LECTURE III.] HISTORY OF CHEMISTRY. 41 



taining different quantities of oxygen. He afterwards develops 

 his views upon this point still more clearly. 25 He says : " I 

 have to show, then, that the proportions of oxygen in the 

 oxides depend upon the same conditions as those in other 

 compounds; that these proportions may vary progressively 

 from the limit, at which combination becomes possible up to 

 that at which it attains the highest degree." The limits them- 

 selves he regards as determined by circumstances of cohesion. 

 In the same way he believes in the existence of salts with vary- 

 ing quantities of base. If the base is precipitated, by means of 

 an alkali, from a salt with an insoluble base, he supposes that a 

 certain quantity of acid is precipitated along with the base, and 

 that this quantity is variable.' 26 In short, compounds with con- 

 stant proportions are, according to Berthollet, exceptions, and 

 the proportions in which substances combine are, as a rule, 

 dependent upon the conditions of the experiment. 



If we summarise Berthollet's views once more, we may 

 say that affinity appeared to him to be a force identical with 

 gravity, 27 whose phenomena are more varied only because it 

 sets the molecules themselves into motion, and its effects are 

 dependent upon the size and shape of the particles. In apply- 

 ing these physical principles to chemical reactions, he arrives 

 at the conception of chemical mass, which he defines as the 

 product of affinity and quantity present. The chemical effects 

 depend upon the magnitude of this chemical mass, and upon 

 the cohesion of the substance, that is, upon its solubility and 

 its greater or less volatility. This then leads him further to 

 two general conclusions : 



1. Tables of affinity are useless, since in them affinity is 

 assumed to be constant and independent of physical conditions, 



2. There are compounds with varying, and progressively 

 increasing proportions of their constituents. 



The first of these conclusions is adopted generally, and we 

 find that the tables of affinity disappear soon after the appear- 



23 Stat. Chim, 2, 370; E. 2, 316. 2(! Ibid. I, .86; E. I, 58. '* Ibid, 

 I, i ; E. x, vii 



