352 PROCEEDINGS OF THE AMERICAN ACADEMY. 



even if the material is finely divided. In the case of the oxide there is 

 no means of deciding whether or not the last traces of oxygen have been 

 removed, while in the case of the bromide a residue of halogen is easily 

 detected. The complete reduction of nickelous bromide described in the 

 preceding paper shows that some such operations are in fact possible. 

 It is easy to see how the increase in volume involved in oxidation might 

 cause a part of the metal to be permanently protected by the growing 

 coat of oxide ; on the other hand, an oxide or salt which is being reduced 

 leaves only its skeleton behind, the innermost meshes of which may be 

 penetrated by the reducing agent. 



Cobaltous bromide when heated in a current of dry hydrogen begins to 

 be reduced at a temperature of about 350°. At higher temperatures the 

 reduction goes on more rapidly, but is then accompanied by partial sub- 

 limation of the bromide. Schiitzenberger's * observations on the sublima- 

 tion of nickelous chloride were similar to these, but he seemed to think 

 that some strange compound must have been the medium of the change, 

 instead of realizing that in the presence of hydrochloric acid a trace of 

 the vapor of the metallic salt might easily exist even in an atmosphere 

 of hydrogen. 



If the hydrogen is moist, however, the action begins at a lower tem- 

 perature, about 250° ; but even under these conditions sublimation can be 

 completely avoided only with the greatest difficulty. The reduction of 

 nickelous bromide offered fewer difficulties ; this process could be effected 

 in a current of dry hydrogen, and without danger of sublimation of a 

 trace of the material. The fact that moist hydrogen is more efficient 

 than the dry gas is easily explained by the hypothesis that the oxide is 

 formed as the first step in the reaction. 



Purification of Materials. 



Pi'eparation of Cobaltous Bromide. — The cobaltous bromide used in 

 this work was prepared by methods similar to those used in our previous 

 investigation, and for details the previous paper should be consulted. 

 Pure metallic cobalt was heated in a current of pure bromine and hydro- 

 bromic acid, and the sublimed bromide was preserved in weighing bottles 

 contained in desiccators until used for analysis. Samples I. and II. were 

 essentially the same as in the earlier investigation, even greater pains 

 having been taken, however, in the purification of the reagents and 

 water used in their preparation. Sample III. likewise was purified by 



* Compt. Rend., CXIII. 177. 



