356 0. G. KNOTT AND S. KIMURA. 



the tensi(iii ol'S, whioli corresponds to :i \on(\ of between 9 nnd 10 kilos 

 — a resnJt in fair agreement witli some of IVofessor Ewing's. 



The 2'eneral conehision that may be deduced is that the effect of 

 tension on the tliermoelectric position (^f an iron wn-e is a complex 

 function of the temperature. Not only does the line on the ther- 

 moelectric diagram suffer dis])lacement u]) or down hui it also suffers 

 rotation. In other words the Peltier Effect and Thomson Effect are 

 both changed. 



These results can only be regarded as preliminary. They are 

 sufficient to show that the method is workable, and they have a dis- 

 tinct value in themselves. Tt would be advisal)le to re])eat and extend 

 the experiments with a much more massi\e irt^n tulx' tlian that liere 

 used. A sniîdler gradient of tem])eratu]'e would l)e therein' obtained, 

 and it would not be necessary to keep tlie one end of the wire at a very 

 high temperature. JW such a modific:ition, mucli highei- tensions 

 might 1)e applied. 



