14 
ASTRONOMY: H. SHAPLEY 
include a consideration of the reliability of the magnitudes and color 
indices and an extended analysis of the observational data. For the 
present communication a part of the results bearing on the absorption of 
light in space are shown in a condensed form in the following table. The 
magnitudes of the stars in the denser regions of the cluster, within 2' 
of the center, are omitted, as none of the polar standards upon which the 
determination of the magnitude is based falls within that area, and 
consequently no control of the possible systematic errors of a photo- 
Frequency of Colors in Messier 13 — Number of Stars Tabulated 
COLOR CLASS 
FROM CENTER OF 
CLUSTER 
bO 
b5 
aO 
a5 
fo 
fs 
gO 
g5 
kO 
k5 
to 
to 
to 
to 
to 
to 
to 
to 
to 
to 
b5 
aO 
a5 
fO 
fS 
go 
gs 
kO 
kS 
mO 
2' to 3' 
7 
39 
10 
16 
20 
50 
50 
17 
5 
1 
3' to 5' 
8 
20 
21 
7 
23 
59 
44 
9 
4 
2 
>5' 
1 
11 
5 
4 
7 
26 
21 
7 
0 
1 
Totals 
16 
70 
36 
27 
50 
135 
115 
33 
9 
4 
graphic nature is assured. The color class has been defined in the Sep- 
tember, 1915, number of these Proceedings; in brief we may say that, 
under the average conditions of luminosity and distance that obtain 
for the brighter nearby stars, color class and spectral class are practically 
identical, but if considerable absorption exists the two do not corre- 
spond for more distant objects, redder color being associated with bluer 
spectrum. 
The most remarkable feature of the distribution of color class exhibited 
in the table is not the relative paucity of a's and early /'s, nor the great 
range in color, but rather the highly significant fact that there are any 
negative color indices {h0-h9). Of the 495 stars more than 17 % are of 
color class h, and of these one-fifth are bluer than represented by hS. 
Moreover, we find no excessively large color indices, and none unusually 
small. A comparison of these colors with those derived by Parkhurst 
for bright stars near the north pole shows an entirely analagous distri- 
bution among the various classes. There seems to be no reason to doubt 
the present result, and we are left with the conviction that, so far as 
present data go, normal negative color indices exist in the Hercules cluster 
in large numbers. From the remarks of the preceding paragraphs it is 
at once obvious that such a condition could not co-exist with measurable 
selective absorption of light in space. 
The distance of the Hercules cluster, we may be sure, is not less than 
