Effect of Electrical Convection. 189 



on ebonite plates c c c c 40 cm. square. The external dia- 

 meter of these rings is 40 cm., internal diameter 10 cm. By 

 means of adjusting-screws it is possible to vary the distance 

 between the disks and condensing-plates, and to adjust the 

 latter accurately parallel to the disks. Fixed to the frame 

 of the apparatus is a test-coil T, through which a conduction- 

 current can be sent at will. The frame carrying each disk is 

 fixed separately to the base-plate of the apparatus, so that the 

 two disks can be placed at any desired distance apart. To 

 each disk is attached a speed-counter formed by a cog-wheel 

 with 199 teeth, gearing directly with a worm on the shaft of 

 the disk. The bed-plate was bolted to a heavy sandstone 

 block 40 x 40 X 75 cm., placed at one end of a stout table. 

 At the other end of the table was placed a counter-shaft 

 driven by a 1 H.P. motor 5 metres distant. From large 

 pulleys on the counter-shaft ran small belts to the pulleys on 

 the shafts of the disks. 



On the side of the ebonite cores next the frame are fixed 

 brass rings, diameter 6 cm., width 5 cm., making contact with 

 the gilded surfaces by strips of tinfoil. Contact is made 

 with these rings by brushes of thin brass mounted on the 

 ebonite plates c c c c. The brushes nowhere come within 

 1£ cm. of the frame or the tinfoil condensing-plates. 



At first ebonite disks were employed, but even though they 

 were 3*5 mm. thick they soon became so badly warped that 

 their use was extremely unsatisfactory. Besides the ex- 

 cessive jarring produced when they were set in rapid rotation, 

 it was impossible to make accurate measurements of the dis- 

 tance between disks and condensing-plates. The micanite 

 disks have served admirably. They are as plane and run as 

 true now as when first put up two months ago. 



In the first experiments, the disks were charged by points 

 brought up close to their edge. The effect sought could be 

 observed with this arrangement, but the needle of the galvano- 

 meter, when the latter was connected in series with the coil, 

 was exceedingly unstable when the disks were behw charged 

 and discharged, and often even became demagnetized, so that 

 it was frequently necessary to take the galvanometer apart 

 and re-astatize the needle. After weeks of experimenting it 

 was found that this was due to the sparking at the points, 

 unavoidable in this method of charging. On charging by 

 means of the brushes and rings above describad, so that there 

 was no sparking at the disks, the spot of light, though it 

 never remained perfectly steady, rarely oscillated over 

 10 mm. 



