October 28, 1922] 



NA TURE 



579 



Since the computed distances depend equally on 

 absolute and apparent magnitudes, they could be con- 

 siderably changed, if the apparent magnitudes are 

 widely wrong, without disturbing present ideas about 

 stellar luminosity. The fundamental work on mag- 

 nitude standards at Mount Wilson, Harvard, Green- 

 wich, and elsewhere supply, however, a basis of un- 

 questioned value for the cluster work. The apparent 

 magnitudes I have observed in clusters cannot, I 

 believe, be far wrong ; 19 but has the light of these 

 distant stars been reduced in transit so that the ap- 

 parent magnitudes as observed are not a true index of 

 distance ? This question should be kept in mind, but 

 the following points seem to show that the observed 

 apparent magnitudes do not differ seriously from true 

 apparent values because of a hypothecated diminution 

 of light during its passage through space : 



(1) The absence of measurable differential light 

 scattering in space, which would appear as a dependence 

 of star-colour on distance. 



(2) The apparent restriction of known obstructing 

 matter to regions near the planes of the local cloud 

 and the Galaxy ; 20 the globular clusters we study are 

 practically all outside these regions. 



(3) The diameter-magnitude correlation for globular 

 clusters, which shows, almost without exception, that 

 the globular clusters with large angular diameters are 

 bright, and that the faint globular clusters are of small 

 angular diameter. 21 



(4) The absence of observable proper motion for 

 clusters, notwithstanding large space velocities. 



Incidental Results 



In the course of the investigation on the scale of the 

 galactic system a number of incidental contributions of 

 general scientific interest have been made. 



I. The great distances of globular clusters provide a 

 much more sensitive test of the degree of selective 

 scattering of light in space than was formerly available 

 from the studies of the colour of nearby stars. Results 

 from many clusters, including the most remote, agree 

 in showing no certainly measurable effect of distance 

 on colour. We conclude that much less than one per 

 cent, of the starlight is scattered while travelling for 

 one thousand years through space. 22 This result, 

 which does not hold of course for some restricted 

 nebulous ■ regions, indicates the extreme vacuity of 

 interstellar space. 



II. This absence of a measurable effect of distance on 

 colour contributes an additional fact of some interest 

 with regard to the nature of light. It is direct ob- 

 servational evidence that the amplitude of the light 

 pulses of different wave lengths has suffered no differ- 

 ential alteration while travelling for more than 100,000 

 years. The age of this incoming stellar radiation, 

 compared with that of the radiation used in laboratory 

 experiments, is uncommonly impressive. 



III. In a more definitely quantitative manner we 

 can again use the base line in space and in time afforded 

 by the globular clusters to derive another property of 

 radiation. The times of maxima of several short-period 

 Cepheid variable stars in the globular cluster Messier 5 

 (distance 12,000 parsecs) have been measured con- 

 currently with blue and yellow light. Within the 



NO. 2765, VOL. I IO] 



errors of observation no difference for the two colours 

 is found in the time of these stellar outbursts. 23 

 That is, in travelling for 40,000 years, radiation that 

 differs in wave length by 20 per cent, differs in time 

 of arrival at the earth by less than two minutes, if at 

 all. This is equivalent to a difference of less than one 

 millimetre in a distance of 5000 miles. Stated other- 

 wise, blue and yellow light travel with the same velocity 

 with an uncertainty of less than one part in ten thousand 

 million. 



IV. In still another way can we make valuable use of 

 the long base line provided by the remote clusters. A 

 considerable analysis of the distribution of spectral types 

 among the giant stars shows no measurable difference 

 for near and distant globular clusters. This strongly 

 suggests of course that the nearest systems are not 

 appreciably more advanced in their evolution ; but 

 because of the finite velocity of light and the great 

 differences in distance, they are, in our records, nearly 

 200,000 years older than the farthest ones. With these 

 globular clusters we can, in effect, examine the process 

 of stellar evolution throughout an interval of 2000 

 centuries. We find no evidence of change in that 

 interval of time. 24 



Now Eddington has shown that very conspicuous 

 advances in the evolution of a giant star would occur 

 in less than 50,000 years, if gravitation is the main 

 source of radiant energy. 2 '' We are led to believe, 

 therefore, that gravitational contraction is not the 

 main source of the energy that maintains the radiation 

 of stars ; it appears that the energy must come from 

 the atom, and probably is released in the course of 

 the transformation of the chemical elements. 



The evidence for a slow stellar evolution is strongly 

 supported, I believe, by the existence of stars, which 

 are still in their giant stage, in the open clusters that 

 move along the galactic plane. Dimensions now 

 assigned the galactic system are so large that a single 

 oscillation of a cluster must require millions of years. 

 In fact, a greater space scale for the Galaxy practically 

 makes necessary a long time scale and a slow stellar 

 development. The extreme slowness with which the 

 periods of Cepheid variables change, as Eddington has 

 pointed out, also demands the new source of energy. 26 



V. In this connexion it may be observed that the 

 question of the dependence of the speed of evolution 

 on mass for a giant star is probably answered observa- 

 tionally by the regularly occurring phenomenon in 

 clusters of increasing blueness with decreasing bright- 

 ness. The well-known investigations by Eddington 

 and Jeans indicate that high absolute brightness is 

 associated with great mass, but the theory is not 

 definite in regard to the relative rates of evolution for 

 different masses. From the clusters we would conclude 

 that the greater the mass the slower the develop- 

 ment. 27 



VI. The most luminous stars in globular clusters, 

 the spectra of which, by the way, appear to have the 

 " c-characteristics " which are associated with extra- 

 ordinary brightness, are more concentrated to the 

 centre than fainter stars. This condition independ- 

 ently supports the inference that the most luminous 

 are the most massive stars. 



VII. The remotest object for which a definite esti- 

 mate of distance has yet been made is one of these 



