630 
PROFESSOR J. A. EWING ON EXPERIMENTAL 
The susceptibility is at first greater with 4 kilos, than with no load, but afterwards 
becomes less with 4 kilos, than with no load (cf. § 89). Moreover, the residual 
magnetism, and also the ratio of residual to induced magnetism, are with low 
magnetising forces greater, and with high magnetising forces less when there is 
load than when there is not load. 
The ratio of residual to induced magnetism passes a distinct maximum in both 
cases; but the value of that maximum is greatest in the condition of no load, where 
it is as great as 0'93. My experiments have presented only one other instance in 
which the ratio reached so high a figure (cf. § 37). 
§ 113. The manner in which curves showing the relation of 3 to -§ cross each other, 
when magnetisation is performed under various constant loads, has been very fully 
illustrated in a number of the experiments described above. As these, however, 
with the exception of the test of steel, referred almost exclusively to specimens of 
iron wire whose mechanical quality was pretty much the same, I judged it desirable 
to repeat some of the observations, using specimens of iron of very different quality. 
The results so obtained were in complete agreement with those which have been 
described above; and for that reason it is needless to repeat them. The general rule, 
apparently true for any piece of iron is, that the susceptibility, under any one load, is 
for low values of 3 greater, and for high values of 3 less, than under a slightly less 
load, the crossing point occurring at higher values of 3 the smaller the load in 
question is. 
Effects of Temperature on Magnetism. 
§ 114. If a piece of magnetised iron or steel be subjected to a cyclic series of 
changes of temperature, it is well known, from the experiments of Wiedemann* and 
others that the changes of magnetism which it undergoes are, in general, not cyclic. 
If the magnetism which is being dealt with is the residual magnetism left after 
withdrawal of any magnetising force, then any cyclic process of heating and cooling, 
or of cooling and heating, results in a fall of magnetism. If, on the other hand, the 
magnetism dealt with is that which has been reached by the application of a mag¬ 
netising force that is kept in action during the experiment, then any cyclic change 
of temperature results in a rise of magnetism. In both cases it is only after any 
cyclic change of temperature has been many times repeated that the accompanying 
changes of magnetism become even approximately cyclic. 
In this respect the effects of temperature are closely analogous to those of stress 
(cf. § 77). Superposed upon the differential effects of heating and cooling there 
are progressive permanent changes of the nature of a shaking in of magnetism, when 
that is induced, or of a shaking out of magnetism, when that is residual. 
By repeating any cycle of temperature changes often enough, however, we get rid 
* “ Galvanlsmus,” II., § 522, et seq. 
