Origin of Molecular Attraction. 665 



internal latent heat, then at all temperatures up to the 

 critical 



L— E 



-j~ — jyj d = constant characteristic of liquid. . (2) 



This has been verified by a most exhaustive examination 

 of the data for 31 liquids of diverse chemical types, com- 

 pounds but not elements (Journal of Physical Chemistry, vi. 

 1902, p. 209, viii. 1904, pp. 383, 593, ix. 1905, p. 402, 

 x. 1906, p. 1). Mills interprets his relation by treating 

 cohesion as due to gravitation, taking d l/3 — D 1/3 as a measure 

 of the change of the mutual gravitational potential energy 

 of the molecules when they pass from the state of density d 

 to that of density D. There are two well-known and per- 

 fectly definite reasons why gravitation cannot account for 

 cohesion : first, the forces between neighbour molecules are 

 enormously greater than their mutual gravitation ; and, 

 second, the latent heat required to evaporate molecules 

 against their gravitation varies for unit mass with the size 

 of the mass evaporated. This variation has never been dis- 

 covered because the latent heat of evaporation against gravi- 

 tation is so minute a fraction of the latent heat of evaporation 

 against cohesional force, that its variations are beyond the 

 reach of existing experimental refinements to detect. If Mills 

 had traced his relation to the law of the inverse square as it 

 operates in a uniform mixture of equal numbers of equal 

 opposite charges of electricity, he would have been on the 

 right track with Fessenden. In this case, the repulsions 

 between the like charges introduce an element unlike gravi- 

 tation. This mixture of electric charges has been shown 

 above to be a convenient and proper schematic representation 

 sometimes for a collection of uniformly electrized molecules 

 in contact. Let us waive the stipulation about the molecules 

 being in contact at absolute zero, and let us take (1) for the 

 electrostatic energy of the molecules in unit volume of a 

 liquid of density p or d. Then for a molecule of volume m/d 

 the electrostatic energy will be 27r/3K(m/<f) 1 3 . If with 

 Fessenden we assume that exactly a similar condition of 

 polarity prevails in the saturated vapour of density D, we 

 can write 27rD 13 /3K»i 1 3 for the electrostatic energy of the 

 vapour, assuming for the moment that K does not change. 

 The change is 27r(d 1 < 3 — D^/SKm 1 ' 3 . If this is equivalent 

 to the internal latent heat, then for a given substance 



L — E 



,, a -pn :3 = constant at all temperatures. . . (3) 



This is the result which appears to justify Fessenden's 



