776 
408 l. c. where the values of ifj ml and tp m2 are assumed to be equal. 
At first sight, the discrepancy seems perplexing; on closer examin¬ 
ation, however, it must be admitted that the data actually obser¬ 
ved would fit equally well with a form of curve departing from 
that adopted by M. Cotton and which might indeed be made to 
satisfy condition (5). 
Let us now examine to what extent the observed variation of 
the rotation can be represented by formula (2) of Article 8. In 
the following table the wave-lengths (in 10 -5 cm) are given in the 
first column, the observed rotations in the second column, in the 
third the corresponding values of the constant D computed from 
equation (2) in Art. 8., F having the value assigned to it in (3) above. 
X 
W\ 
D 
5,22 
2° 50' 
4,97.10" 11 cm' 
5,40 
2 ° 10' *). 
4,92 . 
5,81 
1° 46' 
5,27 . 
5,89 
Assuming 
2 ° 30' 
5.94 . 
(6) 
D = 5,05.10" 11 
cm 2 
equation 
(2) in Art. 1 
X 
8 gives the following results: 
calculated observed by 
from (2) § 8 M. Cotton 
4,75 
2° 32' 
1 ° 52' 
5,00 
2° 51' 
2° 39' * 
5,22 
2° 53' 
2 ° 50' 
5,40 
2° 13' 
2 °10'* 
5,62 
0° 07' 
0 ° 21' 
5,81 
1° 42' 
1° 46' 
5,89 
2 ° 08' 
2° 30' 
6,00 
2 ° 26' 
2° 48' * 
6,57 
1 ° 56' 
1 ° 26' 
q In M. Cotton’s table for X = 5,62 we have the rotation 1° 21'; but this 
is no doubt a misprint and should probably be 0° 21'. Since X = 5,62, however, is 
in the immediate vicinity of X = X 0 , the corresponding value of <1/ is open to consider¬ 
able uncertainty. For this reason another value, that corresponding to X = 5,40, 
was read from the curve. 
