536 
ME. W. CEOOKES ON EEPULSION EESITLTING FEOM EADIATION. 
maximum in about 11 seconds ; the spot of light now returns a few degrees, and then 
proceeds in the first direction to a greater extent than at first. So it goes on, by alter- 
nate steps, advancing a little each oscillation, until, if the light be feeble, the index 
takes up a nearly fixed position ; if, however, the light be strong, the beam is driven 
against the side of the tube. In illustration of this I select the following series of obser- 
vations from a large number recorded in my note-book. The horizontal figures represent 
Fig. 9. 
Degrees on scale, representing repulsion. 
the degrees on the scale, starting from zero, where the spot of light normally rests. 
The vertical figures represent the seconds during which the experiment lasted. The 
zigzag line represents the oscillations of the spot of light, and shows the movement of 
the pith surface under the influence of a uniform source of radiation. The time was 
recorded by a chronograph. Starting from zero the spot of light is seen to have tra- 
velled to 97° in 11*5 seconds ; at the end of 11 more seconds, or 22'5 seconds altogether, 
it had comeback to 50°; at the end of 34 seconds the light had advanced again to 109°, 
and so on. The movements are tolerably uniform as to time, taking about 11*5 seconds 
for the half oscillation, but the amplitude of vibratioD is continually diminishing. 
107. If, however, the light is only allowed to shine on the pith surface for 11*5 seconds 
(or for as long as the spot of light takes to perform its first half oscillation), and if it is 
then instantly cut off, the spot of light almost invariably returns to zero and stops there, 
