1427 
if the fall below the vanishing point is not too small, and if we 
leave out of account a term for Joure heat which only appears 
distinctly at a higher current strength. This naturally suggests that 
we are dealing with a Prrrimre-effect raising the temperature to the 
vanishing point of resistance (e.g. connected with different forms of 
crystallisation or tensions); (the simultaneous cooling of the opposite 
contact has no effect on the resistance which is already practically 
zero and remains zero when further cooled). As regards the threshold 
value of the density at a given temperature for different threads this 
appears (comp. § 6 and Table IV) to be rendered uncertain by 
accidental circumstances. But it deserves notice that it was also found | 
very high in very narrow capillaries. 
§ 9. Experiments on impurities as a possible source of disturbances. 
Although the greatest care was always bestowed upon the purifica- 
tion of the mercury, the explanation of the appearance of a residual 
resistance that offered itself the first for closer investigation 
was the influence of impurities. These may give an “additive mixture 
resistance” to the metal which changes little with the temperature 
and is proportional to the amount of impurity. To such an additive 
resistance 1 ascribed the fact (Comm. N°. 119 and Leiden Suppl. N°. 29) 
that the resistance of very pure platinum and very pure gold did not 
disappear at helium temperatures as I expected with absolutely pure 
metals. Now the experiments had realized the expectation, that mer- 
_eury could be so far freed from impurities, as to make the resistance 
practically nothing. But if one may judge by the additive resistance 
which even very pure gold exhibits, then with the residual resistance 
of mercury which is only perceptible at the threshold value or 
current density for the lowest temperatures, it would be a question 
of an impurity of the order of a millionth of the trace that could 
possibly be present in the most carefully purified gold. And it was 
a priori doubtful if the mercury could be procured in so much 
greater a state of purity than gold. *) 
The experiment was therefore repeated with solid mercury in 
which I believed a very small quantity of an other metal to be 
present. After being distilled in a vacuum by means of liquid air, 
the mercury was in one case brought into contact with gold and 
the other time with cadmium, after which it was mixed with a 
larger quantity of pure mercury. To my surprise with the mercury 
1) For difficulties inherent in the supposition of a resistance equally distributed 
throughout the thread which apply also to our present case of additive mixture 
resistance see § 11, 
