Dec. 1, 1925 
Revised Net-Energy Values of Feeding Stuffs 
1091 
resenting heat production, but also upon the periods in which the 
errors occurred, and upon the differences in feed between the periods 
compared. This may be illustrated by the following hypothetical 
example involving three experimental periods: 
Let H ± , H 2 , and J? 3 represent the heat production of Periods I, II, 
and III, respectively. The ration of Period III is the smallest. 
The difference in dry matter consumed between Periods II and III 
is 1 kg.; between I and II, 3 kg.; and between I and III, 4 kg. The 
heat increments per kilogram of dry matter would therefore be as 
follows: 
Periods compared 
Heat in¬ 
crement 
per kilo¬ 
gram dry 
matter 
I and II.... 
Hi—Hi 
3 
Hi—Hz 
II and IU_... 
I and III.. 
1 
Ih-Hz 
4 
Let three cases be considered: 
(1) If H t alone is in error by ±300 Cals. 
(2) If H 2 alone is in error by ±300 Cals. 
(3) If Ug alone is in error by ±300 Cals. 
If h represents the true heat increment per kilogram of dry matter 
of the feed, we will have the following heat-increment values in the 
three cases just mentioned: 
Periods compared 
Differ¬ 
ence in 
feed 
(dry 
matter) 
Heat increments per kilogram of 
dry matter 
Case 1 
Case 2 
Case 3 
land II____ 
Kg. 
3 
1 
4 
Cals. 
A±100 
n 
h± 75 
Cals, 
h- F100 
h± 300 
h 
Cals. 
h 
ft=F300 
h- F 75 
II and III____ 
I and III____ 
Average___ 
hzh 58 
h± 67 
fc=F125 
The difference between the extreme heat-increment values in case 
1 is 100 Cals., while it is 400 Cals, in case 2, and 300 Cals, in case 3, 
although the magnitude of the assumed error in the heat production 
is the same in each case. The effect of this assumed error on the 
average heat-increment value is also different in the three cases. 
Especial attention is called to the relative magnitude of the devia¬ 
tions from the true value (h) in the different comparisons of the 
periods in which the error is involved. The deviation is greatest 
where the difference in feed is smallest. On this account it seems 
very desirable that the feed of the different periods to be compared 
shall differ in quantity as widely as practicable; and in some cases in 
which the evidence is clear as to the location of the errors it seems 
permissible to exclude from the average the heat-increment values 
based upon small differences in feed. 
