November 6, 1903.] 



SCIENCE. 



599 



the star remains at minimum during half of 

 the period, perliaps with very slight fluctua- 

 tions. Moreover, the light curve is uniform, 

 the increase of light very rapid and the de- 

 crease noticeably slower. The amplitude of 

 the light variations attains in general 1 mag- 

 nitude and a little more, and the periods lie 

 between 12 and 15 hours. The second class 

 embraces the relatively small number of 19 

 stars. They differ from the first only in the 

 fact that the long duration of minimum is 

 absent, and that in general the increase is 

 much slower. The decrease continues with 

 ever-lessening rapidity till the beginning of 

 the following increase. In many cases on 

 the descending arm of the curve a ' stillstand ' 

 appears to exist. The whole range in bright- 

 ness is generally less than 1 magnitude, and 

 the periods range between fifteen and twenty 

 hours. In the third class are counted about 

 34 stars. The peculiarities of these stars are 

 that the durations of increase and decrease 

 are not very different ; also, the increase is in 

 general somewhat shorter than the decrease, 

 but in some cases they are equal, and it even 

 happens that the opposite is true. The ampli- 

 tude of the variation is for the most part not 

 much greater than 0.5 magnitude, and the 

 periods are between eight and ten hours. 



As one sees, the three classes are not 

 separated very widely from one another; 

 especially, the second is only a little different 

 from the first, and there are transitions from 

 one to the other, but as a whole the differences 

 of the changes in light, especially between the 

 extreme members of the classes a and c, are 

 marked. The circumstances by which the 

 variations of light take place among the 

 different members of the cluster, whether due 

 to occultation, rotation or to other causes, 

 must in any case be very different. 



In regard to many peculiarities of the light 

 curves of the different stars the remarks at 

 the end of Chapter IV. give information. 

 Notes are also found in regard to the doubt- 

 ful objects for which no elements are given 

 in the table. 



The fifth chapter is devoted to a special 

 study of four variables of the cluster. The 

 author aimed to prove by this special in- 



vestigation whether, noticeably better results 

 could be obtained if the number of comparisons 

 of brightness was increased, and especially if 

 comparison stars, greater in number, and in 

 the immediate vicinity of the variable, were 

 employed. For each of the selected variables 

 a list of comparison stars was chosen, no one 

 of which was more than 2' distant from it. 

 The variable was then compared with as many 

 of these as possible, and the difference in 

 brightness determined each time in grades. 

 All the new measures were made indepen- 

 dently by the author and by Miss Leland. The 

 magnitudes of the new comparison stars were 

 determined on four plates by joining them to 

 the previously chosen fundamental stars, and 

 also in this case by both observers indepen- 

 dently. 



In addition to these direct determinations, 

 the brightness of the new comparison stars 

 was also found by an indirect method, by 

 which, as is done in visual comparisons of 

 variable stars, by the comparison of the 

 variables with the comparison stars the differ- 

 ences in brightness of these stars in grades 

 were derived; and then by the union of the 

 various differences a scale of grades was 

 formed. The union of the values of the com- 

 parison stars, obtained in these two different 

 ways, furnished at length definite values for 

 them, from which the results of both observi-r 

 were united to form mean values. The differ- 

 ences between the two observers are in general 

 small, although there are indications of sys- 

 tematic differences. The magnitudes of the 

 four variables were derived from the observa- 

 tions simply by the Argelander method, by the 

 help of the scale of grades, first in grades and 

 then in magnitudes. 



The result of this somewhat complicated 

 and prolix research was, in the case of three 

 of the selected variables, a complete ac- 

 cordance with the previously derived elements, 

 in which, naturally, as might be exiiected 

 from the large number of the measures and 

 especially from the greater nearness of the 

 comparison stars, the residuals of the single 

 observations from the light curve resulted 

 somewhat smaller. The case was different 

 with the fourth star, which of all the variables 



