﻿(86) 



Molecular Thermodynamics. 029 



an 1 so on, for we then obtain 

 'i=~; W = t+~t s ^ 



t 2 = ? + h^ t 2 ' = (ft, + J?) + (h + ih s ) 



+ $ + £*.) 



and so on. 



Clearly the coefficients of J M alone, determined term by 

 term (with increasing concentration) will yield the quantities 

 t y /i, etc., one by one, while, if the coefficients of both J M and 

 J s are determined, the pairs (t u ti), (t 2 , t 2 '), etc., will yield 

 in succession the pairs of quantities [t, t s ), (Ji, h s ), etc. 



It must be observed, however, that while t s , h s , etc., are 

 constants peculiar to the solute s (in a given solvent), on the 

 other hand, t, h, etc., are the corresponding average quantities 

 for the whole solution, and so depend upon the relative pro- 

 portion in which these solutes are present [see (56 a)]. Some 

 cases where t, h, etc., vary under the conditions of experi- 

 ment can in fact be dealt with, without difficulty, as will be 

 shown later, but, of course, it is simpler if they are constant. 



Two important cases where this is so present themselves, 

 viz., the case where there is only one solute and the case 

 where there are two, but the concentration of the one is 

 negligible in comparison with that of the " second," the 

 former being the solute s. These two cases gain further 

 importance from the fact that the form of G, the " true " 

 general terms, will here probably be simplest, depending in 

 the second case almost entirely on the " second " solute. 



The last point suggests, however, that in the matter of 

 separating J from Gc, cases where t, h, etc., vary under the 

 conditions of experiment may prove more useful. 



The Quantities a 50 , a , a 1? etc. 



(85) shows that the evaluation of t, t s , etc., would not give 

 us these quantities (in general), but since the differences 

 shown in (85) will generally be much smaller than the 

 (corresponding) quantities a , a, , etc., a rough estimate of 

 these differences (all fairly nearly expressible in terms or a u 

 which is something slightly greater than the mean molecular 

 weight of the pure solvent) will suffice to determine a , «„ , 

 etc., with a relative error much less than that of the estimate. 



Actually, to determine the solvent constants experimentally 



