EFFECT OF PRESSURE OX COXDUCTIVITY OF METALS. 99 



reached equilibrium, the thermo-couple reached equilibrium in a 

 fraction of a minute in the radial flow method, and in at most two or 

 three minutes in the longitudinal flow method. 



While the bath was coming to temperature equilibrium, I usually 

 made a preliminary application of 12000 kg. to be sure that the joints 

 were tight, and that the electrical insulation resistance between the 

 heating element and the thermo-couple circuit was sufficiently high. 

 The insulation resistance was measured independently of the potentio- 

 meter, and had in all cases to be as high as 100 megohms if satisfactory 

 results were to be obtained. The pressure transmitting medium in all 

 these experiments was petroleum ether. The use of this was neces- 

 sary to avoid stresses in the couples introduced by viscosity effects. 

 The mechanical difficulties of retaining this liquid under pressure 

 without leak are very materially greater than for the mixture of ether 

 and kerosene which I had used before. Slight mechanical imperfec- 

 tions in the bearing surfaces become much more important, so that the 

 preliminary test for leak on the whole saved time. There did not 

 seem to be any seasoning effect proper produced in the specimen itself 

 by the initial application of pressure, as indeed there ought not to be 

 if the specimen is homogeneous and well annealed. In a number of 

 cases when the radial flow method was used and the specimen was 

 mounted in a copper sheath, the specimen was seasoned by several 

 applications of the maximum pressure before any of the electrical 

 connections were made. Seasoning in this case was necessary because 

 of unequal compressibility of the specimen and the copper of the 

 sheath. 



After reaching equilibrium at atmospheric pressure, the pressure 

 zero was read. The magnitude of the current in the heating element 

 and its shunt was then read on the potentiometer, both for the direct 

 and the reversed direction of the heating current. The reversal of the 

 heating current was accomplished with a double pole double throw 

 switch and took place so quickly that there was no appreciable inter- 

 ruption in the heat input. The mean of readings with du-ect and 

 reverse heating current eliminates the effect of any parasitic thermal 

 e.m.f.'s in the heating circuit. The final reading made was that of 

 temperature difference on the thermo-couple. This reading was also 

 made with both direct and reverse direction of the heating current. 

 Any error from electrical leakage from the heating circuit (in which 

 the current is of the order of 0.5 amperes) into the thermo-couple 

 circuit (in which the e.m.f. is of the order of a few microvolts) because 

 of defective insulation, is thereby eliminated. The reversal of [the 



