160 TRANSACTIONS OF THE [APrR. 1, 
air, there may be less resistance to vibration in a direction away 
from the surface than in a direction parallel to the surface, and 
therefore that the light emitted is composed mainly of vibra- 
tions in a direction normal to the surface. If this were the case 
the light emitted normally would be unpolarized, while that 
emitted at oblique angles would be polarized in the plane per- 
pendicular to the plane of emission. Furthermore, the polariza- 
tion would. increase with the angle and might be very great at 
large angles, in case the difference in density between the two 
media were very great—conclusions all of which are in accord- 
ance with the facts. In view, then, of the inability to account, 
by Arago’s assumption, for the extreme polarization at large 
angles of emergence, and in view of the plausibility of the other 
explanation, the following qualitative experiments were made in 
order to determine with more certainty the nature of the phe- 
nomenon. 
IV. 
QUALITATIVE EXPERIMENTS. 
The object of this part of the research was: 
(1) To make certain that the property of polarization is due to 
the incandescent body itself,and is not caused by the refraction 
of the light as it passes through the layers of air of varying 
density which rest upon the luminous surface; and, 
(2) To make observations upon as wide a range of substances 
as could be made to emit light without combustion, in order to 
ascertain whether any substance could be found which does not 
possess the characteristic, and also in order to determine in a 
general way the relations of different bodies with reference to 
this property. 
For these purposes two instruments were employed; the first, 
a polariscope similar to that of Arago, save that the simple: 
quartz crystal was replaced by a bi-quartz plate, and the crystal 
of calc-spar by a double Wollaston prism. This is the same in- 
strument which was afterwards used by Arago in his polarime- 
ter, and it has an advantage over the first form in that the two 
colors to be compared are brought into immediate juxtaposition. 
It is delicate to the extent of detecting a polarization of about 
3% (as was shown by succeeding experiments), when white light 
is under examination. When the light to be tested is mono- 
chromatic, as was the case in some of the following experiments, 
the second form of polariscope was found to be preferable. 
In this instrument the bi-quartz is replaced by a cube of 
glass which has been subjected to strain in cooling. A Nicol 
