450 Messrs. Y. Cremieu and H. Pender on 



o£ nodes of electric oscillations such as formed by the disks. 

 a description of which Cremieu has recently given *. He 

 was therefore not absolutely certain that the deflexions he 

 observed were not due to such oscillations, the effects of 

 which might be easily confused with those due to a permanent 

 magnetic field presumably due to the convection-current. 

 These considerations led us to modifv both our mode of 

 operation and the system used to detect the magnetic field. 



New Mode of Oijeration. — In all the experiments performed 

 up to the present by other experimenters and ourselves, the 

 effect observed was always that resulting from charging or 

 reversing the sign of the charge on the disks already in 

 rotation. Of course care was always taken that there was 

 never an effect produced by charging or reversing the charge 

 on the disks at rest, which condition can be realized by 

 properly inclosing the astatic system (or galvanometer- 

 circuit) in a suitable conducting screen. However, the action 

 of such a screen is both electrostatic and electromagnetic. 

 Besides, it is well known that in the neighbourhood of a 

 node of electrical oscillations, there are produced rapidly 

 damped magnetic waves capable of demagnetizing a magnet; 

 and such a demagnetization can result in a permanent change 

 of the position of equilibrium of the astatic system. More- 

 over, it is unknown what effect the movement of the body 

 forming the node for these oscillations can have upon the 

 oscillations themselves. Consequently one cannot assume 

 a i^riori that a screen which is sufficient to suppress any 

 effect of these oscillations when the disk is at rest will also be 

 sufficient when the disk is in motion. 



The following mode of operation appears to us to avoid all 

 these inconveniences : — First, charge the disk at rest, care 

 being taken that this operation produces no effect on the 

 magnetic system. Then insulate the disk from the charge- 

 source and set it in motion. The magnetic system should 

 then, if the effect of convection exists, take a deflexion 

 increasing with the velocity, permanent for any given velocitv, 

 and returning to zero at the moment the disk is stopped. 

 An electrometric measurement of the potential of the disk 

 before and after movement will show if there has been anv 

 sensible leakage. 



As ordinary astatic systems are poorly suited for such an 

 experiment, we attempted to construct a system of a different 

 kind. 



Experiments until New Magnetic Systems. — It is difficult to 

 realize, in most laboratories, a magnetic stability sufficient to 



* Comptes Rendus, toL cxxxv. p. 153 (1902;. 



