IN QUARTZ OF LIGHT IN THE VISIBLE REGION OF THE SPECTRUM. 
279 
After rejecting R 20 Lgo, these cylinders with and R 4 were used for several 
complete series of measurements with light of all the available wave-lengths. As the 
whole of these readings, covering a period of nearly three years, were subsequently 
rejected, they need not be referred to in detail, except in the case of the mercury 
green readings which are of importance as bearing on the tests by which the validity 
of the values finally given was established. 
After a number of preliminary experiments, the two groups of cylinders, each 
cemented into a rod, were found to give the following values {a) by setting to the 
position of minimum rotation {h) by setting with the help of a reflecting eye-piece :— 
7 L Cylinders . 
5 li Cylinders . 
[{a) 448r-781 
\{h) 448179/■ 
3136-51. 
Rotation per millimetre 25°'5350. 
25-5342. 
After re-cementing, somewhat lower values were obtained :—■ 
7 L Cylinders .... 4481-61. Rotation per millimetre 25-5340. 
5 R Cylinders .... 3136-41. ,, ,, ,, 25-5336. 
In trying to determine the efiects of the satellites upon the reading of the violet 
mercury line, the mechanical arrangement of the polariser was modified so that the 
half-shadow angle could be thrown either to the left or to the right. This was found 
to have a marked effect on the readings, which was subsequently recognised as 
occurring even in the case of the very pure mercury green line which gave :— 
7 L Cylinders . 
5 R Cylinders . 
The method of reading with both positive and negative half-shadow angles had thus 
brought the values of the dextro- and Imvo-cylinders into complete agreement and 
appeared to have provided the last device needed to ensure accurate readings. With 
the experience gained during the two preceding years, yet another series of readings 
was undertaken, therefore, in which every line was read with both a positive and a 
negative half-shadow angle. These readings, which were practically complete when 
the quartz was discarded, are referred to later as proving that faulty quartz gives 
incorrect figures for rotatory dispersion even when allowance has been made for errors 
in the absolute values. 
At the close of this series, the cement was found to have been separated a little in 
some of the joints ; it was therefore all cleaned off and the cylinders were put together 
in optical contact, giving a greatly improved parallelism to the end faces of the column. 
After this alteration the cylinders were retested by setting each group in eight or 
H.S.-f 
4481-61 
H.S.- 
4481-53 
H.S.-i- 
3136-41 
H.S.- 
3136-53 
4481°-57 = 25°-5338 per mm. 
3136-47 == 25-5338 per mm. 
