880 



SCIENCE 



[N. S. Vol. XXXII. No. 833 



of Eros, at Arequipa, during the opposition of 

 1903. The instrument employed was one of the 

 Rumford photometers devised by Professor E. C. 

 Pickering and described by Mr. J. A. Parkhurst 

 in the Astrophysical Journal, 13, 249. Observa- 

 tions were carried on from March 30 to August 

 19, 1910. On many nights the observations were 

 made to cover the full double period of the light 

 changes. The results give a mean period of 5.270 

 hours, for the double period, or, since the two 

 halves of the curve appear to be precisely equal, a 

 period of 2.635 hours. The period would change 

 by a small amount during this time, but the above 

 period satisfies well all the observations. The 

 range of variation appeared to vary from five 

 tenths to eight tenths of a magnitude, the mean 

 being about six tenths. The brightness of Eros 

 during these observations varied from about mag- 

 nitude 11 to magnitude 13.5. 



The Division Errors of the Nine-inch Transit 

 Circle of the Naval Observatory, and the Effect 

 upon the Division Errors of Refilling the Divi- 

 sions: W. S. ElCHEI^EEGEE. 

 Two independent determinations of the division 

 errors were made by Professor Eichelberger for 

 each of the 10,800 graduations. A comparison of 

 these indicates a probable error of 0".04 for the 

 2' marks. For the degree marks the probable 

 error is 0".018. These determinations were made 

 after the graduations had been refilled; to de- 

 termine whether they would apply to the 20,000 

 observations previously made, the observed correc- 

 tions to the ephemeris declinations of the sun 

 were collected and it was shown that the recently 

 determined division errors were in all probability 

 applicable. A similar result was obtained by dis- 

 cussing the earlier observations of 400 zodiacal 

 stars, and computing the probable error of a 

 declination (1) when the same division had been 

 used for the same star, and (2) when different 

 divisions had been used. These became practically 

 the same only after the recently determined divi- 

 sion errors had been applied, and indicated that 

 these errors could have been only slightly altered, 

 if at all, by the refilling of the marks. Professor 

 Eichelberger finds that this circle, like those of 

 the 6-inch transit-circle of the Naval Observatory, 

 shows a periodic error in the 2' marks that re- 

 peats itself every 10'. 



The Eclipsing Variable u Herculis: Fbank 

 SCHLESINGEK and ROBEBT H. Bakeb. 

 In the usual case presented by spectroscopic 

 binaries the masses of the two components, and 



therefore their densities, remain indeterminate. 

 This is because (1) the inclination of the orbit 

 can not be computed from measures of the radial 

 velocities alone; and (2) it is necessary to know 

 the ratio of the two masses involved before either 

 becomes determinate. Both of these obstacles are 

 removed in the case of such of the eclipsing va- 

 riables for which both spectra appear upon the 

 plates. Among the stars readily accessible to 

 present-day instruments there are only three that 

 fulfill these conditions: p Lyrae, V Puppis and 

 u Herculis, and only for the last have the neces- 

 sary observations been made. A discussion of 

 these data shows that the two stars must be very 

 nearly the same size, but that one has a density 

 2.6 times that of the other and is about 2.5 times 

 as bright. These results have an important bear- 

 ing on questions of double-star evolution. It was 

 also shown that if the parallax were accurately 

 determined, it would be possible to state whether 

 the surface brightness of these helium stars is 

 greater or less than that of our sun, a question 

 that has been the subject of considerable debate 

 in recent years. This paper will soon appear in 

 volume II. of the Publications of the Allegheny 

 Observatory. 

 The Rotation of the Sun for Different Substances 



in the Reversing Layer: Fbank Schlesikgeb. 



A series of spectrograms were secured in the 

 fall of 1909 with the new Porter spectrograph of 

 the Allegheny Observatory. This is fed by a ver- 

 tical coelostat that forms part of the Keeler re- 

 flecting telescope. The dispersing piece is a large 

 Michelson grating with 500 lines to the milli- 

 meter. The photographs were taken in the third 

 order and yield a linear dispersion of 0.82 milli- 

 meter to the Angstrom. Thirty-seven good lines 

 from \ 4059 to X 4147 were selected and the dis- 

 placements due to rotation measured on eighteen 

 plates. No difference from the mean greater than 

 2.4 per cent, was found and a discussion of the 

 residuals made it very probable that differences 

 due to rotation must be much smaller than this. 

 The results showed no systematic tendency for the 

 various lines due to the same substance. There 

 seemed to be a small systematic increase of dis- 

 placement with increase of wave-length, and the 

 recent observations at Mount Wilson seem to 

 show the same tendency. 

 TJie Orbit and Spectrum of o Persei: Fbank C. 



Jordan. (Bead by Professor Schlesinger. ) 



The spectrum of the principal star is of the 

 type B 2, and that of the secondary seems to be 



