] 90 DR. S. W. J. SMITH AND MR. J. GUILD : A THERMOMAGNETIC STUDY OF 



per cent, steel than in that first considered on account of the greater difference 

 between the ordinates MP and MQ. But the discrepancy between the calculated and 

 observed values cannot be ascribed to this cause alone if we assume a constant 

 demagnetising coefficient as before. For, in that case, it would appear that the 

 calculated position of Q should be below, not above, that observed. 



It is not impossible, however, that the demagnetising coefficient is greater when 

 the material is in the state corresponding with Q than when the rod is more com- 

 pletely magnetic. Under such circumstances the difference between the demagnetising 

 fields would be less than a calculation like that in 7 would give. 



Apart from such uncertainties, which only observations in very strong fields could 

 remove, there are reasons why, even in the strongest fields, the calculated position 

 of Q should, as in figs. G and 7, be higher than that observed. 



To make these reasons clear, it is necessary to consider the conditions under which 

 the eutectoid forms in the different alloys. 



9. The Effects of Incomplete Equilibrium Arising out of the Slowness of 



Diffusion. 



Suppose that the cooling alloy contains c per cent, of carbon, where c is less than 

 the eutectoid percentage e. We need consider only the changes which occur below 

 900 0. At some temperature between this and the eutectoid temperature, which 

 is lower the greater c is, iron " crystals " (easily magnetisable below about 780 C.) 

 begin to separate from the homogeneous " solid solution " of iron and carbide. 



As the temperature falls these crystals grow, and simultaneously the remaining 

 solution becomes richer in carbide. 



The conditions of equilibrium at any temperature 6 require that the solid solution 

 in immediate contact with the separated iron crystals should contain a percentage 

 of carbon c g which is a definite function of 9, intermediate between c and e, increasing 

 in magnitude as the temperature falls. 



The separated iron crystals grow around nuclei distributed throughout the solid 

 solution and their growth gradually restricts the regions within which the carbide 

 is contained. But it will be noticed that the concentration of the carbide within 

 these regions is not necessarily uniform. It is c e where contact with the separated 

 crystals of iron occurs ; but it is only by diffusion inwards from the contact layers 

 that the concentration in carbide can rise throughout the rest of the solution. 

 Simultaneously with this diffusion, the crystals of separated iron grow. 



Thus we see that, unless we suppose a continuous separation of fresh nuclei as the 

 temperature falls (which surface effects tend to prevent), the rate of crystallisation 

 of the iron (and of rise in concentration of carbide in the solution remaining) depends 

 very largely upon the rate of diffusion of the carbide within the solid solution. 



Thus also, we see that the amount of iron which has separated at any given 



