600 



SCIENCE. 



[N. S. Vol. XVIII. No. 402. 



in the cluster lias the smallest range of 

 brightness. For this variable, at first, a proof 

 of 0.275 days had been derived, which, in- 

 deed, satisfied a large part of the observations. 

 The new investigation gave a distinctly dif- 

 ferent value, 0.379S7, and, moreover, made it 

 probable that the period is not constant, but, 

 within the time which the observations cover, 

 varies between 0.3796 and 0.3801 day. At 

 least with this assumption a noticeably better 

 representation is obtained. Since in the case 

 of this star the changes of brightness are 

 not much greater than the unavoidable errors 

 of measurement, it is not very surprising 

 that the new determination, made with other 

 comparison stars, has led to an entirely dif- 

 ferent result. The limit is here reached 

 where the precision of the method employed 

 in the determination of the brightness fails, 

 and it appears more than doubtful whether 

 the new result deserves more confidence than 

 the old. It would, indeed, be wiser, where 

 the light fluctuations are so small, not to at- 

 tempt the determination of the period. 



Under the heading, ' Miscellaneous Re- 

 sults,' in the last chapter is discussed a num- 

 ber of questions which are concerned with 

 the precision of the measurements, the distri- 

 bution of the variables in the cluster, the 

 light curves, etc. Some of these investiga- 

 tions deserve to be taken up briefly. 



The distance of the stars from the center 

 of the cluster has a marked eiiect on the pre- 

 cision of the comparisons. The nearer the 

 variables are to the center, and the more, on 

 this account, the images run together on the 

 photograph, so much more difficult also are 

 the comparisons of brightness. "Within a dis- 

 tance of 5' from the center- of the cluster are 

 found 46 variables ; for 16 of these, on account 

 of the difficulty of the observations, no period 

 at all could be found, and for the remaining 

 30 the mean deviation from the correspond- 

 ing curves is for the whole ±0.159 magnitude. 

 On the other hand, for 35 variables whose dis- 

 tance from the center lies between 5' and 10', 

 the corresponding mean deviation is only 

 ±0.114 magnitude, and for 30 stars, whose 

 distance is greater than 10', it is only ± 0.113 

 magnitude. 



When the deviations of the individual 

 observations from the mean light curve at- 

 tained a value of 0.40 magnitude and greater, 

 the observations were repeated. In certain 

 cases it happened that errors of measurement 

 existed, which were due to errors of identifica- 

 tion, probably. Very many large residuals are 

 found, however, in all 287 cases, which are 

 verified by the measurements of the revision. 

 These residuals have their cause, perhaps, in 

 real irregularities of the light curve, perhaps 

 also in defects on the photographic plates. 

 The general quality of the plates has naturally 

 an influence on the precision of the results, 

 and in general the sharpest and best plates 

 give the smallest residuals; while with very 

 experienced observers and with the extremest 

 care the gain is not so important as one might 

 expect beforehand. 



Since it is known that in case of com- 

 parisons of brightness of two stars in the sky 

 their relative position plays an important role, 

 it was of importance to investigate whether 

 in the determinations on photographs the 

 relative position of the two star images might 

 exercise a marked efEect. To find the answer 

 to this question one of the variables was com- 

 pared on several plates with a number of com- 

 parison stars, while, by turning the plate 

 under the microscope, the comparison star 

 was placed first on the right and then on the 

 left of the variable, and each time the differ- 

 ence in brightness was estimated in grades. 

 It was thus shown that the influence of the 

 change in position, if it exists at all, is so 

 small that it can be completely ignored. 



Of special interest is the investigation con- 

 cerning the distribution of the variables in 

 the cluster. The author has for this purpose 

 divided the variables into twelve groups, of 

 which the first includes all to a distance of 2' 

 from the center, the second those whose dis- 

 tance lies between 2' and 4', etc. For each 

 group the number of variables per square 

 minute was computed. The following table 

 contains the results for the different groups. 

 Next to them are given the corresponding 

 numbers for the distribution of all the stars 

 of the cluster, which were taken from a previ- 

 ous publication in the Harvard Annals. The 



