422 PROCEEDINGS OF THE AMERICAN ACADEMY. 



repeated analysis showed to be almost pure oxygen containing only 

 traces of nitrogen and carbon dioxide) was found to possess a tension 

 of two or three tenths of a millimeter of mercury. A portion of the same 

 material was heated for many hours more in a crucible to a full red heat ; 

 but upon subsequently heating it in a vacuum the evolution of gas pro- 

 ceeded just as before. Since all the occluded oxygen except two or three 

 per cent had been expelled by the preliminary heating, this must have 

 been structural oxygen. The conclusion was confirmed by the discovery 

 of red cuprous oxide in the residue. Repetitions of the experiment showed 

 that the tension of the gas increased enormously with the temperature, 

 exceeding a millimeter before the glass became so soft as to collapse. 

 In some of the experiments the best Bohemian glass was used, in others 

 the hardest made in Jena. The nature of the glass seemed to be with- 

 out influence on the result. The amount of gas evolved in a given 

 time naturally depends upon the efficiency of the pump, as well as upon 

 the temperature ; in one of the experiments two cubic centimeters of 

 oxygen were obtained in eight hours. 



Ignition of the substance in an inert atmosphere should j^roduce the 

 same effect. Since cupric oxide is slightly dissociated by heat, percepti- 

 ble amounts of oxygen should be removed by heating it in nitrogen, just 

 as carbonic acid is removed from limestone by heating it in a current 

 of air. This dissociation of cupric oxide must have its effect on any 

 process involving the ignition of cupric oxide in a vacuum or in an inert 

 gas. The determination of organic nitrogen by means of the Sprengel 

 pump, for example, must be affected by it. The use of carbon dioxide 

 as a displacing medium in the Dumas method probably disposes of the 

 error, however; for carbon dioxide is itself dissociated by heat, and 

 it undoubtedly furnishes enough oxygen to diminish greatly the decom- 

 position of the cupric oxide. 



In the light of these results, it is obviously advisable, when one desires 

 to remove oxygen from a mixture of gases by means of hot copper, 

 to maintain the last portion of the tube at a comparatively low tempera- 

 ture. Even then the issuing gas may not be absolutely free from 

 oxygen.. 



Every one knows that auric oxide is excessively unstable, and that 

 argentic oxide is completely decomposed even in an atmosphere of pure 

 oxygen at comparatively low temperatures. The present results merely 

 extend this behavior, in a much smaller degree, to cupric oxide, and the 

 hypothesis carries the idea still further. The dissociation of zincic oxide 

 must of course be yet less, while the tension of oxygen above magnesia 



