150 PROCEEDINGS OP THE AMERICAN ACADEMY 



iiber die Fortscbritte der Chemie." In many cases the observed 

 values were so near whole numbers that no important error in the cal- 

 culation of analyses arose from this practice, the differences neglected 

 being no greater than the uncertainties of analytical method, and this 

 was especially true with regard to the larger atomic weights. 



One exception to the theory, however, was so marked that it could 

 not be overlooked, namely, the atomic weight of chlorine, which was 

 capable of being determined with great accuracy ; and all the deter- 

 minations uniformly gave a result which was closely 35.5. This and 

 a few similar cases suggested the idea, that, if the atomic weights were 

 not even multiples of the received hydrogen atom, they might be mul- 

 tiples of the half or quarter hydrogen atom, which would simply amount 

 to taking as the ultimate atom of material things a still smaller unit. 



The well-known chemist, Dumas of Paris, was led by this view to 

 undertake a redetermination of a large number of atomic weights, and 

 many of the results then obtained are still accepted as authoritative.* 

 As was to be expected, Dumas found a much closer agreement with 

 this modified theory than with the original hypothesis of Prout ; but 

 obviously such evidence could have but little bearing on the general 

 theory that the atoms were all aggregates of some common unit, for 

 by takiug that unit small enough, — even no smaller than the one 

 hundredth of the received hydrogen atom, — all the atomic weights, 

 even those most accurately determined, would be expressed by whole 

 numbers within the limits of probable error. 



Soon after, Stas of Brussels, a former assistant of Dumas, endeav- 

 ored to set the question of Prout's theory at rest by an investigation 

 which will be ever memorable for its extreme accuracy. f He selected 

 for his investigation those elements whose combining proportions were 

 capable of being determined with the greatest accuracy, and, working 

 on large quantities of material, with every refinement which a full 

 knowledge of analytical methods could suggest, he obtained results 

 which it seemed impossible to reconcile with the theory in any way. 

 This investigation, published in 1865, seemed at first to disprove the 

 theory altogether. 



Nevertheless, when Stas's results came to be collated, and as other 

 determinations of similar accuracy came to be published, the fact ap- 

 peared that a large number of the most accurately determined atomic 



* Annates rle Chimie et de Physique, 3d ser., lv. 129 (1859). 

 t Me'moires de l'Acade'mie Royale de Belgique, xxxv. Also Ann. Ch. 

 Pharni., Suppl., iv. 168. 



