292 
MR. W. CROOKES OK REPULSION RESULTING FROM RADIATION. 
slower, and when the speed was uniform, on timing the revolutions I found the mean 
of seven observations to be 11 revolutions a minute in the positive direction. 
The screen was now placed as in fig. 20, C, so as to cut off the light from the retreat- 
ing cones, and allow it to fall only on the concave cones as they were approaching the 
light. A very slight diminution in velocity took place, but the rotation continued in 
tne same direction as in the two previous instances. The speed was 9'6 revolutions a 
minute. Finally the candle was blown out, and when the cones were still, and the 
bulb cold, the candle was again lighted, the arrangement rema in ing as at C. Rotation 
immediately commenced, the hollow cone advancing towards the light, and the speed 
rapidly increasing till it became uniform at 9-, 5 revolutions a minute. 
These results show that the speed of rotation produced when the candle shines on 
the advancing and retreating cones simultaneously, is practically made up of the sum 
of the speeds attained when the radiation is allowed to fall on either side of the cones 
separately. Adding the speed in position B (1 1 revolutions a minute) to the speed in 
position C (9 - 6 revolutions a minute), will give 20 6 revolutions a minute : a speed 
which, allowing for the imperfection of the candle, is sufficiently near that actually 
observed — namely, 21 - 8 revolutions a minute. 
The candle was brought nearer, and the experiments repeated. The velocities 
were : — 
In position A, 49 '2 revolutions a minute. 
„ B, 25'2 
„ C, 267 
310. The apparent attraction observed in position C, at first offered some difficulty 
in its explanation. Before attempting to theorise I tried further experiments. 
Thinking over the problem its solution appeared less difficult, and it seemed likely 
that a few experiments would give me a simple explanation. In the next experiment 
I varied the material of which the cones were made, and instead of choosing a good 
conductor of heat like aluminium, I took a bad conductor like mica. 
Cones made of clear mica were attached to arms and mounted like a two-vaned 
radiometer. With a candle there was no rotation, both sides being equally repelled, 
and the arm always setting at right angles to the line joining the candle and axis of 
rotation. 
311. Cones were next made of silver flake mica, and similar experiments were tried 
with these as with the aluminium cones. With no screen in front, as in fig. 20, A, 
there was moderate rotation in the direction of the arrows, but there was great 
hesitation in passing the candle. In position B, the concave side being screened off, 
rotation continued at about the same speed as when the screen was absent, or if 
anything a little faster. In position C, the convex side screened from the light, there 
was no movement. 
312. The explanation of the apparent attraction is now clear. The effect of bend- 
