on Horizontal Refractions. 31 
The hummock d is a part of the head land, but appeared in- 
sulated or detached from the rest, and considerably elevated 
above the sea, with an open space between. I then came down 
about twenty-five feet, when I had the dry sand of Robin Rigg, 
x y, in the apparent horizon, and lost all that floating appear- 
ance seen from above, and the Abbey Head appeared every 
where distinct to the surface of the sand ; this being in the af- 
ternoon, the wet or moisture on the sand would in a great 
measure be dried up. I have reason, therefore, to conclude 
that evaporation is the cause of a less refraction near the sur- 
face of the sea ; and when so much so as to make an object 
appear elevated wholly above the horizon, (as at d in fig. t.) 
there will from every point of this object issue two pencils of 
rays of light, which enter the eye of the observer; and that 
below the dotted line A B (parallel to the horizon of the sea 
HO), the objects on the land will appear inverted. 
To explain this phaenomenon, I shall propose the following 
theory, and compare it with the observations which I have 
made. Suppose H O, fig. 2. to represent the horizontal surface 
of the sea, and the parallel lines above it, the lamina or strata 
of corpuscles, which next the fluid are most expanded, or 
the rarest; and every lamina upwards increasing in density till 
it arrive at a maximum (and which I shall in future call the 
maximum of density) at the line D C, above which it again 
decreases in density ad infinitum. 
Though this in reality may be the case, I do not wish to ex- 
tend the meaning of the word density farther, than to be taken 
for the refractive power of the atmosphere ; that is, a ray of 
light entering obliquely a denser lamina to be refracted .towards 
a perpendicular to its surface ; and in entering a rarer lamina,. 
