THERMO-ELECTRIC QUALITY UNDER PRESSURE. 283 



error, to ten times the slope of the curves at the mid-temperatures of 



dE 

 the 10° intervals. This gi\es immediately, therefore, -57. Multi- 

 plying this by the absolute temperature gives, by a well known 

 formula, the Peltier heat at the hot junction. The Peltier heat was 

 so calculated and is tabulated in the following. From the tabulation 

 of first differences the second differences were next found by succes- 

 sive subtractions. These were plotted and smoothed graphically, 



d"E 

 giving at once -7.2. This, multiplied by absolute temperature, gives 



the difference of the Thomson heat in the compressed and the uncom- 

 pressed metal at the temperature in question. This is also tabulated 

 in the following. Perhaps to give the results as accurately as possible 

 an additional smoothing for pressure should have been introduced. 

 But this seemed to me an additional refinement not justified by the 

 reproducibility of the results. 



The convention adopted for the sign of the effect requires explana- 

 tion. In the following, the thermal e.m.f. of the circuit is called 

 positive if at the hot junction the current flows from uncompressed 

 to compressed metal. The Peltier heat is considered positive if heat 

 is absorbed by the positive current from the surroundings on flowing 



from uncompressed to compressed metal. A positive j^ means a 



larger Thomson heat in the compressed than the uncompressed metal, 

 and the Thomson heat is itself considered positive if heat is absorbed 

 by the positive current in flowing from cold to hot metal. In other 

 words, a positive Thomson heat corresponds to a positive specific 

 heat of the positive current, or a negative specific heat of the current 

 of electrons. The opposite convention is sometimes used for the 

 Thomson heat. 



The detailed presentation of data follows. The order of arrange- 

 ment of the metals is the same as that in the previous paper on resist- 

 ance. Four metals used there could not Be used in this investigation. 

 These were: Indium and Tantalum, of which I did not have a large 

 enough supply, and Antimony and Tellurium, which are so brittle as 

 to make hopeless any manipulation of continuous pieces four feet 

 long. In addition to the metals of the resistance paper, two alloys, 

 manganin and constantan, are included at the end of the list. The 

 pressure effect on the resistance as well as on the thermal e.m.f. has 

 been measured. 



The data are chiefly given by tables and diagrams. There are in 



