Unipolar Induction. 1 85 



short rings, and they were also flattened on opposite sides to 

 a depth of some 3 mm. Now a rather long calculation by 

 the method of the first article cited shows that an amber 

 shell extending the whole length of the space and from the 

 shaft to within 3 mm. of the inner cylinder would increase 

 the electric flux by something like 40 per cent. In the 

 actual case the amber extended only an eighth of the length 

 of the insulated cylinder, and was flattened over less than 

 half of the circumference, so that the effect could hardly 

 exceed 2 per cent, of the flux ; and this would result in an 

 extra charge of only O02 x 0'140V = , 03V . This correction 

 just about cancels the one due to the connecting wire. 



§ 5. Determination of tlie Theoretical Effect. 



A rough estimate of the deflexions to be expected ean 

 now be made as follows. A current of 25 amp. would pro- 

 duce an induction of 1600 gausses inside the solenoid, and a 

 flux through the central section of the inner surface of the 

 outer cylinder of about 55,400. At a speed of 33 rev. 

 per sec. this gives a difference of V = *0183 volt between 

 outer cylinder and axis ; the charge on the inner cylinder will 

 therefore be 6'22 x -0183 x 1/300 = 0-114 x 1/300 elst. unit. 

 The total capacity was found to be 138 elst. units ; hence 

 with a sensitiveness of 18,000 mm./volt the establishment of 

 the magnetic field will cause a deflexion of 0*117 x 18,000 

 -f- 138 = 14*9 mm. If in fig. 1 the direction of the magnetic 

 field is from left to right and the top of the condenser is 

 moving out of the paper, then the outer cylinder will be at a 

 higher potential than the axis, and the inner cylinder will be 

 negatively charged ; establishment of the field will cause a 

 deflexion of the electrometer in the direction of positive 

 potential. 



For an accurate test of theory the ratio between electro- 

 meter and galvanometer deflexions was employed. For the 

 comparison readings a potential of V = - 02 volt (about) was 

 applied to the electrometer case, which was used as a standard 

 inductor, and simultaneously exactly one-tenth of this was 

 applied through the megohm to the galvanometer; by 

 applying, reversing, and removing the potential, readings 

 were taken in imitation of the principal ones. The electro- 

 meter readings agreed within 2 per cent., and averaged 

 205 mm., while the galvanometer deflexion was 62*2 mm. 

 The coefficient of influence of the electrometer case had been 

 compared previously by a null method with a good guard- 

 ring condenser, whose capacity was computed to be 40'1 elst. 

 units, and was found to be 7 "42. On the other hand, the 



