174 



Messrs. C. T. Heycock and F. H. Neville. 



almost despaired of being able to interpret them, until after reading 

 Professor Eoozeboom's paper on the " Solidification of Mixed Crystals 

 of Two Bodies,' published in the ' Zeitschrift fur Physikalische 

 Chemie' of December, 1899, The beautiful theory contained in this 

 paper made the attempt to decipher the hieroglyphic of the copper-tin 

 alloys more promising; but the experimental method recommended 

 by Koozeboom, that of isolating the first crystals that form when a 

 liquid begins to solidify, is beset with almost insuperable difficulties in 

 the case of metals melting at high temperatures. Cooling curves will, 

 it is true, give the approximate moment of complete solidification of 

 an alloy, and enable us to plot in a rough way the " solidus " curve r 

 as Eoozeboom calls it ; bat the solidus curve thus obtained is not 

 nearly so accurate as the "liquidus" or freezing-point curve. We 

 therefore had recourse to the microscopic examination of chilled 

 alloys, a method which has thrown so much light on the nature of 

 steel. 



eoo 





eoo 



700 



TOO 

 600 







600 





eoo 



600 



400 





400 







Fig. 2.— Cooling curve of the alloy Cu sl Sn 19 . Percentages by weight : Cu 69*56% 

 Sn 30'44. Time is measured horizontally. Equal vertical distances correspond 

 to equal differences in platinum temperatures. The numbers at sides of 

 diagram give temperatures on the Centigrade scale. The numbers on the 

 curve are the points of chilling. 



The first step was to imitate Austen and Stansfield and obtain a 

 cooling curve of an alloy by means of a recording instrument. We 

 used a Callendar recorder in connection with a platinum pyrometer. 

 Fig. 2 is a small scale reproduction of the cooling curve thus 

 obtained in the case of an alloy containing 19 atomic per cents, of tin. 

 In this curve the temperature of the cooling alloy is measured verti- 

 cally, and the time is measured horizontally. It will be seen that 

 evolutions of heat occur during the period MNO and also at P and Q. 

 Below the temperature the alloy was a rigid mass, a solid. The 

 temperatures marked 1, 2, 3, 3a, 4, 5 on the curve were then selected 

 as points at which it seemed well to chill portions of the alloy. The 

 pyrometer was therefore transferred to a bath of molten tin, heated 

 well above the highest freezing-point of the alloy, and small amounts 



