240 Proceedings of the Royal Society of Edinburgh. [Sess. 
Method of Investigation . — In the following experiments the magneto- 
metric method was adopted throughout. The test specimen, in the form 
of a cylindrical rod, was placed in the “ A ” position of Gauss, within a 
magnetising solenoid. The field at the magnetometer needle, due to the 
current, was balanced in the usual way by a compensating coil of large 
radius, connected up in series with the solenoid. The adjustment was 
carried to such a degree of refinement that, on reversing in the circuit a 
current of considerably greater magnitude than the maximum current 
employed in the tests, the movement of the magnetometer needle was 
quite inappreciable. This balance was tested for various points on the 
scale in the manner recommended by Erhard, and found to hold through- 
out. For the thermal treatment a Fletcher gas furnace was employed, the 
temperatures being measured by a platinum, platinum-iridium pyrometer. 
Preliminary Tests . — In the first experiments several specimens were 
prepared from hard steel, completely demagnetised, and annealed at a 
temperature of 900° C. The specimens, when cold, were subjected to an 
I-H test, the fields extending; from 0 to 125 c.G.S. units. In each case the 
test was repeated after the specimen had been demagnetised. The results 
obtained with the various specimens were in excellent agreement. Fig. 1 
exhibits the I-H curves obtained with the specimens in the two states. 
The continuous line shows the initial, or sensitive, condition of the 
