40 SECTIONAL ADDRESSES 
a maximum of experimental accuracy can be obtained, namely, by using 
long columns of concentrated solutions. Under these conditions the 
validity of the equation has been vindicated up to the limits of experi- 
mental accuracy for a single term in octyl alcohol (45) and for two terms 
in ethyl tartrate, drastically purified by crystallising to constant melting- 
point (46). 
An extreme limit has been reached in the case of quartz, where measure- 
ments to six significant figures have been made for twenty-four wave- 
lengths in the visible spectrum, on a column nearly half a metre in length. 
This column gave a rotation of 12,678-96° for the green mercury line 
Hg 5461 ; and when the ten sections of the column were dismantled and 
re-erected, the original reading was reproduced with an error of only 
0:03°, or less than three parts in a million (47). 
In the infra-red region, the rotation per millimetre falls from 25 -539° 
per mm. at 5460 -742A.U. to2°at 18,000 A.U.,1° at 25,000A.U.,ando+74° 
at 28,000 A.U. Observations are then interrupted by an absorption 
band ; but beyond this there is a narrow window through which observa- 
tions can be made before the medium again becomes opaque. Snow’s 
measurements (48) have shown that the rotations in this narrow region 
of transparency (about o «52° per mm. at 32,000 A.U.) fallonthesame curve, 
and can be expressed by the same formula, as those in the infra-red, 
visible and ultra-violet regions. The infra-red absorption band is there- 
fore, as Drude supposed, without influence on the course of the curve of 
rotatory dispersion. 
In the ultra-violet the rotations increase very rapidly. Thus for a 
copper line at 2263 A.U. the observed rotation of the half-metre column 
rose to 101,332, or 202 °328° per mm. (47). 
Throughout the whole range from 32,100 to 2263 ALU. the rotatory 
dispersion of quartz for about 1000 wave-lengths can be expressed within 
very narrow limits by two terms of Drude’s equation, of opposite sign, 
together with a small constant (47): 
9 5639 2°3113 
22 — 00127493 22 —0-000974  ° 7975 
ao 
This equation does not express with equal accuracy the observations 
made by Duclaux (49) at still shorter wave-lengths with a much shorter 
column of quartz, but it predicts with considerable precision the existence 
of two absorption bands, with characteristic frequencies far out in the 
Schumann region at 1130 and 310 A.U. I am still waiting, however, for 
a physicist to carry out the experiments which are needed to disclose the 
presence of these two bands, the existence of which has already been 
predicted for nearly a quarter of a century. 
NoRMAL AND ANOMALOUS ROTATORY DISPERSION. 
Experiments such as these have demonstrated, beyond the possibility 
of controversy, the ability of Drude’s equation to express the rotatory 
dispersion of transparent media up to the extreme limits of accuracy 
