Aluminum from Iron, 419 



experiment (5) the ignition was made with great care in an 

 atmosphere of superheated steam ; and in experiments (6) to 

 (10) the chloride was dissolved, precipitated as the hydroxide, 

 and weighed as the oxide. 



The experiments in which the chloride was converted to the 

 hydroxide before ignition show upon the average an absolute 

 loss of about 0-0006 grm.; the single experiment in which the 

 ignition took place in steam shows about the same loss — 0.0.005 

 grm.; while in those experiments in which the chloride was 

 dried and then ignited directly, the average loss amounts to 

 about 0'0020 grm. The error of the process which involves 

 the precipitation of the aluminum as the hydroxide, falls within 

 reasonable limits, but it is plain that the direct ignition of the 

 chloride is liable to error, which may possibly be explicable as 

 a mechanical loss occasioned by the too rapid evolution of the 

 hydrochloric acid and water of crystallization, or, possibly, as 

 the result of a very slight volatilization of the aluminum still 

 holding chlorine in spite of the decomposing action of the 

 water upon the chloride. In either case, it would seem to be 

 reasonable to suppose that a layer of some easily volatilizable 

 oxidizer placed upon the aluminum chloride might serve to 

 obviate the difficulty — in the one case, by serving as a screen 

 to diminish mechanical transportation of the non-volatile 

 material ; and in the other, by acting as an agent to promote 

 the exchange of chlorine for oxygen on the part of the alu- 

 minum chloride. 



We have tried, therefore, the expedient of covering tbe 

 aluminum chloride before ignition with a layer of mercuric 

 oxide, which of itself left no appreciable residue when it volat- 

 ilized. The hydrous chloride was collected as usual upon the 

 asbestos in a perforated crucible, dried for a half-hour at 150° C, 

 covered with about 1 grm. of the pure mercuric oxide, gently 

 heated with great care under a suitable ventilating flue, and 

 finally ignited over the blast. The results are given below : 



Table II. 



AlaOa taken in AI2O3 found 



solution as the by ignition Final 



chloride, with tlgO. volume. Error, 



grm, grm. cms grm. 



(1) 0-0761 00758 25 0-0003 — 



(2) 0-0761 0-0754 25 0-0007 — 



(3) 0-0761 0-0751 25 0-0010 — 



It is obvious, therefore, that the precipitation of the crystal- 

 line hydrous aluminum chloride from solutions of the pure salt 

 is perfectly feasible and very complete, when effected by 

 aqueous hydrochloric acid and ether thoroughly saturated with 



