THE UTILIZATION OF ENERGY. 547 
energy required by its digestion hardly affects the computation. 
What the result appears to show is that the estimate of 9 per 
cent. for the digestion and assimilation of the fiber-free nutrients 
is approximately correct. ; 
The difference in the amount of digested crude fiber was some- 
what greater than that in the total amount. If we make the com- 
parison of the two averages on the basis of the fiber-free nutrients 
in the same manner as in previous cases we have— 
Fiber-free nutrients: 
Heavier work... ..............000. 5524 grams 
Lighter work..................... 5086“ 
Difference... ........ 0.00. e eee 438“ 
Equivalent energy .................. 1735 Cals. 
Energy of work....... 20... eee cece 445 “ 
Utilization... 2.0... 0c ccc cece ees 25.65 per cent. 
Apparently a considerable amount of energy was required for 
the work of digestion and assimilation in addition to that equiva- 
lent to the digested crude fiber, a result which seems to conflict 
with the conclusions drawn from a discussion of the same experi- 
ments in the preceding paragraph. The apparent discrepancy lies 
in the determination of the amount of external work equivalent 
to the added nutrients. Wolff, as we have seen, after securing 
an approximate constancy of live weight, corrects the measured 
amount of work in accordance with his judgment of the amount 
which would have been equivalent to the ration given and relies on 
the ‘‘might of averages” to overcome the inherent uncertainties of 
his method. Zuntz & Hagemann, on the other hand, reckon with 
the measured amount of work, but are then compelled to correct 
their final result for the loss of live weight, and unfortunately this 
correction is relatively a very large one (over 50 per cent.) and rests 
upon a rather uncertain basis. While it would perhaps be pre- 
sumptuous to attempt to decide the relative value of the two methods 
and the probability of the divergent conclusions based on them, one 
can hardly avoid feeling that the trained judgment of the actual 
experimenter is a safer reliance than such a relatively large cor- 
rection computed by a critic. 
Digitized by Microsoft® 
