132 



SCIENCE 



[N. S. Vol. XLIX. No. 1258 



ties of these stars with estimates of their 

 masses and surface brightness (which can be 

 made with a fair degree of confidence), it is 

 possible to estimate their parallaxes, and study 

 their distribution in space. The results of 

 Shapley and the speaker show that these stars 

 are distributed through a region nearly sym- 

 metrical about the galactic plane. The great 

 majority of them lie within two or three hun- 

 dred parsecs of this plane, but they extend 

 along the plane in all directions to a distance 

 of at least 1,500 parsecs, with no sign that our 

 observation has reached the limit. 



The statement that the brightness of an 

 eclipsing variable is constant in the intervals 

 between eclipses is only approximately true. 

 In many instances — notably in Beta Lyrse — 

 there are definite maxima half way between 

 the eclipses, indicating that the stars are 

 elongated into ellipsoids by their mutual tidal 

 influence. The amount of this ellipticity de- 

 creases rapidly for pairs in which the com- 

 IK)nents are more widely separated, following 

 very closely the law predicted theoretically by 

 Sir George Darwin. 



Again, there is often a "radiation effect" 

 showing that the companion keeps the same 

 face always toward the primary, and that this 

 face, being heated by the later, is brighter than 

 the opiKJsite side of the secondary (as was ob- 

 served by Dugan in ET Persei, and by Steb- 

 bins in Algol). In a few cases, when the ob- 

 servations are very precise, it has been possible 

 to show, from the form of the light curve, 

 that the disk of the principal star is not uni- 

 formly luminous, but is brighter at the center 

 than at the limb, like that of the sun. 



The theory of eclipsing variables is there- 

 fore now in a satisfactory state: but there is 

 much which still remains to do. Dozens of 

 stars are still to be observed, and the light 

 curves of many more require more accurate 

 determination. The important task of com- 

 paring the visual and photographic light- 

 curves of the same stars is only just begun; 

 and there is a wide field for the physical photo- 

 meter among the recently discovered systems 

 of small range. Theoretically, many problems 



of interest await solution, especially the ex- 

 planation of the small but indubitable, and 

 perplexingly complicated, changes in period 

 which occur in most of the systems which 

 have long been observed. If these changes 

 can be definitely referred in some cases to 

 motion of the line of apsides of the orbit, 

 it should be possible to obtain information 

 about the degree to which the central density 

 of the stars in question exceeds their mean 

 density — a matter about which we now know 

 nothing. There are also curious discrepancies 

 between the times of minima, as observed with 

 light of different colors, which are very puz- 

 zling. Enough is already known, however, to 

 furnish a direct observational proof that the 

 velocities of light of different colors in empty 

 space can not differ by more than a very few 

 meters per second. 



Spectrographic observation of all accessible 

 eclipsing variables for radial velocity is ur- 

 gently to be desired, for this affords the only 

 known way of adding to the few cases in 

 which we can determine the actual diameters 

 of the stars. Such matters as the precise 

 determination of spectral type, especially of 

 the faint companions during total eclipses, 

 and the study of absolute mag-nitudes by 

 spectroscopic means also deserve attention. 



The Cepheid variables, which come next in 

 order in our survey, present one of the most 

 attractive and difficult problems of present day 

 astronomy. There are few other fields in 

 which we know so many facts, and can ex- 

 plain so few. The mass of information is so 

 extensive that it must be rather summarily 

 presented. Observation shows that: 



1. The Cepheids show a regular variation, 

 For each star the period is constant, and the 

 light curve repeats itself with remarkable 

 regularity. (The small deviations from this 

 rule suspected by Shapley and others, though 

 doubtless real, are insignificant in comparison 

 with the irregularities that occur among long- 

 period variables). 



2. The i)eriods range from three hours to 

 more- than a hundred days, but show two well- 

 marked maxima of frequency at about twelve 



