638 
PROFESSOR J. A. EWING ON EXPERIMENTAL 
The same results are also shown graphically in Plate 68, fig. 59, where the full lines 
are the curves for ordinary temperature, and the dotted lines are for 100° C. Both with 
the annealed and hardened specimen the 100° C. curves lie at first higher and afterwards 
lower than the normal curves. The crossing points, at which the effect of this difference 
of temperature on magnetic susceptibility becomes reversed, occur at values of 3 far 
higher than the critical value in the former method of experiment. This difference is 
partly, perhaps, to be ascribed to the fact that here we are dealing with induced 
magnetism, whereas in the former case the magnetism dealt with was chiefly residual; 
but the principal reason for the very much higher values of 3 required to be reached 
here before the effects become reversed is no doubt the hysteresis which exists in the 
relation of magnetisation to magnetising force. The difference in the results of the 
two methods is analogous to that which is found when we determine the Villa ri 
critical point for stress (1) by loading and unloading, (2) by magnetising under 
constant loads (cf. § 107). 
§ 121. The range through which magnetism was altered by heating and cooling in 
the experiment of § 116, and others like it, was so small that I considered it desirable 
to see whether an increased range of temperature variation would not afford evidence 
of hysteresis in the relation of magnetism to temperature. With this view, the effect 
of heating and cooling a steel bar magnet (23 centims. X 2T0centims. X 0‘94 centim.) 
in a vessel full of oil was investigated as follows :—The bar was hung, from a fixed sup¬ 
port above the oil, perpendicular to and with its centre due south of the magnetometer. 
The temperature of the oil was then altered by means of a lamp, by steps and very 
slowly, to allow the bar’s temperature to be sensibly uniform—and the deflection of 
the magnetometer was observed. The temperature of the oil was taken by two 
thermometers, the means of whose readings are given below. 
After heating and cooling the magnet until its magnetic changes became nearly 
cyclic, the following continuous set of observations, extending over 17 hours, was 
made (Feb. 26-27, 1883). The magnetometer readings are approximately proportional 
to the total magnetic moment of the bar. 
