b'20 Mr. K Campbell on the Effect of Change of 



these particles collect on the walls, or on the floor. I£ a 

 source of heat is applied, convection currents are set up 

 which carry the particles, bearing a charge of the same sign 

 as that of the walls, into the body of the gas and into the 

 neighbourhood of the electrode. In approaching the electrode 

 they will induce on it a charge of such a sign that it appears 

 to increase the leak due to the field ; but since they move but 

 slowly under the field, they will not increase the steady 

 current appreciably. When the source of heat is removed, 

 the convection currents become less violent and the particles 

 sink again to the floor of the vessel ; in leaving the neigh- 

 bourhood of the electrode, they induce on it a charge of 

 such a sign that it appears as a decrease of the steady leak. 

 If the number of the particles is sufficiently great, the induced 

 electrification may be great enough to cause the appearance 

 of a temporary reversal of the current. 



18. If this be indeed the true explanation, the heavy 

 particles must have masses of an altogether different order to 

 those of the ions. If X is the electric intensity (seldom more 

 than ^ el. st. unit in these experiments), e the charge, and 

 m the mass of a particle such that its weight is equal to the 

 electrical force on it, 



981 

 Xe = mg, or e/m = — - = 3*9 x 10 3 . 



4 



e/m for an oxygen molecule is about 2 x 13 13 in the same units : 

 hence the particles must contain about 5x .10° molecules, the 

 number which is contained in a sphere of air of 6 X 10 -4 cms. 

 radius. If the effects did not increase with the ionization in 

 the vessel, these particles might be safely viewed as fragments 

 of the metal walls ; but, as it is, it seems probable that they 

 must consist, at least in part, of aggregations of the molecules 

 of the gas. 



19. Three attempts have been made to test the truth of this 

 hypothesis. 



(1) The air in the big vessel was stirred up by working a 

 fan in the vessel by means of a shaft which passed through 

 a stuffing-box in the side. If the heat-changes are due 

 to convection currents, they should be reproduced by the 

 action of the fan. It was thought that small changes of 

 the nature expected were observed. In a series of 50 obser- 

 vations, the mean value of t' (just after starting the fan) was 

 5 per cent, less, and the mean value of i" (just after stopping 

 the fan) was 4 per cent, greater than the mean value of t. 

 It was only to be expected that the effects should be small ; 

 for the currents produced by the fan differ markedly from 



