RICHARDS AND BAXTER. ATOMIC WEIGHT OP COBALT. 365 



resembled those used ia Series II. and III. of last year's work) were 

 found to contain 0.02327 gram of impurity. If, as we may reasonably 

 suppose, all the residues obtained from this sample of material resembled 

 that found in analysis 5, Series V., this residue would have yielded 

 0.0372 gram of argentic bromide. We may now correct last year's 

 results by subtracting from the several weights of cobaltous bromide 

 proportional weights of impurity, and also subtracting from the several 

 weights of argentic bromide amounts of this substance corresponding to 

 the impurity. Making this correction, the atomic weight of cobalt would 

 be loioered 0.008, the averages of Series II. and III. becoming 58.987 

 and 58.979. Our uncertainty regarding the nature and amount of the 

 impurity thus involves an uncertainty of about one part in six thousand in 

 last year's results. In the light of all the circumstances, it is perhaps 

 safest not to attempt any correction of these values, but to accept them 

 uncorrected as subject to this possible error. The results are accord- 

 ingly giveu below in an uncorrected form. 



The data just discussed obviously afford a basis for recording the total 

 percentage composition of the cobaltous bromide analyzed. 



COMPLETE ANALYSIS OF COBALTOUS BROMIDE. 



Based upon Series II., III., and V. 



Per cent. 



Cobalt (Series Y.) = 26.923 



Total Bromine - 



' (II.) 73.050% ' 

 (III.) 73.053% 



Aver. 73.051% 



Bromine combined , -r. «^, 



J . , „ , , y— 72.981 



with Cobalt 



/ Bromine actually [ ^y^. 



r found in impurity C 



i Remainder of -s 



Total impurity . . — 0.103% =z^ impurity '= .033 



( (chiefly Sodium) j 

 Total r= 100.007 



Series VL, perhaps the best of all, is not included in this table because 

 the material used iu it was not quite identical with that used in the three 

 other series. Obviously it is possible to calculate two more ratios in- 

 volving the atomic weight of cobalt, in which the values for the bromine 

 in the bromide are compared with the cobalt found in it. Into this calcu- 

 lation the weio^ht of the bromide itself enters simply as a constant, and 

 an indifferent impurity (such as water) would be eliminated from the 



