39 ^ 
Journal of Agricultural Research 
Vol.XXVI, No. 9 
The results for the alcoholic solutions are the most comparable because 
for both instruments the solution was not diluted before observations 
were made, and, further, alcohol is less volatile, hence any errors from this 
source would be reduced. The alcoholic solution was more advantageous 
for the colorimeter, for experimenters generally agree that readings made 
from dilute solutions are more accurate than those from more concen¬ 
trated ones. It will be observed that for alcohol some of the results for 
the colorimeter are greater and some are less than those for the spectro¬ 
photometer. 
In the case of ether, the pigment oxidized and, consequently, only two 
results were obtained by means of the colorimeter. The other points 
could not be gotten, for the concentration of the pigment became too 
small to be read from the graph in Figure 4. 
By diluting the concentration used it was always possible to obtain 
readings by means of the spectrophotometer. In this respect the spec¬ 
trophotometer was superior to the colorimeter, for it was always possible 
to make dilutions on the solutions used. 
The petroleum ether solution of carotin was diluted 20 times before 
making the determinations on the spectrophotometer, while it was not 
diluted at all for the determinations on the colorimeter. 
In every way the best comparison is in the case of absolute alcohol. 
The results show very clearly what may be expected of the two instru¬ 
ments, under the conditions given. The spectrophotometer gave results 
which were quite consistent, while the colorimeter gave results which 
varied. 
In Table X it is seen that the carotin content as determined in the 
alcoholic solution by means of the spectrophotometer varied between 
3.54 and 3.68 mgm. per liter, with an average of 3.60, which is a maximum 
deviation of + .08 and -.06, or a maximum determination error of 14 parts 
in 360, or 3.9 per cent. In the case of petroleum ether the results varied 
between 49.0 and 52.4 mgm. per liter, with an average of 50.3, which is a 
maximum deviation of +2.1 and —1.3, or a maximum determination 
error of 34 parts in 503, or 6.7 per cent. It must be remembered that the 
petroleum ether solution was diluted 20 times, while the alcoholic solution 
was .not diluted, so that the results as actually measured on the spec¬ 
trophotometer varied between 2.45 and 2.62 mgm. per liter in the case of 
the petroleum ether solution. 
The figures (Table X) which were obtained for the alcoholic solution 
by using the colorimeter vary from 3.2 to 3.9 with an average of 3/55, a 
maximum deviation of +0.25 and —0.35, or a maximum determination 
error of 60 parts in 355, or 16.9 per cent. 
Escher (2, p. 47) is the only investigator who gives the individual 
colorimeter readings. His readings are 82.0, 82,0, 83.8, 84.4, 83.6, 87.5, 
87*9, 87.5, giving an average of 84.8 divisions on the colorimeter. His 
maximum determination error was 5.9, or 6.9 per cent. In other sets 
of readings his maximum errors were 8.4, 4.2, and 5.1 per cent. 
Some figures obtained by three different observers, all of whom were 
more or less familiar with the colorimeter, are submitted herein. A 
carotin solution (1.68 parts in 100,000 of ether) was used, and the same 
solution was put into both cells of the Duboscq colorimeter. The left- 
hand side was set at 50 mm. and each of the three observers, A, B, and C, 
regulated the right-hand side so as to match the solution on the left, 
while the writer observed the readings. Observers A and B by shading 
their eyes with a cloth got the results in AA and BB just a few minutes 
