4l8 PROCEEDINGS OF THE AMERICAN ACADEMY § l6 



Chauvenct, from o to 5 times the probable error; and Mr. 

 Haskell has extended this table in logarithms from 5 to 

 100 times the probable error. It will therefore be an easy 

 matter to calculate, rough!}-, the chances of all the different 

 velocities which are likely to occur, and to see whether 

 the various phenomena, such as have been described, can 

 be attributed to the inequality of their distribution accord- 

 ing to the laws of chance. The results must be accepted 

 with the greatest caution, as indicating the possibility of 

 explaining the phenomena in this way, and not the proba- 

 bility of having found the true solution. 



The internal molecular heat has been so successfully 

 treated of late, by the ordinary assumptions of the Kinetic 

 Theory, that there can be little or no question that the sub- 

 ject is properly a branch of this theory, and consequently, 

 being entirely independent of cohesion, cannot conflict 

 with any supposition as to the nature of the latter. 



The question of vapor tensions needs a special examina- 

 tion. 



Maxwell has pointed out in his Theory of Heat, under 

 the "Molecular Theory of Evaporation and Condensation'' 

 (page 323), that a liquid in contact with its vapor is in 

 equilibrium when the rate of evaporation of the liquid is 

 equal to the rate of condensation of the vapor, both being 

 determined by the laws of chance. 



By assuming that the total energy of a substance varies 

 as the square of a velocit}-, we may at once obtain expres- 

 sions for the probability of a particle of water becoming 

 steam and a particle of steam becoming water, taking into 

 account the interchange across any surface which separates 

 them. The theoretical solution will be of the general form, 



the demonstration of this formula (which is itself of slight 

 importance) will be omitted on account of its length. A 



