39° PRINCIPLES OF ANIMAL NUTRITION. 
above assumptions, be 4.1X2516X0.09=928 Cals. Adding this,- 
as before, to the observed difference of 772 Cals. gives a total of 
1700 Cals. as the effect of the 648 grams of crude fiber, which equals 
2.623 Cals. per gram. With a digestibility of 55 per cent., this 
corresponds to 4.768 Cals. per gram of digested crude fiber, or 
materially more than its metabolizable energy. 
UNCERTAINTIES OF THE CompuTaTION.—The whole method of 
computation, however, is open to serious criticism on at least two 
points, aside from the rather indefinite statements as to the amount 
of hay consumed in Period ¢ and as to the distribution of the ration 
between the three feedings in Period f. 
First, the estimate for the work of digestion of the nutrients 
other than crude fiber which forms the basis of the computation is 
derived chiefly from the experiments of Magnus-Levy on dogs and 
man. Those experiments were not only made with highly digesti- 
ble food, but the digestive work is computed as a percentage of the 
total (gross) energy of the food. The food of the horse contained 
in the dry matter 40.94 per cent. of indigestible substances in 
Period f and 54.37 per cent. in Period c, or if we leave out of account 
the crude fiber the corresponding figures are 31.99 per cent. and 
58.45 per cent. A considerable part of the work of digestion un- 
doubtedly consists of muscular work, which must be performed 
on the indigestible as well as the digestible matter of the food. 
Moreover these indigestible matters, by their mechanical stimulus 
and by acting in a certain sense as diluents, may perhaps cause a 
more abundant secretion of the digestive juices. These facts are 
entirely ignored when the figures for digestive work derived from 
experiments on dogs and man are applied simply to the digested 
food of the horse. 
Second, the method of computation assumes that the difference 
between the metabolism on the two rations which was observed 2.7 
hours after eating would have retained the same absolute (not rela- 
tive) value during the twenty-four hours. The justification for this 
assumption is found in a comparison * of the results of a single res- 
piration experiment, made one half hour after feeding, with the 
average of two experiments in which the excretion of carbon 
* Loc. cit., p. 218. 
