222 



CAKNEGIE INSTITUTION OF WASHINGTON. 



The most important application of the radial-velocity results has 

 been to an investigation of the relationship between absolute magnitude 

 and stellar motions carried out by Mr. Adams and Mr. Stromberg. 

 For this purpose use was made of the absolute-magnitude determina- 

 tions obtained spectroscopically, supplemented by luminosities derived 

 with the aid of a modified formula connecting mean parallax and proper 

 motion. A total of about 1,300 stars was used in the investigation. 

 These were divided into groups situated in concentric zones about the 

 sun within certain definite hmits of parallax. A comparison of the 

 radial motions of the stars within these groups indicates clearly an 

 increase of velocity with decrease in absolute magnitude, independent 

 of distance from the sun, which amounts to about 1.5 km. per unit of 

 magnitude. A combination of the values for all the zones is shown in 

 the accompanying table. M and V are the absolute magnitude and 

 the radial velocity corrected for the solar motion. 



This conclusion was further tested by the use of the velocities at 

 right angles to the line of sight and very similar results were found. 

 The variation with absolute magnitude is independent of the assump- 

 tion of a frequency distribution of velocities in accordance with Max- 

 well's law, provided only that the law of distribution along the three 

 components in space is the same. The influence of stream motion 

 upon the results was also investigated and found to be relatively shght 

 in the case of the stars employed. The fact that the fainter stars move 

 more rapidly than the brighter seems, therefore, to be established 

 definitely by this investigation. 



Reference may be made to two additional products of this study of 

 stellar magnitudes and motions. The first is the evidence that the 

 average space velocity of the K and M type stars is about 1.0 to 1.5 km. 

 greater than that of the F and G type stars of the same absolute magni- 

 tude. The second is the test of the accuracy of the spectroscopic deter- 

 minations of parallax provided by the computation of the values 

 derived from the stellar cross-motions; the comparison indicates close 

 agreement for larger parallaxes and somewhat higher values than those 

 computed in the case of the very small spectroscopic results. 



