278 BRIDGMAN. 



insulating plug, so that it was necessary that the two ends of this stem 

 be at exactly the same temperature. This source of error was re- 

 duced to a minimum by always using the lower bath as an ice bath. 

 The lower tank was kept full of ice and water, well packed down, and 

 a violent circulation of water through the ice was maintained with a 

 stirrer. The bath was also heavily lagged. The result was com- 

 pletely satisfactory, but I have been impressed during this work, as 

 well as in some other, by the absolute necessity of maintaining a vio- 

 lent circulation of water if complete uniformity of temperature is to 

 be secured. An ice bath, even when continuously repacked, develops 

 local inequalities of temperature, unless the water is stirred. Of 

 course the conditions here were somewhat severe, because there was 

 a constant rather large inflow of heat from the upper bath along the 

 pressure tubing. 



The electrical-measurements part of the apparatus now requires 

 description. There were two entirely independent electrical systems 

 involved. One was for measuring the pressure by means of the 

 change of resistance of the manganin coil.* This has already been 

 fully described. The other was to measure the thermal e.m.f. The 

 range of e.m.f. to be measured was from a fraction of a micro volt to 

 0.0007 volts. The galvanometer used was of the Leeds and Northrup 

 high sensitivity D'Arsonval type. It was specially constructed to 

 my specifications ; was critically damped on 5 ohms external resistance, 

 with a period of 7.8 seconds and a sensitivity of 1.5 cm. at 1 m. distance 

 for 10~^ volts. Its internal resistance was 12 ohms, so that as long as 

 the resistance of the external circuit remained under 5 ohms its per- 

 formance was sufficiently constant. The wires used were almost 

 always large enough to meet this requirement. In use the galvanom- 

 eter was set up at 3 m. distance and gave 4.5 cm. deflection for 10~® 

 volts. The method used was a double throw null method, and the 

 steadiness was great enough so that 0.1 mm. could be easily obtained. 

 Readings could therefore be made to 10~^ volts. I have to express 

 special obligations to the Leeds and Northrup Co. for this instrument. 

 At the time, certain of the materials necessary for the construction 

 of a new instrument were not to be obtained in the market, and they 

 were so kind as to take from their own shops one of their experimental 

 instruments and rebuild it to specifications. 



The method of measuring e.m.f. was a potentiometer method, in 

 which the unknown e.m.f. is tapped off around a fixed resistance, and 



4 P. W. Bridgman, Proc. Amer. Acad. 47, 321-343, 1911. 



