406 
ASTRONOMY: H. N. RUSSELL 
spectrum, color, radial velocity, proper motion, distribution in space^ 
etc. (bearing in mind that the limits of telescopic resolution restrict 
our knowledge of the remoter pairs). But Eddington's recent theo- 
retical researches ^2 lead to the hope that it may some day be possible to 
estimate the mass of any star when its absolute magnitude and spectral 
type are accurately known (using the data for double stars as a guide) . 
(a) As regards the determination of the masses of individual stars, it 
should be borne in mind that, for statistical purposes, a pair in which the 
relative motion of the components is known, though the motion in angle 
may be only a few degrees, is very nearly as valuable as one which has 
completed a revolution — while a pair for which the relative motion is 
unknown is of no use at all. The slowly moving pairs which are often, 
but inaccurately, described as 'fixed,^ possess an importance exactly 
analogous to the stars of small proper motion, and give us invaluable 
information about those stars which are bright in proportion to their 
mass — the giant stars, in fact. Now that the discovery of double stars is 
apparently in sight of completion, it is to be hoped that more attention 
may be given to the problem of determining the relative motion in as 
many systems as possible. 
(b) The existing data suffice to show that the masses of the stars differ 
from one another less than any other of their characteristics — the whole 
range among well determined masses being from 20 times the Sun's 
mass to one-sixth of the Sun's, which may be compared with a range in 
luminosity of at least ten million fold. For this very reason, very 
careful observations are required to enable us to say with certainty 
that one star is more or less massive than another. It appears certain 
that the stars of spectrum B are unusually massive,^^ and there is suffi- 
cient evidence to show that, in general, stars of great luminosity are 
more massive than those of small absolute brightness, and that, among 
the dwarf stars, those of 4ater' spectral type are of smaller average 
mass." But there are very few cases in which we can be sure that a 
given star is more or less massive than the average for its type. 
It is very desirable to determine how great is the range of difference 
among the masses of stars of similar spectral class or absolute magnitude. 
Extremely precise determinations of parallax will be needed if this 
problem is to be solved, but the effort will be well worth while. Suffi- 
ciently reliable values of the mean masses of stars of different groups 
have already been determined, to make it possible to estimate the 
parallaxes of all but the nearer binaries and 'physical pairs' more 
accurately than they can at present be observed.^^ This should be of 
aid in the interpretation of other statistical studies of double stars, 
