232 
ASTRONOMY: F. H. SEARES 
diction has been found between the spectroscopic and the trigonometric re- 
sults. This is the more remarkable since we should expect occasional idiosyn- 
cracies in the spectra of stars subject to exceptional physical conditions, and the 
rareness of such cases is interesting evidence for the uniformity of the devel- 
opment of stellar spectra in general. 
RELATION OF COLOR TO INTRINSIC LUMINOSITY IN STARS OF 
THE SAME SPECTRAL TYPE 
By Frederick H. Seares 
Mount Wilson Observatory, Carnegie Institution of Washington 
Communicated by G. E. Hale. Read before the Academy, April 28, 1919 
It has long been known that the color of a star is intimately related to its 
spectral type. More recently it has been found that color is also correlated 
with intrinsic brightness.^ Thus, two stars having different luminosities, but 
the same spectrum so far as the usual criteria are concerned, may differ in 
color-index by more than half a magnitude.^ 
The phenomenon is perhaps to be accounted for as follows: Similarity in 
spectrum implies at least approximate equahty in surface brightness. Differ- 
ence in luminosity is therefore mainly a matter of size, the brighter object 
being the larger. But inequality of dimensions doubtless entails differences in 
atmospheric constitution, and hence also in the selective absorption occurring 
in the atmospheres. Since color is determined by the distribution of intensity 
in the continuous spectrum, while type relates more particularly to the char- 
acteristics of the spectral lines, one star may thus appear redder or bluer than 
another, although both are of the same type. 
But whatever the explanation, an extension of the results of Adams, van 
Rhijn, Monk, and others to include a wide range of luminosity for each spectral 
type is immediately desirable, because of their bearing upon the difficult prob- 
lem of stellar constitution. Further, any phenomenon correlated with stellar 
luminosity should be studied to the utmost, on account of its possible impor- 
tance for the determination of interstellar distances. If the intrinsic bright- 
ness of a star can be found, the problem of its parallax is solved. 
The method employed for the measurement of color in securing the prelim- 
inary results here described is that of exposure-ratios^, which is convenient and 
reliable, and much more expeditious than the direct determination of color-index. 
An isochromatic plate exposed behind a yellow filter records the intensity of 
the 'yellow' light received from the star.- The same plate used without filter 
registers the 'blue' light; the longer wave-lengths are then also active, but the 
image, nevertheless, is essentially blue, owing to the relatively great blue- 
sensitiveness of the isochromatic plate. In neither case are we dealing with 
monochromatic light, or even with a very narrow range of color. That, how- 
