EFFECT OF PRESSURE ON CONDUCTIVITY OF METALS. 83 



tion of the cylinder under internal pressure. If the specimen does not 

 remain accurately centered in the cylinder at all times the flow of 

 heat becomes unsymmetrical and the method is vitiated. Originally 

 I endeavored to meet this condition by surrounding the specimen with 

 a massive cylindrical sheath of copper, approximately | inch thick. 

 The specimen itself was | inch in diameter and 2.5 inches long, and the 

 interior diameter of the pressure cylinder was about | inch. The 

 crack between specimen and pressure cylinder was of the order of 

 0.005 inch initially, and it might increase by double under the extreme 

 pressure. The use of the copper sheath is applicable only to those 

 substances with low conductivity compared with copper. It might 

 be expected to work best for lead and bismuth, and to work less well 

 for tin and zinc. 



Many variations were tried on this method. The chief difficulty 

 proved to be that of getting good thermal contact between the speci- 

 men and the copper sheath. I at first tried to cast the easily fusible 

 metals into the sheath. Unexpected difficulties were encountered in 

 getting good enough contact. Casting into the bright sheath in air 

 always introduced a film of oxide. Casting into the sheath in vacuum 

 was tried, and also casting after making a preliminary coating of the 

 interior of the sheath with solder or tin or the particular metal in 

 question or silver plating, both in vacuum and a protecting atmosphere 

 of illuminating gas. Satisfactory results were not obtained with any 

 metal except lead. It was a surprise that it was so difficult to make 

 good contact with tin. 



A difficulty anticipated with this method because of the unequal 

 compressibility of the copper of the sheath and the metal of the core 

 did not turn out to be formidable; this difficulty could be turned by 

 drilling the sheath with a number of fine holes so as to allow direct 

 access of the pressure to the interior. 



The device was tried of splitting the sheath into two pieces at a plane 

 passing through the axis. In this way the inside of the sheath could 

 be thoroughly tinned, as also the outside of the specimen, and by 

 squeezing the specimen between the two halves of the sheath while 

 still hot, good thermal contact could be secured initially. But with 

 this arrangement the unequal compressibility of sheath and core was 

 prohibitive; there was a progressive change with each application of 

 pressure, the two halves of the sheath separating and working loose 

 from each other. With tin, for example, a progressive change in the 

 apparent effect of pressure on the thermal conductivity of three fold 

 was produced by three applications of pressure. 



