ON ALLOYS. 77 



as the behaviour of the alloys to etching reagents, make it certain that 

 the combs are richer in copper than the average of each alloy or than the 

 mother substance round them. The alloys chilled between the liquidus 

 and the solidus were partially liquid at the moment of chilling, and as 

 the chill was effected by dropping the alloy into water, the result was 

 often to granulate the alloy ; one always finds in these chills more or 

 less of a tin- rich mother substance. Alloys which had been cooled below 

 the solidus before chilling are never granulated, and never show the 

 second crop of primaries ; they must have been solid before the chill. 

 Moreover, in the case of the AB alloys, when chilled below the solidus, 

 the primaries fill the alloy ; a sure proof, as it seems to us, that an alloy 

 becomes solid when its temperature falls below the solidus. This is still 

 more marked in the case of the LCDE alloys, for if these are chilled below 

 the solidus, but above Roberts-Austen and Stansfield's curve, they 

 appear to be homogeneous, though sometimes lines can be seen dividing 

 the area of the etched surfaces into irregular polygons. Below the 

 solidus the primaries are lost, not because they cease to exist, but 

 because they have completely filled the alloy and assimilated the mother 

 substance in which they grew. It appears, therefore, that each of these 

 alloys is an approximately uniform mixed crystal phase when its tem- 

 perature lies between the solidus and Roberts-Austen and Stansfield's 

 curve. On the other hand, alloys whose percentages lie between B and L 

 do not solidify homogeneously. If chilled below the bIC line they are solid, 

 but they contain copper-rich primary combs imbedded in tin-rich mother 

 substance ; near B the combs preponderate, but with more tin the 

 mother substance grows until at the percentage of L it forms the bulk, 

 and in certain chills the whole of the alloy. Moreover, if chilled above 

 C the mother substance appears uniform, while below C it breaks up 

 into a minute eutectic of two bodies. Successive chills of one of these 

 alloys at a series of temperatures from blC to C show a remarkable 

 growth of the primaries. For example, in the chills of Snjg taken close 

 to blC the combs of copper-rich primary are scanty and the lobes are 

 rounded, but as the chilling temperature is lowered the combs grow and 

 become more angular and fantastic. Alloys between L and C show 

 copper-rich primaries if chilled above Ic, but these vanish in the chills 

 between Ic and IC, while when the temperature falls below the curve 

 IC a new copper-rich crystallisation appears. Photographs of the alloy 

 Cu8B.5Sn,3.5 are enclosed which illusti'ate these features. 



In the same way, the CD alloys which show copper-rich primaries if 

 chilled above cd, and are uniform solid solutions between cd and CD', 

 are found to contain a tin-rich crystallisation of bands and rosettes if 

 chilled below the latter curve. The photographs 4, 5, and 6 of the paper 

 published in the ' Royal Society Proceedings,' plate 3, vol. Ixviii., reproduce 

 these facts. The alloys of the branch DE, and beyond, present very 

 similar phenomena. They solidify in the narrow range of temperature 

 between DE and de, but the solid solutions of the region below de are 

 very unstable, and the habit of crystallisation of the solid phase that 

 separates out along D'E' differs from that of the branch XD', a minor 

 change showing itself near Y. 



Thus we see that Roberts- Austen and Stansfield's curve, in its relation 

 to the physical or chemical changes it indicates, closely resembles a freezing- 

 point curve, except that above it there is an unsaturated solid solution, 

 instead of the region of unsaturated liquids that lies above a freezing-point 

 curve. The points on the curve correspond to saturated solids, while 



