ELECTRICAL RESISTANCE UNDER PRESSURE. 115 



second pressure cylinder was led another piece of stout tubing, which 

 passed through a stuffing box in the bottom of the Crisco bath, and 

 below the bath connected with a third pressure cylinder. This third 

 cylinder was kept cool by a bath of water at room temperature. This 

 bath was stirred to maintain the temperature uniform, but it was not 

 necessary to regulate the temperature thermostatically. In the lower 

 pressure cylinder was situated the insulating plug, of the same design 

 as used in all the work with the potentiometer method. The plug was 

 connected with the bismuth in the second cylinder by four insulated 

 leads brought down through the pipe connecting the second and third 

 cylinders. The insulation of these wires was asbestos; asbestos 

 covered copper wire is now a commercial product. In this way the 

 insulating plug was kept cool, so that there was no danger of leakage 

 or failure of insulation because of the high temperature. The only 

 trouble to be anticipated was large parasitic e.m.f.'s because of the 

 large differences of temperature, but the parts in which there were 

 temperature gradients were composed of electrically homogeneous 

 material, and no more trouble was found from this effect than at 

 ordinary temperatures. 



Runs were made on liquid bismuth at 274.6°, 260.0°, and 239.6°, 

 in this order. For fear of damaging the capillary (fused in platinum 

 leads almost always make trouble under pressure) the pressure was 

 not raised to the maximum of this work, 12000 kg., until the last run, 

 so that I did not obtain data for the resistance of the liquid over the 

 entire possible range. At 239.6°, however, pressure was run to the 

 maximum with no bad effects. After the measurements on the liquid, 

 the bismuth was allowed to freeze under pressure, and measurements 

 were attempted on the solid. The results for the solid were not good, 

 however, probably because of strains introduced on freezing in the 

 fine capillary. There was no way of controlling the freezing under 

 pressure and making it take place from the bottom up as had been 

 possible in initially setting up the apparatus. The effects of strains 

 were apparent in two ways; the pressure coefficient of resistance of 

 the solid was negative over part of the range below the solidifying 

 point, whereas that of the unconstrained solid is positive, and the 

 freezing point was depressed a couple of degrees, which is in the 

 direction to be expected if there are internal strains. Irregularities 

 introduced by these strains are of importance, however, only when 

 it was desired to obtain the relatiA'e changes of resistance with changes 

 of pressure, and it was possible to find a value for the change of resist- 

 ance on solidifaction which should not be greatly in error. 



