346 The Variable Star Algol. 
peat what I have there said. When the weather is misty, when 
clouds are flying, in bright twilight or too close vicinity to the 
moon, such observations should not be undertaken. It is especially 
important to avoid the impression of any other light, and, when 
the moon is up, to place one’s self in such a position that the 
moon will be hid by some object. Before pg. the eye 
should be for some time accustomed to t arkness, in order 
that the pupil may be dilated as much as tte Then look 
alternately at the two stars which are to be compared, and en- 
deavor to receive the image _ that part of the eye in which it is 
seen the brightest. On the other hand exercise the most zealous 
can never be seen at once in their full brightness. After the eye 
has thus been directed alternately to the one and to the other sev- 
eral times and a distinct judgment formed as to their relative bril- 
liancy, write this down in Sets I denote a decided difference 
in the brightness of two stars by the expression “a grade,”—set 
down the differences of Tach se as 2 grades when I can im- 
agine a third between them, and of a brilliancy decidedly differ- 
ent from either,—as 3, when the difference is so great that two 
others can be thus interpolated in imagination between them ; 
and soon. A greater difference than 4 grades I do not estimate, 
since this mode then becomes too unsafe; but do on the other 
hand, estimate half, and in some cases bven quarter grades. In 
recording, the letter denoting the brighter star is first written 
own, then the number of grades, and dast the letter of the fainter 
star. The precision which may be obtained in this way, after a 
little practice, is very considerable; the probable uncertainty 
amounts, for a single estimate, to about half a grade, and this 1s 
much diminished by comparing the star which is to be deter- 
mined with many others, especially when some of these are 
brighter and others fainter 
As to the computation of the time of the minimum of Algol 
from the observations at hand, I do not content myself with put- 
ting for this the time of least brilliancy ; : but use for this purpose 
all the observations made for half an hour before and after the 
minimum, ky taking the mean of the times at which Algol man- 
ifested the same difference of brightness from the comparison- 
stars during its seers and increase of brilliancy, os then the 
mean of se means, as the final result. ‘Treated in this 
way, the sheer vatiegel afford mele? itciniot., fi that the 
probable error of an observed minimum does not amount to SO 
much as 6 minutes, and el antes of a quater of an hour from 
the mean are extremely rare. 
_ Since the number of observers has increased within @ short 
time, a series of minima are available which have been inde- 
dently determined by different t observers, and, in part, at 
