290 PROCEEDINGS OF THE AMERICAN ACADEMY 



Upon referring to the other work upon the atomic weight of copper, 

 it is apparent that nearly all of the discrepancies have been explained. 

 A portion of Millou and Comaille's work consisted of weighing the 

 water formed by the reduction of cupric oxide ; the reason of the low 

 result of this part of their work is not even now evident, but at the 

 present day a discussion of the possible causes of its error would be 

 uhprofitable. Shaw's recent result, which formerly ranked among the 

 highest of the observed values, now appears among the lowest. 



Evidently the occlusion of nitrogen by cupric oxide must have a 

 very serious effect upon experiments in which oxygen is determined 

 by the loss of weight of material prepared by the ignition of the basic 

 nitrate ; notably in the determination of the atomic weight of the 

 element after the method of Dumas. When the oxide is obtained 

 by the ignition of the metal in oxygen, the correction is of course in- 

 applicable. Unfortunately, most of the experimenters upon the subject 

 have omitted to state the source of their cupric oxide. The prepara- 

 tion used in the later experiments of Erdmann and Marchand * must 

 have contained nitrogen ; and it will be remembered that the second 

 series of his results gave a much higher result than the first. 



In view of the correction, it is remarkable that the resultant atomic 

 weight of oxygen in this case was not even higher than 16 ; for the cor- 

 rection is about five times as large as the amount necessary to account 

 for the diflTerence between 15.87 and that figure. From a difference 

 either in the mode of preparation or in the temperature of ignition, the 

 oxide used by Erdmann and Marchand must have cont-ained much 

 less nitrogen than the amount found in Experiments 103 to 105; or 

 else some opposite error must have partially counterbalanced this one. 

 Considering the remoteness of all the experiments, a present discussion 

 of the results is unnecessary ; but the conclusions are at least sufficient 

 to throw a serious doubt upon the applicability of cupric oxide for a 

 quantitative source of oxygen, as well as to support the modern low 

 atomic weight of that element. The presence of nitrogen could not 

 seriously influence the results of Keiser or those recently obtained 

 here ; and it is evident that the material used by Noyes could have 

 contained no occluded gas. 



The use of cupric oxide prepared from the nitrate must also intro- 

 duce a serious error into organic nitrogen determinations after the 

 method of Dumas, as Frankland and Armstrong have already pointed 



* J. fiir prakt. Chemie, XXVI. 461. 



