606 
THE EAEL OF EOSSE ON THE EADIATION 
prevalent on many nights, so that any increase of the number of sets was almost 
powerless to obtain a more reliable result. Irrespective of the number of sets and 
excluding, as a matter of course, the first and last determinations, the mean error is 
rhlO‘4 per cent. 
For the sake of comparison we may state that Seidel found the probable error of a 
single complete determination (consisting of four readings of his photometer) of the 
relative brightness of two stars of the first magnitude to be y^, or 8*3 per cent. By 
inspection of the curve it will be perceived that the maximum of heat appears to take 
place a little before full moon, hut that the heat at the time of the first quarter is 
rather less than at the last quarter ; and also that the average heat during the period 
from first quarter to full moon is less than during the corresponding period after full. 
Both these departures of the curve from symmetry on the two sides of the full appear 
somewhat too large to be ascribed entirely to accidental error in the observations. 
Possibly the considerable inequality in the distribution of the mountainous regions 
and the plains and so-called “ seas ” between the preceding and following halves of the 
moon’s visible surface may be looked to as the cause ; and it may be desirable to direct 
attention to this point in a future course of observations, but for the present it would 
evidently be premature to speculate further. As to the amount of the difference, it is 
found by examination of the curve that at the first quarter the amount is 63*3, and at 
the last quarter 78‘0, that just before full moon being 407 - 3. 
In conjunction with the foregoing experiments determinations of the proportion of the 
moon’s heat transmitted by a plate of glass were occasionally made. The following is a 
Table corresponding to those at pp. 589-592 and 600-603 for the moon’s unobstructed 
heat. 
