ELECTRICAL RESISTANCE UNDER PRESSURE. 83 



evident that some of the irregularities remained at all temperatures. 

 In view of the probably much greater magnitude of the effect at still 

 lower temperatures, no readings were attempted at temperatures 

 lower than 28°. 



The capillary containing the potassium was in the form of a U, with 

 open cups at the two ends. Into each cup, two fine platinum wires 

 were sealed for the current and potential leads, and measurements 

 were made by the potentiometer method. The capillary was filled 

 by melting the potassium into it in high vacuum. The filling was 

 accomplished by sealing one of the cups, and connecting to the other 

 cup a succession of bulbs communicating with each other through 

 narrow necks. The potassium was placed in the most remote of the 

 bulbs, and the apparatus was exhausted with a diffusion pump, heat- 

 ing the bulbs and capillary to remove occluded gases. The glass was 

 now sealed off from the pump, and the melted potassium run in suc- 

 cession from one bulb to the next. In this way the scum of oxide was 

 removed. Previous work had shown that further purification of this 

 particular specimen of potassium, as by distilling, was superfluous. 

 When the melted potassium reached the cup of the capillary, into 

 which it could not enter because of capillary action, illuminating gas 

 was admitted to the farther bulb, driving the melted potassium before 

 it and thus completely filling the capillary. The seal of the other 

 cup was now broken, and the capillary mounted as soon as possible in 

 the pressure apparatus under Nujol, the open ends of the capillaries 

 being protected from oxidation with Nujol and paraffine paste during 

 mounting. 



Five runs were made, with two capillaries. The first filling gave 

 measurements of the resistance of the solid alone at 28.6° and 54.2°. 

 The second filling was used at higher temperatures, 95.7°, 132.2° and 

 167.0°, and gave measurements on both solid and liquid. For the 

 second set of runs a special apparatus had to be used by which the 

 insulating plug was kept cold in a third cylinder. This was first used 

 in the measurements on liquid bismuth, and will be found described 

 under that metal. The same apparatus was also used in the high 

 temperature measurements on lithium and sodium. In addition to 

 the pressure runs, the same fillings of the capillaries were used to give 

 the temperature coefficient of resistance by varying the temperature 

 at atmospheric pressure. 



The resistances as measured were smoothed to uniform temperature 

 and pressure intervals, choosing a temperature interval of 35° as being 

 closest to the greatest number of the actual readings. In this smooth- 



