172 PRINCIPLES OF ANIMAL NUTRITION. 
body as compared with a possible 6724 grams from the fat and 
proteids of the food. The feeds used, however, were not well suited 
to young animals and the gain was abnormally small in proportion 
to the food consumed, so that the results could not be expected to 
be decisive. Moreover, the presence of non-proteid nitrogen in the 
food is not considered in the computation. (See the next paragraph.) 
Sourcés of Uncertainty.—Up to this point the results of experi- 
ments on herbivorous.and omnivorous animals had been somewhat 
conflicting. Before taking up the later investigations it is desir- 
able to point out some of the uncertainties attaching to experiments 
such as those above enumerated. These relate, first, to the amount 
of fat actually produced, and second, to the possible sources of 
supply in the food. 
The basis for estimating the amount of fat-actually produced by 
a fattening animal was in two cases a comparison with the amount in 
a supposedly similar animal at the beginning of the fattening, the 
fattened animal being actually analyzed. In the remainder the 
increase in live weight was assumed to have the composition found 
by Lawes & Gilbert. It need scarcely be pointed out that the 
results of such comparisons can be only approximate and are sub- 
ject. to a considerable range of error. Only the most decided 
results one way or the other can be accepted as at all conclusive. 
In experiments on milch cows the production of milk fat can of 
course be determined, but the variations in the weight of such an 
animal often render any conclusions as to gain or loss of body fat 
so difficult that the results as a whole are less satisfactory than 
those on fattening. 
The possible sources of fat in the food, aside from the carbohy- 
drates, are the ether extract and the proteids. As regards the first, 
it is certain that not all the digestible ether extract of stock foods 
is true fat. With the proteids the case is still worse. In particu- 
lar we now know that a portion, and in some cases a considerable 
portion, of the total nitrogenous matter of feeding-stuffs consists of 
non-proteid material, which so far as we know contributes little if 
anything directly to fat production. This is a very important source 
of error. Thus the writer * has shown, as has also Soxhlet,t} that if 
* Manual of Cattle Feeding, p. 182. 
t Compare Soskin, Jour. f. Landw., 42, 203. 
