126 BRIDGMAN. 



and the electrical conductivity increases by 1.8%. The sum of these, 

 16.3%, is an upper limit which the fractional part of the total con- 

 ductivity due to the non-electronic part must not exceed. Now the 

 theoretical value for the Wiedemann-Franz ratio is 4.3 X 10^" (Lo- 

 rentz's value), and the experimental value for nickel is 6.99 X 10^". 

 This allows the possibility of 39% of the thermal conductivity initially 

 being of non-electronic origin, which is more than twice the extreme 

 set experimentally. It would seem that under these conditions, when 

 so comparatively large a part of the atomic conductivity has been 

 wiped out by pressure, that the relation between conductivity and 

 pressure must depart from linearity. The experimental accuracy 

 was not great enough, however, to show such a departure. 



Further consideration of the theoretical significance of these results 

 is reserved for a forthcoming paper in the Physical Review. 



Summary. 



Two methods are described for measuring the thermal conductivity 

 of metals under pressure. The first of these is a radial flow method, 

 which has many theoretical points of advantage, but is of limited 

 applicability in practise because of the difRculty of getting metals in a 

 condition of sufficient homogeneity. The second is a longitudinal 

 flow method, the essential of which is the small size of the specimen. 

 The irregularities of the individual readings by the second method are 

 greater than by the first method, but the* effect of inhomogeneities is 

 less and different specimens of the same metal will give the same result. 



Measurements of the effect of pressures to 12000 kg/cm- on the 

 thermal conductivity of 11 metals have been made by one or the other 

 of these methods. The effect may be either positive or negative, and 

 is more often negative than positive. In only two cases, lead and tin, 

 does the Wiedemann-Franz ratio • increase under pressure; for the 

 other metals it decreases, and sometimes by large amounts. In addi- 

 tion to the metals, the pressure coefficient of thermal conductivity of 

 petroleum ether has been measured. The conductivity increases by a 

 factor of about 2.2. 



The only previous measurements have been by Lussana, who ob- 

 tained results entirely different from those found here. His method is 

 criticised in some detail, chiefly on the basis of the uncertain correc- 

 tion for the lateral loss of heat to the pressure transmitting medium. 



These results indicate that a fairly large part of thermal conduction 



