THE EUTECTOID TRANSITION POINT OF CARBON STEELS. 187 



We may assume tentatively that the intercept PQ is a measure of the amount of 

 iron in the eutectoid and that, similarly, the intercept QM is a measure of the amount 

 of excess iron. The ratio of these two, QM/PQ, can then be compared with the 

 calculated ratio of the excess iron to the eutectoid iron. 



If we assume that the eutectoid contains e per cent, of carbon, then in a steel 

 containing c per cent, of carbon, the ratio of the amount of excess iron to that 

 associated with the carbide in the eutectoid is 



100(e-c)/c(lOO-15e). 



The value of e is not very accurately known. It probably lies between 0'85 

 and 0'9. Accordingly in a steel containing 0'15 per cent, of carbon, the calculated 

 ratio lies between 5'35 and 578. 



The two horizontal lines A and B, near Q in the figure, are at distances above M 

 such that the upper one produced would divide PM in the ratio 578 : 1, whilst 

 the lower one would divide it in the ratio 5 '3 5 : 1. 



It appears, therefore, from the experimental position of Q, that, within the limits of 

 error, the contribution of the eutectoid to the total magnetism corresponds with the 

 amount of iron which it contains. Viewing this result from the opposite standpoint 

 it will be seen that, if we accept it, we obtain a thermomagnetic method of 

 determining the composition of the eutectoid. 



In anticipating such a method the only question was as to how intense the field 

 would require to be in order that the computation might be made with useful accuracy. 

 Here we could only foresee that the steels would be relatively soft magnetically 

 at temperatures near 700 C., and that fields of moderate strength might suffice 

 to produce the necessary approach towards saturation. 



The possibility of error in moderately strong fields, owing to the shortness of the 

 specimen, must of course be borne in mind. For, although the solenoidal field is 

 kept constant, the field within the specimen will vary appreciably with the intensity 

 of magnetisation. In the present case the intensity of magnetisation of the specimen, 

 at the point P, was about 720 C.G.S. units ; while, at the point Q, it was about 

 600 units. 



Assuming a constant demagnetisation coefficient (independent of the distribution 

 of magnetic material within the specimen) of about 0'09, the demagnetising fields 

 in the two cases would be about 65 and 54 C.G.S. units respectively. Thus the 

 effective fields at P and Q would differ by 11 units, being 135 and 146 C.G.S. 

 respectively, and the comparison made above upon the assumption of their equality 

 would require a correction for this difference. 



The correction required is apparently small. It would of course become negligible 

 if very strong fields were used. As a step in this direction, we attempted to obtain 

 curves with a magnetising field of 400 C.G.S. We found, however, that the 

 compensation between the magnetising solenoid and the particular balancing coil used 



2 B 2 



