448 MESSRS. R. S. HUTTON AND J. E. PETAVEL ON ELECTRIC 



aluminium metal, although it is still accompanied hy a considerahle amount of the 

 carbide. 



It therefore became evident that further work at high pressures must be preceded 

 by a more detailed study of the conditions of reduction. The several problems which 

 arise may briefly be stated as follows : 



(1) At what temperature does alumina first show signs of reduction by carbon ? 



(2) In the production of aluminium alloys, what is the function of the auxiliary 

 metal in facilitating the reduction of alumina ? 



(3) What precautions are necessary to limit the formation of carbide and increase 

 the production of metal ? 



Since it is well known that alumina cannot be reduced under ordinary circumstances 

 in the Moissan furnace, it was thought advisable to see whether, by carrying out the 

 reaction in an atmosphere of hydrogen, a definite indication of reduction could be 

 obtained. The Moissan furnace was, of course, modified to exclude the use of limestone 

 and the accompanying production of carbon monoxide. 



As will be seen from Table X., A, a negative result was obtained. 



As a means of limiting the temperature of reaction, calcium fluoride was introduced, 

 but no signs of reduction were apparent at the boiling-point of the bath. From these 

 and similar negative results at lower temperatures, which it is unnecessary to record, 

 we assumed as a working hypothesis that the temperature of reduction of alumina is 

 above the boiling-point of aluminium metal under atmospheric pressure. 



The hypothesis we confirmed by experiments (Table X., B) in which special 

 precautions were taken to protect the material from access of air and to provide a 

 condensing chamber in which the vapours were cooled down before their exit from 

 the furnace. The deposit so obtained showed unmistakeable evidence of the presence 

 of finely divided aluminium.* 



It therefore became necessary to devise some better means for indicating the 

 production of any metal vapour. 



A method which suggested itself to us, and one which has proved of considerable 

 usefulness, was the employment of a bath of molten copper, on the surface of which 

 the reaction mixture was placed. The copper served as an absorbent for any 

 aluminium vapour liberated. 



To determine the lowest temperature at which reduction occurs, a series of 

 experiments was carried out. Small carbon crucibles were used to contain the mixture. 

 These were heated either in a carbon tube furnace, or, for higher temperatures, more 

 conveniently by embedding them in a granular carbon resistance. From the summary 

 of these experiments in Table X., C, it will be seen that the minimum temperature of 

 reduction coincides fairly sharply with the melting-point of alumina, and is not 



appreciably lowered by the introduction of either fluor spar or lime as a flux. 





 * See also C, F, MAPERY, ' Amer. Chem. Journ.,' 1887, vol. 9, pp. 11-15. 



