November G, 1003.] 



SCIENCE. 



GOl 



luiinbers in the last column, which give how 

 many variables are found in the different 

 groups for each 100 stars, are not noticeably 

 different from one another. It follows, there- 

 fore, that in the whole cluster, in a definite 

 area the proportion of variables is about con- 

 stant, and that in no place occurs a very 

 large number. 



In regard to the three classes into which 

 the variables were divided, the mean distances 

 of the stars from the center of the cluster of 

 the different groups are 7.7', 8.1' and 8.9'. 

 It appears, therefore, that no marked differ- 

 ence is indicated in the distribution, on ac- 

 count of the character of the light changes. 

 The mean ma.ximum brightnesses of the three 

 classes are 12.99, 13.10 and 13.33 magnitudes, 

 and the minimum, 14.11, 13.97 and 13.89 

 magnitudes. The amplitudes are thus dis- 

 tinctly different. The mean lengths of the 

 periods are 0.58617, 0.75153 and 0.39463. 



It is of groat value that for the 95 variables 

 whose periods are determined the light curves 

 are given not only graphically in Plates II. 

 to VII., but that the coordinates of the light- 

 curves are given in detail on pages 210 to 222. 

 and in such a way that for each star the 

 period is divided into 24 equal parts, and 

 for each part the corresponding brightness is 

 given. There are also given typical light 

 tables for the three classes, by taking from 

 each class five especially characteristic ex- 

 amples, and taking the mean values for these. 



The chapter closes with some remarks con- 

 cerning the possible causes of the light 

 changes of so many stars in one and the 

 same cluster. The author hesitates to accept 

 widely current hypotheses, and to test them 

 from the available materials of observation. 

 He concludes only that from the complete uni- 

 formity of the periods, which is shown in tu 

 Centaurl for a long time, the variability must 

 be associated with some regularly returning 

 phenomenon, either through regular eclipses 

 by bodies which are in revolution about each 

 other in definite paths, or through the rota- 

 tion of unequally illuminated or irregularly 

 shaped bodies. The explanation by means 

 of occultations, as with Algol stars, involves 

 great difficulties. For, if one would assume, 

 as at first sight might appear plausible, that 

 the planes of the orbits of the different double 

 star systems of the cluster lie about parallel 

 to each other, and that on that account the 

 light-curves of the different systems should 

 show a certain similarity, the form of the 

 light-curves, nevertheless, speaks against the 

 eclipse theory. In u> Centauri are found three 

 different types of light-curves, but in other 

 clusters investigated by the author the first 

 type with the long duration of minimum is so 

 much more common that we may regard this 

 as the characteristic type of variables in star 

 clusters. This type of light-curve is, however, 

 entirely distinct from the Algol type. The 

 eclipse would have to last for a considerable 

 part of the period, and this would hardly be 

 consistent with any orbit system. Also the 

 assumption of an axial rotation with un- 

 equally luminous surfaces seems not very 

 probable, when one considers that such a large 

 number of similar variables is concerned. 

 The whole phenomenon is at this time some- 

 what enigmatical, and we must await the 

 investigation of the occurrences in other 

 clusters before further conclusions are per- 

 missible. 



In an appendix to the volume are given 

 preparatory studies toward the measurements 

 of all those clusters in which more than one 

 variable has been found. There are given the 

 positions of those objects selected to serve as 

 fundamental stars, and the comparison stars 



