246 CARNEGIE INSTITUTION OF WASHINGTON. 



when stars with total velocities exceeding 300 km. are omitted from 

 the computation. The motion of the centroid is almost exactly 

 parallel to the plane of the Galaxy. 



(6) The direction of the major axis of the velocity-ellipsoid is in 

 good agreement with that found from investigations by Stromberg 

 and by Raymond on the motions of dwarf stars. 



(7) If the stars are divided into groups having average magnitudes 

 greater and less than 3.0, it becomes clear that the fainter stars are 

 moving much more rapidly than the brighter stars, and that the apices 

 of their motions lie much more nearly in the plane of the Galaxy. 



(8) The stars with the largest velocities in space are of type F, 

 and among these the earlier types have the larger values. Thus, the 

 stars of types FO to F5 have an average space-motion of 365 km. ; those 

 of type F in general, 307 km. ; those of type G, 156 km. Of the entire 

 list of 37 stars, 26 are of types F and G. 



Spectroscopic Determinations of Luminosity and Parallax. 



The extensive investigation referred to in the last annual report, 

 having as an object the computation of the systematic corrections to 

 be applied to the absolute magnitudes derived spectroscopically from 

 the original curves of reduction, has been nearly completed. The 

 method employed has been to compute the mean absolute magnitudes 

 of groups of stars with but a small range in spectral type and line- 

 intensity, using for this purpose the parallaxes measured trigono- 

 metrically, the parallactic motions, and the peculiar motions. The 

 three methods give results in good agreement with one another, al- 

 though, of course, the degree of accuracy is quite different for stars of 

 different distances and velocities. The mean absolute magnitudes 

 derived in this way are then compared with the absolute magnitudes 

 derived spectroscopically. Three general conclusions may be drawn 

 from this comparison: 



(1) The spectroscopic criteria of absolute magnitude are valid for 

 the stars of highest luminosity, a regular change in absolute magnitude 

 accompanying a change in the intensity of the selected lines. 



(2) The previous reduction-scale employed for the determination of 

 absolute magnitudes obtained by the spectroscopic method gives 

 values which are nearly correct in the case of the fainter stars but 

 systematically too faint in the case of the brighter stars. The cor- 

 rection required is of the order of —0.5 to — 1.0 magnitude. 



(3) The probable error of the absolute magnitude of a single star as 

 derived by the spectroscopic method varies with the spectral type from 

 about ±0.1 for the early F-type stars to ± 0.6 for the later K-type. 



As a result largely of this investigation, it now becomes possible to 

 establish what may be considered a definitive scale for the spectro- 

 scopic determination of absolute magnitudes. This has not been 



