( 646 ) 



f dT,,\ 

 Starting from b = 2b x we find in this way _ , =0.66; 



\-Lkd.vJ x==r[ 



with b^ = 2.5 : 0.65 and with b„ = 3 b l : 0.64. The error which we 

 committed in our choice of b u , will, therefore, bring about no con- 

 siderable modification in the result; it would, indeed, be considerably 

 modified if the critical point of anthraquinone should prove to lie 

 considerably higher than 1120\ This is not in contradiction with 

 our former remark that it is of little importance whether the reduced 

 temperature is \/. 2 or 7s ixt t' ie triple point; for this we started from 

 the supposition of the linear dependenee, whereas here we have 

 abandoned this supposition, and calculate this dependence from the 

 experimental data. So according to the course of reasoning followed 

 here the a lt is given by the experiment, and the smaller value of 

 m would now result in a higher value of a, at given b x , b 2 and a lt . 

 If our estimation may be considered as not too inaccurate, we may 

 conclude that the deviation from rectilinearity does increase the value of 



— — -), but by no means in the degree which would be required to 



TkdxJ x= i 



reach the critical value 0.9. (The value derived from the supposition 

 of rectilinearity is 0,58). 



Though the foregoing calculations teach us hardly anything 

 positive, they fix first of all our attention on the great desirability 

 of more data concerning the values of the quantities a and b of very 

 little volatile substances ; for it appears again that the whole behaviour 

 of all the systems in which such substances appear, is controlled by 

 these quantities, and it would exactly be of great importance for 

 the theory of mixtures, if its results could be tested by such cases 

 where the properties of the two components differ strongly. It is true 

 that it will not be easy to determine the critical point of such sub- 

 stances in the usual way, but we should have gained already much 

 if we could obtain an estimation of the critical temperature by 

 calculation of the a and b from the deviations from the law of 

 Boyle in rarefied gas state, so still some hundreds of degrees below 

 the critical point. 



And further I think that after the foregoing I may be allowed to 

 draw this conclusion, that the appearance of a temperature maximum 

 in the three phase line, far from being thé general case, will be 

 confined to mixtures of very exceptional nature. 



Much more frequently than a temperature maximum will a pres- 

 sure maximum occur. It appears from equation (1) that this will 

 always be the case, when the expression : 



