Microhistological analysis of epidermal fragments in feces is often us
ed to estimate the diet of herbivores but is not generally accepted as
a consistently reliable method, Gross errors arise, especially when d
iets are composed of herbage components with widely different morpholo
gical and structural characteristics. The present study investigated t
he possibility of using such physical characteristics to improve the r
eliability of the method. Over a 7 day period, 4 rumen-fistulated beef
cows were given a fixed diet composed of a shrub, a grass, and a forb
component, On the last 2 days, samples of rumen content and feces a e
re taken for analysis of epidermal fragment, Forbs were under-estimate
d, grasses over-estimated, and shrubs correctly estimated, Correction
factors to estimate true diet composition were defined as the biomass
represented by the specific epidermal fragments (epidermal weight inde
x) and the degree of degradation to which the epidermis is subjected i
n the digestion process (epidermal erodibility factor), These factors
account for characteristic physical features of the different dietary
components and were measured directly or were derived from the calibra
tion experiment, The utility of such factors depends on accurate deter
mination of the component variables and may be overshadowed by samplin
g error and observer bias in the microhistological identification of e
pidermal fragments.