In 18 digestibility experiments with sheep, 9 diets consisting of single-grass hays and 9 diets each consisting of equal quantities of a good and poor quality hay were fed. Correlations between the voluntary intake of organic matter, digested organic matter, total cell wall and digested cell wall in all possible combinations were investigated. There were no significant differences among the simple correlation coefficients for hays fed singly or as mixtures, even though they were eaten in very different amounts. The simple pooled correlations were significant (P < 0-01) among voluntary intake, digested organic matter and cell wall. The cell wall was significantly correlated with digested cell wall, but there was no simple correlation between voluntary intake and digested cell wall. However, when considered together in a multiple linear model they explained 58 % of the variation in voluntary intake compared with 43 % explained by the total cell wall alone. The additional variation accounted for by the digested cell wall was significant (P < 0-05).The relative importance of the total cell wall and digested cell wall is discussed in relation to selection in grass breeding. T , Tr __ __ TT _ lrrlT _-T a comparison of in vivo and in vitro digestible fractions in forage, Moir (1971) considered that this During the growth of plants their cell walls, apparently undigested fraction comprised 6-4 g of comprising cellulose, hemicellulose and lignin, in-protein and protein-free material which was of crease in amount; a physical association of lignin endogenous origin, 3-2 g of undigested protein-free with cellulose and hemicellulose changes in such material, and a variable amount of undigested a way that it becomes increasingly difficult for the protein equivalent to about 15 % of the total rumen-microorganisms to penetrate the cell wall protein in the plant. For a grass containing 10 % (VanSoest, 1969). The result is that the percentage protein the total apparently undigested non-celldigestibilities of cellulose and hemicellulose decline, wall fraction was, therefore, about llg/lOOg The net effect of this increase in amount and grass OM. lowered percentage digestibility is that these cellIn 45 grasses investigated by Moir (1972) variawall carbohydrates tend to be digested in constant tion in the amount of apparently undigested nonamount in the plant. In grasses the mean amount cell-wall fraction was relatively small, but, there of total cell wall digested is about 40 g/100 g grass was much larger variation in the amount of cell organic matter (Moir, 1972(Moir, , 1973(Moir, , 1974.wall digested. In the 45 grasses the mean and The percentage digestibility of the total non-standard error for digested cell wall was 39-5 ±0-6 cell-wall fraction in plants appears to decline as g/100 g grass OM. The range, which was not plants mature, but the effect is rather a decline in reported, was 30-46 g digested cell wall. Although the amount of a fraction that is completely digested the sample of 45 grasses was infinitesimally small (Van...