302 
MR. W. CROOKES ON REPULSION RESULTING FROM RADIATION. 
335. To further test the theory that the direction of rotation of the curved vanes 
depends not on any special effect of curvature as such but on favourable presentation, 
Professor Stokes suggested that I should make a radiometer having the curved 
vanes sloping, as shown in fig. 29, A, so that the tangent at the extremity a would 
Fig. 29. 
fall below the centre, and the same at a, or at most pass through the centre. Such 
a fly ought to move in the direction of the arrows, or in the same direction as a fly 
made like B (fig. 29), under the same circumstances, although the curvatures lie in 
opposite directions. 
On trying the experiment with the fly A, I found that when exhausted to the 
most sensitive point a candle repelled each face with nearly equal force, and therefore 
no rotation took place, although the tendency was in the direction of the arrows. 
When heated with a hot ring round the equator of the bulb, strong rotation took 
place in the direction of the arrows ; there was no reverse movement on cooling. 
A radiometer fitted with a fly like B revolved very well in the direction of 
the arrows. 
EXPERIMENTS WITH DARK AND LUMINOUS HEAT APPLIED INTERNALLY. 
336. The experiments on the action of dark heat on radiometers with “favourably 
presented ” vanes, tried with hot rings applied above, equatorially, and below (298, 
299, 301), could not give results of very great accuracy, as radiation heated a 
considerable portion of the bulb on each side of the hot ring. In some of the 
observations the results scarcely accorded with theory, and although I could explain 
most of the anomalies, there were irregularities which seemed to point to another 
influence which might cause me to modify the theory of the action of dark heat on the 
vanes. For strict investigation of this it was necessary to contrive a very intense 
source of heat always ready to be applied in the same place ; the heat should not pass 
through glass, and it should be completely under control as to intensity and time of 
action. The vanes also should be turned in the most favourable position for rotating 
under the influence of the molecular pressure, and the apparatus should be capable 
of having the exhaustion carried to a very high point and measured when an obser- 
vation was taken. These advantages are all secured in an apparatus represented 
in fig. 30. 
