EFFECT OF TENSION ON CERTAIN ABNORMAL METALS. 49 



balance connected to the antimony by a wire passing through a stuffing 

 box in the side of the oil bath. This stuffing box was very loose, and 

 was without appreciable friction. The current and potential termi- 

 nals were soldered to the antimony wire. It is very difficult to make 

 a soldered connection to this brittle material which shall be sufficiently 

 in axial alignment to permit the application of an appreciable 

 tension. The wires by which the tension was applied were cemented 

 to the antimony wire by DeKhotinski cement. 



Readings were made on three different samples; those on the first 

 were the best. The diameter of the wire was 0.0146 inches. Three 

 runs were made on it; the first two were rough in character and agreed 

 within the limits of error with the third run. Within the limits of 

 error the effect is linear with tension, and is positive, the resistance 

 increasing with increasing tension. The maximum load applied to this 

 specimen without rupture was 80 gm.; it broke at 90 gm. Readings 

 were made at eight loads. The greatest departure from the straight 

 line of any single observation was 8% of the maximum effect, which 

 was a change of resistance of 0.025 %. The tension coefficient of this 

 sample was +4.5 X 10~'' for 1 kg/cm^ tension. 



The second sample was extruded at a different time, and was ex- 

 cessively fragile. Its diameter was 0.024 inches, and it broke at a 

 load of 20 gm. Only two readings were made; not enough to estab- 

 lish the linearity of the effect or to eliminate chance errors. The 

 tension coefficient which would correspond to the mean of the readings 

 with this sample was + 20.0 X lO-'^. 



The third sample was 0.0295 inches diameter. It was much stronger 

 mechanically than the second sample, and allowed loads beyond the 

 elastic limit to be applied. The results were much less regular than 

 for the first sample, however. Within the limits of error the effect is 

 linear with tension up to a load of 140 gm. Ten readings were made; 

 the worst of these departed from the smooth curve by 33% of the 

 maximum effect. The tension coefficient shown by this sample was 

 + 7.5 X 10~^ for a tension of 1 kg/cm-. 



In estimating the most probable coefficient, the first sample must be 

 given considerably more weight; I take as the most probable coeffi- 

 cient +5.0 X 10-«. 



I have previously determined Young's modulus for antimony ;^'' 

 The value 7.8 X 10^^ Abs. C. G. S. units was found. It was possible 

 to definitely establish that the wire was not homogeneous, because the 

 rigidity bears an impossible relation to Young's modulus. It is there- 

 fore not allowable to apply the formula of elasticity to compute 



