82 PROCEEDINGS OF THE AMERICAN ACADEMY 



wards the serious nature of these errors, which he clearly appreciated. 

 Most of the investigations of Stas remain models of precision; and no 

 one who has not attempted practically to follow in his footsteps can 

 form an adequate idea of the greatness of his success. 



Stas's study of the " Statics of Argentic Chloride " has been already 

 referred to. As a result of this long and valuable series of experi- 

 ments, he concluded that the true end-point must be half way between 

 the two end-points obtained by titrating in opposite directions. When 

 this true end-point had been reached, he found that the supernatant 

 liquid was rendered equally opalescent by equivalent amounts of 

 hydrochloric acid and argentic nitrate. These conclusions suggested 

 two possible methods for the solution of the problem before us ; and 

 it is worthy of remark that at least one of these methods had been 

 suggested by Mulder fifteen years before Stas's publication.* 



The unsatisfactory results obtained by the old method inclined one 

 to subject the newer methods also to a careful testing before going 

 further ; accordingly a long series of experiments, not unlike those 

 described by Stas, were instituted to prove them. In the first place, 

 pure thoroughly washed argentic chloride which had been made and 

 kept in darkness was shaken with a dilute solution of pure baric 

 nitrate and nitric acid ; and then the two end-points were determined 

 by adding first hydrochloric acid, and afterward argentic nitrate, un- 

 til no more cloudiness could be observed in either case. The results 

 were not very satisfactory, for a reason which appeared later. Never- 

 theless, many experiments were made in this way, showing that the 

 observation of Stas was at least approximately true. The observation 

 with regard to the equality of opalescence could not be confirmed at 

 once, because the method requires much practice; but in the end the 

 confirmation of it was much more satisfactory than the previous one. 

 It was clear, however, that both methods depended upon a question 

 of judgment rather than upon definite weighing or measuring. 



In default of a more satisfactory method a series of experiments 

 upon the atomic weight of barium was made by determining both 

 end-points and taking the mean as the true one, according to Stas's 

 second method. During this series a curious fact became manifest. 

 As the alternate addition of hydrochloric acid and silver nitrate pro- 

 ceeded, the necessary amount of each solution increased to an enor- 

 mous degree. For instance, in Experiment 33 below, the first 

 difference between the end-points was 16 milligrams of silver, while 



* Silberprobirmethode, p. 94. 



