ASTRONOMY: H. N. RUSSELL 
403 
of statistical investigations be cleared from the effects of the egregious 
observational preference for the brighter and remoter stars. 
(b) Kapteyn^^ has obtained fairly good values of the dispersion among 
the various divisions of Class B, and provisional values for Class A; 
and Russell^^ has given rough estimates for the dwarf stars, and a still 
rougher one for the giants of Class M : but further work is greatly needed. 
Adams' spectroscopic method offers an easy solution of the problem, as 
soon as his present provisional scale of absolute magnitudes for the giant 
stars has been revised with the aid of studies of the parallactic and 
peculiar motions of groups of stars whose spectra indicate that they are 
similar in real brightness. Stromberg^^ has already shown in this way 
that Adams' mean absolute magnitude for all the giant stars, taken 
together, is substantially correct; but there is evidence that the pro- 
visional estimates for the very brightest stars (such as the Cepheid 
variables) make them considerably too faint.^^ 
(c) The existing evidence indicates that the majority of the stars of 
any given spectral class are confined within surprisingly narrow limits of 
absolute magnitude (provided that the giants and dwarfs can be treated 
separately). But there are exceptions of great interest. For example, 
Kapteyn^^ has shown that (3 Orionis is some eight magnitudes brighter 
than the average for its class (B8); and the faint companions of Sirius^^ 
and Eridani'^^ have spectra of class A, although they are at least eight 
magnitudes fainter than normal stars of this class. Exceptional bright- 
ness is probably explicable by unusual size or mass; but the two excep- 
tionally faint stars (which are known to be of normal mass for stars of 
their brightness) present a real puzzle. Something about the physical 
conditions in these stars must be very unusual, and they should be 
studied with the greatest attainable detail. Other such objects may be 
found among the faint stars of large proper motion. 
6. Beyond the limit of direct measures of parallax, our main reliance 
must be placed on proper motions, which are of fundamental importance 
in the study of the galactic system. 
The brighter stars have already been cared for by Boss, and those 
down to magnitude 7.5 are under discussion. The fainter stars can best 
be investigated by photography, carrying the work to objects as faint 
as can be reached with large instruments, in accordance with Kapteyn's 
^Plan of Selected Areas' or some equivalent. For this purpose, it is 
essential to have a set of reference stars, distributed uniformly over the 
sky, and of suitable brightness to serve as photographic standards, and 
to make the observations strictly differential with respect to these, 
using them not merely as reference points for position when reducing a 
