AbstractData from tables of composition of animal feedstuffs in which there were entries for digestibility (digestible dry matter or digestible organic matter) and cell wall were classified into different categories—grass or legume, stage of growth, leaf or stem, and forage species. No conclusions were drawn about legumes, because of insufficient data. After removing the effect of different categories, all the grass species behaved similarly with respect to digested cell walls, In contrast, within‐species variation in digested cell walls was large, as shown by the high residual standard deviation of ± 4.0 in a simple linear regression of in‐vivo digestibility on cell‐wall content. There was a small effect which was due to the lignin content of the cell wall, but the source of most of the variation was not located. While it remains undiscovered, there is a danger that experiments involving forage might merely be testing the effect of treatment on random variation in the digested c