208 MESSRS. C. T. HEYCOCK AND F. H. NEVILLE 



ratio as well as from the gold alone by difference, the two methods agreeing to within 

 a few tenths of an atomic per cent. It must be remembered that at 27 atoms of 

 aluminium the alloy contains only 5 per cent, by weight of this metal, and that an 

 error of 0'3 per cent, in our analyses would shift the point one atomic per cent, along 

 the curve. It is evident that our analyses are consistent with each other to a higher 

 degree of accuracy than this, but they leave the exact position of the intersection D 

 a little uncertain. The microscope shows that the summit of the branch CD is not 

 exactly at the intersection. On the branch rising from D the freezing points become 

 steadier as we go up the curve, and near the summit E, which is reached with 

 33 - 5 atoms, the alloy solidifies wholly at one temperature, like a pure metal. Adding 

 more aluminium after the point E, we have a descending branch ending in a eutectic 

 point F, close to 40 atoms of aluminium. 



The eutectic angle F is associated with a horizontal line of second freezing points 

 as usual. Adding more aluminium after F, we follow a steeply rising branch of the 

 curve to G, at which point the alloy contains about 44 atoms of aluminium. Here 

 there is probably a slight angle, another branch rising very steeply from G, while a 

 horizontal row of second freezing points for alloys with more than 44 atoms of 

 aluminium begins. At the freezing points on this horizontal line the halt in cooling 

 was well marked, but not very prolonged, that is to say, no great amount of metal 

 crystallised at the temperature G. These freezing points are not eutectics like those 

 at C and F, nor are they such steady temperatures as those at B, D, and E. We 

 shall find this an important point when we attempt to interpret the curve. 



The branch rising from G is steep and the first freezing points are very fugitive ; 

 but near the summit H at 6 6 '6 atoms of aluminium they are again very steady 

 temperatures. At H the alloy freezes homogeneously, it being the pure substance 

 AuAl 2 . On adding more aluminium the curve descends rapidly to the eutectic point 

 of gold dissolved in aluminium, the branch being, so far as we know, devoid of 

 singularities. There is a short branch rising from this eutectic angle I to the 

 freezing point of pure aluminium as shown in Curve 5. 



The information given by the curves as to the compounds formed by the two 

 metals may be summed up thus : 



Along the branch AB gold crystallises first. 



Along the branch BC a compound crystallises first which is nearly pure at B, and 

 is probably Au^Al. 



Along the curve CD a body crystallises first which is nearly pure at D, and is 

 probably Au 5 Al 2 , but may be Au 8 Al 3 . 



Along the curve DE the substance crystallising first is Au 2 Al, a body of some 

 stability which occurs pure at the summit E. The same body crystallises first along 

 the branch EF. 



Along the branch FG a new body crystallises first, but its formula is not given 

 even approximately by the point G ; it is perhaps AuAl. 



