110 E.. C. Pickering—Measurements of the 
the result for any unobserved difference in brightness of the 
two images is readily determined, 
If p=0, dp= = , its greatest value, which diminishes as the 
polarization increases, becoming zero when p=0. Hence the 
greater the polarization, the more accurately it can be measured. 
If a=80, dp=,;}, for its greatest value; hence the instrument 
should always give results ,within two-thi oO per cent. 
Observation, however, shows that the error is much greater, a 
difference in brightness of ;'; being by no means perceptible. 
e first series of observations were made on the light of the 
sky. The instrument was screwed into a post and levelled, the 
altitude and azimuth of the sun taken, and the instrument then 
directed toward the points tobe observed. Most of these were 
and ordinates the polarization of the observed points. The 
sun’s altitude vaed E 
Before discussing these observations further, it seemed desira- 
ble to determine the polarization of other parts of the atmos- 
the sun should be compared, the polarimeter was so mounted 
that its principal axis would pass through the sun. The two 
circles would then give solar distances, and the angle 
from the vertical plane through the sun, instead of altitudes 
and azimuths. The second of these angles will be called the 
meridian distance, and will be regarded as positive to the right, 
and negative to the left of the sun. Of course, the direction 
of the axis should continually change, so as to follow the sun; 
but as great accuracy in the determination of the angles was not 
needed, it was found sufficient to re-adjust it every few minutes. 
Another advantage of this arrangement was, that the line 0 
junction, being turned parallel to the axis, would always lie in 
