DIFFUSION AND ENTROPY OF GASES 609 



coming at haphazard, independently of the existing state of 

 the system, effects a new distribution pro tanto. 



26. I know it is said in defence of the theorem that the 

 reversed motion, though possible, is highly improbable. But 

 if the H theorem be true, Maxwell's law of distribution of the 

 velocities must be assumed to prevail ultimately, or anyway 

 cannot be denied by advocates of the theorem. From which 

 it follows that any course and the reversed course are equally 

 probable. To say that one of the two is the more probable 

 because it leads to H diminishing would be a petitio principii. 



27. If, as I think is the case, the complete protection of two 

 diffusing gases from all external influences continuously for a 

 very long time is impossible, the motion is, as I have said, for 

 that reason irreversible, because the external disturbances are 

 equivalent to a new haphazard distribution. But for that same 

 reason the irreversibility does not necessarily involve increase 

 of entropy, and the whole argument falls to the ground. 



28. I have asserted generally that if the molecules of gases 

 are what they are assumed to be in the kinetic theory, no 

 irreversible motion is possible in an isolated system. Very 

 probably it may be necessary to revise our conceptions of a 

 molecule, and Prof. Bryan's work may well be conducive to 

 that end. 



