On the Seventh Section of Bessel's Astronomical Observations. 4 15 



have been erupted from beneath the granite, and intermixed 

 with it on ihe surface ? If the latter, we may well conceive 

 that fragments brought down by the rivers might be washed 

 by the tides and currents as far west as Boulogne. 



Yours truly, 

 Torrington Square, Dec. 10, 1824. RoBT. BAKEWELL. 



P.S. — M. Dutertre had other specimens of a different cha- 

 racter, of which the volcanic origin was more problematical : 

 one semivitreous, containing globules of metallic tin. He had 

 also a very large deep yellow topaz found on the shore. 



LXXIII. Introduction to the Seventh Section of Bessel's 

 Astronomical Observations. 

 [Concluded from p. 349.] 



MESSRS. Rosenberger and Scherck have found the proba- 

 ble error of observation from very numerous comparisons, 

 = 1"-54?1, which determination may be assumed to belong to the 

 zenith distance 45°; applying to this determination the in- 

 crease of the probable error depending on the zenith distance, 

 which has been given in the 7th article, the probable errors of 

 an observation with Cary's circle will be for the zenith distances 

 0° 45° 60° 65° 70° 75° 80° 85° 

 = + l"-517; l''-541; l'-555; l"-5(>2; l"-585; 1''655; 1'775; 2""286. 

 A former determination gave the probable error of a mean 

 of four observations of a. Ursa? Minoris in zenith distance 

 36° = 0'"6845, likewise independent of the errors of division; 

 according to the present determination, it would be = 0"'77: 

 the difference may be accounted for by the uncertainty of the 

 error of collimation involved in the second, and perhaps like- 

 wise by the greater care taken in observing the pole-star. 

 Besides these contingent errors of observation, every zenith 

 distance has the error arising from the peculiar error of the 

 individual divisions on which they depend : this later one 

 might have been entirely avoided, if these divisions had been 

 determined directly by my method, as I first intended, but 

 was prevented from doing by other business and by the near 

 prospect of obtaining Reichenbach's circle. The probable 

 quantity of the remaining error of division is found by the 

 zenith distances of the above-named 38 stars, measured in both 

 positions of the instrument, to be = l"-004; so that the pro- 

 bable error of the mean of an observation made in the same 



position of the instrument, is = s/ \ (l"-004) 2 + £ \ \ By this 



formula, the probable errors of the single determinations and 

 the most probable declinations have been computed as follow: 



■j. Auriga: 



