576 



RUSSELL— RELATIONS BETWEEN SPECTRA [April 20. 



are unusually few very faint stars in the region of the Pleiades, 

 which has been explained as the result of partial opacity of the 

 nebulosity surrounding the cluster. Some of this nebulosity pre- 

 sumably lies between us and the stars of the cluster, and cuts off 

 a part of their light, which would make the distance computed on 

 the assumption that there was no absorption come out too great. 

 If such absorption exists, it should be possible to determine its 

 amount, and allow for it. 



It is of obvious interest to inquire in what other respects besides 

 brightness the giant and dwarf stars of the same spectral class 

 differ from one another. One line of approach is furnished by the 

 visual binary stars. It is well known that, when the orbital ele- 

 ments and apparent brightness of a binary pair are given, we can 

 find what Professor Young calls the " candle-power per ton " — more 

 exactly, the ratio L^/M^ where L is the combined light of the pair, 

 and M the combined mass — without knowing the parallax. The 

 writer has recently shown- that this principle can be extended 

 by simple statistical methods to the stars known to be physically 

 connected whose orbits cannot yet be computed. In this way about 

 350 stars have been investigated, and it is found that they fall into 

 two series, similar in all respects to the giant and dwarf stars, — 

 one marked by high luminosity per unit of mass, nearly the same 

 for all spectral classes, and the other by small luminosity per unit 

 of mass, diminishing very rapidly for the redder stars. By means 

 of the parallactic motions of these groups of stars, an approximate 

 estimate can be made of their distances, absolute magnitudes and 

 masses, with results which may be summarized as follows. 



TABLE IV. 

 Mean Absolute Magnitudes and Masses of Binary Stars. 



' Science, N. S., vol. 34, p. 524, 1910. 

 From directly measured parallaxes. 



