360 
PROFESSOR J. A. EWING ON THE EFFECTS OF STRESS AND 
shows the formation of a loop on the “off” curve at the point 4 kilos., by superposing 
the small cycle 4 —10 — 4. 
§13. The characteristic difference between the “on” and “off” curves is not 
} 
however, a mere lagging of changes of thermoelectric quality behind changes of stress, 
such as the lagging of magnetic change which is observed when the state of stress of 
a magnetised piece 0 f iron is varied. # It is that, but it is something more. An 
inspection of the figures will make it evident that, in general, a reversal from loading 
to unloading, or from unloading to loading, causes at first a continuation of the same 
kind of thermoelectric change as has been going on before the reversal takes place. 
Thus on the “ off” curve, towards the close of the removal of load, a rapid change of 
E.M.F. towards positive is going on, and when loading begins the change towards 
positive continues in the earliest part of the “ on ” curve. And it must be borne in 
mind that this happens irrespective of the speed of unloading and loading, and even if 
a long interval of time be allowed to elapse between the unloading and the next 
loading. Again (figs. 5 to 9), to stop loading at a point on the “ on ’’ curve, when a 
rapid change of E.M.F. towards negative is going on, and to begin unloading, causes the 
metal to continue at first to change towards negative. The only case in which the 
figures do not clearly bear out this remark is where, after the application of a load 
equal or nearly equal to the greatest value used in these experiments, the process of 
unloading begins. But careful experiments made with the express object of elucidating 
this point have given grounds for believing that, there as elsewhere, the very first 
effects of reversal of the process from loading to unloading were of the same sign as 
the effects of the process which was reversed. Certainly the phenomenon in question 
is much less well marked with high than with low values of the load, and is not (with 
high values) easily distinguished from inevitable errors of experiment. But it may be 
affirmed with confidence that generally, if not universally, the earliest effect of reversal 
from loading to unloading, or vice versd, is to continue the same kind of thermo¬ 
electric change as lias been going on just before the reversal takes place. 
§ 14. Effects of Vibration. —These were studied by briskly tapping the wire during 
the application and removal of load, at the same time observing the thermoelectric 
currents. It was found that vibration destroys the effects of hysteresis nearly, though 
not quite, completely, and causes the “on” and “off” curves to approach towards 
coincidence. Fig. 10, Plate 21, illustrates this by showing the effect of applying to the 
same piece of wire as before the cycle of loads 4 — 19 — 4 without and with vibration. 
The wire was kept vibrating by tapping it briskly with a piece of wood, and the load 
was not reduced below 4 kilos., in order that the wire might remain tense enough for 
the tapping to take effect. Fig. 10a shows the changes of thermoelectric quality when 
the cycle 4 — 19 — 4 was gone through in the usual way without vibration, after 
several applications of the same cycle had brought about a cyclic state. Fig. 10b 
shows the effect of the same cycle, with vibration during the application and removal 
* Phil. Trans., voF 176 (1885), p. 587, § 75. Proc. Roy. Soc,, vol. 34 (1882), p. 43. 
