IN QUARTZ OF LIGHT IN THE VISIBLE REGION OF THE SPECTRUH. 267 
before it enters the polarimeter. It was overcome by Pekkin’s device of using a 
direct-vision prism in conjuction with the eye-piece to throw out the stray light. This 
prism forms an essential part of the apparatus and can only be dispensed with under 
exceptional circumstances some of.which are discussed later in the paper. When the 
prism at the eye-piece was required to aid in the resolution of a close doublet or in 
the elimination of a satellite or to throw out stray light from a closely-adjacent line, 
direct-vision prisms were used which had been specially constructed with a centre- 
prism of glass of refractive index 1 ’92 ; three of these prisms were used for the red, 
green and blue portions of the spectrum respectively. A greatly-increased dispersion 
was secured with the help of two prisms, specially constructed from the same dense 
glass, of the Rutherford pattern (see fig. l), in which the angle of tlie central prism is 
so great that light travelling parallel to the base would be totally reflected at an 
air-surface but is actually brought in and out of the prism at grazing incidence by 
prisms of light glass cemented to the dense prism. One of the Rutherford prisms 
transmitted the spectrum to about X 4800 ; the violet lines were here lost by total 
reflection but were transmitted by a second prism constructed with a slightly smaller 
angle ; the actual angles were :— 
For red, yellow and green 
r dense prism 104° 
flight prisms 8° each 
For Idue and violet 
rdense prism 98° 
flight prisms 10° each 
These prisms require more careful setting in front of the eye-piece than prisms of the 
ordinary dh’ect-vision type and are more troublesome for the eye to find and to use ; 
but their very high dispersion, combined with great economy of light, renders them 
of great value in cases where a system of specially high dispersive power is required. 
The primary purpose of the prism in front of the eye-piece of the polarimeter is to 
eliminate stray light rather than to resolve the lines of the spectrum ; this resolution 
must be effected mainly by the constant-deviation spectroscope, in order that adjacent 
lines and satellites may be blocked out by the polarimeter slit and so prevented from 
producing a distracting glare of light in the neighbourhood of the line under observa¬ 
tion. A high resolution is of value, both in separating the lines of a doublet, such as 
the yellow doublet of mercury or of sodium, and in increasing the width of the lines 
that can be used without overlapping, so that in many cases it is possible to cover the 
whole width of the aperture of the polarimeter with a block of pure monochromatic 
light. But high resolution is usually gained at the cost of a serious sacrifice of light, 
and it was only when using bright arc-spectrum lines in the central part of the visible 
spectrum that the most dispersive systems could be used with advantage. 
Successive increases in the dispersive power of the spectroscope were obtained in 
the following way :— 
1 . 
The 11-inch lens of the 
“ eye-piece ” 
Q 
tube of the spectroscope was replaced by an 
M 2 
