236 CARNEGIE INSTITUTION OF WASHINGTON. 



giants and dwarfs of the same spectral type. The adopted values of the 

 surface brightness were checked by comparing Mr. Pease's measures of the 

 angular diameters of 6 stars with their theoretical diameters based upon the 

 data in question. The agreement is excellent and shows that the adopted 

 values of the surface brightness for types later than GO are probably near the 

 truth. The resulting curves of constant mean density as a function of spectral 

 type and absolute magnitude are nearly rectilinear and almost parallel to 

 the line of maximum frequency for the dwarfs in the diagram of Russell. 



The uncertainty involved in the assumption that the masses of the giant 

 stars can be obtained from an application of the principle of equipartition, 

 made an independent test of the results highly desirable. The Cepheid 

 variables were used for this purpose. Almost any type of pulsation theory 

 of variation requires that the period should vary inversely as the square root 

 of the mean density of the star. On the assumption of this type of variation, 

 it is thus possible to calculate relative values of the masses and densities of the 

 Cepheids from their periods and spectral types alone. A comparison of 

 results thus found with the distribution of mass and density obtained from 

 the principle of equipartition showed a variation of the right order of magni- 

 tude. The distribution curves were modified somewhat in order to represent 

 closely the theoretical change of mass and density for the Cepheid variables, 

 and, as thus revised, are adopted provisionally. 



A further indication that the results are of the right order of magnitude is 

 found in the agreement of the densities with those derived from ecHpsing 

 binaries and in the coincidence of the equal-mass line M = 2 with the line of 

 maximum frequency of the late-type giants in Russell's diagram. The signifi- 

 cance of this coincidence lies in the fact that Eddington has found on theoreti- 

 cal grounds that the value of the mass occurring most frequently among the 

 giants should be approximately twice the mass of the sun. The details of the 

 investigation, together with a discussion of its relation to Eddington's theory 

 of stellar constitution, have been published in Contribution No. 226. 



INTERFEROMETER MEASUREMENTS OF STELLAR DIAMETERS. 



The observations made by Mr. Pease with the 20-foot interferometer beam 

 mounted upon the 100-inch telescope have been carried on with two objects 

 in view: first, the determination of maximum visibility for the brighter stars 

 at a separation of the mirrors of 19 feet; second, a study of the influence of 

 conditions of seeing upon the visibility curves. For this purpose early-type 

 stars of small hypothetical angular diameter have been used, and the distance 

 between the mirrors has been varied throughout a wide range. 



The following are the principal results of the direct observations for angular 

 diameter: 



(1) The measures of a Orionis and a Scorpii remain essentially the same 

 as those obtained previously. In the case of a Orionis a slightly smaller dis- 

 tance between the mirrors (indicating a larger angular diameter) is perhaps 

 to be ascribed to poorer average conditions of seeing. 



(2) The visibility curves of a Tauri, a Bootis, and /3 Pegasi are very similar 

 and indicate a disappearance of the fringes at a separation of the mirrors of 

 about 22 feet. The correction for seeing, at present somewhat uncertain, 

 will probably increase this value to 25 feet. 



