[15] CHEMICAL COMPOSITION AND VALUE OF FISH FoR Foop. 245 
represent water-free substance, 7. e., crude extract minus water. I am 
inclined to think it would be better to determine both water and fat in 
the crude extracts in one operation by extracting with ether and noting 
the loss. 
Insoluble protein.—The residue left after the extraction with hot water 
was treated with alcohol or ether, or both, dried and weighed. Water, 
ash, and fat were then determined (except, of course, that fat was not 
determined in the cases where it had been previously extracted). The 
ash was determined by direct burning, it being assumed that the pre- 
vious treatment with cold and hot water had sufficed to remove the 
easily fusible and volatile salts. The removal of the last portions of fat 
4s often extremely difficult, and it is not impossible that in some cases 
traces were left and weighed as insoluble protein. The figures for total 
insoluble protein in the table denote water-free substance. 
Ether extract, fats—From 0.3 to 1.0 gram of “A” (hydrogen dried) was 
extracted with ether until free from fat. The operation was conducted 
in an apparatus similar to that described by Johnson* (Am. Jour. Sci., 
XIU, 1877, p.190). The fat obtained was dried in a current of hydrogen 
before weighing. 
Nitrogen was estimated by the soda-lime process. Some study has 
been made in this laboratory of the conditions under which this method 
gives accurate results, especially with materials rich in nitrogen, the 
outcome of which may be reported elsewhere. It will suffice here to say 
that we find it important to have the substance well ground and 
mixed with the soda-lime and to insure complete ammonification of the 
nitrogen in the decomposition products. This last we attempt to 
secure by a sufficiently long anterior layer of soda-lime, preferably in 
coarse particles, high enough heat and rather slow burning. 
The correctness of the determinations of nitrogen was tested by com- 
parative trials by the absolute method. In this latter, two sources of 
error, often neglected, were taken into account, namely, the air which 
adheres to the interior of the tube and its contents even after long 
exhaustion with the Sprengel pump, and the vapor tension of the 
caustic potash solution over which the nitrogen is collected and meas- 
ured in the eudiometer. 
To determine the correction for residual air in the combustion tube, 
blank determinations were made with pure oxalic acid. The tube was 
well exhausted, the mereury being allowed to run three-quarters of an 
hour after the “click” was heard, before the combustion was begun. 
The amounts of gas obtained in four combustions were, when reduced 
to 0° and 760™™, 0.3°¢, 0.6°¢, 0.5°¢, and 0.6%, averaging 0.5°°, which 
amount was deducted from the volume of gas obtained in each deter- 
mination. 
It is common to disregard the vapor tension of the caustic potash 
solution, especially when the latter is nearly concentrated. In aceu- 
rate work, however, this is hardly allowable. Even a 50 per cent. 
~ *The same as that described by Tollens, Fres. Zeit. Anal. Chem., 17, 1878, s, 320. 
