266 Prof. S. P. Thompson on the 



used nearly twenty years ago by the Committee on Electrical 

 Standards for temporary adjustments, On page 101 of the 

 Reports, as reprinted under the editorship of Prof. Fleeming 

 Jenkin, the following passage occurs : — 



" The idea of using large coils combined with small ones in 

 multiple arc to obtain extremely small minute differences of 

 resistance was suggested to the writer by Professor [now Sir] 

 W. Thomson, and will be found useful in very many ways." 

 But only one way is mentioned in the Reports, namely that 

 of plugging in the high resistance as a temporary shunt during 

 a single experiment. 



2. I will first describe the method pursued in the Physical 

 Laboratory of University College, Bristol, for the adjustment 

 of coils not exceeding 10 ohms in resistance. First, the length, 

 calculated to within 0*5 of a centimetre, is cut off from a 

 specimen of German-silver wire of which the resistance per 

 centimetre is roughly known ; 2 per cent, being added before 

 the wire is cut off. It is soldered to the stout copper terminals 

 prepared to receive it, doubled on itself, twisted, wound up, 

 and when the soldering has cooled, the resistance of the wire 

 is measured with the utmost care by comparison, made on 

 Professor Carey Foster's method, with a standard coil, con- 

 structed at Cambridge in Professor Stuart's workshops, and 

 w T hich has been verified at the Cavendish Laboratory and cer- 

 tified in terms of Lord Rayleigh^s determination of the ohm. 

 The " rough resistance " of the measured coil being thus very 

 accurately known, a calculation is made to ascertain what 

 resistance, connected as a shunt, will reduce the resistance 

 between terminals to one exact Rayleigh ohm. The resistance 

 required for the shunt is in practice from 12 to 50, 60, or 

 even 80 ohms. The nearer the first approximation, the higher 

 the shunt-resistance required, as the formula given below 

 shows at a glance. The calculation then having been made, 

 the requisite length to approximate to this is cut off from a 

 much finer wire, which is then soldered across the terminals. 

 At first we used to measure the resistance of these shunts 

 before putting them on. In practice we find this quite 

 unnecessary, as they can always, if there is any error of im- 

 portance, be readjusted after soldering them to the copper 

 terminals. In nine cases out of ten, however, no further 

 adjustment is needed. 



3. The following is the theory of, and formula for, the 

 value of the shunt required. 



Let ?R be the exact resistance of the first rough approxi- 

 mately-cut coil, being larger by from 1 to 4 per cent, than 



