]\gt ; I. On , the of the Kinetic Theory to Dense S. H. BURBURY , F.R. Received January 12 , \mdash ; Read February 7 , 1895 .
THE motion of a great number of elastic spheres , when their aggregate volume does not bear an evanescent ratio to the containing space , has received little attention from writers on the kinetic theory .
In what respect , beyond the shortening of the mean free path , will it differ from that of the rare medium usually discussed ?
I think that the answer to this question is that there exists in all systems , dense or a tendency for the spheres to move together in masses or streams , and ao to diminish the mean pressure per unit of area , and the number of collisions per unit of volume and time .
And this tendency has an appreciable influence on the form of the motion as soon as the ratio of the aggregate volume of the spheres to the containing space becomes appreciable .
If a part of the system , say spheres , be at any instant contained a volume they have energy , , of the motion of their common centre of gravity .
And they have energy , , of relative motion .
As the spheres increase in diameter , the ratio , will be found to diminish on average .
But the number of collisions per unit of volume and time , given , or , depends on , and therefore dimir.ishes by the diminution of 1 .
Let be the mass , the diameter of a sphere , the number of spheres per unit of volume , the pressure per unit of surface .
Also let us now denote by tk energy of the motion of the spheres ] ative to their common centre of gravity , so that is now the mean value per sphere of this energy .
We have then , as is well known , . . . . . . . . .
( 1 ) , in which is the repulsive force , the distance between a pair of spheres , and the summation includes all pairs in unit of time .
We must first evaluate on the assumption that no forces act except during collisions .
2 .
Let be the relative velocity of two spheres .
Let be the angle made with by the line of centres at collision , if a collision takes place .
The angle may have any value from zero .
As the effect , of collision the velocity , a , in the line of centres is reversed for each sphere .
We may assume this reversal to be effected 27.2.96 . .
S. H. BURBURY ON THlJ APPLICATION by the constant finite force acting on each sphere in the line of centres during the small time Let us define this short but finite time as the time during which the two spheres in collision , or the duration of collision .
Then , during collision , if be small enough , the virial of the supposed force is sensibly constant , and we calculate its value as follows .
Let , be the direction ines of the line of centres referred to any axes .
The coordinates of the point of contact shall be .
Then those of the centres of the two spheres are , being the diameter of either sphere , for one sphere , and for the other .
The component forces acting at the centre of the first sphere are Those acting at the centre of the second sphere are the same with reversed signs .
For two spheres colliding with relative velocity we find that the virial is at each instant during collision .
We have to multiply this by the chance that , given two spheres A and with relative velocity , they shall be in collision at any given instant .
It is assumed that we are dealing with a space throughout which is constant , and therefore the fact that the relative velocity is , affords no presumption with regard to the relative position of the two spheres .
About the centre of sphere suppose a spherical surface described with radius .
An element of that surface is .
Upon that element of surface form the element of volume .
And form a similar element of volume on the other side of the sphere , that is , using for .
Then the two spheres are at this instant in collision if the centre of the sphere is within either of those elements .
Let V be the volume in which spheres are moving with .
constant .
Then by our assumption is as likely to be in any part of V as in any other .
Therefore the ohance that A and , having relative velocity , are in c.ollision , is .
The average virial for two spheres with relative velocity is then at each instant OF THE KINETIC THEORY TO DENSE GASES .
.3 if , be the component velocities of the two spheres .
Now since is the energy of relative motion of spheres in volume V , M2 the summation including ery pair , and therefore . . . . .
( 2 ) , p. , since .
Then for spheres in V , for spheres in unit of volume .
Substituting this.value of in ( 1 ) we obtain . . . . .
( 3 ) .
It is assumed in these results that we are dealing with a space throughout which is sensibly constant .
3 .
We see then that is proportionai to The analogy between this expression and BOLTZMANN'S , iu which denotes potential energy , suggests that the law of distribution of velocities among our spheres should be , instead of as in the rare medium , , or rather , since there may be stream motion as well as relative motion , Let us further develop this analogy .
In the Clausian equation where .
is the kinetic energy of the motion of spheres in volume V relative to their common centre of inertia .
Hence If the spheres , being initially contained in volume , be compressed into volume V , and be maintained constant during the process , the work donc in compression is MR. : H. BURBURY ON THE APPhIOATION The .
first term .
the right-hand side expresses the work which would be done the process if the spheres were material points ; and no between them .
The second term expresses the amount by which this work is increased by the spheres having finite diameter and collisions .
And if be infinite , this additional work is or We see then that the term in the index of represents work done against collisions in sing the system from an infinite volume to its actual volume with constant .
It is analogous to the potential X in the usual expression .
It might not be inappropriate to call .
the potential of collisions .
In order to confirm or otherwise the above suggestion , I to consider\mdash ; 4 .
The distribution ofenergy in column of gas when in equilibrium in a lield of uniform force , the molecules being equal elas:tic spheres of diameter Let the colulnn be an infinite cylin , the parallel to its axir Take a plane perpendicular to the axis as base , and ] be the height of a point above that plane .
Then we have , with the same notation as before , . .
( 4 ) and , as before , : .
Here is the average per sphere of the energy of }is concerned in .
But , in our vertical column , assume nent , hat there is no stream motion that need be taken into account , and , therefore , , we may .
write instead of , and becomes ( 3 ) .
Now contains as a factor .
: But ] now assume that , not , but , is independent of , so that we may write ( with constant .
On that and ( 4 ) give : .
where is the value of when 5 .
Now consider spheres crossing the plane with for vertical component these , say .
planp , } ' .
iop the is struck out , there be another collision by which OF THE KINBTIC THEORY TO DENSE GASES .
uted for it , the substituted sphere having the same vertical .
velocity , but not quite the position on average , as the original one ad at the instant of its collision .
6 .
If the collision be direct , i.e. , the line of centres coincide with the relative velocity , the substituted sphere is advanced in position through a distance equal to in the direction of the line of centres , and that without any loss of kinetic energy by the action of the force .
If the collision be not direct , there is an advance , and we proceed to calculate its value .
Let A be the centre of a sphere which comes out of collision with for vertical component of velocity , the centre of the sphere in collision with it , the point of contact .
Let denote the vector line of centres and ( ul ) the cosine of the angle between and .
Then is the projection of HA on the vertical , and is its average value for all the collisions in question .
7 .
Again , let be the centre of a sphere which enters collision with vertical velocity that of the sphere colliding with it , the point of contact , and let denote the .
Then , ) There will be as many collisions per unit of lume and time of the one class as of the other , and the height of the point of contact , above the se , is on average the same for one class as for the other .
Therefore , taking the collisions in pairs , one from each class , each pair substitutes A for as the sphere with vertical velocity ; and , on average , the substituted sphere A is at the instant of its collision above the original sphere at the instant of its collision by the distance ( ul ) .
We have next to show that on average of all collisions of spheres taking place between and , where 8 .
Let be the actual velocity of the sphere A as it issues from collision , so that is the vertical component of , and .
Let be the velocity of the other sphere as it issues from collision with , and the angle between and Also let be their relative velocity , so that Whatever be the values of , and and therefore .
Again whatever } ) the values of therefore We have to tiPly this expression by the number of collisions which spheres , each having absolute velocity , undergo in time with other spheres having absolute velocity ; the angle between and being E. . .
, and then integrate according to and E. That number is if be the number per unit ofvolume of spheres whose veloci is so that Therefore the complete average value of ( ul ) , that is the average vertical displacement of the substituted spheres , is , or , since But also so our result is We know ( WATSON'S 'Kinetic Theory of Gases , ' 2nd edition , p. 56 ) that the quantity of vertical momentum trannferred across the plane per unit of area OF rIHE KINETIC THEORY TO GASES .
and time is proportional to , and , therefore , increases in the ratio 1 : , when the spheres , from being material points , acquire diameter .
We now see that this is true for each separate class of the vertical- momenta .
9 .
Corresponding to spheres with vertical velocicy at the plane at the beginning of , we have at the end of spheres , the same or ostituted , whose average height is or .
But their loss of energy due to the ascent is on average Mfds for each sphere .
It follows that the average loss due to an ascent is , allowing for substitutions , 10 .
Let then the number per unit of volume of spheres whose energy of vertical velocity is . .
be at the base , where in which ( A ) .
Then the number per unit of volume which at height have for energy of their vertical velocity is ( remembering ( 5 ) ) that is ( B ) .
The two classes A and are equally numerous per unit of volume , and since , allowing for collisions and substitutions , the loss or gain of energy due to the force in passing from the base to or vice versed is either class can by ascending or descending ( the proper number of substitutions taking place ) replace the other .
And the assumed law of distribution of velocities is not disturbed by the force , as spheres pass up and down the column .
Now , make in the above reasoning , and we find it is exactly the reasoning from which , in the ordinary case of a rare medium , we conclude that is constant throughout the column .
The same reasoning leads in the general case to the conclusion that is constant throughout .
11 .
The above results are obtained on the hypothesis that no account need be taken of stream motion among our spheres .
If , however , there be such stream motion , we have to suppose that the spheres crossing the plane were members of a large group having a vertical velocity of their common centre of gravity , and that the spheres have vertical velocity relative to this common centre of gravity .
Then in time the spheres will have risen on average the distance by virtue of their relative velocity , where , and a distance by virtue of the common velocity U. Their loss of etic energy by the action of the force in this ascent is or if And now writin for T. for the energy of the stream motion whose vertical component is , we find that is constant throughout the column , and we may now write 12 .
It follows that we cannot express the -law of distribution of velocities the spheres in the form , or , as in the rare medium , with law mus , be , in which is some quadratic function of the velocities .
Suppose that for spheres it is \amp ; c. , with corresponding expressions for the components and If we find the mean value of by integrating for all vaIues of , \amp ; c. , between the limits , we find , there being variables or 3 component velocities for each sphere , But therefore , This result might be considered to justify the assumption that the law of distnbution of velocities is in all cases , including the field of no forces formed by making , accurately expressed as follows .
The chance that the velocities of spheres , forming a group together , shall be . .
\amp ; c. , is proportional to in which M2 M2 But we must remember that the whole treatment is based on the consideration of a great number of spheres , so that we cannot safely assume the law to hold when is small .
Let us , then , consider the subject from yet another point of view .
have shown elsewhere ( ' Science Progress , ' November , 1894 ) , that in a dense medium the velocities of contiguous spheres cannot be independent of one another , because there is a presumption that recent collisions of two spheres near to one another have been with the same third sphere , and they have , so to speak , inherited some common ocity from it .
In other words contiguous spheres have been exposed to the same environment , and , in the dense medium , environment does not change rapidly ; therefore their velocities are not indepen\amp ; nt .
If , therefore , the spneres contained in a volume V be in number , and their positions known .
the chance that their locities shall be OF THE KINETIC THEORY TO DENSE GASES . . .
\amp ; .
, . .
\amp ; .
, must be of the form in which the index is a quadratic function of \amp ; c. , \amp ; c. , and \amp ; c. , but contains no products of the form , or , and the coefficients , \amp ; c. , are functions of the positions of the spheres .
14 .
With regard to the forms of these coefficients , we observe that the quadratic fnnction in the index must always be positive , because the chance cannot become infinite for infinite values of bhe variables , .
The condition for this is that the determinant and all its coaxial minors , must be positive , and , therefore , every positive .
15 .
Again , not zero , expresses the fact that and are not independent .
But it is also a fact that they are more likely to be of ihe same than of opposite signs .
Therefore , , and similarly every , must be negative or zero .
16 .
Evidently also the coefficients must generally diminish in absolute magnitude as the distance between the spheres to which they relate increases , and must become inappreciable at some distance , small compared with the dimensions of our system , but possibly large compared with the diameter of a sphere .
The 's must be functions of the positions of the spheres within V having this property .
Again we may consider either a volume V containing spheres , or a smaller volume containing out of the spheres , that is , all the spheres except one , which one belongs to the outer layer .
If the velocities of that one be denoted by , and if be the quadratic function for the spheres , for the spheres , we must have If we actually perform the integrations indicated for one variable , we get with \amp ; c. \amp ; c. , in which \mdash ; \amp ; c. As the number of spheres included in our group diminishes , the coefficients diminish , and , since every is negative , the 's increase in absolute magnitude. .
On the other hand , as increases the 's increase , and every , if changed at all , diminishes .
Therefore , as increases the function tends to a limiting form .
But that limiting form must be if , because in that case we know the law to be .
We must then have u]timately , where is a quadratic function of the velocities .
And we may now assume that for sufficiently number of spberes comprised in a group , throughout which is sensibly constant , the law is expressed by the function , in which being the determinant of the coefficients of the quadratic function We have thus obtained certain conditions which the coefficients \amp ; c. , must satisfy .
Another condition is that th ' , assumed law of distribution of velocities expressed by the function not be disturbed by a collision taking place between any two of the spheres , which colJision changes the velocities , but not the positions , of the two spheres in question .
To find the values of \amp ; c. , to satisfy this condition presents considerable difficulty .
It is not , however , at present necessary to solve the problem in that form , as will be seen later .
For we have only to consider the positions of the spheres as and we obtain a solution .
sufficient for our purpose .
If , namely , it be given that there are at any instant heres within a spherical space , bnt nothing is known of their positions within , we have only to assume that the chance of their having at that instant velocities , \amp ; c. , is with containing only one coefficient and one , and we shall find that all necessary conditions are satisfied , including the condition that the assumed distribution shall be unaffected by collisions .
For let be the component of two spheres before collision .
A collision between the two converts these components into in the following manner .
Let be the direction cosines of the of centres at collision .
The velocities of the two spheres resolved in the line of centres are , before collision , and respectively .
And we have OF THE KINETIC THEORY TO DENSE GASES .
, \amp ; c. That is , Call these equations A. But inasmuch as the motion might take place the reverse way with the same values of , it must by the same reasoning be true that \mdash ; , \amp ; c. . . . .
which are the same as equations A with the accents interchanged between the right and left-hand members .
If we solve either set of equations we get the other set .
17 .
Now it is given that at this instant the chance of the spheres having velocities . .
\amp ; c. , is .
in which terms containing squares and products of the velocities The chance that after this collision the velocities of the spheres shall have the values , and \amp ; c. , is found by substituting for , in the index , their values in terms of , \ldquo ; 1 , , as given by equations .
Effecting this substitution we find that the coefficient of is that is .
Similarly , we find that the coefficient of in the new index is , and the coefficient of every product in the new index is the same as that of in the original index .
The new index is then the same function of that the original one was of .
The assumed distribution of velocities is therefore not affected by any one , and therefore not by any number of ] lisions .
This , it will be remembered , is on the assumption that all the coefficients are the same , and all the coefficients are the same , so far as t.hey have to do with the colliding spheres .
18 .
Now , if and and if we use this for we find that all necessary conditions are satisfied , including the condition for permanence notwithstanding coIIisions .
19 .
If , however , the coefficients were not the same for both colliding spheres , but the form were , then we ehould find that the coefficient of in the new index is ; the coefficient of is ; the coefficient of is as before , ; but that of is ; and that of is The assumed law of distribution cannot in this case be unaffected by collisions , unless ( 1 ) all the coefficients are the same ; ( 2 ) if the velocities of the two colliding spheres be and , and those of any third sphere be , then that is , the coefficients must be such functions of the position , that if spheres 1 and 2 are close together \amp ; c. To return to the case of the positions of the spheres being unknown .
20 .
If we form the determinant of the system we find Hence we find .
therefore OF THE KINETIC THEORY TO DENSE GASES .
1.3 Also , \amp ; .
[ These results are easily obtained by considering the general determinant of constituents in which all the axial constituents are , and all the non-axial ituents are It will be found for , and thence by induction , that Replacing by , and by , we get the result above stated .
] Also and have the same value as .
And the whole kinetic energy of the spheres , or , is , on average , , or Again , -\amp ; c. , \mdash ; \amp ; .
, \mdash ; \amp ; c. , with similar expressions for the 's and .
But by the of determinants .
Therefore and and therefore and therefore , which increases as increases , that is , coeteris , as the diameter increases .
Again , the pressure per unit of area on a plane moving with the stream , and therefore the mean pressure , is , which is independent of { , he diameter Now , the spheres being material points with , the mean pressure is ; that is , as we may write it , , because in this case .
As the spheres increase in diameter with unaltered , the mean pressure per unit of area remains unaltered .
In other words , it is exactly as much diminished by the conversion of part of the energy of relative motion of contiguous spheres into energy of stream motion as it is increased by the introduction of the term as the eres acquire diameter aring the actual value of with what it would be if , with the same total kinetic energy , the spheres were material points , we see that it is diminished in the proportion :1 .
The number of collisions per unit of volume and time is proportional to ; that is , to .
It is less in the proportion than it would be if , with the same total kinetic energy , the spheres had velocities independent of one another .
21 .
From the fact that is independent of , it follows that local variations of density , i.e. , of , do not involve the expenditure of any work on the whole , and therefore such variations may and will come into being .
22 .
BOLTZMANN'S minimum function continues to diminish by collisions , finally attaining its minimum constant value when the distribution of velocities defined by - our assumption in 13 is established , but its actual value when minimum differs by from what it would be if the spheres were material points .

