242 CARNEGIE INSTITUTION OF WASHINGTON. 



A further indication of the complexity of the energy relations in the 

 spectrum is indicated by the fact that the curves found by plotting 

 exposure ratios for different regions of the spectrum against each other, 

 for example, violet-yellow against violet-red, seem to be smoother than 

 those found by plotting either ratio against the spectral types of the 

 stars in question. The deviations are larger than the uncertainties in 

 the spectral classification, and suggest that there are other elements 

 besides luminosity and spectral type which determine the color of a 

 star. In fact, the correlation with luminosity is probably more or 

 less artificial, atmospheric absorption doubtless being the fundamental 

 factor. 



The method of exposure ratios has also been used by Mr. Scares and 

 Mr. Hubble to determine the colors of the nebulous stars of Herschel. 

 An account of the results is given on page 236. 



Investigations of Star-Cltjsters. 



As stated in earlier reports, the distribution of stellar clusters in 

 space has led Mr. Shapley to the conclusion that the galactic system is 

 very much larger than was generally supposed before the inception of 

 work on clusters at Mount Wilson. Some students of the subject have 

 naturally hesitated to accept a pronounced modification of long-stand- 

 ing interpretations, and it is therefore important to develop new 

 methods of determining the distance of clusters. To establish the new 

 view of galactic dimensions, it is sufficient to show that the cluster 

 stars are of high intrinsic luminosity, because the brightest stars in 

 globular clusters must be giants if the size of the sidereal universe is of 

 the order assigned by Mr. Shapley, while if the older view holds, 

 these stars must be dwarfs. 



An examination of the spectra of some of these stars of types G and K 

 by Mr. Adams has already shown that they have the characteristics of 

 giants. A comparison of the distribution of energy in the spectra with 

 the corresponding spectral types of these objects by Mr. Shapley by 

 an extension of the method used by Adams and Kohlschiitter also shows 

 that this conclusion must be accepted. A 6° prism, 14 inches in diam- 

 eter, was placed in the converging beam about 20 inches from the 

 secondary (134-foot) focus of the 100-inch telescope. Small stellar 

 spectta (chromatically elongated images) maybe photographed through 

 the prism, over a field of about 10' radius. "V\Tien specially sensitized 

 jilates are used, the whole interval from X3800 to X7500 is recorded, 

 with a break in the middle due to the relative insensitiveness of the 

 plate to yellow-green light. The blue-violet half of one of these chro- 

 matic images represents that part of the starlight photographed on 

 ordinary plates; the other half lies in the red and infra-red, a region 

 that has been comparatively little studied in stellar spectra. The 

 relative intensity of the two halves depends on spectral type, but for 



