232 Messrs. Lodge and Clark on Dusty Air in the 



we still believe it modifies the effects, and that sometimes 

 notably. We call it the settling hypothesis, and it may be 

 stated thus : — 



Dust falls through fluids at a pace depending on the size 

 of the particle and the viscosity of the fluid, but the relative 

 settling velocity is not affected by the motion of the fluid, 

 whether horizontal or vertical. We may regard all dust as 

 constantly settling relatively to the fluid in which it is. 



By so settling, it may leave dust-free spaces. 



Under a horizontal plate, therefore, the dust might settle, 

 and leave a clear dust-free coat; so it might also perhaps im- 

 mediately under the middle of a round rod; and it is just 

 conceivable that air from this clear space might get partially 

 carried up and round the rod by convection-currents no faster 

 than fresh settling kept on renewing it below. 



It is undeniable that gravitation must assist the formation 

 of coats on the under surface of plates, and must do its best 

 to spoil any coat on the top. The action of gravitation in 

 this way is described in various parts of the paper. It is im- 

 possible, however, for any one witnessing a really good dark 

 coat under a warm rod, and seeing the rushing-up of the dust 

 against the dust-free coat, to go away with the notion that 

 gravitative settling has produced that sharp, definite, and 

 thick coat against which the up-rushing dust-particles almost 

 seem to rebound. The idea is irresistibly suggested of some- 

 thing keeping them forcibly off. 



Moreover, suppose a plate to be originally at the tempera- 

 ture of the air: it has no coat. Turn on a powerful light : 

 a coat instantly forms, and rapidly thickens under the eye 

 before any convection has time to begin. It looks just as if 

 the dust were driven out from its surface by some action 

 which depends upon the temperature. 



Again and again have we been driven to the notion of a 

 molecular bombardment carried on from the surface of the 

 body, by which the dust-particles are driven away like the 

 vanes in Mr. Crookes's radiometer, or like drops in the sphe- 

 roidal condition. But we were unable to take up precisely 

 this position because of the enormous thickness of the dust- 

 free coat as compared with the mean free path of an air- 

 molecule at the ordinary pressure. 



We then tried to develop a notion of an extreme free path : 

 we thought whether the dust-particles might not be so easily 

 moved that the impact of even a few molecules on their sur- 

 face would be sufficient to drive them back; and we tried to 

 imagine that, though the great bulk of the flying molecules 

 only shoot a distance from the surface reckoned in hundred 



