( 353 ) 
9 Influence of reflexions from the walls of the vacuum, tube. 
In the course of the investigation different mercury tubes were 
used With one of these I had the opportunity of repeating an 
observation, made on a former occasion'), under very less favourable 
circumstances. I will give some details for a case of simple magnetic 
separation. , , 
The yellow mercury lines split into two components, it the light 
is examined parallel to the field. Observing in the latter direction 
and while using a new tube I was astonished to see the yellow 
mercury lines divided into triplets, the three components being ot 
equal intensity. The plane of incidence of the rhomb was under this 
observation vertical. The intensity of the central line of the triplet 
was a maximum when the plane of vibration of the Nicol was 
horizontal, being zero when it was vertical. The central line appar¬ 
ently is linearly polarized. It is easily seen by means of a figure 
that this central line may be due to light, emitted nearly at right 
angles to the direction of the field, and, after being reflected once 
or twice from the inner surface of the capillary, emerging m the 
direction of the axis of the electromagnet. 
Apparently in the direction of the axis a doublet, circularly pola¬ 
rized, is seen superposed upon a triplet with linear polarization. In 
the specified position of the Fresnel rhomb, at an azimuth of 45 o 
the Nicol one of the components of the doublet, after a rotation of 
135° the other component must vanish; then, however, the vertical 
component of the reflected triplet still must remain visible, its inten¬ 
sity being halved, because the principal section of the Nicol makes 
in the case now considered an angle of 45° with the vertical line. 
The outer components of the yellow mercury triplets only have 
about one half of the intensity of the central component, hence, if 
expressed in the intensity of the latter only 7< remains for the 
vertical vibrations. Observation wholly confirms this. 
If the Fresnel rhomb be brought into such a position that the 
plane of incidence makes an angle of 45° with the horizontal plane, 
the central component must be expected to exhibit circular polarization. 
This also was confirmed by observation, as other conclusions to 
be drawn in the case now under review. 
I may be permitted to mention, that the central component in the 
Ease of the mercury line 5790 has an asymmetrical position; sym¬ 
metrical, however, in the case of 5771. This appearance corresponds 
b Zeeman, These Proceedings 29 Febr. 1908, p. 573. 
Lohmann, Diss. Halle S. 62, 1907. 
