1896.] Scientific Instruments used by the late Dr. Joule. 347 



into the position it occupied before the passage of the current — that 

 is, into a position midway between the two fixed coils — and c be the 

 current as measured by the " Kelvin Balance," then, since the attract- 

 ing force between two coils depends on the square of the current, 



c = Kv^p or K — c/y/iv, 



where K is the constant of the instrument. 

 The following observations were made : — 



Current c. 



"Weight to. 



\/ to. 



Constant. 



4-469 



20-92 grams 



4-574 



0'9770 ampere 



3-938 



16-25 „ 



4-030 



0-9769 



3-455 



12-52 „ 



3-338 



0-9767 



3-108 



10-13 „ 



3-183 



0-9764 



2-604 



711 „ 



2-6665 



0-9765 



It will be noticed that the value of the constant slightly, but fairly 

 regularly, decreases as the current decreases. 



I noticed this decrease in all the experiments I made with both this 

 balance and the newer one, but was unable to find any explanation. 



I do not think it could be due to leakage through the wood of 

 Joule's balances, for I tested them thoroughly for leakage and could 

 find none, nor could the electrostatic force between the coils be the 

 cause, for this, on calculation, appears to be quite inappreciable. 



For reading the balance a needle is provided fixed horizontally into 

 the end of the balance beam, the end of this needle oscillates in front 

 of a graduated scale attached to one of the sides of the box, the 

 observer's eye being placed to a small hole cut in the opposite side of 

 the box and thus any errors of parallax avoided. 



By this means the weight on the movable coil could be easily 

 adjusted to less than 5 milligrams, and as the weight varies with the 

 square of the current, the half only of any error made in the deter- 

 mination of the weight appears in the estimation of the current. 



On reference to the above table it appears that with this balance 

 10 grams means about 3 amperes, so that 3 amperes may be measured 

 correctly to 1 part in 4,000; it is also evident that as the weight 

 varies with the square of the current the balance is much more 

 accurate for large than for small currents. 



A source of some trouble is a constant changing of the zero, that 

 is to say, the counterpoise at the end of the beam has to be frequently 

 adjusted in order to bring the movable coil into its zero position 

 halfway between the two fixed coils. I may mention that there is 

 no accurate method of bringing the suspended coil exactly into this 

 mean position, but by experiment I found that the exact point fixed 

 upon as zero is immaterial, the value obtained for the constant being 

 the same, whatever point, within certain limits, be chosen as zero. 



vol. lix. 2 B 



