346 PROCEEDINGS OF THE AMERICAN ACADEMY 



Relation of Silver to Copper. 



The mean value, 63.437, has a greatest variation of ±.01, and a 

 probable error of ±.0023. 



The value obtained by Shaw was 63.333 



« « " corrected .... 63.360 

 The value given by Clarke is 63.173 



It remains only to prove the absolute purity of the copper used and 

 the silver formed. The only metals which could affect the atomic 

 weight of copper to an appreciable extent, if present in traces, are 

 bismuth and silver. The former was tested for with great care, using 

 a delicate spectroscope, and not the faintest trace of the very well- 

 defined blue bismuth line was apparent ; further, ten grams of the 

 copper were dissolved in nitric acid, a little potassic hydrate added to 

 the diluted solution, and the whole boiled and shaken for three hours. 

 According to Hampe, all the bismuth present will be found in the 

 basic precipitate ; this was tested qualitatively for the metal, and not 

 a trace was found. The solution of a portion of the copper in pure 

 nitric acid gave not the faintest trace of opalescence with hydro- 

 bromic acid, excess of ammonic hydrate, or baric chloride. A gram 

 of the copper was tested for arsenic and antimony in a Marsh-Berzelius 

 apparatus, and no trace of a mirror was obtained. As the presence of 

 so much copper would cause the generator to run too rapidly, the 

 solution was^treated with just enough potassic ferrocyanide to effect 

 precipitation, and the colorless filtrate was run into the apparatus. 

 This method has been found to give good results. The copper and 

 the precipitated silver each dissolves in nitric acid without leaving a 

 trace of residue. 



If the copi^er is pure, the only possible impurity which the silver 

 could contain is copper ; and this was tested for with great care in each 



