456 Messrs. Y. Cremieu and H. Pender on 



a resistance of 4*38 ohms and sensitive to 10~' ampere. 

 Hence for a convection-current of 4x10"'^ ampere, the 

 current flowing through the galvanometer will Ije of the order 

 of 1*7 X 10~'^ ampere, and the resultant deflexion about 

 170 mm., a deflexion susceptible of great accuracy of 

 measurement. 



The first experiment was to place two fixed metallic 

 brushes so as to rub on the ring under the two points S and 

 Si- However, the friction of the brushes against the un- 

 charged disk was sufficient to produce a deflexion of about 

 100 mm. Moreover, the gold was rapidly rubbed away and 

 the resistance therefore rapidly increased^ so that all we 

 could obtain from this method of procedure were rough 

 qualitative results. 



But, making yet another slight modification, we were 

 enabled to make quite accurate measurements. This was 

 accomplished by employing the following exceedingly simj^le 

 and effective contacts. At the extremities AA of the moving 

 axle (fig. 4), coinciding with the axis of rotation, were fixed 



Fio-. 4. 



aT 



3- 



two very fine copper wires insulated from the axle. Two 

 small glass tubes TTi, all but sealed up at Ti so that the 

 opening there was just a little larger than the diameter of 

 the wire aa, were so placed that the wires turned in the 

 openings T^ without touching the walls of the tube. Before 

 being placed in position these tubes were filled with mercury 

 and electrodes sealed in the outer ends. These contacts 

 proved extremely satisfactory — their resistance when the axle 

 was in motion remained practically the same as when the 

 axle was at rest, an hour''s running producing an increase of 

 resistance of less than five hundredths of an ohm. 



To return to the convection experiment. Two points 

 C and D, diametrically opposite, wxre connected permanently 

 to the wires aa, which in turn were connected through the 

 contacts to the galvanometer. H now SSi is charged and 

 the disk set in motion, there will be produced in the disk 

 conduction-currents distributed as explained above. But as 

 C and D move with the disk, between these two points the 

 difference of potential will be alternating, and consequently 

 in the galvanometer there will be an alternating current and 



