126 SCIENCE PROGRESS 



low temperatures to trace this curve. But we may confidently 

 predict that for an aqueous solution of a salt which separates in 

 the anhydrous condition — or for a mixture of any two substances 

 which do not form compounds or isomorphous mixtures — the 

 usual V-shaped freezing-point curve of the two components 

 (ME;« of fig. 2) is accompanied by a V-shaped supersolubility 

 curve (SHs of fig. 2) representing their temperatures of 

 spontaneous crystallisation. 



[Since writing the above I hear from Mr. Hartley that he 

 and Mr. Barrett have realised the ice supersolubility curve in 

 studying the crystallisation of dilute solutions of sodium sulphite.] 



In general, therefore, the cooling of two substances dissolved 

 in each other, whether an ordinary solution, a mixture of two 

 fusible substances, a rock, or an alloy, will be represented by 

 fig. 2, in which M and m are the melting points of the two sub- 

 stances A and B, ME and wE their freezing-point curves meeting 

 in the eutectic E ; S and a are the temperature of spontaneous 

 crystallisation of A and B when molten and supercooled; SH 

 and o-H constitute their supersolubility curve, meeting in what 

 we call the " hypertectic " point H ; this is the highest tempera- 

 ture at which they can crystallise spontaneously together. 



The process of cooling of a liquid mixture initially of com- 

 position X is generally supposed to be represented by the line 

 xa/E. The liquid cools down till it becomes saturated at a and 

 then cools, with crystallisation of A, always remaining saturated 

 (as represented by a/E) until at E it reaches the eutectic com- 

 position, when B and A can crystallise together without change 

 of temperature, and the whole mass solidifies as the eutectic 

 surrounding the crystals of A. 



It will now be clear that this only occurs if the liquid be 

 inoculated with A, be kept stirred, and be cooled extremely 

 slowly. Otherwise supersaturation will occur, and the process 

 is, again, more faithfully represented by the line XabfE ; crystals 

 may make their appearance by inoculation at the temperature a 

 — but at the temperature b there will be a sudden access of 

 crystallisation by reason of the spontaneous growth, and some 

 time will elapse before the solution again falls to saturation 

 point at /. Or, before this happens, the solution may become 

 labile for B, and that substance may crystallise spontaneously, 

 and so the eutectic mixture may never be attained. 



In order to test these predictions we have made experiments 



