358 Proceedings of Royal Society of Edinburgh. [sess. 
have spoken of the “ comparison ” beam as entering the face AB, 
while the absorbed beam was supposed to enter the face DE. 
But obviously these are merely the special circumstances of a 
particular mode of application of the apparatus, and the latter 
might be attached to a spectrophotometer in any other way, and 
would work equally well, provided that one of the beams — it does 
not matter which — is made to fall normally on the face AB, and 
the other to fall normally on the face DE ; and provided also that 
the point 0 (fig. 2) be in a plane optically conjugate to the retina 
of the observer’s eye. The latter condition is necessary to avoid 
the appearance of diffraction effects caused by the cutting off of 
the edges of the two beams at the edge of the silvering on the 
interface ; and also because the juxtapositor cannot be so exactly 
made that the two beams emerge from it quite parallel to each 
other ; but as can easily be seen, their edges in such a case will 
once more be brought in contact in any plane where a real image 
of the point 0 is produced by the parts of the optical train of the 
spectrophotometer. 
An important, and indeed one may almost say essential, principle 
of such an apparatus has been successfully observed, namely, that 
each of the two beams of light should pass through exactly the 
same length of glass. When this condition is not fulfilled the 
light from one beam will be absorbed to a greater extent than the 
light from the other, and an error will thus be introduced. Of 
course, in theory at least, an appliance faulty in this respect might 
be used correctly were its differential absorption found accurately 
beforehand ; but the correction would have to be ascertained for 
a great number of different wave-lengths throughout the visible 
spectrum, and every observation made with the spectrophotometer 
when the appliance was in use would have to be individually 
corrected. That the passage through even a short length of glass 
causes marked absorption in a beam of light, particularly at the 
blue end of the spectrum, has been shown by various workers, 
among others by Nichols and Snow;* and the knowledge of this 
fact caused the author to reject an earlier design which, though 
* “ Note on the Selective Absorption of Light by Optical Glass and Calc- 
spar.” By Edward L. Nichols and Benjamin W. Snow. Phil Mag. (5), 
No. 203, pp. 379-382, April 1892. 
