In livestock feeding experiments the problem of recursiveness arises because the quantity of feed consumed by an animal is a function of its past history of feeding. The problem is most acute where experiments are designed to analyse sub ad lib feeding. The analysis of such experiments encounters two important problems. The first is that the actual quantity of feed consumed is an endogenous variable and is not directly under the control of the experimenter or the livestock producer. The second problem is the mathematical complexity of the relationships involved. A solution is suggested as a quasi reduced-form model.