568 Wisconsin Academy of Sciences, Arts, and Letters, 
Determining the values of 6^ and 6^ by the comparison method, 
the coefficient of absorption per meter length was found for 
certain wave-lengths by use of this equation. 
With this form of spectrophotometer settings can be made 
to a high degree of accuracy, so that with good illumination 
check settings can be made to within the limits of error of the 
scale reading itself. Under the working conditions which 
were found to be necessary, however, such accuracy was • not 
obtainable. The intensity of the source for difterent wave¬ 
lengths varies greatly and it falls off rapidly toward the blue 
end of the spectrum. The retina of the eye reacts to light in 
a way that is also a function of wave-length. The rate at 
which the visual impression develops varies with both the in¬ 
tensity and the wave-length of the light that produces it as 
well as with the previous treatment of the retina. This was 
considered in the number of readings necessary so that for a 
single determination of the angle of the analyser the average 
of a number of settings varying from five to fifteen and in ex¬ 
treme cases even twenty or more were taken depending upon 
the reading of the wave-length scale. In order to eliminate any 
personal equation every one of the twenty-eight thousand 
spectrophotometer settings necessary to complete this investiga¬ 
tion was made by the writer. 
Since the wave-length scale of the spectrophotometer is en¬ 
tirely arbitrary, a calibration was made from certain standard 
wave-lengths. For this purpose were used the sodium, the 
lithium, and the argon lines, the violet, green, and yellow 
mercury lines, and the blue, green, yellow, and red helium lines. 
To correct for a noticeable width of the lines as seen in the 
instrument, an average of the readings of both edges of the 
line was taken as the position of the line on the scale. 
Due to the very small absorption coefficient of distilled 
water and some of the lake water measured, a long absorbing 
column is necessary. After careful experimenting, the ap¬ 
paratus was used as shown in Fig. 3. N is a Nernst glower 
which was run at full intensity and a narrow portion of which 
was cut out by the slit S so as to make it as nearly as possible 
a point source. The diverging beam was made parallel by an 
achromatic lens of 50 cm. focal length and then divided 
by the plate glass, P. The transmitted ray entered the brass 
tube, T, through the opening 0, which was one cm. in dia- 
