OF ARTS AND SCIENCES. 23 



metal serving as electrodes. This method for the preparation of 

 pure silver is a very satisfactory one. Since the silver was allowed 

 to cool in an atmosphere of hydrogen, it could have contained no oxy- 

 gen. The agreement of Experiment 19 with the others is satisfac- 

 tory proof that the amount of sulphur contained in the first samples 

 of silver must have been infinitesimal, if appreciable at all. 



Other Materials. — The methods used for the preparation of pure 

 water, pure nitric and sulphuric acids, and pure sodic carbonate, 

 have been discussed at length in a previous paper.* Precautions 

 taken with regard to carbon dioxide, hydrobromic acid, and many 

 other substances, are to be found under earlier heads. Alcohol was 

 purified for the present investigation by repeated distillation in ap- 

 paratus wholly free from cork or rubber connections. In some cases 

 a platinum still was used. 



The large mass of platinum used in the first experiments was 

 kindly loaned by Professor Cooke, but subsequently a quantity was 

 purchased especially for the work. The methods used in freeing 

 the surface of these vessels from iron are described in the fourth 

 paper upon the revision of the atomic weight of copper, f 



The Method of Analysis. 



Thus far it has been found possible to determine accurately only 

 the ratio of baric bromide to silver and argentic bromide. Un- 

 fortunately no accurate method for the direct determination of the 

 amount of metal present is known to exist; hence a complete 

 analysis is impossible. 



The usual scheme of operations was very simple. The baric 

 bromide, after having been pulverized in an agate mortar, was 

 heated for a long time at 200-400° ; it was then gradually raised to 

 dull redness, and maintained for some time at that temperature. 

 Repeated heating sometimes caused a very slight loss, due to in- 

 creased decomposition; but more usually the weight remained con- 

 stant. The drying oven was a large porcelain crucible, and at first 

 illuminating gas was used as the source of heat. Afterwards, 

 when a faint trace of cloudiness found in the solution of the baric 

 bromide was traced to the formation of baric sulphate from the sul- 

 phur in the illuminating gas, an alcohol lamp was used exclusively. 

 In Analyses 3, 4, 15, 16, and 17 the amount of this insoluble 

 residue was determined, and appropriate correction was made. In 



* These Proceedings, XXVI. 245-249. t Ibid., 249. 



