400 



Prof. J. A. Ewing. Effects of Stress on [June 



The method consisted in applying load to wires by the weight of a 

 hanging tank containing water, which could be rnn in and run out at 

 any desired rate, the junction between the stressed and unstressed 

 parts of the wire being kept at 100° 0. and 18° C. approximately. 

 The thermoelectric effects were measured by the deflections produced 

 in a short-coil mirror galvanometer, adjusted for great sensibility. 



The effect of applying a moderate amount of longitudinal pull to iron 

 wires was found to be negative (that is to say, the position of iron in 

 the thermoelectric series, with the given condition of temperature, was 

 shifted towards antimony). This result was merely a confirmation of 

 the observations of Sir William Thomson (" On the Electrodynamic 

 Qualities of Metals," "Phil. Trans.," 1856). But, instead of 

 increasing up to the breaking point, the effect reached a negative 

 maximum, after which it decreased and sometimes even changed to 

 positive before the wire broke. This result was confirmed by a large 

 number of tests of different wires. 



When loading was stopped shortly before the wire broke, and the 

 load was gradually withdrawn, the thermoelectric effect passed back 

 through a negative maximum, different from the first, and, finally, 

 with no load, it reached a positive value as a consequence of the 

 stretching which had taken place. (This positive effect due to pre- 

 vious stretching when the load was withdrawn had been observed 

 before by Magnus and by Thomson.) The fact that the thermoelectric 

 quality passed back through a negative maximum during unloading 

 showed that the maximum which had occurred during loading was not 

 the result of an antagonism between the influences of stress and strain. 



A second loading of the same wire gave a series of thermoelectric 

 values greatly different from those got in either of the two former 

 processes of first loading and unloading, and showing another negative 

 maximum long before the new limit of elasticity was reached. When 

 that limit was passed, the subsequent drawing out of the wire was 

 associated with a relatively rapid change in the thermoelectric quality 

 towards positive. 



When, after a wire had once been stretched, any given load was 

 gradually applied and removed successively within the new elastic 

 limit, the thermoelectric effects for equal intermediate amounts of stress 

 during loading and during unloading were widely different, but passed 

 through a cyclic series of values for each repetition of the cyclic 

 change of stress. This effect is shown in the paper by curves which 

 give the relation of stress to thermoelectric current. The curves got 

 by putting on load (not exceeding the elastic limit) and by taking it 

 off are far from coincident, but form a closed figure containing a wide 

 area. 



This cyclic phenomenon is experimentally studied at length in the 

 paper, and the thermoelectric effects of various cycles of loads are 



