Magnetic Effect of Electric Convection. 149 



good enough, and consequently he failed to obtain the calcu- 

 lated magnetic field. 



Rowland & Hutchinson, in their experiments*, used an 

 arrangement which is equivalent to Cremieu's without the 

 drum-ends. It is shown diagram matically in fig. 5. Two 



Fio-. 5. 



charged ebonite disks, carrying metal sectors, were rotated 

 on the same axis GH. Two glass plates, C and D, were 

 fixed between them, and carried the opposite charges to those 

 on the disks. These plates also carried metal sectors. The 

 magnetic field was measured by means of an astatic needle 

 system E, placed between the plates so that the lower needle 

 was on the axis of rotation of the disks. This needle system 

 was enclosed in a metal tube. The observed field agreed well 

 with that theoretically predicted. 



It is evident that this arrangement is practically the same 

 as Cremieu's without the drum-ends, so that we may conclude 

 that the insulation between the sectors in Rowland & 

 Hutchinson's apparatus was good enough to prevent the lines 

 of force slipping across the gaps, which was probably not the 

 case in Cremieu's apparatus. 



It may be worth while to point out that the drum-ends in 

 Cremieu's apparatus ought not to stop the magnetic effect 

 except when they are close up to the gaps between the sectors 

 on the mica. For if they are some distance away, then the 



* Phil. Mag. xxvii. p. 445 (1889). 



