MAGNETISATION ON THE THERMOELECTRIC QUALITY OF IRON. 
377 
But suppose that no disturbance takes place, and let the process of loading he begun. 
The observed effect is, first, a change towards positive, then soon a positive maximum, 
and then a great and continued change towards negative. It is open to conjecture 
that the first (positive) effect of loading is due to the mere molecular disturbance 
associated with change of stress ; that though the proper effect of loading is negative, 
that effect lags, and superposed on it there is a positive effect due to the simple 
disturbance caused by loading, acting on the unstable molecular structure to which 
the previous unloading has given rise. This view, were it tenable, would certainly 
make the fact less extraordinary than it otherwise is, that in general a change from 
loading to unloading, or vice versd, continues the same kind of thermoelectric change 
as has been going on before. 
But this view appears to be negatived by the experiment of fig. 31, and by similar 
experiments made with mechanical shaking substituted for magnetic shaking. For we 
find that at a, for instance, in spite of the violent molecular commotion which preceded 
the loading, nevertheless the first effect of loading is to carry the wire through a 
positive maximum of thermoelectric quality, before beginning that change towards 
negative which is the main consequence of any moderately great load. And after 
every other shaking up at intermediate points in the loading the same curious positive 
maximum comes in before the main trend of the “ on ” curve is resumed. It is clear 
that this puzzling characteristic of the curves cannot be ascribed to any ordinary 
mechanical agitation which may be assumed to accompany changes of stress; for not 
only does it appear after the metal has been thoroughly shaken into what may be 
presumed to be a stable state, but at points, such as c, e, g, &c. (fig. 31), it asserts itself 
in the direction opposite to that in which the thermoelectric quality has been altered 
by mechanical or magnetic disturbance. And it is not a little remarkable that the 
phenomenon under discussion has no counterpart in the (otherwise similar) hysteresis 
which the induced or residual magnetism of a piece of iron exhibits when it is forced 
to vary by application and variation of stress. 
§42. Up to this point all the experiments of the new series (from §21 onwards) 
were made with the same sample of iron wire, which, as has been stated (§ 21), was 
a piece that had, after annealing, been moderately stretched beyond its elastic limit. 
Other experiments with other samples, treated in the same way, confirmed the 
results which have been described. The examples which have been cited are chosen 
out of a much larger number of tests, which need not be particularly described, as 
they exhibited no features which have not been already referred to. In the experi¬ 
ments which have been given (§§ 21 to 38) the hot junction had a temperature of 
from 130° C. to 160° C., and the cold junction a temperature of from 15° C. to 20° C. 
Care was taken to preserve the temperatures nearly constant during any one experi¬ 
ment, but it was not considered necessary to give them strictly the same values from 
day to day, as the aim of the research was to study, not the dependence of thermo¬ 
electric effects on temperature and temperature differences, but those peculiar features 
MDCCCLXXXYI. 3 C 
