MAGNETISATION ON THE THERMOELECTRIC QUALITY OF IRON. 
373 
to avoid confusion.) After each of these loads was applied, the wire was subjected to 
the process of demagnetising by reversals before the magnetisation, whose effects were 
to be noted, was begun. The importance of this will appear in the sequel. (§§ 32-34.) 
In all four figures the greatest magnetising force applied was nearly the same, hut the 
magnetisation in figs. 24 and 25 (and to a less degree in fig. 26) was greater than in 
fig. 23 on account of the fact that the presence of a moderate amount of load increases 
the magnetic susceptibility of iron, especially of iron which is treated as the specimen 
under test was treated.'"' 
The figures show that the presence of load diminishes the general thermoelectric 
effect of magnetisation, and finally reverses it when the load is great. For the reason 
already given, the form of these curves is not very precisely defined, and the evidence 
of hysteresis in the relation of E.M.F. to magnetism is, at the best, inconclusive. 
§ 32. It has been said that in figs. 24, 25, and 26, after the load was applied, the 
wire (although then not at all magnetic) was subjected to the process of demagnetising 
by reversals before the observations were taken. This was because it was found that 
the process in question was very effective in wiping out what may be called the 
historical traces of previous actions that are left in consequence of hysteresis. This 
is a very interesting part of the subject, which may easily be explained by a few 
examples. Thus, if in the experiment of fig. 24 the wire had simply been demag¬ 
netised by reversals with no load, and then loaded to 3 kilos., before the observations 
were begun, the results would have been (during the first application of magnetising 
force) very different from those actually obtained. The difference of conditions would 
have been that, owing to the accumulated effects of hysteresis during the application 
of load from 0 to 3 kilos., the wire would have possessed a distinctly altered molecular 
structure from that which it possessed when (as in the actual conditions of the experi¬ 
ment) this accumulated result cf hysteresis was removed by repeating the process of 
demagnetising by reversals after the load was on. Precisely analogous phenomena 
have been noticed in the writer’s paper on magnetism,! where it has been shown that 
the magnetic susceptibility of iron depends not only on the load present, but on past 
values of the load, until by the process of demagnetising by reversals the metal is 
made indifferent to liow the actual state of load has been reached. 
§ 33. In fig. 27, Plate 23, the wire, previously loaded to 12 kilos, and unloaded many 
times, was subjected to the process of demagnetising by reversals, while the load was off. 
The thermoelectric E.M.F. was then -j-6'2 microvolts (see a, fig. 27). Then load was 
applied, up to 12 kilos. (Curve ah) and removed (Curve he). The want of coincidence 
between c and a represents a difference of molecular structure, caused by the application 
and removal of load. If the process of demagnetising by reversals had then been 
repeated, the E.M.F. would have risen to the value shown by a. It must be under¬ 
stood that throughout the loading and unloading there was no magnetic field in action 
and no magnetism in the wire. Next, loading was resumed (Curve cd) and continued 
up to 3 kilos. That load was then kept constant, while magnetisation of the wire 
* Phil. Trans., vol. 176 (188-5), p. 609, § 93. f Ibid., p. 612, §§ 96-101. 
