420 Gooch and Havens — Separation of Aluminum, etc. 



the gaseous acid and kept cool ; and that the conversion of the 

 chloride into the weighable form of the oxide is best effected 

 by ignition under a layer of mercuric oxide, or by dissolving 

 it in water and precipitating it as the hydroxide to be after- 

 vrard washed, dried, and ignited. Of the two methods the 

 former is by far the more convenient. 



The precipitation of the aluminum chloride in pure condi- 

 tion from solutions containing ferric chloride ought not, it 

 would seem, to present any difficulty, providing only that the 

 precaution is taken to have present a sufficient excess of ether. 

 The question was put to the test of experiment with the results 

 recorded in Table III. 



Measured portions of the standardized solution of aluminum 

 chloride were evaporated nearly to dryness in a platinum dish, 

 an amount of pure ferric chloride equivalent to about 0*16 

 grm. of the oxide was added in a very little water, 15 cm^ of 

 the mixture of strong hydrochloric acid and ether in equal 

 parts were introduced, the liquid was saturated* at 15° C. with 

 gaseous hydrochloric acid (the dish being held in a convenient 

 device for cooling it by running water), 5 cm^ more of ether 

 were added to secure complete miscibility of the solutions, and 

 more gas passed to perfect saturation. The aluminum chloride 

 was collected upon asbestos in a perforated crucible, washed 

 with a mixture of ether and aqueous hydrochloric acid thor- 

 oughly saturated with the gaseous acid, dried at 150° C. for a 

 half-hour, covered with 1 grm. of pure mercuric oxide, and 

 ignited at first gently and finally over the blast. 







Table III. 









AI2O3 taken in 



AI2O3 found FeaOa 









solution as the 



by ignition present as 



Final 







chloride. 



with HgO. chloride. 



volume. 



Error 





grm. 



grm. grm. 



cm'^ 



grm. 



(1) 



o-ovei 



0-0757 0-15 



25-30 



0-0004 



(2) 



0-0Y61 



0-0756 0-15 



25-30 



0-0005 



(3) 



o-o^ei 



0-0755 0-15 



25-30 



0-0006 



(4) 



0-0761 



0-0755 0-15 



25-30 



0-0006 



The results show plainly a very satisfactory limit of error. 



