130 



KNOWLEDGE. 



April, 1913. 



adjacent stars are of nearly equal magnitudes, and it 

 appears to be the universal rule that when there is 

 a contrast of colour the tint of the fainter star lies 

 more towards the violet (more refrangible) end of 

 the spectrum than that of the other. Thus, the 

 brighter star being reddish or yellowish will have a 

 green or blue companion. Sir John Herschel, who 

 worthily continued his father's work, and also con- 

 tributed what he modestly calls his " mite" towards 

 double-star astronomy, has suggested that this com- 

 plementary colouration is " probably in virtue of that 

 general law of optics that when the retina is under 

 the influence of excitement by any bright light, 

 feebler lights, which when seen alone would produce 

 no sensation but of whiteness, shall for the time 

 appear coloured with the tint complementary to that 

 of the brighter." There are, however, difficulties in 

 the way of accepting this explanation in all cases, 

 and others are inclined to regard this contrast of 

 colours as being, in some cases at least, due to a 

 real difference in the physical nature of the stars. 

 However this may be, the beauty of the sights 

 visible to the telescopist is greatly enhanced by the 

 wonderful display of colours given by different 

 celestial objects. It is often found that if in a 

 double-star system the coloured star be much less 

 bright than the other, it will not affect the latter's 

 colour. Thus, for example, >/ Cassiopeiae is com- 

 posed of a large white star and a fainter one of a 

 " rich ruddy purple." A pleasing picture of the 

 curious alternations of illumination that would be 

 produced for the inhabitants of a planet circulating 

 round a pair of coloured double stars may be drawn, 

 and is given by Sir John Herschel (" Outlines of 

 Astronomy," p. 851). Suppose a planet revolving 

 round a red and and green sun. When the red sun 

 rises there will be daylight, and " all will be red." 

 Bye and bye the green sun will rise and mount 

 higher above the horizon. The light will gradually 

 change from a reddish tint to pure white. Later on, 

 the red sun will set, and the remaining illumination 

 will consequently be green. Last of all, the green 

 sun will set, and darkness will set in. Thus we 

 have the alternations red-day, white-day, green-day, 

 and night, the colours of all objects undergoing 

 corresponding variations. 



It is a remarkable fact that though isolated red 

 stars are found in most parts of the sky, no decided 

 green or blue star has ever been noticed unassociated 

 with a brighter companion (Herschel). 



We have already stated that the orbits of binary 

 stars round their centre of mass are ellipses, usually 

 much more oval than the planetary orbits in our 

 solar system, the average eccentricity of the visual 

 binaries being about 0-5 (See " Researches," Vol. II, 

 ch. 20), but it is a remarkable and interesting fact 

 that the closer " spectroscopic " binaries move in 

 much more nearly circular orbits. The application 

 of the spectroscope to stellar astronomy has not only 

 given us information otherwise unobtainable as to 



the chemical nature of the heavenly bodies, but it 

 has also enabled us to detect hitherto unsuspected 

 motions, and has rendered known the existence of 

 bodies perhaps for ever invisible to our telescopes. 

 By means of this instrument it has been ascertained 

 that the number of binary stars is far greater than 

 anyone had previously imagined, but that in most 

 cases the components are too near together to be 

 separated by the most powerful telescope. Accord- 

 ing to estimates based on the work done at 

 the Lick Observatory, Campbell found that about 

 one star in five of those examined proved to be a 

 spectroscopic binary, and in certain groups this ratio 

 was found by Frost to be as high as one third (See). 

 The telescope discloses only the widely-separated 

 and luminous companions amongst the systems 

 nearest to us in space, but the spectroscope enables 

 us to detect all attendant masses which are large 

 enough perceptibly to disturb their luminous 

 " fellows," whatever be their distances, thus enor- 

 mously increasing our knowledge of stellar systems. 

 When a star is approaching us the dark and bright 

 lines in its spectrum are shifted slightly towards the 

 violet; when it is receding they are shifted in the 

 opposite direction, and by the comparison of well- 

 known lines thus changed in position with their 

 ordinary position as seen in terrestrial spectra it is 

 possible to determine the speed of their motion. 

 Thus, the well-known variable star, Algol, exhibits 

 changes in its spectrum indicating that the velocity 

 in the line of sight undergoes variations, being alter- 

 nately towards and away from the Earth, and thus 

 is confirmed the view that its variability is due to 

 partial eclipse by a revolving dark satellite, the 

 " stupendous dark globe." It has been shown, too, 

 that a Virginis, like Algol, has a massive dark com- 

 panion which, however, does not eclipse, as it does 

 not come between the star and our position. Other 

 double stars have been discovered in which both 

 components are bright, so that at one part of their 

 orbit the lines common to the spectra of the two stars 

 appear double and separated, gradually closing up 

 till the\- appear single and then opening out once 

 more. Since visual binaries with known orbits are 

 found to give variations of a similar character in 

 their spectra and have thus come to be included in 

 the class of spectroscopic binaries as well as "visual" 

 ones, but so far no spectroscopic binaries first dis- 

 covered as such have been resolved telescopically, 

 we see that the difference merely consists in the 

 smaller size of their orbits and consequent shorter 

 periodic times of revolution of the latter. Periods 

 of a few days, or even hours, are known for these, 

 whilst the shortest periods for a visual binary yet 

 known is that of f5 Equulei (5 • 7 years) and orbits 

 with periodic times of hundreds of years have been 

 calculated for some of the more widely-separated 

 binaries (y Virginis, 182 years; rr Coronae, 340 years, 

 Lewis). When the angular dimensions of the orbit 

 and the parallax of the system are known, the real 

 dimensions (in miles, kilometres, and so on) are 

 easily calculable, and from a knowledge of the 



