November 6, 1903.] 



SCIENCE. 



597 



the scale less than 1.2 magiiitudes, aud yet 

 more desirable would it have been for the 

 formation of the scale to have made use, not 

 of change in the time of exposure, but rather 

 of the cutting down of the objective (which 

 Pickering has already made use of on another 

 occasion), not, however, through the use of 

 circular caps, but by sectors, or to have made 

 use of polarization methods for decreasing the 

 light for the formation of the scale. It is 

 thus only indicated by what method it may 

 be possible to obtain greater accuracy. The 

 Pickering method for the determination of 

 photographic stellar magnitudes in and for 

 itself in the present instance the reviewer 

 regards as thoroughly commendable. It is 

 without doubt preferable to various methods, 

 especially to that of measuring the diameters 

 of the stellar images. 



In the second and third chapters are given 

 the measurements of brightness of the vari- 

 ables in the cluster w Cenfauri. The photo- 

 graphic plates employed for this purpose 

 cover the time from May 15, 1892, to August 

 16, 1898. In all 128 variables were measured. 

 At first 132 objects were selected as variable, 

 and were designated with the numbers 1-132. 

 But later the four stars Nos. 28, 31, 37 and 

 93, which appeared not sufficiently sure, were 

 rejected. The extremely wearisome compari- 

 sons, which made the greatest demands on the 

 endurance and skill of the observer, were 

 made for the most part by Miss Leland. A 

 smaller number of plates were measured by 

 the author himself, especially for the purpose 

 of determining the provisional periods. The 

 measurements were made in such a manner 

 that the variable was estimated between two 

 comparison stars, one of which was brighter, 

 the other fainter than the variable. The 

 difference in brightness was estimated in 

 grades. The method is thus the same as the 

 well-known Argelander method of comparison 

 by grades, which has proved so effective in 

 direct observations of brightness in the sky. 

 The value of a grade was not far from 0.1 

 magnitude. Generally only two comparisons 

 were made. Only when the variable ap- 

 peared equal to one of the comparison 

 stars, two other comparisons were made, 



one with the fainter and the otlier with the 

 brighter comparison star. All the obser- 

 vations for each of the 128 variables are 

 given in tabular form on pages 17-124 of the 

 volume. In all there are 15,000 determina- 

 tions of brightness, and since on the average 

 each determination consists of two compari- 

 sons, there are about 30,000 comparisons for 

 bringing the results into form. In the tables 

 are give the individual comparisons in grades, 

 and also the derived magnitudes of the 

 variables, as well as the residuals of the single 

 estimates from the mean, and also the phase, 

 that is, the time between the mean time of 

 the exposure and the preceding computed 

 maximum, and finally the residuals of the 

 observed magnitudes from the mean light 

 curve found for the star in question. When 

 this residual was 0.40 or more the measure- 

 ment was repeated, and in case the new 

 measurement brought no sure conclusion it 

 was checked still a second or third time, 

 eventually also by another observer. 



Most interesting is the fourth chapter, 

 which contains the results of all the observa- 

 tions. In the tabulated grouping of the ele- 

 ments of the light changes are found only 95 

 examples out of the 128 variables. For 13 

 variables no periods at all were found, and 

 for 20 objects the results obtained were so 

 doubtful that they are not given in the table 

 itself, but are mentioned only in the remarks 

 which follow the table. In regard to the 

 method of determining the periods, the epochs, 

 the extremes of brightness and the mean curves, 

 no description at all is given in this chapter, 

 and one only finds in the whole volume 

 some brief notes on page 232. The method 

 of computation is, however, so well known in 

 such cases that a detailed description would 

 perhaps be superfluous. The plates made on 

 the successive days, May 3 and May 4, 1898, 

 eight plates on the first, and six on the second 

 date, which cover periods of seven and five 

 hours, respectively, give for nearly all the 

 variables a sufficient indication of the char- 

 acter of the light changes, and render possible 

 in most cases the finding of an approximate 

 value of the period. An improvement on this 

 approximate period can generally be obtained 



