116 



the development of lieat above that which produces locally the 

 vanishing point temperature ; it would be further determined by the 

 circumstances under which the excess of the heat developed would 

 be given off. When we compare the potential difference observed in 

 the different cases, there are one or two things that seem to confirm 

 this supposition '). 



Taking all this together we are brought back to the idea that 

 the potential phenomena must be ascribed to "bad places" , although 

 in a different sense to that in '^ 9. But Ihe regularity of the pheno- 

 mena remains a weighty objection to this hypothesis'^). For although, 

 with the explanation of the local development of heat by a difference 

 in the states of crystallization, the difficulty disappears which in the 

 explanation by foreign resistances arose out of the circumstance that 

 the whole section must be blocked up, still the appearance of a 

 dividing surface between two states of crystallization is govei-ned 

 by chance. In any case, to come to an explanation on this principle, 

 we should have to assume, that there are various PELTiER-places of 

 the kind meant in each mercury thread of any length and that they 

 are not too unevenly distributed. 



But in this manner we should add a new indefinite hypothesis 

 to the one which has to be tested and it is only by a complete 

 quantitative working out of a perfectly definite theory that the 

 question with which we are dealing can be answered : for the 

 answer involves some far-reaching inferences. If we might assume 

 that the potential phenomena in mercury-threads at a cui'rent density 

 exceeding' the threshold value are entirely due to disturbances then, 

 on account of the systematic connection of the potential phenomena, 

 there would be every reason to assume that we get a truer idea of 

 the actual degree of conductivity of the superconductive mercury, 

 the lower the temperature at which we determine the threshold 

 value of current densiry of a thread^). And as at the lowest tempe- 

 ratures the disturbances still have an influence, although a smaller 

 one, the actual conductivity would therefore have to be placed 

 higher, perhaps a good deal higher, than the value found in § 7, 

 which was already 0,5.10^" times that at the ordinary temperature, 



1) Too indeiinite to be published. 



2) The existence of a real microresidual resistance is also made probable by that 

 the ratios between the resistances for the mercury in the capillary tube and the 

 frozen mercury thread at 4°.25 K. seems to run parallel to the threshold values, 

 so that the difference of the threshold values might be ascribed to differences of 

 the local deviations of the cross sections from the mean. 



5) In this train of thought there is no reason for not supposing that the con- 

 ductivity assumes its large value immediately below the vanishing point. 



