62 PROCEEDINGS OF THE AMERICAN ACADEMY. 



as the others. The precautions taken were somewhat greater, aud the 

 fractional precipitation with amnionic oxalate was continued long after 

 the last traces of calcium discoverable by the spectroscope had disap- 

 peared from the precipitates of magnesic oxalate. The ammonic mag- 

 nesic chloride, already very pure, prepared from this sample, was then 

 crystallized eight or ten times, the last six or eight recrystallizations 

 being conducted in platinum. From over a kilogram of magnesic 

 chloride used in the beginning, the portion finally separated out consisted 

 only of a few grams. This sample showed no traces of the sodium line 

 when tested with the spectroscope ; indeed, several other samples, ob- 

 tained from the mother liquors of the purest sample, gave equally satis- 

 factory negative spectroscopic results. Since the magnesic chloride had 

 contained in the first place a very noticeable amount of sodic chloride, 

 the fact of the complete elimination of the impurity seemed a satisfactory 

 indication of the elimination of other foreiijn materials. The double 

 chloride was dried over an alcohol lamp, and treated in the same manner 

 as the other samples. 



Purification of Silver. 



No very great labor was expended iipon the purification of the first 

 quantity of silver, as the chlorine in magnesic chloride was to be pre- 

 cipitated with an undetermined excess of silver nitrate. Residues were 

 therefore worked up by dissolving silver (obtained by reduction with 

 zinc) in nitric acid, precipitating the metal as chloride, and converting 

 tlie chloride into metallic silver by means of invert sugar. The reduced 

 silver, after having been fused into buttons, was thoroughly washed and 

 dissolved in nitric acid. The solution of argentic nitrate thus obtained 

 was diluted very much with water, allowed to stand, aud filtered just 

 previous to using. 



With the second sample, on the other hand, much greater care was 

 taken, as it was designed in this case to ascertain the direct ratio between 

 silver and magnesic chloride. The material came partly from some 

 refined silver, purchased in the market, aud partly from some pure silver 

 residues remaining from previous work. The silver was jirecipitated 

 from a solution of the nitrate with pure hydrochloric acid, and reduced 

 by means of invert sugar and pure sodic hydrate, the sodic hydrate 

 having been previously freed from heavy metals by electrolysis. Both 

 the chloride and reduced silver were very thoroughly washed, the silver 

 was dissolved in pure nitric acid, and the process was repeated. After 

 this cycle of operations had been performed four or five times, the re- 



