18 Mr Zeleny, On Convection Currents and on the [Oct. 31, 



the rays are turned on both streams of ammonium chloride par- 

 ticles move towards the plate from which the rays are screened. 



By cutting off the rays from one side of the field the gas in 

 the exposed part, which normally moves towards its adjacent plate, 

 now has its motion reversed so that all the gas between the plates 

 is moving towards the screened plate. 



If the distribution of the ions in this case is considered it is 

 seen that this is a necessary consequence. For, of the ions pro- 

 duced in the space exposed to the rays, those which under the 

 electric field move towards the screened side have a longer 

 distance to travel, and while they are in the screened space they 

 constitute a much larger free charge than is normal, for here there 

 are not present any of the ions of the opposite sign which usually 

 neutralize the greater part of the effect. 



The motion of this abnormally large charge produces an air 

 current which sets into motion all of the air between the plates 

 and even overcomes the force near the exposed plate which tends 

 to produce motion in the opposite direction. 



It has been stated that the velocity of the convection currents 

 is accelerated for an appreciable time, so that by finding this 

 acceleration it is possible to get a rough idea of the magnitude of 

 the free charge necessary to produce the observed movements. 



Thus in a particular case where the field was uniformly ex- 

 posed to the rays and where the potential gradient was about 

 60 volts per centimetre, the stream of particles was observed to 

 move 5 millimetres from rest in two seconds. 



Assuming the acceleration uniform for this short period of 

 time, this gives its value as '25 cm. per sec. per sec. 



Assuming further that the charge in each cubic centimetre 

 has only to move this volume of air although it actually has to 

 move rather more, and calling 8 the average density of electrifi- 

 cation, we have approximately by equating the two expressions for 

 the force acting 



8 x J^ = 25 x 1-2 x 10-*. 



This gives for 8 a value of about 10 -3 , a number of the same order 

 as has been observed for the free charges existing in these cases. 



II. The Potential Gradient at the Electrodes. 



It has recently been shown independently by the writer* and 

 by Child f that when conduction is going on between two plates 

 the potential gradient between them is not uniform and that at 

 the plates there is a rapid fall of potential. 



* J. Zeleny, Phil. Mag. July 1898, p. 147. 

 f C. D. Child, Wied. Ann. hxv. p. 152, 1898. 



