446 
Proceedings of the Poycd Society 
of the filter was connected with a glass flask, in which the condition 
of the air was tested. So long as the difference of temperature was 
kept up, and the current not too rapid, the air passing through the 
apparatus showed no signs of producing cloudy condensation on the 
pressure being reduced, showing that the filter had trapped all, even 
the invisible dust particles. 
Some experiments on the effect of diffusion on the distribution of 
dust at the surface of a diaphragm are described. Where carbonic 
acid diffuses into a space, the dust comes close to the diffusing 
surface ; but if hydrogen is the diffusing gas, a clear space is formed 
in front of the diaphragm. 
An explanation is then offered of the repulsion of dust by hot 
surfaces, and its attraction by cold ones. It seemed possible that 
the dust might be repelled in the same way as the vanes of a 
Crooke’s radiometer, by a radiation effect. That this is not the 
true explanation was, however, proved by placing in the dust-box 
a polished silver flat test-surface, one half of which was coated with 
lamp black, when it was found that the dark space in front of the 
lamp black was no thicker than that in front of the polished 
metal. It is thought that the repulsion is due to the diffusion of 
the hot and cold air molecules. The hot surface repels, because the 
outward diffusing molecules are hot, and have greater kinetic energy 
than the inward moving ones ; and as the side of the dust particle 
next the hot surface is bombarded by a larger number of hot 
molecules than the other side, it is driven away from the hot 
surface. The attraction of a cold surface is explained by the less 
kinetic energy of the outward than of the inward diffusing 
molecules. Some experiments are referred to, to show that the rate 
at which gas molecules diffuse indicate that tliis diffusion effect is 
sufficient to account for the repulsion and attraction of the dust. 
If the explanation here given is correct, then the dust is repelled 
in the same way as the vanes of a radiometer when placed in front 
of a surface fixed inside the radiometer bulb, and hotter than the 
residual gas, — the principal part of the energy producing the motion 
being transferred from the hot surface to the repelled surface by 
the kinetic energy of the molecules, and not by radiation. 
Ill illustration of the tendency of dust to move from hot, and to 
