270 TERRESTRIAL REFRACTION AND MIRAGE. 
at the point D and the point H is in fact seen by the eye at C in 
the direction CD in which the ray enters the eye ; it thus appears in 
the position H,. On the whole the rod appears bent into the form 
KH). The property possessed by different media, which causes the 
tay to be bent in passing from one medium to another, is termed 
the refractive power of the medium, and in passing from one medium 
to another whose refractive power or refractive index is greater 
the bending of the ray is always towards the normal; but in passing 
from one medium to another which is less refracting, the bending 
of the ray is always away from the normal. In particular if a ray 
passes from a vacuum into air of a certain density then the ray is bent 
towards the normal, and is thus bent to an extent which depends upon 
the density of the air; and if a ray passes from air of one density to air 
of greater density then again the bending is towards the normal, where- 
as if the path of the ray is from air of a certain density to air of less 
density, then the bending of the ray is away from the normal. It 
happens that the earth is surrounded by a spherical envelope of air 
which is not all of the same density. There are two principal causes 
which militate against the uniform density of this air; one is barometric 
pressure and the other is temperature ; there are other possible causes 
that we will touch upon later. The barometric pressure diminishes 
from the surface of the earth upwards, and if the question of tempera- 
ture did not enter at all, there would be a gradual diminution of the 
density of the air from the surface of the earth upwards. Asa matter 
of fact the temperature largely modifies the state of the air in many 
cases. As a general rule—thatis to say under normal circumstances— 
the temperature diminishes as we rise from the surface of the earth and 
this has the effect, to a certain extent, of diminishing the variation in 
density, but it has nota sufficiently counteracting effect to cause the 
density to be anything like uniform. The density of a given volume 
of air depends upon the pressure to which it is subjected and upon its 
temperature, and any work on heat shows how the density may be de- 
termined when once the pressure and the temperature are known. The 
refractive power of the air, or the refractive index, can be shown to be 
sensibly proportional to the excess of the number expressing its density 
over unity; if the letter p represent the density of air, the refractive 
power can be shown to be proportional to p—1. 
We may consider the air to be stratified in horizontal layers; it is in 
reality, of course, stratified in spherical layers, but it will suffice for our 
present purpose to consider it to be stratified in horizontal layers, be- 
cause the sphericity of the earth is not at all the principal cause of the 
terrestrial refraction ; terrestrial refraction would still exist if the earth 
had no sphericity, or if its surface were perfectly plane. What we may 
regard as being the normal state of the atmosphere in Great Britain is 
a gradual diminution of temperature upwards at the rate of about 
34> degree Fahr. per foot, and under those circumstances the density of 
the air gradually diminishes as we rise from the surface of the earth. 
The diagram (Fig. 2) may represent, roughly, by the closeness of the 
lines which indicate density, the normal state of the atmosphere. 
When a ray of light passes from one point of such an atmosphere to 
