110 BRIDGMAN. 



the best of the specimens gave the higher results. Because of the 

 high conductivity of copper, the correction for the transmitting 

 medium is small, being only 1.2%. 



The pressure coefficient of thermal conductivity deduced from the 

 data above is — 0.0o75. No departure from linearity could be de- 

 tected. I have found for the effect of pressure on electrical con- 

 ductivity over this range the mean value +O.O5I8I. Lussana's 

 results for copper run only to 1000 kg. He found the effect to be 

 linear, and the conductivity to increase under pressure, the opposite 

 sign from my results. His coefficient is +O.O5IO. The relation be- 

 tween pressure and electrical resistance he also found to be linear, and 

 the coefficient to be O.O5212. 



Silver. Only the longitudinal flow method was used on this metal. 

 The material was obtained from Baker, and was said to be of high 

 purity, but I have no analysis. Three different samples were used. 

 The two runs made on the first two samples apparently indicated an 

 increase of conductivity- under pressure. The reason for this has been 

 discussed in detail previously; not enough readings were taken, and 

 there was a tendency for the points to shift from a lower lying to a 

 higher lying curve with increase of pressure, thus simulating a false 

 effect. IMore readings were made on the third sample, and an effort 

 made to control the position of the points on one or another of the 

 possible lines. The results with this sample are reproduced in Figure 

 8. The tendency of the points to lie on one or another of three dis- 

 tinct lines is obvious, as also the fact that the effect is negative, the 

 conductivity decreasing with increasing pressure, instead of increasing, 

 as was indicated by the results first obtained. The two first samples 

 were now set up again, and the measurements repeated, with the 

 precautions which had been gained from the intervening experience. 

 The results shown by these two samples were now also unmistakably 

 negative, and of nearly the same numerical value as shown by the 

 third sample. The numerical magnitude of the total decrease of 

 conductivity under 12000 kg/cm^ shown by the three samples was 

 4.1, 4.4, and 4.6% respectively. The correction for the pressure 

 effect on the transmitting medium was 1%. 



The pressure coefficient of thermal conductivity given by the above 

 data is —O.O^Sl. The average pressure coefficient of electrical 

 conductivity between and 12000 kg. I have found to be +0.05334. 

 This metal was not investigated by Lussana, so there are no previous 

 results for comparison as to the thermal conductivity. 



Nickel. Both methods were used for this metal, and material from 



