ON THE STRUCTURE OF THE UNIVERSE 7 



velocities under the general attraction of the whole stellar system 

 — the law of attraction being doubtless somewhere between an 

 attraction varying directly as the distance (the law of attraction 

 due to a uniform distribution of matter) and an attraction varying 

 inversely as the square of the distance (the law of attraction 

 when the attracting mass is concentrated at one point — as in the 

 motion of the earth around the sun). The stars of early types 

 O and B, owing to the comparatively small time which has 

 elapsed since their birth, have small velocities and are strongly 

 condensed in the galactic plane. The stars of later types having 

 been born earlier have had time to acquire larger velocities and 

 to stray further from the galactic plane, so that we find the type 

 M stars very uniformly distributed. If this be the true explana- 

 tion, it is not free from difficulties. If the stars are, as we have 

 supposed, born in the galactic plane, the attraction under which 

 they move will be in the same plane, so that it is not easy to see 

 how they can move out of it. To avoid this difficulty, Prof. 

 Eddington has conjectured that the stars of types K and M may 

 have been formed originally in a more spherical and uniform 

 distribution than the stars of early type, and that possibly, for 

 some reason which we do not know, the birth of stars was 

 retarded in the galactic plane. It is a well-known fact that the 

 stars of early type do on the whole move parallel to the galactic 

 plane, as we should expect, and this fact has been used by Prof. 

 Kapteyn to predict their parallaxes with a considerable degree 

 of accuracy. Another alternative explanation which Prof. 

 Eddington has suggested is that in the galactic regions, richly 

 supplied with star-forming material, the tendency has been to 

 form large stars which have developed very slowly, whilst away 

 from the galaxy small stars have been formed, which have run 

 through their course of development quickly. This would account 

 for the predominance ol early-type stars in the galactic plane ; 

 but it raises other difficulties, and does not appear to explain the 

 increase of average velocity with the progression in type. One 

 feels that one is treading on uncertain ground, and that these 

 hypotheses are merely speculative, with little solid foundation. 

 The facts, nevertheless, must be of fundamental importance in 

 any theory of the structure of the stellar universe. 



A few further facts must be considered before we can form 

 a general idea of the structure of our sidereal system. So far 

 nothing has been said as to the average distances of stars of 



