260 
Proceedings of the Royal Society of Edinburgh. [Sess. 
away with by using a double-image prism so constructed that the two 
beams are not, as is usual, polarised vertically and horizontally, but instead 
in two directions inclined at 45° to the vertical. I would add the general 
remark that, while polarisation spectrophotometers are popular on account 
of their great convenience, they possess the grave fault of being wasteful of 
light. This is due, of course, to the fact that in polarising a beam of light 
more than half the intensity is inevitably lost. And although polarisation 
affords a ready means of altering light intensities, the accuracy of the 
measurement obtained when a double-image prism is employed is some- 
what uncertain, because such prisms do not really divide a beam of light 
into two equally intense beams, but give rise also to faint additional beams. 
This may easily be verified by examining any bright light-source through 
such a prism, when, in addition to the two chief images that are formed, 
various others will be seen. 
III. These reasons induced me to turn my attention to the construction 
of a flicker spectrophotometer which should be independent of polarisation. 
Two different types of arrangement suggest themselves, one depending 
on the deviation caused by a prism, the other on reflection by a mirror. 
(1) An instrument employing the first of these principles is diagrammati- 
cally shown in fig. 3, where W is a sort of glass wedge wheel, consisting 
of two semicircular discs A and B, fig. 4, A being one half of a convergent 
lens, and B one half of a divergent lens of equal power. The axle of W 
is situated parallel to the collimator tube intersecting the (imaginary) 
prolongation of the slit, and (indifferently) either above or below the 
latter. The axle is rotated by means of a small pulley fixed on it, driven 
by a belt from the little water motor mentioned before. It will be seen 
that the transition of the passage of the light entering the collimator slit 
from the one half lens A to the other B takes place very rapidly. When 
A is in front of the slit the beams X and Y are deflected downwards ; when 
B is in front of the slit they are deflected upwards. The result is that 
the aperture 0 in the screen S placed at the end of the collimator is lighted 
alternately first by one beam and then by the other ; and therefore, unless 
the two beams are of equal brightness, the observer, on looking through 
the small telescope T, which is focussed on the aperture O, is conscious of 
a flickering effect. Evidently it is not essential that the lens wheel should 
be placed at the collimator slit, and as a matter of fact I originally put 
it in a position in front of the observing telescope T ; but there are two 
drawbacks to this. First of all, it is less convenient because it means 
mounting both lens wheel and driving motor on the movable arm of 
the spectroscope, instead of having the lens wheel alone on the immovable 
