82 
MR. S. W. J. SMITH ON THE THERMOMAGNETIC ANALYSIS OF 
tration gradient due to deposition and of the increase of the rate of diffusion due to 
rise of temperature. During cooling from 9\ to 0' 2 , the growth is represented by the 
curve wx, which is flat compared with vw, because the rate of diffusion gets less 
as the temperature falls. The next curve xy may be steeper than wx ; but the 
succeeding element yz will be flatter. Finally, on account of the continuous decrease 
in the gradient, crystallisation will proceed so slowly that alternation of temperature 
between 9' 2 and 9\ will be without appreciable effect upon its amount. 
Assuming the effect of alternations of temperature carried out comparatively 
rapidly to be represented by vwxyz, the effect of allowing the material to cool below 
6' 2 may now be considered. With the exception of the layers surrounding the 
crystals, the solution will have the same concentration as it possessed when the 
cooling was first interrupted at 0' 2 . It is metastable at 0' 2 , but a small lowering of 
temperature is sufficient to reduce it to the labile condition. Hence crystallisation 
will begin at 9' 2 , and will continue at lower temperatures in this solution exactly 
as in the case of uninterrupted cooling. Through p, the point in which the ordinate 
at 6 '2 cuts the curve ABC, let pp' be drawn parallel to the axis of temperatures 
and cutting the curve A'B'C' in p. Then below the temperature corresponding 
to p' (# 3 ) the crystallisation curve will coincide with the corresponding part of 
A'B'C'; between 9' 2 and 9 S the curve will follow a course represented diagram- 
matically by zp'. This follows because, by hypothesis, the process represented by 
vwxyz caused the strength of the solution round the crystals to become practically 
equal to the saturation value at 9' 2 . From the curves ABC and A'B'C' it is seen that 
a solution of this strength does not pass into the labile condition until the tempera¬ 
ture 9 3 is reached. Hence there is no further deposition of nuclei from this solution 
until 9 2 is reached. Further, a solution labile at 9' 2 deposits crystals during cooling at 
such a rate that it also is labile at 9 3 . Hence both the solution which lost salt by 
deposition round crystals already present and that which lost salt by the continuous 
formation of nuclei reach 9 S in the labile condition, and thus, at the end of the 
process represented by vwxyzp’, there will be practically the same amount of labile 
solution remaining as after uninterrupted cooling along C'B'p'. All the solution is 
again labile at 9 3 , and the curves of interrupted and uninterrupted cooling will 
coincide below this temperature, From the figure (26,1.) it is clear that if the cooling 
is first interrupted at some temperature lower than 9' 2 , like that corresponding with 
v', the solution round the crystals present, originally, at v' may still be metastable 
when the temperature 9 0 is reached. 
Other effects of thermal treatment may be exhibited diagrammatically in a similar 
way. Thus, to consider the possible effects of interrupted heating following the 
saturation curve ABC, suppose first that the heating is discontinued before the 
temperature reaches the value corresponding to the line A!a ( e.g ., at q, fig. 26, II.), i.e., 
before the solution is concentrated enough to become labile, during cooling, at 9 0 . 
Then it is clear that no crystallisation by formation of nuclei will occur during 
