-ii4 



of Dee. 1911 in table I. We find from the threshold value of the 

 current at 4°. 19 K., that the resistance of the thread at this tem- 

 perature may be put at <[ 3.10-^ £^. In the experiment at 3°. 65 K. 

 we find that when the strength of current rises to 1 amp. the resi- 

 stance, 11.5.10— ^i2, was already distinctly greater than when the 

 Avhole thread was at 4°.iy K., while the ends must still be at 3°. 65 K. 

 The portion that comes above the vanishing point by this heating, 

 as it assumes ordinary resistance, need only be very small to pro- 

 duce the potential differences observed ; in the case in point only 

 0.1 mm. If we assume that the giving off of heat to the bath may 

 be calculated by the same data as were found for it above the 

 vanishing i)oint in § 7, then we find that, if the whole surface of 

 the thread were at the mean of the temperature of the bath and 

 of the vanishing point, the loss of heat })er second should be about 

 20000 microjoules, while in reality only 14.0 microjoules, or about 

 1400 ti7nes less, are given off. 



We conclude from this that the rise of temperature in the thread, 

 which is in a bath of a temperature below the vanishing point is 

 only local. If there were anywhei'e else a rise of temperature 

 (although of a smaller amount) the thread must have ceased to give 

 off heat to the glass to a perceptible degree, except at certain points. 

 The heat could therefore only flow to the extremities or the remain- 

 ing points of conduction. This might be the consequence, for instance, 

 of the mercury having come away from the glass everywhere except 

 at the places indicated. But this is contradicted by the fact that in 

 freezing the mercury adheres to the glass, and that immediately 

 above the vanishing point the contact has not yet ceased. The sup- 

 position that everywhere where the temperature remains above the 

 vanishing point (and perhaps close to it) the mercury thread gives 

 off heat, and that it does not where the temperature is lower, is confirmed 

 by the way in which the resistance disappears below the vanishing 

 point (see Table II and fig. 7). If we determine, from the propor- 

 tion of the resistance remaining to that just by the vanishing point, 

 the length of the portion of the thread which is at the temperature 

 of the vanishing point, then the JouLE-heat that it must give off at 

 the existing strength of current corresponds more or less to that 

 which is to be expected at the assumed difference of temperature 

 of bath and vanishing point if the heat is given off to the glass over 

 the whole length of that portion; more or less, for there remain 

 unexplained and apparently systematic differences, with which 

 perhaps the difference of the curves for different strengths of current 

 in fig. 7 is connected. 



