500 Scientific Proceedings, Royal JDublin Societp. 



bright double or multiple star, the brighter stars seem to keep 

 aloof from their inferiors, and when they form systems they keep 

 these systems to themselves. 



I may further remark that the principle that faintness is a 

 mark of great distance has been acted on by astronomers in their 

 researches after parallax, and in several instances it has been 

 verified by the results thus obtained. This was the mode in which 

 Bessel first determined the parallax of 61 Oygni. Suspecting from 

 its large proper motion that this star was near us, he selected two 

 fainter stars in the neighbourhood, which he assumed to be so re- 

 mote that for practical purposes their distances might be regarded 

 as infinite. The same operation was repeated by Struve on another 

 faint star in the neighbourhood, and the near coincidence of their 

 results seems to prove that the stars thus selected are really much 

 more remote than 61 Cygni. Struve's star was fainter than either 

 of Bessel's, and the slight discrepancy in their results would be 

 explained by supposing that it was also more distant. The same 

 method has subsequently been adopted by other observers. Had 

 the stars which Bessel and Struve selected for reference been much 

 brighter than 61 Cygni, their results might not have met with such 

 universal acceptance. Possibly, too, these results would have been 

 different. 



On these grounds, I am inclined to think that there is a suffi- 

 cient average correspondence between the magnitude of a star and 

 its distance to justify the conclusions which I draw from that 

 assumption as approximations to the truth. Of course it is only 

 when we are dealing with wide averages that such results are of 

 any value, and therefore their value increases with each successive 

 magnitude, provided that we have sufficient data to work upon. 

 When we take in a wide region of space, distributed impartially in 

 every longitude and latitude, it seems probable that the peculiari- 

 ties of different portions of this region will compensate each other, 

 and that the general average, computed on the hypothesis of uni- 

 formity, will not be far from the truth. I already noticed that as 

 the twenty brightest stars are certainly not the twenty nearest 

 stars, the average distance of a first magnitude star is almost 

 certainly greater than it would be on the hypothesis of uniformity. 

 But this is only partly true of the stars of succeeding magnitudes. 

 All subsequent magnitudes will indeed contain stars which owe 



