30 Further Remarks on Dr. Bradley’s Theorem 
It is well known that showers of rain often proceed in zones 
along the surface of the earth, in various directions; and should 
this have taken place across any part of the are P Q, a short 
time previous to the observation being made, the evaporation 
that follows, by producing a greater degree of cold, will cause 
a considerable augmentation in the refractive power of the eir- 
cumambient air. 
A large surface of sand, situated any where along the arc P Q 
will, by absorbing a considerable portion of heat from the sun’s 
rays, and afterwards continuing to restore it to the atmosphere, 
longer than the neighbouring soils that have absorbed less, by 
keeping up the rarefaction, bend the ray out of its course, in 
an opposite direction to that of the former cause. 
A large wood or a marsh will also be the cause of deviation ~ 
in the trajectory of the ray: and all these actions on its course 
will be the stronger the nearer it lies to the horizon, or in other 
words the less the altitude of the object. 
Mr. Delambre, in the excellent ‘T'reatise of Astronomy which 
he has lately published, makes the following remarks on the re- 
fraction at low altitudes. 
*¢ It follows from all this that Simpson’s Formula and Brad- 
ley’s Table will not agree with the observations. There is an 
error of about 8” at 821°, and no known table will agree with 
them. That of the Board of Longitude computed from Laplace’s 
formula, and on the value which | found for the constant quan- 
tity «, (the refraction at 45° which is p of our formula) agrees 
better, but the error is still from 2” to 3”, 
“¢ J have already spoken of the uncertainty of observations of 
refraction in the neighbourhood of the horizon. I have re- 
marked, from one day to the next, and in circumstances which 
were the same in appearance, that the refraction would vary from 
15” to 20”, without my being able to assign any cause; but the 
variations are still more sensible in the horizon, as we may judge 
from the following statement : 
Computed Observed ; Therm, 
zenith distance.| zenith distance.} Refraction, | Batom.| of 80. 
90°44’ 5-4’'| 90° 8’36:8”| 357 26°8' |27 6-0} 16-64 
90 33 39:2 | 90 2 43-6 30 55:6 |27 6:0} 16:64 
90.33 --9°:1.} 90. 2 12:7 30 57> 27 7°4| 20-64 
90 33 13°0 | 90 1 53 31 19°6 |27 6:5) 20:32 
90 27 506 | 89 54 36 33 14-6 |27 8-1) 11-84 
90 39 24:5 | 90 4 27 34 57:4 [27 6:3} 19:20 
/ 
«© All these observations were made in June at sunrise. From 
the 
