In supposing, however, that tlic development of heat which 

 brings a part of the thread to tlie temperature of the vanishing 

 point is of a local nature, we give up the supposition that the 

 microrcsidual resistance is evenly distributed over the thread. Assu- 

 ming the whole of the path of the current to be of pure mercury, 

 there could possibly oidy be an apparent microresidual resistance, 

 in consequence, for instance, of the mercury not being homogeneous, 

 or not free from mechanical tension. These disturbances would then 

 be the cause of threads showing a resistance throughout, while the 

 pure homogeneous tension-free mercury would have an impercep- 

 tible microresidual resistance. 



If we remember that with lead the increase of resistance by 

 pressure ') becomes less at low temperatures, and has almost disap- 

 peared at hydrogen temperatures, then it is not probable that tensions, 

 although they could cause PELTiER-etTects, and although their regularity 

 corresponds to that of the phenomena, should really play a part in 

 the disturbances. 



It would be more natural to suppose a lack of homogeneity in 

 the thread, which might be the consequence of difference of the 

 state of crystallization. When we turn down a block of very pure 

 KAHLBAUM-lead on the lathe, we can sometimes see a moiré effect 

 on the surface, which indicates different alternating states of 

 crystallization, each of which extends over more than a centimetre. 

 In this way a thread of solid mercury might consist of a series of 

 differently crystaUized portions, the dividing surfaces of which would 

 be at the same time usually cross sections of the thread. 



At a dividing surface of this kind, a local heating such as we 

 have treated above, might take place, at the expense of current 

 energy. For instance a transitional resistance might give an apparent 

 microresidual resistance to such a dividing surface. But the relation 

 between the threshold value of current density and the temperature 

 of the bath, points (see § 8) rather to a PELTiER-effect at this 

 transitional place. We should then have to imagine that when the 

 current density reaches the threshold value, the temperature at the 

 dividing surface between two states of crystallisation, even if not 

 high enough to occasion a thermoelectric force equal to the potential 

 difference observed, yet reaches the vanishing point, and that, there- 

 fore, by further increase of the current density ordinary resistance 

 must appear at this dividing surface. The length of the thread which 

 takes an ordinary resistance would then increase with the excess of 



1) H. Kamerungh Onnes and Bengt Beckman. Gomm. No. 1326, Nov. 1912, 



8* 



