ON GOLD-ALUMLNIUM ALLOYS. 203 



cases the aluminium was determined by evaporating the filtrate from the gold in a 

 porcelain dish to remove the sulphurous acid, adding a few drops of methyl-orange 

 and precipitating the aluminium by a very slight excess of ammonia. The liquid 

 was then heated just to boiling and the alumina filtered off on the pump, washed, 

 dried and strongly ignited in a platinum crucible. The gold and aluminium add up 

 to the weight of the pure alloy, except when we are dealing with alloys not far from 

 AuAl 2 . But in the complete analyses of alloys in this region the gold and aluminium 

 usually add up to only a little over 99 per cent, of what we call the pure alloy. It 

 was noticed that in these alloys the residue was always larger than in alloys formed 

 at lower temperatures. We do not feel absolutely certain as to the nature of the 

 missing 1 per cent. It was certainly not due to errors in the estimation of the 

 gold or aluminium, but we are inclined to attribute it to the presence of a consider- 

 able amount of carbon in the alloy weighed. In most of these cases the alloy hud 

 been exposed for several hours to a temperature of over 1000 in contact with the 

 carbon of the crucible and of the stirrer. Under these circumstances the alloy 

 certainly took up carbon mechanically, and a portion of this carbon, which would be 

 in a very fine state of division, probably disappeared either during the solution of 

 the alloy in aqua regia, or, more probably, during the subsequent ignition of the 

 residue. 



In those cases where both the gold and the aluminium were determined, it was 

 easy to ascertain with certainty the atomic percentage.* Curve 4 gives a small 

 number of points near AuAL, which were determined in this way ; they are free 

 from any source of error known to us. In all the other points determined in this 

 region the aluminium was arrived at by difference only, and it is not improbable 

 that some of the alloys may have contained quite one atomic per cent, less aluminium 

 than our tables indicate. We have not attempted to correct for this error on the 

 main curve, but we feel quite sure that it is sufficient to account for the summit in 

 Curve 3 lying a little to the right of the formula AuAU 



Near the lower summit of the curve, corresponding to Au^Al, it will be seen from 

 the tables that it is immaterial whether the composition is arrived at from the per- 

 centage of gold alone, or from a complete analysis. We believe that this is true for 

 all alloys containing less than 50 atomic per cents, of aluminium. The aluminium 

 itself was, as analysis shows, very pure, and it, apparently, only takes up carbon and 

 other impurities after a prolonged exposure to a very high temj)erature. 



* For example, take the alloy in which the percentages directly determined were 



Au 93-21 per cent., Al 6'86 per cent. 



l>ividing these numbers by 197'2 and 27'Otf, the atomic weights of gold and aluminium resiwctively, we 

 get the formula 



Au 0-4727, Al 0-2533. 



Hence the atomic percentage of aluminium is 



472 2533 1 * 100, that is, 34*9. 

 2 D 2 



