354 Prof. W. C. Roberts-Austen. [Mar. 12, 



being recorded in the usual way by a curve. The gold was then 

 re-melted, and a weighed quantity of aluminium added, the mass being 

 stirred and the temperature of its freezing point measured and re- 

 corded in a curve. 



The addition of 0'2 per cent, of aluminium produced an appreciable 

 fall in the freezing point, but this initial fall is only indicated by a 

 change in direction of the curve. The fall as measured upon the 

 photographic plate is only 1 mm., which nevertheless corresponds to 

 a difference of temperature of 7'68 C. 



A further addition of 0'2 per cent, of aluminium (making 0'4 per 

 cent.) increased the fall to 1'8 mm., corresponding to 14'28 C. It 

 may be urged that these measurements are small, but the observa- 

 tions were repeated with a scale some distance from the galvano- 

 meter, and chronographic records gave results having the same 

 values. 



An experiment with gold in which 1 per cent, of aluminium was 

 present also conGnns this ; the fall in temperature of the freezing 

 point was in this case 33'66 C., but there were indications that the 

 gold, the solvent, was becoming saturated. 



Now 0-2 per cent, of aluminium corresponds to 2/10 x 196/27'5 = 

 1-42 atom per 100 atoms of gold.* 



Hence the fall per atom present per hundred atoms of gold = 

 7-68/1-42 = 5-4C. 



Similarly, a percentage of 0'4 corresponds to 4/10 x 196/27'5 = 

 2'85 atoms per hundred of gold, and the fall per atom = 14'28/2'85 

 = 5-OC. 



With 1 per cent, of aluminium the total fall will be due to 

 7*16 atoms per hundred of gold; hence the atomic fall will be 

 33-66/7-16 = 4-7 C. 



It may be added that experiments (as yet incomplete) seem to 

 show that lead, bismuth, silicon, and platinum cause a much greater 

 " atomic fall " in the freezing point of gold than aluminium does. 



The relations of aluminium to gold would, therefore, appear to be 

 peculiar in more ways than one. The curve (fig. 3) clearly indi 

 that aluminium has a remarkable influence on the cooling of a 

 of gold, and in view of this it would seem strange that calculatio: 

 based on the atomic weight of aluminium should show that it delays 

 the initial solidification of gold less than other elements. The com- 

 plete solidification is, however, much retarded ; for merely stirring a 

 mass of gold contaminated with very little aluminium reveals the 

 fact that the added element has set up during the solidification of the 

 mass a " pasty stage " which continues through an unusually long 

 range of temperature. 



In the metallurgy of iron, aluminium is known to play an im- 

 * 196 is the atomic weight of gold, 27*5 that of aluminium. 





