1894.] Specific Resistance of Pure Electrolytic Copper. 67 



lengths were hanging dbvvn. The double bent wire was then wound 

 up in the double screw thread, and secured at the top by means of 

 string ; the influence of self-induction was thus avoided. Each wire 

 had four terminals, the main current terminals, and the shunt ter- 

 minals, which were soldered to pieces of stout copper, the distance 

 between which determined the length of the wire under examination. 

 These reels saved the wires from being hardened or otherwise in- 

 jured. The length of the wire was taken both before and after the 

 electrical measurement. In some earlier experiments when mica 

 strips were used for coiling the wire upon, great differences were 

 observed. When the ebonite reels were employed hardly any differ- 

 ence could be observed. 



The electrical measurements were made by the fall of potential 

 method, refined as much as is possible. A D'Arsonval galvanometer 

 was employed as an indicator of potential difference. The " dead 

 beat " property of this instrument is an advantage which in this 

 class of measurement cannot be over-rated. Finding in the first 

 experiments with the D'Arsonval galvanometer, that its sensitive- 

 ness was not sufficient for our purpose, the upper suspension was 

 lengthened very considerably. The coil was made to hang on the 

 upper wire, the wire below the coil being left slack. By this means 

 the required degree of sensitiveness was obtained. The galvanometer 

 readings were made by means of a telescope and scale, a plane mirror 

 being attached to the galvanometer coil. To gain the advantage of 

 using very small deflections, the scale and galvanometer were widely 

 separated, the distance between them was 8 m. The largest deflec- 

 tions used were 600 divisions of the scale (each division = 1/30 in.), 

 corresponding to (at that distance from the mirror) an angle of 

 1 49', which falls within the limits for : 



(1.) tan 2 angle = 2 tan angle + O'OOl x 2 tan angle. 



The readings of the deflections were found to be proportional to 

 the P.D. by direct and carefully made measurements, they were 

 always read in two directions, and the sum taken as the deflection ; 

 the accuracy was thus doubled, and possible error by displacement of 

 the zero point avoided. 



The limit of sensitiveness of the galvanometer was : 



1 scale division deflection = g-oo'oVoo'o am P ere (approximately), or 

 = 0-00000003101 ampere (accurately). 



The resistance of the suspended coil was 10 ohms. The time re- 

 quired for the coil to come to rest was about two seconds. The 

 accompanying diagram shows the arrangement for making the 

 measurements. They were made by direct comparison between a 

 standard ohm and the various coils to be measured, and in such a 

 manner that error due to the resistance of the leads was eliminated. 



r 2 





