ASTRONOMY: H. N. RUSSELL 
407 
such as the proportion of double stars among all the stars of a given 
magnitude, the relative numbers of close and wide pairs, etc. 
The determination of the relative masses of the components of binary 
systems will soon also be possible in many cases which have previously 
been somewhat neglected. 
When a sufficient number of accurate determinations of mass have been 
made, a detailed study of the spectra of stars differing in mass should 
be made, in the hope of finding peculiarities depending directly on the 
mass, which might make it possible to estimate the masses of isolated 
stars. 
(c) A great number of spectroscopic binaries await investigation, and 
more are continually being discovered. In the determination of orbits, 
preference should be given to those which show the spectra of both 
components, as it is only in this case that definite information can be 
obtained about the masses. Eclipsing and Cepheid variables are also 
worthy of special attention, and also stars of large proper motion, or 
others which appear to be dwarf stars. 
It is very desirable that some method should be found for observing 
the spectrum of the secondary component when it is too faint to be di- 
rectly seen. Perhaps Koch^s spectromicrometer might furnish a solu- 
tion. Favorable cases for trial, in which the brightness of the invisible 
secondary spectrum is known, may be found among eclipsing variables. 
10. The densities of stars can so far be determined only when they are 
eclipsing variables. In this case, when both spectra can be photo- 
graphed, the diameters of the components can also be found. Several 
systems of this sort, which have not yet been investigated spectro- 
graphically, are within the reach of existing instruments. 
If, however, the relations between spectral type, color index, and 
surface brightness can be so well determined that it is possible to esti- 
mate the last of the three when the other two are known, it will then 
be possible to determine the densities of all visual binary stars, the linear 
diameters of all stars of known parallax, and the angular diameters of 
all the stars in the sky. The known eclipsing variables should afford 
sufficient material for a first investigation of the problem, if only 
sufficiently accurate information can be obtained regarding the color- 
equation of the visual and photographic methods of observation which 
have been employed at various observatories. 
1 1 . All that can be said at present regarding the internal constitution 
of the stars depends on Eddington's theoretical work,^^ which indicates 
that, in the stars of low density, the mass should be greatly condensed 
toward the center — the central density being 54 times the mean den- 
