]\gt ; DESCRIPTION OF PLATES .
Water at rest .
Fig. 3.\mdash ; Obstacle with Pressure and Stream Indicators in Position in Tank , with the Fig. 4.\mdash ; Swing from Left to Right\mdash ; First Part .
Fig. 5.\mdash ; Swing from Left to Right\mdash ; Second Part .
Fig. 6.\mdash ; Swing from Right to Left .
Some Problems the Forms of Nebnloe .
By GEOBGE W. WALKER , A.R.C.Sc .
, , F.B.S. , formerly Fellow of Trinity College , Cambridge .
( Received March 27 , 1915 .
) The possible forms of distribution of a mass of gaseous material under the influence of its own gravitation are of considerable interest in the nebular theory .
The law of density which it appears most reasonable to assume is Boyle 's Law , in which the pressure is proportional to the density , unless the pressure becomes so great that the material begins to resemble an incompressible substance .
Althougb it is unlikely that the temperature is uniform throughout , still the solution under this restriction would be of value as a step in the direction of greater knowledge as ards possibilities in astronomical phenomena .
The equations be formed and lead to a differential equation for the surfaces of equal density .
This equation is not linear , and in the threedimensional case little progress to a general solution has been made .
In the two-dimensional case , however , considerable progress be made .
A number of years ago I obtained the exact solution of the statical case of symmetry about the origin .
Shortly after I found that Pockels had obtained the complete solution of the statical two-dimensional equation , and particular cases and have arrived at results indicating such a close analogy with actual astronomical forms that I venture to hope they may be of interest as gesting types that may be expected in three dimensions .
We begin by consideration of the fundamental equations .
Let be the gravitational potential , the gravitational constant , the pressure at any point of the gas , the density at any point of the gas , , the Cartesian co-ordinates of any point .
The hydrostatic equations of equilibrium are and the potential must further the equation If , where is a constant , we get as an constant , or Hence This is the equation of which Pockels has obtained the solution in terms of two arbitrary functions as follows .
As is the solution of , then the solution for and is We require to put this in a which gives real positive values to , and I find that this can be secured by taking Otherwise , if and are conjugate functions , so that , we get and hence of interest to pure mathematicians to investigate this point , just as it would be of value to applied mathematicians to know the most eneral form .
I\mdash ; If we take we This is the most general case of circular sy1nmetry .
We note that when the density becomes infinite at the elsewhere finite , and vanishes at co .
When the density is finite at the origin , and elsewhere is finite , at co .
When the density is zero at the origin , rises to a finite maximum .
and then diminishes and vanishes at infinity .
Writing the equation in the form the curves in fig. 1 show as a function of , for the four special values , and 2 ; is still at our disposal while is determined when and are fixed .
ring nebula .
Case II\mdash ; Take we get The surfaces of equal density are given by The curve , fig. 2 , shows the particular surface drawn to a scale of 10 .
This case is somewhat peculiar .
Starting with when , the curve for positive values of proceeds as shown , continually approaching the origin by a succession of diminishing spirals .
The locus is , however , symmetrical about .
If now we proceed to draw another surface of different density it would be found to cross the original curve at a succession of points for which .
This would mean that at such points two densities are possible , and we cannot admit this .
We must , therefore , stop the curve at the nodal point , and then the complete series of curves of equal density form a series of non-intersecting curves of this rem.arkable researches of Darwin , Poincare , and Jeans .
Case llI.\mdash ; Let us take the elliptic co-ordinate transformation , and let Then we get sin2 .
Thus the surfaces of equal density may be written in the form For , we get Fig. 3 shows the curves obtained for , 1 , 2 .
The density falls off to zero towards infinity , while for values the locus breaks up into two oval curves about and , which are the foci of the original ellipses in the transformation .
and are singularities at which would be infinite .
The general character of the loci is not altered by giving some finite value .
For comparison fig. 4 shows on the same scale the result of putting .
The equation is and the curve drawn is for is seen to be rather flatter than the corresponding curve for This case then , as a whole , has some analogy in the case of a nebula with two fundamental equal nuclei .
: these three cases are sufficient to indicate the generality of the method , and [ the remarkable interest of the forms that can be obtained .
I pass to the consideration of an important related problem .
The material of an actual nebula may be moving and not at rest , and in particular it may be rotating .
As the rotation , if it exists at all , is very slow , the divergence from the statical equilibrium in such a case would be extremely small , so that the statical solutions are themselves of value .
It is , however , of interest to know what the effect of motion may be .
The simplest case we can consider is the final state in which the material rotates about an axis like a rigid body , with steady angular velocity There the problem reduces to a statical one , when we take axes rotating with the material and add to the gravitational forces the centrifugal effect .
In case of a gas such a rotation , however small , can hardly be expected to extend indefinitely , and we should rather expect the motion to fall off and finally cease at a great distance .
While keeping this in view , it is not without interest to examine what effect such an imposed uniform rotation would have .
Mr. Jeans*has referred to this problem , but without giving details .
Some of my results are in agreement with his conclusions and others apparently diverge .
If the components of velocity are and , and we neglect viscosity , the equations for a steady motion referred to fixed axes 'Phil .
Trans vol. 213 , p. 462 ( 1914 ) .
Mr. G. .
Walker .
In the case of , where is a function of where is a function of only .
' The equations then reduce to If , we get the case of body rotation , and If , we get an " " irrotational\ldquo ; motion , and In the latter case the velocity falls off as increases but is infinite at the origin .
We might make any function of ; and among these we may select one , e.g. , which makes the velocity zero at the origin and also zero at great distances .
But such cases are not steady and the influence of viscosity would arise .
We cannot , in general , integrate in finite terms the equation ' but if suppose so small that differs from its value in the statical case by a small quantity we have approximately while Let , and , then the equation may be transformed to Now the equation has a solution A Hence the equation may be written* and the particular integral which we require is *Cf .
Boole , ' Differential Equations , ' p. 205 .
roblems Illustrating tVeor m. buloe .
417 In the case this gives for \ldquo ; where These curves are shown to scale in figs. 5 and 6 , and we note that in both cases z when , is everywhere positive , and finally becomes infinite and proportional to when is reat .
The general effect of the rotation is then to increase the density from what would obtain in the statical case , and by increasing proportion as increases from the origin , but even if it were permissible to pass to large values of the density would still diminish to zero .
We cannot , however , extend our #$ value and to move the point of maximum density further out with increasing ; but there is no indication of a tendency to throw off the outer layers .
This latter result disagrees with the general conclusion stated by Jeans .
cit. cmte ) .
For " " irrotational\ldquo ; motion and the particular integral we require is where In this case is egative and infinite when , rises to a maximum positiye value , and again declines to a negative and infi1lite value as becomes indefinitely great .
As before , we cannot extend our solution to reat values of , but we may say that the tendency of the motion is to make the distribution hollow near origin , to increase the concentration towards the maxirnum density at a finite value of and finally to tend to throw off the outer layers when .
becomes great .
This latter conclusion is in agreement with Jeans .
When we pass to other cases such as Cases II and III , where the original distribution of density is not symmetrical about the origin , it is rather difficult to think of any steady motion other than uniform rotation which would leave the distribution in anything like a permanent form .
We must , therefore , confine our attention at present to this case , artificial though it may be .
.or , if In Case II of the distribution of pear-shaped character I find that the .solution will have to be obtained by a somewhat complicated series , and as I have not yet gone very far with it I pass to Case III , which can be integrated in finite terms .
We had as the statical solution Hence the equation we have to solve is - or , on changing to elliptic co-ordinates , we get .
The solution is , where and are functions of only , and satisfy and respectively .
A solution for , when , is .
Hence the particular integral is A solution for , when , is .
Hence the particular is - taJlh 2 This function is - at , increases and becomes positive as we increase , and becomes infinite and oive with as becomes infinite .
Again , along the axis of , so that sech \mdash ; log cosh At the origin this function is - , it increases as increases and becomes positive , finally being infinite and positive with as becomes infinite .
The effect of the rotation is then .
to diminish the density near the origin and to increase it at greater distances , but there is no clear tendency to throw off the outer layers , since the function becomes positive and infinite and not negative and infinite .
This is , again , in apparent conflict with Jeans ' conclusion .
It would thus appear from the cases considered here that a finite boundary of a gaseous mass is not to be looked for as a consequence of rigid body rotation , but rather as a consequence of some other type of motion in which viscosity may play a part .
My sincere thanks due to Sir Joseph Larmor for kindly criticism and advice .
[ Note added April , 1915.\mdash ; It has been pointed out to me that in Case II discontinuous the line .
Hence this is a line of singularities .
In accordance with the principle already stated we must , therefore , suppose this lirle to be occupied by so.lid matter of the proper density at any point , determined by the change of in crossiIJg this line .
Theoretically this distribution would extend indefinitely , but practically the density of solid matter required at any point on becomes very small as the distance from the origin increases .
]

