106 



O. V. L. Charlier 



The axis of revolution (which is the major axis) of the ellipsoid is directed 

 against the true vertex. Its position was found to be 



A = 18''51™, . 



n^ = — 19«.35. 



The two axes (a^' and n^') of the ellipsoid give the mean peculiar velocities of 

 the stars parallel to the axis of revolution (o/) and in directions perpendicular to 

 it (Og'). They are expressed in the velocity (s) of the sun as unit. The numerical 

 computation gave 



a,' = (1.5676 + 0.05B0) S , 

 Og' = (0.7921 ± 0.0278) S . 



With a value of ,ç equal to 20 km we get: 



= 31.35 ± 1.10 km, 

 0,' = 15.84 ± 0 56 km. 



The excentricity (e) of the velocity ellipsoid has the value e = + 0.8626. 



The higher characteristics (^^^J of the frequency function (<i>J of the linear 

 velocities have sensible values, which not inconsiderably modify the »ellipsoidal 

 hypothesis*. 



An easy suggestion was to explain these higher characteristics, as well as 

 the ellipsoidal form of the velocity surface, by the hypothesis of two star-streams. 

 For testing it, I have elaborated a method for dissecting a given correlation surface 

 into two spherical components. 



Applying this method to the actual problem it was found that the hypothesis 

 of two star-streams does not account for the deviations from the ellipsoidal velocity 

 distribution. This result was confirmed by discussing the problem according to two 

 other independent methods. 



At the end of the memoir I give some considerations regarding an hypothesis 

 of Turner for explaining the velocity distributions of the stars and regarding some 

 consequencies of the kinetic theory of gases. 



Summing up, the resuU of the investigation of this memoir may be formulated 

 in the following way: 



1) It is necessary to base the discussion of the stellar velocities on the linear 

 velocities of the stars and not, directly on the apparent proper motions; 



2) The distribution of the linear velocities may be found from the observed proper 

 motions, as soon as the value of a certain parameter [q) is know; 



3) Though the value of thts parameter is, as yet, rather uncertain, it is possible 

 to assign to it certain limits, viz. 1 <C 2 <C 1-27 ^). It is assumed that q does not differ 

 considerably from, this maximum value (1.27); 



^) The value of the upper limit is, however, still open to some uncertainty. 



