﻿Attraction of Unlike Molecules. 5 



stream is proportional to 



^dn 1 n 2 v~iki-{-n l V2\2 

 dx n i -\-n 2 ' 



and the diffusion-coefficient to 



The defect of Meyer's theory is that it takes no account of 

 the actually existing diffusion velocities in the two sets. 

 Supposing the process of diffusion to be arrested suddenly by 

 some cause which then suddenly ceases to act, Meyer's 

 method shows how the diffusion-streams would begin to flow 

 again, but it cannot follow the process after that, because it 

 takes no cognizance of the bodily motions existing in the two 

 sets. 



Tait's theory takes account of both Stefan's and Meyer's 

 causes ; he supposes the molecules of each medium besides 

 their velocities of agitation to have velocities of translation 

 en masse ol x and a 2 , an( l then calculates the quantities of each 

 flowing in unit time across unit section, these quantities 

 depending on oc l and a 2 and on expressions similar to Meyer's. 

 The velocities a x and a 2 are determined exactly as in Stefan's 

 method, so that Tait's method labours under this difficulty, 

 that he supposes each molecule of each set to have a certain 

 velocity combined with the velocity of agitation, and yet 

 this velocity is different from that of the set as a whole. 



On theoretical grounds, therefore, Stefan's theory appears 

 not to have been improved by the later attempts, and, further, 

 it seems to me not to have been sufficiently recognized that 

 Stefan has given satisfactory experimental proof of the sound- 

 ness of his method of treating the diffusion problem ; for 

 exactly on the lines of his theory of the diffusion of two gases 

 into one another he constructed a theory of the evaporation of 

 a liquid into a gas, which led to a striking formula for the 

 velocity of evaporation of a liquid, a formula verified by his 

 own and Winkelmann's experiments. As Stefan's elegant 

 theory of evaporation will only take a few lines to reproduce 

 here, and as it gives a valuable method of determining dif- 

 fusion-coefficients; it may as well be reproduced in the present 

 connexion. 



Suppose a tube half filled with a liquid whose properties 

 are to be denoted by suffix 1 evaporating into an atmosphere 

 with suffix 2, but with fresh liquid added from below so as 

 always to keep the free surface of the liquid at a fixed mark 

 on the tube; then, when a stationary state is established, there 

 is a steady diffusion-stream of the vapour through the upper 



