44 



Walter Gyllenberg 



The difference seems to indicate the reality of a peculiar velocity distribution 

 of the class B stars. However, if there exists a difference in the value of K in 

 different parts of the sky this circumstance will largely influence the results. 



41. It seems very remarkable that the stars of type B have like the stars of type, 

 ill", a circular distribution in the plane of the Milky Way. This is unexpected and 

 we should indeed expect to find the largest divergence between these two groups. 

 Regarding however the proportion of giant and dwarf stars in these spectral classes, 

 a certain similarity is to be found, which perpaps will help to explain the fact. The 

 class 7?stars, having small dispersion of the absolute magnitude, contain exclusevily 

 stars of the giant type. In the following classes the relative number of entering 

 dwarf stars increases and has its maximum in one of the intervening classes. As 

 the material may approximative^ be considered to contain stars brighter than a cer- 

 tain magnitude the giant stars belonging to the class M will be relatively favoured, 

 while only the nearest of the dwarf stars enter. The catalogue contains further 

 only seven stars of type M, whose parallaxes are determined. Also, the determination 

 of the mean parallax of the class M stars gives a small value pointing to the presence 

 of large number giant stars. 



Returning to the examination of the constant K, this was shown to be almost 

 equal in sign and magnitude for stars of classes B and M. This was found to be 

 also the case both for the sun's relative velocity as for the mean distances. And finally 

 looking back to the figures 2 — 8 we fouud a new similarity concerning the variation 

 of the constant K in terms of angular distances from the principal vertex. The 

 two spectral types thus give a distinct positive displacement of the mean in the 

 direction of the principal vertices, without giving any tendency to a larger mean 

 velocity in the same directions. 



Unfortunately there are too few stars to confirm any explanations. 



The mean errors in the moments of the second order. 



42. The quantities in table XIII are computed by a linear transformation 

 of the observed values of a. Let the mean error in one observation of the left 

 membrum of (38) be 



0" = £(0 2 ), 



then the mean errors in the unknown quantities are expressed through the formula 



Observing that 



s (o 2 ) = 2 as (a) 



and 



