Joty—Ldentification of Crystals by the use of Birefringence. 491 
as possible to the crystal plate. A small galvanometer mirror of 
long focus also serves the purpose, but not so well. With diffused 
daylight, and without aid of condenser, the use of a No. 3 Leitz 
gives ample illumination. Considerably higher powers may be 
used with concentrated rays, solar or artificial. The application of 
this mode of examination to crystals of very small dimensions is, 
with the foregoing arrangements, not satisfactory owing to a 
certain amount of paralactic displacement of the ray returning from 
the mirror: the illumination not being quite vertical. 
The change of phase, introduced by reflection at the surface of 
the metal, of the component vibrations into which the crystal plate 
resolves the rays from the illuminator, appears to be sufficiently 
small to be negligible. It can be shown that the change of phase 
produced by metallic reflection of a polarized ray at normal inci- 
dence is zero, and will be small at the higher angles. In the 
present case the ray is nearly normal. The loss of phase arising 
on the total reflection of the beam within the prism of the illumi- 
nator, and before the rays reach the rock-section, might be expected 
to create a more marked effect. However, when elliptic polariza- 
tion is sought for by a double image prism of Iceland spar replac- 
ing the eye-piece of the microscope, the extinction of the one 
image is found to be very perfect ; as complete, nearly or quite, as 
when the normally polarized beam of the sub-stage nicol is 
examined in the same manner. ‘The disposition in which the ray 
reflected in the illuminator is polarized at right angles to the angle 
of incidences should be the best. 
A simple experiment further serves to show that for thin sec- 
tions at least, this method of examining a crystal plate truly 
affords an interference colour proper to a section of double the 
thickness. A thin cleavage flake is taken from a limpid crystal of 
selenite, and of such a thinness as to show a vivid interference 
colour in transmitted light, and uniformly over a sufficient area. 
The plate is then cut across the area of uniform illumination and the 
two halves attached upon a glass slip, the cut ends being overlapped 
by a millimetre or thereabouts. Examined now, first in light once 
transmitted through the plate, the doubled thickness where the 
plates overlap should present the same interference tint as the 
single thickness will reveal when examined in polarized light 
twice transmitted through the crystal. It is found to prove 
