THERMO-ELECTRIC QUALITY UNDER PRESSURE. 281 



the other circuits but not the galvanometer circuit, eliminated this 

 as a source of error. An eliminating switch about the 0.1, 1, and 10 

 ohm coils, as described by White, showed that there were no parasitic 

 e.m.f s. in this part of the circuit. The construction of the rest of the 

 circuits assures that any other parasites, except in the pressure part, 

 are overwhelmed and masked by the applied e.m.f., and parasites in 

 the pressure part, provided they stay constant, are eliminated by the 

 zero correction. All switches and the galvanometer were further 

 protected by a leakage shield, as advocated by White. The whole 

 apparatus was surprisingly satisfactory in its performance, and 

 except with some of the more inhomogeneous metals the steadiness 

 was such that even over the maximum temperature range individual 

 readings could be made to the limit of sensitiveness, about 10~* 

 volts. 



The accuracy of the readings was of course not usually as high as 

 10"^ volts, but varied greatly for the different metals; the details 

 will be given later. In general, the precautions necessary to ensure 

 reproducible results were to previously season the metal by several 

 preliminary applications of pressure and to change pressure slowly, in 

 order to avoid viscosity effects from the transmitting medium. Most 

 of the metals showed no hysteresis, and the pressure measurements 

 might have been made in any order, but they were, as a matter of 

 fact, nearly all made alternately with rising and falling pressure. 



Measurements on a single metal consisted of readings at four 

 different temperatures of the variable bath (25°, 50°, 75°, and 95°) 

 and seven different pressures (0, 2000, 4000, 6000, 8000, 10000, and 

 12000 kg./cm.^). The procedure was usually as follows. The 

 apparatus was set up and seasoned by several preliminary applications 

 of pressure to the maximum at room temperature. The lower bath 

 was then filled with ice and the upper bath adjusted at 25°. E.m.f. 

 measurements were then made at 0, 4000, 8000, 12000, 10000, 6000, 

 2000, and kg. This might occupy two hours. The time required 

 for dissipation of heat of compression after every change of pressure 

 was cut down by secondary variations of pressure, as has already been 

 described in previous papers.^ The upper bath was then changed to 

 50°, the lower one still being packed with ice, and e.m.f. measurements 

 were made at the same pressures as before. This was repeated at 

 75° and 95°. The temperature of the bath was read on a Tonnelot 

 thermometer calibrated at the Paris Bureau of Weights and Measures. 



6 P. W. Bridgman, Proc. Amer. Acad. 49, 14, 1913. 



