108 



Tims it ap|>ears that observations of circumpolar stars are 

 not adapted for obtaining extreme accuracy* and that tli^ 

 quantity of mean refraction at 45" so determined cannot rea- 

 sonably be depended on to less than a quarter of a second. 



The direct experiment for determining the refractive force 

 of ait may be made independently of the divisions of an in- 

 strument. The whole quantity of refraction is ascertained, 

 instead of the differences of refractions as in circumpolar 

 stars. There are also other sources of accuracy by which the 

 result may berendered very exact. 



For the above reasons, the determination -7— n;= 57",82 or 



' ««. 1" ' 



the mean refraction at 45" (bar. 29, t>0 and therm. 60) = 57,67 



appear to me more to be relied on. 



16. In deducing the above value of ^^^-^ from the observations 

 of circumpolar stars, 1 only used such stars as were less than 

 80" from the zenith when below the pole. 



It is well known to those conversant in observations made 

 with good instruments that near the horizon an irregularity 

 in refraction hitherto unexplained shews itself. This com- 

 mencing even at less zenith distances than 80°, is at first very 

 small, but increases to a very considerable irregularity as we 

 approach the horizon. 



The bright star Capella being within the limits of this irre- 

 gularity has not been used for the co-latitude. A considerable 

 number of observations of this star below the pole have how- 

 ever beenvmade by me, which may serve for two purposes. 



