100 BRIDGMAN. 



heating current does not eliminate the effect of spurious e.m.f.'s in the 

 thermo-couple circuit. There were no such spurious e.m.f.'s of de- 

 tectible magnitude in the circuit in the absence of the heating current, 

 as could be checked by the complete vanishing of the thermal e.m.f. 

 on breaking the heating current. The only assurance against spurious 

 e.m.f.'s introduced by the heating current itself was in the careful 

 design of the apparatus. T could think of no way of making a direct 

 check of this point without being able to introduce a cooling equal to 

 the heating. In addition to the measurements above, as a rough 

 check, the total current output from the storage battery into the 

 ballast lamp was read on a commercial ammeter. 



The pressure was then changed, usually by a step of 2000 kg., and 

 after attaining equilibrium, the same succession of readings was again 

 made. An interval of about 15 minutes was usually sufficient for 

 equilibrium. This time is required to dissipate the heat of com- 

 pression in the liquid and to again attain complete equality of tempera- 

 ture at the outer and inner ends of the three terminal plug. The 

 usual routine consisted of two complete sets of readings to 12000 kg. 

 and back. The results were plotted roughly as soon as obtained, 

 making correction for any slight changes in the heating current and 

 for fluctuation in the temperature of the copper coils of the potentio- 

 meter. This last correction was usually not necessary. When using 

 the longitudinal flow method, after the cause of the irregularity of the 

 readings was discovered, the plotted results were used to best direct 

 the subsequent changes of pressure. The final computation of the 

 results and the application of the other corrections which have been 

 described above was made later. 



The detailed data for the different metals follows. 



Detailed Data. 



Lead. More readings were made on this metal than on any other, 

 and in general the results are better and the final result is probably 

 more accurate than for any other. Nearly all the readings were made 

 by the radial flow method, to which this metal is best adapted, be- 

 cause of its comparatively low conductivit3^ 



Three grades of metal were used. A first rough trial was made with 

 ordinary commercial lead. Several of the early measurements, in 

 which a copper sheath was used, were made with so-called test lead 

 from Eimer and Amend. The majority of the measurements, how- 



