100 PROCEEDINGS OP THE AMERICAN ACADEMY. 



experiments carried out to test this point it appeared that freshly sub- 

 limed nickelous bromide exposed to the free air of the room absorbed 

 about 0.1 milligram per gram in ten minutes. 



Since the value of the selection of nickelous bromide as the starting 

 point in the determination of the atomic weight of nickel depends on the 

 possibility of dissolving the salt in hot water without the slightest loss of 

 bromine, it seemed necessary to investigate this point. To this end, a 

 mass of several grams of the sublimed compound was suspended in 

 a flask arranged in such a way that the atmosphere of the flask during 

 the solution could be swept through a bulb-tube containing a mixture of 

 potassic iodide and starch. Any formation of nickelous oxide or oxy- 

 bromide during the heating of the solution could have taken place only by 

 disengagement of bromine, which would have been shown by the blueing of 

 the indicator in the bulb. Not the slightest tinge of blue color appeai-ed 

 at any time during the experiment, in spite of tlie fact that the solution 

 of nickel bromide was finally subjected to prolonged boiling. That the 

 iodo-starch mixture was exceedingly sensitive to very minute traces of 

 free bromine was assured by experiment in the first place. This experi- 

 ment is detailed here, however, merely as corroborative testimony, and is 

 not advanced as being final on the subject ; the best evidence is yielded 

 by the quantitative results of our series of analyses. 



Being now sure that we could analyze the salt if we could obtain it in 

 a pure state, we next proceeded to determine whether it could be pre- 

 pared in a condition altogether suitable for weighing and analysis. The 

 first sublimations were performed in an ordinary combustion tube in an 

 atmosphere of carbon dioxide. These sublimations yielded in no case 

 material that was above suspicion, and none of it was used in our anal- 

 yses, the experiments being carried out merely as a study of the proper- 

 ties of the substance under examination. Since it seemed possible that 

 the carbon dioxi^ used might suffer partial dissociation at the high tem- 

 perature necessary for the sublimation, nitrogen was substituted for this 

 gas. After many experiments with nitrogen, both alone and when mixed 

 with bromine vapor, it was found best to carry on the sublimation of the 

 nickelous bromide in a stream of nitrogen mixed with hydrobromic acid. 

 The elaborate apparatus constructed by Mr. G. P. Baxter for supplying 

 suitable mixtures of any or all of these gases and vapors for preparing 

 and subliming the salt will be described in a later paper, upon the atomic 

 weight of cobalt. 



The temperature at which the salt sublimes lies not far from that at 

 which the hardest glass begins to soften, hence we found it advantageous 



