igo^\ Mills — Molecular Attraction. 173 



where T must be the critical temperature and d must be the 

 critical density. 



If therefore the values of — given in Table 25 be divided 



by the values of —7- given in the same table, the results should 



prove constant for all non-associated substances. The result 

 of this division is given in Table 25. 



The mean value for the non-associated substances is 10.76. 

 Since the values for // are uncertain by about two per cent., 

 and since the critical data cannot be measured accurately, the 

 close agreement can be regarded as exceedingly satisfactory. 

 A review of the data leads the author to believe that isopen- 

 tane and normal octane are the only variations that are not 

 due largely to the values adopted for \x! . 



Associated substances, a priori, could not agree with the 

 equations deduced and they do not. But considering them 

 more particularly, it will be seen that if at the critical tem- 

 perature the molecular association had vanished (as is said to 

 be the case for the alcohols), equation 24 would hold. If 

 instead of the average value obtained for y! at the lower tem- 

 peratures, we use the values for the constant obtained near 

 the critical temperature, equation 22 should also hold, simul- 

 taneously with equation 24. For water the data near the 

 critical temperature is lacking, but making use of the proper 

 values for the alcohols, we obtain the results obtained in 

 Table 25. These results would evidently be somewhat better 

 if the observtions of /*' could have been obtained yet nearer 

 the critical point, and we are thus led to regard these asso- 

 ciated substances as giving a very remarkable confirmation of 

 the theory. 



It should be noted that the constant obtained from equa- 

 tion 25 and given in Table 25 is just one-half of the constant 



obtained in Trouton's formula, -=- = constant, where T is 



the boiling point of a substance. That there is a reason for 



