October 20, 1911] 



SCIENCE 



527 



dueed with the sine flexure determined from the 

 collimators. The corrections A0 = + 0".2o for 

 circle A and + 0".14 for circle B, based on direct 

 observations of circumpolars observed at both cul- 

 minations have been applied. 

 The Alt-azimuth Instrument of the V. S. Naval 



Observatory : F. B. Littell. 



The telescope of this instrument is of 5 inches 

 aperture and 50 inches focal length. The gradu- 

 ated circle is 23 inches in diameter, read by 4 

 microscopes magnifying 30 times. Two levels are 

 used on the alidade. For stars at more than 10° 

 zenith distance, the Pulkova vertical circle method 

 has been used in observing the declinations of 

 stars. For stars at less than 10° zenith distance, 

 the instrument has been used in the meridian, and 

 a double observation has been secured by bisecting 

 with the micrometer at a side thread before and 

 after reversal about the vertical axis. 



The essential differences between the use of this 

 instrument for declinations and that of a meridian 

 circle are: (1) that the double zenith distance of 

 the object is measured; (2) that each observation 

 is complete in itself and does not depend on the 

 stability of the instrument except for the short 

 interval, 5 or 6 minutes, covered by the observa- 

 tion; (3) that cosine flexures are immediately 

 eliminated; (4) that spirit levels are substituted 

 for the mercury horizon in determining the refer- 

 ence point of the circle, and (5) that in the 

 Pulkova method a time observation is substituted 

 for a micrometer observation. This latter method, 

 however, may be adapted to meridian circle work 

 by slightly inclining the zenith distance thread, 

 and it has been thus used by some observers. 



In general the differences seem to be favorable 

 to the use of the vertical circle and it is probably 

 the best type of instrument yet devised for obtain- 

 ing fundamental declinations. The use of the 

 micrometer to obtain additional bisections may 

 lead to the securing of greater accuracy. 



About 5,000 observations made by the writer 

 from December, 1903, to July, 1907, have been 

 reduced. One of the levels used in the early part 

 of the work did not give satisfactory results, but 

 for the last 3 years of the work, the levels per- 

 formed in a fairly satisfactory manner. During 

 that period the element of probable error due to 

 levels (mean of two) in a declination resulting 

 from a single observation was 0".035. 



The division errors were not measured, but their 

 effect was reduced by reading on two divisions 

 under each microscope and by shifting the circle. 



The probable error of a single observed declina- 

 tion due to division error is 0".15 



The sine flexure was measured in 1910 after the 

 installation of horizontal collimators. From 14 

 sets of measures the value + 0".79 ± 0".063 has 

 been adopted. 



The variation of latitude as given by the Inter- 

 national Geodetic Commission has been applied. 



From 1,190 observations on 134 ciroumpola* 

 stars at both culminations, the following correc- 

 tion to the latitude, A<p, and correction to the 

 refraction constant, AB, were obtained: 

 A0 = + 0".50 ± 0".090, 

 AB=— 0".22 ± 0".056. 



As the separation of A0 and AB was not well 

 determined, it was assumed that AB = and 

 the resulting value A0 = -f 0".15 ± 0".017 was 

 adopted. 



The accidental error of a single observation, not 

 including division error, is shown in the following 

 table : 



On account of the construction of the instru- 

 ment, it is diffieult to observe stars near the zenith 

 and this probably accounts for the larger probable 

 error of stars from 0° to 10° as compared with 

 those from 10° to 50° zenith distance. 



The comparisons of the observed declinations 

 with those of Newcomb (N) and those of Boss 

 (B) after the application of the corrections for 



