42 
CONTRIBUTIONS TO CHEMISTRY AND MINERALOGY. 
[BULL. 167. 
light box that can be easily held in one band, about 35 centimeter long 
and 13 to 14 centimeter square, painted black inside and out and with 
one end closed by a piece of ground glass, the other open. For a space 
equal to the width of the glasses the cover is removed at the top next the 
glass end to permit of the insertion of the glasses, side by side, in such a 
way that no light shall penetrate around their sides or between them. A 
stiffly sliding, black cardboard shutter is movable up and down im- 
mediately back of the glasses, so that all light can be cut off except that 
which comes through the liquid. 
Precautions of this kind are necessary if accurate results are to be 
counted on. Except for mere traces this simple combination of glasses 
and darkened box insures greater accuracy and rapidity of work than 
Nessler tubes and is preferable likewise, so far as my own experience 
goes, to expensive instruments like the colorimeter of Soleil-Duboscq, 
etc. The glasses should, of course, have exactly parallel sides and 
equal diameters, though they may perhaps be made with advantage 
rather narrow in one direction to permit of using smaller bulks of 
liquid. 
In making the color comparisons the box is best held close to a win- 
dow, so as to get a full, strong light. The condition of the light seems 
to make an appreciable difference in-'the accuracy of the comparisons, 
all of which were made by daylight. 
Accuracy of Results. 
Especially for comparatively and very small amounts of chromium 
the method gives exact results, better than can ordinarily be hoped for 
from any gravimetric method, considering the fact that as a rule other 
substances are present which it is extremely difficult to remove 
completely. 
A few comparisons between colorimetric and gravimetric determi- 
nations of chromium in a few rocks are here given to show the order of 
agreement, the former having been made several months and even 
years after the latter : 
Gravimetric. 
Colorimetric. 
Per cent. 
Per cent. 
trace 
0.018 
0. 05 
.051 
.14 
.12 
.08 
.083 
trace 
.013 
none 
.0086 
none 
.0067 
