EFFECT OF PKESSTJRE ON CONDUCTIVITY OF METALS. 113 



separated by much less than 5%, as is to be expected. A partial 

 run with this same specimen, which had to be discontinued because 

 of leak and also because of short circuit in the three-terminal plug, 

 gave exactly the same slope for those readings which could be obtained 

 as the final run. The change shown by this second sample was a 

 decrease of conductivity of 14.5% for 12000 kg., agreeing fairly well 

 with the first sample. Since the second sample gave somewhat more 

 regular results, it is given greater weight in the mean, which I take as 

 14.1%. The total correction for the effect of pressure on the trans- 

 mitting medium was 5% of the total conductivity, amounting to 

 about 33% of the observed change under 12000 kg. 



The results found above give for the pressure coefficient of thermal 

 conductivity — O.O4I2. There are no previous results for comparison. 



An incidental result obtained from the measurements with com- 

 mercial and pure nickel was a comparison of the absolute thermal 

 conductivities. The longitudinal flow method is not well adapted to 

 give the absolute conductivity because of the uncertainty in the cor- 

 rections for loss through the leads, etc. (the absolute conductivities 

 directly calculated average about 5% higher than the values of Jaeger 

 and Diesselhorst), but the comparative values of absolute conductivity 

 of different materials should be nearly correct. The thermal conduc- 

 tivity of the two samples of pure nickel was found to be 37% higher 

 than that of the two samples of commercial nickel. Considerable 

 confidence may be put in these values, as the individual readings were 

 very consistent; the two samples of pure nickel gave results differing 

 by less than 0.5%, and the results on the two samples of commercial 

 nickel were identical to three significant figures. 



Platinum. Measurements were made on two different samples by 

 the longitudinal flow method. The material was obtained from 

 Baker, the purest which they could supply. I have no analysis, but 

 have the statement of Baker that the impurity is guaranteed to be 

 less than 0.1% and is probably not greater than 0.01%. 



The measurements on platinum were scattered on three lines of a 

 maximum separation of about 5%, as is usual with this method. 

 The two specimens gave identical results, a decrease of conductivity 

 of 1.9% for a pressure change of 12000 kg/cm^. The observed effect 

 was positive, but the correction for the effect of the transmitting 

 medium is so large, 5.2%, as to alter the sign of the result. 



The pressure coefficient of thermal conductivity as given by the 

 above measurements is — O.O5I6. The pressure coefficient of electrical 

 conductivity at 30° between 0° and 12000 kg. is +O.O5I86. 



