378 BRIDGMAN. 



tension range was high, 2800 kg./cm.^ Under the most unfavorable 

 conditions the two wires differed 50% from the mean. The results 

 are given by the formulas: 



At 52° e.m.f. = (0.00022T-0.073T2) X 10-« volts, 

 at 77° = (0.00013T-0.0712T2) X 10-«, 



at 94° = (0.00047T-0.075T2) X 10-^. 



At 52° the departure from linearity is so great that the e.m.f. 

 passes through a maximum. This disappears at the higher tempera- 

 tures. E.m.f. is from stretched to unstretched at the hot junction, 

 corresponding to from uncompressed to compressed for the pure 

 pressure effect. This is the first example met of an effect of this sign. 

 The pure pressure effect was not measured. 



Manganin. This was from the same stock as the pressure sample, 

 but not from the same piece. The range of tension was 1300 kg./cm.^ 

 The two samples, contiguous lengths, did not give the same sign for 

 the effect. The one showing the smaller effect reversed in sign with 

 changing temperature, at the lower temperatures being opposite in 

 sign from the more active piece, but at 98° becoming of the same sign, 

 although only 10% of it. The effect for the more active specimen was 

 not linear, but is less proportionally at the higher temperatures, and 

 at 32° actually passes through a flat maximum near f of the maximum 

 tension. The maximum values of e.m.f. for this piece were: 0.1, 

 0.25, 0.38, and 1.33 X lO^^ volts at 32°, 50°, 75°, and 95° respectively. 

 The direction is from unstretched to stretched at the hot junction, 

 corresponding to a pressure effect from compressed to uncompressed. 

 This agrees with the pure pressure effect. 



Dependence of Thomson Heat on Temperature Gradient. 



Although it is not directly connected with the immediate object 

 of this work, I nevertheless made experimental examination of one 

 other point, both because I was in a position to make the experiment 

 with comparatively little effort, and because this point is indirectly 

 involved as an assumption made in deriving the formulas used in 

 deducing the Peltier and the Thomson heats from the total e.m.f. 

 The assumption is always made that the Thomson heat depends only 

 on the temperature at a point, and not on the temperature gradient, 

 that is, on the rate of flow of heat. The existence of such a depend- 

 ence on gradients would mean the possibility of the existence of thermo- 

 electric currents in unequally heated circuits of a single homogeneous 



