198 Mr. H. Pender on the Magnetic 



M and ^M.rdr can be calculated from the dimensions of the 

 apparatus by means of elliptic integrals, provided the coil I is 

 perfectly uniform. However, there were considerable varia- 

 tions in the dimensions of the coil from point to point ; so 

 that this method was not deemed sufficiently accurate. 

 Hence the following scheme, which amounts practically to 

 an experimental determination of M and a graphical calcula- 

 tion of \Mrdvj was adopted. Twelve concentric coils of one 

 turn each were mounted on a plane board. This board could 

 be clamped in turn up against each surface of each disk. 

 The radii of the coils varied from 4*3 to 16*0 cms. By a 

 simple device, any one of these coils or the test-coil T 

 could be connected at will to two leads from the reverser, 

 The lead wires from the Yoss machine to the switch S were 

 disconnected and the terminals of a storage-battery, 110 

 volts, in series with 150 ohms' resistance substituted in their 

 stead. In this manner a current of f of an ampere could 

 be sent at will through the reverser and any one of the coils 

 on the disk, and the current induced in the coil I on 

 running the reverser measured. Instead of employing for 

 this purpose the sensitive galvanometer above described, an 

 ordinary D'Arsonval about 1000 times less sensitive was used. 

 (Hence the large value of the inducing current.) The 

 D'Arsonval was employed because it could be used during the 

 day ; also its readings were much more reliable than those 

 of the astatic instrument. The ''calibration of the disk 

 apparatus " then consisted in sending a given current through 

 the test-coil T, with the reverser running, and noting the 

 deflexion of the D'Arsonval connected in series with the coil 

 I. The same current was then sent through coil No. 1, on 

 the disk, the deflexion noted ; then again through coil T, the 

 deflexion noted, and so on for each of the twelve coils when 

 clamped up in turn against each face of each disk. The 

 deflexion due to the coil T was determined before and after 

 every determination for a coil on the disk, in order to elimi- 

 nate any error due to the variation in speed of the commutator. 

 The ratio p of the deflexion due to the given current in each 

 coil to the deflexion due to the same current in coil T was 

 determined from these readings. To find the value of the 

 factor 2MK it is theu necessary to know only the deflexion 

 of the astatic galvanometer produced by a unit current in coil 

 T with the reverser running. Call this A. Then 2MK = Ap. 

 (This of course depends on the sensibility of the galvanometer 

 and the number of reversals per second ; see below.) It is 

 hardly necessary to state that the relative positions of the 



