60 



SCIENCE 



[N. S. Vol. XLII. No. 1071 



the majority of cases. As practically every- 

 thing else depends upon this guess, the further 

 discussion is for many stars only an indica- 

 tion of "what may be true. This limitation is 

 well enough understood by the author, but 

 there is danger that a casual reader, on seeing 

 the array of tabular results, will infer that aU 

 of the numbers rest upon a secure observa- 

 tional basis. 



The title of the work is something of a mis- 

 nomer, as in the table of so-called orbits we 

 find for each system, in addition to the char- 

 acteristics of the light variation and relative 

 paths of the bodies, the radius of each in terms 

 of the distance between them; the proportion 

 of light given by each component; their den- 

 sities ; constants giving the approximate shapes 

 of the ellipsoids produced by tidal action; and 

 finally the hypothetical dimensions and paral- 

 lax of each system, based upon certain assump- 

 tions. There is omitted, however, the quantity 

 which may be rigorously derived, namely the 

 mean density of the system, which does not de- 

 pend in any way upon an assumption as to the 

 total mass, or how it is divided between the two 

 bodies. Russell and Shapley were the first 

 to take into account generally the probable 

 darkening of stellar disks toward the limb, 

 such as exists in the case of the sun, and 

 throughout this work the results are given as 

 based upon the alternative or limiting eases 

 of apparently uniform disks, and of bodies 

 completely dark at the apparent edge. The 

 convenience of the description leads to such 

 anomalous terms as a " uniform orbit " and a 

 " darkened density." 



It has been fairly well known that the fainter 

 component of an eclipsing system is likely to 

 be the larger, but Shapley now points out that 

 this is due to the manner of observational 

 selection. The most usual cases actually 

 existing consist of systems with small faint 

 companions, the apparent preponderance of the 

 other type being due to the fact that large 

 companions produce greater variation in light, 

 and hence are more readily discovered. 



Thanks to the work of the Harvard Obser- 

 vatory, the spectra of most of the eclipsing 

 stars have now been classified, and one of the 

 most interesting results comes in the relation 



between spectral class and density. Shapley 

 finds that the white stars of the Orion and 

 Sirian type range from slightly less than the 

 solar density to 1/100 of this amount; but 

 among the yellow stars, having the same 

 spectrum and presumably the same surface 

 conditions as the sun, there are densities rang- 

 ing from three times larger than the sun's to 

 only 1/100,000 on the solar standard. This 

 extreme range in eight stars needs to be sup- 

 plemented by more data, but so far as it goes 

 the evidence supports Russell's contention that 

 there are two classes of yellow and red stars, 

 the so-called giants and dwarfs, the latter be- 

 ing much more condensed than the former. 



The use of hypothetical parallaxes may not 

 appeal to some students of exact science, but, 

 after all, the assumption that a double system 

 has twice the sun's mass, combined with a rea- 

 sonable estimate of the surface intensity of a 

 star with known spectrum, gives valuable in- 

 formation. Most of these eclipsing systems 

 are at such great distances that direct meas- 

 ures of parallax are absolutely hopeless; and 

 since the computed distance varies only as the 

 cube root of the assumed mass, good estimates 

 of probable average distance are obtained. The 

 general inference from the eclipsing binaries 

 confirms the conventional view that the uni- 

 verse of stars is more extended in the direction 

 of the Milky "Way than at right angles to 

 that plane. 



The general discussion of results covers 

 twenty quarto pages, but a summary on a 

 single page would have been welcome. In fact, 

 there is some diificulty in the reader's getting' 

 aU out of the work that is actually there; for 

 instance, there is no discussion showing what 

 is a normal or average eclipsing binary. 

 Using Shapley's data, the reviewer finds it 

 probable that the preponderant type of eclips- 

 ing system consists of two bodies with a dis- 

 tance between centers of about five times their 

 average radius, a period of four days, and a 

 mean density 1/20 that of the sun. 



In view of the present rate of discovery of 

 eclipsing stars, and the prospect of many more 

 additions from the list of spectroscopic bi- 

 naries, the new methods of treating the obser- 

 vational data are especially welcome, and all 



