588 



SCIENCE PROGRESS 



ture the number of electrons per unit volume increases, the 

 resistance will increase from this cause. Hence it might reach 

 a minimum at a low temperature and thence increase to large 

 values. The only way to decide which of these possibilities 

 is the more correct was by direct experiment at extremely low 

 temperatures. 



With very, but not absolutely, pure platinum wire it was 

 found that the resistance decreased to a very small constant 

 minimum value at about 5 K. The same kind of phenomena 

 was observed with gold, and in each case the purer wires gave 

 smaller values for the constant, which was clearly due to small 

 residual impurities. 



This result was very interesting and important but not deci- 

 sive, and hence it was necessary to employ the purest obtain- 

 able metal, i.e. mercury. This was most carefully distilled and 

 drawn into fine capillaries, which were connected together so 

 as to give sufficient resistance to be measurable. The result 

 obtained was extremely definite and remarkable. A sudden fall 

 in resistance occurs at 4-25° K., and at 4-2 K. the resistance is less 

 than io~ 6 times its value at o - o°C. Further, at the lowest attain- 

 able temperatures this extremely low value was maintained. 

 Subsequently it was found that lead exhibited the same effect at 

 about 6° K., a temperature which is not easily maintained con- 

 stant as it is above the boiling-point of liquid helium. With 

 tin the corresponding value was 3-8° K., which could be deter- 

 mined exactly. These are the values obtained with small 

 currents as the highest temperatures of the point of change from 

 ordinary to super-conduction. 



TABLE I 



Relative resistances of metals, Rt\Ro. 



