EFFECT OF PRESSURE ON CONDUCTIVITY OF METALS. 81 



involved a knowledge of current and resistance, and of temperature 

 difference, which demanded a measurement of the electrO-motive 

 force of a thermo-couple. Since it was necessary to find only the 

 relative changes in thermal conductivity, a source of heat of unknown 

 magnitude would have been sufficient, provided that it remained 

 perfectly constant. The arrangement of the circuits, to be described 

 in the next paragraph, ensured approximate constancy of the heat 

 input, so that the measurements of the heating current degenerated 

 to check readings from which a small correction was determined. The 

 measurements of both heating current and e.m.f. were made on the 

 same potentiometer as was used in previous measurements of the 

 effect of pressure on thermo-electromotive force.'* This apparatus 

 has already been described in sufficient detail. By an arrangement 

 of suitably protected switches either the thermo-couple or a small 

 potential tap from the heating circuit could be connected in place of 

 the couple as previously used. The tap in the heating circuit was 

 constructed of heavy manganin wire of a resistance of approximately 

 1/13000 ohm. 



Correction for the change of resistance of the heating element with 

 pressure was avoided by using a similar heating element in shunt with 

 the one exposed to pressure. The shunt unit was mounted in the 

 same thermostated bath as the pressure cylinder, but was not exposed 

 to pressure, so that its resistance remained constant during changes of 

 pressure. By writing down the equations for the divided circuit, one 

 can see in an instant that if the total input of current into the two 

 heating elements in parallel is maintained constant, the rate of genera- 

 tion of heat in that one of the two elements which is exposed to small 

 fluctuations of resistance remains constant. The total input of cur- 

 rent was maintained approximately constant by using for the source 

 of supply a storage battery in series with a ballast lamp (iron filament 

 in an atmosphere of hydrogen) obtained through the courtesy of the 

 General Electric Company. The total fluctuation of resistance in 

 series with the lamp due to changes of pressure was only 0.08%, and 

 wath constant e.m.f. the lamp reduces the fluctuations of current due 

 to this to negligible proportions. However, the battery itself was not 

 always sufficiently constant, and in order to determine the small 

 correction due to its fluctuations, the total current delivered by the 

 battery was measured by determining the current in the two branches 

 of the shunt. This was done with the potentiometer and a low re- 

 sistance tap, precisely as for the heating circuit. The total corrections 

 due to fluctuations in the heat input were always a small fraction of 1%. 



