66 PROCEEDINGS OF THE AMERICAN ACADEMY. 



Of these figures (2), (5), and (fi) represent single results, and 

 the other figures represent averages. Since a number of samples of 

 baric chloride and a number of different methods of comparison 

 were used in the individual cases, the averages, which to some extent 

 eliminate the differences, are more reliable than the single results. It 

 is evident that the different varieties of silver were quite as nearly 

 alike as the accuracy of our present processes demands. 



The investigation of the silver closed with a direct determination of 

 the amount of sulphur present in a sample which was purposely made 

 under conditions as favorable as possible for the retention of this 

 impurity. Twelve grams of silver, which had been prepared by the 

 sulphite method and fused three times successively for a long period 

 in the flame of an ordinary blast lamp, were dissolved and precipi- 

 tated by electrolysis in a solution of silver nitrate prepared from the 

 same silver. The residual electrolyte, which must have contained all 

 the sulphur, was much diluted, freed from silver by the addition of 

 hydrochloric acid, and evaporated on the steam bath to free it from 

 nitric acid. This solution yielded 1.4 milligrams of baric sulphate, 

 corresponding to 0.2 milligram of sulphur. Hence under these con- 

 ditions the silver held only g-Q.V 07 °f i* 8 weight of sulphur ; and it is 

 fair to conclude that according to the usual method of subjecting the 

 silver only to a single brief fusion not more than 200,000 °f lts weight 

 would be absorbed. This result is in entire accordance with the 

 quantitative and qualitative results already described ; it shows that 

 silver fused once in illuminating gas may be used without appreciable 

 error for all ordinary work. 



The preparation of pure water, alcohol, sulphuric and nitric acids, 

 and all other necessary materials, has been discussed in previous 

 papers.* 



Standard Solutions. — Several standard solutions were so con- 

 stantly used during the research that they may well be described at 

 the beginning. In the first place, exactly a gram of silver was dis- 

 solved in a slight excess of nitric acid, with the usual precautions,! 

 and diluted exactly to a litre. A solution of pure hydrochloric acid 

 was then made of equivalent strength by comparison with pure sodic 

 carbonate, and verified by comparison with the silver solution. This 

 last comparison is not easily made, for the argentic chloride separates 

 but slowly from a solution so dilute. It was found most convenient 



* These Proceedings, XXVI. 245-249 ; XXVIII. 23. 

 t Ibid., XXV. 198; XXVIII. 24. 



