EFFECT OF PRESSURE ON CONDUCTIVITY OF METALS. 115 



was now so kind as to make a quantitative determination of the silver, 

 and found 0.03%, confirming the conservative estimate of the U. S. S. 

 Metals Refining Co. I now succeeded in finding a small residue of 

 my original electrolytic bismuth, and Professor Saunders made a 

 spectroscopic analysis of this. He could find only traces of Cu and 

 Pb, the Cu being stronger than in the commercial electrolytic bismuth, 

 and some silver, evidently considerably less than in the commercial. 

 The conclusion seems forced that a quantity of silver as small as 

 0.03% can depress the temperature coefficient to half the normal 

 value, thus exerting an effect very much greater than such impurities 

 as Pb and Sn, which are present in ordinary commercial bismuth. 

 That difficulty would be expected in removing the silver by recrystalli- 

 zation is evident on an inspection of the mixed crystals diagram for 

 these two metals. This would also be indicated by the energetic 

 alloying of silver and bismuth, which made impossible the preparation 

 of the radial flow specimens. 



Under the circumstances it seemed that the best thing to do was to 

 use the commercial electrolytic bismuth, with its known analysis of 

 0.03% of silver, in the expectation that the effect of this small impurity 

 is abnormally high on the temperature coefficient of resistance. I 

 had previously found that the effect of impurity on the pressure 

 coefficient of resistance is much less than on the temperature coefficient 

 of resistance. 



The samples were made from | inch wire which had been formed 

 by hot extrusion in the regular way. One advantage of forming the 

 specimen by extrusion is that the crystalline structure is very much 

 finer than when the specimen is cast, and so the results are much more 

 likely to give the average for all the directions of a single crystal. 



The thermo-couple holes were drilled in these specimens in the 

 regular way, but a modification was necessary in mounting the heating 

 element. Previously the heating element was mounted in a copper 

 capsule, which was cemented into a hole drilled in the end of the 

 specimen. This was no longer possible, because the capsule was so 

 large that it was not possible to drill a hole to receive it without 

 breaking out the walls in so brittle a material as bismuth. The 

 heating element was accordingly placed in a smaller hole drilled 

 directly in the end of the specimen. This has the disadvantage that 

 the terminal conditions of temperature are not so accurately defined 

 as with the other metals, and the motion of the heating element in 

 its receptacle may produce other irregularities. This was indeed the 

 fact; the points were more scattered than with the other metals, and 



