EFFECT OF TENSION ON CERTAIN ABNORMAL METALS. 61 



hence means that its mechanism is the same kind as that of Hthium, 

 and verifies the evidence from crystalhne structure. The positive 

 coefficient of antimony indicates the same thing. In this regard it 

 may be said that recent work has cast considerable doubt on the 

 reahty of the supposed expansion when antimony freezes, ^^ so that my 

 former expectation would now lose its chief ground for support. Also 

 against my former argument I may mention that there is a polymor- 

 phic transition of antimony at 135°, which I had previously failed to 

 take into account. Even if the relations are abnormal for the modifica- 

 tion which is stable up to the melting point, there seems no reason why 

 we should expect the same abnormalities in the other modification, 

 which is stable at room temperatures. 



Strontium seems to require special consideration. Its tension 

 coefficient is not abnormally' high numerically; it is nearly the same 

 as that of calcium, and much less than that of lithium or bismuth. 

 On the other hand its pressure coefficient is unique in being at least 

 three fold greater than that of any other abnormal metal. It seems 

 reasonable to suggest that the mechanism of conduction in strontium 

 may be a combination of both types. These would conspire to give 

 an abnormally high pressure coefficient, and oppose each other, giving 

 by difference a relatively small . tension coefficient, of a sign which 

 could not be predicted without further evidence. 



These views of the conduction mechanism receive support from a 

 numerical discussion. We confine ourselves to changes of resistance 

 at constant temperature, that is to the changes under pressure and the 

 longitudinal changes under tension. We assume that the change of 

 resistance may be written down in terms of the changes of dimensions. 

 The transverse and longitudinal changes of dimensions will affect the 

 resistance in different ways, which are different for the two different 

 types of mechanism. We write the equation 



1 A5/ A8r 



where Av denotes the change of resistance per unit strain transverse to 

 the direction of the current, ki denotes the change of resistance per 

 unit strain longitudinally, and A8r and A5i are the change in the trans- 

 verse and longitudinal distance of separation of the atoms. Now if 

 the strains are produced by tension, we have 



Adi AT A8r (X 



8i~ E' 8r~ ~ E^^ 



19 M. Toepler, Wied. Ann. 53, 343-378, 1894. 



