( 593 ) 
remains as with the transversal effect. Yet the presence of elliptic 
polarization ought to manifest itself by the appearance in the lowest 
stripe of lines, corresponding to the outer components. 
With very dilute vapour and with that of intermediate density as 
good as no trace of it is seen. 
Fig. 7, Plate II shows the first phase with dilute vapour, Fig. 8 
the second phase with denser vapour. Only traces of absorption, 
indicative of elliptic polarization can be seen near D*, Fig. 8. 
The ellipticity is, however, undoubtedly proved by means of the 
Fresnel rhomb, placed with its principal plane at an azimuth of 45° 
with the horizon. Fig. 9 shows the appearance. 
The outer components of the quartet towards the red or towards 
the violet, dependent upon the stripe and the direction of the field, 
are now considerably weakened; in the case of the sextet they have 
vanished altogether. All this proves the elliptical polarization of the 
outer components. For, if the polarisation were linear, as might be 
inferred from observations with the calcspar alone, then the obser¬ 
vation with calcspar and rhomb combined, ought to show no difference 
between the upmost and lowest stripe. The light of all plane polarized 
components would issue circularly polarized from the rhomb and, 
the calcspar making no selection between right-handed and left- 
handed polarizations, the components towards red and towards violet 
would all be alike. Such a condition is disproved by photographs 
such as Fig. 9. 
14. One point must be considered somewhat more in detail. What 
is the reason that the ellipticity is not shown by the calcspar 
rhomb alone, whereas its existence is most clearly demonstrated by 
means of the Fresnel rhomb? 
Let an elliptic vibration with vertical axis b, horizontal axis a, 
be incident upon the rhomb, the principal plane of which is at an 
azimuth of 45°. 
It is easily proved that the elliptic vibration issuing from the 
Fresnel rhomb has its axes in the same direction as the original 
~ h _« a 
motion and a ratio of the ; 
If a be small in relation to 
issues from the Fresnel as a r 
easily analysed. 
— r= —-, the original ratio being —. 
b x b-\-a b 
(an elongated ellipse), then, the light 
ore circular vibration, which is more 
It depends upon the magnitude of a, whether - is greater or less than 
HV 
