ASTRONOMY: F. H. SEARES 
233 
ever, is of no great consequence; it is only important that the mean wave- 
lengths for 'blue' and 'yellow' should differ by a considerable amount. The 
relative intensity for the two spectral regions wilPthen vary with the color of 
the star, and in consequence may be used as an index of its color. 
Practically, however, it is simpler to use as a measure of color the ratio of 
the exposure times which will yield blue and yellow images of equal intensity, 
or, better still, the logarithm of this ratio. Thus we make a series of exposures 
to blue, usually 2^ 4^ and 8% and then one or more to yellow, such that the 
resulting images fall within the limits of size fixed by the first and last blue 
exposures. For an early-type star the yellow exposures are generally 32^ and 
64^, for an intermediate type 16^ and 32^, and for a late type 8^ and 16^. The 
logarithm of the exposure-ratio for equal blue and yellow images is then esti- 
mated directly from the plate. With a normal emulsion the logarithms for 
the principal spectral types (ratio of blue to yellow for stars of zero absolute 
magnitude) are approximately as follows: Bo, 8.79; Ao, 8.94; Fo, 9.10; Go, 
9.26; Ko, 9.44; Ma, 9.55. 
Troublesome photographic difficulties are avoided by using images of con- 
stant size produced by a constant series of exposure times, compensation for 
differences in apparent magnitude being made by varying the aperture of the 
telescope. With the 60-inch reflector stars as faint as the eleventh magnitude 
can be observed with the exposures given above. 
The photographic emulsion is changed as infrequently as possible, for every 
change necessitates a systematic correction, and indeed each plate usually re- 
quires a small correction. From five to ten stars are ordinarily observed on 
each plate; and, since the adopted value for a star is based on several plates, 
the measures themselves afford a means of reducing the individual plates to a 
homogeneous standard system of colors. 
The measures require a correction for atmospheric extinction, but since all 
the observations on a plate are made at aboUt the same zenith distance, this 
is assumed to be constant and is determined as a part of the systematic cor- 
rection for the plate. Errors depending on seeing and atmospheric conditions 
generally also enter into the plate correction and are thus largely eliminated 
from the results. 
The internal consistency under favorable conditions is illustrated by the 
accompanying observations of the two components of 61 Cygni which ex- 
tend over an interval of about two years. The logarithms of the exposure- 
ratios for the two stars, which were photographed simultaneously, are in the 
first two columns of the table. The average deviation of the differences 
in the third column corresponds to about 0.08 magnitude, which indicates the 
uncertainty in a single determination of the difference in color of two stars on 
.the same plate. This differential result, which is free from systematic plate- 
errors, is equivalent to an uncertainty of 0.05 or 0.06 magnitude in one meas- 
urement of the color of a single star, a quantity of the same order as the aver- 
age deviations found directly from the measures of each of the two components, 
I, 
