RICHARDS AND CUSHMAN. — ATOMIC WEIGHT OF NICKEL. 329 



dently this silica had prevented the cohering or " sintering " of the metal 

 during its original reduction from the oxide, and hence caused the pul- 

 verized state of the impure metal, — a fact which is interesting as 

 showing the great change in properties produced by a small amount of 

 impurities. 



These two experiments emphasize the well known facts that glass is 

 wholly unsuitable for accurate work, and that material prepared even in 

 platinum is extremely difficult to I'ender wholly free from silica, unless 

 it is vaporized.* While however our purest nickel sometimes contained 

 traces of silica, the bromide prepared from it by sublimation was un- 

 doubtedly as free as possible from this impurity. Further light upon 

 this question will be given in a following paper upon cobalt. 



Although a possible contamination with silica was thus little to be 

 feared, some other constituents of glass or porcelain were much more 

 dangerous. All the material actually used in our analyses had been 

 prepared wholly in platinum vessels at every stage excepting at the very 

 end, when it had been sublimed in a porcelain tube. Obviously this 

 tube might be attacked by the hot mixture of hydrobromic acid, bromine, 

 and nickelous bromide vapor ; but since nothing beside sodic bromide 

 would probably sublime with the nickelous salt, and the "equivalent" of 

 sodic bromide is almost equal to that of nickelous bromide, the slight 

 impurity could produce no important effect upon our last year's work. 

 This was realized at the time ; and the discovery of the presence or 

 absence of this error was one of the prearranged objects of the present 

 paper. Since the work recorded below was completed, Professor Wink- 

 ler, by a kind personal letter as well as by a recent article, wisely called 

 attention to this flaw ; and it will be seen that his objection had been both 

 substantiated and answered before he wrote about it. 



The easiest method of detecting sodic bromide in nickelous bromide 

 is obviously to reduce the latter and then to extract the former with 

 water from the spongy metal. Moist hydrogen easily divorces the halo- 

 gen from its none too stable metallic union at a temperature of not much 

 over 300°, at which temperature sodic bromide is essentially non-volatile. 

 In this way repeated experiments showed that all our nickelous bromide 

 had contained on the average not far from one tenth of one per cent of 

 sodic bromide. The particulars concerning the determination of this 

 serious impurity naturally form an essential point in the method of anal- 

 ysis of the nickel salt, hence they will be found later under that head. 



* Compare Stas's " Untersuchungen " (Aronstein), pp. 269 and 279. 



