A NEW CURRKNT WEIGHER, KTC. 523 



a very desirable material for weights, o \vini; to its density being so small, but in 

 our experiments the effective mass <>f the weights never varied by so much as 

 8 in 1,000,000 from the mean, and if no correction had been made for variable air 

 displacement, the error in the measurement of the current would never have exceeded 



4 in 1,000,000. Of course, the corrections were applied. The probable error of the 

 effective mass is of the order 1 in 1,000,000. 



Preliminary Difficulties. 



(A) Defects in flexible Concentric Cable. The cable originally used consisted of 

 an inner conductor of 30 copper wires of diameter 0'0048 inch and an outer tubular 

 conductor of 74 strands of the same diameter. After connecting the balance coils 

 to the multiple commutator and plug board the cable was found to be faulty, and 

 subsequent examination showed that many of the internal strands were broken. 

 The cable was therefore replaced by a concentric one having an inner conductor of 

 three copper wires of diameter 0'022 inch, and an outer tubular conductor of sixteen, 

 of diameter - 0148 inch. This proved to be entirely satisfactory. 



(B) Unsteady Current and Convection Currents of Air Produced by the Heating 

 of the Flexible Leada. Originally the current was led into and out of each pair of 

 suspended coils by two silver-gilt strips, each 13 centima long, 0'37 inillim. broad, 

 0'035 millim. deep, and of 0'15 ohm resistance. In each pair of silver strips there 

 was 1 calorie of heat produced every 14 seconds when a current of 1 ampere passed 

 through them, and the maximum increase in temperature of the strips was about 20 C. 

 The temperature coefficient of electric resistance of silver is 0*36 per cent., hence the 

 increase in resistance of the four strips was 0'04 ohm, and a fluctuation in temperature 

 of 1 C. corresponded to a change in resistance of the circuit (110 ohms) of 0'002 ohm. 

 Such a change in temperature frequently resulted, as was proved by including the 

 silver strips in a circuit containing 110 ohms of manganin ; with a current of 

 1 ampere the fluctuations in current were of tmT order 5 in 1,000,000. When the 

 strips were removed from the circuit, the corresponding changes were 1 in 1,000,000. 



The energy of motion of the air particles in the immediate neighbourhood of each 

 pair of strips was increased at the rate of about 3x 10* ergs per second. The effect 

 of the convection currents of air thus produced was tested by passing a current of 

 1 ampere through one pair of the strips inside the balance case when the balance 

 coils were not included in the circuit. After the circuit had been completed for 



5 minutes the resting point of the balance changed by an amount equivalent to an 

 added load of 9 milligrammes on that side of the balance with the heated strips ; 

 after 10 minutes the change corresponded to 24 milligrammes ; 15 minutes afterwards 

 to 39 milligrammes, after which the resting point of the balance was approximately 

 constant. The circuit was broken for 1 5 seconds and the change noted ; it corre- 

 sponded to 0'4 milligramme; equilibrium was restored after 5 minutes. The length 



a x '.' 



