334 BRIDGMAN. 



At atmospheric pressure the thermal electric behavior against lead 

 is given by the formulas: 



E = (-17.32^-0.0390^2) >< iQ-e volts, 



P = (-17.32 -0.078 (.t + 273) X 10-^ volts, 



a = - 0.078 (f + 273) X 10-6 volts/°C. 



For the pressure measurements it was seasoned twice at room 

 temperature to 12000 kg. Successful runs were made at 25°, 50°, 75°, 

 and 97°. At 97° the zero was displaced by an unusually large amount 

 after the run. This displacement apparently had no connection with 

 the other readings, and was probably accidental; this point was 

 discarded. Except for this point, the maximum zero correction was 

 1.5% of the total effect, the maximum departure of any point from a 

 smooth curve was 4.5%, and the average arithmetical departure 

 0.81%. The maximum readjustment in going from curves at con- 

 stant temperature to those at constant pressure was 1%. 



The numerical results are shown in Tables XXIV and XXV and 

 Figures 25 and 26. The effect is negative and large, increasing regu- 

 larly with pressure and temperature to —20.61 X 10^^ at 12000 kg. 

 and 100°. The Peltier heat is negative, increasing in magnitude 

 with pressure and temperature. The Thomson heat is also negative, 

 in general increasing in magnitude with pressure and temperature, 

 but at the highest temperature it passes through a pressure minimum 

 near 8000 kg. 



There are no previous results for comparison. Cobalt has the 

 largest negative effect of any metal measured. One would expect 

 anomalies because of the unusually large Thomson heat at atmos- 

 pheric pressure. 



Iron. Three different samples of iron were used; the effects were 

 complicated as for tin and aluminum, and it therefore seemed desirable 

 to find how the effect varied with different material. The first sample 

 was of American Ingot Iron, from the same piece as that of the re- 

 sistance measurements under pressure. It was drawn down to 

 0.020 inch diameter, and annealed to redness in the air. Except for 

 the difference of diameter, this was the same treatment as the resist- 

 ance specimen received. Another specimen of the same wire was 

 used, but it was left unannealed after drawing from 0.017 to 0.0105 

 inches in diameter. Its average temperature coefficient between 0° 

 and 100° at atmospheric pressure was 0.006080, against 0.006206 for 

 the same material when annealed. The third specimen was a com- 

 mercial soft iron wire such as is used for binding hay bales etc. It 



