OF ARTS AND SCIENCES. 197 



The average rule ia first-class observatories is about this : — 



Stars observed en masse, or by zones, should be twice observed, three 

 times if the two observations disagree much. 



Stars observed for ordinary catalogues of objects, interesting as bright 

 or having proper motions, should ha four times observed, twice in each 

 of two opposite positions of the instrument. Stars requiring special 

 accuracy, not of the very first order, should be observed eight or ten 

 times. For fundamental determinations, it is not so much the number 

 of single observations that is important, as the number of separate 

 determinations (in different years) of eight or ten observations apiece ; 

 and also the number of variations in the position of the instrument 

 and its parts. 



For some months I have been bringing together all available material 

 for a catalogue of latitude stars for the United States Enjiineers. Tal- 

 cott's method puts a heavy strain on the catalogue, as it is very simple, 

 and easily made accurate with a good instrument, but employs quite 

 small and ill-known stars, for want of well-proportioned pairs enough 

 at any given time and place. 



Hence the British Association Catalogue, which contains stars 

 enough, was found almost from the day of its publication to be far 

 from precise. It could hardly be used with a two-inch zenith telescope 

 in indifferent condition, on the boundary between this country and 

 Mexico. 



In supplying star-places for this purpose from time to time, the 

 gradual increase of material has been greatly encouraging. But, on 

 the other hand, there is a troublesome want of uniformity in modern 

 star-catalogues. To say nothing of unreduced observations, it often 

 happens that there is much carelessness in settling the zero-points of the 

 instrument, and the clock- and other corrections ; so that the errors in 

 these are often larger than those of observations. Moreover, the star- 

 places used as fundamental are often less accurate than those obtained 

 by their help ; and the various flexures and errors of division, and the 

 variations, in clock-rate, are neglected or ill-determined. 



As a result, it becomes very difficult to assign the proper weights to 

 the separate determinations, and to settle upon those w^hich are to be 

 excluded. The simplest rule would be to exclude certain doubtful 

 catalogues altogether ; but unfortunately they are sometimes indispen- 

 sable, where the star is wanted, and no other authority of the same 

 epoch is at hand. 



