ASTRONOMY: H. N. RUSSELL 405 
In extending the list of observed radial velocities, much advantage 
has been gained by a policy of selective observation of classes of stars of 
special interest — such as stars of unusually large and small proper mo- 
tion, absolute magnitude, and the like, variable stars, and stars of the 
rarer spectral types. A similar investigation of double stars showing 
evidence of physical connection would be worth while. 
8. Statistical discussions of the motions of the stars and of the Sun, and 
their relation to spectral type, etc., offer an extensive and very intricate 
field. Among the matters demanding further investigation may be 
mentioned the reason for the differences in the direction and velocity 
of the solar motion derived from stars of different spectral types, and 
from proper motions and radial velocities separately; the origin of the 
constant term in radial velocities (Campbell's K term) ; the existence of 
tendencies toward common motion among the stars in particular regions 
of the sky; the dependence of the mean peculiar velocities of the stars 
upon spectral type and absolute magnitude, and the real cause of this 
dependence (possibly a correlation between large velocity and small 
mass); the true nature of preferential motion, and whether it really 
gives evidence of the existence of two physically different ^streams;' 
the dependence of preferential motion upon spectral type, absolute 
magnitude (the latter an unworked field) and perhaps upon the region 
of the sky considered; the devising of a rapid method for the detection 
of moving clusters, and the identification of their members; and so on. 
The discussion of most of these problems should be based simultaneously 
on proper motions and radial velocities. Results derived from either 
one alone may fall into errors which the combination of both would 
detect. 
One practical matter deserves specific mention. When it appears 
desirable to exclude certain stars from a statistical discussion (for ex- 
ample, those of very large proper motion), the limits of exclusion should 
be clearly and precisely stated. Neglect to do so may cause great 
trouble to other workers who wish to make a comparison with their own 
results, and has sometimes led to very serious errors of interpretation. 
9. Another set of data of fundamental importance depend upon rela- 
tions involving the masses of the stars. Here there appears the grave 
difficulty that nothing at all can at present be found out concerning 
the mass of a star unless it is double. There are plenty of double stars, 
to be sure ; but what certainty have we that they are similar in mass to 
stars which are not double? Only an indirect answer is possible, by 
means of the statistical comparison of single and double stars with re- 
spect to as many characteristics as may be — absolute magnitude^ 
