I9I5.] OBSERVATION OF ECLIPSING VARIABLES. 53 



primary minimum, Shapley has recently explained this asymmetry 

 as a Periastron effect — an increased brightness of the stars when 

 nearest each other. There is another possible explanation which I 

 shall consider in connection with the third star. 



In 1909 the series of 18,384 measures of s Draconis was com- 

 pleted.- Evidence of a very shallow secondary minimum, only six 

 or seven hundredths of a magnitude in depth, displaced slightly 

 toward the following primary minimum, was clearly found. Ellip- 

 ticity of figure and exchange of radiation were again demonstrated. 

 These two effects have been abundantly verified in other cases. 



The theoretical representation of the observations of both RT 

 Persei and 2 Draconis just at the beginning and end of eclipse is a 

 little unsatisfactory — the theoretical curve starts to drop more 

 rapidly than the observed curve. This is possibly evidence that the 

 stars are darkened toward the limb — like the sun. 



It was hoped that the observations of RV Ophiuchi would not 

 only demonstrate the existence of darkening toward the limb but 

 would determine approximately the degree of darkening. But 

 darkening seems to be very elusive sometimes. The observed 

 secondary minimum is barely deep enough for the uniform solution, 

 and a darkened solution requires a still greater depth. The well- 

 determined primary minimum is strongly asymmetrical. The inter- 

 radiation effect is about the same as in the other two stars, while 

 the ellipticity is much greater. This latter fact when considered 

 with the anticipation that the ellipticity effect would in this case be 

 too small to detect — on account of the large distance between the 

 component stars — is somewhat disconcerting. Then too there are 

 several well-defined hollows in the curve — a conspicuous one just 

 after primary and another just after secondary — and a long stretch 

 before secondary when the star is much brighter than before pri- 

 mary. The only satisfying method of solving the curve seemed, 

 therefore, to be a least square solution of the whole curve — a pro- 

 cedure hitherto avoided. The final theoretical curve, representing 

 our present knowledge of the causes of light variation of these stars, 

 is the result of the solution of forty-two equations with seven 



2 See Contributions from the Princeton Univ. Obs., No. 2. 



