g 4 SCIENCE PROGRESS. 



the rate of cooling of any substance is necessarily greater, 

 the greater the difference in temperature between it and 

 surrounding bodies, and this feature in the curves has no 

 special meaning. 



On examining these curves we see that the upper part 

 of numbers i to 7 contains a well-marked flat due to the 

 Raoult effect, and that this becomes fainter as the percent- 

 age of aluminium increases. From 9 to 12 we see a series 

 of flats occurring at a nearly constant temperature, cor- 

 responding probably to the formation of conjugate liquids, 

 while from 12 onwards in the alloys rich in aluminium 

 we see a well-marked flat rather below the melting point 

 of aluminium (654°). This occurs at exactly the same tem- 

 perature in each case, and is certainly the eutectic flat for 

 aluminium saturated with iron. The copper-tin and copper- 

 bismuth curves given by Roberts Austen show similar 

 features. 



If the preceding analysis of the process of solidification 

 be correct, then during the whole of the period from M to 

 P (fig. 4) crystals of A or of a solid solution of B in A will 

 be separating out and forming nuclei of solid matter, while 

 from P onwards a liquid of constant composition will be 

 solidifying in the spaces between the crystals already formed. 

 This liquid which solidifies last is independent of the original 

 percentage composition of the alloy, so that there is some 

 danger of mistaking it for a definite compound. A structure 

 corresponding" to the above has been observed in the micro- 

 scopical examination of alloys. 



The cooling curve may be much modified in special 

 cases. For example, it is conceivable that the freezing 

 point of the eutectic solution of A in B might be above 

 N instead of below it. In this case the flat at N will be 

 prolonged, and the second flat will disappear. The same 

 change in the diagram, though from different causes, will 

 occur if the two metals are miscible in all proportions, like 

 lead and tin or zinc and cadmium. Surfusion may some- 

 times cause a rise at the point M, and the formation of 

 chemical compounds might profoundly alter the curve. 



It appears probable that mechanical strength will largely 



