RICHARDS AND BAXTER. — ATOMIC WEIGHT OF COBALT. 363 



It seemed highly important to us at this point to prepare cobaltous 

 bromide which should be free from every impurity. In the first place 

 the drying apparatus was slightly modified, no strong sulphuric acid 

 being used except in drying the air necessary for sweeping the nitrogen 

 out of the weighing bottle. Dilute sulphuric acid was substituted in 

 every case, and two columns of stick potash followed by one of phosphoric 

 anhydride were inserted beyond this dilute acid. As the porcelain tubes 

 had evidently been the source of the alkaline impurities found, a platinum 

 lining, made by bending a large piece of platinum foil into the form of a 

 cylinder, was provided for the outside porcelain tube.* The smaller 

 porcelain tube was not used at all, the sublimed matei'ial being removed 

 by taking out the foil and unfolding it. Cobalt from Sample III. was 

 then sublimed in the remodelled apparatus from a platinum boat. The 

 material obtained in this way gave results for the atomic weight alto- 

 gether too high, a circumstance due to large quantities of ^ilatinum 

 actually found in the sublimed bromide. Even here a small amount of 

 alkaline impurity existed, having crept in through the crack in the pla- 

 tinum foil. Hence, no more work was done with this material ; but 

 renewed precautions were taken to prepare by the older method cobaltous 

 bromide which should contain the smallest possible number and quantity 

 of impurities. 



Four analyses were made with this new material, which proved in 

 spite of all our care to contain as much soluble matter as before. The 

 water extracts from these analyses were evaporated in a flat platinum 

 dish, which exposed a large surface of the solution to the air. This served 

 to oxidize completely the dissolved cobaltous hydrate, for the residues did 

 not become gray when heated, and upon the addition of water gave per- 

 fectly clear solutions. These residues were faintly blue, the color being 

 due doubtless to a trace of unreduced cobaltous bromide. For some un- 

 discovered reason, the amount of silver bromide obtained from the 

 residues was still too small to correspond to the weight of the residue, 

 if calculated as sodic bromide. 



Tests were repeated for sulphuric, phosphoric, and silicic acids with 

 the greatest possible care, but still with negative results. In one analysis 

 the cobalt was determined in the residue and found to be only 0.00013 

 gram. It is possible, however, that these few tenths of a milligram dis- 



argentic bromide. Hence 0.0010 gram of argentic bromide corresponds to 0.000626 

 gram of residue. This factor is used in calculating the results in Series I. 

 * Compare the preceding paper on Kickel, page 331. 



