448 Dr. W. C. Roberts-Austen. 



also pointed out that surf usion could be either stimulated or hindered 

 by suitably modifying the conditions, but he made no attempt at 

 thermal measurements. It was not until ten years after the publica- 

 tion of van Riemsdijk's work that the recording pyrometer, which I 

 devised and submitted to the Royal Society in 1891,* enabled such 

 measurements to be readily effected. 



It consists of a camera, enclosing a dead-beat galvanometer, to 

 which the free ends of a thermo- junction are attached. The thermo- 

 junction is suitably protected, and, as it only consists of two wires, 

 twisted together and covered with a fine clay tube, it can be placed 

 in the cooling mass of molten metal or alloy, and the cooling curve 

 of the mass may be traced by a spot of light from the galvanometer 

 which falls on a moving sensitised plate. A ready method for 

 studying all the phenomena of the solidification of metals and alloys 

 is thus afforded. 



The freezing point of a metal, or the initial freezing point of an 

 alloy, for a fluid mass of two metals may possess many points of 

 solidification, is represented by one or other of three typical curves. 

 These are shown in the accompanying figures which indicate the 



FIG. l. 





nature of the curves, traced by the recording pyrometer. Fig. 1 

 shows the freezing point curve of a pure metal, the horizontal 

 portion, ob, indicating the actual solidification of the mass, the 

 sharpness of the angles at a and b attesting the purity of the metal. 

 The initial freezing point of most alloys would resemble fig. 1 in 

 having the corner a sharp, while the point b is generally rounded 

 off. If the alloyed metals form an isomorphous mixture, neither 

 angle is sharp, and in many cases there is no true freezing point, the 

 curve being of the form shown in fig. 2. This represents the 

 freezing of the gold-silver alloy containing 28 per cent, of gold in 

 which the fluid mass, as a whole, passes through a long pasty range 



* '.Proc. Boy. Soc.,' 1891, vol. 49, p. 347. 



