70 



PROCEEDINGS OF THE AMERICAN ACADEMY. 



Jour. Chem. Soe. (2), I. 51. 58.7 



Sitz. Wien. Akad., LIV. (11.), 50. 60. 



Zeit. Anal. Chem., VI. 18. 59.4 



Jour. Chem. Soc. (2.), VII. 294. 59.1 ? 



Ber. d. d. Chem. Gesell , II. 592. 59.1 



Am. Jour. Sci. & Arts, (3.), II. 44. 59.2 



Ann. der Chem., CCXXXII. 321. 58.9 



Ber. d. d. Chem. Gesell., XXII. 11. 



XXII. 891. 



Zeit. Anorg. Chem., II. 221. 58.8 



Compt. Rend., CXIV. 1149. 60.1 



Zeit. Anorg. Chem., IV. 10. 59.8 



" IV. 462. 59.5 



" VIII. 1. 



" VIII. 291. 



" XL 73. 58.9 



" XVI. 362. 59.0 



" XVIL 236. 



Proc. Amer. Acad., XXXIV. 351. 59.0 



The present paper. 59.0 



All except four of these investigations were acconnpanied by similar 

 ones upon the atomic weight of nickel, and in general the criticisms made 

 by Richards and Cushman in a recent paper ui)on this subject may be 

 applied unchanged to the jiresent argument. It is only necessary to call 

 attention here to the features in which diftereuces existed. 



The favorite method for the determination of both constants has been 

 the reduction of the oxides. It is evident from our work upon this 

 method that Russell's cobaltous oxide (1863) must have been contami- 

 nated with a higher oxide; that Zimmerraann's material (1886), treated 

 with greater care in more perfect apparatus, was more nearly normal, 

 although still impure ; and that Schiitzenberger (1892) must have driven 

 off some of the oxygen, which should have been weighed, by his employ- 

 ment of a high temperature. It was reserved for Herapel and Thiele 

 (1895) to discover by varying the conditions of work that cobaltous 

 oxide was incapable of giving constant results, although they did not 

 attempt to find the reason of the inconstancy. Their irregular observa- 

 tions were really a proof of the thoroughness and care of their work, 

 while the wonderful constancy of Zimmermanu's results simply shows a 

 constancy of imperfect conditions. 



The far greater dissociation tension of the higher oxide of nickel makes 

 the method much less dangerous in this case ; hence the similar results 

 for the atomic weight of nickel are both more concordant and more nearly 

 correct than those for the atomic weiirht of cobalt. 



