102 



DIFFUSION OF GASES 



275 



tabulated together with those observed, the centimetre and 

 second being taken as units : 



Coefficients of Diffusion. 



The agreement of the theory with experiment is of 

 course not entirely perfect, but is sufficiently good for us to 

 see in it a further important argument for the correctness 

 of the hypotheses of the kinetic theory. For it remains, 

 indeed, a striking and remarkable circumstance that the 

 coefficient of diffusion may be calculated from that of 

 viscosity with so near an approximation. 



The differences that yet remain between the calculated 

 and observed numbers may be explained on different grounds. 

 It seems to me most probable that Stefan's hypothesis of 



the simple relation , , x 



' = i(*i + * 2 ) 



between the radii of the different spheres of action does 

 not hold with accuracy. For the forces between similar 

 molecules on the one hand and dissimilar molecules on the 

 other may have very different intensities, and there perhaps 

 exists a different relation between the shortest distances to 

 which two molecules can approach each other. 



A second ground for the want of agreement undoubtedly 

 resides in the accumulation of the errors of the manifold 

 data of experiment employed one after another, which are 

 all combined in the result and may give rise to great 

 uncertainty. The values of the molecular speeds have been 

 calculated from the observed densities, the free paths from 

 the coefficients of viscosity. Lastly, we have also used the 

 molecular weights and, therefore, stoichiometrical measures. 

 Consequently, at least six different observations, of which 



T 2 



