employed for determining the Ohm. 261 



metre thickness to the tenth of a millimetre, as in the older 

 experiments of the British Association, was useless. 



It is not difficult to determine the internal diameter, in 

 various directions, of the coils of wire (that is, the diameter of 

 the frame on which they are wound), either by means of a 

 kathetometer, or with an inextensible steel tape of constant 

 temperature; but, on the other hand, the measurement of the 

 external diameter or circumference is much more difficult, in 

 consequence of the inequality of the covering of the wire and 

 the unevenness of the surface. 



If the error in the determination of the mean diameter 

 only amounted to 0'5 millim., which, in view of the circum- 

 stances mentioned, is certainly not an extreme estimate, then 

 in the case of the coil of 314 millim. diameter employed by the 

 Committee of the British Association the estimate of the space 

 enclosed by the coils would be wrong by 3|^=0"32 per cent. 

 In order to reduce this error as much as possible, it is neces- 

 sary, as both W. Weber * and Lord Rayleighf recommended, 

 to take the diameter of the coil as great as possible consistently 

 with accuracy of rotation. Also the wire should be wound with 

 a tension as uniform as possible ; and the exterior diameter 

 should be controlled after each layer has been wound. 



A much more important source of error lies in the uncer- 

 tainty of determining the mutual position of the separate coils, 

 depending upon the conditions explained above, upon which 

 the intensity of the extra current induced in the coil when put 

 into rotation depends. Since the inductive action of the coils 

 upon each other takes place at very small distances, a very 

 small error in measuring the distance apart is of great im- 

 portance. 



A further source of inaccuracy is introduced by making the 

 coil of two parts parallel to each other, but with a space between 

 in order to admit the thread by which the magnet is sus- 

 pended; so that here also the parallelism and distance apart of 

 the two portions must be very exactly determined. That the 

 data may be greatly altered by the extra currents is seen from 

 the fact that, in an experiment of a Committee of the British 

 Association, the position of maximum induction of the coil in 

 rapid rotation was displaced by not less than 20°; and the 

 correction for this amounted to some 8 per cent. 



If we seek to determine the self-induction by opposing the coil 

 to another of known coefficient as a YVheatstone's bridge % , 



* Berichte der Koniglich Scicksischen Gesellschaft der Wissenschaften. 

 1880, p. 77. 



t Proceedings of the Royal Society, 1881, vol. xxxii. p. 122. 



% Compare Maxwell's ''Treatise,' vol. ii. p. 357; Brillouin, Comptes 

 Rendus, vol. xciii. p. 1010 (1881) ; Beibldtter, vol. vi. p. 39. 



