The first number of our earlier series, A Programme for Growth, carried a notice of forthcoming papers. Five were announced but eventually only four were published. The fifth, which was intended to deal with consumption functions, never appeared; now it takes its place as number one in the new series. It is not that ten years ago we had nothing to say on the subject of consumers' behaviour. The crude estimation method that I had used in my original (1954) paper on the linear expenditure system gave interesting and in many respects satisfactory results, some of which were published outside our series, for instance in Stone, Brown and ). With this method the parameter estimates changed Rowe ( 1964 very little after the first few iterations. Nevertheless they did change, and with the computing resources then at our disposal we failed to reach convergence. It was mainly for this reason that we decided to wait.