and on the Isomeric Modifications of Selenium. 495 



The product of this specific heat by the equivalent 170*98, 

 which is usually given to aluminium, is 35*15. This number is 

 a little less than the analogous products obtained for other simple 

 bodies; but this seems to indicate that the metal on which I 

 operated was far from being pure. In fact, according to an ana- 

 lysis which M. Salvetat was good enough to make at my request, 

 the metal contained, — 



Silicium 2*87 



Iron 2-40 



Copper 6*38 



Lead traces 



Aluminium (by difi'erence) , 88*35 



100*00 



We can calculate very closely the specific heat of pure alumi- 

 nium from that which we have found for the impure metal, and 

 from the specific heats of the metals which are mixed with it. 

 It is merely necessary to make a supposition on the specific heat 

 of silicium which has never yet been determined. Let us as- 

 sume that it is 0*1425, because the product of this specific heat 

 by the equivalent 266*7 of silicium is 38. But this presupposes 

 that the formula of silicic acid is SiO^, which is far from being 

 demonstrated. Whatever it may be, the error which would result 

 from the falsity of this assumption would be of little importance 

 for the object which at present occupies us. We should find 

 thus for the specific heat of pure aluminium 0*2181 ; and the 

 product of this number by the equivalent of aluminium is 37*29, 

 which is entirely within the prescribed limits. 



Hence the value which we have found for the specific heat of 

 pure aluminium confirms the formula Al^O^ which chemists 

 have been led to adopt for alumina, from the isomorphism of 

 this substance with the sesquioxide of iron. 



Since then I have succeeded in procuring purer aluminium 

 which had been prepared by M. Rousseau; it only contained 

 traces of silicon, and 2 per cent, of metallic iron. Two deter- 



