366 H, P. Armsby—Digestion Experiments. 
. Nitrogen. Crude fiber. 
Probable error of 1 determinati 0-052 per cent, 0°22 per cent, 
These figures differ somewhat, but not very greatly, from 
those computed directly from the analyses in question, and 
serve to show that the latter include no gross errors. 
The determinations of dry matter, protein and crude fiber 
in the dung were made in precisely the same manner as with 
the fodders. As only duplicate determinations were made on 
any one sample there are no sufficient data for a computation 
of probable errors, but there is no evident reason why they 
should differ essentially from those of the same determinations 
it must be remembered that the percentages of dry matter are 
to be calculated upon the amount of air-dry hay fed, or of fresh 
dung excreted, while the percentages of protein an erud 
fiber are to be reckoned upon the amount of dry matter fed, or 
excreted. e may, for convenience, first compute the proba- 
ble errors on the basis of a single determination each to be a 
follows : 
Probable Errors.—Single Determination. 
Average consumption. Average excretion. 
Grms, Grms. 
Fresh 700 622 ; 
WA 588 +0°84 287 0°75 
Protein (N x 6°25) .....-__- 1841-47 sos 
rade fiber 1894191 99 40°93 
These probable errors have now to be reduced in accordance 
with the number of observations upon which the value of each 
quantity depends. These are as follows: 
For dry matter, protein and crude fiber of fodder, four deter- 
minations ; for protein and crude fiber of dung, two determina 
e value for dry matter excreted rests upon duplicate 
daily determinations, repeated for six days. The Siow 
above computed, then, should be divided by W/2 XW 6=3'4 6. 
For the actual probable errors, then, we find the following 
amounts : 
Probable Errors. ak 
Average consumption. Average excretion. Average oe 
Grms. Grms. 4? 
aoe oS a ae 588 + 0-42 287 + 0-22 per 
Protein (N x 6°25) -__ BL0-74 3940-51 39+ hat 
rude her: - .. 22. 189+0-96 99 + 0°66 90+ 
By dividing the probable errors of the last column by the 
amounts of the corresponding ingredients fed, we obtain the Por 
centage error, V1Z: 
