470 MASSES OF VISUAL I'.IXARV STARS. 



period? There are certainly no such pairs in the northern sky 

 where they would most likely be found, and it is doubtful if there 

 are such elsewhere. We are forced to assume that solar-type 

 Stars cannot exist in binary combination when their mean dis- 

 tance apart is much below five times the distance of the Earth 

 from the Sun. In this connection the problem of the Cepheid 

 variable stars may have some bearing ; these are all solar-type 

 stars which vary in brightness; their brightness increases when 

 they are apj)roaching the Rarth. and vice versa. When it is 

 assumed that the changes in their spectra can be dealt with upon 

 dynamical principles —that is, when the changes are sup])Osed to 

 be due to the revolution of one body around the other — it is 

 found that their distance apart is under one unit. The double 

 star of known smallest mean distance amongst the G stars is ^ 

 Hercules, and it is interesting to recall that although it has com- 

 pleted more than three revohitions since its discovery, and it is 

 never a difficult pair to measure, its orbit cannot be found with 

 any accuracy. Mr. T. Lewis, of the (jreenwich Observatory, 

 considers its period to be increasing. 



It is unfortunate that so many of the short-period binary 

 stars of Table I have not had their parallaxes measured, as they 

 are almost certainly amongst the nearest stars. 



Reflecting upon the above conclusions, and u])on the failure 

 to detect parallax in many stars of large proper motion, one is 

 tem[)ted to speculate both tliat gravitative power does not play a 

 leading role in the movement of the stellar universe, and that the 

 star clouds of the Milky Way are as near or nearer to the Sun 

 than many of the proper motion stars are — or. in other words, 

 that the stars of the Milky Way are faint, not because they are 

 very distant, but because they are very small, and that many 

 of the bright stars we see are really beyond the main girdle of 

 the Milky Way. A very rough analogy would compare the sn^-all 

 stars of the Milky Way to the zone of small planets travelling 

 between the orbits of Mars and Jupiter, and the lucid stars to 

 the major planets, the Sun again being near the centre of the 

 system. 



If we could carry out the operation of dividing a star mto 

 two equal masses as suggested earlier in the paper, would the 

 spectrum of the new bodies be similar to that of the old body? 

 This is hardly likely. It seems to be more probable that the 

 spectrum varies with the mass or quantity of matter and its 

 gravitative power or pressure. It would be too dogmatic to 

 assert that if the spectra of two stars were absolutely identical, 

 then the stars themselves would be absolutely identical in every 

 other respect, but it is worth considering as a working hypothesis. 



In the solar system, evidence of light-])ressure is only 

 exhibited by comets, and then only when they approach closely 

 to the Sun. It is doubtful if there is any manifestation of this 

 pressure when comets are at greater distances than five units. 

 The planets, being composed of matter in a cohesive state, show 

 no evidence of light pressure. If they were gaseous bodies of 



