ASTRONOMY: H. SHAPLEY 
459 
It is expected that the complete paper will appear as a publication of 
the Carnegie Institution of Washington. 
^ Rosancff, A. J., and Florence Orr, A study of insanity in the light of the Mendelian 
theory, Amer. J. Insanity, 68, 221-261 (1911), also Eugenics Record Office Bull., No. 5. 
2 Riidin, E,, Einige Wege und Ziele der Familienforschung, mit Rucksicht aiif die Psy- 
chiatrie, Zs. ges. Neurol. Psychiat., 7 (Hft. 5), 487-585 (1911). 
2 Galton, F., History of Twins, in Inquiries into human faculty and its development (1883), 
reprinted in 'Everyman's Library,' London, Dent, N. Y., Button. 
SECOND TYPE STARS OF LOW MEAN DENSITY 
By Harlow Shapley 
MOUNT WILSON SOLAR OBSERVATORY. CARNEGIE INSTITUTION OF WASHINGTON 
Presented to the Academy, June 29, 1915 
The role played by stellar densities in the theories of the order of 
stellar evolution is easily recognized. If the development of a star 
uniformly follows the spectral sequence B, A, F, G, K, M, that is, if 
in contracting through the agency of its own gravitation its Hght changes 
in color steadily from bluish white to red, then the mean density of the 
B and A type stars must naturally be less than that of the redder stars. 
But, on the other hand, if the earlier as well as the later stages in a 
star's evolution are characterized by the second and third type spectra 
(F, G, K, M), the density of the redder stars should have both small 
values and large values relative to the bluer stars. 
There are two principal Hues of observational evidence that indicate 
the existence of two orders of densities for second type stars : 
1. The recognized existence of enormous differences in the absolute 
brightness of red or yellow stars of essentially identical spectra leads 
immediately to the conclusion that the volumes also differ greatly. 
Now since stellar masses so far as known have not very greatly differing 
values, we are lead to believe that the disparity in size must be accounted 
for by large differences in mean density. For instance, the density 
of our sun, which on this interpretation must be considered a body 
relatively advanced in age, may be from a thousand to a million times 
greater than that of a star of great intrinsic luminosity, such as Alpha 
Aurigae, which has a similar spectrum and presumably therefore a 
comparable suf ace temperature. An objection to this argument in favor 
of very low densities for some of the redder stars may be that our in- 
formation relative to stellar masses is insufficient to prove that the 
large volumes are not to be attributed to large masses with normal 
densities. 
2. The densities of eclipsing binaries, however, can be determined 
directly, independently of the imknown masses; and recent studies of 
