268 ASTRONOMY: H. SHAPLEY 
hypothesis into the conclusion that such light scattering is practically 
negligible in all parts of the sky awaits, first, the verification of the origi- 
nal observational result; second, the confirmation of the proposed phys- 
ical interpretation of such data; and, finally, an investigation of the faint 
and most distant stars in a wide range of galactic latitudes and longi- 
tudes. The present note contains a further contribution to the subject 
and permits a fairly definite solution. 
In the first study of the magnitudes of the stars in the Hercules 
cluster normal values of color indices were obtained. Among a thou- 
sand objects no abnormally large or small values were found; and 
among the fainter stars, photo visual magnitude 14 to 16, a great number 
of negative indices appeared. The essential correctness of this result 
has now been checked in two different ways — by a new series of polar 
comparisons, and through a determination of the color indices by the 
method of exposure ratios.^ 
Assuming the observed colors to be reliable, the presence of negative 
color indicates two significant conditions, namely, the absence of light 
scattering, and the great distance and size of the cluster. There ap- 
pears to be no adequate reason for modifying this adopted interpreta- 
tion of the blue faint stars. The decrease in the intensity of light by 
terrestrial atmospheric scattering is known to vary inversely as the 
fourth power of the wave length. The dissipated atmospheres and 
meteoric dust, to whose agency spatial light attenuation is commonly 
ascribed, are recognized as similarly capable of the preferential scatter- 
ing of blue light. Such a differential light absorption certainly would 
make all stars appear redder. The presence of very blue stars, there- 
fore, points to the inefficiency of the light-scattering residue in space, 
provided the objects involved are sufficiently remote. As all studies of 
B-type stars (negative color indices) indicate a high absolute luminosity 
and relatively small dispersion around a mean value, such stars, when 
apparently faint, are an immediate and apparently reliable index of 
great distance, and distance is the most important factor in testing the 
scattering of light in space. ^ 
The systematic increase of redness observed toward the center of 
the Hercules cluster* may be interpreted as a spurious effect caused 
by the crowding of images on the photographic plate, or as a scattering 
of light in the denser portion of the cluster itself. The former is almost 
surely the true explanation. The possibility, however, that the latter 
cause contributes, suggests that the absence of light absorption in the 
direction of the Hercules cluster does not imply an inappreciable 
absorption in the lower galactic latitudes where stellar material appar- 
