Mammalian herbivores cannot break down cellulose except by fermentatio
n, and may have fermentation chambers at either end of the gut: rumina
nts have their principal fermentation chamber in the stomach but horse
s ferment only in the hindgut. A mathematical model (Alexander, 1991)
predicted that foregut fermenters should do better than hindgut fermen
ters on poor foods, and the reverse on richer, less fibrous foods. Fur
ther, the optimum gut for poor foods would have the hindgut fermentati
on chamber only a little smaller than the foregut chamber. However, it
has been claimed that horses do better than ruminants on poor food, a
nd the hindgut of ruminants is much smaller than the rumen. In this pa
per, the basic model is modified in ways designed to make it more real
istic, and the effects are investigated. None of the modifications alt
ers the conclusion that the optimum gut for poor food has a large fore
gut fermentation chamber. However, the optimum proportions of fore-to
hindgut, for pool diets, become more like those of real ruminants when
account is taken of the diminishing volume of the food passing throug
h the gut, and of incomplete mixing in the rumen.