480 
ASTRONOMY: H. SHAPLEY 
As much of Bailey's work preceded the establishment of the North 
Polar Sequence, his values of the magnitudes and ranges of variation 
may now be standardized. A sample of the revised data is given in 
table 1. The variables differ very little from each other in any respect, 
and particularly significant is the striking similarity of the median 
magnitudes. Bailey derived Hght curves and periods for 110 variables 
in this cluster, and 54 of them were chosen as fairly free from uncer- 
tainty. The median magnitude (defined as the mean of maximum 
and minimum) of these 54 stars is 15.49, on the Mount Wilson system, 
with a probable error of less than a hundredth of a magnitude. The 
average deviation from this mean for a single star is =±=0.07, and the 
largest deviation is less than two-tenths of a magnitude. If all 110 
variables are used, the mean is 15.50=^0.006, and the average deviation 
is =±=0.08. The distribution of the residuals follows the law of error as 
closely as could be expected for a small number of values. 
TABLE 1 
STAR 
PERIOD 
bailey's magnitude 
mount WILSON PHOTOGRAPHIC MAGNITUDE 
median 
magnitude 
Maximum 
Minimum 
Maximum 
Minimum 
Range 
days 
9 
0.542 
14.60 
16.76 
14.82 
16.15 
1.33 
15.48 
10 
0.570 
14.75 
16.70 
14.92 
16.11 
1.19 
15.52 
27 
0.580 
15.00 
16.68 
15.06 
16.10 
1.04 
15.58 
34 
0.559 
15.00 
16.70 
15.06 
16.11 
1.05 
15.58 
38 
0.561 
14.80 
16.83 
14.95 
16.19 
1.24 
15.57 
40 
0.552 
14.70 
16.83 
14.88 
16.19 
1.31 
15.54 
49 
0.548 
14.76 
16.76 
14.93 
16.15 
1.22 
15.54 
63 
0.570 
14.80 
16.62 
14.95 
16.06 
1.11 
15.50 
80 
0.539 
14.74 
16.70 
14.91 
16.11 
1.20 
15.51 
Great extremes of absolute brightness are known to exist in Messier 
3 — nine or ten magnitudes, at least. More than 20,000 stars fainter 
than these variables have been photographed at Mount Wilson, but 
no other variables of this or other types have been found among them. 
The result is confirmed by an examination of long-exposure plates at 
Harvard. The periodic light variations are apparently confined to a 
narrow interval of brightness. 
Since the deviations of the median magnitudes from their mean are 
far within the errors of observation, the conclusion is forced upon us 
that in Messier 3 short-period variation is associated with stars of a 
very definite intrinsic luminosity. The situation of the variables in a 
distant globular system necessitates the equality of absolute as well as 
of apparent magnitudes. Moreover, the work on color, so far as it has 
