184 TRANSACTIONS OF .THE [APR: 1, 
that all of the light emitted by the glowing body had under- 
gone a refraction. Considering the closeness of agreement be- 
tween the calculated and observed values, it is difficult to escape 
the conclusion that this assumption is correct, and that no 
particles whatever of the incandescent solid send out into the air 
natural light, save in the case in which the angle of emergence 
is zero. This simply means that all of the particles of the light 
emitting body, including the so-called surface layers, lie within 
the denser medium, and beneath the plane at which reflection 
and refraction take place. This relieves the refraction theory 
of the causes of the phenomenon of its greatest difficulty : viz., 
the difficulty of conceiving that, in the case of an exceedingly 
opague body like platinum, the uppermost molecules send out 
but a very small proportion of the whole light emitted. If we 
follow the explanation of Arago and Verdet we are obliged by 
the results of this research to conclude that the emitted light 
originates almost entirely in molecules other than those of the 
uppermost layer. On the contrary it seems much more reason- 
able to assume that, in the case of such a body as platinum, the 
light emitted is due mainly to this topmost layer, but that the 
reflection process takes place entirely above the platinum. 
Quincke has shown that when light from an external source 
is reflected at the surface of a metal the reflection does not take 
place in the geometrical plane between the two media, but 
rather takes place in the metal itself, the vibration penetrating 
for a certain depth into the denser medium. The converse is 
also doubtless true that the vibration originating in the metal - 
is not reflected instantaneously at the surface of the rarer 
medium, but is reflected in the layer of air of finite thickness 
which borders upon the metal. Thus all light originating in the 
platinum, whether in the surface layer or the sub-surface layers, 
must undergo the process of reflection and refraction before it 
can emerge into the air. 
Lastly, the calculated values were all obtained under the 
assumption that the optical constants of the metals are the same 
for high temperatures as for low; that is, that the reflecting 
properties of an incandescent metallic surface are precisely the 
same as the reflecting properties of a cold metallic surface. 
The closeness of agreement between the results given by this 
assumption and the facts as determined by experiment seems to 
warrant the conclusion, that the change in the optical properties 
of metals due to incandescence is exceedingly slight; a conclu- 
sion to which the somewhat inexact experiments of Grove upon 
the reflecting properties of incandescent platinum would also 
lead. 
