The preceding 65 observations give the mean correction 

 = - 0",49. This would give J:^, == 57",74 and the ref. at 



45" =i: 5J",58 very nearly the same as the; French tables, but 

 this exactness cannot be depended on, even if we supposed 

 the irregularity of refraction to disappear in the mean, be- 

 cause the zenith distances of Capella above and below the 

 pole may be affected by errors of division. If we suppose 

 the co-laiitude exact, and take the error of the mean of the 

 six microscopes in each position of Capella = 0",5 and also 

 take the error of refraction arising from using the mean be- 

 tween uniform temperature and uniform density = 0",25. 

 The above correction may become = — (0"49i-l,00+ 0",25) 

 =-- 1",74 



or it may become + 0,76. 

 •The first will make the ref. at 45' = 57",37 

 the second - - =57 ,79 



These are probably two limits. 



•^ 'Limits of Refraction. — Observed Refractions of a- L'yi-iB 



below the Pole. 



17. It has been stated' in art. 4. tliaCtlie quantity or at- 

 mospherical refraction is less than would result from an uni- 

 form temperature in the atmosphere and greater than what 

 would result from a density decreasing uniformly. 



