122 BRIDGMAN. 



The effect of pressure is seen to be a large increase of conductivity, 

 amounting at 12000 kg. to an increase of 2.22 fold. The increase is 

 not linear with pressure, but there is a departure from linearity in the 

 normal direction, in that the change becomes proportionally less at 

 the higher pressures. 



The initial rate corresponds to an increase of conductivity of about 

 20% per thousand kg. So far as order of magnitude goes, this agrees 

 with Lussana, who found the correction for his transmitting medium 

 to be at the rate of 30% per thousand kg. He did not find a departure 

 from linearity. Of course there is no reason to expect more than 

 agreement as to order of magnitude, because his transmitting medium 

 was a comparatively heavy oil, quite different in properties from mine. 



The measurements of the effect of pressure on the thermal conduc- 

 tivity of liquid is a thing worth doing for its own sake, and I hope to 

 get the chance to make measurements on a number of others. In fact 

 I already have results for two alcohols and kerosene. Suffice it here 

 to mention that there seems to be an intimate connection between the 

 pressure effect on thermal conductivity and the pressure effect on the 

 velocity' of propagation of sound. 



General Comment on Lussana's Results. 



The only previous measurements of the effect of pressure on thermal 

 conductivity are those of Lussana. Since his results often differ 

 essentially from mine, even as to sign, and since this is a master of 

 considerable importance for theoretical considerations, some critical 

 survey of his results seems called for. In general, Lussana finds that 

 the thermal conductivity of all metals increases under pressure, and 

 this increase is nearly the same as that of the electrical conductivity, 

 so that the Wiedemann-Franz ratio remains nearly constant under 

 changes of pressure. 



Lussana's method was an adaptation to high pressures of one origi- 

 nally due to Depretz and Biot. A long bar of metal has a source of 

 heat at one end and is immersed in a medium through which the heat 

 may flow away laterally. The temperature of the bar, which is as- 

 sumed constant across the section, is measured at three equi-distant 

 points along it, and in terms of the two differences of temperature thus 

 obtained, a relation can be found between a certain geometrical factor 

 and the ratio between the thermal conductivity and the lateral con- 

 ductivity into the surrounding medium. The essential difference 

 between this method and mine is that in mine there is a source at one 

 end of the bar and a sink at the other, so that nearly all the heat input 



