quantitative potential of the technique of iden- 
tifying fragments of grass leaf epidermis in 
feces. The study entailed feeding experiments 
with captive animals. Feces weighing 70 grams 
(fresh weight) were collected either twice or 
three times a day. The feces were kept in for- 
malin acetic alcohol (85 parts 70 percent alco- 
hol, 10 parts 40 percent formaldehyde, and 5 
parts glacial acetic acid). From each 70-gram 
fecal collection, a 1-gram sample was placed in 
4 ml. of concentrated nitric acid in a round- 
bottomed flask and heated for 2—3 minutes over 
a water bath. This treatment clears the epider- 
mal fragments and facilitates identification. 
The sample was then made up to 100 ml. with 
water, and boiled and stirred to complete the 
clearing process. After removal of the superna- 
tant fluid, the fragments were stored in a con- 
stant volume of 1 part formalin acetic alcohol 
and 1 part 30 percent aqueous glycerine. From 
each 1-gram sample, subsamples were spread 
out on six slides under 3.8- by 1.9-cm. cover- 
slips with a suitable density of fragments. 
Slides were analyzed both by counting frag- 
ments and by assessing their cover. Counts 
were made by traversing the slide systemati- 
cally at 100X magnification, and counting 
those fragments which fell partly or entirely 
between parallel lines marked on the slide. 
Area measurements were made, with an eye- 
piece micrometer, along similar traverses; 
those fragments or parts of fragments lying 
between parallel lines were measured. 
The results clearly invalidate the estimation 
of proportion of grasses ingested by counting 
all fragments. This invalidation occurred be- 
cause some species broke into smaller frag- 
ments and the total number of fragments was 
thus more numerous. Analysis by measuring 
the area of 100 fragments in each of three 
fecal samples showed considerable improvement 
in accuracy compared with counts. The vari- 
ance was high, however, and possible causes 
seemed to be the presence of occasional, very 
large fragments, or differential separation of 
fragments before or after these were placed on 
the slide. The high variance, and the length of 
time required to make analysis with point 
quadrats, mean that it would be extremely time 
consuming to make sufficient analysis to reduce 
the standard error to within 10 percent of the 
mean. 
Since the major constituents of the diet can 
be identified in the feces, quantitative data on 
a frequency basis, indicating the relative im- 
portance of different grasses in the diet, can be 
obtained. 
Croker (1959) collected droppings from one 
sheep for each sample. These samples were di- 
luted with an equivalent volume of water and 
subsampled. A thin film of material was spread 
between two slides, and 50 identifications were 
made on each of four slides, for a total of 200 
identifications per sample. No special treat- 
ment or staining was necessary. The objective 
was to determine which species were being 
grazed at that particular time. The results 
were not intended to be a quantitative esti- 
mate. 
Hercus (1960) reported on further work in 
New Zealand. She collected unweathered feces 
from the field and preserved them in 
formalin/acetic acid/alcohol. Three-gram sam- 
ples were suspended in 100 ml. of water, and 3 
subsamples were withdrawn for microscopic 
examination. The number of fragments of each 
cuticle pattern was counted and identified. She 
found some variation in the total count per 
unit volume on duplicate samples, because the 
cuticle fragments were not of even size or 
shape. To allow a quantitative estimate of in- 
take and utilization, she felt it was necessary 
to relate amounts of cuticle in feces to the ac- 
tual amounts of each plant species eaten, either 
in terms of weight or number of leaves. 
Storr (1961) used the microscopic analysis 
technique on feces of kangaroos and wallabies 
in Australia to determine their diet. Feces 
were first dried and thoroughly ground. The 
sample was then boiled in 10 percent nitric and 
chromic acid, and fitted to a reflux condenser 
for 1 minute. The material was washed with 
water and transferred to a centrifuge tube, 
drained, and dyed with an alcohol solution of 
gentian violet. The fecal residues were then 
washed with alcohol and centrifuged for at 
least 5 minutes and drained. Slides were exam- 
ined under low power (45X) by systemati- 
cally traversing zones 2.8 mm. wide, whose 
centers were 5 mm. apart. Fragments of epi- 
dermis were identified, and their area was esti- 
mated in hundredths of a square millimeter by 
using a graduated eyepiece. The percentage by 
area of each species present was then obtained 
for each slide. Analysis of many slides pre- 
pared from a single sample of feces was neces- 
sary when precise information was required 
about the diet of individual animals. More 
often, information is required about a popula- 
tion. In population studies, only one fecal pellet 
from each quokka was prepared, and the com- 
position of the preparation was based on the 
mean of the proportions observed in two slides. 
Data obtained were qualitative and quantita- 
tive because (1) there is little or no digestion 
of epidermis that is encased in cutin, (2) the 
epidermis is usually identifiable to species 
under low-power microscopes, and (38) the re- 
lation between the surface area and dry weight 
of the foliage can be determined for each spe- 
cies. : 
Hegg (1961) adapted procedures of Dusi 
(1949). For grass examinations he softened 
the fecal samples in water and then washed 
153 
