THERMO-ELECTRIC QUALITY UNDER PRESSURE. 



285 



by small amounts which might rise to as much as 0.04 X 10"^ volts 

 at the highest temperature. Furthermore, all runs showed consider- 

 able hysteresis, which of course is characteristic of a substance not 

 in perfect equilibrium. The zero was usually recovered with fair 

 accuracy; at 98° the zero error was hardly 1% of the total eflFect, but 

 at 75° there was some sort of progressive change during the run, 

 which of course produced the largest effect at zero. The readings 

 at 75° failed by about 30% to fall on a smooth curve with those at 

 other temperatures, and the 75° points were accordingly disregarded 

 in drawing the curves. 



In spite of the irregularities, the uncertainties never became great 



II IIM I I! 



^ 



^mmmmmm 



0° 10° 20^ 30° 40° 50° 60° 70" 80° 90° OC 



Temperature 

 Tin 



Figure 2. Tin. Thermal E. M. F. of a couple composed of one branch 

 of uncompressed metal, the other branch compressed to the pressure in 

 kg./cm.- indicated on the curves, the junctions being at 0°C and the tempera- 

 ture plotted as abscissae. 



enough to obscure the essential character of the results, which are most 

 curious. The results are shown in Figure 2, and Table I, in which 

 e.m.f. at constant pressure is plotted against temperature. The 

 effect is small, rising at the most to 0.4 X 10"^ volts, and reverses in 

 sign with temperature. At 12000 kg. the e.m.f. passes through^a 

 negative maximum of 0.18 X 10~^ at about 25°, and then reverses 

 sign, rising to 0.4 X 10~® volts at 100°. Or if the results are plotted 

 as a function of pressure at different constant temperatures (which 



