EFFECT OF TENSION ON CERTAIN ABNORMAL METALS. 63 



amplitude is abnormal in sign, and more than counterbalances the 

 effect of changing distance between atoms, ki and k.^ should both be 

 negative, and k^. less numerically than ki. On the other hand, for 

 those metals whose conduction mechanism is by the passage of elec- 

 trons in channels between the atoms, we expect k^ to be negative, since 

 increasing the transverse separation of the atoms decreases the 

 resistance, and ki to be relatively small. For strontium, which has 

 a combination of both types of mechanism, we expect both ki and kr 

 to be negative, k^. being numerically larger than A7. 



An inspection of the table shows that in every case these anticipa- 

 tions are strikingly verified, and the probable essential correctness 

 of the theory receives strong support. 



The values of the coefficients found for manganin and therlo indi- 

 cate that for these the mechanism is for the most part like that of 

 lithium, calcium, and antimony, but that there is in addition a small 

 contribution by a mechanism of the bismuth type. Of course the 

 phenomena for alloys are most complicated and varied in their types 

 of behavior, but it is not difficult to picture to oneself that under some 

 conditions when two different kinds of atoms crystallize side by side 

 into the same space lattice that there should be channels left between 

 the atoms for the passage of conduction electrons, or that the law of 

 force between atoms of different kinds should show the same sort of 

 abnormality that the atoms of bismuth show. 



So far as I know there has been no previous attempt to make con- 

 nection between any theory of conduction and the tension effects. 

 In the light of the success of the above for abnormal metals it would 

 now be of much interest to accurately determine the tension coeffi- 

 cients of the normal metals. If in addition the coefficient of trans- 

 verse resistance could be determined, a most valuable check would be 

 obtained, for we would then have three independent experimental 

 coefficients, which must be expressible in terms of the two quantities 

 hi and kr. 



Discussion of the Effect in Nickel. 



The peculiar nature of the phenomena for nickel makes it evident 

 that there must be some unusual mechanism involved. It seems to 

 me that there is probably an intimate connection with the poly- 

 morphic transition at 360°. Under a tension the transition tempera- 

 ture will be displaced by an amount proportional to the square of the 

 tension, and if a certain function of the compressibility and Young's 

 modulus has the right sign, the transition temperature will be 



