52 BRIDGMAN. 



had dropped to one quarter of its initial value, which was at the rate 

 of an increase of resistance of |% per hour. After the removal of a 

 load which has exceeded the elastic limit there are also time effects, 

 not nearly so large in magnitude as the effects on applying load, but 

 in the same direction; the resistance continues to increase for some 

 time after the removal of load. In this particular the effect is like 

 an ordinary elastic after effect. 



Nickel. The sign of the pressure coefficient of this metal is normal; 

 I have already mentioned in the introduction my reasons for repeating 

 the measurements of Tomlinson. The metal was provided by the 

 Research Laboratory of the Leeds and Northrup Co. in the form of 

 bars approximately 6 cm. long, L5 cm. wide, and 3 mm. thick. I cut 

 from this bar a piece of suitable dimensions, and drew it down in steel 

 dies to 0.0101 inches. It was annealed after drawing by heating to 

 redness. 



The temperature coefficient of resistance between 0° and 100° was 

 found to be 0.00634. This is higher than the highest previous value 

 with which I am acquainted, 0.00618 by Fleming, and is evidence of 

 the unusual purity. 



Tomlinson had found that nickel is abnormal in two particulars. In 

 the first place the resistance decreases under tension, and in the second 

 place the decrease is not linear with tension, but passes through a 

 minimum, so that at a tension of the order of two thirds of the elastic 

 limit the resistance begins to increase with increasing tension. 



I verified both these particulars of behavior, and made a few addi- 

 tional observations. If instead of increasing the tension steadily to 

 the elastic limit, measuring the resistance as a function of tension on 

 the way, as Tomlinson did, the wire is subjected to a seasoning for 

 some fixed load by a number of applications and removals of the same 

 load, and then a cycle of resistance measurements is made, it will be 

 found tliat the decreasing measurements do not follow the increasing 

 measurements, but an open hysteresis loop is described. This loop 

 differs in one important particular from ordinary hysteresis loops. 

 By hysteresis we usually mean an efPect that for some reason lags 

 behind. In this case, this would mean that with decreasing tension 

 the resistances correspond to some greater tension on the increasing 

 run. The curious fact here is that at the initial stages of the decreas- 

 ing limb of the loop the resistance may be lower than that correspond- 

 ing to any value of the tension on the increasing run. The mechanism 

 of the loop must be something quite different from the ordinary hyster- 

 esis effects. 



