43 8 SCIENCE PROGRESS 



was also derived in the course of a determination of the syste- 

 matic motions of the stars from their radial velocities by 

 A. S. Eddington and W. E. Hartley (M.N., R.A.S. vol. lxxv. 

 p. 521). In this paper also, the position of the vertex of the 

 preferential motion of the stars was found to be R.A. 94°- 6, 

 Dec. + i2 0, 5, in very good agreement with the position 

 obtained from the study of proper-motions. The relation 

 between the radial velocity and apparent brightness of stars 

 has been discussed by C. D. Perrine (Astroph. Jonrn. vol. xli. 

 PP- 3 l S> 395)- He finds that, in general, there is a small 

 but regular increase in speed with decreasing apparent bright- 

 ness, which holds both for galactic and for non-galactic stars. 

 This result, taken in connection with the preceding, indicates 

 that part at least of the progression found in the above in- 

 vestigations must be due to one or both of the first two men- 

 tioned causes, and a little consideration shows that the only 

 possible explanation is that the average speed of the stars 

 increases with decreasing absolute luminosity. The same 

 conclusion is found by W. S. Adams (Astroph. Jonrn. vol. xlii. 

 p. 172) from a discussion of the radial velocities of 500 stars 

 determined at Mt. Wilson Observatory ; he concludes also 

 that the magnitude of the effect is of greater range than the 

 progression of speed with spectral type. The bearing of these 

 results on the general question of stellar evolution has been 

 discussed by A. S. Eddington (Observatory, vol. xxxviii. p. 392) 

 who suggests that there is a correlation between the speed 

 and mass of a star, from which both the progression of speed 

 with type and also with absolute luminosity arises. 



Stellar Evolution. — H. Shapley (National Acad, of Sciences 

 (Washington) Proceedings, vol. i. p. 459, and Astroph. Jonrn. 

 vol. xlii. p. 271) discusses the densities and dimensions of five 

 eclipsing binary systems of late spectral type, and finds for the 

 upper limit of the mean densities in the five cases the respective 

 values o-02, 0-0005, 0-0005, 0*00012 and 0-00005, the density 

 of the Sun being the unit, and that the dimensions of the 

 systems are very large compared with those of the Sun. This 

 proof of the existence of stars of types G and K of very small 

 density is of great importance from its bearing upon the 

 question of the order of stellar evolution. The formerly 

 accepted view was that, starting from nebulas, stars gradually 

 passed in succession through the spectral types B, A, F, G, K, M, 



