T)r, Brewster on the Polarisation of Light. 227 
rings, I was surprised to observe the lower half of all the 
rings completely different from the upper half, like the sketch 
in fig. 3, the half of the rings above the conjugate diameter 
of each was the very same as the first set described in my 
former paper, while the other half below the conjugate dia- 
meter contained the complementary rings, such as those which 
form the second set in the same paper. 
When we polarise a beam of light by refraction, at the same 
angle at which it is polarised by reflection, we obtain very in- 
teresting results. The number of plates necessary for this pur- 
pose is thirty, as represented by ABCD, fig. 4. If a ray of light 
RS is incident at S, so that the angle of incidence is 54° 35', 
then, according to the observations of Malus, the reflected 
ray ST will be wholly polarised, while according to the pre- 
ceding experiments, the transmitted beam EF will also be 
completely polarised. The pencil EF, however, is polarised 
in the opposite manner to ST, and the two pencils have the 
same relation to each other, as the two images formed by 
double refraction. If we, therefore, suppose the thickness AC 
of the thirty plates diminished to such a degree, that the eye 
could receive both the pencils ST and EF, and if we examine 
these two images by a prism of Iceland spar, they will com- 
port themselves in every respect, as if they had been produced 
by a doubly refracting crystal, vanishing and reappearing al- 
ternately in every quadrant of the circular motion of the spar. 
As a portion of the transmitted beam SEF is polarised at its 
emergence from each plate, and as this portion possesses a 
polarisation opposite to that of the reflected beam ST, and falls 
upon the subjacent plates at an angle of 54° 35', not one par-- 
tide of it will stiff er refection, but each minute portion of light 
G g 2 
