RICHARDS AND BAXTER. — ATOMIC WEIGHT OF COBALT. 65 



nor the ammonia could have introduced either of these impurities, the 

 oxide must have been as pure as possible. It has already been shown 

 that specimens of cobalt prepared by any of the methods described in the 

 preceding parts of this paper are essentially identical.* 



A weighed platinum boat, containing several grams of cobaltic oxide, 

 was placed in the porcelain tube, and, after the air had been exhausted, 

 the portion of the tube containing the boat was heated to full redness. 

 The expelled oxygen was removed with the air-pump ; and the tension 

 of the gases in the tube was finally reduced to a small fraction of a 

 millimeter of mercury. The tube was then cooled while the vacuum was 

 still maintained, air was admitted, and the boat was transferred to the 

 weighing bottle and weighed. Successive ignition for periods of about 

 two hours each reduced the weight of the cobaltous oxide slowly by 

 amounts varying from two to five tenths of a milligram, until constant 

 weight within one teoth of a milligram was obtained. During this 

 process the cobaltous oxide, which was light brown at first, gradually 

 assumed a darker tint. 



Since complete reduction of this oxide by hydrogen at a temperature 

 which could be borne by hard glass is very slow, the rednction was con- 

 ducted in a porcelain tube heated by the furnace. The metallic cobalt, 

 which sintered together in such a fashion as to occupy not more than one 

 fifth the original volume of the oxide, was cooled in a vacuum, and then 

 transferred to the weighincr bottle and weighed. Constant weight was 

 obtained without difficulty. f 



In the following table are given the results obtained by this method of 

 procedure. The correction to a vacuum standard of +-000069 gram per 

 gram of cobaltous oxide was applied, the specific gravity of the oxide 

 being taken as 5.68 (Joule and PIayfair).$ = 16.000. 



* These Proceedings, XXXIII. 120 (1897). 



t Sintered cobalt does not occlude important amounts of hydrogen, even when 

 cooled in this gas. A paper showing this fact will soon be published. 

 I Landolt and Bi'.rnstein, Tabellen (189i), p. 13i. 

 vor-. XXXV. — 5 



