IN QUARTZ OF LIGHT IN THE VISIBLE REGION OF THE SPECTRUM. 
29] 
The equation given above also shows a remarkably close agreement with a pre¬ 
liminary series of observations in tlie infra-red. It is, therefore, evident that tlie 
infra-red absorption is an essential factor in determining the optical rotation in quartz, 
l)oth in the infra-red and in the visible region of the spectrum, and that Drude 
was in error in supposing “ that the kinds of ions whose natural rotations lie in the 
infra-red are inactive” {loc. cit., p. 413). 
The equation does not agree so well with preliminary measurements in the ultra¬ 
violet region and cannot be made to do so without sacrificing the concordance obtained 
in the values for the visible region of the spectrum. It is considered probable that 
further experience may demand either (l) some modification in the value \i= 0'010627 
at present accepted for the wave-length of tlie ultra-violet absorption, or (2) substitu¬ 
tion for the term ^ of a term of the ordinary form 
lijo 
The exact nature of the 
modification needed to bring the values for the ultra-violet region into agreement 
with the observed figures will be discussed only when the experimental work in this 
region has been completed. 
For the present the calculated values serve mainly to place a limit upon tlie 
maximum error from all sources that can be ascribed to the measurements now 
recorded. The readings for the red zinc line appear to be quite abnormal, as they 
differ from the calculated value by 0°'006 per mm. and cannot be brought into agree¬ 
ment by any modification of the constants without spoiling the concordance of the 
rest of the figures. Deviations amounting to nearly 0°‘004 per mm. are observed in 
the case of the blue cadmium line; but these form part of a group of negative 
differences and must be attributed, at least to some extent, to imperfections in the 
formula, which (as has already been suggested) requires the addition of a fourth 
arbitrary constant to bring it into complete accord with the experimental figures. 
The necessity for a fourth term is shown by the fact that whilst the concordance 
between the observed and calculated values is very good at the ends and in the 
middle of the series, the differences being 
Li 6708, +0-0003; Hg 5461, +0-0009; Hg 4359, +0-0006; 
there is a predominance of positive differences in the first twelve and of negative 
differences in the last twelve values. 
Thus, if all the values be included, the average differences are 
XVave-length 6708 to 5461 +0-0007 
Wave-length 5351 to 4359 —0-0007 ; 
if the reading which shows the largest difference in each group be excluded, the 
average differences are 
Wave-length 6708 to 5461 +0-00025 
Wave-length 5351 to 4359 —0-00036. 
2 r 2 
