0. N. Rood—Force involved in Onoulier Radiometer. 409 
its true position as much as 20°. The cause of this phenomenon 
I take to be, that the dark heat from the face of the observer 
warms mainly that part of the flask directly opposite him ; this 
causes a feeble ascending current of air in that region, and this 
again a horizontal current toward the heated wall, the latter 
current acting on the disc precisely as the wind does on a 
weather-cock. The deflection from this cause was pretty con- 
stant, and did not prevent qualitative observations from bein 
made, though in the more important cases I avoided it alto- 
gether by quickly taking the reading with the aid of a screen, 
and then retiring for some minutes to the other side of the 
toom before taking a second reading. 
Experiment with gas flame and a solution of alum.—The light 
ofa luminous gas flame at a distance of a foot, after passing 
through a saturated solution of alum 41 mm. thick, was allowed 
to act on the blackened side of the disc, and forty readings 
were taken at intervals of about twenty seconds. A repulsion 
from the light was produced; below is given the deviation 
indicated by the mean of each successive set of ten readings. 
oo. aA 3°01 4°]. 
glass by the naked gas-flame, causing thus a current to set 
toward it, and drawing the disc in that direction after the 
manner of a weather-cock, I moistened the side of the glass 
hext to the gas flame with water at. the temperature of the 
toom, by the aid of a small camels’ hair brush. For a second 
ke confirm the idea I entertained. 
hile making the observations just mentioned, I was seated 
minutes, t 
eapring each time to a considerable distance from the appara- 
us. Below are the results of three experiments, consisting each 
of only two d 
flection ; 16°, 15°-4, 15°-8 away from light. 
Pcl with these conditions, the dise under the full atmos- 
Pheric pressure imitates the behavior of one suspended in a vacuum. 
