Economic history in general and agrarian history in particular is seriously hampered by the lack of production and income data before the middle of the 19th century. Analysis of British agriculture therefore has to rely on indirect evidence. This paper discusses previous attempts and suggests a novel solution to agrarian ourput and productivity measurement, using new types of data and explicit theoretical constructs. The results reported here suggest that the first systematic attempt to measure agrarian production, i.e. Deane & Cole (1962), underestimates growth while the most recent analysis by Crafts (1985) seems to overestimate growth.