THERMO-ELECTRIC QUALITY UNDER PRESSURE. 305 



pressure were made before the runs at 25°, 50°, and 75°, instead of at 

 25° only, as usual, but this seemed to make no difference. Within 

 the limits of error, the relation between thermal e.m.f. and pressure 

 at any constant temperature is linear. This is surprising for so large 

 an effect, which rises to + 80 X 10"^ volts, larger than for any metal 

 except bismuth. The maximum departure from linearity of any 

 single point observed was 6% of the maximum effect. This point 

 was situated at 8000 kg. and 75°; the discrepancy is almost entirely 

 due to the width of the hysteresis loop. On changing from the curves 

 at constant temperature to those at constant pressure, readjustments 

 of as much as 4% were necessary. By a curious accident, the points 

 as first plotted at each pressure lay accurately on two straight lines, 

 intersecting at a negative temperature, the points at 0°, 25°, and 50° 

 forming one group, and those at 75° and 100° another. I thought at 

 first that this was due to a polymorphic transition, but a moment's 

 consideration shows that a transition cannot produce a discontinuity 

 in the actual ordinates of a thermal e.m.f. curve, but only an abrupt 

 change in the direction. Such was not the case here. Zinc is also 

 one of the substances for which Cohen ^° supposes polymorphic 

 transitions. His transition point is at 25°, whereas here evidence of 

 internal instability was found at every temperature above 0°. 



Wagner found -f 40 X 10"'- volts per degree per kg. between 0° 

 and 100° at 300 kg. This is to be compared with 70 X 10"i- deduced 

 from the measurements above. The lack of agreement is not surpris- 

 ing in view of the hysteresis and its dependence on pressure. Wagner 

 found the same hysteresis effects as were noted above, and says that 

 the results are doubtful for this reason. With respect to this be- 

 havior he puts Zn and Cd in a class by themselves. 



Magnesium. This was from the same piece of commercial magne- 

 sium as the resistance specimen, but was extruded at a different time 

 into wire 0.020 inch diameter. The specimen whose pressure coeffi- 

 cient of resistance had been measured was too small and too much 

 oxidized to use again. It was not annealed in the electric oven, as 

 were the other newly extruded wires, for fear of too great oxidation. 



The thermal e.m.f. against lead at atmospheric pressure is very 

 small, and is given by the formulas: 



E = (-0.095 i + 0.00004^2) >< iq'S volts, 



P = (-0.095 + 0.00008 (t + 273) X lO"*^ volts, 



0- = 0.00008(^ + 273) X 10-« volts/°C. 



10 E. Cohen and W. D. Helderman, ZS. phys. Chem. 89, 742-747, 1915. 



