EFFECT OF PRESSURE ON CONDUCTIVITY OF METALS. 107 



would be lowered by about 5% if my correction for the effect of pressure 

 on the thermo-couple is applied. Lussana applied no such correction. 



Zinc. Runs were made on three samples by the radial flow method, 

 and two by the longitudinal method. The radial flow results were 

 valueless; there is considerable difficulty in making homogeneous 

 castings. The longitudinal flow specimens were extruded hot, and 

 should be sufficiently homogeneous. The material was obtained from 

 the Bureau of Standards, one of their melting point samples, and had 

 the following analysis: Fe .005%, Pb .0004%, Cd .0018%o, As, S 

 trace; Sb, Sn none; Zinc (by difference) 99.993%. Melting point 

 419°. 4. The results obtained with the two longitudinal samples were 

 scattering, the points lying on several distinct lines, as already ex- 

 plained, but the correct slope could be picked out with some assurance. 

 The change produced by 12000 kg. was 2.4% for one sample, and 2.7% 

 for the other. Take as the mean 2.5%. The correction for the pres- 

 sure effect on the transmitting medium was 3.0%, reducing a directly 

 observed effect of 5.5% to 2.5%. 



The pressure coefficient of thermal conductivity given by the 

 above is -)-0.0521. I had previously found the pressure coefficient of 

 electrical resistance for a pressure range of 12000 kg. to be —0.05463. 

 Lussana, over a pressure range of 2600 atmospheres, found the effect 

 of pressure on thermal conductivity to fall off slightly from linearity, 

 and the mean coefficient to be O.O54I. The pressure coefficient of 

 electrical resistance he found to be — O.O56O2; this relation he found 

 to be sensibly linear. 



Iro7i. Two samples were made for the radial flow method, and four 

 for the longitudinal. The radial flow results were unsatisfactory. 

 The thermo-couples and heating element were placed in copper tubes 

 sweated into larger holes drilled in the specimen. The results were 

 irregular, and the irregularities repeated themselves, showing that the 

 effect is real, and denotes some defect in the specimen, not in the 

 accuracy of the measurements. 



Two of the longitudinal flow specimens were made from American 

 Ingot Iron, from the same piece as the wires whose pressure coefficient 

 of resistance and thermal e.m.f. have been previously measured.* 

 The total impurity in this iron was not over 0.03%. The other two 

 samples were made from a small sample of electrolytic iron of high 

 purity for which I am indebted to the Bureau of Standards. The 

 analysis of this iron is as follows: 



Carbon 0.005 per cent. 

 Silicon 0.007 " 

 Sulfur 0.011 " 



