522 
ASTRONOMY: F. H. SHARES 
these involves only the use of the spectroscope; the determination of a 
star's spectral type at once fixes its color, at least within narrow limits. 
But usually spectroscopic observations are concerned with other physi- 
cal factors and we ordinarily think of them as only incidentally determi- 
nations of color. 
The second method depends upon a comparison of photographic 
and visual or photo visual magnitudes. A red star, owing to a defi- 
ciency of light of shorter wave-lengths, makes little impression on the 
ordinary photographic plate. It will therefore be faint photographically 
even though visually bright. The light of a white star, on the other 
hand, affects strongly both the eye and the plate, and its photographic 
and visual brightness are essentially the same. The difference in the 
photographic and visual magnitudes is therefore frequently used as a 
measure of a star's color. This difference is commonly known as the 
color-index. 
The third method determines the color of stars by finding the effective 
or predominating wave-lengths in the light which they emit.^ The 
observations are made by means of a coarse objective grating attached 
to the end of the telescope. The central diffraction image of each star 
in the field will have adjoining it on either side tiny images of the first- 
order spectra. For a red star the blue ends of the spectral images will 
be weak; the distance separating the points of maximum photographic 
density is accordingly greater than for a bluer object. From the ob- 
served value of this distance the wave-length for the point of maximum 
density can be calculated. The result is the effective wave-length. 
Since spectral type is determined mainly by the number, intensity, 
and position of the Fraunhofer lines, its relation to color, which depends 
upon the distribution of intensity in the continuous spectrum back- 
ground, is not necessarily constant. Thus stars of the same type show 
differences in color whose origin may be traced to differences in their 
intrinsic luminosity or absolute magnitude; and any selective absorp- 
tion or scattering of light during its passage through space must also 
introduce an element into the star's color which escapes a determin- 
ation based upon spectral classification.^ The first method, therefore, 
does not necessarily give exact values of the color. Moreover, it can- 
not be used for very faint stars. 
The second method is free from the objections attached to the first, 
but it is not always easy of application. Both the photographic and 
photovisual magnitudes must be known on the absolute scale. Aside 
from the difficulties arising from systematic errors in the slope of the 
scales, there are others, connected with the determination of their zero 
