SUN'S PLACE AMONG THE STARS — ADAMS 



147 



opportunities for a detached electron to reunite with an ionized 

 atom. So we should expect the lines due to the ionized atoms to be 

 strong in the spectra of stars of low density and weak in those of 

 high density. This is just what observations show. An interesting 

 application of this theory is found in the case of the spectrum of 

 the sun's atmosphere (pi. 5, fig. 1). In the upper levels of the 

 atmosphere the lines due to the ionized atoms are much stronger 

 than at lower levels and this in spite of the fact that the tempera- 

 ture is lower. The lower density is the significant factor and leads 

 to a great predominance of ionized atoms. 



The application of these principles to the problem of determining 

 the luminosities of stars is very simple. If we select two stars of 

 the same spectral type whose distances have been measured by any 

 one of the methods I have described, one of which has a high 



How Intrinsic Brightness Affects Spectral Lines 

 Spectrophotometric Curves 



01 Tauri = 175 Sunt briqhtnets 



I I 



445^ -+461 



Polaris =• 3000 Sun's brightness 



I I 



4E26 4653 



42 i 5 42 2 6 4253 424& 4250 



Figure 4. — In such tracings the height of the curve shows the intensity of the line. 



luminosity and the other a low luminosity, we find the lines due 

 to the ionized atom strong in the star of high luminosity because 

 the density is low. By taking a sufficient number of stars of known 

 distance and luminosity a correlation can be established between 

 the intensities of ionized and neutral lines and the luminosity. 

 Then in the case of any other star whose spectrum can be photo- 

 graphed it is merely necessary to measure the intensities of the 

 selected lines and from the correlation-curve to read off the corre- 

 sponding luminosity or absolute magnitude. In practice, pairs of 

 neighboring lines are selected, one of which varies greatly with 

 luminosity whereas the other is comparatively insensitive (pi. 5, 

 fig. 2). In this way each spectrum of a star can be treated inde- 

 pendently, and no absolute scale of intensities is necessary (fig. 4). 



This is, very briefly, the method for deriving the luminosities of 

 stars directly from their spectra. It is not so accurate as the method 

 of angular measurement, the so-called " trigonometric method ", for 



36923— 3e 



-11 



