EFFECT OF TENSION ON CERTAIN ABNORMAL METALS. 65 



was at first thought to be continuous through a transition point is 

 found, on increasing the purity, to be really discontinuous. It is true 

 that the direction of the difference of temperature coefficients found 

 by Werner is the reverse of that demanded by the above explanation, 

 but in view of the extreme sensitiveness of the temperature coefficient 

 to small impurities, I do not believe that a great deal of importance 

 should be attached to this. 



In support of the suggested explanation is in the first place the fact 

 that neither iron nor cobalt show similar effects. These metals and 

 nickel are similar in many respects, but are unlike in regard to their 

 transitions. Cobalt does not have any polymorphic transitions, and 

 the first transition of iron is at so much a higher temperature than that 

 of nickel that it may well be without effect. The decrease of the 

 tension of the minimum resistance of nickel with increasing tempera- 

 ture is also in accord with this view; at a higher temperature a smaller 

 tension is necessary to make the transition take place. It is of course 

 not possible to make any very exact numerical comparisons in view 

 of the flatness of the minimum of the curves, but the data are at least 

 not inconsistent with a depression of the transition point of a single 

 homogeneous (Crystal by an amount proportional to the square of the 

 tension. My explanation also demands that the resistance of the high 

 temperature phase when it is subcooled be less than that of the low 

 temperature phase. In view of Werner's failure to find a large dis- 

 continuity in the transition point, this probably means that the tem- 

 perature coefficient of the high temperature phase is greater than that 

 of the low temperature phase, and this is in accord with the fact that 

 the initial rate of decrease of resistance with tension is less at high 

 temperature than at low. 



The abnormal character of the hysteresis loop on the falling branch 

 is to be explained as follows. The primary effect of a tension is to 

 force a transition from one phase to another. This transition would be 

 expected to show hysteresis on reversing the direction of the change of 

 tension. On the other hand, the sum of the pure tension effect in all 

 the individual grains shows no hysteresis, because this is determined by 

 the average tension, which is equal to the applied tension itself. 

 Hence on decreasing the direction of the change of tension, the abnor- 

 mal effects are decreased in magnitude, whereas the normal effects are 

 unaltered. An application of this analysis to the upper end of the 

 loop beyond the minimum of resistance (maximum on the diagrams) 

 shows that on release of tension the resistance may drop to less than 

 under any increasing tension. 



