December 1, 1898.] 



KNOWLEDGE 



281 



loses again in about four hours. Figure 2 represents 

 the maximum of Xo. 12 in this chister. The further 

 sections of its curve appear to inchide a protracted 

 minimum, but had not been traced out at the date of 

 publication. A specific resemblance is evident between 

 Nos. 18 and 12, as between Nos. 42 and 81. 



Prof. Barnard was greatly struck, in the course of his 

 scrutiny of M 5, with a number of "inky black spots" 

 near, but not in its densest parts, recalling the analogy of 

 the model northern cluster, M 13 Herculis. Even the 

 minor group is described as " a gorgeous object," faintly 

 visible to the naked eye, I with a yross diameter, so to 

 speak, of about nineteen minutes of arc. ] 



M 3, the beautiful radiated star-throng in Canes Vena- 

 tici, contains the largest proportion of variables of any 

 cluster yet examined. One in seven — one hundred and 

 thirty-two in all — of the nine hundred components 

 separately discernible on the Arequipa plates fluctuate 

 extensively. Their laws of change, however, have received 

 up to the present only preliminary attention. The same 

 may be said of the fifty-one variables in M 15. This is a 

 cluster in Pegasus, considered by Dr. Roberts to be strongly 

 nebulous. The stars, moreover, imprinted on a negative 

 taken November 4th, 1890, exhibit an arrangement " in 

 curve lines, and patterns of various forms, with lanes or 

 spaces between them." : 



The absence of variables from most globular clusters 

 accentuates the wonder of their abundance in others. Nor 

 is it possible to discover any corresponding differences of 

 state or aspect. Thus in 47 Touoani, an exact analogue on 

 a slightly reduced scale of m Centauri, the causes of vari- 

 ability seem to be strictly localized. A nest of six changing 

 stars was at once brouglit to light by comparisons of Prof. 

 Bailey's plates, but none are scattered at large through 

 the assemblage, which is otherwise made up of perfectly 

 stable components. Still more remarkably, only two out 

 of two thousand stars rigidly tested in the great Hercules 

 globe have proved variable, and that to an insignificant 

 extent ; whereas nearly one per cent, of the visible populace 

 of heaven shine more or less unsteadily. Now, this last 

 cluster is not only generally nebulous, but many of its 

 outlying members are separately provided with luminous 

 appendages; so that the disconnection of nebulosity from 

 variability in light, already tolerably obvious, is, by these 

 fresh experiences, emphatically re-affirmed. Another fact 

 of interest, albeit likewise of negative import, is that M 80, 

 the cluster in Scorpio illumined in 1860 by the blaze of a 

 " new star," is exempt from the slightest suspicion of 

 actual variability. Nor has the phenomenon been met 

 with in any " irregular " group, such as the Pleiades, the 

 " Beehive," the chromatic cluster about k Crucis, or the 

 blazonry in the sword-handle of Perseus. 



To what, then, can it be ascribed ? Prof. Pickering 

 makes the "provisional ' suggestion that the key to the 

 enigma may be found in the relation to the line of sight of 

 a common plane of revolution, axial or orbital. Irregular 

 collections, in this view, possess no such fundamental 

 plane ; while spherical assemblages show effects of 

 variability depending upon its approach to coincidence 

 with our visual ray. The hint is of tempting significance ; 

 it opens up possibilities of insight into clugter-mechanism 

 such as might have been deemed desperate of attamment 

 from any other point of view. Yet there are fatal objections 

 to its unconditional adoption. It implies two rationales 



* The co-ordinates are given by Pickering in Astr. Xack., No. iJ354. 



f B. Sprague, " Popular Astronomy," Vol. I., p. 408. 



X Barnard, Astr. Sack., No. 3519. 



§ " Photographs of Star-Clusters and N'cbuUc," p. 119. 



of stellar variability — the spot-theory, and the eclipse- 

 theory. In the first, the period is determined by the 

 rotation of a single body, in the second, by the mutual 

 revolutions of a pair. Bodies variable through axial 

 movement are necessarily assumed to be brilliant on one 

 side, comparatively obscure on the other. Piebald suns, 

 however, may be dismissed from consideration as mere 

 mathematical postulates. They serve conveniently as the 

 basis of formulie, but lie apart from physical reality. A 

 degree of interior mobility, indeed, utterly inconsistent 

 with the presence of fixed dark areas, is indispensable to 

 the maintenance of white radiation. For it must be 

 remembered that these clustered stars are unvarying in 

 their variability. They do not brighten unawares, or 

 casually " hide their diminished heads." The phases of 

 each are settled once for all by unalterable law. 



The eclipse-hypothesis of short-period variability stands 

 on a very different footing. There is at any rate good 

 reason for holding stars of the 5 Cephei class to be 

 genuine spectroscopic binaries, with synchronizing light- 

 and-motion periods. But no agreement between their 

 epochs of minimum and of possible eclipse has been 

 established, to say nothing of other glaring incongruities 

 and improbabilities. In addition, eclipsing couples of the 

 authentic Algol stamp are not forthcomimj nmoni/ iiiiijfiijated 

 stills. Yet they should, on the geometrical theory, abound 

 and super-abound. Their apparent absence must be 

 accounted for in any plausible speculation as to the 

 causes of variability in globular clusters. 



VARIABLE STARS IN CLUSTERS. 



CIRCULAR No. 33 of the Harvard College Observa- 

 tory deals with the results of a systematic search 

 by Prof. S. I. Bailey for variable stars in globular 

 clusters. The whole number of stars examined 

 was nineteen thousand and fifty, of which five 

 hundred and nine are variable. This amounts to one 

 variable in thirty-seven stars, or nearly three per cent. 

 It does not follow, however, that clusters in general con- 

 tain more variable stars than occur elsewhere, for, if we 

 except the four clusters, m Centauri, Messier 3, Messier 5, 

 and Messier 15, which together contain three hundred and 

 ninety-three variables, an average of seven per cent., the 

 remaining nineteen clusters have one hundred and sixteen 

 variables among thirteen thousand three hundred and fifty 

 stars, or less than one per cent. There is a very striking 

 difference between the results obtained in clusters equally 

 rich in stars, as, for example, between Messier 13, the great 

 cluster in Hercules, where an examination of one thousand 

 stars shows two variables, one in five hundred ; and 

 Messier 3, where, among nine hundred stars, one hundred 

 and thirty-two are variable, one in seven. A common 

 plane of revolution, orbital or axial, of the different 

 systems or individuals of star clusters, and the relation of 

 that plane to the line of sight, might provisionally be 

 suggested as a possible explanation. 



The periods and light curves of several variables in 

 other clusters have been determined, but the study of those 

 in iu Centauri is well advanced. This cluster may be 

 called the finest in the sky. It lies just within the borders 

 of the Milky Way. There 'are no bright stars near. To 

 the naked eye it appears as a hazy stir of the fourth 

 magnitude. It has a diameter of about forty minutes. The 

 brightest individual stars in this region are between the 

 eighth and ninth magnitude. Over six thousand stars 

 have been counted on one of the photographs, and the 

 whole number is much greater. Only about three thousand. 



