ATOMIC WEIGHTS 



421 



A few determinations of the atomic weight of iron by Winkler ' still 

 need to be mentioned, not as directly significant, but as relating to the 

 validity of a method which he applied to nickel and cobalt. Iron, not 

 absolutely pure, was dissolved in a solution of iodine and potassium iodide. 

 The quantity of iodine was known, and after the reaction ended the 

 amount unconsumed was measured by titration with thiosulphate solu- 

 tion. A ratio between iodine and iron was thus determined, which can 

 be expressed as lo : Fe : : 100 : x. Two series are given, one with iron 

 cleaned by scrubbing, the other with iron which had been heated in 

 hydrogen. The weights of iron given below are corrected for known 

 impurities. 



Mean, 22.149, ± .0012 



Hence Fe = 56.323. 



Mean, 22.136, ± .0017 



Hence Fe=: 56.190. The weighted mean of both series is 22.145,± 

 .0010, which gives Fe = 56.213. This value is high, and so are the 

 values found for cobalt and nickel by the same method. The process 

 is probably affected by serious constant errors, and the results obtained 

 by it are not good. For comparative purposes, however, the iodine ratio 

 is included in the following tabulation of ratios : 



(1). FeA:2Pe: : 100: 69.9728, ± .00083 

 (2). 2Ag:FeBrj: : 100: 99.960, ± .0027 

 (3). 2AgBr:FeBr,: : 100: 57.4195, ±: .00044 

 (4). 2Ag: Feci. :: 100: 58.866, ± .0530 

 (5). 3Ag:FeCls:: 100: 50.2435, ±.0132 

 (6). l2:Fe::100:22.145, ±.0010 

 (7). H:Fe: :1:55.608, ± .0532 



^Zeitsch. anorg. Chem., 8, 29]. 1895. 



