]\gt ; INDEX SLIP .
JE , J. H.\mdash ; On the VIbrations and Stability of { ating Planet .
Phil. Irans .
, , vol. 201 , 1903 , pp. , igure of the .
JEANS , J. H. Phil. Trans. , , vol. 201 , 1903 , pp. 157-184 .
Elastic Solid\mdash ; Vibrations and of a Gravitating Spherical .
JEANS , J. H. Phil. ans .
, , vol. 201 , 1903 , pp. 167-184 .
Gravitational Instability , as a factor in the Evolution of Solar System .
, J. H. Phil. Trans. , , vol. , 1903 , pp. 157-184 .
Planetary Systems , Evolution of .
JEANS , J. H. Phil. , vol. 201 , 1903 , pp. 157-184 .
V. On th Vibratio Stability of a Grauitaliw Planet .
J. H. of Collegc , G. H. , F. ovembcr 8cceml ) , 1903 .
S 1 .
IN a former ' I have considered the effect of gravitation as a factor tending towards instability , the case of a spherical ebula of gas .
The object of the present paper is to investigate the problem in the case of a spherical planet , the planet bein supposed composed of solid or fluid matter .
The main question at issue is the following .
S2 .
So long as gravitation is neglected there can be no doubt as to the stability of an elastic solid ; any displacement increases the potential energy , and an unstressed configuration of equilibrium is therefore necessarily stable .
But when grayitation is taken into account , the gravitational energy may be either increased or decreased by a displacement from equilibrium , and if a displacement can be found which effects a the gravitational potential of ount sufficient to outweigh the increase in the potential of the elastic forces , then the configuration will be unstable .
Now , in S 2 of the preyious paper already referred to , it was shown that for any spherical body displacements can be found such that ) is a decrease in the potential .
is sufficient to prove that a spherical configuration of equilibrium may be unstable .
In the terminology of POINCABfi it appears that on any " " linear series\ldquo ; of spherical configul.ations there may be " " points of bifurcation We must , therefore , attempt to settle the of these points of bifurcation .
In particular , it will be of rest to examine whether a sphere of the size and material of the earth may be regarded as being aIlywhere in the od of a point of bifurcation .
' ' The Stability of a ) \ldquo ; ' Phil. ' , p. 1 .
' Acta , Math vol. 7 , p. 259 .
33 , 5 .
4.4.03 .
J. ON ATIONS ANT ) S 3 .
A rough and very simple tion will give an approximate answer to this latter question .
Let be the radius of a sphere , which will ultimately be taken to be the earth , its , and the mean density given by Let us nse the elastic constants and let be the mean value of .
Since the sphere is supposed to be spherically symmetrical , , and will be functions of the single -ordinate , the distance from the centre .
Imagine , and each expressed as functions of let , .
be the coefficients which occur in these functions ; coefficients being mere numbers and independent of the system of units in which , and are measured .
Imagine a linear series of equilibrium configul.ations obtained by varying any one of the quantities , or , while keeping the remaining two quantities and the coefficients , . .
constants .
The points of bifurcation on this series will occur when the varying becomes equal to some definite function of the remaining quantities and of , the gravitational constant .
Hence , however the linear series are al.rived at , the points of bifurcation will be given by an equation of the form . . . . .
( 1 ) .
Now the coefficients , . .
are mere numbers , and the only way in which , and can be combined so as to give a mere number is through the term .
Hence equation ( 1 ) can be expressed in the form ( 2 ) .
We have seen that the spherical configuration must be unstable for some values of , and always unstable , hence equation ( 2 ) must have at least one real root between and .
Let the lowest root be . . . . .
( 3 ) , where is a function of , . .
; then a spherical configuration is stable so long as , and becomes unstable as soon as The coefficients , . .
will , on the average , be comparable with unity , because are referl.ed to their values ; they are as likely ( speaking somewhat loosely ) to be above as to be below unity .
Hence itself will be comparable with unity , and * The notation is that of 'Theory of Elasticity .
' The of THOMSON and TAlT are given by STABILITY OF A GRAVITATING PLANET .
it is not at present possible to say whether it is more likely to be greater or less than unity .
S4 .
Now , in the case of the earth ( THOMSON and TAIT , S838 ) , we have centims. , and the value of in C.G.S. units is known to be This gives for the value whence it appears that for a sphere of the size and mass of the earth the spherical configuration will be unsbable unless has a comparable with Now for steel ( cf. THOMSON and TAIT , p. 435 ) the values of the elastic constants in absolute units are ] ] ence .
We therefore see that the critical values of the elastic constants in the case of the earth are comparable with those of steel .
The foregoing calculation is , of course , very rough , but it shows that the critical values for the earth are at least in the neighbourhood of what must be supposed to be the actual values , so that we are driven to mpting a more accurate determination of these values .
If the view of the present paper is sound , this approximate equality is more than a mere coincidence ; we shall see that it could have been predicted from our hypotheses of planetary evolution .
We now attempt a rigorous mathematical investigation of certain problems which have a bearing upon the astronomical questions in hand .
Those readers whose interest lies in the application of the results rather than in the processes by which they are obtained may be recommended to turn at once to S22 .
THE A GRAVITATING ELASTIC SOLID .
The Equations of Small Vibrations .
S5 .
We shall begin by discussing the principal vibrations and the frequency equation of a spherically symmetrical solid .
The case of a non-gravitating has been fully discussed by Professor but the inclusion of the gravitational terms , as will be seen later , brings about a considerable complication in the analysis .
The case of a gravitating but incompressible sphere has been considered by BROMWICH , but this has no bearing on the present problem , in which the whole " " On the vibrations of an Elastic Sphere ' Proc. Lond. Math. ' vol. 13 , p. 189 .
" " On the Influence of Gravity on Elastic Waves , \amp ; c ' Proc. Lond. Math. Soc vol. 30 , p. 98 .
interest turns upon the compressibility .
The solution which follows is , in its main points , very similar to that of Professor LAMB , so that I have not thought it to give the steps of the argument in great detail .
From the symnletry of the solid it follows that the elastic constants , and the density , will be fUnctions of the single co-ordinate , the distance the centre .
o1igin , we shall use rectangular -ordinates , , and shall the solid to execute small vation , such that the ) acelnent of the element itially at , has components , .
The component of displacement along the radius will be denoted by and the dilatation by , so that S C. After the density is , or , since is a function of only , Hence potential , is changed by displacement from V into V is the potential of the following distribution of matter:\mdash ; A volume distribution of density ( 4 ) , ( ii .
) A surface distribution of which the surface density is . . . . .
( 5 ) , this ) eing taken ovel surface at which the density changes abl.uptly , the change frol1l to in crossing surface in the direction of increasing .
In particular will occur at the outer surface of the solid , the value of in this case being zero .
* S7 .
The after )acement being , that at will be *In tho tions on ting spheres given in } and TAIT'S ' Natural Philosophy , ' the course of procedure is tantamount to neglecting the volume distribution ( 4 ) , and regarding as the potential of a surface diqtribution ( 5 ) alone .
For this reason the result obtained ditfers from that of tho present paper .
STABILITY OF Hence the force at in the direction of increasing , found by differentiating the foregoing expression with respect to , is , ectino .
squares of the displacements , ( 6 ) .
Let us suppos3 that , in addition to its gravitation , the sphere is acted upon by an external field of force of potential let us , in the usual notation , denote the six stresses by , U. Then the equations of motion of the element at in the displaced .
are three of the form in which W , and all the terms such as , . .
are uated a ) , but . .
are calculated in the displaced configuration at S 8 .
No the only case in which we have any accurate as to the values of is when hole strain is snnall , when is small .
In the case of the earth , V is not , in this sense , small .
* The way in which we can proceed with any certainty is , therefore , by taking is to say , we must artificially annul gravitation in the equilibrium confignration , so that this equilibrium configuration may be completely unstressed , and each element of matter be in its normal state .
In this case it seems justifiable to suppose both the density and rigidity to be constant throughout the sphere , and , indeed , it is only with the help of this simplification that the equations become at all manage The vibrations of this system will be of kinds .
First there are " " sphel.ical\ldquo ; vibrations in which the displacement is purely radial at every point , so that the solid remains spherically symmetrical after displacement , and , secondly , there is larger class of vibrations in which the displacement is not of this simple type , so that the displaced tion is not one of spherical symmetry .
We hope , by discussing the vibrations of this system , to obtain some into the corresponding vibrations of a natural non-homogeneous solid , say the earth .
Now it is extremely doubtful whether the spherical vibrations of our artificial system ] much in common with those of the natural system , but it will be seen later that is of no importance .
We shall not be in any way concerned with these vibrations .
What we shall require is a knowledge of the unsymmetrical vibrations , and this , it is hoped , can be obtained with fair accuracy from a consideration of the corresponding vibrations in the artificial case .
There must be some uncertainty even in the case of unsymmetrical vibrations , and , unfortunately , this seems to be levitable ; our Lovr ' sticity , ' I. , p. 220 .
VOL. CCI.\mdash ; A. MR. J. H. JFANS ON VIBRATIONS case appears to be the only case in which the equations can be solved by ordinary analysis .
We now replace by their ordinarily assumed values , and equation ( ) , putting , takes the form ( 8 ) , and there are two similar equations for Principa7 S lferentiate these three equations of motion with respect to and add ; then ( 9 ) .
Now , from ) definition of we have , in the case in which is constant , . . . . .
( 10 ) , and hence equation ( 9 ) becomes ( 11 ) .
If we suppose proportional to pt , this equation assumes the form , where . . . . .
( 12 ) .
There is , therefore , a particular solution of ( 11 ) of the form . . . . .
( 13 ) , where is a surface harmonic of order , and the general solution found by summation of solutions of this type is . . . .
( 14 ) , where the summation extends over all possible harmonics , and over all values of It will appear later that each term in this solution can be made to satisfy the boundary conditions , and , therefore , that each term represents a normal vibration .
The vibrations may , therefore , be classified into vibrations of order , 1 , \amp ; c. , the order being that of the harmonic which occurs in the expression for .
The vibrations of order are the spherical vibrations already referred to .
STABILITY OF A GRAVITATING PLANET .
We shall assume this provisionally , in order to avoid the continual repetition of double summation , and now proceed to evaluate and to form the boundary equations for the simple vibration given by equation ( 13 ) .
S10 .
From equation ( 8 ) it appears that the displacement is given by ( 15 ) .
The solution is . . . . .
( 16 ) , where is solution of \mdash ; . . . . . .
( 17 ) , and is the most general solution of . . . . .
( 18 ) .
It can easily be verified that a solution of equation ( 17 ) is ( 19 ) .
There will be solutions simil to ( 1G ) , but the three solutions for must be such that . . . . . .
The left-hand member of ( 20 ) is , from ( 16 ) , and from and ( 17 ) , .
Hence ( 20 ) is satisfied if ( 21 ) .
S 11 .
Write for as before , and for Then we shall verify that the solutions for and are . . . . .
( 22 ) , in which are functions of , as yet unknown .
Assuming these solutions , the value of , calculated as explained in S6 , is ( 23 ) , where denotes .
J. TIIE the value of the which occur in this expression , and the sum of the first terms inside the curled brackets is found to be .
JIence we may write in the form where ( 24 ) , We now have , from equation ( 16 ) , , hence .
S12 .
There are three boundary-conditions to be satisfied , expressing that the normal pressure and he two tangential tractions shall vanish at every point of the surface .
As LAMB shows , these may be represented by three symmetrical equations , to be satisfied at the surface , each of the type for and from and this becomcs . .
Write so that the right-hand members are solid harmonics of degrees and ; , .
191 .
OF , From these identities it is clear if the terms in ( 27 ) do not depend on or are expanded in spherical harmonics , they will contain no harmonics othel than and .
We therefore see that the form of may be assumed to be ( 28 ) , where $ and \copy ; are functions of .
The value of is ( 29 ) , whence . . . . . .
( 30 ) .
S 14 . .
for in ( 27 ) and equating the coefficients of co and , we obtain the two following equations which must be satisfied at ( 31 ) , and a second of similar kind , of which the first line can be obtained from the first line of the above by writing for , and the second line is ( 32 ) .
The expression hich occurs in cul.led brackets in ( 31 ) can be sformed into ( 33 ) , while the corresponding expression in ( 32 ) is seen to From the value of by equation ( 24 ) , , Hence expression ( 33 ) becomes of which the value at is .
This is the value at of the term which occurs in curled brackets in equation ( 31 ) .
The value of the similar term in ( 32 ) , namely expression ( 34 ) , is seen to be ( 35 ) .
Write ( 36 ) , , ions ( ud econle , ( 38 ) , ( 39 ) .
equation ( 2G ) , ritc then last biun becomes Now , equations ( 38 ) and ( 39 ) become .
( 40 ) , .
( 41 ) , in which must be put equal to STABILITY OF A GRAVITATING PLANET .
S 15 .
Now is known to be a solution of equation ( 18 ) , and hence , from equation 28 ) , we may assume ( h.r ) , , ( hr ) , where . . .
( 42 ) .
It follows that at we may write where ( 43 ) , , ( 44 ) .
Hence , from ( 30 ) , we have at ( 45 ) .
Lastly , we have from ( 28 ) .
In order that ( 21 ) may be satisfied , we must have ( 46 ) .
Substituting for and , we find that this is satisfied for all values of if ( 47 ) , i. e. , if we have , at . . .
( 48 ) , where ( 40 ) .
Equations ( 40 ) and ( 41 ) now assume the forms . . . . .
( 50 ) , $ . . .
( 51 ) , in which must be put equal to The general frequency equation may be found at once by the elimination of $ , the values of and being given by equations ( 36 ) and ( 37 ) . .
J. H. ON THE ANY ) of $ 16 .
The interest of the question lies in the position of the points of bifurcation ; to find these we must put in the frequency equation .
The reason why it was not possible to put earlier stage will be understood by those who have read the former paper " " On the Stability of a Spherical Nebula In the present instance it is , perhaps , sufficient to say that putting at an earlier stage would have led to an entirely misleading result .
Upon ) utting p in equations ) we find that the two brackets multiplying $ vanish , and we therefore see that $ must be treated as an infinite quantity of the order of Expanding these brackets as far as , and then 'utting , we find that the two equations become , ( 5 ) , where The equation giving points of bifurcation is , of course , . . . . . .
( 54 ) .
The values of are found , after some , to be .
( 55 ) , .
Now , it has already been seen that ( p. 1 ( i4 ) .
substitute this value for , write for , and simplify equations as far as possible .
we have ( 57 ) , .
( 58 ) , : A ATING while the value of may be written in the form .
( 59 ) , ( ) .
The equation giving points of bifurcation can now be found by substituting these values in equation ( 54 ) .
S 17 .
This equation will have roots corresponding to the different ) ' values of , 1 , 2 . .
; these determine points of bifurcation such that the vibrations of orders , 1 , 2 .
respectively .
Of these the points of bifurcation of zero order are of no importance .
The reason is exactly similar to that given in the case of a spherical nebula ( S2S of the paper already quoted ) , namely , that a point of bifurcation of order does not indicate a depal.ture from the spherical shape .
We therefore will only discuss values of different from zero .
of S 18 .
discussing the general form assumed by equation ( 54 ) , it will be well to consider the simple case of .
Putting , we obtain from equations ( 57 ) and ( 58 ) .
to equations and ( 53 ) we see that the equation giving points of bifurcation is . . .
( 61 ) .
The oots of various , , \amp ; .
, the roots continually ' with .
Thus the point of cation is iven by , the vibration is of ordel of Zero .
S J. The genel.al equation in which is not put equal to zero is much more complicated than equation ( 61 ) , which has just been considered .
If we , it will be seen that the equation giving points of bifurcation of order is of the algebraic function of and of V0L . .
J. H. JEANS ON THE ATIONS AND To oxinlate numerical solutions , my plan has been to of the functions in this way obtain a solution of the equations for rent values of .
There is no difficulty in drawing graphs of the functions ; these are functions , we have the successive connected by the elation save essed all details of this somewhat tedious of the .
The results for are given in the following LOWEST Values of For large values of it will be found that equation ( 54 ) reduces approxinlately to , and hence that for any value of the lowest value of is slightly less than the value in the case in which Tlw First Point of .
It therefore appears that the first point of bifurcation may be safely assumed to be of order .
The value of for which it occurs will have some value between and , according to the value of .
Now , and the value of is .
Hence the first point of bifurcation is given by In equation ( 3 ) we this poin of bifurcation to be given by present analysis we have already taken ; if we take to be identical with .
former , we see that the actual values of , wheu , when *It be found that the first point of bifurontion is given , with accuracy , by the single equation for all values of between and .
This is of interest , as showing the relative importance of and in maintaining stability .
As might be foreseen , the importance of relatively to increases as increases , and for , the factor must be replaced ) STABILITY OF A GRAyITATING PLANET .
We have now found a closer approximation to the value of than that which was given in S3 , and have obtained the additional information that instability first enters through a vibration of order .
It must , however , be borne in mind that these results are only true of the special and somewhat artificial case specified in S8 .
with the of Spherical S 21 .
It will be seen that the general argument of S 3 will apply to the case of a gaseous planet or nebula if be taken to mean the pressure in the gas .
In this case , however , the laws of distribution of density and pressure are not independent .
If the gas is in conductiye equilibrium throughout , the planet or nebula must be supposed to extend to infinity , and for these conditions the criterion of stability was worked out in the former paper already referred to .
Calling the elasticity of the gas , the first point of bifurcation was found to be reached when the function attains a certain finite value .
vanishes in comparison with the mean density , so that for the radius of the nebula , and for the mean pressure , we have , at this first point of bifurcation Comparing this with the general result obtained in S 3 , we see that in this extreme case the value of becomes infinite .
This result is only of importance to the present investigation as showing the tendency of a concentration of density about the centre .
It seems to show that as the density becomes more concentrated about the centre , the value of may be expected to increase .
We are therefore led to expect that in general will have a value rather greater than that found for it upon the assumption of homogeneity of density .
RECAPITULATION AND DISCUSSION RESULTS .
S22 .
It will be well to recapitulate our results before attempting to draw any deductions from them .
We consider a spherically symmetrical mass of solid , liquid , or gaseous matter .
We denote the radius of this by , the mean density by , and the .
mean value of by , where denotes an elastic constant or the pressure of the fluid , according as the matter is solid or fluid .
We have seen that the stability of this dynamical system depends upon the value of the function , a pure number .
When when we deal with artificial matter which is totally devoid ol gravitation ) there can be no doubt that the system is stable .
We have seen that a point of bifurcation occurs when the number has a certain value .
It has not been proved in the present paper that an exchange of stabilities accompanies this point of bifurcation , but it will be seen that , with slight alterations , the proof of the exchange of stabilities for the spherical nebula , which was given in S28 of the earlier paper , can be made to apply to the present case .
Admitting this , it appears that the spherical system which is at present under discussion will be stable so long as is less than an becomes unstable so soon as exceeds S 23 .
The next question is as to the exact value of , and as to the vibration through which instability enters at the point of bifurcation .
To the first part of the question we not been able to obtain a very definite answer .
This matters the less , since numerical data which would have to be used in making any applications of our results are not themselves very definite .
On the whole , the uncertainty in the value of is not much greater than the uncertainty in the value of the numerical data ( or , what comes to the same thing for our present purpose , .
the uncertainty in our knowledge of the law of compressibility and distribution of density in the planets of our system ) .
The general argument of S 3 showed that must , except in extreme cases , be comparable with unity .
We then examined an artificial case : a planet in which the density and elasticity were constant throughout\mdash ; this system bein made mechanically possible by introducing an external field of force , of amount just sufficient to annul in the equilibrium configuration .
For this system was , of course , taken equal to , the uniform density , and was taken to be equal to in the notation of LOVE , or in the notation of THOMSON and TAIT .
The value of depends , of course , on the ratio or .
For we found ; for we found ; for intermediate value of we saw that the value of was intermediate between these two values .
The planets to which we wish to apply our results do not possess uniform density : it is almost certain that in every case the mean density.is much greater than the surface density .
The general argument of S3 shows that there is still a point of bifurcation corresponding to a value of which is comparable with unity , but affords no evidence as to the change which a concentration of density will effect in the value of .
We therefore examined a case in which there is an infinite concentration of density\mdash ; the case of a spherical nebula extending to infinity\mdash ; and found that in this extreme case the value of becomes infinite .
It therefore seems probable that a concentration of density is attended by an increase in the value of .
As a working hypothesis we shall assume for the planets of the solar system the uniform value .
It must be left to the reader to form ajudgment as to the amount of error involved in this assumption , but it will , perhaps , be admitted that results depending upon it will at least be as regards order of magnitude .
It will be seen later that considerable variation in the value of is possible before the astronomical evidence we are going to brmg forward is seriously invalidated .
S24 .
As regards the nature of the vibration through which instability of the spherical configuration enters , we are able to colne to a more definite conclusion .
In STABILITY OF A GRAVITATING PLANET .
each of the cases referred to in the last section this vibration is found to be one of order , i. e. , one in which the displacement at every point is proportional to the first harmonic .
This is the result which we should naturally expect\mdash ; just as we expect a mass ofliquid to become unstable through long surface waves sooner than through short ones .
We shall , therefore , suppose it to be true of the planets in general .
It is conceivable that planets could be artificially constructed for which this assumption would not be true , but , at present , since we have not complete knowledge of the structure of the planets : are therefore compelled to make some assunptions , it seems as if the assumption just made is far and the best to tnke as a working hypothesis .
TO THE HyPOTIIERTS .
6 .
25 .
In the former paper , already referred to , the suggestion has been put vard that the instability of a nebula , sun or planet , which , upon the nebular hypothesis , is supposed ultimately to result in the ejection of a satellite , may be largely brought about by a gravitational tendency to instability of the kind we have been investigating .
Let us , for the moment , take an extreme hypothesis , and imagine that this agency is the preponderating agency , and that the rotational tendency to instability may be disregarded in comparison .
Upon this hypothesis let us consider the case of an approximately spherical planet or sun which is known to have thrown off a satellite .
Before the ejection of this satellite commenced , the primary mass would have an approximately spherical for which ) would be below the critical value .
When this critical value is reached , a divergence from the spherical fornl occurs , and a series of new processes begins .
We are not now concerned with the details of these processes , but they must be supposed ultimately to result in the ejection of a satellite .
It must be noticed that we are not supposing the pri mary to be devoid of rotation\mdash ; for this would be inconsistent with the ejection of a satellite\mdash ; but are supposing the rotation to be so small that the rotational tendency to instability is small in comparison with the gravitational .
If we suppose one or more satellites to have been ejected , and the primary to have regained an approximately spherical form , new value of must be less than .
Now every satellite of which we have any knowledge , in our own system at any rate , is small in comparison with the primary .
A legitimate inference seems to be that the ejection of a satellite is only a small part of the life-history of the primary .
We shall not , however , need to make any assumption so definite as this , but shall.suppose only that the values of for the imary after ejection are nearly equal equal , except multiplying factor to the values of the same quantities before the process of ejection commenced .
All this may be summed up , with sufficient accuracy for the present investigation , in the statement that wheTl the ejection of a satellite is taking place the value of must ) nearly equal to : Tested by the .
To a certain extent the sohr system supplies material for testing this conclusion .
Let denote the ent mean density and radius of any member of the system which is known to have thrown off a satellite , and let be the mean value of , whether this denotes a pressure or an elastic constant , then we may tyrite At the time at which the satellite was ejected the same equation ought to hold with nearly equal to unity .
Unffortunately we have no knowledge as to the changes which will have taken place in since the ejection of the last satellite .
We shall , for a rough approximation , neglect these changes altogether , take , and examine to extent the equation . . . .
( 62 ) holds for the solar system as it now stands .
will be remembered that there are three sources of error in this equation:\mdash ; We are neglecting the effect of rotation in bringing about the ejection of a satellite .
( ii .
) We are neglecting the changes which have taken place since the ejection of the last satellite .
( iii .
) We are using an arbitrarily chosen value for , and applying this to every planet , while we know , from the difference in the physical constitutions of the planets , that the value of must be different for each .
These three sources of error would each be serious if we were attempting to get accurate results , but as our calculations are necessarily only of the roughest kind , we may be content to neglect them .
S27 .
In the accompanying table the masses and radii of the sun , and of those planets which possess satellites , are given in the first two columns ; the value of mass/ ( radius)2 is given in the third .
The units are so chosen that the earth is measured by unity in each of these columns .
Venus is included for the sake of comparison , although the existence of a satellite is extremely doubtful .
STABILiTY OF A AVITATING P If our hypotheses a fair account of the facts the numbers in this colunlll will be proportional to calculate the actual values of , and these are give in the fourth column .
S28 .
Knowing nothing about the variation in , we shall be content preliminary hypothesis to suppose it to have the same value for each planet .
Combining this with the hypotheses already formulated , we llotice to have the same value each planet , as therefore ought also the function mass/ ( radius)2 , which is tabulated in column ( 3 ) .
It will be seen at once that there is a certain anlount of mity tl nnmbers in this column , but it equires some considel.ation to determine much significance is to be attached to this uniformity .
Now , apart any hypothesis as to how the solal system ated oched its present configuration\mdash ; i. , regarding the solar system as a fortuitous collection of bodies of varying sizes\mdash ; we should expect the mean density to be greatest in the greatest planets .
We should , therefore , expect the quantity mass)adius ) to be more variable thau the radius .
other words , we should , priori , exlJect less uniformity in the third column than in the second .
Judged by this , the uniformity of the numbers in the column would bo very there , the variation in these numbers is of the kind we should exlJect .
, it is known that the density of Jupiter is very much greatel the centre than neal the sul.face ; we should accol.dingly expect a large value of , and it large entry in the third column .
The same argument would apply to the Sun , but the physical constitution of the Sun is probably so diffel'ent from that of the planets that there could be no surprise at the Sun as an exceptional case .
Another exception is that of Mars .
Part of the discrepancy might , perhaps , be attributed to the I .
H. Any ) smalll of the planet , but the figure in the fourth column would to rest that instability must have ayed c large part in the creation of the satellites .
If , on the other hand , we begin by regarding the planets not as a fortuitous collection of bodies , but as a series of satellites all ejected from the same primary , the cafie is different .
For here we should expect the smaller planets to have cooled more the heavier ones , and therefore to be at a lower temperature .
Against must be set the fact that the heavier planets will probably have the greatest concentration of density about the centre , and the greatest mean pressure .
The first tion tends to increase the value which we should expect for the mean density of the smaller planets as compared with that of the Teater ones ; the second consideration tends in the opposite direction .
We can hardly profess to estimate the relative weights of these two considerations with any approach to accuracy ; perhaps it is best to revert to the argument given in the last paragraph , while bearing in mind that the approximate equality of our numbers may become considerably less significant as soon as the question of relative temperature is taken into account . .
We now consider the evidence afforded by the absolute value of our figures .
After allowing the exceptional cases , it appears that the value of for the earth most of the planets is about ] .
In other words , if we suppose these nets suddenly to resume the molten state , while retaining their present mass and radius , the spherical will be stable or unstable according qs the mean value of is greater or less than .
In the molten state we may take the value corresponds to a value equal to about half of that of steel , ' which .
If , however , we attempt to trace the history of a planet backward in time , we cannot suppose the mass and radius kept constant : the mass may be constant , but the radius will increase .
Under these conditions we find that the critical value of will be inversely proportional to the fourth power of the dius , will , thel.efore , be somewhat less than the value .
It would be extremely difficult to a reliable estimate of what this corrected value to be , and equally diflicult to estimate what would be the actual of molten similar to that of which our planets must have been composed when in the ltelt state .
Our rgurIlent is that the two values of least of the order of tude , ) equad , except ccuracies in our tions .
ofth 30 .
We conclude this part of our vork by coull ) aring two extreme hypotheses : the first referring the phenomena of planetary evolution solely to rotational , and the second solely to gravitational instability .
Given the approximate values of and for a planet , and the fact that this STABILITY OF A GRAyITATING PLANET .
planet has thrown off a satellite , the former hypothesis leads to a certain inference as to the angular momentum of the system ; the latter to an inference as to the radius of the primary .
The former hypothesis leads to no inference at all as to the size of planets which are to be expected\mdash ; they are as likely to be of the size of billiard balls as of the size of the planets of our system\mdash ; while the latter leads to no inference as to the angular momentum of the system , but presupposes it to be small .
The contention of the present paper is that the inferences which are drawn from the former hypothesis are not borne out by observations on the planets of our system , while those which are drawn from the latter are borne out as closely as could ) expected .
The true hypothesis must of necessity lie somewhere between the two extremes which we are compal.ing , and our evidence seems to show that it is much nearer to the latter ( gravitational ) than to the former otational ) STRESSES 1N THE EARTH . .
It has already been seen that in dealing with a sphere of the size of the earth it is necessary to take into account terms which are omitted by Lord KELVIN and others\mdash ; the terms which introduce into equations the tendency to instability .
It is of some importance to know whether the existing solution for the tions and displacements of the earth would be altered to an appreciable extent by the inclusion of these terms .
The genel.al frequency-equation which is p. is too complicated for manipulation , and is , moreover , open to the objection thal it does not represent the facts of the case ; for , inside the , the ) caused by permanent gravitation cannot legitimately be treated as small .
S 32 .
onsiderations of a general nature will , howeyer , give us some insight into the question .
In an imaginary earth , in which are infinitely great , the gravitational terms will be of no importance in comparison with those representing the elastic stresses .
The true solution will , therefore , become identical ] classical solution in which the terms are neglected .
For smaller values of the error will become appreciable , and if continue to decrease ecome infinite as sooll as the first is eached ; for at of bifurcation the application of an infinitesimally small exterllal force will finite displacement in the solid . .
intermediate values of the will small if are great compared with the critical values of at the point of bifurcation , and great if are near to these critical values .
* In addition to the inference as to the size of the planets , the hypothesis of gravitational instability leads to a further inference as to the distances of the fixed stars .
This has been discussed in my former paper , " " On the Stability of a Spherical Nebula\ldquo ; S48 ) , and hcre also the results seem to agrce with observation as closely as could reasonably be ) ected .
CHREE , ' Camb .
Phil. Soc. Proc vol. 14 , or LovE , ' asticity , , p. 220 .
VOL. CCI.\mdash ; A. 2 .
J. H. ON THE yIBRATIONS AND .
The most reliable evidence as to the actual values of is to be obtained from the nomena of earthquake } ) From the " " time curves\ldquo ; given in the Associ tion Report presented at the 1902 meeting , there seems to be little doubt that the so-called " " large-waves\ldquo ; are propagated merely through a thin crust on the earth 's surface , while the " " preliminary tremor\ldquo ; is propagated in a sensibly line through the itself The average velocity of propagation is found to be about kilolDs . .
second , and this is independent of the length of the path .
The inference is that ) is neal .
lie constant hout the earth 's interior , that value is about or .
If we suppose the vadue of to be gives for the mean value of Now , the critical value of which corresponds to the first point of been seen to be )bout 4 .
It would , therefore , ) the introduced in the solution for the cements and stresses is uot is comparable with the error tu which attention ] OF THE S 34 .
the evidence of the last section it will be seen that there is an overwhelmin probability that the values of the elastic constants of the earth.are such that a state of symmetry would be one of stable Whether or not the earth is at present in a state of spherical symm metry is a stion ; various indications , in particuku , the inequality iu the ibution ofland between the two of the rest that it is not so .
Now , even if material of the earth is at the present nloment of sufficient strength to maintain a spherical configuration in spite of the ravitational tendency to instability , it does not seem that it has always been so .
Looking backwards in time we must come to a stage in which the material of the earth was plastic , and , further back still , fluid .
At this time the value of would be much smaller than its present value , and , as already pointed out in S29 , would probably be about equal to the critical value for the planet at that period of its existence .
There would , therefore , seem to be a sufficient reason for considering the possibility that the earth , at the which consolidation set in , was not in a state of spherical symmetry .
Let us mine some of the consequences of this conjecture .
Professor MILNE has kindly assisted me in this question . .
ante .
STABILITY OF A PLANET .
It is easy to see that enormous stresses would be set up in the interior of the earth after consolidation .
An equilibrium configuration depends in general upon the compressibility of the material , and a guration which was one of equilibrium for the compressibility which obtained at the moment of solidification would not remain so after the incompressibility and rigidity of the material had increased by cooling .
If we suppose the earth to cool in unsymmetrical configuration the stresses set up will soon become very great .
In fact , Professor DARWIN has shown that the stresses which would be produced by the weights of our continents in an earth initially homogeneous irregularity less than thousandth part radius ) would be so great that the material would be near the breaking point .
* We must therefore suppose that as the earth cools and the elastic constants change there will be a series of ruptures resulting from the stresses set up in the interior .
The configuration will become approximately spherical ( spheroidal if rotation is taken into account ) as soon as the poin of bifurcation is passed .
The fact that the ultimate collfiguration is reached only as the result of a long succession of ruptures puts the whole question outside the range of exact mathematical treatment .
We however , see that the final tion ( disregarding rotation ) will probably not be quite spherical , but will retain traces of the initial unsymmetrical configuration . .
Before we can attempt to decide whether or not the earth shows traces of a process such as that just described , it will be necessary to form some idea of the with ] which the process must have .
We cannot accurately calculate the " " linear series\ldquo ; of unsymmetrical configurations except in the immediate .
of the point of bifurcation .
Near to this point the configuration is spherical except for terms proportional to the first harmonic .
The free surface will , therefore , be strictly , and it ] , of course , be an equibut its centre will llot coincide with the centres of other surfaces of equal potential .
If we suppose a fluid mass of this kind to solidify , and then to shrink by cooling , the shrinking being accompanied by a series of ruptures of the kind already explained , we can easily imagine that the free surface would retain an approximately spherical form , but that when the final state is reached this surface would not be quite an equipotential , and the centre of gravity would not quite coincide with the centre of figure .
If water is placed on the surface of a of this kind , it will form a circular sea , of which the centre will be on the axis of harmonics , while the dry will form a spherical cap. Evidence the Distribution of Seas and Land .
S 36 .
Now this is not observed on the earth , and it could not be expect .
, since we have ignored all the agencies which have contributed to the figure of the earth , 'Phil .
Trans vol. 173 , 1882 , p. 187 .
except the one } which this paper is specially concerned .
The question is not whether we observe the statejust described , but whether we can detect ally approach to this state , and this , I believe , can be done .
Professor DARWIN " " It is well known that the earth may be divided into two hemispheres , one of which consists almost entirely of land and the other of sea .
If the south of England be taken as the pole of a , it will be found that almost the whole of the land , excepting Australia , lies in that hemisphere , whilst the antipodal hemisphere consists almost entirely of sea .
This proves that the centre of gravity of the earth 's mass more emote from ngland t the centre of figure of the solid globe .
deformation of this kind is expressed by a surface harmonic of the first order S 37 .
We can carry our calculations a step further .
The divergence from the initial uration is only represented by a first harmonic so long as squares of this dive i may cted .
If these are taken into account , we must Fig. 1 .
include a term ) to the second harnlonic a , s well as that proportional to tfi .
Thi , s process of successive tion might be continued to any extent , so a complete series of unsymmetrical urations m be calculated in the explained in my former paper .
We may , however , be content to stop at the second harmonic .
The free surface will now be of the form G. H. , ' Phil. Trans vol. 173 , , p. 230 .
'Phil .
Trans , vol. 199 , p. 41 .
STABILITY OF A RAVITATING PLANET .
and we therefore examine whether any traces of the second harmonic term can be found in the earth 's surface .
Now , if we take positive in equation ( 63 ) , equation is that of the pear-shaped curye which was on p. 46 of this earlier paper .
This differs fionl the spherical mainly in possessing a protuberance\mdash ; the stalk end of the of which the centre is on the axis of harmonics .
aces of this protuberance.may , I think , ps ) found in the continent , the arrangement being that shown in .
It is true that the centre of does not coincide with the antipodes of , but the discrepancy becomes less when we take into account the region of ocean shallows which lies to the east of Australia .
discrepancy can be ful .
there reduced taking the rotation of the earth into account .
When the rotation of the earth was greater than at present the ellipticity of would be greater , and the transition from this to the present ellipticity -ould take place through a series of uptures snilar to those already described .
The rotation ( assumed small ) of the pear can be allowed for by adding a term to the -hand side of equation ( 63 ) , this .
a second harmonic deformation hayin the axis of rotation for axis of harmonics .
The present rotation of the earth can similarly be represented by a term where .
The equation to the present surface of the may be taken to be and hence the above the present sea-level of the surface of the primaeval pear , if restored , would be It will be found that the of the rotational term is to move the theoretically predicted Australia nearer to the equator of the earth , and to change its shape from a spherical cap to a conic .
Again , we should expect the highest land to be on the of harmonics , and , therefore , in or near England .
Here , , the agreement of facts with theory might be closer if we could suppose the continent , which shows to existed at one time in mid-Atlantic , to be restored to its former position .
But the agreement of facts with theory can only be expected to ) of the roughest kind , and we must always bear in mind that our theory does not lead us to expect that the present figure of the earth will be pear-shaped , but only that it will enlble a pear disfigured by a long series of ruptures .
* Added JanuaIy 3 , 1903 .
I am indebted to the referee for suggesting this addition .
MR. J. H. JEANS ON THF VIBliATIONS AND from the Distribution of Centres .
S 38 .
It can be seen that the earthquake regions of the world have a reference , as their distribution on the earth 's surface , to this pear-shaped figure , and this , ust be considered as evidence .
Let ns first examine the facts .
divides the earthquake-areas of the globe into distinct regions , and a ) map of these is given in the British Association Report ' These regions are given in the following table .
The first figure denotes the ltlm)er of large earthquakes which have occurred in these regions in the three .
The earthquakes from the three regions printed in italics small in comparison with the others .
In the last column is giyen the approximate latitude of the centre of each region , referred to Greenwich as pole ( the latitude of Greenwich being taken to be ) .
of Earthquake Regions .
A 14 A 16 12 Andean 29 panesc 5 4 ] 16 12 Andean 29 pallGSC 5 4 ] 16 12 Andean 29 pallGSC 5 4 ] 16 12 Andean 29 pallGSC 5 4 ] 16 12 Andean 29 pallGSC 5 4 ] 16 12 Andean 29 pallGSC 5 4 ] 16 12 Andean 29 pallGSC 5 4 ] 16 12 Andean 29 pallGSC 5 4 ] 16 12 Andean 29 pallGSC 5 4 ] .
, 14 Asiatic ) retic Now , it will be at once noticed that for most of these regions the latitude is small .
If weight the regions according to the corresponding number of earthquakes , iving half-weight to the small earthquakes in regions , we find as the mean of the numerical values of the latitude about , whereas if the regions were distributed at random we should expect the mean latitude to be radians , or about We therefore see that the earthquake regions tend to lie near the equator of our pear .
The evidence can be put in a more striking way as follows:\mdash ; Exactly half of the surface of the globe is of a latitude less than The half for which the latitude is less than , measured from Greenwich as pole , responsible for 156 earthquakes ; the remaining half was responsible only for 42 , of which 28 were the small earthquakes from regions , J. There is , therefore , no doubt that the earthquakes tend to emanate from points near to the equator of the supposed pear .
Now , if we look back to fig. 1 , we see that this is equivalent to saying that earthquakes occur where the " " slope\ldquo ; in the figure of the earth is steepest .
This conclusion is the same as that to which the British Association Committee were led from a 'Brit .
Assoc. , Report , ' Belfast , 1902 , " " Seismological Investigations p. 4 .
STABILITY OF A GRAyITATING consideration of the actual figure of the earth , and it is that which might naturally be expected .
The theory put forward in this paper may , ) , suggest a reason why these regions should lie )pproximately on a circle of the earth , and why this great circle should approximately divide the earth into two ispheres of sea and land .
S39 .
conclusion it may be well to sunmarise those of the paper which refer to the figure of the We saw that at the molnent of solidification the earth might be either spherical ( except in so far as it was deformed by its rotation ) or pear-shaped .
Our theoretical calculations and our knowledge of the constants of the earth at the time of solidi- fication were not sufficiently lrate to enable us to decide of the two alternatives is the Jllore probable .
The of the earth , there spherical or pearshaped , could not be maintained long against the enormous strains which would be set up in the earth as the process of cooling ) oceeded , and this shal ) would radually give place to an } ) tely sshal ) , the in bein produced by a long succession of uptures .
The suggestion of this is that the earth , in spite of this series of ruptul.es , still shows tl.aces of pear-shaped configuration .
Such a configuration would possess a single axis of , and this , it is suggested , is an axis which meets the earth 's sul.face somewhere in the neighbourhood of England ( or , possibly , some hundreds of miles to the S.W. of .
Engand ) .
Starting from England we have in the first place a hemisphere which is practically all land ; this would be the blunt eIld of our .
Boundin this hemisphere we a great circle of which England is ) pole , and it is ovel this circle that earthquakes and volcanoes of most frequent occurl.ence .
If we supl ) our pear contracting to a spherical shape we notice that it would probably be in the neighbourhood of its equator that the change in curvature and the relative displacements would be greatest , and hence we should expect to find earthquakes and volcanoes in greatest numbers near to this circle .
Passing still further from England we come to a great region of deep seas\mdash ; the Pacific Ocean , the South Atlantic Ocean , and the Indian Ocean : these may mark the place where the " " waist\ldquo ; of the pear occurred .
Lastly we come , almost at the antipodes of England , to the Australian continent and the shallow seas which lie to the east of it ; these } be the remain of the stalk-end of the pear .
S 40 .
It may , I all afraid , be thought that the hypotheses upon which the paper is based are too speculative and the results , consequently , too uncertain .
In defence it may be said that the object of the paper is not so much to establish new doctrines as to point out possibilities , and that these possibilities seem to be of such a kind that it may be useful to keep them in mind in discussing questions connected with the AND STABILITy OF A PLANET .
and structure of the earth , as well as the more general questions of planetary evolution .
In condusion I to indebtedness to Professor G. H. DARWIN and Professor A. H. ssistance which I have received from them .
[ NOTL.\mdash ; Added Otl , , 1903 .
While the above paper was in the press , Professor W. J. SOLLAS a before the Geological Society in which the Figure of the was discussed from a geological standpoint .
Professor SOLLAS had arrived , an examination of the features of the earth , at a conclusion similar to that to which I had been led from theoretical considerations : he had detected an axis of symmetry , other than the axis of rotation , the earth 's , and ) ressed the opinion that " " the pear-shaped form , now that it was pointed out , became obvious to mere inspection : it was a geographical fact , and not a speculation The axis of Professor ' pear does not , however , coincide with that which I tentatively ) forward in the above paper , and the object of this note is to accept the alteration suggested by his paper .
The conclusion reached in his paper is that the axis of symmetry of the pear-shaped figure passes through a point of latitude and longitude about N. by E. Thus \mdash ; the continent whose mean ight sea-level is be taken to be the centre of the " " Land Hemisphere\ldquo ; in of my paper , while the protuberance which formed the stalk of the peal is nerged i the Pacific Ocean , which now forms the " " Water ' Almost the only remaining evidence of the existence of this protuberance is the fact that the axis of the pear coincides with the earth 's greatest diameter .
The great of earthquake-centres suggested in S 38 of my paper is to be replaced by the line of Pacific ; this approximately forms a small circle ( of radius about ) which almost coincides with the ) osed great-circle in the therll h. Further details of Professor view will be found in his The Figure of the Earth ' Quart . .
lix .
, Part 2 ) .
The that is surrounded by a belt of seas , and this again by belt of before tlJe 1 is ched , laps to a bodily subsidence of the blunt end of the peal , the of .
possibly been the line of Pacific folding .
fracture would , of course , displace the of glavity of the pear , and prob bly this would account not only the feature just mentioned , but also for the non.appearance of the protubel.ance .
It will be noticed that the of tho latitude of the extremities of the axis agl.ees well with the theory of planetary evolutio put in SS of the ) ]

