THERMO-ELECTRIC QUALITY UNDER PRESSURE. 313 



the maximum effect, and the average departure from a smooth curve 

 was 1/2%. At the lower temperatures, 25° and 50°, where the effect 

 is very small and the curve passes through a flat minimum, the regu- 

 larity of the points is much greater. No adjustment whatever was 

 necessary in passing from the curves at constant temperature to those 

 at constant pressure. The numerical results are shown in Tables 

 XIII and XIV and Figures 14 and 15. The reversals in sign and the 

 maximiun remind one of the behavior of tin, but a detailed examina- 

 tion shows many differences. The effect at 25° is small and negative 

 throughout the entire range, passing through a flat minimum of 

 about -0.1 X 10"« volts at 6000 kg. At 50° the effect for the first 

 5000 kg. is much the same as at 25°, but above this it becomes rapidly 

 different, rising through zero and reaching + 0.3 X 10"^ volts at 

 12000 kg. At 75° the effect is positive for most of the range, and the 

 negative minimum has been pushed back to 1500 kg. At 100° the 

 curve probably starts out negative, but almost immediately reverses, 

 rising to 2 X 10"^ volts at 12000 kg. At all temperatures the curves 

 remain concave upwards. It is evident that no comparison betw^een 

 these results and those of Wagner is possible, because he used only 

 one pressure, 300 kg., and one temperature interval. The effect he 

 found was very small and negative, —0.59 X 10"^^ volts per degree 

 per kg. Since the effect is changing so rapidly it is not safe to use 

 my results by interpolation in this region to find figures to compare 

 with his. 



The behavior of the Peltier heat is on the whole simpler than that 

 of e.m.f. itself. The curves change sign and cross in a complicated 

 way, but there are no maxima or minima, and the curvature is slight. 

 The Thomson heat is still more simple; it is throughout positive. 

 At the lower temperatures, it increases regularly with pressure, but 

 at the higher temperatures there is a rather complicated crossing of 

 the curves. 



Measurements with commercial aluminum were made only at 25°, 

 50°, and 75°. I would have been glad to complete the series at 95°, 

 but after 75° the inside connection, w^hich was merely wrapped, 

 pulled apart, and since the purpose of the readings with this specimen 

 was merely to see whether the effects were qualitatively the same, I 

 did not feel justified in taking the time to set up the apparatus again. 

 The wire was seasoned for the runs by a preliminary application of 

 12000 kg. at room temperature as usual. The results, as far as the 

 broad outlines go in which aluminum is so strikingly different from 

 other metals, were the same as for the purer specimen, but the numer- 



