128 PROCEEDINGS OF THE AMERICAN ACADEMY. 



ing ten results, the second three, and the third, and fourth each two 



results. 



Sample I. 58.987 



Sample II. 58.992 



Sample III. 58.995 



Sample IV. 59.004 



In this case, as in the case of nickel, a slight rise in the atomic weight 

 is to be observed with the increasing jaurity of the materials. Although 

 larger here than before, the rise cannot be considered greater than the 

 probable experimental error, — especially since one or two of the indi- 

 vidual results obtained from Sample I. were slightly greater than any 

 obtained from Sample IV. We must therefore conclude that, if " gno- 

 mium " exists, it must have an atomic weight about equal to that of 

 nickel and cobalt, and hence that the wide variations to be observed in 

 the results of other experimenters cannot be considered a valid argument 

 in favor of the late Professor Kriiss's doubtful discovery. 



According to the present investigation, then, the atomic weight of 

 cobalt seems to be very close to 59. Upon comparing this result with 

 the earlier ones of other experimenters, we see that this result is as 

 much in accord with their general verdict as almost any other would be. 

 In other words, the values are so various in magnitude that no satisfac- 

 tory conclusion can be drawn from them. It is worth noticing that one 

 of the results obtained by Thiele, under Professor Hempel's direction, 

 agrees exactly with ours, however. 



Our work thus has this outcome, at its present stage : — 



If Oxygen — 16.000, Cobalt = 58.99. 

 If Oxygen = 15.88, Cobalt = 58.55. 



Cambridge, Mass., June, 1897. 



