]\gt ; 397 ] VIII .
The Distribution of Motecular Energy .
J. H. JEANS , B.A. , Scholar of College , and Newton Student in the University of Cambridgc .
by Profe , J. J. THOMSON , F.R. Received June 14 , \mdash ; Read , 1900 .
INTRODUCTION .
S 1 .
THIS paper is primarily an attempt to deal with certain points connected with the application to the Kinetic Theory of Gases of BOLTZMANN'S Theorem on the partition of energy in a dynamical system .
It is found by experiment that the ratio of the two specific heats of certain monatomic gases mel.cury , argon is .
If we admit that the energy of these gases is distributed in the manner indicated by Theorem , then this theorem leaves no escape from the conclusion ) the molecules of these gases must be rigid and geometrically perfect spheres .
A similar difficulty arises in connection with other gases : the number of degl.ees of freedom which a consideration of the ratio in question leads us to expect a molecule of a gas to possess , is always less than the number which the spectrum of ] ) glowing gas shows to actually exist .
Further , BOLTZMANN'S Theorem excludes the possibility of the ratio of the two specific heats having any values except one of a certain sel.ies of values , whereas experiment shows that the ratio is not always equal to one of this series , although it is generally very near to such a value .
Finally Theorem leaves no room for a variation of this ratio with the temperature , although such a variation is known to exist .
In the present paper I have tried to suggest a way by which it is possible to escape from this dilemma .
As there is not sufficient known about the constitution of a molecule to enable it to be completely ) ecified as a dynamical f , ystem , the paper is limited to the consideration of two imaginary types of molecules .
The conclusioroe arrived at are the same in each case .
In the first place the distribution of energy which is given by BOLTZMANN'S Theorem is the only distribution which is permanent under the conditions postulated by this theorem .
And in the second place , this law of distribution may break down entirely as soon as we 22.5.1901 admit an interaction , no matter how small , between the molecules and the surrounding ether .
That such an interaction must exist is shown by the fact that a gas is capable of radiating energy .
In fact , BOLTZMANN'S Theorem rests on the assumption that the molecules of a gas form a conservative dynamical system , and it will appear that the introduction of a small dissipation function may entirely invalidate the conclusions of the theorem .
* Thus we may regard the BOLTZMANN distribution as unstable , in the sense that a slight deviation from perfect conservation of energy may result in a complete redistribution of the total energy , and it will appear that this new distribution of energy will lead to values for the ratios of the two specific heats which are not open to the objections mentioned above .
S2 .
A second difficulty , of a mathematical rather than physical nature , may be mentioned here , as it will occur incidentally in the course of the analysis .
It is well illustrated by the consideration of an imaginary type of molecule which has been suggested by BOLTZMANN .
A loaded sphere , that is to say , a sphere of which the centre of gravity is at a small distance from the geometrical centre , will , for the present purpose , possess five degrees of freedom , and this is true , however , small may be .
The ratio of the specific heats of an ideal gas whose molecules are of this type ought accordingly to be .
If , however , actually vanishes , the molecules must be regarded as completely symmetrical , and possessing only three degrees of freedom , so that the ratio ought now to be .
There is thus seen to be a discontinuity when has a zero value , and this requires investigation .
It must be borne in min that a degree of freedom , for purposes of BOLTZMANN'S Theorem , is not idendical with the usual dynamical degree of freedom .
In the strict dynamical sense every sphere possesses six degrees of freedom , the principal momenta being the momenta of the of gravity in three rectangular directions and the three rotatory momenta about the principal axes of inertia .
But if the sphere is perfectly smooth , rigid , and symmetrical , it is impossible to change the three latter momenta by the application of any forces which are at our disposal in the kinetic theory of gases , and for this reason the corresponding degrees of freedom must be left out of account , when applying BOLTZMANN'S Theorem .
Similar remarks apply , in the case of the loaded sphere , to the degree of freedom which arises from rotation about the axis of symmetry , so that the loaded sphere must be supposed to possess five degrees of freedom , and not six .
* The matter may be looked at from a slightly different point of view as follows : If an interaction between matter and ethel exists , no matter how small this interaction may be , the complete dynamical system will consist of the molecules of the gas , togethel with the ether , and must therefore be regarded as a system possessing an infinite number of degrees of fresdom .
Applying BOLTZMANN'S Theorem to this system we arc merely led to the conclusion that no steady state is possible until all the energy of the gas has been dissipated by radiation into the ether .
This application of theorem may or may not be is , I think , certain that no other application is legitimate .
'Vorlesungen uber Gastheorie , ' II .
, p. 129 .
MR. J. H. JEANS ON THE DISTRIBUTION OF MOLECULAR ENERGy .
It is obvious that the degrees 01 freedom which have been rejected do not in any way influence the motion of the sphere , whereas if they are counted as separate degrees of freedom , the series of states through which the sphere is made to pass by varying all the co-ordinates , is no longer such as to satisfy MAXWELL'S condition of Continuity of Path ( See S 11 of the present paper .
) PART I. THE DISTRIBUTION 0F ENERGY 1N A GAS 0F WH1CH THE MOLECULES ARE LOADED SPHERES .
The of Energy caused by Collisions .
S 3 .
We may begin with the consideration of a gas of which the molecules are loaded spheres of the kind mentioned in the last section .
These spheres are to be perfectly elastic , each of radius , and the centre of gravity of each is to be at a small distance from the geometrical centre .
We require to find equations giving the transfer of energy between the various degrees of freedom in such a gas .
If we know the law of distribution of the various co-ordinates of the molecules , we shall be able , upon making the usual assumptions of the kinetic theory , to calculate the number of collisions which are such that the values of the variables , which are required to completely specify a collision , lie within certain specified small ranges of value .
At each of these collisions the transfer of energy is the same , a function of the variables which specify the collision ; so that by multiplying this quantity by the number of collisions of the kind under consideration which occur during the interval , and integrating over all possible values of the yariables which specify the collision , we shall obtain an expression for the transfer of energy during the time Let us denote the mean energy of translation of all the molecules at any specified instant by , the mean energy of rotation by H. If we regard the number of molecules in the gas as infinite , we may regard the quantities and as varying continuously with the time , and the expressions which have just been found for the transfer of energy will enable us to calculate and corresponding to any state of the gas .
The values of will in general depend on ( the density of the gas ) and the coefficients which occur in the various laws of distribution of co-ordinates .
If we consider the case in which , we find that no transfer of energy is possible , so that must vanish with .
We further notice that these differential coefficients must remain unaltered , if be written for , so that , assuming for the moment that they can be expanded in ascending powers of , we see that the lowest power of which can occur is .
We shall suppose to be so small that terms in may be neglected in comparison with terms contaimng Since the total energy remains unaltered at every collision , we must have ; it is therefore only necessary to calculate one of these differential coefficients .
S 4 .
The state of a molecule at any instant will be determined by the following 12 variables , The co-ordinates in space of its centre of gravity , and their time-rate of change ( ii .
) Any three independent variables , , specifyin the orientation in space of molecule .
( iii .
) , the rotations about three principal axes , the last of these being the rotation about the axis of symmetry of the molecule .
Let accented letters refer to a second molecule ; then a collision between these two molecules , if possible , is completely specified by the whole 24 variables , but these are not all independent , and the collision will , as regards transfer of energy , be sufficiently specified by the independent variables and six other variables to determine the direction in space of the axes of the molecules , and the line of centres .
Let the variables after collision be distinguished from those before collision by placing a bar over them , then we can from the ordinary equation of impact calculate the value of where , in terms of the variables before collision .
This expression must be a quadratic function of the velocities , and cannot enter .
If we write it is easily seen that the expression must be of the form . . . . .
, where are nctions of the six variables determining orientations in space , and are algebraical functions of , in which the lowest power is IslR .
J. H. JEANS ON THE DISTRIBtlTION OF MOLECULAR ENEItGY .
S 5 .
In calculating the number of collisions of this kind which are to be expected in the interval of time , a consideration enters , which does not enter in the simpler case in which the spheres are symmetrically loaded .
From the co-ordinates of the two molecules just before collision , we can trace back , as far as the previous collisions , the paths by which the molecules arrived at this position .
If these paths are such that the spaces occupied by the two molecules , at any two corresponding points of these paths , are found to overlap , then it is clear that a collision of the kind we are investigating can only occur , either when the same two molecules have previously collided , or when one of them has collided with a third molecule within a certain small interval previous to the collision in question .
In either case it would be wrong to calculate the probability of such a collision upon the assumption that the molecules of the gas .
, in BOLTZMANN'S sense , When , however , terms of degree higher than are neglected , it will be legitimate to ignore this consideration altogether .
For the number of collisions to which it applies will vanish with , so that if equation be summed oyer all collisions , the terms on the right-hand side which are influenced by this consideration will be of a higher order in than ?
, and may accordingly be inaccurately calculated , without invalidating the result as far as terms in S6 .
When we agree to ignore this consideration , we may at once average equation over all values of the six variables of orientation .
The probability of these val.iables having specified values at a collision is not independent of the velocities of the collision , but will be the same for all collisions such as we are now considering , in which these velocities have specified .
In this way we find that the mean increase in at a collision at which the yelocities are , is of the form terms of a higher order in . . .
( ii .
) , in which are constants .
Now if we suppose that the gas has reached its present state through a series of natural processes , the law of distribution of velocities will depend only on and .
In the case in which , this law is known to be . . . . .
( iii .
) .
Hence in the case in which is small , it may be taken to be . . . . . . .
, where is a function of which the coefficients involve , but is such ( iv .
) reduces to ( iii .
) when * Direct calculation shows that the values of are , where is the radins of gyration of a molecule bout a line perpendicular to the axis of uetry .
VOL. CXCVI .
} .
J. H. JEANS ON THE OF MOLECULAR Calculated upon the usual assumption , the number of collisions which occur in a volume of the gas and within a time , between pairs of molecules of which the velocities lie within a range du surrounding the values , 1 , is Hence referring to expression we see that the total increase in the translational velocity of the gas , in time , is If we reject all terms of a degree fher than in this expression becomes .
Now the functional form represented by is unknown , but the part of the above which contains depends only upon and this can be seen to be proportional to and to involve .
Let us denote by , so that I is a function of only ; then the part of which contains contains multiplied by .
and a function of The part of which contains depends on I and also on .
If we write ( so that is the mean value of taken over all the molecules of the gas ) , then this part of will be multiplied by and by a function of Hence determining the functions of from a consideration of , we find .
( vi .
) , in which are constants ; or in terms of and ( vii .
) , in which the constants do not in any way depend upon the law of distribution of velocities .
the vnlues for given in the footnote on p. 401 , we find for the Yitlues In this we can prove the relation , instead of assuming it .
MR. J. H. JEANS ON THE DISTRIBUTION OF In BOLTZMANN'S steady state , , and , since the energy is equally divided the five of freedom .
This leads to the relation and equation ( vii .
) may be Wl.itten ( viii .
) .
The rate of tion of account of collisions is therefore given by .
( ix .
) .
Introduction of S7 . .
the changes in the values of and arise solely from collisions , the equations just found enable us to these changes , starting any initial values .
Let us , }yever , suppose ] there is second cause of change in and ; suppose that the angular velocity is by a reaction between the molecules and the s medium , the a1nonnt of retardation being .
It is 1lot suggested reaction this type exists in nature , ) this reaction may be taken ils being ] ) reaction ) ossible , just a has taken as the ) .
Due to a reaction of this kind , mean rotational will suffer a decrease of amount unit The .
equations ) of change of and therefore be . . . . .
, .
S 8 .
A steady state is ) ossible , on account of the dissipation of energy takes place .
If , , we ) posetl } energy to be in some such ils exactly to loss of energy , rapidly steady tate given by , or , by equation ( xi .
) , . . . . .
( xii .
) .
The value of ponding to given value of is tlJerefore ( xiii .
) .
Hence if have values that is very great in compal.ison with then will be very small in comparison with K. The rate at which the e.xterIlal agency introduces enel.gy is , and this may be madevel-.y small by supposn to be very small , although in vitlu .
If this rency is removed J. ON OF CULAR ENERGy .
the state of the gas will only very slowly , so that the state specified by ( xiii .
) may be the " " steady\ldquo ; state .
vanishes iu coml ) arison with giving this state takes the form . . . . .
( xiv .
) conditions which have been found to be necessary in .
that this state exist , are should be small , and that should be so small that 11 be ossible for all temperatures below certain ture , namely , the temperature which to be comwith .
Below this temperature vanishes in comparisoll with , and the rate of dissipation of energy is a small quantity ) the second order .
It follows that if ) erimentsal.e conducted ) el.atures so low to be this critical ture , no value of possibly be observed except . . . . . .
At temperatures , there is no definite ratio between which tends to establish itself In , if conducted with a view to determining the value obseryed will depend on the past history of the the duration of the experiment , so that may have any value between Thus it ppears t if , under the conditions we are , a consistent value is obtained for from nents on the in question , this value can be no other than , and hc ture at which the ) el.iments are conducted must been referred to as a low ture .
It must be particularly noticed , that } ) is only low relatively to the other temperatures considered : no knowledge as to its bsolute value is possible so and remain unknown quantities .
If , however , for the momeut , we assume that the molecules are .
representation of molecules of an actual , and that the dissipation of energry used by olllssumed fctions s ) a true analogy to tion of , then we can some e as to a " " low\ldquo ; , must .
It is which , and the that is to , it is it } ) ture at which the gas is ndescent .
Dislribution ) Steady State . .
To SU111 up , it appears that if we willing to cudmit that our present dynamisystem ) a sufficiently good analogue to a real gas , then the introduction of a alion function will supply an explanation of the difficulties mentioned in the introduction , at any rate for the case of a non-luminous gas .
Part IL of this consists of an effort to show that our present system is a fair analogy , if not MR. J. H. JEANS ON THE DISTRIBUTION OF ULAR 1 of an actual molecule , yet at any rate , of a dyninnical system which contains the features we believe to be essential to molecule .
Referring back to tions ( , ( xiv .
) , it is ) luminous } ' , the equations deterlnini n will be , and . . . . .
( xvii .
) .
Equation ( xyii .
) is the between ) equation of XWLLL dzMANN , .
: It therefore ) , in the presenlt c total radiation will be ) to and in the more general case discussed in Il .
, the radiation will be seen to increase still more rapidly with the temperature .
it is easy to see how it is possible { the total to very } ) idly n the of candescence , whereas if we ) the energy divided in between different degrees of , it ; difficult to see tion c ) , but directly ] ) } ) to ] letell } ) Extension of S 10 .
It is , under certain conditions , to the above metbods to more general type of molecule .
Let the energy the molecule consist of partly of various kinds of internal energy , potential well kinetic .
The only case considered will be that in which the internal energy is small : potential will thel.efore arise from small oscillations about a position of these tions will be of definite period , and such as be supposed to esult in the emission of light ) a line-spectrum .
Thus the total to any such mode of vibration , will , when yeraged ovel nulnbe of molecules , be half ) half kinetic .
It is success of the present method the of a collision between two molecules should depend solely on elative velocity , not on .
internal -ordinates .
Now a is to be rded c energy , and a rotation will be equivalent to volume , itnd will therefore increase the probability of a collision unless the molecules of invariable radius , itnd of which the centres moye in lines .
the molecules must either be ) of which the centre of gravity the centre coincide , or else as in the former case , they must differ by so little this , that the divergence no effect un the final result .
We slJallful .
there suppose for ) resent that the internal energies are only slightly altered by collision , as was the case in the former )roblem .
The law of distribution of will now be dependent of the velocity of translation , so that mean value of the energy of any specified mode , be the same when ) the is taken over all collisions or over all molecudes .
The same notation as before will be used in with the motion of translatio energies of the ions i ] nodes will be denoted by . . .
and their mean values taken ovel molecules by . . . .
If are the potential kinetic energies of which the snm is then the mean value of will be We begin by calculating the increase at any single collision in , \amp ; c. of these quantities will be a quadratic function of the velocities concerned , ( is the nlolecules .
We next assume the of distl.ibutio of velocities to be and average the values we have found oyer all collisio1Js , the procedure being reitdy f in the ) given velocities we an equation .
to equation , p. 400 .
It is to be articularly noticed that the translational elocities can only enter tljrough the terln We now continue in the malmer of S 6 .
The factor V occurs ying cvery intg.and , and giving rise to the term in the final result .
By this means we arrive at equations nilar to equations ( viii .
) and , p. 403 , giving the of of , . . .
cullisions .
By what has been already said , these Dlust be of the and nilar equations , together with ] . . . . . . .
( xviii .
) .
It is here assumed that a specified value of internal velocity is just as probable as the equal tive velocity , otherwise the mean value of products of different velocities could not be supl ) osed to vanish .
The equations determining the steady state are . . . . . .
( xix .
) .
MR. J. H. JEANS ON THE DISTRIBUTION OF These equations are not independent ; when they are added together the resulting equation vanishes identically in virtue of the fact that the mean total energy is , under all circumstances , unchanged by collisions .
Hence the equations can all be satisfied when the variables .
are in a definite ratio .
The distribution of indicated by this ratio will therefore be permanent , and since the equations which determinle it are linear , it will be unique .
This is the distribution discovered by , in hich the energy is equally divided between the various degrees of S11 .
Any of the coefficients in the ) system of equations vanish ; so that it will be possible the equations to into two groul ) , in such a that no variable occurs in both groups .
The motion will in this case be steady all the variables of the first group are in a give ratio , and all the variables of the second group are in a given ratio , but there need be no fixed between the two groups .
Thus the total energy of the first roup will be divided to } Law , and the same applies to the second roup , but the distribution the groups will not follow this law .
This is the essio of condition as to " " Continuity of Path * The Kinds of Internrd Co-ordinates .
S 12 .
Let us suppose , as before , certain velocities arc )ject to a retardation proportional to the velocities .
The mean energies from these degrees of freedom will be denoted by , . . .
, the reserved for those energies which are not by friction .
The system of equations ( xviii .
, p. 406 ) must be replaced . . . . .
( .
) .
If we suppose that at a collision only a small amount of can be between the modes and the remaining modes , then all the coefficients will be small .
It is immediately obvious that equations ( xx .
) may be treated exactly as equations , ' Camb .
Phil. Soc. vol. 12 , p. 548 ; or ' Collected lVorks , ' vol. 2 , p. 714 .
and ( xi .
) were treated .
At low temperatures the ratios all tend to zero , and the equations may be replaced by and . . . .
( xxi .
) .
A steady state would be possible , if we could simultaneously satisfy all equations of type , together with . . . . .
( xxii .
) , by values of , \amp ; c. , which were different from zero .
Let us use to denote summation with respect to degrees of freedom .
Then since we know the solution of equations ( xix .
, p. 406 ) we can , by substitution , arrive at a system of relations between the coefficients .
Making allowance for the alteration in notation equations these relations can be written in the form and If we attribute an alnount of energy to every degree of freedom , we have and Thus the tions ( xxii .
) are very approximately satisfied , in virtue of the smallness of the and coefficients .
The solution we have found will therefore give a state which is approximately steady .
S13 .
Thns at a sufficiently low temperature the energy of the gas which we have been considering will distribute itself in such a manner that an equal amount of energy will correspond to each degree of freedom which is not retarded by friction .
The amount of energy corresponding to a degree of freedom which is retarded by friction lvill be vanishingly small .
The amount of such energy is given by the equations , ( xxiii .
) .
If the degrees of freedom included in the 's are in number , and those counted in the are in number , it is obvious that the ratio of the specific heats must be taken to be ( xxiv .
) , MR. J. H. JEANS ON THE DISTRIBUTION OF MOLECULAR ENERGy .
whereas BoLTZlfANN'S theorem would lead to The equation ( xxiv .
) will only hold in the limit when the ture .
At other temperatures 7 will have a slightly different value , since the energies cannot be entirely neglected .
Our results as before only hold up to the temperature at which the gas begins to emit an appreciable amount of radiant energy , and this temperature may be supposed to be somewhat above the point of incandescence .
Up to this point , , . . .
will always be in the same proportion to one another , so that the brightness of the various lines in the spectrum will be in a constant ratio , each being proportional to S 14 .
We have been working on the assumption that there is a complete absence of frictional forces acting on , and on , . . . .
These assumptions , however , are not necessary .
In the steady state we have from equation ( xxiii .
) , where is a quantity which depends only on the construction of the molecule .
The temperatures which have been considered have been those for which \amp ; c. , are all very small .
But if for any single degree of freedom , say that for which the energy co-ordinate is , either is exceptionally snlall or exceptional]y great , the range of temperature will be greatly restricted on this account .
At temperatures at which is large while the remaining similar quantities are small , it is clear that must be treated as an co-ordinate .
At zero temperature all the energy co-ordinates to which friction corresponds must be regarded as co-ordinates .
As the temperature increases we must suppose these co-ordinates one by one to change from being co-ordinates , and after occupying a position intermediate between that of an and that of an co-ordinate to finally become co-ordinates .
If there is a co-ordinate for which is extremely small , or very great , that is to say , a co-ordinate corresponding to a degree of freedom which is only very slightly retarded by friction , or to one from which passes freely , then such a co-ordinate will an co-ordinate at such a low temperature that it be regarded as always being an co-ordinate .
* It may be noticed that the value of supplies a measure of the facility which is exchanged between the mode and the other modes .
If , it is impossible for such an exchange to take place , and the mode does not satisfy the condition of continuity of path .
Thus if friction dissipates the energy of the mode , the value of will finally be zero .
If , however , we have together with , the value of is indeterminate .
The rotation of the loitded sphere about the axis of symmetry supplied a good illustration of a mode of energy for which VOL , CXCVI .
Variation of with Temperature .
S15 .
Thus our conclusion is that it is not permissible to count degrees of freedom ; they must be weighted as well .
We can write , where is the " " \ldquo ; of the of freedom .
So .
as is small , we may put .
When this quantity is very great , either owing to the greatness of or the absence of , we put .
For intermediate values of , will be a proper fraction , the value of which depends not only on the temperature but also on series of changes through which the gas has passed .
As a consequence of this , it is clear that may be expected to vary with the temperature , and that it is no longer restricted to having one of the values given by the formula .
It would be going too far to expect any agreement with experiment at present , since we are considering a purely arbitrary type of molecule such as certainly does not exist in nature .
Case of Diatomic Jfolec S 16 .
As an illustration of the foregoing theory , we examine the case of molecule which is composed of two atoms held ether by an attractive force .
The atoms can execute internal vibrations iving rise to the emission of light ; for these vibrations we can suppose to be very small , so that .
The molecule can rotate about its axis of symmetry , but we can suppose to be zero , so that again .
For the rotation about the two remaining axes will be large , so that we may take .
We can suppose that the oscillationoe of the atoms as a whole about their position of equilibrium are dissipated by radiation so that This leads us to the value , a value which it is impossible to arrive at by means of BOLTZMANN'S theorem when we are considering a molecule which is made up of two separable parts , but which is nevertheless known to be the true value for many diatomic ases , such as hydrogen , nitrogen , and oxygen .
If the present theory in of this kind would give rise to definite lines in the spectrum of a frequency which might , and would , be very different from the frequencies of the light vibrations given out by the internal vibrations of the atoms .
In this connection it is of interest to remember that experiments with IIcrtzian vibrators have demonstrated the existence , in certain substances , of free periods of which the frequency is only about of the frequency of the sodium lines .
( P. DRUDE , ' Wied .
Ann , ' vol. 68 , , p. 17 ; vol. 64 , p. 131 .
) MR. J. H. JEANS ON THE DISTRIBUTION OF MOLECULAR ENERGy .
auy way corresponds to the facts , molecules of these gases must possess a symmetry similar to that possessed by ures of revolution .
PART II .
THE DISTRIBUTION ENERGY 1N A GAS 0F VHICH THE MOLECULES ARE 0F Statement of Problem to be Discussed .
S 17 .
Having discovered , by means of the ) illustration discussed in Part I. , what sort of results are to be ) ected , it ow becomes possible to nine the case in which the molecules form a more complex dynamical system , and as this may be done by an entirely method from that previously followed , it is now possible to emove the restrictions which it was found necessary to impose on the nature of the molecules .
The molecules , as , supposed to be all exactly similar , but intermoleculnr forces are no excluded , and the radiation is supposed to be of a type .
Let us suppose that each molecule is a dynamical system , possessing in itself degrees of freedom in addition to the freedom of the molecule to moyc in space .
There will therefore be 2 co-ordinates required to specify the condition of a molecule from its position in space , and 4 quantities are required to specify a collision .
The co-ordinates of position of molecule will be , the co-ordinates of its centre of gravity referred to axes fixed in space , , . . .
, the co-ordinates which do not occur in the expression for the potential energy ; as , for example , the co-ordinates which determine the orientation of a rigid body .
, . . .
, the co-ordinates which do occur in the expression for the potential The co-ordinates of velocity lvill be the time-difIerentials of . . .
, , , , , , . .
, . . .
, , , , , . . .
We shall write for and it will frequently be necessary , for the sake of brevity , to denote all co-ordinates of the same type by a single representative letter without a suffix .
Thus , \amp ; c. We shall assume that these co-ordinates are principal co-ordinates of the system , so that.both the kinetic and potential energies will be the sums of squares .
We may , in fact , write where . . . . . .
We shall suppose the oscillations of the -ordinates to be so small as to be isochronous , and in this case the and coefficients will be constants .
Since the co-ordinates are to be very small , the " " configuration\ldquo ; of a molecule may be supposed to be determined by its co-ordinates .
With a view to simplifying subsequent analysis , we shall assume that the 's also are constants .
It will be seen that the character of our results is not materially modified by this simplification , and the assumption is , of course , legitimate if we suppose the molecule , except as regards small oscillations , to behave like a rigid body , the atoms never far from certain equilibrium positions .
We shall suppose that the vibrations of the molecule result in a radiation of energy , and we assume a dissipation function G. This will be supposed to be a quadratic function of the -ordinates with constant coefficients ; it will not in general be reducible to the sum of squares .
The existence of implies an interaction between matter and ether .
The assumption that contains no terms in or is in ictness only legitimate if we suppose the and velocities to be uninfluenced by the ether , but it is easy to see , as in S 14 , that even if these velocities are acted upon by the ether , the neglect of these actions is of no importance so long as they are sufficiently small .
We have spoken of and V as kinetic and potential energy , but there is no reason why these energies should not be regarded as electro-magnetic and electrostatic energy , or indeed as energy of any other kind , provided only that it is always possible to deduce the equations of motion from the energy function by LAGRANGE'S method .
But it is probably best to regard the system just specified as simply a dynamical system in the strictest sense , this system being capable of illustrating all the properties experiment shows to be possessed by a molecule .
S 18 .
Corresponding to a collision in the case of two spheres , we shall suppose that it is possible for an action to take place between two molecules , and this action win be spoken of as an " " encounter For the present , it is not necessary to specify the exact nature of an encounter , but it will be supposed that the duration of an encounter is infinitesimal , so that an encounter causes no direct change in the co-ordinates of position of a molecule , MR. J. H. JEANS ON THE DISTRIBUTION OF MOLECULAR ENERGY .
41.3 ( ii .
) that an encounter is similar to an impact in the former problem , in that it may or may not entirely change the translational velocities of the two molecules concerned , but that the internal velocities are only changed by a small amount .
The sudden increase in any quantity consequent on an encounter will be denoted by In virtue of the above assumptions and are small , and will iu general be comparable with " " collision\ldquo ; will be a special case of an encounter , and may be described as follows .
Suppose that every molecule is surrounded by a small sphere , of the centre coincides with the centre of gravity of the molecule , which moves as though it were rigidly attached to the molecule .
The radius of the sphere is not yet fixed , but it must be such that the sphere entirely encloses the mattel oi which the molecule is composed .
Then a collision will be defined as an encounter which is such that the spheres of the two molecules which are engaged , intersect ; the " " duration\ldquo ; of a collision will be taken to be sufficiently long to include the whole interval from the instant at which the spheres first intersect to the instant at which they separate .
The assumptions as to the nature of the gas , which usually expressed by saying that the is geordnet , and that the number of collisionoe in which three or more molecules are engaged , is infinitely small in comparisoll with the number of binary collisions , will be replaced by the following assumptions : The duration of a collision is so short , that the co-ordinates may be treated as constant throughout the collision , while the velocity co-ordi- nates are abruptly changed .
( ii .
) The chance of any molecular sphere intersecting two other spheres at once , vanishes in comparison with the chance of its intersecting one other sphere .
( iii .
) The chance that a molecule A is found with all its -ordinates within certain small ranges of values , which are such that the sphere of the molecule does not intersect any other sphere , depends solely upon the co-ordinates of the molecule , and upon the potential upon A of the field of intermolecular force ; it does not depend upon the arrangement of the other molecules .
The Charaeteristic Equation .
S19 .
Starting from the state of the gas at the time , we can arrive at the state after an interval , by imagining the following succession of events .
Suppose in the first place that each molecule is to move under ' 414 MR. J. H. JEANS ON ] OF MOLECULAR ENERGY .
no force except its own internal forces for a time , and let all the collisions which would occur in this tinle be supposed to occur .
After this imagine the molecules divided into pairs in every possible way and suppose an encounter to occur between the two molecules of every such pair .
The duration of the encounter is to be ; during it the O -ordinates of each molecule are to change only on account of the forces of the encounter ; that is , on account of the intermolecular forces existing between the two molecules under consideratio1l .
These encounters are to take place consecutively , not simultaneously .
It is easily seen that each molecule has now been acted upon by exactly the same forces by which in the actual course of events it would have been acted upon in the interval .
Hence , since there is no limit to the smallness of , the final state of the is independent of the order in which this series of events takes place , and is identical with the state in which the gas would have been found if the forces had acted simultaneously .
Two points deserve attention in connection with this argument .
Firstly , it might be objected the changes in the co-ordinates of molecules which expel.ience an actual collision are not additive , inasmuch as one of these changes is not infinitesimal .
It is , however , clear that there is no necessity to take the infinitesimal changes into account at all in the case of these molecules , for the number of these molecules vanishes in comparison with the total number when is made to vanish .
Secondly , it is true the number of molecules within any specified limits will not always consist of the same individual molecules .
But it is a fundamental assumption of the kinetic theory that any molecules hich have nothing common except that certain co-ordinates specified values , will behave exactly like other similarly conditioned molecules .
We can therefore reduce the continuous changes of the co-ordinates of molecules which arise from the action of intermolecular forces , to a series of encounters of the kind described in S 18 .
S20 .
To completely specify an encounter , we require the values of all the nates enumerated in S17 , of both molecules .
It will , however , be convenient to write with a similar notation for \ldquo ; , and to specify a collision by the values of . . .
instead of the values of As regards the law of distribution of the various co-ordinates , we notice that whatever the values of the , and co-ordinates may be , the probability that the co-ordinates lie within the limits may , in the absence of external forces , be taken to be where .
Since the internal energy is only slightly changed by ncounters , we shall again suppose that the distribution of internal MR. J. H. JEANS ON THE DISTRIBUTION OF CULAR ENERGY .
energy is independent of the translational energy , and hence we may suppose that the number of molecules per unit volume which lie within a range dpdqdr.dsdu is or , as it will be frequently written , being written for dqdrds .
We shall suppose that the has adjusted itself so that the distribution of the energy of translation is the permanent distribution .
Thus at a point at which he potential energy of a molecule is we shall have As we are going to the existence of intermolecular forces , the potential of a molecule at a point will depend on the co-ordinates of the molecules as well as the position of the point .
Thus will in general be a function of and ?
: Let molecules of which the co-ordinates lie within limits be called molecules of class ; if the limits are , let the molecules be described as molecules of class .
Each of these classes will consist of a number of molecules which is indefinitely small in comparison with the total number of molecules present .
At any moment , let us imagine all the molecules placed in position , except those belonging to one or other of these two classes .
Let them produce a field of force such that if a molecule of class is placed with its centre at the point , then the potential of this molecule will be Then the probability that a molecule of class will be found with its centre within an element dxdydz at is Hence the total number of such molecules to be found in the whole unit volume may be obtained by integrating this expression over the whole volume , and be written as . . . . .
, where IJI . . . . . . . .
( ii .
) .
The integral is taken over the whole unit volume , since the integrand is supposed to vanish if the point is such that the centre of a molecule of class be found there .
The quantity will be called the mean intermolecular potential for a molecule of class .
It is clearly a function of all the coefficients which occur in the law of istribution as well as of the co-ordinates of molecules of class .
If we remember that the co-ordinates are supposed to be very small , it is clear that it will be sufficient to imagine that is a function of , on account of the way in which this coefficient enters in equation ( ii .
) , ( ii .
) the coefficients occurring in ( these have , however , already been assumed to be invariable ) , ( iii .
) the co-ordinates of molecules of class Thus , for our purpose , is a function of and only .
Fixing our attention on any one of these molecules of class , the probability that the centre of a molecule of class may be found within the limits measured relatively to the first molecule will be where is the potential of a molecule of class at this point .
Now can be made up of two parts , the part due to the presence of the single molecule of the first class , and the part due to all the other molecules combined .
It is clear that will only depend on the two molecules of the encounter , and is therefore a function of and The total number of encounters of the type we are now considering , namely those within limits . . . .
( iii .
) , will be where the summation extends to all the molecules of class This number may be written as ' Now it is obvious that the mean value of , taken over all the elements of volume included in the summation , will be , where is the mean intermolecular potential of a molecule of class , and is therefore a function of and only .
Thus , since the summation extends to elements of volume , This gives us for the total number of encounters of the type we are considering , or , if we write .
( iv .
) , the number is MR. J. ON THE DISTRIBUTION OF MOLECULAR ENhRGY .
Let us call encounters of this type , encounters of class , and denote their number ( expression ) There will be a second class of encounters which will be called class , such that the co-ordinates after the encounter lie within the limits ( iii .
) , ( p. 416 ) .
The co-ordinates before the encounter will accordingly lie within certain other limits , . . . . .
, surrounding certain values 1 .
\amp ; c. By Theorem , ) complete difTerential is equal to the complete differential ( iii .
) , hence the number of encounters of class will , by comparison with , ) seen to be the positional co-ordinates and therefore also , remaining unaltered by the encounter .
Let be the total number of molecules lying within a range .
Then a certain number of encounters of class A will result a unit loss to , a cel.tain number of encounters of class in a unit .
Thus , if for one at least of the -ordinates which are changed by the encounter , say , we have , where denotes an increase due to an encounter , then it is certain that the co-ordinate will be placed outside the limit , and will accordingly be diminished by inity .
It is , howeyer , conceivable that for every co-ordinate we may have , and in this case there is a probability that no single co-ordinate passes outside , limits , and therefore that the molecule after encounter must still be counted in .
It is easy to see that the probability that the co-ordinate remains the limits is 1 - , and therefore that , where denotes continued multiplication ling to all the dinates .
The loss experienced on account of encounters of class A will therefore ) where for certain values of and is a proper fraction oyer the remainder , and the boundary of these regions depends on the co-ordinates of encounter .
It follows that as the result of encounters of classes A and ined , there is a net to of The expression in square brackets may be written as ] the total gain to arising from all classes of encounters will be . . .
VOL. CXCVI .
\mdash ; A. Here , as throughout the paper , a single tall integral sign denotes integration over all values of the variables , of which the differentials occur after the sign of integration .
For those values of for which is different from zero , we have seen that is , for every co-ordinate of the same order of small quantities as , and hence it follows that is of the same order of small quantities .
In the limit , when the differentials such as are supposed to vanish , this integral will vanish also , so that we may put in expression ( vii .
) .
The total gain to from all encounters is therefore , where . . .
( viii .
) .
S21 .
Now we have seen ( S 19 ) that the in in time will arise from three canses\mdash ; ( a ) the from collisions , the dlange arising encounters other than collisions , the change arising from the change in the co-ordinates of the various molecules during the time , which would occur if there was no interaction between different molecules .
Now any molecule , , will collide with a second molecule , , in the interval of time , provided that at the beginning of this time the centre of A lies within a certain region of space ; provided , that is , that lie within certain limits .
It is easily seen that these limits are arrived at by writing , and allowing for of values corresponding to points inside a certain circle of diameter equal that of a molecular sphere .
Hence the increase in arising lrom cause will be where ( .
) , and denotes the increase due to a collision .
The increase arising from encounters other than collisions has already been found to be ( see viii where must denote the increase due to an encounter of which the duration is .
Since is to be very .
small , we may replace by where denotes the rate of increase under the influence of termolecular forces only .
The increase to arising from cause will therefore be , where ' . .
J. H. JEANS ON THE OF hNERGY .
S .
In addition to the gain or loss which has just been calculated , will ence a further gain or loss on account of the changes in the co-ordinates ol the various molecules .
Let be a typical co-ordinate , and the limits of the values of for molecules counted in .
Aftel time will to , so that some molecules will enter within the above limits , and others will out , in course of the time The molecules which enter within the limits will be those for which , at the ) ning of the interval , had a value which between and .
The number of such molecules is accordingly Similarly the number of molecules which escape from between these limits is so that the resultant to on account of changes in the co-ordinates , the other co-ordinates being supposed to remain constant , will be There is no limit to the of if all the -ordinates vary simultaneously , the gain to will be or in which the summation extends to all the co-ordinates , and The value of must be found from the equations of , motion of the molecule when under the action of no external forces .
If be any co-ordinate of ) osition , and be written for the total , then these equations will be of the This leads to the following scheme of values for the various time-differentials Now it appears from the scheme of values just found that terms of the form only arise connection with that may be replaced where . . .
to from the cause will be .
We have counted ) nll sources of change in ; we therefore have , as the tion to be .
S .
We have found the equation , the summation is such as to cover all ranges of We herefore rite HF where is .
by ( xiii .
) , and is therefore function of only .
Substituting this value of in .
tion ( xii .
) , we have the equation ( HF ) ( xiv ) Let rite 1 and substitute this value for in the integrals and K. We to equation ( iv .
) we find that the tion of energy ( see p. 412 ) , H. ON THE DISTRIBUTION OF Hence where In virtue of the ssunlptions which have beenl made , i , s small quantity , so that we put and therefore The same transformation holds if is by .
Hence we have ( see equations ( ix .
) and , p. 418 ) .
where , .
Making these substitutions , equation becomes \mdash ; or dividing throughout by HF , ( xvii .
) .
This is the characteristic equation satisfied by of Solution .
S24 .
We must first examine in what way the integrals and involve , and .
In these -ordinates are only involved through the ctor which o the integrand .
Now , from the equations of impact , be expressed linear functions of all the velocities concerned .
The coefficients will be functions of 1\ldquo ; but we may as usual put It follows that is a function of , and of degree equa ] to that of , and involve as well as these variables .
If , then , is of degree in we may regard as a function of , and of degree , of which the coefficients are functions of the variables with respect to which integration is performed in uating L. Hence fter integration shall have fuuctiun of , and free n MR. J. H. JEANS ON THE STIiIBUTION OF MOLECULAR ENERGY .
In the same way , the variables , and only enter the integral , through the term , and we may write Now denotes the rate of increase in owing to the action of forces of which the potential energy is Write , then in which is any one of the or co-ordinates , and is the corresponding velocity co-ordinate .
Hence ; ; The right-hand members of these equatio1ls will be functions of the coordinates , but it is clearly legitimate to put all equal to zero , and regard the expressions as functions of and only .
It therefore appears that will be a function of , and , of which the degree is , and upon integration , that is a function of , and of the same degree .
The terms and 2 which occur in equation ( xvii .
) will be fimctions of and s of degree ?
S 25 .
It is therefore clear that the correct form to assume for is a rational integral function of the co-ordinates , and If we assume to be the most general function of degree in these co-ordinates , the coefficients being functions of the time , and if we substitute this assumed value in equation ( xvii .
) , we shall get , on each side of the equation , a function of , and of degree If therefore we equate the coefficients 01 every term on the two sides of the equation , we shall have found solution of equation ( xvii .
) , p. 421 , inasmuch as equation is now satisfied identically every value of , and The process of equating these coefficients leads to a series of diffel.ential equations , in which the time-rate of increase of every coefficient is given explicitly in terms of the othel coefTicients and of .
If , therefore , we suppose the coefficients to vary with the time in the manner giveu by these equations , the value of so obtained will be a solution of equation ( xvii .
) all time .
the equations involve , and varies with the time , one ful .
there equation is required before we can express the co-ordinates at any time in terms of the initial values of the co-ordinates and the time .
This additional equation is supplied by the fact that , the total number of molecules , remaius constant . .
J. H. JEANS ON THE DISTRIBUTION OF MOLECULAR ENERGY .
We have ( see S 23 , equation ( xiii .
) ) , .
( xviii .
) , and the equation is the equation required .
With the help of the other equations , it can be written so as to giye explicitly as a function of and the other co-ordinates .
Thus if we have the initial form of given , we have obtained sufficient equations to enable us theoretically to determine at any subsequent time .
It is not proposed to attempt the solution of the system of equations in the nlost general case ; the discussion is confined to the modified forms which these assume in the two states of which the physical interest is greatest , na mely the steady state , and the state in which the gas is non-luminous .
in State . .
The mathematical condition that a steady state my possible , is that it may be possible for the rates of variations of the coefficients to simultaneously .
From the equations found by equating to zero the time rates of all the coefficients except , it is possible to find these coefficients in terlns of so that the condition for a steady state to be possible is that the function of obtained ) substituting these values in the expression for shall vanish identically for all values of .
It is , however , known that the condition that a steady state shall be possible is that shall be identically zero , and we may therefore begin by putting and neglecting the equation Thus all the equations necessary are contained in the characteristic equation satisfied by , and this is now ( cf. equation xvii .
, p. 421 ) or substituting from the scheme of p. 419 , .
Let us assume as a possible value for the most general expression of in and , the coefficients now being independent of the time .
Consider the system of equations which is obtained when we equate to zero the coefficients of terms of degree in equation ( xix .
) .
The terms of degree in the integral arise entirely ) the ternoe of degree ( see S24 ) .
These terms will each be mu]tiplied by linear functions of the coefficients of terms of degree in , and the whole expression will be multiplied by where Now is the number of collisions per unit time which are experienced by the molecules of which the , co-ordinates lie within limits .
Since the collision not depend on the co-ordinates , it follows that the total number of collisions per unit time is .
It follows that is a function of only , does not depend on the coefficients which occur in In second integral , terms of degree do not occur at all .
In elnaining terms of ( xix .
) terms occur of degree in , the coefficients being of the same form as those occurring in except that the factor does not occur .
We have thus found as many equations as there are coefficients ; in these equations every term is a coefficient multiplied by a constant .
The only solution of this system of equations is that every coefficient vanishes .
result depends 011 the assumption that is greater than two .
Hence in the state will contain no terms of a degree higher than the second in and At the same time can contain no terms which are in or .
The introduction of these terms would give a law of distribution such that an infinite number of molecules would have an infinite value , and We may , therefore , suppose that , except for an additive constant , is a quadratic function of .
and , in which square terms occur .
The equations between the coefficients of this quadratic expression must be linear , since they are the coefficients of the terms of highest degree in , and hence must lead to unique value for these coefficien ts .
Thus for the type of molecule which we have been considering , there is only one steady state possible on the assumptions we have made .
It is easily verified that . . . . . . . .
( xx .
) is a tion of equation ( xix .
) .
For , with this Yalue and both vanish , so thnt and nish , we also Hence term of eqnation ( xix .
) vanishes separately , and the value of given by equation ( xx .
) a tion which , as we have seen , nrust be unique .
This is the solution of BOLTZMANN .
* The ility of ) eing an iltfinite series been disregarded in the ) sections .
may , however , considcr that series which for finito values of the would lead to an of distribntion , hilst $t series which is convergent for all finite be treated as the limit of a finitc serics in the nnmI ) of is made intinite .
MR. J. H. JEANS ON THE DISTRIBUTION OF MOLECULAR ENERGY Solution for Low S27 .
In the approximately steady state which was found to be possible at low temperatures , for the previous system of molecules , it was found that the.various mean energies varied very slowly with the time .
But if such a state were possible for the present system , it would not follow that would only vary slowly with the time , so that an approximate solution of , even if it could be found , would be useless .
Let us , however , examine under what circumstances we could have equal to zero , without approximation .
From the remarks at the beginning of the last section , it is clear that the coefficients in can be so chosen as to make equal to zero , but that it is only in the event of being absent , that these values will also make O. Let us suppose that by some external agency is caused to increase uniformlv throughout the whole gas , at a rate exactly equal to the rate at which it decreases in consequence of the value of , found in the manner described at the beginning of S26 , being different from zero .
Then a completely steady state will have become possible , and this is because the imaginary agency introduces exactly sufficient energy to compensate that lost by radiation .
In a state such as that which , in the first part , was described as approximately steady , the radiation was very small .
If a similar state can be shown to be possible in the present case , the radiation will be very small , and therefore the energy introduced from outside will be very small .
Hence it will be legitimate to describe the state which would be arrived at by checking the external flow of energy as approximately steady .
The equation which leads to such a state is , or ( xxi .
) .
The last term in equation ( xxi .
) is of the same degree as , so that just as in the last section it may be shown that cannot contain terms of a degree higher than the second .
And for the same reason as before , cannot contain terms linear in any of the coefficients , so that we may assume Now at low temperatures all the terms in equation ( xxi .
) are small , except terms of the form These terms must therefore vanish approximately for all values of the variable , and this requires the relation VOL. CXCVI .
\mdash ; A. 3 I MR. J. H. JEANS ON THE DISTRIBUTION OF MOLECULAR ENERGy .
This merely shows that , neglecting small quantities of the order of the coefficients in , the energy of any , of the modes of vibration must be equally divided between kinetic and potential energy .
It is now clear , that at high temperatures the last term in equation ( xxi .
) is smah in comparison with the others , so that the law of distribution will be very little altered by the presence of a dissipation function , whereas low temperatures , the term arising from the dissipation function becomes as important as any other term in the equation , and , therefore , the presence of a dissipation function , however small , will be sufficient to entirely alter the law of distribution .
And without investigating the solution of the system of equations which determine the coefficients in , it is clear that since they are all linear , every coefficient must be a single-valued function of only .
Hence , as before , there is only one approximately steady state for a given temperature , but it is no longer true that the various lines of the spectrum increase in brightness in the same ratio when the temperature is increased .
Since the modes of energy suffer very little loss of energy , and since energy passes freely between these modes , it follows that the energies of these modes will very approximately be distributed according to BOLTZbIANN'S Law .
Hence all that was said about the ratio of the two specific heats with reference to the former type of molecule considered , will apply also to molecules of the type which we have just been discussing .
PART III .
PHYSICAL C0NSEQUENCES OF THE FOREGOING THEORy .
S28 .
We have thus been led to the same results in both parts of this paper .
It seems natural to suppose that results which are qualitatively the same will be found to be valid for any gas , and , assuming this to be the case , to examine some of the consequences of these results .
Radiation a given Temperature .
S29 .
In the first dynamical illustration , the radiation from each degree of freedom at temperature was found to be proportional to .
In the second illustration , the expression for the radiation would have been too complicated for the calculation of it to have been profitable .
A modification of the system discussed in Part II .
, leads to an interesting expression for the radiation .
In this modified system the molecules are to be sphel.es with modes of internal vibration to and from which energy only passes with difficulty .
Each sphere is surrounded by a field of force , such that when two molecules MR J. H. JEANS ON THE DISTRIBUTION OF MOLECULAR ENERGY .
are in contact in any position , their potential has a constant value .
When two molecules are not very near , let us suppose that the transfer of energy between different modes may be neglected , so that the main transfer of energy takes place through collisions.$ At any collision , the energy of the colJiding molecules will be small , so that the amount by which the energy is increased will depend only on the and energres of the collidin molecules ; that is to say , on the average , on Thus if collisions take place per unit time , the total transfer from the modes of energy to any specified mode may be taken to be , where is a constant for any particular mode .
The sum of such amounts must be equal to the amount radiated per unit time , and may therefore be supposed to be roughly proportional to the energies of the modes .
lt follows that for a single molecule the energy is proportional to Now we may take for the number of collisions where A is a constant , Hence and if be the density , the radiation will be proportional to , or , in terms of the temperature , to This example is of such a special nature that not much importance can be attached to the actual result obtained .
It is , however , of interest , as showing that it is at any rate possible for the radiation to increase very rapidly with the temperature .
comparison with the result of S 9 , shows that the introduction of a field of force has introduced a factor into the expression for the radiation , and a factor of this form figures in every formula for radiation .
The presence of the factor multiplying the expression for the radiation , is an essential feature of the present theory .
If the exponential factor changes very rapidly with the temperature , so that the point of incandescence is sharply defined , then this point will olearly be almost independent of , and variations in radiation on This assumption , although not stated explicitly in the investigation of Part II .
, is implied in the assumptions made there .
WIEN and PLANCK give for the radiation in the part of the spectrum between and , ths formula .
LORD RAYLEIGH , on theoretical grounds , suggests as an emendation ( ' Phil. Mag June , 1900 .
) [ Added March 19 , 1901.\mdash ; I was not aware , when writing this , that the presence of the factor had been detected experimentally .
See LIVEING and DEWAR , ' Roy .
Soc. Proc 49 , p. 217 , or KAYSER , ' Handbuch der Spectroscopie ' ( 1900 ) , I. , p. 143 .
] account of density will be unnoticeable in comparison with variations on account of temperature .
It is , however , clear that in considering tions from gases of great density , such as for example occur in the sun , the factor would be of considerable importance .
of Temperature .
S30 .
The state of a gas may be regarded as depending upon a principal temperature , and also on a Ilumber of subsidiary temperatures , .
, each of these temperatures corresponding to one ( or possibly more ) of the degrees of freedom of the molecule .
The principal temperature is to correspond to the three degrees of freedom implied by the possibility of translation through the ether , and to any other degrees of freedom which are such that their mean energy is at all temperatures equal to a third of the mean energy of translation .
The principal temperature is to be proportional to the mean energy of translation of a molecule and each subsidiary temperature proportional to the mean energy of each of the modes to which it corresponds .
Thus two modes can only have the same subsidiary temperature when their mean energies are , under all circumstances , equal , as , for example , when they are the kinetic and potential energies of the same vibration .
When the energy is equally distributed between all the degrees of freedom all these temperatures are to become equal .
We have found that at temperatures below the temperature of incandescence there is an approximately steady state in which , \amp ; c. , where , \amp ; c. , are functions of , which at these temperatures are very small in comparison with T. At higher temperatures we have not investigated the forms of , .
, but at infinite temperatures , . . .
T. S31 .
The steady state specified above was arrived at on the assumption that external agencies could only influe1tce the energy of translation , and that the other energies were only influenced indirectly through changes in the energy of translation .
Thus the above equations will not hold in the presence of agencies which exert a direct on the subsidiary temperatures .
Such influences may be looked for in the forces of chemical action , disturbances in the ether , and ) ossibly in the cathode rays , if we suppose these rays to be streams of charged ions which are so small as to penetrate inside a molecule rather than act on the molecule as a whole .
When such agencies are present , the above equations must give place to others .
The subsidiary temperatures which are most directly concerned may attain to MR. J. H. JEANS ON THE DISTRIBUTION OF MOLECULAR ENERGY .
abnormally high values , and this may result in the phenomena of phosphorescence , chemi-luminescence , \amp ; c. To take a definite instance , suppose that corresponds to a vibration in the molecule of frequency .
If a ray of light passes through the substance , those components of this wave of which the frequency is nearly equal to will ) energy to the mode of the molecules , and this will be distributed from ) mode to the other modes , and so through the substance .
Thus the result is a heating of the substance , and an absorption band in the spectrum of the light transmitted through it .
The illustration might be varied by supposing that energy could not easily distribute itself from to all the other temperatures , but that it passed freely to a second temperature .
In this case the temperature might conceivably attain to such a high value as to emit its own spectrum , and so set up fluorescence or calorescence .
The spectrum of the gas in any condition whatever will be arrived at by the superposition of the various spectra of the subsidiary temperatures , and the state of the gas as regards the emission of radiation will be completely specified by the values of the various subsidiary temperatures .
Thermodynamics .
S 32 .
At temperatures at which the gas is dark , we may take . . .
Thus at these temperatures we are only concerned with the principal temperature , and the total energy of the gas is proportional to this temperature .
If degrees of freedom correspond to this temperature , the ratio of the specific heats will be both specific heats being constant as regards the temperature .
The view which we have put forward does not clash with the ordinary thermodynamics as regards dark gases .
When the subsidiary temperatures begin to have appreciable values the case is different .
The total internal energy is now given by where is a constant , and is the number of modes of energy of which the subsidiary temperature is .
The specific heat at constant volume is given by and therefore depends on both the temperature and density .
If a quantity of work be absorbed by a gas , If the gas be made to pass through any succession of states so as to return to the same temperature and density , Thus the second law of thermodynamics will only hold in special cases for a gas which is emitting radiation of any kind .
This part of the subject of luminosity has , however , been develnped by WIEDEMANN , that it seems unnecessary to pursue it any further here .
* E. , " " Mechanics of Luminosity ' Phil. Mag vol. 28 , 1889 , p. 162 , or ' Wied .
Annalen ol .
37 , p. 181 .

