now in progress.

[Illustration: Fig. 1 and Fig. 2]

Fig. 1 shows the appearance when looking along a copper or carbon rod
laterally illuminated; the paths of the dust particles are roughly
indicated. Fig. 2 shows the coat on a semi-cylinder of sheet copper
with the concave side turned toward the light.

It is difficult to give the full explanation of the dust free spaces
in a few words, but we may say roughly that there is a molecular
bombardment from all warm surfaces by means of which small suspended
bodies get driven outward and kept away from the surface. It is a sort
of differential bombardment of the gas molecules on the two faces of a
dust particle somewhat analogous to the action on Mr. Crookes'
radiometer vanes. Near cold surfaces the bombardment is very feeble,
and if they are cold enough it appears to act toward the body, driving
the dust inward--at any rate, there is no outward bombardment
sufficient to keep the dust away, and bodies colder than the
atmosphere surrounding them soon get dusty. Thus if I hold this piece
of glass in a magnesium flame, or in a turpentine or camphor flame, it
quickly gets covered with smoke--white in the one case, black in the
other. I take two conical flasks with their surfaces blackened with
camphor black, and filling one with ice, the other with boiling water,
I cork them and put a bell jar over them, under which I burn some
magnesium wire; in a quarter of an hour or so we find that the cold
one is white and hoary, the hot one has only a few larger specks of
dust on it, these being of such size that the bombardment was unable
to sustain their weight, and they have settled by gravitation. We thus
see that when the air in a room is warmer than the solids in it--as