530 SCIENCE PROGRESS 



to be little doubt but that the system of globular clusters is 

 of the wide extent which he attributes to it. What, then, are 

 the arguments of Curtis in favour of the smaller scale of distance? 



Curtis denies, in the first place, the validity of the as- 

 sumption that the Cepheid variables in clusters and in the 

 immediate neighbourhood of our stellar system must obey 

 the same luminosity-period relationship as the Cepheids in 

 the Smaller Magellanic Cloud, and he further asserts that the 

 method of fixing the zero of this relationship, by means of the 

 parallactic motion of the nearer Cepheids, is based upon in- 

 adequate material — the small proper motions being uncertain 

 and the number of stars too few to be reliable. But unless the 

 Magellanic Cloud is a unique region of space, it is difficult to 

 believe that the cause of Cepheid variation is other than an 

 intrinsic property of the star itself, conditioned by its tempera- 

 ture and density. If a luminosity-period relationship holds for 

 any one part of space, there seems no escape from the conclusion 

 that it will also hold universally. Further, as already men- 

 tioned, the distances assigned by Curtis would require an 

 impossibly large parallactic motion for the Cepheids. 



Curtis further denies that it is legitimate to use the value 

 -1^5, found by Shapley as the mean luminosity of the twenty- 

 five brightest stars in Messier 3, to deduce the distances of other 

 clusters. He considers that the Cepheid variables are a small 

 and possibly exceptional class, and that they are not typical 

 of the stars in general. He discusses the brightness of the 

 stars in several nearer open clusters, and obtains a mean value 

 of about + 2**. Assuming this value to hold for the globular 

 clusters, he deduces a much smaller distance than Shapley. 

 But this assumption is open to objection, for the number of 

 stars in a globular cluster is much greater than in the open 

 clusters, and amongst the relatively small number of stars in 

 the latter it is hardly surprising that the mean magnitude of 

 the brightest stars should be fainter than for the globular 

 clusters. Moreover, Shapley uses the argument based upon the 

 brightest stars in the cluster mainly to corroborate the dis- 

 tances found from the Cepheid variables, and the results obtained 

 in the two ways support one another. The various lines of 

 argument developed by Shapley are so concordant in their 

 conclusions that there seems little room to doubt their validity. 

 Both authors discuss the status of the spiral nebulae. Curtis 

 is strongly in favour of the view that these constitute island 

 universes, and obviously the large scale found by Shapley for 

 our Galactic System would militate against this view. There 

 is very little reliable evidence as to the distances of the spiral 

 nebulae. From the radial velocities found for spirals, which 

 are of the order of 700 kms. per second, it may be deduced that 



