CRYSTALLOGRAPHICAL RELATIONSHIP 245 



a temperature of 850 C. was sufficiently high to transform 

 In(OH) 3 into In 2 3 . When, however, there is excess of 

 NH 4 NO 3 , a certain amount of nitric oxide remains absorbed 

 in the oxide at 850. Here the amount of ammonia used 

 for precipitating the hydroxide was the actual minimum 

 sufficient to ensure complete precipitation, since indium 

 hydroxide passes to some extent into colloidal solution in 

 presence of much ammonia. Consequently there was small 

 likelihood of error due to absorption of gas by In 2 O 3 ; the 

 hydroxide was, however, heated to 900, to ensure the elimina- 

 tion of any residual gas. There was no evidence of sublima- 

 tion at this temperature. 



To estimate thallium, the precipitated thallic hydroxide 

 was redissolved in sulphuric acid, and reduced to the thallous 

 state by passing a current of sulphurous acid through the solu- 

 tion. This was then evaporated to dryness to drive off the last 

 trace of sulphurous acid, and the thallium was now estimated 

 by the bromine method (14). As any sulphurous acid in the 

 solution would reduce the bromate and lead to too high a result 

 for the thallium-content, it was essential that all traces of the 

 gas be first removed. Although the bromine method involved 

 some rather troublesome processes, it was, on the whole, more 

 trustworthy than the peroxide method (1), since it took into 

 account any thallium that might be reduced to the thallous 

 state during the precipitation of the hydroxide. 



The solution from which the hydroxide had been pre- 

 cipitated was then acidified with nitric acid, and the halide 

 was precipitated and estimated as silver halide. In the case 

 of bromides excess of silver nitrate was added and the solution 

 boiled. The excess of silver was then got rid of by addition 

 of hydrochloric acid ; the filtrate was evaporated to dryness 

 with sulphuric acid, and the alkali metal estimated as sul- 

 phate. 



