346 BRIDGMAN. 



nated. The zero corrections were negligible; the maximum devia- 

 tion of any single reading from a smooth curve was 3% of the total 

 effect, and the average numerical departure was 0.47%. The maxi- 

 mum readjustment in passing from constant temperature to constant 

 pressure curves was 0.27%. 



The numerical results are shown in Tables XXX and XXXI and 

 Figures 31 and 32. Both at constant pressure and constant tempera- 

 ture the e.m.f. curves are nearly linear. The e.m.f . is positive and one 

 of the largest found, rising to 26.5 X 10"« volts at 100° and 12000 kg. 

 The Peltier heat is positive, rising regularly with pressure and tempera- 

 ture. The Thomson heat above 5000 kg. is positive, falling with 

 rising temperature at the higher pressures and rising at the lower 

 pressures. Below 5000 kg. it is approximately zero. 



The value given by Wagner at 300 kg. between 0° and 100° was 

 +23.7 X 10"i2 volts per degree per kg. against 21.9 X lO"'^ inter- 

 polated from the data above. 



Platinum. Readings were made on the same piece of Heraeus 

 platinum, loaned by Professor H. N. Davis, as were the resistance 

 measurements under pressure. I wanted to procure especially good 

 data for platinum, because so much work has been done on its other 

 physical properties, but unfortunately the results are among the most 

 unsatisfactory. The piece was so short that only one strand could be 

 used, and even then an especial procedure had to be adopted to make 

 the outside connections. Furthermore, the wire had been previously 

 used for a resistance thermometer, wound on a small mica frame, and 

 it was consequently full of small tortuosities which could not be re- 

 moved without danger of altering the properties of the wire. The 

 roughness of the wire and its small size conspired to make the tension 

 effects of viscous drag particularly high. The same effect was re- 

 sponsible for the wire breaking during the application of seasoning 

 pressure at room temperature. It was fused together again by arcing. 

 The viscous drag might give rise to an error of as much as 20% in 

 the total effect, but this can be very greatly cut down by careful 

 manipulation. 



In addition to the measurements on Heraeus platinum, a complete 

 series of readings was also made on an impurer sample from Baker. 

 The pressure effect on the 'resistance of this has also been measured 

 previously. This wire was used bare, four strands in parallel. No 

 such trouble was found from viscous drag as with Heraeus platinum, 

 and the results were considerably more satisfactory. 



At atmospheric pressure the thermo-electric behavior of the two 

 grades of platinum against lead is gi^■en by the following formulas. 



