44 THE TRUE VALUE OF a OF VAN DER WAALS' 



(e), a = 3 MN' ir3 T r C/d ^ = 1.3392 when calculated as usual 

 from the formula d = (d—J))T c ' li /(T c —Tf l \ C, cal- 

 culated from Morgan and Daghlian (Jo urn. Amer. Chem. 

 Soc, 33, p. GSO, 1911) without correction according to 

 Lohnstein makes C = 2.160. From this a = 21.99 X 10 12 . 

 If we substitute the proper C for benzene of Richards 

 and Coombs in place of that of Ramsay and Aston used 

 by Morgan and Daghlian, G would be 2.181 and this 

 would make a = 22.21 X 10 12 . 



There is in the case of fluor-benzene a wide variation in the 

 values of a calculated by these different methods, from which I 

 believe we may infer that probably the critical data need some 

 revision. It would appear as if the critical density were a little 

 greater than it ought to be. The first two methods give a concor- 

 dant result of a mean of 21.55. The latent heat formulae give a 

 mean value of 23.09. From the surface tension determinations of 

 Morgan and Daghlian by the drop weight method using their 

 results uncorrected by the Lohnstein method we have 21.99, which 

 by the substitution of Richards and Coombs' value for C in place 

 of that of Ramsay and Aston for Benzene, is raised to 22.21. 

 This value cannot, I think, be very far in error. It is possibly still 

 a little low as the Lohnstein correction generally has the effect of 

 raising the surface tension values.- 



The value calculated for a from the molecular weight and 

 valences, taking 30 valences to the molecule, fluorine being consi- 

 dered to be monovalent, is 23.81. This is about 7 °/ higher than 

 the surface tension value and higher than any other determination 

 except that from the latent heat by Dieterici's formula. It is 

 6 — 7 °/ above the mean value. One would expect that fluorine would 

 be more than monovalent, but it may be mentioned that Pascal by 

 his method of measuring the valences also found fluorine to be 

 monovalent in fluorbenzene. The agreement, therefore, in the case 

 of fluorbenzene is not so good as in many other substances, but 

 whether this is due to the fact that there are fewer valences in the 

 molecule than we usually assume to be there, or whether it means 

 that the critical data need some revision, T do not know. In any 

 ease the determination from the valences and molecular weisht is 

 only G — 7 7o différent from the mean and lies between the extremes 

 of the values found. On the other hand van Laar's value, which 

 corresponds to 17.90 X 10 12 > is certainly 20 °/ in error. 



