f>2 PROCEEDINGS OP THE AMERICAN ACADEMY. 



conditions ; henoe the method finally adopted was the ignition of care- 

 fully pur i tied parpureo-cobaltic chloride, which may l>e dried before it 

 d( compi 



A considerable quantity of this compound was prepared by passing air 

 through a Btrongly ammoniacal solution of cobaltoua chloride and subse- 

 quently acidifying the mixture with hydrochloric acid. The precipil 

 was then recrystallized ?-ix. times by solution in pure redistilled ammonia 

 in a platinum dish ami precipitation with redistilled hydrochloric acid in 

 a Jena glass fiask. This latter operation could not be performed in 

 platinum, because of the liberation of small quantities of chlorine. The 

 final precipitate was collected on a pure wash d filter and dried. In 

 order to convert the amine into CObaltOUS chloride it was heated ill an ait- 

 bath to about 200°, until it was entirely decomposed without fusing. 

 Owing to tli*- fact that the dried salt melts a! a temperature very slightly 

 above that at which it breaks up. tin; heat was applied vi iy gradually. 

 Here again the liberation of traces of chlorine prevented the use of 

 platinum, a porcelain crucible being used to contain the salt. The cor- 

 responding purpureo-bromide oi cobalt melts at a temperature too near 

 its decomposing point to have allowed the use of a similar method for 

 preparing cobaltous bromide in the earlier part of this research. 



The cobaltous chloride thus prepared, still containing a large amount 

 of amnionic chloride, was now heated in the platinum boat in a rapid 

 current of dry nitrogen and hydrochloric acid gas until amnionic chloride 

 ceased to lie expelled. The drying apparatus used in the work upon the 

 bromide was refilled, and easily altered for this purpose by substituting a 

 strong solution of hydrochloric acid tor the bromine and red phosphorus. 

 The hydrochloric acid gas was displaced by pure nitrogen, and this in 

 turn by pure air ; and finally the dry cobaltous chloride was transferred to 

 the weighing bottle by means of the usual bottling apparatus. 



After having been weighed, the chloride was reduced in hydrogen in 

 the manner already described under the analysis of cobaltous bromide. 



During the reduction no trace of cobaltous chloride sublimed from the 

 boat, but a slight sublimate of <iiiiiin>iiir chloride appeared in the tube. 

 The amount of this sublimate was determined by washing out the tube 

 with pure water and ,\r.v ilerizing the solution, and this amount was sub- 

 tracted from the original weight of cobaltous chloride. Probably the 

 correction was somewhat too great, for traces of ammonic bromide 

 appeared in every reduction of cobaltous bromide, when no ammonia 

 could have been present in the salt ; but the error arising from this 

 source must have been small. 



