MAGNETISATION ON THE THERMOELECTRIC QUALITY OF IRON. 
365 
§ 11. One conspicuous feature in this difference of thermoelectric quality during 
loading and unloading is an apparent lagging of the thermoelectric change behind, 
the change of stress. This aspect of the action may justify the use, for the sake of 
brevity in referring to it, of the name hysteresis, introduced by the writer in speaking 
of a similar phenomenon which presents itself whenever changes of magnetisation are 
caused by changes of magnetising force or by changes of stress.* 
This hysteresis is (as Herr Cohn has already pointed out) an essentially static 
phenomenon. It is in no way affected by the speed at which load is applied and 
removed. In some of the writer’s experiments the rate of loading and unloading was 
varied tenfold without causing any perceptible change in the form of the curves or in 
the area included between them. In other experiments a particular value of the load 
was kept for a long time constant, but the mere lapse of time appeared to be without 
effect. The value of the E.M.F. reached by any process of loading or unloading 
remains constant if the wire be undisturbed. The E.M.F. associated with any 
particular load is, of course, capable of assuming any value within a wide range, in 
dependence on the particular mode of loading by which the assigned load is reached. 
Any point within the area enclosed by the curves expresses a possible relation between 
E.M.F. and load, and the actual relation depends not only on the actual load, but on 
all the preceding states of load, especially on those which have immediately preceded 
the actual state. The effect of adding any load is by no means necessarily the same 
as the effect of removing that load, unless the operation be repeated, by itself, often 
enough to reduce the corresponding thermoelectric changes to a cyclic state. 
§ 12. Figs. 5, 6, 7, 8, and 9, Plate 21, further illustrate the character of this hysteresis 
in the relation of thermoelectric quality to stress. They all represent experiments made 
on the wire of fig. 4. In fig. 5 the cycle 0—14 — 0 was performed, with the result 
that although the “ on ” branch had stopped short of the negative maximum (at 
18 kilos, in fig. 4), nevertheless there was a very distinct negative maximum on the 
“off” branch. Fig. 5 is similar to the figure given by Cohn as representative of the 
behaviour of annealed iron. Other experiments showed the same thing to be true if 
the process of loading was stopped anywhere on the steeply-descending portion of the 
“on” curve. Fig. 6 shows the loop formed by superposing on the main cycle, 
0—21—0, the small cycle, 11—3—11, starting at the point 11 kilos, on the main 
“ on ’’ curve. The dotted line in fig. 6 gives the result of a separate experiment, 
showing the effect of stopping unloading at 12 kilos, on the “off - ” curve and reapplying 
the full load. It illustrates how the points which define the relation of E.M.F. to load 
are not even limited to fall within the main “'on” and “off” curves. Fig. 7 shows 
the curious changes of E.M.F. which occurred when at the point 16 kilos, on the “ on ” 
curve, the load was reduced to 7-g, then reapplied, and the loading continued to 
21 kilos. In the same way, fig. 8 shows the effect of loads 0 — 12 — 6 — 15. Fig. 9 
* Hysteresis, from itnepew, vide Proc. Roy. Soc., vol. 33 (1881), p. 22; vol. 34 (1882), p. 39; 
vol. 36 (1883), p. 123 ; and Phil. Trans., vol. 176 (1885), p. 524. 
