33 ^ G'koou^^wgv's further Observations 
perature of the atmosphere ; and from these causes may be 
computed the effect they should have on a ray of light passing 
through the same : yet we must resort to observation, for the 
verification of the theory ; and reduce the quantity so found, 
to the most simple and convenient formula. I shall proceed 
to deduce, from this course of observations, such formulae as 
will appear to result, for the computation of the refraction ; 
from the zenith, to the lowest star which I have observed : 
these may be considered as sufficient for the observation of the 
sun at the winter solstice, in high latitudes ; since those of the 
moon, from its great parallax, and the planets from their 
general invisibility, would probably not be attempted. Never- 
theless, it is to be wished, as a matter of curiosity, or from 
which some useful deductions might be made, that in those 
Observatories, wherein from their elevated situations it might 
be practicable, the true quantity of refraction should be 
ascertained to the horizon. 
Of all the formulae for computing the mean refraction, that 
proposed and used by Dr. Bradley, is the most convenient 
and applicable for the practical astronomer. But as it is now 
acknowdedged, that the numbers he had assumed for the co- 
efficient of r (the refraction ;) and of x (the quantity at 45®) 
were too small : their real values will appear to be the mean 
of several arcs, and such as I now propose to be adopted. I 
have found, that the same formula will serve to 87® of zenith 
distance ; possibly this might not happen in low situations, 
where the height of the vapours would form a greater angle 
with the horizon : yet in more elevated places, we may rea- 
sonably suppose, that a general formula might be carried 
nearly to the horizon. 
J 
