Stellar velocity distribution 



21 



to differ very little, at least concerning the direction of the solar motion, in spite 

 of the uncertainty that will occur on account of the small angular distance of the 

 apex from the plane of the Milky Way. 



Concerning the relative velocity of the sun, it seems to he larger for stars in 

 the galactic plane than for stars surrounding the pole. This fact may, indeed, largely 

 he due to the relatively large number of class B stars in the plane of the Milky 

 Way. These stars show, as seen in table IV a, the largest value of S. 



16. A general wiew of the results for stars of different spectral classes seems 

 to indicate exceptionally high velocities for classes B and M. The mean error is 

 however large, as will be shown in a following paragraph. 



For the different spectral classes Campbell has computed the sun's relative velo- 

 city 1 . It is, however, to be observed that the solutions are different so far as Campbell 

 determines the solar velocities, assuming the position of the apices to be known for 

 each type. For class B stars he accepts the apex direction a — 270°.5 § = -f 34° .3 

 found from Boss' general solution from all spectral types. For the other types he 

 assumes the apex direction: a = 270°.o S = -f- 30°.o. 



From the equation 



cos dV 0 — V = 0, 



where V 0 is the velocity of the solar system with reference to the stars, d the 

 angular distances of the stars from the assumed apex and further V the observed 

 velocities of the stars or the groups into which the stars were divided, the unknown 

 V 0 was solved with aid of the method of least squares. 



The results of Campbell do not differ very much from the present. On the 

 whole the stars used are also the same. 



The constant K. 



17. Concerning the constant K it enters 

 the general equations for the apex elements, 

 and is solved with the method of least squares. 

 Campbell gets his values from a simple mean 

 of the residual velocities, or the velocities cor- 

 rected for the influence of the solar motion. 

 To get a comparison with his values I have in 

 table V reprinted the two series, adding the 

 mean error to the present results. 



The agreement seems to be good when 

 taking account of the mean errors. 



In fig. 1 I have made a graphical re- 

 presentation of this constant. As remarked by 



' Lick. Obs. Bull. VI. 196. p, 127. 



K 



Fig. 1. 



The constant K for each spectral type. 



