EICHARDS AND BAXTER. ATOMIC WEIGHT OF COBALT. 69 



ous chloride lias no immediate effect on hydriodic acid. The presence 

 of the imprisoned higher oxide is undoubtedly the reason i'or the low 

 value for the atomic weight of cobalt obtained in this way. 



The cobalt itself obtained from the pure oxide by reduction at a high 

 temperature is very permanent in the air, and occludes only traces of 

 hydrogen. 



With o-reat reluctance the work upon the monoxide was abandoned at 

 this point; for a substance whose composition varies so widely under 

 conditions which vary so sligiitly is obviously unfitted for woi-k of the 

 hio-hest accuracy. As in the case of cobaltous bromide, the preparation 

 of this substance even in a reasonably pure state appears to be an impos- 

 sibility. Unfortunately the accurate estimation of the worst impurity in 

 this case, cobakic oxide, is a very difficult matter. An attempt to use 

 pure cobaltic oxide itself as the basis of the determination proved to be 

 equally unsatisfactory. 



The results of this part of the investigation are nevertheless of con- 

 siderable import in the light they throw upon previous analyses of the 

 monoxides of both nickel and cobalt as well as upon the accuracy of the 

 results obtained in the analyses of cobaltous halides described in this 

 paper. The material which gave the lowest result contained oxides of 

 cobalt higher than the monoxide, hence these results are too low. On 

 the other hand, the material which gave the highest result was heated 

 under conditions which theoretically should have caused, and evidently 

 did cause, partial reduction of the oxide to metal. The true value of the 

 atomic weight of cobalt must lie between these two extremes, 58.93 and 

 59.07 ; and until trustworthy evidence to the contrary has been produced, 

 the most probable value seems to be the mean 58.995, obtained from the 

 analysis of the bromide. 



Critical Review of Earlier "Work. 



In the light of the information gained during the protracted investiga- 

 tion which has just been described, it; is possible to discover the reasons 

 for many of the unusual discrepancies in the older work upon the atomic 

 weight of cobalt. Following is a chronological list of this work, with 

 references to the original articles. The atomic weight of oxygen, upon 

 which the appended values are based, is taken as 16.000. 



Pogg. Ann., VIII. 184. 59 



Pogg. Ann., CI. 387. 60.0 



Bibl. Univ. fie Geneve. I. 372. 58.8 



Ann. Cheni. Pharm., CXIII. 25. 59.1 



