458 SCIENCE PROGRESS 



disintegration of the clusters. In this paper he attempts to 

 trace the early history of such a collision. Considering first 

 the simpler case of a cluster meeting a uniform shower of stars, 

 it is found that it will thereby be spread out so as to be almost 

 disc-shaped, reminiscent of the form of the Ursa Major cluster. 

 If now there are two equal collisions in perpendicular directions, 

 star streaming is set up in the mutually perpendicular directions, 

 and the resulting law of distribution of velocities is Schwarz- 

 schild's law. Under less simple conditions this law is still 

 found to hold. If again the second colliding cluster is smaller 

 than the disc on which it impinges, there will be a flat outer 

 annulus of stars more or less unaffected by the second encounter, 

 and the effect will be to produce a lens-shaped centre surrounded 

 by a galactic ring, somewhat similar to what we find in our 

 own universe. 



This paper gives a remarkable view of the ways in which 

 our universe might have been formed, explaining also the 

 observed phenomena of star-streaming. Although but a 

 rough model, it is extremely suggestive and comprehensive. 

 It can of course only be accepted if we are willing to accept 

 the island-universe theory (that our stellar system is but one 

 of many universes scattered through space), but there are 

 many other reasons to believe that this is a correct view ; the 

 fact also that only on this view has it been possible to formulate 

 a logical theory of the formation of our stellar system may 

 be regarded as considerably strengthening the theory. 



A. S. Eddington, M.N., R.A.S., lxxvi. p. 525, 1916, proves 

 that in any star cluster in a steady state the internal kinetic 

 energy is one-half the exhaustion of potential energy, a useful 

 result which has an analogy in gas theory. This result is 

 applied to determine the rate of dissolution of a cluster, which 

 naturally tends to expand, taking into account the mutual 

 attraction of the cluster. Eddington has shown that the 

 Taurus cluster could not have existed more than 57 million 

 years unless the scattering had been counteracted in some 

 way ; the present considerations indicate that the time taken 

 by the cluster to increase from one-half its present size to its 

 present size is of the order of not less than 200 million years. 



A. S. Eddington, M.N., R.A.S., lxxvi. p. 572, 1916, also 

 discusses the significance of the law of density p = (c 2 + r 2 )~ &/i 

 found by C. Plummer and H. von Zeipel to govern the dis- 



