342 



C. G. KNOTT AND S. KIMURA.. 



by which to rlrnw the correct conclusion. The nature of tlie prol:)]em 

 is most sini])ly expressed in terms of the Jnng-nnge of the thermoelec- 

 tric (liag'rnm. In this diagram tlie thermoelectric relations of the 

 ditfererit metals are represented l)y liries (usnally straigdit) in such a 

 manricr that the electromotive force existing in any circnit of two 

 metals is cqnnl to the area included between the appropriate metal 

 lines arid the two lines drawn perpendicular to the temperature axis 

 and throngli the j^oints representing the temperatures of the two junc- 

 tions. The (pie;!tion pr(^pounded aliove is tlicn this. AYhat chp.nge 

 does stress applied to a given metal produce upon tlie jjosition of the 

 line in the tliermoelectric diagram ? Does it translate it as a Avliole u\) 

 or down ; d<X's it rotate it as a whole about some definite point ; or 

 does it effect a cond)ination of these so that the line is deformed as 

 well as shifted ? In other words does stress cliange the Thomson 

 .Effect in a wii-e, or does it sim])ly change the l\'ltier Effect with 

 reference to an unaffected second wire ? 



Now it is (piite clear that the only way to answer this problem is 

 to arrange an apparatus in which the electromotive forces due to 

 (lifcrnit differences of temperature can l)e measured siinnlfaneouslij on 

 a wire under given conditions of stress. This could be accomplished 

 only by having tlie gradient of temperature along the wire both 

 steady and gradual. Junctions could tlien be made at several points 

 along tlie Avire ; and the electromotive forces due to the several cir- 

 cuits so olitainable could be easily measured and compared, once the 

 temperatures were steady. The sim])lest way of realizing these condi- 

 tions seemed to be to stretch the wire inside a metal tube, and then to 

 heat the metal tube as in the Forbes Experiment on the conduction of 

 heat along bars. 



For ease in manipulation the tube, which was of iron, was made 

 in two semi-cylindrical parts. The upper part or lid fitted accurately 



