98 MR. W. CROOKES OK REPULSION RESULTING FROM RADIATION. 
These experiments show that the inner half of the concavity of the cups sends 
a considerable amount of molecular disturbance to react on the glass case, although 
it is not so effective in this respect as the outer half of the concavity. 
406. The metal cups used in these experiments were now lampblacked on the 
inside, and their speed compared with what it had been when they were bright, as 
at fig. 11 , A. It was increased from 35 to 75 revolutions per minute. A mica disk, 
lampblacked on the inside, was now fixed close to the concave side of the cup, 
when the speed rose from 75 to 150 revolutions per minute. Had the blacked mica 
screens been some little distance from the cups, this doubling of the speed would have 
been easily understood, as the screens would have acted as additional driving vanes 
to the radiometer ; but being close to the cup they can only act by absorbing and 
radiating back the heat given out from the bright concave surface of the cups, and 
thereby increasing the difference between the temperature of the black convex and 
bright mica surfaces. 
407. The experiments hitherto described have been with mica screens. Mica 
being a bad conductor and radiator of heat, I next tried replacing it by aluminium, 
to see if the employment of a good conductor as a screen made any radical change 
in the phenomena. A radiometer was constructed like the one shown at fig. 8, 
with the exception that the screens as well as the cups were of aluminium, and the 
outsides of the cups were coated with lampblack, as at fig. 7, E. The diameters of 
the cups and the disks were the same. 
The experiments were tried with the screens in the same positions as are shown 
in fig. 8, so I will refer to that figure in illustrating the results. The screens could 
not be got closer than 4 millims. to the black convex surfaces (intermediate hr posi¬ 
tion between C and D, fig. 8), so I could not ascertain if there was a point of 
neutrality ; hr this position the speed was at the rate of 6 per minute in the positive 
direction. In position G, the screens being as close as possible to the concave sur¬ 
faces of the cups, the positive rotation was at the rate of 120 revolutions a minute ; 
and at any intermediate point between these two extreme positions the speed, in 
the positive direction, was proportionally intermediate between 6 and 120. 
Making allowance for the increased sensitiveness communicated to the cups by 
blacking their convex surfaces, these results agree fairly well with those illustrated 
in fig. 8, and prove that the action of the screens is ahnost entirely one of mechanical 
obstruction, and does not depend on their special power of conducting, absorbing, or 
radiating heat. 
408. In most of the instruments hitherto experimented with, the screens have been 
in such a position in respect to the glass bulb that the secondary action which the 
candle may set up by reason of its shining on the bulb has tended somewhat to 
complicate the results. The cup-shape of the vanes is also not the simplest form with 
which to investigate these reactions. A careful examination of the results obtained 
in the foregoing experiments with screens, throws much light on the physics of the 
