( *94 ) 
We distinguish, the following eases 
1. a very small, then - . 
b-\-a b 
'& + « b 
3. 
2 . 
We shall apply these results to the interpretation of our observations. 
Two eases dependent upon the magnitude of a are of principal 
importance. 
In the first case we can observe the effect of both the axes of the ellipse 
by means of the combination of the Fresnel rhomb and the calcspar 
{this is the case of the quartet) (D, Fig. 9), whereas without Fresnel 
rhomb; no effect of the small axis is visible. In. the second ' case the 
effect of the small axis becomes apparent by the use of the calc- 
spar, whereas its existence cannot be demonstrated with the Fresnel, 
the value of -q— being too small. This case is represented by the 
sextet , (D 2 Fig. 9). 
If the observation is made by means of the calcspar rhomb, we 
indeed see with dense vapour new components in the lowest stripe 
(see Fig. 8, D s ). The theoretical import of this result will be discussed 
on another occasion. 
v After introduction of the Fresnel rhomb the component to the 
left of the central line (small axis of the ellipse) remains invisible. 
(Fig 9, D„ inferior stripe). 
Hence we may conclude that at the angle now investigated the 
ellipticity of the outer components of the sextet (the ratio -) exceeds 
that of the quartet (and is also larger than 0,414). 
15. Observations at & = 45°. 
The photographs taken with the calcspar alone, show very dearly 
the ellipticity of the outer components. 
With vapour of intermediate density the phenomenon is already 
very marked, especially in the case of D, (Plate II, Fig. 10). Very 
remarkable is the slight diminution of intensity of the inner com¬ 
ponents of the quartet. According to the elementary theory the inten¬ 
sity of the central component of a triplet ought to have diminished 
already to less than half the original value. 
