110 BRIDGMAN. 



the unit of resistance is the resistance of the soUd at atmospheric 

 pressure and 0° C. There may be some question as to the accuracy 

 of the value for the ratio of resistance of soUd to liquid, but this can- 

 not affect the relative accuracy of the values listed in the table for 

 the resistance of the liquid. 



The values listed in the table have been corrected for the thermal 

 expansion and compressibility of the glass, so that the values are 

 proportional to the specific resistances. In making this correction the 

 thermal expansion of the glass was assumed to be O.O58, and the linear 

 compressibility O.Oe?. The correction for thermal expansion is small, 

 and any error in the value assumed for the coefficient cannot appreci- 

 ably affect the results, but because of the smallness of the pressure 

 coefficient of resistance, the correction for the compressibility of the 

 glass is rather important. The correction for compressibility, as given 

 in the table, amounts initially to 1 1% of the pressure coefficient. The 

 compressibility of different varieties of glass varies a good deal; it 

 is perhaps conceivable that the value assumed is as much as 20% 

 different from the correct one, so that the possibility must be recog- 

 nized that the values given in the table for the pressure coefficient of 

 resistance may be in error by as much as 2%. However, any correc- 

 tion of this sort will not affect the relative curvature, since the com- 

 pressibility of glass is sensibly linear, and it seems justified to retain 

 all the significant figures of the table. The eft'ect of the corrections 

 for the glass is to increase both the observed temperature and pressure 

 coefficients of resistance. 



The average pressure coefficient of the liquid to 12000 kg. is seen 

 to vary little with temperature. In absolute value it is somewhat 

 less than one half that for bismuth, and one sixth that for liquid mer- 

 cury. The curvature is in the normal direction, that is, the coefficient 

 becomes less at the higher pressures. 



The pressure coefficient of resistance of the solid was measured 

 only at 0°; at this temperature the entire pressure range of 12000 kg. 

 was aA'ailable, whereas at higher temperatures the pressure range was 

 restricted by the melting, and the accuracy was proportionally less. 

 The absolute resistance of the specimen was low, so that measurements 

 could not be made with as much accuracy as usual. Beside the final 

 run from which the tabulated results were taken, several rougher runs 

 were made with other samples; these agreed with the final results 

 within the accuracy of the measurements. Within the limits of error 

 the relation between resistance and pressure is linear at 0° to 12000 

 kg., and the coefficient is — 0.05247. The maximum departure of 



