284 SCIENCE PROGRESS 



stars, which brushes back the atmosphere from the front sur- 

 face of the secondary. Harlow Shapley (Astroph. Journ. 40, 

 p. 448), in a thorough discussion of the problem of Cepheid 

 variation, advances strong arguments against the possibility 

 of these stars being binary systems. The Cepheids are all 

 stars of very great absolute luminosity and of large dimensions, 

 and the small dimensions of their orbits, obtained on the binary 

 theory, seem very improbable, being often of smaller dimensions 

 than the stars themselves. Shapley favours the hypothesis 

 that the variations in brightness are due to regular periodic 

 pulsations in the atmosphere of a single star. This would 

 explain the connection between maximum brightness and 

 maximum radial velocity. The periods of free vibrations and 

 the variations of the radial velocity obtained on this hypo- 

 thesis are of about the right order of magnitude, but the neces- 

 sary displacements in the atmosphere appear somewhat large. 

 The Cepheids are all solar-type stars, and if this hypothesis is 

 true, we may regard our Sun as a long-period Cepheid variable, 

 the period being that of the sun-spot cycle, but the range of 

 variation being very small. 



C. D. Perrine {Astroph. Journ. 41, p. 307), has also ad- 

 vanced a tentative explanation, assuming the stars to be 

 binary systems. He shows that the orbits are of small eccen- 

 tricity and small dimensions and that the mass of the secondary 

 is small, and suggests that the variation in the light is caused 

 mainly by variations in the light of the secondary, due to the 

 disturbance in its orbit near periastron. 



Orbits of Eclipsing Binaries. — An important volume on 

 this subject has recently appeared {A Study of the Orbits of 

 Eclipsing Binaries, by Harlow Shapley, Contrib. from Princeton 

 Univ. Obs. No. 3). Within the last few years new methods for 

 the computation of the orbit of an eclipsing binary from the 

 observations of its brightness have been introduced and de- 

 veloped by Prof. H. N. Russell and Dr. Shapley. In this 

 publication a brief summary of the theory underlying these 

 methods is given, together with some applications of them to 

 stars typical of the various classes which arise, and also the 

 results of the treatment of the observations of ninety eclipsing 

 variables. In this treatment, practically the whole of the 

 observational material available — to which Dr. Shapley 

 himself has largely contributed — has been discussed. In 



