342 Proceedings of Royal Society of Edinburgh. [sess. 
liquid in the tube, of these three errors just noted only the first 
would be done away with, besides which such a plan would give 
less light intensity than the arrangement of the collimator described 
above. 
In a spectrophotometer as ordinarily made there is no room 
between the collimator and the prism for an absorption vessel, 
and to comply with the above parallel light condition it becomes 
necessary to take off the collimator and to mount it by itself in 
front of the spectrophotometer at such a distance as permits of 
inserting the absorption vessel between the two. 
In the ordinary type of spectrophotometer there are two difficulties 
that would arise were parallel light to be used. It will be 
seen that as all the rays of both the beams of light which emerge 
from the absorption vessel are parallel to the general optic axis of 
the instrument, these two beams of light, after duly passing 
through the prism and the object glass of the telescope, will give 
rise to one and the same spectrum; and that the width of this 
spectrum will be very small. 
Taking the latter difficulty first, the width of the spectrum 
produced by any spectroscope must be equal to the length of the 
collimator slit multiplied by the focal length of the telescope 
objective and divided by the focal length of the collimator lens. 
Now in the above arrangement the “length” of the small hole 
which acts as a collimator slit may be about ^th of an inch, so 
that the spectrum formed by the two beams of light will have a 
quite insufficient width for our purpose. Besides, we require 
each beam to give rise to a separate spectrum, and we must not 
have the two spectra formed in the same position one upon the 
other. Both difficulties, however, are readily solved by using as 
the telescope objective a cylindrical lens (C, fig. 4) whose axis of 
figure is placed vertically : the focal length of the lens being 
identical with that of the spherical lens whose place it has taken. 
Tn this way, while using a strictly parallel beam of light to pass 
into the absorbing vessel, we obtain two separate spectra placed 
one above the other, and formed respectively by the “ comparison ” 
and by the “ absorbed ” beams of light ; and the widths of these 
spectra are amply sufficient, for they are respectively equal to 
the heights of the cross sections of each beam of light. See 
