RICHARDS AND BAXTER. ATOMIC WEIGHT OF COBALT. 367 



porcelain tubes might have been attacked during the sublimation of the 

 bromides, with the introduction of foreign bromides into the nickelous 

 and cobaltous bromides. He goes on to state that the bromides were 

 dried in an acid atmosphere, and probably retained hydrobromic acid 

 after this gas had been displaced by air. His third criticism is that the 

 nickelous bromide in the earlier analyses contained nickelous oxide 

 which had to be determined and subtracted ; and his final objection 

 applies to the use of the Gooch crucible. 



Answers to the greater part of his criticisms can be found in the very 

 articles which he criticises. In one case only can his view be substan- 

 tiated; — the porcelain tubes are really attacked. That this flaw was 

 a possibility we realized at the time ; but we also realized the smallness 

 of the error introduced by even a comparatively large amount of such 

 impurity. This matter has been already discussed in detail, both in this 

 paper and in the paper upon the atomic weight of nickel. 



There are two possible ways in which hydrobromic acid could have 

 been retained: — by adsorption and by inclusion. At the high tempera- 

 tures employed the adsorption must have been very slight, and the 

 long process of washing witli an indifferent gas was favorable to the 

 elimination of any tendency in that direction. "While the inclusion of 

 liquids is a very serious possible cause of error, that of gases is usually 

 negligible because of the small mass involved. For this reason crys- 

 tallization from solutions is far less satisfactory than sublimation as a 

 means of purification. 



That as a matter of fact our bromides were neutral there is no lack 

 of evidence. The possibility of acidity had occurred to us also, but rea- 

 soning from analogy we had decided that this possibility was rather an 

 improbability. Bromides and chlorides of barium and strontium, heated 

 in the same way in a dry acid atmosphere, after the acid has been dis- 

 placed by dry air, give absolutely neutral reactions with methyl orange.* 

 With cobaltous bromide the end point is not as easy to detect as with the 

 before mentioned halides on account of the color of the dissolved salt, 

 but colorimetric comparison makes it possible to distinguish the change 

 very accurately. A solution of our cobaltous bromide containing methyl 

 orange perceptibly changed color upon the addition of the minimum 

 amount of hundredth normal acid necessary to produce a change of color 

 in pure water containing methyl orange, showing that the salt must have 

 been very near if not at the turning point. As a final test, potassium 



* These Proceedings, XXIX. 59, XXX. 373. 



