118 PROCEEDINGS OF THE AMERICAN ACADEMY. 



The boat containing the cobalt was of course placed in the hottest part 

 of the furnace, and the open end of the inner tube was adjusted just 

 beyond this point. At first, the glaze of the receiver tube was attacked 

 by the hot bromine and bromides, and the first sample of sublimate was 

 found to be contaminated by the decomposition products of the glaze. 

 Later the porcelain became covered with a brilliant-colored coating of a 

 cobalt glaze which seemed to be permanent. The sublimate used in the 

 analyses was taken more especially from the central portions of the sub- 

 limed masses, so that even if a small amount of alkali had been extracted 

 from the porcelain it must have remained clinging to the walls of the 

 tube. All the sublimates excepting the very first gave a wholly clear 

 aqueous solution. 



As has been already stated, anhydrous cobaltous bromide thus pre- 

 pared consists of brilliant green crystalline plates, which quickly absorb 

 water from the air to form the pink hydrated salt, and dissolve with 

 great ease in cold water. 



The specific gravity of anhydrous cobaltous bromide being unknown, 

 the following experiments were made to determine it. (a). 2.0915 

 grams of the salt which had been dried at about 200° displaced 

 0.3285 gram of kerosene (dried over soda and redistilled) at 25°. 

 (b). 1.3965 grams of the salt displaced 0.2186 gram of kerosene. The 

 kerosene, which was used in this case because cobaltous bromide was 

 found to be slightly soluble in toluol, had at 25° a specific gravity of 

 0.7698 referred to water at 4°. Hence we have for the specific gravity 

 sought, (a) 4.901, (b) 4.917. The mean of these two values is 4.909. 



Since brass weights were employed, in order to correct the weiglits of 

 cobaltous bromide found in air during the subsequent accurate analyses 

 to that which would have been observed in a vacuum, 0.00010 gram 

 should be added for each gram of this material. In the same way 

 0.000045 gram must be add^^d to every apparent gram of argentic bro- 

 mide, and 0.000031 gram must be subtracted from every apparent gram 

 of silver. These corrections are all applied below. 



The balance used in this research was the one used in those upon 

 copper, barium, strontium, zinc, and nickel : it needs no further de- 

 scription. The weights were of brass, gold plated ; they were stan- 

 dardized with great care, and were used for no other work. 



Purification of Materials. 



Cobalt. — The doubt which has been thrown on the elementary char- 

 acter of cobalt by several experimenters led us to the application of un- 



