( 302 ) 



succeeded in stopping that process. Yet at any rate this experience 

 obliged us to perform measurements at low tenipei'atnres, like tliose 

 treated here for the gold wire, in a very shorl lime in I he case of 

 copper wire. Willi gold and silver this jdienomenon did not appear. 



§ 4. ComiKU'ison of iIlc re.slstance.s. In order to in\estigate the 

 variation of tlie resistance of the wire, wound as described in § 2, 

 the ratio of this resistance to that of [)latiiHnn had to be deter- 

 mined in different baths. To this end previous investigators have 

 always measured the resistance of the two wires alternately and 

 hence derived the mean i-atio. The lirst experiments made by me 

 were also arranged in that way. 



In order to attain a higher degree of accuracy I ha\e followed 

 the ad\ice of Prof. Kamerlingh Onnks and arranged the measurements 

 so that at a definite moment the ratio itself can be read. 



If we use the Whkatstone's bridge, it seems that this may be 

 attained by arranging the wire lo be com])ared in [)arallel to the 

 box of Hartmann and P>rau.\ (/j*,) instead of arranging it in parallel 

 to the platinum resistance. 



An insuperable obstacle for this simultaneous determination is, that 

 at any rate a connection is required between the two wires, which 

 connection may be made by two of tiie leads from the l)ath, in which 

 we measure, to one of the angles of the l»ridge. Elimination of all the 

 resistances of the coimections excej)t that of the stops used in the 

 measurement, as it was oblaiiu'd iu the pi-evious paper, is impossible. 

 Errors may then creep in of which the amount may only be estimated. 

 Resistances of the coimections may occui- to a considerable amount 

 and be bi-ought about by minor causes. If we cannot consfantly test 

 their amount the results remain uncertain. 



Moreover in the Wheatstonk's bridge the unavoidable resistances 

 of the connections are so large that we cannot reach the accuracv 

 l>roposed in the outset with wires of so little specific resistance as 

 gold. For in that case, even though we succeeded in winding a wire 

 of more than 20 meters in the necessarily small compass of the bath, 

 a resistance of no more tiian 35 Ohms at 0^ i\ was obtained, which 

 in liquid oxygen or nitrogen fell to below about 6 Ohms. 



All these reasons led us to choose the method of the differential 

 galvanometer for the comparison of the metals inter se. It enaldes 

 us to determine at once the ratio of two resistances, while resi- 

 stances of the connections have no -appreciable influence. It is true 

 that a measurement cannot be made in such a short time as 

 with the WhisATstonk's bridge but this inconvenience is sulFiciently 

 balanced by the advantages mentioned. 



