114 BRIDGMAN. 



course if I ever repeat this work I shall make connections by soldering. 



A measurement of the temperature coefficient of resistance showed 

 a discouragingly high probable impurity; the average coefficient 

 between 0° and 95° was 0.00076, whereas Matthiesen ^^ had found 

 0.0038. The high probable impurity of this sample did not make it 

 worth while to make any very extended pressure measurements. 

 One run was made. There were considerable irregularities, but within 

 the limits of error the relation between pressure and resistance is 

 linear to 12000 kg., and the value of the coefficient is —0.05.326. 



This coefficient is similar to that of a number of metals both as 

 regards magnitude and sign. Arsenic is seen therefore to acquire 

 neither the abnormal sign of the coefficient of its neighbors bismuth 

 and antimony on the one side, nor the abnormally high numerical 

 value of the coefficient of black phosphorus on the other. 



Liquid Bismuth. The pressure coefficient of resistance of solid 

 bismuth is abnormal in being positive; it was of particular interest to 

 find whether the same abnormal behavior holds for the liquid. The 

 bismuth used for these measurements was from the same lot of electro- 

 lytic bismuth as that whose pressure coefficient was previously meas- 

 ured. It was melted into a U-shaped fine glass capillary provided 

 with four sealed-in platinum terminals for use with the potentiometer 

 method. Special precautions were necessary to prevent the bismuth 

 from cracking the capillary on freezing; this was accomplished by. 

 very slow cooling from the bottom up after the capillary had been 

 filled with liquid bismuth. In this v^ay congealing ran upward from 

 the bottom of the capillary toward the open top, and no liquid was 

 entrapped by the solid to crack the glass by its expansion on freezing. 



A special arrangement of the pressure apparatus was necessary to 

 permit the electrical measurements. The same arrangement was also 

 used with lithium, sodium, and potassium, but since the apparatus 

 was first used with bismuth, it will be described here. The difficulty 

 was with the insulating plug, which was packed with soft rubber. 

 This would have been carbonized by the temperature of melting 

 bismuth. The pressure apparatus was accordingly constructed in 

 three parts, instead of the customary two. There was an upper 

 cylinder, as usual, in which pressure was produced, and in which was 

 located the measuring coil of manganin wire. This upper cylinder 

 was connected by a stout tube with the cylinder below it, in which was 

 placed the bismuth in the glass capillary. This second cylinder was 

 surrounded with a bath of Crisco, by which the desired temperature 

 was maintained by thermostatic regulation. Out of the bottom of the 



