MOUNT WILSON OBSERVATORY. 231 



tude. If, as seems probable, the absolute magnitude derived by the spectro- 

 scopic method depends only on the temperature and density of the stars, it 

 may be shown that the masses of binary stars computed from spectroscopic 

 parallaxes will be the same for the same values of the absolute magnitude. 

 Spectroscopic parallaxes, therefore, though they will give good values for the 

 mean masses of groups of stars, are of little value in the detection of individual 

 differences from the mean. 



A list of about 1,500 dynamical parallaxes prepared by Professor Russell 

 at Princeton has been furnished to the Observatory in advance of publication. 

 These results are of especial value in comparisons of spectroscopic with dyna- 

 mical parallaxes and for the selection of binary systems for spectroscopic 

 observation. 



An extension of the spectroscopic method of obtaining absolute magnitudes 

 has been made to the stars of the A type of spectrum. Investigations of the 

 A-type stars belonging to the Taurus and the Ursa Major streams showed that 

 when these stars had been classified as closely as possible according to spectral 

 type and the sharp or diffuse character of their lines, a definite relationship 

 could be established between absolute magnitude and spectral subdivision. 

 The stars with sharp lines were found to be systematically brighter than those 

 with diffuse lines. A similar relationship proved to hold for all A-type stars 

 with absolute magnitudes determined from trigonometric parallaxes as well 

 as from group motion for which spectrograms were available. In short, it 

 was found that, within the limits of error of the determinations, there was no 

 dispersion in the absolute magnitudes of stars of the same spectral subdivision 

 showing lines of the same character. 



The relationships derived in this way were applied to the determination of 

 the individual parallaxes of 82 stars belonging to recognized stellar groups, 

 and to 104 stars with parallaxes determined trigonometrically. The following 

 comparison shows the agreement between the parallaxes obtained spectro- 

 scopically and those from group-motion or trigonometric measures: 



Stars with parallaxes derived from group-motion: 



Systematic difference, -0''0014. Average deviation, ^O'OO??. 



Stars with trigonometric parallaxes: 



Systematic difference, 0T0000. Average deviation, =t0!0131. 



These values compare favorably with those found in the case of stars of 

 types F to M, where the average deviation between the parallaxes derived 

 trigonometrically and spectroscopically is about ±0''015. 



This method has been applied to the determination of the parallaxes of 

 544 stars with spectral types ranging from B7 to F2 and the results have 

 been published recently. An excellent degree of correlation has been found 

 to exist between the average values of the parallaxes and the proper motions 

 for these stars, a result which furnishes a valuable check upon the validity of 

 the method. 



It seems probable that quite similar relationships will prove to exist between 

 absolute magnitude and spectral type in the case of stars of the B type of 

 spectrum. It is well known that the stars of this type which belong to the 

 earlier subdivisions, such as BO and Bl, are intrinsically brighter than those 

 of later type, such as B8 and B9. If it should be found to be possible to 

 employ for these stars methods similar to those used for stars of type A, the 



