ASTRONOMY: H. N. RUSSELL 
395 
matter, even in the interval between G and K5, where the differences 
between consecutive types are least prominent. The publication of a 
detailed descriptive 'key' with good reproductions of photographs of 
spectra of each successive class would however be a great boon to 
isolated workers. 
Of much greater importance is the devising of some method for pho- 
tographing the spectra of stars fainter than the tenth magnitude — 
which are now about at the limit of accessibility. Long exposures 
with the objective prism are greatly embarrassed by difficulties in 
guiding, but the problem is doubtless soluble in some way, and ought 
to be solved. 
(c) There is now good reason to believe that the differences between 
the main classes of spectra arise from differences in the effective surface 
temperatures of the stars, and that differences in their other physical 
characteristics play only a minor role in the spectra, but reveal them- 
selves in differences in detail, formerly described as 'peculiarities' 
when they were noticed at all. The investigation of these finer differ- 
ences is to-day the most promising field in stellar spectroscopy. 
What valuable results may be obtained was shown by Hertzsprung's^ 
work on Miss Maury's 'c-stars' (with unusually sharp spectral lines) 
which prove to be of greater real brightness than any other class of 
stars so far known; and later, and still more remarkably, by Adams' 
and Kohlschiitter's^^ discovery that the absolute magnitudes of stars 
(of the 'later' spectral classes, at least) can be predicted with surprising 
accuracy from the relative intensity of a few pairs of lines in their spectra. 
The data for stars of great luminosity are still scanty, but should be 
easily obtainable, using the hundreds of spectrograms now available at 
the great observatories, and determining the mean absolute magnitude 
of groups of stars, which the spectroscopic method indicates as being 
of similar brightness, by means of their parallactic motions. When 
this has been done, our knowledge of the distribution of the naked-eye 
stars in space will be very greatly advanced. 
The careful comparison of the spectra of pairs of stars otherwise simi- 
lar, but known to differ in other characteristics than absolute magnitude, 
may yield results of importance. Many recognizable spectral 'peculi- 
arities' too, such as the diffuseness or sharpness of the lines, the pres- 
ence of bright lines, the abnormal strength or weakness of certain 
lines, etc., have as yet been very incompletely studied, especially as 
regards their relation to other characteristics of the stars. For ex- 
ample, it should be possible to distinguish between widening of spectral 
lines due to a star's rotation, (which would affect all lines alike), and 
