The Rev. Dr. Robinson on the Constant of Refraction. 



193 



mined In a few minutes at the very time of observation, this is of no conse- 

 quence. 



The index correction of this instrument is deduced from observations of 

 Polaris. The star is observed five times near the meridian, and reduced to it by 

 a table computed from the formula, 



rfo =: A 4- A* X tang 8 X sin 1", 

 where, 



sin X cos . 8 



sinl' 



X versme p. 



These, compared with the mean places of Bessel brought up by the constants of 

 Baily's catalogue (for the time) and corrected for the term 2 3) , give the approxi- 

 mate correction. When conjugate observations (above and below the pole) can 

 be obtained, the mean is independent of any error of the assumed declinations ; 

 but at other times the difference between Bessel's place and my own is applied as 

 a correction.* As long as the difference of individual results is manifestly mere 

 error of observation, it is assumed that the mean is the index correction during 

 that period. Its changes are slow, having an annual period, and a given extent 

 of variation during the eight years that the instrument has been used. The 

 most probable cause of this appears to be some influence of temperature on the 

 hill, for the transit instrument, and a telescopic meridian mark about fifty feet 

 south, suffer analogous variations. As the fact is curious, I annex a table of the 

 index corrections during 1839, which will also show that no error can arise from 

 its occurrence.f 



• Equal to -I- 0".21 by 700 conjugate observations, 

 t 1838, Dec. 18, , _ ^ ^ 

 1839, Feb. 24, \ /^^ 

 April 7, 

 ,, 24, 

 May 16, , _ 3 g^ 

 Junes, |_i63 



OK > 



' ] — 4.75 

 ] — 5.20 

 t — 4.19 



e '''}-0.14 



Sept. 11, -J ggg 



0'='-i8'|_a49 



1840, Feb. 28, 



VOL. XIX. 



80 obs. 



40 



50 



55 

 115 



10 



75 



45 

 105 



25 



2c 



