of Bessel's Astronomical Observations. 441 



on the 11th of May 1820 a convenient form. This mark is a 

 parallelepiped of granite, which has towards the observatory a 

 face 24 inches high and 18 inches broad : this face is furnished 

 with 5 sets of black rectangles on white ground, of 4'' breadth 

 and 2"*6 in height, which are contiguous in the direction of 

 their diagonal lines ; so that the centres of two contiguous 

 rectangles differ 4" in azimuth. The first set begins at the 

 (apparent) upper edge of the mark in the middle of its 

 breadth, and determines the azimuths — 4"— 8"; the se- 

 cond set has five rectangles for 7", 3", — 1 ", — 5", — 9" ; the 

 third lias four rectangles for 6". 2", — 2", — 6'' ; the fourth has 

 five rectangles for 9", 5", 1", — 3'' — 7"; the fifth has three 

 rectangles for 8'', 4", 0. By this construction the single se- 

 conds of the deviations of the. wire are immediately perceived. 

 When the atmosphere is at rest and the illumination good, 

 one may estimate parts of a second ; and as one may wait for 

 these favourable circumstances, it may be assumed, that the 

 deviations of the wire from the mark, as given when re- 

 versing the instrument, or in the last column of the observa- 

 tions, are correct to ,,- 2 or 0''*3. Notwithstanding this cer- 

 tainty in observing the deviations, I have found that this 

 method does not perfectly determine the error of collimation ; 

 by reversing the instrument I found from 11th June to 12th 

 October, 



These observations would appear to prove a sensible al- 

 teration, which, however, as I shall show below, did not take 

 place. 



The other method of determining the error of collimation 

 by the pole star cannot be employed with instruments of this 

 kind, as the reversing cannot be effected quick enough to ob- 

 tain a satisfactory result by one passage even of a Urs. min. : 

 and in combining different passages one would run the 

 risk of transferring small alterations of the pillars, &c. to the 

 error of collimation. The same cause which in the former 

 method causes the uncertainty, might besides be apprehended 

 in both the latter cases. 



An accurate determination of the collimation being, how- 

 ever, necessary, I have employed a third method, which has 

 the advantage of being deduced from repeated observations, 

 and at the same time considerably diminishing the contingent 

 errors. It consists in employing for determining the colli- 



Vol. 63. No. 311. June 1824. 3 K mation, 



