ON GOLD-ALUMINIUM ALLOYS. 



200 



Along the whole of the curve GHI the substance crystallising first is AuAl^ the 

 remarkable and beautiful purple substance discovered and studied by Sir W. 



KOBERTS-AUSTEN. 



So far as we have been able to see, the above are the only compounds indicated by 

 the curve ; but, owing to the rapidity with which the excess of red-hot aluminium 

 attacks the porcelain cases of our pyrometers, we have not been able to determine 

 the points on the curve HI with as great accuracy as the rest of the curve, and 

 there might possibly be singularities in this branch, if it were not for the micro- 

 scopical evidence against such a supposition. 



We had much difficulty in determining the exact freezing point at the summit H, 

 l)ecause the pyrometer tubes, after some hours' immersion in the alloy at 1000", were 

 very apt to become perforated, and hence to cause a change in the constants of the 

 coils. But our best experiments, carried out with freshly made alloy, and with 

 pyrometers in perfect order, which had immediately before been checked by the 

 determination of the freezing point of gold, give the freezing point of the purple 

 alloy at H, as identical with that of gold (see Curve 4). This coincidence is a 

 remarkable feature in the relations of gold to aluminium, perhaps more remarkable 

 than if the compound had frozen at a higher temperature than gold. It is 

 impossible to misread the freezing points at H on account of the steady tem- 

 perature. 



SECTION II. 

 The Tables. 



In column 1 of the table we give the number of the alloy in chronological order, 

 in column 2 we state the percentage by weight of aluminium in the alloy as given by 

 analysis, and in column 3 the atomic percentage. The numbers in square brackets 

 are the percentages of aluminium based on a direct determination of both the gold 

 and the aluminium. Column 4 contains the temperature of the freezing point on 

 the platinum scale ; this is CALLENDAK'S pt. Column 5 contains the temperature of 

 the freezing point on the Centigrade-air scale. In columns 4 and 5 the successive 

 freezing points of the same alloy are placed under each other. In calculating the 

 Centigrade temperature we strictly followed the method of calculation described in our 

 paper " On the determination of high temperature " (' Journ. Chem. Soc.,' 1895, p. 160). 

 Aasuming, as we did, CALLENDAR and GRIFFITHS' value for the boiling point of 

 sulphur, the 8 of our platinum wire proved to be 1 '50. If CHAPPUIS and MARKER'S 

 value for sulphur be taken, the S will be slightly greater, and the melting point of 

 gold will l>e raised a few degrees above 1062, lower temperatures being raised by a 

 corresponding amount. This correction can be made at any time, but as it does not 

 affect the argument of the present paper, we have not thought it necessary to apply 



p T> 



it. We give the platinum temperature pt, that is, 100 X TT- -)r because this, 



"100 "o 



VOL. CXCIV. A. 2 E 



