]\gt ; [ 407 ] X. On the Collision of Etastic Bodies .
S. H. BURBURY , Received October 24 , \mdash ; Read November 19 , 1891 .
Revised June 20 , 1892 .
IN a paper read before the Society on June 11 , Sir WILLIAM THOMSON expressed a douOt as to the general truth of the doctrine concerning the distribution of energy among a great number of mutually acting bodies , and suggested that certain test cases should be investigated .
The test that he proposed on that occasion was a number of hollow elastic spheres , each of mass , and each containing a smaller elastic sphere of mass , free to move within a larger one .
This he calls a doublet .
This case is within the general proof of the doctrine given below .
It is , however , I amenable to a simpler treatment , which has been applied to the case of elastic spheres external to one another .
1 .
Every doublet has a centre of inertia of the sphere and its imprisoned Let V be the velocity of that centre of inertia , the relative velocity of and If V and be given in magnitude , given in direction , V may have any direction , and in MAXWELL'S distribution , for given direction of , all directions of V are equally probable .
Conversely , if , whatever be the values of V and , for given direction of all directions of V are equally probable , MAXWELL'S law prevails .
Now consider a very great number of doublets , all having their reIative velocity and the velocity of centre of inertia within limits , and V , .
Consider them before and after collisions between and .
Nothing is changed by collision except the direction of , and that change of direction is independent of the direction of V. Therefore after collision for given direction of all directions of V are equally probable , and therefore distribution prevails after as well as before collision , and is therefore not affected by collisions .
To proceed to more general cases .
2 .
The characteristic property of collisions of conventional elastic bodies is that with continuous variation of the coordinates , and without variation of kinetic energy , the velocities at a certain instant change discontinuously .
The general treatment adapted to systems of this kind is as follows : there be a system defined by coordinates , the corresponding velocities bein . .
, and the generalised components of momentum . . .
At a certain instant a , discontinuity of occurs .
After the collision , let the velocities and 3,8.92 momenta be denoted by .
Let be the potential .
In the configuration .
there are independent linear functions of the forces , , each equal to zero .
We might then find new coordinates . .
such that .
, and therefore if , be the components of momentum corresponding \amp ; c. , and are finite or zero .
In the limit when the relnaining force becomes infinite and acts for an infinitely short time or , and restoring the original coordinaes .
( A ) , in which the coefficients are ions of the ates .
And therefore -llinear fnnctions of the 's are unaltered by the collision , namely:\mdash ; suppose . .
( B ) .
Again , since the kinetic energy is unchanged , and by the properties of generalised coordinates Therefore The last equation , with the equations ( A ) , a system of equations , all linear as regards , \amp ; c. Since \amp ; c. , are not all zero , we must equate the determinant of the system ( A ) and ( C)to zero .
That gives us a linear equation in \amp ; c. We can now substitute for the 's their values in lerms of , and so obtain a linear equation MR. S. H. BURBURY ON THE COLLISION ELASTIC BODILS .
\amp ; c. where \amp ; c. are functions of the coordinates and constants .
That is , we have in all linear functions of the velocities , namely , which remain unaltered by the collision , and one other linear fumction , , which remains unaltered in value , but .
That must be the case on every collision of elastic bodies .
3 .
The kinetic energy may be expressed in terms of the variables and , in lieu of the velocities . .
, and since it is not aItered by the collision , which changes the sign of unaltered , it must be of the form in which is a quadratic function of with coefficients functions of the coordinates and constants of the system , and is a metion of the coordinates and constants .
4 .
The system , after collision , has velocities , \amp ; c. , which we will call the second state .
We may conceive a system with the same coordinates velocities - \amp ; c. , and call this the second state with reversed velocities .
In this state , will have opposite signs to those they have in the first state , and has the same sign as in the first state .
The system its coulse , and a collision occurs changing into - , \amp ; c. , leaving . .
unaltered , and changing into \mdash ; R. 5 .
To define a collision , we may suppose that a certain function , , of the coordinates and constants is prevented by the physical conditions of the system from becoming positive .
When becomes zero , from being positiv becomes discontinuously negative , and a collision is said to take place .
We may take for one of our generalised coordinates in lieu of , and , or , for the corresponding component of velocity .
The kinetic energy is a function of , and we may express it as a function of . .
, and , where . .
are the constants found above .
Since the kinetic energy is not altered by the discontinuous change in , whatevel the values of , it must be of the form .
That is is reversed in sign , but unaltered in magnitude by the collision , and is , therefore , equal or proportional to found above .
6 .
What we have proved for a system is of course lrue if for system we wlite of systems .
For instance , let there be two sets of systems : ( 1 ) systems defined by coordinates .
, and velocities , and ( 2 ) systems defined by COOl.di nates and velocities .
and .
If is a function of .
such that when a collision\mdash ; i.e. , a discontinuity of the velocities\mdash ; occurs , we may treat the pair of systems in all respects as a single system within the preceding investigat.ion .
7 .
All those systems , or pairs of systems , for which any instant lies between zero and St , being positive , will undergo collision within the time after that ant .
We may , therefore , take or as measuring the frequency of collisions for given values of \amp ; c. 8 .
From the linear equations ( B ) above given we can find any of the velocities , for , as a linear function of , and will be the same function with written for R. Hence where and the 's are functions of the coordirlates and constants .
Aiso Now , without altering or , or the coordinates , let us make pass through the whole range of values consistent with . . . . .
( ) .
Also let be the number of systems for which these variables lie between the limits and and and and the coordinates being constant .
Then R. .
1 the integrations being over all values consistent with ( E ) .
Now , in the MAXWELL-BOLTZMANN distribution is a function of the kinetic energy only , and is therefore constant throughout this integration .
fore the right-hand member of the last equation is zero , because for any given set of values of 1 satisfying ( E ) , there is a corresponding set with the opposite signs also satisfying it .
Therefore .
and , therefore , the average value of for all collisions given and is zero .
The same is true for , \amp ; c. ( 1 ) , ( 2 ) , ( 3 ) , ( 4 ) .
Four tangential components of velocity .
( 5 ) .
The angular velocity of the spheroid round an axis dicular to in the plane of the ) And if be normal components of velocity , and the angular velocity round an axis perpendicular to the plane of the paper , the following two , viz. :\mdash ; By conservation of momentum , ( 6 .
) By conservation of moment of momentum round the axis through perpendicular to the plane of the paper , ( 7 .
) hIcU We then form the equations and by conseryation of energy , ; and equating the determinant to zero we obtain , neglecting common factors , and , using V , , and for and in the general equations , we see that if the MAXWELL-BOLTZMANN distribution of energy it is not disturbed by collisions between spheres and spheroids .
III .
Professor BURNSIDE'S problem ( see his paper , Roy .
Soc , Edinburgh , July 18 , 1887 ) .
He supposes a number of similar and equal spheres , each of unib mass , but each sphere , instead of being homogeneous , has its centre of inertia at a distance from its centre , being supposed very small compared with the radius .
The principal moments of inertia are for each sphere , and the direction cosines of referred to the principal axes through the centre of inertia are for each sphere the same , viz. , .
The direction cosines of the line of centres at impact referred to the principal axes are .
one sphere , and for the other Further the normal velocities are , and the angular velocities round the principal axes are for the two spheres respectively .
lally we write , MR. S. H. BURBURY ON THE COLLISION OF ELASTIC BODIES .
The system of two spheres has twelve degrees of freedom .
We require , therefore , eleVen linear functions of th velocities to be invariable .
They are as follows , viz. , four components of velocity in the common tangent plane , , and seven others , viz. , cru As before , we form the equations , and by the conservation of energy , and equating the determinant to zero , We can now substitute V , for in the general equation , and we obtain , as before , the result that the distribution , given existing , is not affected by collisions .
10 .
Professor BURNSIDE obtains the same eight equations as above given , and I acknowledge my obligation to him , but he originally deduced the lesult that the energy of rotation is twice the energy of translation , instead of equal to it , as , according to the theory , it should be .
He has since seen reason to .
his ) with regard to this problem .
ME . .
H. BUli13 ON COLLISION LASTIC BODIES .
volume and time of colliding pairs for which the angle between V and before collision is veen and is , and number for which after collision it lies between and is .
Hence we have to multiply by , and integrate in each case to .
The result is And so we get Nn where V , are usual ioal coordinates Also , as we have seen and therefore suppose .
And if initial values , the rate of subsidence is directly proportional to the density and to the square root of the absolute temperature .
* Since writing the bovo I Iind this esult has are , ady been obtained for the case of elastio pheres by Professor , by an independent method .
[ 407 ] X. On the Collision of Etastic S. H. BURBURY , F.R. Received October 24 , \mdash ; Read November 19 , 1891 .
Revised June 20 , 1892 .
IN a paper read before the Society on June 11 , Sir THOMSON expressed a as to the general truth of the MAXWELL-BOLTZMANN doctrine concerning the distribution of energy among a great number of mutually acting bodies , and suggested that certain test cases should be investigated .
The test that he proposed on that occasion was a number of hollow elastic spheres , each of mass , and each containing a smaller elastic sphere of mass , free to move within a larger one .
This he calls a doublet .
This case is within the general proof of the doctrine given below .
It is , however , I think amenable to a simpler treatment , which has been applied to the case of elastic spheres extelnal to one another .
1 .
Every doublet has a centre of inertia of the sphere and its imprisoned Let V be the velocity of centre of inertia , the relative velocity of and If V and be given in magnitude , given in direction , V may have any direction , and in MAXWELL'S distribution , for given direction of , all directions of V are equally probable .
Conversely , if , whatever be the values of V and , for given direction of all directions of V are equally probable , MAXWELL'S law prevails .
Now consider a very great number of doublets , all having their relative velocity and the velocity of centre of inertia within limits , and V , .
Consider them before and after collisions between and .
Nothing is changed by collision except the direction of , and that change of direction is independent of the direction of V. Therefore after collision for given direction of all directions of V are equally probable , and therefore MAXWELL'S distribution prevails after as well as before collision , and is therefore not affected by collisions .
To proceed to more general cases .
2 .
The characteristic property of collisions of conventional elastic bodies is that with continuous variation of the coordinates , and without variation of kinetic energy , the velocities at a certain instant change discontinuously .
The general treatment adapted to systems of this kind is as follows : there be a system defined by coordinates , the corresponding velocities bein . .
, and the generalised components of momentum . . .
At a certain instant a , discontinuity of occurs .
After the collision , let the velocities and 3,8.92 momenta be denoted by .
Let be the potential .
In the configuration .
there are independent linear functions of the forces , , each equal to zero .
We might then find new coordinates . .
such that .
, and therefore if be the components of momentum corresponding \amp ; c. , and are finite or zero .
In the limit when the remaining force becomes infinite and acts for an infinitely short time or . .
, and restoring the original coordinates . .
, .
( A ) , in which the coefficients are functions of the coordinates .
And therefore -llinear functions of the 's are unaltered by the collision , namely:\mdash ; . . . . .
suppose . . . . .
( B ) . . .
Again , since the kinetic energy is unchanged , and by the properties of generalised coordinates Therefore The last equation , with the equations ( A ) , a system of equations , all linear as regards , \amp ; c. Since \amp ; c. , are not all zero , we must equate the determinant of the system ( A ) and ( C)to zero .
That gives us a linear equation in \amp ; c. We can now substitute for the 's their values in terms of , and so obtain a linear equation MR. S. H. BURBURY ON THE COLLISION ELASTIC BODILS .
, \amp ; c. . . . . .
where , \amp ; c. are functions of the coordinates and constants .
That is , we have in all \mdash ; llinear functions of the velocities , namely , which remain unaltered by the collision , and one other linear fumction , , which remains unaltered in value , but changes .
That must be the case on every collision of elastic bodies .
3 .
The kinetic energy may be expressed in terms of the variables and , in lieu of the velocities . .
, and since it is not altered by the collision , which changes the sign of unaltered , it must be of the form in which is a quadratic function of . .
with coefficients functions of the coordinates and constants of the system , and is a fimction of the coordinates and constants .
4 .
The system , after collision , has velocities , \amp ; c. , which we will call the second state .
We may conceive a system with the same coordinates having velocities - \amp ; c. , and call this the second state with reversed velocities .
In this state , will have opposite signs to those they have in the first state , and has the same sign as in the first state .
The system its course , and a collision occurs changing into - , \amp ; c. , leaving . .
unaltered , and changing into \mdash ; R. 5 .
To define a collision , we may suppose that a certain function , , of the coordinates and constants is prevented by the physical conditions of the system from becoming positive .
When becomes zero , from being positiv becomes discontinuously negative , and a collision is said to take place .
We may take for one of our generalised coordinates in lieu of , and , or , for the component of velocity .
The kinetic energy is a function of , and we may express it as a function of . .
, and , where . .
are the constants found above .
Since the kinetic energy is not altered by the discontinuous change in , whatevel the values of , it must be of the form .
That is is reversed in sign , but unaltered in magnitude by the collision , and is , thel'efore , equal or proportional to found above .
6 .
What we have proved for a system is of course true if for system we write pair of systems .
For instance , let there be two sets of systems : ( 1 ) systems defined by coordinates .
, and velocities , and ( 2 ) systems defined by COOl.di nates and velocities .
and .
If is a function of .
such that when a collision\mdash ; i.e. , a discontinuity of the velocities\mdash ; occurs , we may treat the pair of systems in all respects as a single system within the preceding investigat.ion .
7 .
All those systems , or pairs of systems , for which any instant lies between zero and St , being positive , will undergo collision within the time after that ant .
We may , therefore , take or as measuring the frequency of collisions for given values of \amp ; c. 8 .
From the linear equations ( B ) above given we can find any of the velocities , for , as a linear function of , and will be the same function with written for R. Hence where and the 's are functions of the coordirlates and constants .
Aiso Now , without altering or , or the coordinates , let us make pass through the whole range of values consistent with . . . . .
( ) .
Also let be the number of systems for which these variables lie between the limits and and and and the coordinates being constant .
Then R. .
1 the integrations being over all values consistent with ( E ) .
Now , in the MAXWELL-BOLTZMANN distribution is a function of the kinetic energy only , and is therefore constant throughout this integration .
fore the right-hand member of the last equation is zero , because for any given set of values of 1 satisfying ( E ) , there is a corresponding set with the opposite signs also satisfying it .
Therefore .
and , therefore , the average value of for all collisions given and is zero .
The same is true for , \amp ; c. MR. S. H. BURBURY ON THE COLLISION OF And , therefore , since and are arbitrary , , \amp ; c. , are not altered by collision at all , that is , the distribution , given existing , is not altered by collisions .
The above proof also shows that is zero without the factor , that is , the average value for all systems is zero , as well as for all collisions ; and in proving that , it does not matter whether we introduce the factor or not .
9 .
We will give certain examples of the functions . .
R. I. Elastic spheres of masses and respectively .
Here a colliding pair , which corresponds to a system in the general treatment , has six degrees of freedom .
There should , therefore , be five linear functions of the velocity unaltered by collision .
They are , the tangential components of velocity at the instant of collision , and where are the normal components .
We have from the last equation and by the equation of energy whence or II .
The system consists of a sphere of mass colliding with a spheroid of mass M. It is assumed thab the spheroid will acquire no rotation about its axis of figure , but may have rotation about any other principal axis .
It has then five degrees of freedom , and the system of spheroid eight .
We require , then , seven linear functions of the velocities to be invariable .
Let be the centre of the spheroid , in the plane of the paper , its axis of figure , the point of collision , PN normal at , ON perpendicular to , and ON , A the moment of inertia of the spheroid round an axis through perpendicular to .
Then our seven constants are i3 ( 1 ) , ( 2 ) , ( 3 ) , ( 4 ) .
Four tangential components of velocity .
( 5 ) .
The angular velocity of the spheroid round an axis dicular to in the plane of the ) And if be normal components of velocity , and the angular velocity round an axis perpendicular to the plane of the paper , the following two , viz. :\mdash ; By conservation of momentum , ( 6 .
) By conservation of moment of momentum round the axis through perpendicular to the plane of the paper , ( 7 .
) hIcU We then form the equations and by conseryation of energy , ; and equating the determinant to zero we obtain , neglecting common factors , and , using V , , and for and in the general equations , we see that if the MAXWELL-BOLTZMANN distribution of energy it is not disturbed by collisions between spheres and spheroids .
III .
Professor BURNSIDE'S problem ( see his paper , Roy .
Soc , Edinburgh , July 18 , 1887 ) .
He supposes a number of similar and equal spheres , each of unib mass , but each sphere , instead of being homogeneous , has its centre of inertia at a distance from its centre , being supposed very small compared with the radius .
The principal moments of inertia are for each sphere , and the direction cosines of referred to the principal axes through the centre of inertia are for each sphere the same , viz. , .
The direction cosines of the line of centres at impact referred to the principal axes are .
one sphere , and for the other Further the normal velocities are , and the angular velocities round the principal axes are for the two spheres respectively .
lally we write , MR. S. H. BURBURY ON THE COLLISION OF ELASTIC BODIES .
The system of two spheres has twelve degrees of freedom .
We require , therefore , eleVen linear functions of th velocities to be invariable .
They are as follows , viz. , four components of velocity in the common tangent plane , , and seven others , viz. , cru As before , we form the equations , and by the conservation of energy , and equating the determinant to zero , We can now substitute V , for in the general equation , and we obtain , as before , the result that the distribution , given existing , is not affected by collisions .
10 .
Professor BURNSIDE obtains the same eight equations as above given , and I acknowledge my obligation to him , but he originally deduced the lesult that the energy of rotation is twice the energy of translation , instead of equal to it , as , according to the theory , it should be .
He has since seen reason to .
his ) with regard to this problem .
The equations given by Professor BUBNSIDE can easily be modified so as to meet the case of elastic bodies of any shape .
11 .
lt would not be difficult to extend the method of ( 8 ) and show that the MAXWELL-BOLTZMANN distribution is a necessary , as well as a sufficient condition for stationary motion .
But that is more completely done by following or extending BOLTZMANN'S method .
Let there be a set of systems which we will call systems , whose co-ordinates and velocities are ] .
and .
Let the number of stlch systems for which at any instant the co-ordinates lie between the limits ( A ) , and the velocities between the limits ( A ) , be or , shortly , Let there be another set of systems , which we will call systems , whose coordinates are .
and velocities .
And let the number of systems for which at any instant the coordinates lie between the limits and ( B ) , and and the velocities between the limits and ( B ) , and be . .
MR. S. H. BURBURY ON THE COLLISION ELASTIC BODIES .
or , shortly , The number of pairs of systems , each consisting of one system from each set , whose coordinates and velocities at any instant lie within the above limits , is . .
Now let be a function of the coordinates which cannot become positive , and such that when the velocities change discontinuously , and a collision occurs .
We may use for one of our coordinates , expressing in terms of and , and in like manner in terms of .
and .
All those pairs for which at the given instant lies between zero and being positive , will undergo collision within the time St after that instant .
And so the number of such collisions in unit time is , writing for .
Ff Here is a function of .
only , but , by virtue of the elimination of and , is now a funclion and . .
, R. Each of these collisions changes into , \amp ; c. , that is , changes the system from the first into the second state .
The number in unit time of collisions , which with the same coordinates change the system from the second state with reversed velocities into the first state with reversed velocities , is . .
in which are the same nctions of \amp ; c. , as are of \amp ; c. By a known theorem , .
and in distribution .
And so the number of reverse collisions is in that distribution equal to the number of direct collisions .
And this insures the permanence of the distribution .
It is assumed that there are always as many systems with any given set of velocities with those velocities reversed .
And so we speak of the second state with reversed velocities as equivalent to the second state .
12 .
If , the number of reverse collisions is not equal to the number of direct collisions .
And , therefore , more ( or fewer ) pairs of pass out of the first state into the second than vice versed .
In that case the number of systems whose coordinates and velocities are within the limits , is increased by collisions .
S. H. BURY ON TH1 COLLISION OF ELASTIC BODIES .
with systems , whose coordinates and velocities lie within the linits , by the quantity ( per unit of time , and is increased per unit of time by collisions with systems for all values of .
by the quantity all of , \amp ; c. , being included in the integration .
We will now assume ( see p. 418 , post ) that the velocities of and are not on average altered except by collision between and .
And so the above-mentioned increments are the only increments by which the class of systems within the limits is ffected .
In that case and , therefore , ( over all values of ) By symmetry , as the right-hand member includes all possible collision between and .
and , therefore , .
By symmetry , as we may interchange the accents , MR. S. H. BURBURY ON TH COLLISION OF LLASTIC BODY b therefore , . .
Now let . .
and , therefore , ' as we have seen .
Now , this expression is necessarily tive , unless whenever the of systems having coordin ates and velocities .
can pass by collision , clnd , efore , with unchanged kinetic energy , into the state in which they that is , unless MAXWELL-BOLTZMANN distribution exist , and is then zero .
tends to a minimum which it reaches when We will now where is the minimum value ssumed by when .
Then and , and is always negative .
We may define the function to be tl , disturbance , and to be the ate of subsidencc of the disturbance by collision .
In certain cases , we can calculate the rate .
13 .
We have assumed that varies only as the result of collisions .
is , if denote the time variation of due to causes other than collisions , and be formed from as from , then , on average .
It orth while to consider on what condition this may be safely Ssumed .
Let As we are dealing with elastic bodies under the action of no forccs , we treat as function of translation velocities , and three angular velocities , , about three principal axes of the solids .
Let , be the moments of inertia .
Evidently , there being no forces , the ) velocities cannot vary except as the result of collisions .
But for each solid , the law of their variations being EULER'S equations .
We may , , in ) treat as a fnnction of ) only .
Then ' and with two other corresponding terms , the limits being in each , with these limits , Now , we may assume and , when any one of the three variables is infinite , whether positive or negative .
And this assumption is sufficient to justify the statement 14 .
It is possible to calculate , in a simple case , the rate at which a disturbance ubsides by collisions .
For example , two sets of ela , stic spheres , of mass , and of mass , in unit of volume .
In the normal state , the number in unit of volume , whose velocities are represented by lines drawn from an origin to points within an element of volume is for the spheres where are usual } ) herical coordinates ; or , let us say , Similarly , for the spheres , ' expresscs the law of distribution of in the undisturbed state .
We will and for these expressions .
We will now suppose there is a small turbance consisting in having different values for the two sets .
Let be written for the spheres , and for the ?
spheres .
We shall neglect third and higher powers of .
Then becomes in the disturbed state MR. S. H. BURBUR ON LISION OF ELASTIC BODY that is , terms in will appear that terms required .
Similarly becomes \amp ; c. .
We will further suppose that the disturbance is introduced without the total energy .
That gives the ] abion .
, or 15 .
The disturbance will subside by collisions between and ? .
And we will treat of the case in which it subsides in such manner that the above values of and apply at every instant with the values that and that instant have .
Such a mode of subsidence is possible , at all events if our equations lead ( as they do ) bo a relation of the form constant .
Let us then form the function where , and are usual spherical coordinntes , where and have the values above given .
That is siu terms independent of 1 ) and , terms independent of and because .3 112 .
S. H. COLLISION OF LASTIC The tel'lns independent of and are equivalent to , the minimum value of when there is no disturbance .
And so Nlog ) .
That is .
Now since and N. Iu order to find we will transform our coordinates thus : Let V denote the velocity of the centre of inertia of a pair of spheres and their relative velocity , the angle between and .
Then \mdash ; - and and so where are the values of after collision .
MR. S. H. BURBURY ON THE OF } Therefore Nn Now , if be what becomes after collision , Also by the relation ' and hese substitutions Nn .
Also forming we see that the last ctor is squared , and so the product contains the factor .
We , therefore , now write 1 for and , and also write , otherwise should have terms in Therefore Again , contains the factor denoting the frequency of collision .
Now number of collisions given V and in unit of volume and time is to , where is the sum of the radii of and 7/ .
Also all directions of collision are equally , and given the direction before collision , all directions after collision are equally probable .
Therefore , given and , the number in unit ME . .
H. BUli13 ON COLLISION LASTIC BODIES .
volume and time of colliding pairs for which the angle between V and before collision is veen and is , and number for which after collision it lies between and is .
Hence we have to multiply by , and integrate in each case to .
The result is And so we get Nn where V , are usual ioal coordinates Also , as we have seen and therefore suppose .
And if initial values , the rate of subsidence is directly proportional to the density and to the square root of the absolute temperature .
* Since writing the bovo I Iind this esult has are , ady been obtained for the case of elastio pheres by Professor , by an independent method .

