THERMO-ELECTRIC QUALITY UNDER PRESSURE. 297 



question, it is to be noticed that the effects were found at all tempera- 

 tures of the range, increasing at tlie higher temperature. If the 

 effect were due to a polymorphic transition, we would expect to find 

 it above a definite temi)crature only. Cohen located the transition 

 temperature at about 67°. 



The mean curves of thermal e.m.f., that is the curves through the 

 mean of the points with ascending and descending pressure, disre- 

 garding hysteresis, when plotted against pressure at constant tempera- 

 ture, are somewhat unusual in that they are concave upward, the 

 proportional effect of a given increment of pressure thus becoming 

 greater at higher pressures. This is not what one would expect, and 

 is the reverse of the behavior of most metals. 



The uncertainty introduced by hysteresis was so large that rather 

 large readjustments were necessary in order to obtain smooth curves 

 when the curves of thermal e.m.f. at constant pressure, plotted against 

 temperature, were read from those at constant temperature against 

 pressure. The greatest readjustment necessary was at 8000 kg. 

 and 75°, where it was 4% of the total effect. The probable error of 

 these readjusted curves, computing by the mean square formula 

 from the departure of the indi^•idual points, was 0.34% of the maxi- 

 mum effect. 



The numerical results are shown in Figures 6 and 7 and Tables Y 

 and VI. The thermal e.m.f. is large and positive, rising with 

 pressure and temperature somewhat more steeply than normal; the 

 Peltier heat is positive and rises with temperature and pressure, the 

 Thomson heat is positive, and rises with temperature at the low pres- 

 sures, but falls at the high pressures. 



The value found by Wagner for 0° to 100° to 300 kg. was + 36.3 X 

 10"'- volts per degree per kg., against 35.6 X 10"'^ above. The 

 agreement is much better than could be expected in view of the large 

 hysteresis. 



Lead. This was Kahlbaum's "K" lead, freshlv extruded into wire 

 0.0288 inch diameter. It would have been desirable if I could have 

 used the same excessively pure lead from Professor Richard's labora- 

 tory that was used for the resistance measurements, but no more was 

 available, and the wire of the resistance measurements was too fine 

 for this. However, this " K" lead is exceedingly pure, its temperature 

 coefficient having been found to be only 0.2% lower than that of the 

 purest, and it is doubtless as good as the inherently less accuracy of 

 this work justifies. 



This lead is used as the standard in this paper against which the 



