RESEARCHES IN MAGNETISM. 
619 
It is interesting to notice (as was shown by the curves of figs. 44 and 46, &c., 
and is now confirmed here) that whereas the presence of a moderate stress, such as 
12 kilos, per square mm., greatly increases the susceptibility of a previously stretched 
wire, it greatly reduces the susceptibility of a soft annealed wire, although, as is 
evidenced by the recovery of susceptibility after the stress is removed, the stress is 
too small to produce any sensible hardening of the metal. 
§ 105. I have now shown that both in the soft annealed and in the hard-drawn 
state iron exhibits variations of magnetic susceptibility under varying loads which 
depend not merely on the existing state of stress, but on the states of stress which 
have preceded that which is selected as a condition of experiment, although these pre¬ 
ceding stresses produce no permanent mechanical effects, and, indeed, leave no directly 
visible traces of their having been applied, and although the piece under test is free 
from visible magnetism during these changes of stress. But, with respect to this last 
point, it occurred to me that the conditions of the foregoing experiment were not 
altogether satisfactory, for this reason, that the wires were to some extent affected by 
the horizontal component of the earth’s magnetic force. They hung vertically, and 
the vertical component of the terrestrial field was neutralised by a solenoid provided 
for that purpose. The only longitudinal magnetisation was therefore that given by the 
other or magnetising solenoid. But this arrangement left the horizontal component 
of the terrestrial field to act transversely on the wires, producing, of course, some 
small amount of transverse magnetisation, which must have been altered with every 
change of load. Now it was conceivable, though it appeared very unlikely, that the 
alterations of this transverse magnetism produced by changes of load should (even in 
the absence of all longitudinal magnetism) produce changes of susceptibility to sub¬ 
sequent longitudinal magnetisation sufficiently considerable to account for the pheno¬ 
menon discussed in the preceding paragraphs (§§ 96-104). For this reason I judged 
it desirable to repeat some ol the above experiments under conditions which would 
make any action of the kind impossible. Accordingly the wire to be magnetised was 
suspended along the line of dip, and the current in the neutralising solenoid was 
increased to the value necessary to make it neutralise the full terrestrial field. The 
wire was now free from all magnetising force, and, after being completely demagnetised, 
it was subjected to stress as before. It was then found that the peculiar residual 
effects of stress described in §§96-104 were as conspicuous as before. The suscepti¬ 
bility depended, just as much as when the wire was in the usual vertical position, on 
preceding as well as on actual states of stress. The experiment showed conclusively 
that the transverse magnetism which must have existed in the wire when set vertically 
did not materially affect the results, and could not have been the cause of the residual 
differences of susceptibility which were found when a wire, free from longitudinal 
magnetism, was brought to a given state of stress through different preceding states. 
§ 106. Magnetisation under constant Loads in a high Field. —It appeared desirable 
to examine the form which curves of magnetisation under constant load took when 
O 
