360 PROCEEDINGS OF THE AMERICAN ACADEMY. 



cobalt was undoubtedly true from the fact that they were colored a pale 

 blue both after evaporation and after heating. During the heating 

 Iiowever a slight blackening took place which was due to the oxidation of 

 traces of unoxidized cobaltous hydrate. In order to make sure that all 

 unreduced bromides had been renaoved from the cobalt by the process of 

 leaching, in each analysis the metal was dissolved by cold very weak 

 nitric acid* and treated with silver nitrate. As not even the faintest 

 cloudiness was ever again visible in these solutions, it was assumed that 

 the soluble matter had been completely removed. 



In analysis 5 the bromine in the residue was determined and the result 

 found to be considerably too low to correspond with the weight of the 

 residue, if calculated as sodium bromide. In order to discover the cause 

 of this discrepancy an elaborate series of experiments was carried out at 

 the expense of much time and labor. To describe these experiments in 

 full would result only in confusing the mind of the reader. It is sufficient 

 to say that 2:)ure spongy cobalt was treated with varying amounts of pure 

 sodium bromide in solution, the conditions being regulated so as to be as 

 nearly as possible like those in the analyses. The following conclusions 

 drawn from these experiments are of great importance. 



In the first place, no bromine is lost by the residues either during 

 evaporation or in heating to 130°. The cobalt in the residues may 

 be present in three forms, as cobaltous hydrate which has escaped oxi- 

 dation, as cobaltic hydrate, and as unreduced cobaltous bromide,t each 

 in exceedingly small amounts. The doubt as to the quantity of each 

 present makes it impossible to apply the correction for dissolved cobalt 

 with any degree of accuracy and in the table of results no attempt has 

 been made to do so. In the later analyses this cause of uncertainty was 

 removed. t This correction within a correction is however an infinitesi- 

 mal of the second order ; neglecting it can produce no serious effect upon 

 the accuracy of the final result. § In this respect the research upon 

 cobalt differs from that upon nickel, where the amount of hydroxide in 

 the residue was relatively great, owing to the fact that it had not been 

 chiefly eliminated by oxidation during the evaporation. 



* In a platinum dish this solution takes place with great ease, and with no danger 

 of a loss of bromine. The galvanic action is a great assistance. 



t Cobaltous bromide could only have found its way into this residue by having 

 been protected from reduction by enclosure in crystals of sodic bromide. It must 

 have been exceedingly small in amount, if present at all. 



f Compare page 3G3. 



§ This extra cobalt may partly explain why the sum of the total analysis slightly 

 exceeds 100.000 per cent. See page 365. 



