398 
ASTRONOMY: H. N. RUSSELL 
luminosity curve), and would enable us to express our stellar photo- 
metric data in absolute units. 
It is also desirable that methods for measuring the brightness of the 
stars with red and ultra-violet light should be developed, with careful 
determination of the luminosity curve in each case, and of the color 
equation which (for normal stars) makes it possible to reduce color- 
indices obtained on any of these systems to a standard scale. 
The determination of the colors of faint stars by other methods 
affords a promising field, as is shown by the success of the method of 
effective wave-lengths, and of that of exposure ratios, recently 
developed at Mount Wilson. 
Such a determination of exact scales of magnitude and color index is 
evidently a necessary condition for the full utilization of the great mass 
of material which is in process of collection concerning the numbers of 
stars of different magnitudes, their concentration towards the Galaxy, etc. 
(c) The statistical investigation of the relations between color index 
and spectral type, and between both and absolute magnitude, have 
already opened up possibilities of estimating the distances of stars far 
too remote to be reached in any other way. Such investigations 
should be extended, with special reference to stars of great and small 
absolute brightness, and to those having peculiar spectra. 
Closely connected with this is the question of possible selective ab- 
sorption of light in space. Shapley's results,^^ and the theoretical 
work of L. V. King,23 appear to negative the existence of any general 
absorption of this sort. But local selective absorption may occur, and 
it would be well worth while to study intensively the color indices and 
spectra of stars in regions where the existence of absorbing matter is 
suspected, such as Barnard's dark lanes in Scorpius. It is interesting 
in this connection to note that the three most abnormally yellow stars 
of Class B (f, o and ^ Persei)^^ lie within 5° of one another, in a region 
full of diffuse nebulosity. A survey of the stars in this region for 
color-index and spectral type would be well worth while. 
(d) Another interesting problem is presented by the extreme infre- 
quency of very red stars. Color-indices up to about + 1 .8 on the Harvard 
scale are fairly common; but greater values are very unusual, and are 
practically confined to the 'side chain' which includes Class N. In 
this subsidiary sequence the color-indices increase to about -|-4, as 
might be expected as a result of decreasing temperature; but in the 
main series, ending in Class M, this does not happen. Are all the 
stars of Class M of about the same temperature, or is an increase of 
redness in Classes Mb and Mc masked by increasing absorption in the 
