OF ARTS AND SCIENCES. 



175 



These systematic discordances, especially in right ascension, are so 

 alarmingly large that, unless they can be reconciled, the heliometer ob- 

 servations are comparatively worthless. Mr. Gill, therefore, proposed 

 a second list of 12 stars, one half comparatively bright and the other 

 half faint. The observations of these stars are now completed, but 

 the only series yet at hand, are those of Konigsberg and Harvard 

 College. Here the discordance is very large, and varies with the mag- 

 nitude of the star observed. Professor Pickering, the Director of 

 Harvard College Observatory, early in this investigation, proposed 

 the artificial reduction of the magnitude of the bright stars by holding 

 circular diaphragms of varying diameters in front of the object-glass 

 of the telescope. By alternating between bright and faint images of 

 the same star, on different groups of the transit threads, the personal 

 equation between bright and faint stars can be found. This plan was 

 followed in the investigation at Harvard College Observatory, at 

 Leiden, and probably at some other observatories. At Harvard Col- 

 lege Observatory, also, a sensible difference was found between results 

 obtained with bright and faint fields of the telescope, this difference 

 varying with the magnitude of the star. 



A similar investigation is now being made in another class of obser- 

 vations, viz. the measurement of the position angle and distance of 

 double stars with the filar-micrometer. The range of systematic dis- 

 cordances between the measures of different observers is of course here 

 far less than will always be found in the determination of position in 

 space, for such observations are entirely relative in their character ; but 

 the outstanding errors are still so large as to demand a special in- 

 vestigation. Even with observers of skill and long experience, such 

 as Struve, Hall, Dembowski, Burnham, and Stone, there are residual 

 errors in the measurements of the same components far exceeding the 

 limits indicated by the magnitude of the probable error of any single 

 observer. 



Finally, it is even an open question whether any real advance has 





