of the Principal Fixed Stars. 129 



less than \". The difficulty of avoiding so small an error, in the 

 value of A-, as 0",S5 will be presently adverted to. 



But it will be asked, Can it be possible that, in one place, 

 the constant of refraction, for the internal thermometer, should 

 be 58",01 ; and in another, for the external thermometer 

 57",33? The mean external thermometer at Konigsberg is 

 probably rather more than 5° lower than the interior. For 5" 

 the reduction to the interior is =^|- x 57",3 = 0",60. This 

 is very nearly the difference Mr. Groombridge found when he 

 determined the constants of I'efraction by the internal and ex- 

 ternal thermometers respectively. Thus the two constants 

 will be 



K ' 'cTh ' '57 'q^ r^°^ the external tliermometer. 

 A difference much greater than this may arise from the de- 

 viations of the heights of the barometer and thermometer, at 

 each place, from exact standards. Thus we have only a mean 

 discordance of 1" to account for. This might easily arise from 

 errors, diflficult to avoid, of the co-latitude in one place, and 

 of the determination of the Polar point on the circle, in the 

 other. On the supposition of these errors, and the above er- 

 rors of the constants of refraction, the differences would become 

 those in column D. 



But still the errors in the co-lat., and in the Polar point, 

 are more easy to be avoided than much greater than those 

 above supposed in the constants of refraction. Therefore, let 

 us suppose that the whole discordance is to be attributed to the 

 errors of the constants of refraction : then we shall have 



(27 tan. co-lat. + 27,85) 8^-=-59",69 

 or, taking the lat. of Konigsberg = 54° 48' 



8;t-=f£!^= _i",2 



47,02 



If we suppose this error equally divided between the two con- 

 stants, we shall have 58",32 for the constant at Dublin for the 

 internal thermometer ; and 57",08 for the constant at Konigs- 

 berg for the external thermometer : or probably 57''',68 for the 

 internal thermometer. On this hypothesis the differences would 

 be as in column E. 



The difference between 5S'',32 and 57",68 is less than might 

 arise from the deviations of the respective bai'ometers and ther- 

 mometers, from exact stjmdards. This appears to be a subject 

 that has not been at all attended to. The difTcrence of the 

 specific gravities of the mercury in the barometers, and other 

 causes, might occasion a difleience of heiglit : want of exact- 

 ness in the scales of the tliermometer might also occasion other 



Vol. 61, No. 298. /r6. 1823. 11 differences. 



