DIAMETERS OF THE STARS — DAN JON". 171 



As we have just seen, the masses found in the case of double sys- 

 tems fall within rather narrow limits, and the ratio between the ex- 

 treme values is not above 1,000. We are thus forced to conclude that 

 the amount of condensation attained by the stars varies very much 

 from one to another for the range of masses stretches from 1 to 1,000 

 whereas that of the volumes is from 1 to 1,000,000,000. - We must 

 conclude that certain stars have a density a million times smaller 

 than others. A specific comparison will emphasize the difference: 

 platinum is only 250,000 times as dense as hydrogen under normal 

 circumstances. 



We can foresee the importance of such a result for cosmogony and 

 the theory of evolution. Despite its acknowledged dependence to 

 some extent on hypotheses it conforms to all the known facts as well 

 as to the theoretical researches of Eddington. 



The immense diversity of stellar densities seems to be thus well 

 established. 



DWARF AND GIANT STARS. 



So far we have been forced, in order to estimate stellar diameters, 

 to get first the angular diameter and then to correct for the distance of 

 the star from the earth. We can perform the operation in a single 

 step by starting from the absolute magnitudes since these are referred 

 to a standard distance. 



A list of 1,64G stars whose absolute magnitudes were directly ob- 

 tained by the remarkable method of Adams has just been published. 

 This carries to nearly 2,500 the number of stars for which we have 

 this datum. 



Interesting consequences follow from a statistical study of these 

 stars. In order legitimately to infer diameters from these results we 

 must first assort the stars into classes. It is only with stars of the 

 same spectrum type that we may infer equality of surface bright- 

 ness. 



Figure 1 shows graphically the results. The abscissae refer to 

 absolute magnitudes, the or-dinates, the numbers of stars of the 

 different spectrum classes. A capital result at once catches the eye. 

 It is the separation of the stars in each of the advanced classes into 

 two perfectly distinct groups. While the curves for classes B, A, 

 and F show only one summit, corresponding to the most frequent 

 magnitude, the other curves each have two maxima. The first cor- 

 responds to giant stars, the second to dwarfs. Thus we here find 

 verified with an unexpected definiteness the existence of stars of 

 very different volumes. Further there is not a continuous variation 

 in the number of stars as we pass from the giants of one class to the 

 dwarfs of the same class. 



