ELECTRICAL RESISTANCE UNDER PRESSURE. 103 



a capillary of very thin glass with the liquifl, exposing to a pressure 

 sufficient to freeze it, and making measurements on the solid con- 

 tained in the glass without allowing the pressure to fall low enough to 

 melt the mercury. By the use of thin glass I hoped to eliminate error 

 due to the constraining action of the container. It was necessary to 

 use a capillary of rather large section, because of the mechanical diffi- 

 culties of blowing and handling a very fine capillary with excessively 

 thin walls, and the potentiometer method of measurement was 

 therefore used. The leads were of fine platinum sealed through the 

 glass. Under pressure the glass cracked around the seals, but of 

 course this introduced no error in the measurements of the solid. 

 Only one set of measurements was made on the resistance of the solid, 

 at 0°. Measurements were made on the liquid up to the freezing 

 pressure, pressure was then increased very cautiously beyond the 

 freezing point so that freezing took place slowly, and the solid so 

 formed was seasoned by an exciu'sion to 12000 kg. and back nearly 

 to the freezing pressure. Readings were now made in the domain of 

 the solid to 12000 kg. These readings were exceedingly regular; they 

 showed no departure from the smooth curve within the sensitiveness 

 of measurement, which amounted to one part in 1500 on the total 

 effect for the range 7640 to 12000 kg. This was gratifying, because 

 it showed that the very thin glass exerted no perceptible constraining 

 effect. In the prcA'ious work with the lic^uid in heavy glass capil- 

 laries A'cry irregular results were found after the metal had frozen. 



The relative values for the resistance of the solid are shown in 

 Table X in terms of the resistance of the solid at 7G40 kg. and 0° as 

 unity. In comparing these values with those of the liquid it must be 

 remembered that the values for the solid are relative values of the 

 "observed" resistance, and must be corrected by a factor equal to the 

 linear compressibility in order to give relative values of specific resist- 

 ance. The pressure coefficient of "observed" resistance may be 

 found from the table to be —0.04236, and within the sensitiveness of 

 the measurements it is constant over the pressure range from 7640 to 

 12000. This value for the pressure coefficient is somewhat higher 

 than the minimum value set in the previous work, which was — O.O52. 

 The accuracy of the previous work for the solid was so low that it was 

 stated that the' maximum A-alue for the solid might not impossibly be 

 ten times the minimum. The pressure coefficient of the solid is very 

 nearly that of the liquid, which is — O.O5224 at 6500 kg., when corrected 

 by one third the \-olume compressibility of the liquid mercury so as 

 to be strictly comparable with the value for the solid. It is surprising 



