780 
It is of interest to compare these figures with the value of R which 
is obtained by making direct use of the results of Art. 11. Substi¬ 
tuting (3) and (6), § 11., in (2) above, we find 
(3) R = 4,20.10 -16 cm 3 . 
The accuracy with which equation (4) of Art. 8. represents the 
variation of ellipticity with wave-length can be seen from the 
following table; in order to calculate the values given in the second 
column, the following estimate of the constant R has been adopted: 
(4) R = 3,85 . IO" 16 cm 3 
X (p 
calculated observed by 
from (4) § 8 M. Cotton 
5,00.10-5 cm io 52 / 0°18'* 
5,22 2° 57' 1° 25' 
5,40 . 4° 07' 2° 33'* 
5,62 4° 55' 4° 46' x ) 
5,81 4° 11' 40 54' 
5,89 . 3° 40' 3° 40' 
6,00 . 2° 57' 2° 45'* 
From the table as well as from the appended graphical represen¬ 
tation (Fig. 3) it appears that the agreement is anything but satis¬ 
factory. This negative result might indeed have been anticipated. 
In equation (4) § 8 the constant quantity D may be positive or 
negative but in a given absorption band it is either one or the 
other. Hence equation (4) § 8 precludes the possibility of the curve 
cp=zçp(X) intersecting the axis <p = 0; yet this intersection is a not 
uncommon occurrence in M. Cotton’s experiments. 
§ 13. The formula (5) of § 8 above allows an estimate of the 
magnitude of the constant k being made from any pair of conjugate 
values of tp and cp. Let us endeavour to get in this way some 
idea of the magnitude of the constant k in our equations. It will 
be perceived at once that this method and that explained and used, 
with the same object in view, in §§ 9 and 11, are independent 
of one another. 
A. The following table shows the results obtained from equa- 
*) A misprint in M. Cotton’s table is here corrected. The values marked* 
are read from M. Cotton’s curve. 
