EQUATION AND THE NATURE OF COHESION. 7 



well founded. This law of Young's, as I found afterwards, is 

 simply a special case, at the limit of absolute zero, of the well 

 known Eötvös surface tension law, and the calculation of a can 

 just as well be made from Eötvös' law at ordinary temperatures, 

 as I shall show in a moment, but T did not know that this was 

 the case when I published my first paper. The values of a which 

 were thus obtained from Young's law, were on the whole some- 

 what higher than those generally computed for a assuming that 

 a and b are constant, and particularly they were higher in 

 the diatomic gases. In making this calculation I had to make use 

 of the Eötvös constant, C, in the equation: sV' 2li =C(T c — -T). 

 I supposed that C was practically the same constant for all non- 

 associating substances, at. any rate for all normal substances with 

 more than three atoms in the molecule. This assumption is not in 

 reality true, but in making it I was following the opinion of Eötvös 

 and Ramsay and Shields. I took as the mean of the constant the 

 value 2.19. Eötvös gave 2.27 and Ramsay and Shields about 2.12. 

 With this value of the constant I found a for the 2 G substances 

 of which the critical data ware most accurately known, namely the 

 substances studied by S. Young. With these values of a I tried 

 to find out what it was in a molecule which determined its value 

 of a, and after many unsuccesfull attempts the importance of 

 valence occurred to me and on trying I found that these values of 

 a gave a constant in each case when divided by the two thirds 

 power of the product of the molecular weight and the number of 

 valences. The only important deviations were iodobenzene and brom- 

 benzene, substances in which the critical data had been determined 

 by extrapolation over a wide temperature interval, and were pre- 

 sumably on this account less certain than in the other substances. 

 Further study showed that the constant I had thus found, when 

 a was for a single pair of molecules, was nothing else than the 

 value (w^ 2 /)-' 3 of a substance of molecular weight of unity and with 

 one valence; m x being the gravitational mass and k the gravita- 

 tional constant. This permitted me to write the value of a in 

 the more general form given at the beginning of this paper : 

 a = N 2 (m 2 K X Val(Mf 13 . In making this discovery there was no 

 juggling with the valences at all, as van Laar implies. Everywhere 

 carbon was tetravalent, oxygen bivalent, hydrogen monovalent. The 

 only elements of uncertain valence were CI, Fl, Br, I. These might 

 have 1, 3, 5, or 7 valences. It was impossible to know which to 

 take. So I followed the most recent determinations of the valence 

 by Pascal taking Fl as monovalent and CI as trivalent. These values 



