RICHARDS AND CUSHMAN. — ATOMIC WEIGHT OP NICKEL. 347 



remain in solution for twenty-four hours after weighing and before 

 titration, and when this circumstance is complicated by the presence 

 of a metal capable of acting to a slight extent even upon pure water 

 in the presence of air, one can hardly contend that the conditions are 

 the best. The chance of side reactions seems to be too great to admit 

 of infallibility in the results. One is surprised, indeed, that Winkler's 

 results approach as near to those of Zimmermann as they do, and this 

 close approach is evidence of great accuracy of manipulation on Wink- 

 ler's part. In short, viewed from the standpoint of ordinary analytical 

 experience, Winkler's last work is admirable, while from the standpoint 

 of atomic weight research it is inadmissible. In justice to Professor 

 Winkler it is only fair to add that he realizes this fact himself.* One 

 need not dwell upon possible inaccuracies, however; for Winkler himself 

 has furnished us with data for computing the error of his method. In a 

 short paper he uses the same method for determining the atomic weight 

 of iron, and finds for this quantity the value 5G.174, if O = IG.OOO.f 

 Now according to the fairly consistent work of Berzelius, Erdmann and 

 Marchand, Svauberg and Norlin, and Maumene, the atomic weight of 

 iron cannot be far fi-om 56.02 ; and there is no contradictory evidence 

 of serious value.t Winkler's method then gave him a result 0.275 per 

 cent § too high in the case of iron, and it is fair to conclude that the 

 error could not have been far different in the case of nickel. Making 

 the corresponding subtraction, Winkler's corrected result approaches 

 astoundiugly near to those obtained by Zimmermann and by us.|l 



Winkler's corrected value 58. G9 



Zimmermaun's value 58.694 



Richards and Cushman's value 58.706 



Average 58.70 



Owing to a slight uncertainty in the atomic weight of iron, as well as 

 to the possibility that iron may behave somewhat differently from nickel 



* Zeitschr. Anorg. Chem., XVII. 239. 



t Zeitschr. Anorg. Chem., VIII. 291. 



} Clarke's recalculation, p. 289. The atomic weight of iron is now being further 

 studied in this Laboratory. 



§ It is possible that a small part of this error is due to the omission of the 

 reduction to the vacuum standard, which would afiect the final value by about 

 0.01 per cent. This correction may have been applied, but there is no evidence 

 of such application. 



li Mr. Baxter first called oiir attention to this remarkable unanimity. 



