PRESSURE ON RESISTANCE OF METALS. 625 



sample alloyed with the solder and the leads dropped off, so that the 

 run at 100° was not made; the connections to the two impurer samples 

 were undamaged at 100°. The initial resistances at 0° were 3.8, 4.0, 

 and 6.9 ohms respectively with decreasing purity; the lengths of the 

 wires w^ere in all cases about the same. 



The effect of pressure on bismuth is abnormal, as it is for antimony, 

 the pressure coefficient being large and positive instead of negative 

 as usual. The positive coefficient was shown by all three samples. 

 The most marked effect of impurity is that after a cycle of changes 

 of temperature or pressure resistance does not return to its original 

 value. Furthermore, the runs at constant temperature show great 

 pressure hysteresis; this hysteresis becomes less as the impurity 

 decreases. At 0° the most impure specimen showed a hysteresis of 

 12% of the total effect, and a permanent change of resistance after 

 the run of 8% of the effect. The hysteresis is without doubt due to 

 internal viscosit}', which prevents the constitution of the mixed crystals 

 keeping exact pace with the changes of pressure. This is suggested 

 strongly by the fact that after every change of pressure an abnormally 

 long time elapsed before the resistance became approximately con- 

 stant. This view is also consistent with the observation that at 

 higher temperatures the hysteresis rapidly became less. On making a 

 run at 0° again after the run at 100° the original hysteresis had returned 

 to almost its original value; the decreasing hysteresis at higher tem- 

 peratures cannot, therefore, be an effect of accommodation. The 

 average pressure coefficient between and 12000 kg. decreased from 

 +O.O428I at 0° to +O.O42OI at 100°. The sample of grade "K" 

 showed much less hysteresis, and did not require an appreciably 

 longer time than any other metal to reach equilibrium after a change 

 of pressure. The maximum permanent change of zero was 4% of the 

 pressure effect at 0°, and the maximum hysteresis was also at 0° and 

 was 3.5% of the total effect. The average pressure coefficient between 

 and 12000 kg. decreased from +0.04269 at 0° to +O.O42O5 at 100°. 



The electrolytic bismuth was seasoned l)efore the runs by several 

 excursions between 0° and 130° at atmospheric pressure and by one 

 application of 12000 kg. at 0°. It showed no peculiarity of behavior 

 suggesting impurity except a hysteresis amounting at the most to 3% 

 of the total effect at 0°. The permanent change of zero after a run was 

 not over 0.3% of the total effect, which is not larger than found for 

 several other metals, and no longer time was required for the attain- 

 ment of steady conditions than was necessary for dissipation of the 

 heat of compression. The curves of actual resistance against pressure 



