i6 Meldkum, L)evelopmc7it of tJic Atomic Theory. 



Bostock, Clement, Dtisormes and Proust. Yet the failure 

 of these chemists to discover the law of multiple propor- 

 tion, despite their immense labours, was complete. 



An incorrect expla}iatio II of the failure. 



The reason usually offered for this failure is, that the- 

 data for the composition of substances were calculated in 

 such a way as to hide the law.'* Plainly the implication 

 is, that the data calculated in a suitable way must reveal 

 the law at once. This is mere guess-work, for as a matter 

 of fact, data were frequently stated in precisely the way 

 required. Proust, for instance, gives practically all his 

 data for the oxides and sulphides of a metal, in terms of 

 lOO parts of the metal. ^' 



The true explanation. 



The true explanation is twofold. In the first place, 

 accurate chemical analysis is impossible without a check of 

 some kind. That the analyst should have good intentions, 

 even the best intentions, is not enough. In the absence 

 of a guiding principle, chemists cannot tell when a 

 substance is pure, or when an analysis is correct. 

 As explained in the first paper of this series, it was this 

 state of uncertainty which contributed at the beginning 

 of the nineteenth century, more than anything else, to the 

 spread of C. L. Berthollet's ideas regarding combination 

 in indefinite proportions. Arrhenius has pointed out 

 that every chemist noiv prepares his substances so that 



^* E. von Meyer, "Hist, of Cliem.,"' Eng. trans., pp. 195-196, 1906, 

 and Arrhenius, "Theories of Chem.,"' Eng. trans., p. 16, 1907. 



15 Ann. de Chitii., vol. 28, p, 214, 1798 ; Jour, de P/iys.^ vol. 54, p. 92, 

 1802 ; vol. 55, p. 330 ; vol. 59, p. 324, 326, 330, 352, 1804 ; vol. 62, p. 136^ 

 138, 139, 1806 : vol. 63, p. 431, 1806. 



