An improved method of ultra-violet anomalous rotary dispersion of sodium tartrate

Abstract

A large number of ultra-violet polarimetric measurements have been made by a method described by one of us in 1908.* In this method a triple-field polarimeter, with Foucault in place of Nicol prisms, is arranged in series with a quartz or quartz-calcite spectrograph. A quartz-calcite lens, replacing the eyepiece of the polarimeter, casts a real image of the triple field on the slit of the spectrograph, and thus gives rise to a triply divided spectrum, intersected by dark bands. The wave-length corresponding with a given rotation is determined by finding a line which is of equal intensity in the three fields. The approximate form of the curve of rotatory dispersion can be determined by setting the analyser in a series of positions separated by 5° or 10° ; but it is then advisable to “bracket” some of the more conspicuous lines by making fresh exposures at intervals of perhaps 0·2°, when the rotation corresponding with the wave-length of a given line can be determined within about 0·1°. This degree of accuracy is less than that which can be reached in the central part of the visible spectrum, where the readings may be reproduced under favourable conditions within about 0·01° ; but it is not appreciably less than the accuracy of visual readings in the red and violet regions, and for many purposes is quite satisfactory. Thus, with a column of quartz 496 mm. in length, readings to 1° sufficed to give the rotatory power in degrees per milli­metre within 0·002°, corresponding with an error of about 1 part in 100,000 at wave-length 2327. The present investigation was undertaken in order to find out whether the same apparatus could be used to record with sufficient accuracy the much smaller rotations of solutions which had been largely diluted, in order to render them transparent in a region near to (or covered by) an absorption band. For this purpose the concentration is often reduced to I per cent. : the observed rotations may then be of the order of 0·5°, and must be read to 0·01° or less in order to give a true impression of the form of the curve of rotatory dispersion.