134 VARIABLE STARS IN- SOUTHERN HEMISPHERE, 



with several stars of the ninth or tenth magnitude. That all three of 

 the stars formerly existed there can be no doubt, for B.A.C. 3679 

 and 3683 were observed by Taylor and Brisbane, and B.A.C. 

 3680 by those two observers and also by Lacaille. 



III. I will now direct your attention to the desirability of a 

 complete naked eye examination of all the stars in the Southern 

 Hemisphere to the fifth magnitude. As a general rule the best 

 months for comparison observations in this latitude are May, 

 June, and July ; the atmosphere at that period of the year being 

 peculiarly steady and transparent ; observations should, however, 

 be made at all available opportunities, as variable stars of short 

 period might thus be earlier detected. The stars at the time of 

 comparison should be at a considerable altitude. In my short 

 experience I have found that a star, as it approached the horizon, 

 sometimes became fainter and at other times brighter. The 

 former effect is probably produced by thin clouds near the hori- 

 zon, the latter by diffused particles of matter in the atmosphere 

 causing the well-known phenomenon of scintillation, which is 

 very apt to produce an exaggerated impression of its lustre on 

 the judgment of the observer. Care should be exercised that 

 the portions of the heavens occupied by the compared stars be 

 equally clear, as inequality in that respect might seriously affect 

 the results. Comparisons should also be repeated at different 

 hours on the same night, in order to the elimination of any errors 

 which might arise from the source just mentioned. Attention 

 should also be paid to the position of the moon on the nights of 

 observation. It is well known that an observer's estimate of the 

 brilliancy of a star is considerably influenced by the degree of 

 illumination of the sky on which it is projected ; consequently he 

 should not compare stars in the neighbourhood of the moon with 

 those more remote from her. An inexperienced observer would 

 hardly credit the amount of error incurred by a neglect of these 

 precautions. The observer must, in fact, in this, as in every 

 other department of Astronomy, pay the most scrupulous atten- 

 tion to all probable sources of error. In instituting a set of 

 naked-eye comparisons, a number of standard stars must be 

 selected for the epoch, in a descending scale of magnitude, from 

 the star of the greatest brilliancy to one that can only be 



