October 21, 1922] 



NA TURE 



545 



I. 

 ^PHROUGHOUT the known sidereal universe there 

 J- is, among material bodies, an obvious associative 

 tendency, which we see well illustrated in meteor 

 showers, in satellite and planetary systems, in binary 

 stars, and in larger stellar groups such as the Pleiades. 

 These various products of gravitational ordering are 

 clearly but parts of still greater systems, and one of the 

 most fascinating of astronomical studies is to attempt 

 to seek out the structure of an all-inclusive sidereal 

 organisation. 



It is proposed in this communication to discuss the 

 structure and extent of the sidereal system as indicated 

 by recent studies of stellar clusters and variable stars. 

 My own observational investigations of these objects, 

 and the deductions based upon them, have been mainly 

 published in Contributions and Communications of 

 the Mount Wilson Observatory 1 from 1914 to 1918. 

 The present discussion is made in the light of criticisms 

 and numerous tests to which the conclusions have been 

 subjected during the past four or five years. 



It appears that we have three principal types of 

 celestial objects to consider — the diffuse nebulae, the 

 stars, and the nebulae of the spiral family. The first 

 two are generally thought to be related as parent and 

 offspring. The stars, having formed, as we think, 

 out of nebulous pre-stellar states, are, apparently, 

 largely organised into groups, a common, possibly 

 prevailing form being the globular cluster. It is from 

 combinations of these clusters that I believe our galactic 

 system has developed. 2 From the work on clusters 

 there can be little doubt of the enormous mass and 

 dimensions of the galactic system as compared with 

 clusters and nebulas. Its flat form and heterogeneity, 

 its content of numerous fragmentary systems (open 

 clusters, wide binaries, spectrally-similar groups) of 

 apparently different ages and separate origins, and 

 its control over the motions of the clusters and near-by 

 spirals, have led me for some years to advocate the 

 hypothesis that the Galaxy is a growing composite of 

 disintegrating minor systems. The Galax}' appears to 

 include all the common sidereal types, with the probable 

 exception of most nebulae of the spiral family. But 

 the latter are apparently not stellar in composition, 

 nor galactic in size. I think present evidence favours 

 but does not establish the hypothesis that typical 

 spiral nebulas represent a sidereal evolution not directly 

 connected with that of stars. 



The foregoing paragraph may serve as a brief 

 outline. Some of the details may now be considered, 

 but, before proceeding with the discussion, I should 

 like to point out that the proposed interpretations 

 involve the following somewhat fundamental assump- 

 tions, if we choose to call them assumptions : (1) that 

 gravitation directs the organisation and motions of 

 celestial bodies ; (2) that the physical laws we know 

 are equally valid in all parts of the space with which 

 we are familiar ; (3) that the Russell-Eddington theory 

 of stellar evolution is correct in its general features. 



1 Adapted from an 

 Association on May 31 



Address given before the British Astronomical 



The Galactic System. 1 

 By Dr. Harlow Shapley. 



Certainly these three are not serious restrii tions. 

 On the first I need make no comment here. The second 

 is the basis of our belief in the general uniformity oi 

 conditions throughout the stellar system. It insists 

 that our stellar neighbourhood is not operated by local 

 laws. It is a highly reasonable but necessary assump- 

 tion before we can safely compare the luminosities 

 and other properties of stars near the sun with those 

 of stars in distant parts of the galactic system. The 

 third assumption, the Russell-Eddington theory, is 

 not necessary for my conclusions concerning the 

 dimensions of the galactic system, but is essential in 

 putting together the general scheme, and also in trying 

 to interpret some anomalies of the spiral nebula-. 

 We might call the evolutionary scheme the Lane- 

 Lockyer - Ritter - Sampson - Emden - Schwarzschild - 

 Hertzsprung - Russell - Eddington - Jeans - Eggert 

 theory, but Russell and Eddington have been the most 

 important contributors to the theory in its present 

 form. 



Concerning Star Clusters. 



Clusters " - of stars can be placed in two fairly distinct 

 categories, the globular cluster, of which nearly ninety 

 arc now known, and the open or loose cluster, of which 

 there are several hundred. 



Most globular clusters (but not quite all) appear to 

 be remarkably alike in general structure. Compared 

 with naked-eye objects they are extremely remote ; 

 hence their stars, though apparently very faint, are 

 actually of high intrinsic luminosity. Few stars in 

 globular clusters are brighter than the eleventh apparent 

 magnitude. Each globular cluster contains some tens 

 of thousands of these intrinsically bright stars, and 

 possibly a far greater number of dwarfs, which at present 

 are beyond the reach of our telescopes. Of high 

 importance is the fact that the cluster stars appear 

 to be remarkably similar to the stars in the solar 

 neighbourhood in spectral type, colour, variability, 

 and other properties, notwithstanding the much higher 

 stellar density near the centres of globular clusters. 



Open clusters are of great variety. They range in 

 brightness from naked-eye systems, such as the Hyades, 

 to small, dim groupings that may be nothing more 

 than chance aggregations of faint Milky Way stars. 

 Open clusters vary also in richness, in apparent and 

 real dimensions, in stellar content. One property 

 they have in common : they are all near the plane of 

 the Milky Way. The distance of the average open 

 (luster is smaller than that of globular clusters, but 

 the determination of distances for the former is gener- 

 allv subject to much uncertainty. This fact is due to 

 variety of form and content, and to the absence from 

 open clusters of peculiar types of highly luminous stars, 

 which for globular clusters serve to determine positions 

 in space. 



The estimation of the distances of globular clusters, 

 which has been the most important part of the work 

 on the scale of the sidereal universe, must be based 

 on the newer methods of measuring space. The various 

 trigonometrical methods, when applied to globular 

 clusters, so far give negative results, indicating only 

 that the distances are very great. The various photo- 



NO. 2764, VOL. I IO] 



