368 PROCEEDIiNGS OP THE AMERICAN ACADEMY. 



bromide was sublimed iu the apparatus which had been used for the prep- 

 aration of the cobaltous bromide. The sublimate was then heated in a 

 current of dry nitrogen and hydrobromic acid gas, and finally, when cool, 

 the nitrogen and acid were displaced by dry air, just as in preparing the 

 cobaltous bromide for analysis. This potassic bromide upon solution 

 gave an absolutely neutral reaction with methyl orange. Taking into 

 consideration these three points, one cannot believe that enough hydro- 

 bromic acid was retained to have had an appreciable influence on our 

 results. 



The third criticism, objecting to the ftict that in some analyses a small 

 amount of nickelous oxide was found in the bromide, is an unfortunate 

 one. A careful perusal of the work would have shown that only in the 

 preliminary series of results was this the case, and that this series does 

 not enter into the final average, although its results differ only by a very 

 small amount from those subsequently obtained with material free from 

 oxide. As far as the nickel is concerned, a conclusive proof of the ab- 

 sence of acid is afforded by this very fact that the earlier determinations, 

 in which it was necessary to filter off a residue of finely divided nickelous 

 oxide, gave results no higher for nickel than the later results.* Hence 

 the second and the third criticisms are obviously inconsistent with one 

 another. 



The advantaces of the Gooch crucible are too well known to need 

 mention. Professor Winkler's specific objection to the collecting of dis- 

 placed asbestos upon an ordinary filter affects only an amount of a few 

 tenths of a milligram ; and a i^roof that no error was introduced in this 

 way lies in the fact that in every case the amount of silver bromide found 

 agreed very closely with the amount of silver necessary to complete the 

 reaction, where all but a few tenths of a per cent of the silver was 

 weighed out, and the remainder was added volumetrically. The following 

 brief table will make this clear. 



Tlius, in the cobalt work, 18.1G302 grams of silver yielded 31.61642 

 grams of silver bromide, — a ratio of 57.448 to 100.000, — while in the 

 case of nickel, 15.51556 grams of bromide gave 26.67078 grams of silver 

 bromide, — a ratio of 57.444 to 100.000, while Stas found 57.445. 



From the cobalt work AgBr : Ag = 100.000 : 57.448 



From the nickel work AgBr : Ag = 100.000 : 57.444 



From Stas's work AgBr : Ag = 100.000 : 57.445 



* The slight colloidal solubilitj' of nickelous hydroxide was evidently destroyed 

 by the presence of large amounts of nickelous bromide, as one would expect. 



