362 



BRIDGMAN. 



easily made, there was no hysteresis or other evidence of incomplete 

 internal equililuium. The effect was so large that it was necessary 

 to use the potentiometer direct connected. The maximum departure 

 of any point from a smooth curve was 2.1% of the total effect, and 

 the average numerical departure was 0.53%. As with the resistance 

 specimen, the run at 100° was made impossible by breaking of the 

 connection because of the low melting point of the alloy of bismuth 

 with solder. In passing from curves at constant temperature to those 

 at constant pressure no appreciable readjustment Avas necessary. 



Temperature 



Bismuth 



40^ 60° 

 Temperature 



Figure 41. Bismuth. On the left, the heat absorbed by unit quantity of 

 electricity in flowing from uncompressed metal, to metal compressed to the 

 pressure indicated on the curves, as a function of temperature. On the right, 

 the excess of Thomson heat in metal compressed to the pressure indicated on 

 the curves over uncompressed metal, as a function of temperature. 



The numerical results are shown in Tables XXXIX and XL and 

 Figures 40 and 41. The e.m.f. is positive and very large, increasing 

 regularly with pressure and temperature to 710 X 10"^ volts at 100° 

 and 12000 kg. The largest 'effect found for any other metal was for 

 thallium, which gave only 52.5. Except for the largeness of the value, 

 the only unusual feature is the upward curvature of the e.m.f. curves 

 at constant temperature plotted against pressure; this denotes a 

 greater proportional eft'ect for the same pressure increment at the 



