190 



KNOWLEDGE 



[October 1, 1892. 



compare the actual amount of liglit sent oat by each, we 

 have to multiply the numbers representing the areas by 

 the numbers representing the intrinsic brightness. Making 

 the calculation, we see that the light of the sun is to the 

 light of Saturn as 2082 x 2082 x 105 x 105 is to unity. 

 Now, the square root of the product just given is 218,210 ; 

 which means that we should have to remove the sun to 

 about 218,210 times its present distance before it would 

 shine as a first magnitude star. 



This, of course, is only an approximation, and only 

 useful (but very useful, nevertheless) in framing first and 

 elementary ideas as to the extent of the visible universe. 

 Michell was cai-eful to point out that although his method 

 might be trusted to give trustworthy ijencral data to start 

 with, it would not be applicable to special cases ; for, as he 

 warns us with one of bis touches of inspiration, most likely 

 the stars are not by any means of an equal size, and the 

 surface brightness may differ very greatly in different stars. 



The paper then goes on to discuss methods for deter- 

 mining the actual sizes of the stars, and the difficulty — 

 nay, the impossibility — of ever solving this great problem : 

 '■ There seems to be little chance," says Michell, " of dis- 

 covering with certainty the real size of any of the fixed 

 stars, and we must consequently be content to deduce it 

 from their parallax, if that should ever be found, and the 

 quantity of light which they afford us compared with that 

 of the sun. And in the meantime, till this parallax can be 

 found, or something else may arise to furnish us with a 

 more general analogy, we can only suppose them, ' at a 

 medium,' to be equal in size to the sun, this being the 

 best means which we have at present of forming some 

 probable conjecture concerning the extent of the visible 

 universe. That we may be the better enabled to do this, 

 it seems to be an object worth the attention of astronomers 

 to inquire into the exact quantity of light which each star 

 affords us separately when compared with the sun ; that 

 instead of distributing them, as has hitherto been done, 

 into a few ill-defined classes they may be ranked with pre- 

 cision, both according to their respective brightness and 

 the exact degree of it." 



Let us pause for a moment to see what modern astronomy 

 has to say to Micliell's results. Seventy years had to pass 

 before Mr. Henderson and Sir Thomas Maclean could prove 

 to the world from observations on the parallax of Alpha 

 Centauri that Michell's theoretical estimate was not far 

 from the truth. Moreover, the qualifications he had laid 

 down as to the various sizes of the stars were also shown 

 to be justified by the results of a comparison between the 

 discoveries of Henderson and Maclean, and those of Bessel 

 on the star 61 Cygni ; for this latter star, although only 

 about three times as far away, gives us scarcely a two- 

 hundredth part of the light of Alpha Centaiu-i, whereas if 

 all the stars were equal in size and brightness it would 

 send us nearly thirty times as much light as it does. If it 

 should be urged that Michells speculations were nothing 

 but lucky guesses, it may be retorted that such luck comes 

 only to genius. For the rest, it may be worth noting 

 that Michell's dream of a star- catalogue, in which all the 

 stars should be "ranked with precision," has only begun 

 to be realized within the last twenty years or so. 



The best part of Michell's extraordinary paper is his 

 application of the theory of probabilities to the distribution 

 of the stars. It has always been the custom to divide the 

 stars mto gi-oups or constellations, for religious or historical 

 purposes, or for convenience of reference. Generally 

 speaking, these groups or constellations are well-defined : 

 Orion, Ursa Major, and the Tleiades, for example. From 

 the time of Copernicus this grouping had been regarded 

 as accidental, and the constellations were supposed i 



to result from the projection of a number of bright 

 stars, differing greatly in distance, upon small portions 

 of the imaginary sphere of the heavens. Michell then 

 took up the parable, and applied the doctrines of 

 probability to the question for the first time in the 

 history of astronomy. Dealing with the brighter stars 

 in the Pleiades, as being the most convenient for 

 his purpose, he was able to show that the odds were 

 500,000 to one against any six physically unconnected stars 

 scattered at random through space being found projected 

 upon the sky so closely as these are.'' Hence it is 

 nearly certain that the Pleiades (and other similar groups, 

 of course, by the same reasoning) are in reality physically, 

 as well as optically, associated. Unfortunately, this grand 

 result, so valuable in itself, and so full of promise to 

 further research along the same lines, was practically 

 neglected, and astronomers took over a century to redis- 

 cover that which ^lichell with good reason regarded as 

 demonstrated. The observations of Bessel, Wolf, and 

 Elkin shows that the greater number of stars in the 

 Pleiades group have a common small proper motion, which 

 seems to indicate that the whole group is a connected 

 system ; and the more recent application of photography 

 has confirmed this theory by revealing the remarkable 

 fact that a gigantic nebula envelopes the whole or the 

 greater part of the whole group. An additional and strong 

 argument was also advanced on the same side by the late 

 Mr. Proctor, founded upon the small number of stars in 

 the sky-space adjacent to the Pleiades, suggesting, so great 

 is the contrast, a sea of darkness enclosing an island of 

 light. Surely, if the greatness of Michell needed demon- 

 stration these facts ought to be enough. It may be added 

 that the laws of probability aj)ply in exactly the same 

 way to cases where two, or ten, or any number of stars 

 are seen close together on the sky, though the odds in 

 favour of a physical connection in such instances are much 

 reduced as the number of stars so situated is less. 



The remainder of the paper is taken up with a discussion 

 of the relations presented by particular circumstances of 

 star-grouping, and the possible position of the sim amidst 

 the stars. The results in the main are nothing less than 

 what astronomers during the last twenty or thirty years 

 have begun to see clearly. But singular injustice has 

 been done to Michell in this as in all things else. 

 Kecent writers have either appropriated his work with- 

 out acknowledgment, or have too generously handed it 

 over to others. So the intimation that stars optically 

 close are almost certain to be physically connected has been 

 placed to the credit of Sir WiUiam Herschel, who simply 

 determined the orbital niations of some of these close stars, 

 and describes himself as surprised at the discovery of their 

 connection. In speaking of the discovery, he says that 

 he went out like Saul to seek his father's asses, and found 

 a kingdom — the dominion of gra\itation extending to the 

 stars. It had been known for a very long time that many 

 of the stars which appear single to the naked eye are 

 double or triple when seen in the telescope ; and Michell's 

 paper was a satisfactory proof of the significance of the 

 phenomena. The great telescopic powers which Herschel 

 used enabled him to split up a great many more stars 

 which were not previously known to be double, but so far 

 was he from grasping the physical explanation of his facts 

 that, in his first catalogue (published in 1782), he sees no 

 other reason for a change of position in the components of 

 a double star than " by admitting a proper motion in either 



* Sii" John Herschel points out that when the stars of the 

 southern lien\isphere are included in the investigation these figures 

 have to be considerably reduced. The correctness of the argument 

 is unaltered, however. 



