ATOMIC AVEIGHTS 



443 



In a second memoir/ Richards and Cnslmian describe a series of de- 

 terminations based upon the reduction of nickel bromide by heating in 

 hydrogen. The corrected data appear in the next table : 



Mean, 26.855, ± .0005 



Hence Ni = 58.685. 



In this series a correction was applied for traces of sodium bromide 

 contained in the nickel salt. A similar correction, applied to the former 

 series of determinations, would raise the atomic weight of nickel by 0.015. 



The three memoirs upon cobalt, by Richards and Baxter,^ contain 

 data relative to the bromide, the chloride and the oxide. It is hardly 

 necessary to state that all of the materials employed in the investigation 

 were scrupulously purified, and that all weights were reduced to a 

 vacuum basis. First, as in the case of nickel, the two silver ratios to 

 the bromide were determined. A preliminary set of analyses gave results 

 as follows: 



Mean, 58.252. ± .0040 



The second and third series of analyses gave both ratios, and may be 

 tabulated together : 



^ Proc. Amer. Acad., 34, 327. 1899. This memoir contains a very full criticism of all the earlier 

 work on nickel. 



"Proc. Amer. Acad., 33, 115. 1897. Ibid., 34, 351. 1899. IMd., 35, 61. 1899. For a criticism 

 of Richards, Cushman and Baxter, see Winkler, Zeitsch. anorg. Chem., 17, 236. 1898. 



