374 
PROFESSOR J. A. EWING ON THE EFFECTS OF STRESS AND 
was begun. The results of magnetising are shown in the upper part of fig. 28, Plate 23, 
which is a curve of E.M.F, and 3- Starting from d (fig. 28)—a state identical with d 
in fig. 27—the wire was magnetised until with a value of 3 = 780 the point e was 
reached. Then gradual reversal of the magnetisation carried the E.M.F. through the 
values shown by the dotted line ef. 
§ 34. Next, the wire was demagnetised by reversals and the whole process of 
loading and unloading repeated twice, but at the second time the unloading was 
stopped while 3 kilos, remained on the wire. The E.M.F. was then found to be 
— 2’3 microvolts—a stake shown by the point g in fig. 28. Then, with the load 
constant, magnetisation gave the Curve gh, and reversal of magnetisation gave the 
Curve hi. 
Both parts of fig. 28, therefore, represent the thermoelectric effects of magnetising, 
while a constant stress of 3 kilos, is maintained. But they differ in this, that in the 
upper part of the figure that load was reached by increment from 0 to 3 kilos., 
whereas in the lower part of the figure the load was reached by decrement from 
12 kilos. This difference in the history of the loads causes a difference of molecular 
structure which is nearly, though perhaps not altogether,* obliterated by a single 
application of a strong magnetising force, and in its obliteration gives rise to the 
widely different curves de and gli, both of which represent the first effect of magneti¬ 
sation on the thermoelectric quality of the same piece of iron in one and the same 
constant state of stress. 
Fig. 28 will make it clear why in the series of experiments shown in figs. 23-26 the 
process of demagnetising by reversals was performed after each of the several loads 
was ajtplied. 
§ 35. The experiment w T as then extended, in the same manner as before, to 
determine the thermoelectric effects of magnetising under 6 kilos, of load, starting in 
one case from the point 6 kilos, on the “ on” curve (reached, after demagnetisation, 
by the loads 0 —12 — 0 — 6), and in the other case from 6 kilos, on the “ off” curve 
(reached, after demagnetisation, by the loads 0 — 12 — 0 —12 — 6). The results are 
shown in fig. 29, Plate 23, from which it will be seen that the first magnetisation 
gives widely different curves of E.M.F. and 3 in these two cases, but the subsequent 
reversal of magnetisation gives effects so nearly similar in the two cases that they 
have been drawn by a single line (the dotted line of fig. 29). 
§ 36. Finally, fig. 30, Plate 23, shows in the same way the effect of a first magnetisa¬ 
tion and subsequent reversal of magnetism, under a constant load of 12 kilos., reached, 
after demagnetisation, by loading to 12, unloading to 0, and loading again to 12 kilos. 
§ 37. The experiments of figs. 28, 2.9, and 30 differ from those of figs. 23-26 in this 
respect, that in the latter the effects of hysteresis, accumulated during the process of 
loading or unloading which preceded magnetisation, were wiped out by the process of 
* The Curves ef and hi differ by an amount which may possibly be due to variation of thermal 
conditions in this protracted experiment. 
