﻿Attraction of Unlike Molecules, 7 



In the case where the level of the liquid is allowed to fall the 

 velocity of the gas is not exactly 0, and in a later paper 

 (Sitz. Akad. Wien, xcviii. 1890) Stefan has given a cal- 

 culation wherein the small value of a 2 is taken account of. 

 By very simple experiments on the evaporation of ethyl 

 oxide and carbon disulphide in test-tubes, Stefan verified his 

 expression first as regards the relation of h and t at constant 

 temperature, that is at constant p s , and then as regards the 

 very characteristic factor log p/(p —p s ) in which the satura- 

 tion-pressure enters, by studying the evaporation of ethyl 

 oxide at temperatures ranging from 11 0, 3C. to 28°'7, where 

 the range in the saturation-pressure p s is from 302 millim. of 

 mercury to 605. 



Winkelmann has still more thoroughly verified Stefan's 

 evaporation theory in applying it in an extended series of 

 experiments to the determination of the diffusion-coefficients 

 of a number of vapours into air, hydrogen, and carbon 

 dioxide (Wied. Ann. xxii., xxiii.,xxxiii.,xxxvi.). As regards 

 the formula (5) the most important part of Winkelmann' s 

 work is his further verification of the soundness of the factor 

 log p/{p—p s ) by varying p in the case of water from 61 

 millim. to 749, while p s was about 1*5 millim. 



The correctness of Stefan's formula (4) for the diffusion of 

 gases composed of forceless molecules seems to me therefore 

 to be well assured by the successful application of the 

 principles involved in it to the details of the process of 

 evaporation; and the foregoing brief sketch of his theory 

 serves as the most natural introduction to a theory of gaseous 

 diffusion wherein the attractions of molecules are taken 

 account of. 



Let us first see wherein the expression 



^ = / m l + m 2 \ i 3^ 



\ m 2 / 87r^ai+a 2 ) 2 (?ii + ra 2 ) 



for the diffusion-coefficient of forceless molecules applies 

 to natural gases and wherein it fails to apply. As m V K x 2 is 

 proportional to absolute temperature T, and n x + n 2 is pro- 

 portional to p/T, where p is the pressure at which the diffusion 

 goes on, 



T f / 1 IV 1 

 Doc-f— + L) * (6) 



p \m 1 m 2 J [a 1 + a 2 y v J 



Thus the theoretical diffusion-coefficient varies inversely 

 as the pressure, which has been proved experimentally by 

 Loschmidt to be the case for the natural gases (Sitz. Akad. 



