96 MR. W. CROOKES ON REPULSION RESULTING FROM RADIATION. 
being greater than that from the concave side. The proportion is roughly 50 to 6, for 
when the molecular pressure is as completely as possible cut off between the convex 
side of the cup and the glass bulb, as in position A, fig. 8, the rotation due to the 
repulsion generated by the concave side is at the rate of 6 revolutions a minute; 
but when the screen is in position G, fig. 8, cutting off the pressure between the 
concave side and the bulb, and allowing that from the convex side fully to act, the 
revolutions become 50 a minute. In position C, the screen is at such a point that 
the active pressure is nearly equal from each side of the cups, and the result is 
neutrality. In the other positions a balance of pressure, represented by the stated 
revolutions per minute, is allowed by the screens to escape and react against the glass 
bulb. 
402. I now endeavoured to ascertain if there was a difference between clear mica 
and blacked mica screens in front of metal cups. A radiometer was made similar to 
the one last experimented with (fig. 8), but the mica screens were lampblacked on the 
sides facing the concave surfaces of the cups. This altered the action considerably. 
At an exhaustion of 47 M, and with the candle 3 inches off, there was no move¬ 
ment when the screen was close to the concave surface, as at fig. 9. When the 
Fig. 9. 
distance was increased to 2 millims., there was very slow positive rotation at the rate 
of 1 in 1-| minute. With the screen midway between the disks, the speed was scarcely 
increased, and when it was 2 millims. from the convex surface the rotation was only at 
the rate of 1 per minute. 
The exhaustion of 47 M was chosen, as that is near the point of maximum sensi¬ 
tiveness for ordinary radiometers (334, 382). I next tried the effect of exhausting 
the instrument to a higher point, at which an ordinary radiometer would begin to 
lose sensitiveness. At an exhaustion of 26 M, the speed was found to be 5 revolu¬ 
tions per minute in the negative direction, when the screens were in the position 
of fig. 9. With the screens 2 millims. from the concave side there was no motion, and 
beyond this point the more the screens approached the convex sides of the cups the 
more decided was the positive rotation. 
403. The mica screens were now arranged in a horizontal plane, the metal cups 
remaining vertical (fig. 10). The exhaustion was 8 M, the candle remaining as before. 
In the position shown by the black lines the speed in the positive direction was 15 
revolutions per minute. When the screens were moved to the positions shown by the 
dotted lines, the speed was reduced to 6 a minute ; and at any intermediate distance 
between these two positions, the velocity of rotation likewise varied in proportion. 
