50 BRIDGMAN. 



Poisson's ratio. The above Young's modulus, combined with Rich- 

 ard's value for the compressibility, 2.4 X 10~^- C. G. S.,® would give 

 0.18 for Poisson's ratio, which is improbably low. I shall in the fol- 

 lowing use 0.30, which is an average value for many metals, as more 

 probably correct. 



The third of the specimens above allowed some examination of the 

 behavior beyond the elastic limit. Permanent increases of resistance 

 were produced by loads in excess of 140 gm. ; rupture took place at 270 

 gm. The phenomena were somewhat unusual in that there were no 

 time effects. The permanent change of resistance under a given load 

 assumed at once its final value, and there was no creep, as is usually 

 the case. This does not seem very surprising in so brittle a material. 



Bismuth. Measurements were made on three different grades of 

 material, of three different grades of purity. The first was ordinary 

 commercial metal, which has about 3% impurity. The second was 

 electrolytic bismuth, for which I am indebted to the kindness of the 

 United States Metals Refining Co. x\nalysis showed only 0.03% 

 impurity of silver, and only traces of anything else. In spite of this 

 very small impurity, however, the temperature coefficient of resistance 

 is only about one half normal; apparently the silver exerts some very 

 large specific effect. The third sample was electrolytic bismuth of my 

 own preparation, which I had made several years ago for a determina- 

 tion of the pressure coefficient of resistance. ^^ I have no chemical 

 analysis, but spectroscopic analysis by Professor F. A. Saunders shows 

 less impurity of silver than the other electrolytic bismuth. I was not 

 able to repeat the preparation of this material, and had only a small 

 quantity available. I verified the high temperature coefficient on the 

 special piece used in this work. The question of the curious behavior 

 caused by the small quantity of silver is more fully discussed in my 

 paper on the effect of pressure on thermal conductivity. 



Tension decreased the resistance of all the samples of bismuth. The 

 effect is comparatively large, and with little care it was possible to 

 obtain very good sets of readings, with deviations from the smooth 

 curves by individual points of not more than 1%. The effects beyond 

 the elastic limit are complicated. There is a very considerable initial 

 range, however, within which the eft'ect is linear, and there is no evi- 

 dence for departure from the usual relations of perfect elasticity. The 

 coefficients quoted in the following were determined within this range. 



Three determinations were made on commercial bismuth at room 



11 Reference 4, p. 624. 



