264 PHENOMENA DEPENDENT ON MOLECULAR PATHS 99 



only on the nature of the two diffusing gases, and on the 

 temperature and pressure at which they are ; its value also 

 increases as the temperature rises, and it varies inversely as 

 the pressure, just as in the theory first developed and in the 

 observations that Loschmidt has made. 



But this agreement with experiment does not prove the 

 accuracy of the simplified theory. It is still open to the 

 objection that the resistance offered by the other kind of gas 

 is not the only one experienced by a molecule ; molecules of 

 the same kind also obstruct a particle in its forward motion, 

 since they, too, can get in its way just like those of the 

 other kind. This resistance of the same medium, and the 

 consequently ensuing shortening of the molecular free path, 

 cannot be left out of account. 



The assumption that the current of diffusion of a gas 

 should be resisted only by molecules of the other kind which 

 are moving in the opposite direction rests on a fallacy only. 

 For the velocity of the current is much too little for there 

 to be found in it a sufficient cause why the particles should 

 collide so much oftener or more forcibly with the particles 

 of different kind which meet them than with particles of 

 their own kind, that the influence of the latter should appear 

 vanishingly small. 



10O. Observations in Test of the Theories 



Led by these considerations, I spoke out, in the first 

 edition of this book, which appeared in 1877, for the view 

 that the coefficient of diffusion is not constant, but is 

 variable with the ratio of mixture of the gases. But I 

 could not prove the correctness of this assumption, because 

 no decisive experiments had then been made. I could 

 adduce in favour of my conjecture only one observation 

 of Graham's, 1 to which Maxwell 2 had drawn attention; 

 for Maxwell's calculation gave smaller values of the co- 

 efficient when the experiment lasted long, and larger values 

 when the period of diffusion was shorter. 



1 Brandos Journal, 1829, pt. 2, p. 74. 



2 Phil Mag. 1868 [4] xxxv. p. 202 ; Scientific Papers, ii. p. 61. 



