Mabch 23, 1906.] 



SCIENCE. 



445 



and (&) the average distances of the stars 

 of each individual magnitude. These dis- 

 tances are obtained from a combination of 

 the velocity with "which the solar system 

 moves through space and the resulting ap- 

 parent displacement of the stars in the sky, 

 the solar motion being determined spectro- 

 seopically and the displacement of the 

 stars by means of the meridian circle. 

 Kapteyn has obtained in this way the mean 

 distances of stars from the third to the 

 ninth magnitude; the present paper sup- 

 plements this by a new determination of 

 distances for stars between the ninth and 

 eleventh magnitudes. 



Based upon a part only of the above 

 material, current astronomical doctrine 

 affirms that the visible universe is a stellar 

 system of limited extent whose boundaries, 

 at least in some directions, lie within reach 

 of our present telescopes. This view is 

 opposed in the present paper, where it is 

 pointed oiit that the evidence in favor of 

 limited extent is obtained by ignoring two 

 important factors of the problem; when 

 these factors are taken into account the 

 supposed evidence vanishes. 



The first, and more important, of these 

 factors relates to the intrinsic brightness 

 of the stars. It has been commonly as- 

 sumed that the fainter stars appear faint 

 because of their greater distance from the 

 earth and that there is no reason to sup- 

 pose them to be intrinsically less luminous 

 than the bright ones. The first part of 

 this statement is unquestionably correct as 

 far as it goes; the second part is wholly 

 wrong. The faint stars, according to the 

 writer, are less luminous than the bright 

 ones, and stars of any given magnitude 

 emit, on the average, only ninety per cent, 

 as much light as do stars a magnitude 

 brighter. 



The second factor omitted in previous 

 discussions is not here ignored, but re- 



.jected as useless. The known existence of 

 dark matter or cosmic dust scattered 

 through space leads naturally to the sup- 

 position that it may in some manner affect 

 the transparency of interstellar spaces and 

 cause an appreciable diminution in the 

 light of fainter stars. Although this view 

 has been rejected from current theories, it 

 is evident from mathematical discussion of 

 the data here considered that there is such 

 an absorption and that appx'oximately five 

 per cent, of light is lost in transmission 

 over a distance equal to a million times the 

 diameter of the earth's orbit. 



In abandoning the concept of a stellar 

 system of limited and measurable extent 

 the writer considers, provisionally, the eon- 

 sequences that result from the hypothesis 

 of a system indefinitely extended on every 

 side, with substantial uniformity in the 

 direction of the Milky Way but thinning 

 out on either side of the Milky Way, be- 

 cause the stars are here less numerous, or 

 less brilliant, or because a denser cosmic 

 dust more effectually quenches their light. 

 It is impossible, now, to decide between 

 these alternatives, but something of the 

 kind is operative, and any one of the alter- 

 natives or a combination of all suffices to 

 explain every feature of the stellar system 

 for which the current theory accounts. 

 The new hypothesis also brings out features 

 hitherto unrecognized or unexplained. 



The stars are not all of one kind, but 

 differ among themselves in physical condi- 

 tion and properties which the spectroscope 

 is able to detect and analyze. By far the 

 larger part of the stars fall into one of 

 two classes which the spectroscopists desig- 

 nate as type I. and type II., and which 

 they regard as different stages of develop- 

 ment of the individual, a star of type I. 

 passing over with increasing age into type 

 II. Now Pickering has shown recently 

 that these stellar types are not scattered in- 



