204 CARNEGIE INSTITUTION OF WASHINGTON. 



from that derived from the general spectrum. The behavior of the 

 color in such cases is of interest, and probably of importance, in view 

 of its relation to temperature. 



In this connection it may be remarked that Mr. Seares and Mr. 

 Shapley have found that in certain Cepheid variables the color seems 

 to be closely correlated with the spectral type as indicated by the 

 hydrogen lines. Both vary periodically with the changes in brightness, 

 while the general spectrum, as shown by the observations of Mr. 

 Adams and Mr. Joy, remains nearly or quite constant. 



The Period-Luminosity Curve of Cepheid Variation. 



A very remarkable relation between the apparent magnitudes and 

 periods of Cepheid variables was first pointed out by Miss Leavitt for 

 25 stars of the small Magellanic cloud. Mr. Shapley has further 

 investigated the relation, which is in fact a correlation of period with 

 absolute luminosity, and finds that it holds for the isolated variables 

 of the galactic system and for the Cepheids, whether of the ordinary 

 kind or of the "Antalgol" type, in at least six globular clusters. On 

 the basis of light curves of more than 200 variables, a definitive period- 

 luminosity curve has been obtained which gives uniquely the absolute 

 magnitude of a typical Cepheid when its period of light variation is 

 known. Since the parallax of a star is a simple function of its absolute 

 and apparent brightness, the distance of a Cepheid, or of any stellar 

 system containing such variables, is immediately obtainable from 

 determinations of period and apparent magnitude. 



The linear formula connecting logarithm of period and magnitude, 

 as suggested by Miss Leavitt, is found to be a first approximation 

 only. The adopted curve indicates a definite lower limit for the 

 absolute brightness of Cepheid variables; and from a study of the 

 luminosity curves in globular clusters this same limit appears to be 

 closely related to the blue end of the giant series. In other words, 

 typical Cepheid variation is apparently impossible, except for giant 

 stars of low density. While the double-star theories of Cepheid vari- 

 ation afford no plausible explanation of the period-luminosity curve, 

 a pulsation hypothesis appears competent to interpret at least its 

 general features. 



Photometry of Star-Clusters. 



In last year's report on Mr. Shapley's investigation of star-clusters, 

 reference was made to the determination of the relative distances of 

 globular systems that contain typical short-period variable stars. 

 The period-luminosity curve described above is at the basis of the 

 most reliable method for the determination of their absolute distances. 

 Through the development of methods for other classes of stars and 

 the use of the marked correlations between the parallax, the brightness, 

 and the diameter of cluster images on photographic plates and charts. 



