ELECTRICAL RESISTANCE UNDER PRESSLTRE. 117 



The positive coefficient of tlie solid is therefore presumably due to 

 its crystalline structure. The liquid behaves in other ways also like 

 normal metals. When resistance is plotted against pressure, the curve 

 is convex toward the pressure axis; that is, the pressure coefficient 

 decreases relatively (and also alxsolutely) at the higher pressures. The 

 pressure coefficient is little affected by temperature, within this range, 

 and also the temperature coefficient is little affected by pressure. The 

 initial pressure coefficient of the "observed" resistance at 275° is 

 — O.O4I23, which corrects, using the constants above, to — 0.0.il32 

 for the specific resistance. Both of these coefficients are to be dis- 

 tinguished from the pressure coefficient of the "observed" resistance 

 of a solid. The pressure coefficient of the liquid is of the same magni- 

 tude as that shown by the softer solid metals, such as lead, and is also 

 very nearly the same numerically, although of opposite sign, as that 

 of solid bismuth. 



The temperature coefficient of the "observed" resistance at 275° 

 is 0.00047, which corrects to 0.00048 for the coefficient of the specific 

 resistance. This is about five times less than the value for a normal 

 solid at the same temperature. It is almost always true that the 

 temperature coefficient of the liquid is materially less than that of 

 the solid. 



At 7000 kg. at the equilibrium point, the resistance of the liquid 

 is approximately 45% of that of the solid. At atmospheric pressure 

 Northrup and Sherw^ood ^'^ found for the ratio 43%. There was con- 

 siderable preliminary rounding of their melting curve, so their results 

 are probably not any more accurate than mine, but it is at any rate 

 evident that this ratio does not suffer any large change with increasing 

 pressure. 



Tungsten. In the preceding paper ^^ results were given for the 

 pressure coefficient of resistance of tungsten, but the value of the 

 temperature coefficient of resistance of the sample used was so low 

 (0.00322) that it was prol)able that the tungsten was not very pure. 

 Since the publication of my earlier paper Beckman ^^ has measured 

 the effect of pressure to 1600 kg. on the resistance of a sample of 

 tungsten having a considerably higher temperature coefficient than 

 my original piece, and has found a higher initial value of the pressure 

 coefficient than I did. 



The sample of tungsten on which I previously experimented was 

 the purest which the General Electric Co. was at that time in a posi- 

 tion to ofl'er me. I h&ve since learned that it was probably " doped," 

 that is, thoriated, the impurity of thorium being 0.2 or 0.3%. Through 



