MAGNETISATION ON THE THERMOELECTRIC QUALITY OF IRON. 
371 
c.g.s. units. Then, while the current was kept up, the process of loading and un¬ 
loading was gone through twice. The resulting changes of magnetism are shown in 
fig. 13, Curve II, and the simultaneous changes of thermoelectric quality in Curve 
Ila, fig. 15. In order to make the curves of magnetism and thermoelectric quality 
have a similar aspect, the former are drawn as they would appear if the sign of 
magnetisation (which is arbitrary) were reckoned negative. 
§ 26. Next, the magnetisation was raised to 3 = 228 by raising the field to a higher 
value, which was maintained while loads were applied and removed twice. The 
magnetic results are shown in Curve III, fig. 13, and the thermoelectric results in 
Curve Ilia, fig. 1G. 
§27. Next, in the same way, the magnetisation was raised to 3 = 598, and 
the same procedure followed. The corresponding curves are IY, fig. 13, and IVa, 
fig. 17. 
§ 28. Finally 3 was further raised to 696, and the wire loaded. The magnetic and 
thermoelectric curves for this last case are V, fig. 13, and Va, fig. 18, respectively. 
§ 29. The magnetic Curves H, III, IY, and Y are similar to others published by 
the writer in a former paper,* and their characteristics need not be dwelt on here. A 
comparison of the thermoelectric curves with them and with each other shows well the 
gradual transformation which these undergo in consequence of increased magnetism, 
and the curious likeness which subsists between the effects of hysteresis in the two 
cases. In both, the effects of the second loading and unloading differ from the effects 
of the first in much the same manner, except that there is little in the thermoelectric 
curves to correspond to the immense changes of magnetism which take place when 
loading begins. In both, the ascending limb of the “on” curve (corresponding to 
comparatively high values of load) is a small feature when the magnetisation is weak, 
and becomes transformed into the principal feature as the magnetisation becomes 
intense. It is exceedingly interesting to trace the change from the familiar form 
shown of la to that of Ya, whose connexion with the other would scarcely be 
suspected were the intermedia,te links absent. This transformation, although occurring 
to some extent in both sets of curves, is much more complete in the thermoelectric 
than in the magnetic group. One of its results is that, whereas in ordinary conditions 
the general effect of stress on thermoelectric quality is that a moderate pull makes 
the wire more negative, this effect is reversed when the wire is strongly magnetised. 
Another general effect of strong magnetisation is that it reduces the influence of 
hysteresis, making the “on” and “off” curves less widely asunder than they other¬ 
wise are. Another is that it makes the wire, when there is no load, thermo-electrically 
negative. The positive maximum which occurs at an early stage in the first loading 
is very conspicuous in la, less so in Ila, and absent in the remaining figures. The 
similar positive maximum which occurs during the second loading is present in I«, 
Ila, and Ilia, but has become a mere inflection of the curve in IYa and Ya. The 
* Phil. Trans., vol. 176 (1885), p. 523. 
3 b 2 
