EICHARDS. — RETENTION AND RELEASE OF GASES. 409 



ences the relatiou of the vohimes of oxygen and nitrogen retained, or 

 else that the total amount of the enclosed gases is not a safe measure 

 of the temperature. It will be shown that both of these inferences are 

 correct, and hence that the logic of Morse and Arbuckle's point, correct 

 in itself but unsafe in its premises, falls to the ground. 



A careful study of all the data, Morse and Arbuckle's as well as ours, 

 convinced me that, while the two gases must be evolved together, they 

 behave subsequently in quite a different fashion. It is apparent imme- 

 diately that the oxygen must escape at first much faster tban the nitro- 

 gen, for the ratio of the two volumes in the nitric anhydride set free 

 must be originally 5 : 2, while the average of Morse and Arbuckle's 

 result showed that less than a fifth of the finally retained mixture con- 

 sisted of oxygen. Part of the loss might be due to the preliminary for- 

 mation of lower oxides, and to unequal diffusion, but this cannot account 

 for so great a loss of oxygen. Even nitrous oxide should yield upon 

 decomposition a mixture one third of the volume of which must consist 

 of oxygen ; and the fractional diffusion of this mechanically enclosed 

 mixture should allow the lighter of its two constituents, nitrogen, to 

 escape the more quickly of the two ; hence the mixture should contain 

 never less than thirty-three per cent of oxygen. The same anomaly 

 is even more manifest in the case of cupric oxide, for from this sub- 

 stance nearly all the oxygen soon escapes even at comparatively low 

 temperatures. 



Why then does the oxygen escape the more easily of the two gases ? 

 This question introduces us at once to the second section of the paper, 

 which deals with the composition of the occluded gases. 



The first step toward answering the question is a certain determina- 

 tion of the facts, which have been doubted by Morse and Arbuckle. 

 The three conditions which would be most likely to affect the amount 

 and composition of the gaseous mixture are the temperature of ignition, 

 the time of ignition, and the compactness of the solid material. In 

 order to test the effect of these possible modifying causes experimentally, 

 several series of analyses were made, both with zincic and with cupric 

 oxide. 



The zincic oxide was made by heating pure zincic nitrate for a long 

 time at 280°. It was more compact than the cupric oxide, and un- 

 doubtedly contained residual traces of nitrate until further ignited. 



