364 



PROCEEDINGS OF THE AMERICAN ACADEMY. 



crepaucy are due to the presence of all or at least several of the above 

 mentioned acids combined with sodium, each iu quantity too minute for 

 detection ; and for the present this will have to rest as the explanation. 



Sixth Series. CoBr.i : Co. 



Average of Series V. and VI, 



Co = 58.998. 



This final average differs from that published before by only about 

 one part in ten thousand ; but iu comparing the two one must remember 

 that the material used last year must have been contaminated by the 

 same impurities which have been discussed in this paper. If the im- 

 purity contained as much bromine as cobaltous bromide contains, it would 

 have had no effect upon last year's results. In the case of nickel, where 

 the impurity consisted wholly of sodic bromide, the effect of correct- 

 ing the observed results in the paper of 1897 was to raise the atomic 

 weight of nickel from 58.688 to 58.703. The impurity from our cobalt- 

 ous bromide, on the other hand, contained unknown substances in quau- 

 tities so small as to elude detection, but large enough to change the 

 sign of the corresponding correction. Thus 22.63 grams of cobaltous 

 bromide in Series V. (the series in which the materials most nearly 



