MB. W. CROOKES ON REPULSION RESULTING FROM RADIATION. 
103 
size of the bulb. The positive pressure between a and b is now overweighted by the 
negative pressure between c and cl, and rotation is therefore produced in the direc¬ 
tion of the arrow, or negatively. If we imagine the arm carrying the screens to be 
rotated on the axis, it will be seen that as one screen leaves the convex side of the 
vane, more of the positive rays of force are rendered active, whilst negative rays are cut 
off by the lower screen. Hence, as the screens are moved round, the fly will gradually 
rotate less strongly in the negative direction ; it will become neutral when the screens 
are in such a position that the opposing forces balance; and ultimately it will rotate 
positively. This is what experiments prove to be the case (413). 
I have tested this explanation by comparing with it the whole of the phenomena 
obtained with movable screens in various positions and at different pressures as des¬ 
cribed in pars. 391 to 413, and in no case does it fail. There is therefore high 
probability that it is true. 
417. An apparatus (fig. 15) was constructed not differing in principle from the last, 
but having, in addition to the aluminium hemi-cylinder and movable mica screen, a small 
rotating fly made of clear mica, mounted in such a way that it could be fixed, by means 
of an exterior magnet, in any desired position inside the bulb ; the screen was also 
capable of adjustment by means of another magnet. The aluminium hemi-cylinder 
was in this apparatus immovable. The adjustable indicator, being very small in 
diameter, in comparison to the other parts of the apparatus, and being easily placed in 
any part of the bulb, was expected to afford information as to the intensity and direc¬ 
tion of the lines of pressure when a candle was brought near the bulb. 
It will be impossible within reasonable limits to give more than a few general results 
which I have obtained with this apparatus. Experiments have been tried :— 
a. With the screen in different positions in respect to the hemi-cylinder ; 
b. With the indicator in different parts of the bulb ; 
c. With the candle at different distances from the hemi-cylinder, on one side or the 
other ; and 
d. With the degree of exhaustion varying between wide limits. 
These four conditions have been varied in numerous ways, and each time experi¬ 
ments have been tried both with and without an alteration in some or all of the other 
conditions. Although much time has been spent in this mode of experiment, it will 
be understood that, with so vast a number of possible combinations, only a few of the 
most obvious can be thoroughly investigated. 
418. I will neglect the results at inferior exhaustions, and will confine myself to 
results obtained with a rarefaction of 12 M. 
Fig. 15 gives a perspective view of the working parts of the apparatus; a is the 
aluminium hemi-cylinder rigidly fastened to an arm fixed to the bulb ; s s is the mica 
screen, drawn up close to the concave side of the hemi-cylinder; s' s' shows the same 
screen when moved some distance away. The screen can be moved right round the 
bulb, till it comes into position s", where it is shown touching the convex surface of the 
