on Horizontal Refractions. 33 
time, and represents the proportion of the inverted to the erect 
object, as near as I could take it by the eye, the former being 
about two-thirds of the latter in height, and the same breadth 
respectively; though at one time during' my observation, which 
I continued for about an hour, I thought the inverted nearly as 
tall as the erect object. The day was fine and clear, with a very 
light air of wind, and I found very little tremor or oscillation 
in viewing her through the telescope. 
I have laid down fig. 4. for the explanation of the above phae- 
nomena, in which A represents the window I viewed B the 
vessel from ; H O, the curved surface of the sea ; C D parallel 
to H O, the height of the maximum of density of the atmo- 
sphere ; the lines marked with the small letters a a, b b, c c, dd, 
the pencils of rays under their various refractions from the ves- 
sel to the eye, or object glass of the telescope. 
The pencil of rays a a , from a point near the head of the 
mainsail, is wholly refracted in a curve convex upwards, being 
every where above the maximum of density ; and the pencil of 
rays d d, which issues from the same point in the sail, and passes 
near the horizon of the sea at x, is convex upwards from the 
sail to W, where it passes the line of maximum of density, which 
is the point of inflection ; there it becomes convex downwards, 
passing near the horizon at x to y, where it is again inflected, 
and becomes convex upwards from thence to the eye. The 
pencil of rays b b, from the end of the boom, passing nearly pa- 
rallel to the horizon, and near the maximum of density, suffers 
very little deviation from a right fine in the first part; but in 
ascending (from the curvature of the sea) will be convex up- 
wards to the eye. The pencil of rays c c , from the same point 
in the boom, may have the small part to c convex upwards, 
MDCCXCVII. F 
