109 
dicular to their refractive edges, could be given any value. When by 
this means the same deviation is given to the ray of light thrown 
on the polariser, as is caused by the polariser itself but in opposite 
direction, the deviation is neutralized. 
3. For the observation of the lines in the compensator, a slit is 
placed before it and the threads, to shut off the strong light, that 
passes on the side of the threads through the compensator. A slit 
of 1 mm. is sufficient. A narrower slit gives troublesome diffraction 
phenomena. A shifting of this slit before the compensator gives a change 
in the position of the compensator-lines. Therefore the slit must 
always be symmetrical with regard to the threads. As, however, 
slit and threads are not in the same plane, this symmetrical position 
is disturbed, when the rays run no longer in the direction of the 
axis of the telescope behind the analyzer. Thus before every determination 
of the position of the compensator-lines the telescope must be carefully 
placed in the direction of the reflected rays. If this precaution is 
observed, the positions of the compensator lines in the different pola- 
riser positions differ at most so much as corresponds to a phase- 
difference of 0,0013 x22, the limit of observability being 0.0006 x 22. 
+. Finally Mr. Rerser has still applied the following expedient 
to make the method of observation more precise. With a telescope 
behind the compensator, focussed for parallel rays, we see besides 
the direct slit image, two more side images, formed through internal 
reflection in each of the wedges. By focussing on the compensator 
lines, hence on the threads, the side images coincide, however, with 
the central, direct one. It is self-evident that this is not conducive 
to an accurate measurement of phase-difference and tbe ratio of the 
components of the examined elliptically polarized light. When the 
rays of these side-images are shut off by a screen with a slit in the focal 
plane of the object-lens of the telescope, the compensator line becomes 
perfectly black, narrower and more sharply defined. The accuracy 
of only one determination of the position of the line corresponds to 
0.0006° 22; the accuracy of the restored azimuth, which was before 
about 20’ for 64 adjustments, now amounts already to 5’—9’ for 
16 adjustments. We have found this increased accuracy to be abso- 
lutely necessary for the investigation of the surface-layers. 
5. In order to prevent vibrations of the mercury surface, the 
mercury vessel is mounted on a small adjustable stand, free from 
the goniometer. The displacements of the nicols and the compensator 
can then take place without danger of vibrations of the liquid surface. 
