66 PROCEEDINGS OF THE AMERICAN ACADEMY. 



From the fact that the above results agree neither among themselves 

 nor with the results obtained from the bromide (58.995), it seemed highly 

 probable that all excess of oxygen had not been removed from the 

 cobaltous oxide. It has been shown that oxides made by ignition of 

 nitrates give off their extra " included " oxygen only upon long continued 

 heating, the gas gradually working its way out by a process of dissocia- 

 tion and recombination.* Here also the oxygen combined with the 

 innermost parts of the monoxide, too tightly enclosed to escape easily, 

 must have required time to force its way out by the same process. 



Accordingly, in the next analysis the oxide was heated for much longer 

 periods in the highest vacuum obtainable. Under these conditions the 

 loss of weight between each two ignitions increased very materially, 

 amounting to more than a milligram in each case, and culminated in a 

 loss of three milligrams accompanied by a decided darkening of the 

 oxide. This called to mind an experiment made by Dr. Cushman in his 

 work upon the atomic weight of nickel. f In this experiment he ignited 

 nickelous oxide to bright redness in a current of nitrogen. Upon 

 removal from the ignition tube the oxide was found to have been partially 

 converted into metallic nickel, owing doubtless to the dissociation of the 

 oxide into metal and oxygen, and the subsequent carrying away of the 

 oxygen by the current of nitrogen. Cupric oxide, when heated only to 

 700°-800°, is known to dissociate perceptibly into cuprous oxide and 

 oxygen. $ As a matter of fact, when the cobaltous oxide was removed 

 from the boat, it was found to have been reduced to such an extent that 

 enough metallic cobalt had been formed to alloy with the platinum ; thus 

 cobaltous oxide also, at about 800°, dissociates very considerably into 

 cobalt and oxygen. In the zeal to obviate oxidation, the opposite error 

 had been encountered. Obviously the only method of obtaiuing the 

 monoxide free from the metal was to determine the tension of the oxygen 

 evolved from it at the temperature to which the oxide was heated, and 

 always to keep the tension above this point during the ignition. 



In the earlier experiments, an exceedingly slight known leakage of 

 water through the brass stoppers into the tube may explain why the 

 reduction of the monoxide did not occur, for metallic cobalt decomposes 

 water at a high temperature. 



Some of the brown monoxide which had suffered partial reduction in 

 the experiment described above, and which consequently must have been 



* T. W. Ricliards, These Proceedings, XXXIII. 399. 

 t These Proceedings, XXXIII. 424. 

 J Ibid., 421. 



