238 CARNEGIE INSTITUTION OF WASHINGTON. 



(5) A comparison with a system of mean parallaxes based on the di- 

 rectly measured parallaxes of all observers indicates an error of +0''002. 



From the five different methods used, Mr. van Maanen finds that the 

 systematic error of the Mount Wilson trigonometrical parallaxes does 

 not appear to exceed 0''002. For the last comparison it was thought 

 worth while to derive the systematic deviations of other series of paral- 

 laxes as well. A discussion of all the available material led him to 

 conclude that while in most of the modern series of parallaxes the sys- 

 tematic deviations are small, they can not be neglected when the highest 

 accuracy is required. The method followed is, in brief, thus: All stars 

 for which the parallax, proper motion, magnitude, and spectral type are 

 known, 901 in all, with 1,583 observations of parallax, were made to fit 

 an exponential formula of the type proposed by Kapteyn. The con- 

 stants of the formula were determined for each spectral type separately, 

 using two different systems of weighting. The deviation from the 

 mean system thus established was then determined for each observer. 

 The results show that for practically all series of observations the 

 systematic de\'iation from the mean system is not a function of spec- 

 tral type or of proper motion, but in most cases is dependent on the 

 magnitude. The only exceptions are the Allegheny, Greenwich, 

 Mount Wilson, McCormick, and Yerkes parallaxes. The application 

 of these systematic deviations to the individual parallaxes of the stars 

 measured by more than one observer reduces the differences in 65 per 

 cent of the cases, while for 19 stars with six or more available observa- 

 vations their use improves the agreement in 72 per cent of the cases. 



The search for faint stars of large proper motion was continued. Al- 

 though 23 fields have been examined, for which the interval is 4 or 5 

 years, only two cases were found, both on parallax plates: near Boss 

 1182, a star of photographic magnitude 13.2 with an annual motion of 

 0''38, and near W Ursse Majoris a star of magnitude 14.1 with a motion 

 of 0''92. With this interval motions of half a second or more annually 

 can be detected on photographs taken at the 80-foot focus by simply 

 superposing the plates. As each plate of 15 minutes' exposure contains 

 from 35 to 200 stars, it is clear that the number of faint stars with mo- 

 tions exceeding 0''50 annually must be very small. 



A series of photographs duplicating some of the old parallax-plates 

 has been begun, from which it is hoped that it will be possible to derive 

 the parallactic motion, and possibly the stream-motion, of the stars 

 as faint as the twelfth magnitude. 



LUNAR PHOTOGRAPHY. 



Mr. Pease has photographed the moon at the 134-foot focus of the 

 100-inch telescope on 9 nights. The photographs of September 15, 

 1919, when combined stereoscopically with those of August 5, 1920, 

 which are at practically the same phase (21.5 days), show strong relief, 

 because of the displacement produced by libration. The small details 



