REACTIONS TO LIGHT AND GRAVITY IN DROSOPHILA ol 
length, to the end of which was fastened an oblong piece of lead 
weighing 410 grams and heavily padded with felt. This pendu- 
lum was then suspended in such a manner that at the end of each 
complete vibration the padded end would strike the tube at its 
middle point. The length of the swing was limited by the tube, 
upon the one hand, and a shelf against which the pendulum hit 
on its backward stroke. This shelf was back of the tube a hori- 
- gontal distance of 50 em. and 22 cm. above it. When the tube 
was placed in position for operation, it was held in place by the 
rubber bands mentioned above. These bands were fastened to 
a couple of tacks on the side of the tube toward the pendulum 
and at equal distances from its center. These were then passed 
under the tube toward the operator, and then over it to the edge 
of the board toward the pendulum, where they were again fas- 
tened. Thus when the pendulum was set in motion it. would 
strike the tube and push it a slight distance away from the nails 
against which the bands tightly held it. These would immediately 
snap it back into position ready to be again displaced by the 
next swing of the pendulum. Thus the tube underwent a con- 
stant jarring of a fixed degree of violence and at regular inter- 
vals. The pendulum was kept in motion by a slight pressure 
of the hand on its upper end, delivered just at the beginning of 
each backward swing. While it might be objected that this pres- 
sure would vary, thus varying the length of the stroke and the 
force of the blow, this variation in practice was found to be very 
slight indeed. This was made possible by the fact that great 
care was used to give exactly the amount of pressure necessary 
to make the pendulum just reach the shelf on its backward 
swing. With a little-practice this action became mechanical and 
exceedingly constant. The degree of accuracy was in fact de- 
termined thus. An assistant counted the number of strokes 
per minute for five separate trials with an interval of a minute 
and a half between each trial. The count each time was 40 
strokes. The strokes were then counted for three consecutive 
minutes, and the count was 118. According to the other test it 
should, of course, have been 120, thus showing the very small 
error of 0.6 of a stroke per minute. No greater care was exer- 
cised during these tests than was normally the case in the experi- 
