Electromagnetic Efect of Convection- Currents. 455 



was abundant ; in addition, the winter here has been very 

 damp. Most of the trouble of this kind has been due to 

 the glass in the apparatus ; in no case where glass was used 

 as an insulator has it proved satisfactory, not even when the 

 air was dry. First, the stand with glass legs, on which the 

 Leyden-jar battery was placed, was found to furnish an 

 excellent earth-connexion. 



Paraffin blocks interposed stopped this. The reversing- 

 key had three glass rods in it, all of which were found to 

 leak ; six different specimens of glass, some bought par- 

 ticularly for this as insulating glass, were all found to allow 

 great leakage. Shellacing had no effect. Hard rubber was 

 finally substituted for glass ; and after that the key insulated 

 very well, even in damp weather. 



On charging the glass plates, the disks being earthed, 

 it seemed almost as if there was a direct earth-connexion, so 

 rapid was the fall of the charge. This was not regarded at 

 the time, as the plates were always kept earthed ; but later, 

 when it became necessary to charge the plates, the insulation 

 bad to be made good. 



Investigation showed that this was caused by leakage 

 directly through the substance of the glass to the brass back- 

 pieces (HH). Hard rubber pieces were substituted, and the 

 trouble was entirely removed. 



There was at first a deflexion in reversing the electrification 

 while the disks were at rest. This was of course due to direct 

 electrostatic effect ; but it was not for some time clear where 

 the point of weakness in the electrostatic screen lay. It was 

 found to be the faulty contact between the tinfoil covering of 

 the glass tube and the brass collar ; the brass had been 

 lacquered. After this was corrected there was never again 

 any deflexion on reversing the charge, although the pre- 

 caution was taken of testing it every day or so. 



The currents induced in the axle by the rotation caused no 

 inconvenience ; if the disks are rotated in the same direction 

 their effect is added, while the effect of the axles is in opposite 

 directions. Even when the disks were rotated oppositely, the 

 deflexion due to the axles was only 3 or 4 cm., and remained 

 perfectly constant. 



On running the disks, unelectrified, without the glass plates 

 between them and the needle, a deflexion of 4 or 5 cm. was 

 noticed. This was perfectly steady deflexion, and could 

 easily be shown to be due to the presence of the plate, as it 

 ceased when the plates were replaced. 



This was very troublesome for a time, especially as the 

 presence of a brass plate in place of the glass was found to 



