mCHAEDS AND BAXTER. — ATOMIC WEIGHT OP COBALT. 357 



the silver alone could do no harm; but unfortunately it was attended by 

 a reduction of the nitric acid also. This was proved by passing the re- 

 sulting gas through a hot tube, when traces of ammonia were formed, 

 capable of easy detection by Nessler's reagent. We had come face to 

 face with one of those frequent cases where an attempt at purification 

 had introduced a flaw as serious as the one it eliminated. The very 

 common use of argentic nitrate as a means of purifying hydrogen is 

 obviously a pernicious one, if accurate results are desired. 



The hydrogen apparatus was then entirely remodelled. Owing to the 

 fact that the amount of hydrogen required for the completion of a reduc- 

 tion was very much larger than the amount actually necessary to combine 

 with the bromine,* a gasometer was constructed which should collect the 

 hydrogen after it had passed through the tube and deliver it repeatedly 

 to the apparatus, after removal of the hydrobromic acid. The hydrogen 

 was generated by a primary battery consisting of zinc amalgam, hydro- 

 chloric acid, and platinized platinum. The gas delivered by this appa- 

 ratus is pure, except foi- the presence of a little hydrochloric acid, which 

 can be removed easily by means of potash. The following cut shows 

 the apparatus in its improved form. 



The bottle B is filled with pure hydrogen generated by the battery 

 C. From the bottle A water is siphoned into B, forcing the hydrogen 

 by way of the stopcock e through the column D, filled with beads 

 moistened with aqueous cupric sulphate to remove sulphur compounds 

 taken from the rubber; through the columns E and .^ which contain 

 dilute sodic hydrate, then either directly through g or through a potash 

 tube G into the reduction tube H. After being freed from hydrobromic 

 acid in the bottle K containing potash, the gas is conducted through the 

 open stopcock h into A. When B is full of water the process can be 

 repeated by interchanging A and B, and opening the stopcocks c and d 

 after closing h and e. The generator C served to keep the pressure 

 always outward. The current of gas could be regulated by a pinchcock a 

 on the rubber siphon tube. This apparatus proved entirely satisfactory, 

 and was not altered during the investigation. 



As in the case of nickel, it was found impossible by reduction alone 

 to remove the last traces of bromine from the spongy cobalt. Even 

 long continued heating to a temperature much above the subliming 

 point of cobaltous bromide failed to give complete reduction, the solu- 

 tions of the reduced cobalt giving decided tests for bromine. In the first 



* See the preceding paper, p. 333. 



