228 PHENOMENA DEPENDENT ON MOLECULAR PATHS 87 



the viscosity and partly in the estimation of the tempera- 

 ture. But against this view there is the circumstance that 

 a fluctuation is shown in the values of the constant H first 

 calculated which occurs in several esters in almost exactly 

 similar fashion. 



We are also forced to the opinion that these fluctua- 

 tions are not entirely fortuitous by the consideration that 

 the theory is not without other underlying causes, which 

 may be adduced, along with that theoretically investigated 

 by Sutherland and justified by observation, in order to 

 explain the large variation of viscosity in vapours with the 

 temperature. 



88. Dissociation 



Among the causes which make vapours deviate from the 

 laws of perfect gases the most important depends upon the 

 dissociation which the molecules undergo as well from rise 

 of temperature as from diminution of pressure. 



In earlier sections of this book we have several times 

 spoken of this dissociation. Especially in 48-51 has its 

 nature been described and the influence which it exerts on 

 the density of a vapour. In a vapour the molecules are not 

 all of the same size and mass as in a perfect gas. The case 

 often occurs, in consequence of the forces of cohesion, in 

 which two molecules which chance to come near enough to 

 each other join together to form one molecule of double 

 size ; and just as frequently the case arises that such a 

 double molecule splits up again into its constituents, either 

 as a result of collision with another molecule, or simply in 

 consequence of an increase of speed, which not only breaks 

 up the whole molecule, but also partly tears apart the atoms 

 bound up in it. Hence it happens that a vapour is always 

 a mixture of large and small molecules. 



The ratio in which the simple and compound molecules 

 are mixed together differs with the temperature; for an 

 increase of speed, as brought about by a rise of tempera- 

 ture, increases the probability of the division of the larger 

 masses. The number of simple molecules therefore in- 



