596 



SCIENCE. 



[N. S. Vol. XVIII. No. 462. 



paratively short part of the whole period 

 (sometimes only one fifth). For, since in the 

 remaining time no noticeable changes occur, 

 the variability is made apparent, in many 

 cases, only by the examination of very many 

 plates. 



In order to determine the brightness of the 

 variables in the cluster w Centauri, in the 

 first place, a system of 38 comparison stars 

 was selected, whose brightness ranged between 

 magnitude 9.0 and 14.6, and which lie either 

 within the cluster itself or at least in its im- 

 mediate neighborhood, in any case not farther 

 from the center than the most distant vari- 

 ables. The mean difference in brightness be- 

 tween two successive sequence stars is about 

 0.15; for the brighter stars the interval is in 

 general larger, while the fainter stars lie closer 

 together. The method which was employed in 

 the determination of the magnitudes of the 

 comparison stars is that which Pickering has 

 described in Volume 26, Part II. A com- 

 parison scale was first made in which by the 

 use of a selected star a line of images was 

 obtained on the same plate by clockwork, and 

 also with exposures of different lengths. The 

 star was first exposed for 810 seconds, then 

 the telescope was moved by a slight amount in 

 right ascension and the star again exposed, 

 this time for 270 seconds. Four other images 

 with exposures of 90, 30, 10, and 3 seconds 

 serve to complete the scale. On the supposi- 

 tion that the photographic effect is propor- 

 tional to the time of exposure, any two suc- 

 cessive images of the six star impression on 

 the plate would differ by about 1.2 magnitudes. 

 That portion of the plate which contained the 

 six images was cut out and, protected by a 

 glass cover, was fastened in a small frame, 

 and could be placed on any other plate. By 

 the help of such a comparison scale the de- 

 termination of the brightness of the 38 

 selected comparison stars was made by plac- 

 ing first the brightest of them between the 

 two images of the scale, which seemed to 

 stand next to it in intensity, by which means 

 the difference of the interval was estimated 

 to tenths. The same process was then per- 

 formed for the second star, and in this way 

 the difference in brightness between the two 



comparisons was determined in magnitudes. 

 In just the same way the second comparison 

 star was joined to the third, the third to the 

 fourth, and so on, each one with the one fol- 

 lowing it in brightness. The whole work was 

 not, moreover, carried out by the use of a 

 single scale, but in all four different scales 

 were used, and, moreover, each pair of stars 

 was measured on several plates, for the most 

 part on four. All the measures were inde- 

 pendently made by Mrs. Fleming as well as by 

 Miss Leland. In order to arrange all the 

 brightnesses m magnitude, values were taken 

 for the first three, which accorded with the 

 photometric system of the Harvard Observa- 

 tory, and the magnitudes of the remaining 35 

 stars were then obtained by the use of the 

 differences found for the single pairs of stars, 

 after the application of an unimportant cor- 

 rection, which was made on the assumption 

 that the first three stars belonged to a dif- 

 ferent spectrum type from the others. The 

 final magnitudes of the 38 comparison stars 

 are given on page nine, in the last column 

 of Table III. 



It is obvious without extended discussion 

 that the whole method of the determination 

 of the brightness of the comparison stars 

 does not permit the attainment of the highest 

 degree of accuracy. Aside from the fact that 

 the ratio between the times of exposure and 

 the brightness of the images is not rigidly 

 exact, as a result of which the different 

 images do not differ from each other by pre- 

 cisely 1.2 magnitudes, the comparisons be- 

 tween two images of the scale is a somewhat 

 uncertain operation. It is, therefore, not 

 to be wondered at that the various columns 

 of determinations show in part very strong 

 systematic differences from one another, not 

 only between the two observers, but also with 

 the same observer on different plates, and es- 

 pecially with the use of different scales. Ap- 

 parently the appearance of the star images 

 played an important role. Also the different 

 scales, on accoimt of dissimilar atmospheric 

 conditions, may give a noticeably different 

 result. Perhaps it would have been better 

 for the precision of the comparisons to have 

 made the steps between successive images of 



