2 
MR. J. REGINALD ASHWORTH: 
change ot‘ intensity in this cyclic state is nearly a linear fnnction of the temperature, 
and tlie I'elatlon is 
I/' = 1/ {1 + — t)\, 
where 1/^ and 1/ are the magnetic intensities at the higher and lower temperatures d 
and t, and a is a coefficient which, in general, is negative. 
For a given range of temperature* the irreversible part of the change may be 
expressed by the et|uation 
+/ 8 ), 
where I, and !/■ are initial and final Intensities. Hitherto, almost without exception, 
for residual magnetism, has been found to be negative, that is to say, there is a 
permanent loss of magnetism as the result of repetitions of heating and cooling. 
The magnitude of both a and (3 varies considerably, but the conditions which 
determine the magnitude have not been exhaustively examined. Some of these 
conditions are investigated here, and it will be shown that under certain well-defined 
circumstances the coefficients a and /3 may change sign. 
2 . In a former paperf it was proved that the dimension ratio of a magnet governing 
its demagnetising factor controls to a large extent the magnitude and even the sign 
of the temperature coefficient. The experiments then made were carried out on 
pianoforte drawn steel in the commercial state, but they have now been extended 
to steel in other conditions, and the results are given in Table I., from which 
Diagrams I. and II. are plotted.;]; 
* In the experiments in this paj^er the range of temperature is from 14° to 100° C. 
i ‘ Roy. Soc. Proc.,’ vol. 62, p. 210. 
I Throughout this paper all numerical results are expressed in c.g.s. units and in degrees Centigrade. 
