150 COSMOS. 



The Arabs had the great merit of showing how tables might 

 be gradually amended by a comparison with observations. 

 Ulugh Beg's catalogue of the stars, originally written in 

 Persian, was entirely completed from original observations 

 made in the Gymnasium at Samarcand, with the exception 

 of a portion of the southern stars enumerated by Ptolemy, 14 

 and not visible in 39 52' lat. (?) It contains only 1019 

 positions of stars, which are reduced to the year 1437. A 

 subsequent commentary gives 300 other stars, observed by 

 Abu-Bekri Altizini in 1533. Thus we pass from Arabs, 

 Persians, and Moguls, to the great epoch of Copernicus, and 

 nearly to that of Tycho Brahe. 



The extension of navigation in the tropical seas, and in 

 high southern latitudes, has, since the beginning of the six- 

 teenth century, exerted a powerful influence on the gradual 

 extension of our knowledge of the firmament, though in 

 a less degree than that effected a century later by the ap- 



14 In my investigations on the relative value of astronomical 

 determinations of position in Central Asia (Asie centrale^ 

 t. iii. pp. 581-596), I have given the latitudes of Samarcand 

 and Bokhara according to the different Arabic and Persian 

 MSS. contained in the Paris Library. I have shown that the 

 former is probably more than 39 52', whilst most of the best 

 manuscripts of Ulugh Beg give 39 37', and the Kitab al-athual 

 of Alfares, and the Kanum of Albyruni give 40. I would 

 again draw attention to the importance, in a geographical no 

 less than an astronomical point of view, of determining the 

 longitude and latitude of Samarcand by new and trustworthy 

 observations. Burnes's Travels have made us acquainted with 

 the latitude of Bokhara, as obtained from observations of 

 culmination of stars; which gave 39 43' 41". There is there- 

 fore only an error of from 7 to 8 minutes in the two fine Persian 

 and Arabic MSS. (Nos. 164 and 2460) of the Paris Library. 

 Major Rennell, whose combinations are generally so suc- 

 cessful, made an error of about 1 9' in determining the latitude 

 of Bokhara. (Humboldt, Asie cmtrale, t. iii. p. 592, and 

 Sedillot in the ProUgomen.es d Oloug-Beg, pp. cxxiii. cxxv.) 



