344 PROCEEDINGS OF THE AMERICAN ACADEMY. 



method of little value, so that the result (58.77 ?) does not carry with it 

 much weight. 



Tiiree chemists in three different decades, Marignac, Baubigny, and 

 Schiitzenberger,* have attempted to solve the question by the quantita- 

 tive ignition of nickelous sulphate. The three investigations agreed 

 fairly well upon an average result, 58.71,t Baubigny's being by far the 

 most satisfactory. This method is one involving two errors which 

 nearly counterbalance each other: — the sulphate has a tendency to 

 retain water, while the oxide almost invariably retains sulphuric acid. 

 For this reason, the method gives results which approximate closely to 

 the truth.! Here again we have a support for the conclusion that the 

 value iu question cannot be far from 58.7. 



All the published work upon the subject has now been referred to 

 except some early work of Eothoff, Erdmann and Marchand, and 

 Deville § (which deserves no more than a passing mention), and the 

 more recent researches of Winkler. Kriiss's misguided work has been 

 sufficiently dissected by Winklei''s able but unsparing criticism || and in 

 the experimental part of this paper. The only points not covered by 

 Winkler, — the rose-colored flame test and the solubility of nickelous 

 hydrate, are explained in the foregoing pages. 



The work of W^inkler is surprising in its variety and in the ingenuity 

 of his methods, but unfortunately it is equally surprising in the wide 

 range of one per cent between his several results. His earliest work,** 

 depending upon the reduction of sodic aurochloride by nickel, giving the 

 extremely high value 59.45, is obviouslj' at fault. Winkler himself 

 ignores it in his discussion,tt so that further criticism of it may well 

 be omitted. 



Winkler's two later investigations, carried out only a few years ago, 

 gave results much lower and more satisfactory. In his first revision he 

 weighed nickel, converted it into chloride, and determined the chlorine 



Greatest Difference 

 from Mean. 



* (1858) Marignac, Arch. Sci. Nat., (nouv. ser.), I. 374, Ni = 58.70 ± 0.15 

 (1883) Baubigny, Compt. Rend., XCVII. 951, Ni = 58.73 ±0.002 



(1892) Schiitzenberger, Compt. Rend., CXIV. 1149, Ni = 58.65 ± 0.075 



t Clarke, Recalculation, top of page 302. 



I See " A Table of Atomic Weiglits," These Proceedings, XXXIII. 297, 298. 

 § See Clarke, Recalculation, p. 291. 



II Zeitschr. Anorg. Chcm., IV. 10. 



** 1867. Zeitschr. Anal. Chem., YI. 18. Ni = 59.45 (Clarke). 

 tt Zeitschr. Anorg. Chem , IV. 10, VIII. 1. 



