376 
PROFESSOR J. A. EWING ON THE EFFECTS OF STRESS AND 
From these results, which have been confirmed by other independent experiments, 
it will be seen that the process of “ demagnetising by reversals ” has an effect closely 
similar to that of mechanical vibration in removing traces of hysteresis. (Cf. §§ 14,15.) 
§ 39. One very remarkable feature in fig. 31 is the positive maximum (or negative 
minimum) which the wire passes through when, after the process of reversals, one 
continues to add more load. This maximum occurs for all values of the load less than 
8 kilos., and is found, generally, when rather more than 1 kilo, has been added to the 
load present during the process of reversals. Further, during the removal of load a 
negative maximum (or positive minimum) occurs in each stage, as is shown by the 
dotted curves. These features are conspicuous in the descending limb of the main 
curve, that is to say, in that part of the diagram which lies between the zero of load 
and a load of about 8 kilos. In the right-hand portion, or ascending limb, the effects 
of hysteresis are comparatively indistinct, and the process of reversals affects the 
actual and subsequent values of the E.M.F. to a much less degree. 
§ 40. Experiments like that of fig. 31 were made, with this difference, that instead 
of applying the process of “ demagnetising by reversals ” at the points b, d, f, h, and 
so on, the wire was briskly tapped, so that a merely mechanical in place of a magnetic 
disturbance was set up. Then, when the vibration had subsided, and the altered 
value of the E.M.F. had been observed, the process of loading was continued. The 
results were entirely similar to those of fig. 31. After tapping (under any load up to 
6 or 8 kilos.) the resumption of loading caused the E.M.F. to have a positive 
maximum or negative minimum, and the resumption of unloading gave a positive 
minimum or negative maximum, as in that figure. The molecular state produced by 
brisk mechanical vibration is like that produced by applying the process of reversals, 
at least in these particulars—that the traces of previous stress-changes are more or 
less completely obliterated, and the next succeeding application or removal of stress 
causes similar complex variations in thermoelectric quality. 
§ 41. Attention has already (§13) been drawn to the fact that the characteristic 
difference between “on ” and “ off” curves of load and thermoelectric quality is not a 
mere lagging in the change of thermoelectric quality. If it were, a reversal, say, 
from unloading to loading would cause in the earliest stages no thermoelectric effect, 
instead of causing (as it generally does) an effect whose sign is the same as that which 
was previously occurring during the unloading. It occurred to the writer to inquire 
whether this might not be accounted for as follows, by the instability of molecular 
structure which, generally .speaking, exists at any point in the process of loading or 
unloading. To fix the ideas, suppose that, after a considerable load has been on, the 
process of unloading is stopped at a point at or near the zero of load. Before the 
stoppage the E.M.F. has been rapidly becoming more positive. But the metal is, 
owing to hysteresis, in a less positive state than it would reach if we were now to 
vibrate the wire, or subject it to the process of “ demagnetising by reversals.” In 
the circumstances assumed any disturbance will produce a change towards positive. 
