RESEARCHES IN MAGNETISM. 
583 
removal of the load the following observations were taken. The readings given are 
the deflections of the magnetometer needle from its zero : they represent on an 
arbitrary scale the intensity of magnetisation of the wire. 
Stretched Iron Wire (March 2, 1882), Plate 62, fig. 31. 
Load in kilos. 
Magnetometer 
Magnetometer 
1 kilo. =2'08 kilos. 
reading. 
Load in kilos. 
reading. 
per sq. mm. 
0 
252 
i 
259 
r 
11-8 
426 
2 
273 
11 
428 
3 
294 
10 
432 
4 
323 
9 
435 
5 
355 
8 
438 
Loading . . 
6 
389 
7 
440 
7 
413 
Unloading .<( 
6 
438 
8 
430 
5 
431 
9 
437 
4 
413 
10 
438 
3 
385 
11-8 
434 
2 
350 
<- 
12-6 
429 
1 
304 
0 
254 
The same experiment is shown graphically in Plate 62, fig. 31, where the arrows 
distinguish between the on curve, or curve of loading, and the off curve, or curve of 
unloading. The whole action is very nearly cyclic, but there is a remarkable difference 
between the values of the magnetism corresponding to equal values of the stress 
during the on and off parts of the operation. The changes of magnetism lag behind 
the changes of stress. 
This hysteresis, or lagging of the changes of magnetism behind the changes of 
stress, is of a purely static character. The magnetism does not change after a load 
has been applied, and while the load remains constant, except to a very insignificant 
extent. The successive steps in the process of loading and unloading may be 
performed at intervals of some hours even without affecting the form of the curves. 
Examining now the on curve, we see that the effect of moderately loading a 
stretched wire is to increase its magnetism, under the conditions we have here, but 
that a maximum is passed, and the later stages of the loading diminish the mag¬ 
netism. A similar maximum occurs on the off curve, but owing to the presence of 
hysteresis it occurs at a smaller value of the load than the maximum point in the on 
curve. Each maximum is, in fact, shifted by hysteresis to a later part ol the 
operation than it would otherwise occupy, and if there were no hysteresis we should 
expect it to take a position between the two actual maximums, or at about 8 kilos, 
of total load. Another characteristic of the curves, obviously attributable to 
hysteresis, is the comparatively easy gradient at the beginning of the on curve and 
again at the beginning of the off curve. 
4 F 2 
