RICHARDS AND CUSHMAN. — ATOMIC WEIGHT OF NICKEL. 341 



It has been already pointed out that this impurity of ^'^j per cent of 

 sodium would make no difference in last year's results if the " equivalent " 

 of sodium equalled that of nickel. Since, however, it is somewhat less, 

 slightly too much argentic bromide was obtained last year, and the 

 atomic weiglit of nickel appeared lower than it really is. Assuming the 

 amount of impurity to have been the same in last year's preparation as 

 in this, 0.0 15 should be added to the atomic weight, in order to correct 

 this error. The results of the four series, which represent the sum and 

 substance of the present research, are then as follows. Ratios (e) and 

 (/) were obtained by cross-reckoning from the earlier ratios. The 

 reason for thus restating the results is because this restatement uses the 

 weight of the nickelous bromide only as a constant, and not as a basis of 

 calculation. 



59.706 



If any assurance is needed that this average indicates very nearly the 

 true atomic weight of nickel, the assurance may be found in the review 

 of older work which follows. This review has been postponed until the 

 close of the paper, in order that the methods might be judged in the light 

 of our own experience with the subject. 



A Brief Criticism of Earlier Work. 



The atomic weights of nickel and cobalt have each been the subject of 

 a score of different researches since the i^roblem was first attacked by 

 Rothoff in 1818. These investigations have not only led to exceedingly 

 discordant results, but have also given rise to several interesting and 

 important controversies. As a chronological list is given in our former 

 papers,* there is no need of repeating it here. For the purpose of 



of finding the exact amount of the sulphuric acid, but in any case it was so small 

 as to produce only a negligible effect on the result. From some experiments of 

 Mr. Baxter's it is safe to assume that this error could not have exceeded 0.005 per 

 cent in tiie worst cases, and in the average it must be still much less. 

 * These Proceedings, XXXIII. 07, 115. 



