systems for which the initial values of pl, gl, . . . pn, gn lie within a range such that

pl is between pl and pl -l-dpl

qi, ., l 91 .. qi-l-dai, and S0 On-Let

the product dpl dql . . . dp, dgn be spoken of as the “extension ” df this range of values.

After a time dt the value of pl will have increased to pl-I-pldt, where pl is given by equations (I), and there will be similar chan es in gl, p2, gl, . . . g... Thus after a time dt ordinates and momenta of the small group

side ration will lie within a range such that pl is between pl-l-pldt and pl-l-dpl+

g

the values of the coof

systems under conlSl+%apl

at

Qi+%dQ1 dt,

and so on. Thus the extension of the range after the interval dt is gl H n gl u

dpi I'l" dl> dgl <1 -i-%dt) . . .

or, expanding as far as first powers of dt, QQ QQ)

dpldql . . . dpndqyl I-j-E1 api-l-aql (ll . From equations (1), we find that

an aéll

af); +6Q1 T O

so that the extension of the new range is seen to be dpldql . dp, .dg, ,, and therefore equal to the initial extension. Since the values of the co-ordinates and momenta at any instant during the motion may be treated as “ initial ” values, it is clear that the “ extension " of the range must remain constant throughout the whole motion. This result at once disposes of the possibility of all the systems acquiring any common characteristic in the course of their motion through a tendency for their co-ordinates or momenta to concentrate about any particular set, or series of sets, of values. But the result goes further than this. Let us imagine that the systems had the initial values of their co-ordinates and momenta so arranged that the number of systems for which the co-ordinates and momenta were within a given range was proportional simply to the extension of the range. Then the result proves that the values of the coordinates and momenta remain distributed in this way throughout the whole motion of the systems. Thus, if there is any characteristic which is common to all the systems after the motion has been in progress for any interval of time, this same characteristic must equally have been common to all the systems initially. It must, in fact, be a characteristic of all possible states of the systems. It is accordingly clear that there can be no property common to all systems, but it can be shown that when the system contains a gas (or any other aggregation of similar molecules) as part of it there are properties whim are common to all possible states, except for a number which form an insignificant fraction of the whole. These properties are found to account for the physical properties of ases.

g Let the whole energy E of the system be supposed equal to El-l-El, where Eg is of the form

El=%E(mu'-l-mv*-l-mwz-I-a1012-I-H202 -l' » - - +¢1»9, l) -|f%2(m'u”+m'1f'2+rn'w'2+6'l4>l”+B2¢>2”+ . +/91./<1>»» ) (2) sr

2 2

where 0l,0l, . . H.. and similarly qSl, ¢>l, . 4>, ,, are any momenta or functions of the co-ordinates and momenta or co-ordinates alone which are subject only to the condition that they do not enter into the coefficients al, al, &c.

In this expression the first line may be supposed to represent the energy (or part of the energy) of s similar molecules of a kind which we shall call the first kind, the terms %(mu”-l-mv” -l-mwz) being the kinetic energy of translation, and the remaining terms arising from energy of rotation or of internal motion, or from the energy, kinetic and potential, of small vibrations. The second line in E2 will represent the energy (or part of the energy) of s' similar molecules of the second kind, and so on. It is not at present necessary to suppose that the molecules are those of substances in the gaseous state. Considering only those states of the system which have a given Value of El, it can be proved, as a theorem in pure mathematics that when s, s', . are very large, then, for all states except an infinitesimal fraction of the whole number, the values of u, 1/, w lie within ranges such that

(i) the values of u (and similarly of 11, 'w) are distributed among the s molecules of the first kind according to the law of trial and error; and similarly of course for the molecules of other kinds: (ii) E§ mu2 Elmo* Etmwi 25111012

3 -i .. i i -...

s s s s

2;im'1/2 § il34>i°

5

=L..? .=...=fS7-=... (3)

See jea ris;-'Dynamical Theory of Gases (1904), ch. v. A state of the system in which these two properties are true will be called a “ normal state "; other states will be spoken of as “ abnormal.” Let all possible states of the system be Th N divided into small ranges of equal extension, and of Sta “rm” these let a number P correspond to normal, and a number 8 e p to abnormal, states. What is proved is that, as s, s', . . . become very great, the ratio P/p becomes infinite. Considering only systems starting in the p abnormal ranges, it is clear, from the fact that the extensions of the ranges do not change with the motion, that after a sufficient time most of these systems must have passed into the P normal ranges. Speaking loosely, we may say that there is a probability P/(P-l-p), amounting to certainty in the limit, that one of these systems, selected at random, will be in the normal state after a sufficient time has' elapsed. Again, considering the systems which start from the P normal t'1t!g'f:s, We see that there is a probability p/ (P +p) which vanishes in the limit, that a system selected at random from these will be in an abnormal state after a sufficient time. Thus, subject to a probability of error which is infinitesimal in the limit, we may state as general laws that- system starting from an abnormal state tends to assume the normal state; while ~,

A system starting from the normal state will remain in the normal state.

It will now be found that the various properties of gases follow from the supposition that the gas is in the normal state. If each of the fractions (3) is put equal to I/4h, it is readily found, from the first property of the normal state, that, of the Lawofpis, s molecules of the first kind, a number tribution of 5/ (Wm3/7|-3)g'hm(“2+vZl'IU2)dud»lldw (4) Velocities. have velocities of which the components lie between u and u-|-du, 2/ and 'u-l-dv, w and 'w-l-dw, while the corresponding number of molecules of the second kind is, similarly,

s'/ (h”m'3/1rf')e"""'<“”+”2*'”2>dudvdw. (5) If o is the resultant velocity of a molecule, so that c2=u2-I-'vi-I-w2, it is readily found from formula (4) that the number of molecules oil- 'Elie first kind of which the resultant velocity lies between c and c c IS

47l'$V (hims/1r3)e“'"””2c2dc. (6)

These formulae express the “ law of distribution of velocities " in I;the normal state: the law is often called Maxwell's Law of Distri ution.

If § mu2 denote the mean value of émuz averaged over the S molecules of the first kind, equations (3) may be written in the form im“2=i"'W2 = im”/2=i;i'lEf= - - ~ = I/471, (7) Equlpartlshowing that the mean energy represented by each H011 of term in El (formula 2) is the same. These equations EH¢1'.&Veécpressl tht; “ lagéifl equipartition of energy, ” commonly spoken o as the axwe - atzmann Law.

The law of equipartition shows that the various mean energies of <£ffe1§§ /nt lkinds aire aél equal, elgch bleing measured by the quantity I 4 e ave a rea y seen t at the mean ener increases with the temperature: it will now be supirilosed T"”P"" that the mean energy is exactly proportional to the mn temperature. The complete justification for this supposition will appear later: a partial justification is obtained as soon as it is seen how many physical laws can be explained by it. We accordingly put I/2h=RT, where T denotes the temperature on the absolute scale, and then have equations (7) in the form mu2=mv2= . . . =RT. (8)

When a system is composed of a mixture of different kinds of molecules, the fact that h is the same for each constituent [cf. formulae. (5) and (6)] shows that in the normal state the different substances are all at the same temperature. For instance, if the system is composed of a gas and a solid boundary, some of the terms in expression (2) may be supposed to represent the kinetic energy of the molecules of the boundary, so that equations (7) show that in the normal state the gas has the same temperature as the boundary. The process of equalization of temperature is now seen to be a special form of the process of motion towards the normal state: the general laws which have been stated above in connexion with the normal state are seen to include as special cases the following laws:- ]l}Iatter originally athnlon-uniform temperature tends to assume a uni orm temperature; w i e

Matter at uniform temperature will remain at uniform temperature. It will at once be apparent that the kinetic theory of matter enables us to place the second law of thermodynamics upon a purely dynamical basis. So far it has not been necessary to suppose the matter to be in the gaseous state. We now pass to the consideration of laws and properties which are peculiar to the gaseous state. A simple approximate calculation of the pressure exerted by a gas on its containing vessel can be made by supposing that the molecules are so small in comparison with their distances apart that they may be treated as of infinitesimal size. pressure °f Let a mixture of gases contain per unit volume if mole- “Gas” cules of the first kind, v' of the second kind, and so on. Let

us fix our attention on a small area dS of the boundary of the