , when changes sign and becomes negative , the period- equation must , at any rate at first , continue to have all its roots real .
If it have a pair of complex , the ratio of must not only be negative , but numerically greater than someJinite limit .
16 .
The roots of the period-equations for the oscillations that are symmetrical about the axis of the spheroid are to be separated in exactly the sa1ne way .
It will be sufficient to state the results here .
We suppose , . .
, are the values of which make . . .
, ( 71 ) .
Let be odd .
One of the above values , will be zero , whilst , and so on .
Also , the ratio : must be positive .
It will be found that the period-equation ( 64 ) has one root between each of the following of ; , . .
, , . . .
, * Acta Mathematica , ' vol. 7 , p. 326 .
LIQUID SPHEROID OF FINITE ELLIPTICITy .
If be even , the least positive and negative roots of ( 71 ) are and , also .
If the ratio : be positive , we find that the positive roots of the period-equation ( 62 ) are situated between the following values:\mdash ; , . .
, while the negative ones which are equal and opposite to them are situated in the intervals between , . . .
, there being no roots between and When vanishes , the roots are the quarltities , . .
, none of which is equal to zero .
If the ratio : now become negative , the roots of ( 62 ) will at first continue to be real , being situated between the values , . .
, , . .
, This will be the case until we arrive at a value of for which the period-equatiol has a pair of equal roots , each equal to zero .
When this is so , we have whence , ( 72 ) , or When the ratio or : becomes greater than two of the roots of the period-equation will become imaginary .
In every case there must be at least one positive root between each of the quantities , . .
, and corresponding negative roots , so that under no circumstances can equation ( 62 ) have more than one pair of imaginary roots .
Solutions of the 17 .
For a spheroid of given eccentricity , , and therefore , are known .
Now , the functions , can be expanded in finite terms of in exactly the same way its the ordinary spherical qrmonics , while , ?
can be expressed in finite terms of , i.e. of ; hence , the function can be calculated for any value of By method we may then approximate to the values of roots of the equation in in the ) cases .
The periods of the waves are corresponding values of , while is expressed in terms of by equation ( 50 ) To obtain some idea of the relacive frequencies of the various waves , I tabulated the values of thus calculated for harmonics of the second , third , and fourth degrees for a spheroid in which 1 or , the eccentricity being , there . .
The results are embodied in the accompanying Table .
As already stated , t , he positive oots correspond to rotating more slowly than the liquid , relatively in the direction opposite to that of rotation of the mass , while those having the double signcotTespond to symmetrical oscillations of the liquid .
TABLES of the alues of for Waves on a Spheroid whose Eccentricity I find that the period of the syrmetrical or zonal harmonic oscillation of the second degree ( in which the surface remains spheroidal ) is , in this spheroid , * ' ' On the pression of Spherical Harmonics of the Second Kind in a Finite Form Cambridge Philosophical Proceedings , ' December , 1888 .
See THOMSON and ' Natural Philosophy , ' vol. 2 , S772 .
LIQUID SPHEROID OF FINITE ELLIPTICITY .
of the corresponding time of oscillation in a llon-rotating spherica ] of liquid of the same density .
Harmonic .
When is numel.ical , and is zero ; thus , the periodequation reduces to . . . .
( 73 ) , of which the roots are given by . . . . .
( 74 ) .
The condition that these roots may be real is that . . . . .
( 75 ) , that is must be positive .
These results have been obtained previously by POINCARE in the special case in which , but in his investigation an extraneous factor has been introduced into the period-equation , giving a third root which does not properly belong to it .
The expression for in ( 21 ) is here proportional to that is , in Cartesian coordinates , to and is independent of Thus , the motion of the liquid is " " two dimensional and takes place in planes parallel to the equatorial plane of the spheroid .
By the laws of vortex motion the molecular rotation or spin of the actual motion of the liquid is therefore eyerywhere constant and equal to , being that due to the rotation of the liquid .
In other words , the wave motion of the iquid / to the rotating axes is irrotational .
Free Precession of the Spheroid .
19 .
Another case of some interest is when the harmonics the small periodic relative motions are of the second degree and rank .
Putting the period-equation ( 66 ) reduces to ( 76 ) .
Now , whatever the value of may be , is always a root of this equation .
For , if we put in the left-hand side , it becomes ( 77 ) , as was to be proved .
Substituting from the relation just found in ( 76 ) , the period-equation becomes or , dividing throughout by other two roots are given by whence . . . . .
( 78 ) .
The expression for is proportional to , that is , to sir ] while the height of the displacement of the surface is proportional to emembeI i that this displacement is so small that ibs square may be neglected , it can be readily shown by the usual methods of analytical geometry that , if , as is here supposed , the ellipticity of the spheroid be finite , the displaced surface is a spheroid of the same form and dimensions as the original spheroid , and can be obtained by turning the latter through a small angle about the line This will , however , no longer be true if the ellipticity of the spheroid is a small quantity comparable with the height of the small displacement , or the surface is spherical or nearly spherical .
In such cases it will be found that the displaced surface LIQUID SPHEROID OF FINITE ELLIPTICITY .
is an ellipsoid , differing in form from the original spheroid by ] quantities of the first order , whose axes make finite , not small , angles with those of the spheroid .
Suppose the liquid spheroid is rotating steadily about its axis of figure with angular velocity , and that this axis does not quite coincide with our fixed axis of , but is inclined to it at a small angle , while the axes of rotate about the axis of with angular velocity .
The coordinates of the fluid particles will now no longer be constant , but will undergo small periodic changes .
In the time both the fluid particles and the axes will come round to their original positions ; thus , the period of the relative oscillations is , although the liquid is in reality rotating steadily .
This accounts for the root , which occurs in the period-equation , result which may be completely verified by rigorous analytical methods , The movements corresponding to the other wo roots are somewhat similar to precession , the axis of figure of the spheroid turning about the axis of , to which it is inclined at a small angle .
Stability of Spheroid .
20 .
We have already alluded to POINCARE'S investigations of the condition that the spheroid , if viscous , may be secularly stable , which requires that the energy of the system for the given ] momentum must be a minimum in the spheroidal form .
The greatest eccentricity corresponds to the least value of which causes any one of the coefficients to vanish , and this is shown to be that given by , whence , as in THOMSON and TAIT ( S 772 ) , If the liquid be perfectly inviscid , the criteria are very different .
So long as the roots of the period-equations for the various waves and oscillations are all real , the spheroid cannot be unstable .
It will , however , become unstable if for any the equation in has a pair of complex or imaginary roots .
For , calling these roots , we get the possible surfa , ce cements compounding into the displacement which indefinitely with the time .
Let us imagine that our spheroid is subj e to constraints such as to freely allow of its sLlrface undergoing harmonic displacements of degree and rank , but which MR. G. H. BRYAN ON ] ] WAVES ON A ROTATING allow of no other displacements ( such constraints are , of course , purely theoretical ) .
The spheroid , if at all viscous , will be secularly stable or unstable according as or and we have seen that the latter condition can only hold if be even , that is , if the displacement be one symmetrical with respect to the equatorial plane .
The critical form is that in which But since , when changes , the roots of the period-equation at first continue real , the limits of eccentricity for which the perfect spheroid is " " ordinarily\ldquo ; stable are in every case wider than those consistent with secular stability if the liquid be viscid .
The critical form is determined by the condition that the period-equation must have a pair of equal roots .
In my paper on\ldquo ; The Waves on a Viscous Rotating Cylinder\ldquo ; I have endeavoured to further elucidate the difference between " " ordinary\ldquo ; and ' ' secular\ldquo ; stability .
Assuming the displacement from relative equilibrium to be proportional to is always complex for viscous liquid , and the condition that the disturbance may not increase with the time is that the real part of must be positive .
Both real and imaginary parts change sign when , the corrugations becoming relatively fixed and the liquid figure becoming a form of " " bifurcation Relative equilibrium is then critical .
But , if there be no viscosity , may be purely imaginary , as in the present case , when is real , and the ways will neither increase nor diminish in amplitude with the time ; thus , a change in ] of one of the roots of the period-equation mereJy implies a change in the relative direction of the wave .
Moreover , it appears that the criteria of ordinary and secular stability will be different only if the angular velocities of the waves be different in the two opposite directions , and this can only be the case if the liquid be rotating .
Reverting to the perfect liquid spheroid , the determination of the greatest eccentricity consistent with ordinary stability involves the question , if be gradually diminished , what is the harmonic displacement for which the period-equation of the waves first commences to have complex roots ?
It appears probable that this happens for .
With this assumption , we see , by ( 75 ) , that the critical value of is given by the , uation this leads to * Cambridge Philosophical Proceedings , ' 1888 .
LIQUID SPHEROID OF FINITE ELLIPTICITY .
whence I find by trial and error that .
This result agrees with that found by RIEMANN , who treated the problem as a special case of the general motion of a liquid ellipsoid .
It does not , however , seem possible ify the above assumption as to the nature of the displacements by a perfectly general rigorous .
The condition that the period-equation should have a pair of equal roots is far too complicated to allow of this point being fully , proved in the way that POINCARE has done for secular stability .
It is certain that the spheroid will be unstable for all values of greater than ; it is , but not certain , that it will be stable for all values less than this limit .
Spheroids of Ellipticity .
21 .
If the eccentricity of the spheroid , and , therefore also , its angular velocity , be small , the period-equations for the waves are much modified .
The value of will become very great , and we shall suppose it to be so great that is a small quantity .
thab can be neglected .
Since is small , we may put The function is proportional to since the other terms in it involve only and lower powers of .
Hence , to this approximation , ( 79 ) , and ( 80 ) ; moreover , ( 81 ) , whence , as in THOMSON and TAIT ( S771 ) , . . . . .
( 82 ) .
Firstly , suppose that the values of remain finite in the limit .
Then ultimately , and , since is negligible in comparison with , equation ( 65 ) gives . . . . .
( 83 ) , having real roots between 1 and \mdash ; 1 .
In the case of the oscillations symmetrical about the axis the equation for is ultimately or . . . . .
( 84 ) , according as is odd or even .
The frequencies of these waves or oscillations are proportional to the angular velocity of the liquid .
As the latter is diminished without limit they become relatively and finally cease to exist , for the limiting case of a mass of liquid without rotation oscillating about the spherical form .
Secondly , suppose the periods of the waves remain finite in the limit .
Put so that is the period .
Since remains finite , will increase without limit as diminishes , and , therefore , equation ( 66 ) gives , to the first order of smaJl quantities , whence , by ( 82 ) , ( 85 ) .
If we put , we get ( 86 ) , the welLknown result for the oscillations of a liquid sphere .
Denoting by the expression we find , whence , substituting in the small terms , we get .
( 87 ) .
Rememberin the expressions found in S 13 for the relative and actual angular velocities of the corresponding waves , this result may be stated as follows : effect of communicating a small angular velocity to a spherical mass of gravitating liquid will be to add an angular velocity to the angular velocities of all the free waves which are determined by harmonics of degree The symmetrical oscillations will be unaffected by rotation to this order of approximation .
If we proceed to a higher approximation by taking into account small quantities of the second order , the equations become much more complicated .
But , for spheroid similar to the Ear th , the above approximation would be practically sufficlent .
LIQUID SPHEROID OF FINITE ELLIPTICITY .
Forced Tides .
22 .
We now revert to the applications of the methods of this paper to the investigation of the tides produced on the surface of the spheroid by the influence of periodic variations of pressure over the surface of the liquid , or by the attractions of disturbiug bodies in the neighbourhood of the spheroid .
In this connexion , the equations found in SS9 , 10 will be required , viz. , if at the surface V ( 88 ) , and . . . .
then will be given in terms of by the relation ( 90 ) , where , as in ( 12 ) , ( 15 ) , and Also , from ( 39 ) , we have ( 91 ) .
( 92 ) .
The value of at any point of the surface being . . . .
( 98 ) , is determined in terms of , by the last equation ( 92 ) .
28 .
An interesting case occurs when .
The period of the tides will then be half that of a complete revolution of the liquid , and they may therefore be called " " semidiurnal\ldquo ; with reference to the spheroid .
Except in the case when equation ( 90 ) gives . . . . .
( 94 ) ; also , from ( 92 ) , , and , therefore , : hence , it is evident that must also vanish throughout the liquid .
MR. G. H. BRYAN ON ] ] ] ON A ROTATING The height of the forced tides is therefore the same as we sIlould get by the " " equilibrium theory i.e. , by ecting the small relative motions of the fluid entirely .
In fact , these motions have no effect on the height of the tides .
It does not llow that they do not exist ; in fact , it is evident , on the contrary , since the tides move relatively to the liquid , that they must exist .
But on referring to equations ( 34 ) we see that , when , the small relative velocity components may be finite , even though vanishes .
The zonal oscillations are , , given by ( 95 ) , since , and therefore .
For these oscillations does not vanish .
24 .
Another interesting application is to determine the height of the permanent corrugations produced by disturbing forces which remain constant and fixed relatively to the liquid .
We now have to take ; therefore , and If is different from zero , then , whether be odd or even , equation ( 90 ) gives us $ and ( 92 ) gives , whence ; and therefore is everywhere ero .
If and is odd , then is finite , and , as before , we find and Lastly , let and let be even .
This is the case of a harmonic disturbance which is symmetrical both about the axis of the spheroid and also with respect to its equatorial plane .
Then , as in S16 , approaches the finite limit when is diminished indefinitely , and therefore ( 96 ) , ( 97 ) .
In the first two cases the height of the corl'ugations is given by the " " equilibrium theory and , since , it follows from equations ( 34 ) that are all zero .
Thus , the fluid continues to rotate as if rigid in a form differing slightly from the LIQUID SPHEROID OF FINITE ELLIPTICITY .
rigin al spheroid , as we should most naturally expect .
But in the last case , since is Bnite , will be a finite relative motion of the liquid with respect to the moving axes .
That this must be the case ' be seen as follows : spheroid is supposed to be deformed from its original form by the action of the given conserva forces and surface pressures .
The displacement does not vanish at the equator ; hence , if we consider the fluid particles at the surface , forming a circle round the equator , the displacement must necessarily increase or diminish the size of this circle .
By THOMSON'S circulation theorem , the circulation in this circuit must remain the same as before , since the liquid is supposed perfect ; hence , the angular velocities of the fluid particles in this circle must be altered , and they can no continue to rotate about the axis of the spheroid with the original angular velocity .
Therefore the disturbance must produce permanent relative motions of the liquid , unless there be any viscosity present , in which case the mass will ultimately rotate as if rigid in the deformed figure , and the " " equilibrium theory\ldquo ; will again become applicable .
Tides due to Action of tellite .
25 .
We shall conclude by showing how to the forced tides due to the presence of a small satellite of mass revolving in orbit about the spheroid .
If we take to be the longitude of any point on the spheroid , measured from a plane fixed in space , with which the moving plane of coincides at time then , being the longitude measured from thelatter plane , we have . . . . .
( 98 ) .
Let be the spheroidal coordinates of the mass at time being measured from the fixed imtial plane .
Then , at any point whose distance from the mass is , we have . . . . .
( 99 ) .
Since can be expanded in spheroidal harmonics by the formula ( 100 ) .
Since the motion of the satellite is supposed known , ) are known functions of .
In order to complete the solution we must suppose the quantities , and expanded by FOURIER'S theorem in simple harmonic functions of the time .
If the period of the satellite in its orbit be , the expansion will only involve circular functions of multiples of .
We then write for , and resolve all products of sines and cosines involving or in terms of circular functions of sums and differences .
The ression V will now be triple series of the same form as in equation ( 88 ) , and the of the tides due to each term of the series may be determined separately by the application of equations ( 89 ) , ( 90 ) .
26 .
Suppose that the attracting mass rotates round the spheroid in the equatorial plane in a circle of radius with angular velocity ( referred to fixed axes ) .
We then have .
Therefore .
We have , ( odd ) , , ( even ) , odd , even , so that the ] ) ansion only involves harmonics which are symmetrical with respect to the equatorial plane .
The first terrn in the above expansion is a harmonic of the first degree and rank , and determines the motion of the spheroid as a whole about the centre of mass of the spheroid and the attractin body .
This motion can be taken separately .
Taking , in the usual way , . . .
( 102 ) , where the summation includes only even values of , we find , by the method of , ( even ) ( 103 ) ; also , if be even , ( 104 ) , LIQUID SPHEROID OF FINITE LLIPTICITY .
where , in this case , ( 105 ) , the corresponding value of being ls .
Equations ( 102 ) , ( 103 ) , ( 104 ) fully determine the height of the forced tides and deformations on the surface of the spheroid due to the attracting body .
If , instead of one attracting mass , we suppose two equal masses on opposite sides of the body , the expression for , and therefore for , will contain only zonal harmonics , and harmonics of even rank , but these will be the same as before .
Harmonic Tides of the Second Orde ?
: 27 .
If the body be very distant from the spheroid , will be great , and the series ( 101 ) will converge vely rapidly , so that the harmonics of the second are the most important .
Taking , we get ( 106 ) , and the corresponding term in the height of the forced tide is If the attracting body be rotating in the same direction the liquid , but with less angular velocity ( as in all cases of astronomical interest ) , lies between and 1 .
If , in addition , is positive , the denominator in the expression for cannot vanish , and the tide produced by this term cannot therefore become very large .
In fact , the angular velocities of both the free harmonic tides will lie beyond the above-mentioned limits , and will neither of them coincide with that of the attracting body .
If , however , , and if , in addition , the attracting body be fixed in space , so that , this forced tide will increase indefinitely .
The spheroid will now have a semidiurnal free tide , which will be fixed in space , and will coincide with the forced tide due to the ttracting body .
The in the spheroidal form will therefore be completely broken up .
If the racting body be rotating very slowly about the spheroid , the same thing will happen if has a certain corresponding small negative value .
Since the spberoid is secularly stable or unstable according as is positive or negative , the mode in which its relative equilibrium will be destroyed if becomes negative will depend on circumstances .
If the liquid possess but little MR. G. H. BRyAN ON THHi ways ON A ROTATING viscosity , the clJanges due to secular instability will take place very slowly , but the efl'ect of tide generating force due to an attracting body when the angular velocity of one of the free tides coincides with that of the body will become very great .
If , ever , the viscosity be considerable , it will prevent the forced tides from becoming large , and will cause the liquid to rapidly assume the form of a JACOBIAN ellipsoid , to the spheroidal form being secularly unstable .
clusion .
28 .
The results of the present paper suggest several considerations , which may possibly throw further light on the past history of the Solar system .
The criteria of stability applicable to the two cases where the spheroid is formed of perfect and of viscous liquid respectively have been are , ady discussed in S 20 .
In the last paragraph I have alluded to the possibility that equilibrium in the spheroidal form may be broken up by an attracting body which causes the harmonic tides of the second order to increase indefinitely , in accordance with Professor DARWIN'S ypothesis .
The present analysis , however , suggests that the same thing may happen in the case of harmonic tides of higher order than the second , and , moreover , the results arrived at concerning the lmber and situation of the roots of the frequencyequation render this hypothesis quite admissible .
Except for harmonics of the second order , many of the free tides will rotate in the same direction as the liquid , but with less angular velocity , even though the spheroid be secularly stable ; and , if an attracting body shou be rotating about the spheroid with the same angular velocity as one of these tides , they would certainly rise to an enormous height , and the liquid might , perhaps , ultimately be broken up into two or more detached masses .
Take , for example , the sectorial harmonic waves of order .
If one of these be fixed in space , we must , from S13 , have the corresponding yalue of , and , by ( 73 ) , this leads to the condition .
If is greater than 2 , this will be satisfied for some secularly stable form of the spheroid .
Under these circumstances , the presence of a fixed attracting body near the spheroid would cause the sectorial harmonic waves of the ?
order to increase indefinitely .
The only obstacle in the way of the present supposition is that when the distance of the attracting body is at all considerable , the harmonic components of tide force of the higher orders become very small in comparison .
Another question of astronomical interest is whether a rotating spheroid can be broken up into one or more ?
of rotating liquid .
This can only happen if one of the zonal harmonic oscillations increase indefinitely or become unstable .
LIQUID SPHEROID OF FINITE ELLIPTICITY .
Now , as long as the spheroid continues stable for such zon al harmonic displacements when the liquid is supposed perfect , the frequencies of these oscillations will none of them vanish ; hen ce , their amplitude cannot be increased indefinitely by the attractions of bodies remaining in the equatorial plane of the spheroid ; and the only way in which this can take place is under the influence of the tide force due to a satellite whose orbit is inclined at a considerable angle to the equatorial plane of the spheroid , the effect being greatest if their planes be perpendicular .
There is still the possibility that , contrary to our hypothesis in , a perfect liquid spheroid may first become unstable for some displacement symmetrical about the ax is ; and , unless this question be ] decided , we not in a position to say that such is not the case , and that LAPLACE'S hypotbesis is wholly unfounded .

