Empirical support for the very close link between physical resources and agriculture is longstanding. For example, Hidore (1963) investigated the relationship, between: (i) flat land, defined as slopes of less than 3; and (ii) cash grain farms. For the sample of 730 counties in the north-central United States, the Pearson product-moment correlation coefficient was statistically significant at +0.652. The only major deviant state was Minnesota where its high percentage of flat land was offset by unsuitable soil conditions,the southern edge of the Laurentian Shield. Studies like that of Hidore are part of a long tradition of environmental studies which have formed a major part of the geographical literature. However,as Found(1971) reminds us in his extensive review of agricultural location theory, we must not allow the fact that land-use variations over large areas can be 'explained' to a considerable extent by variations in land quality (including climate), to persuade us that the role of man as an independent decision maker can be disregarded. Individual environmentally sympathetic decisions at the local scale may be lost when aggregated into land-ues zones which conform to the broader sweep of physical circumtances. At the local, farmers' decisions depend on production functions, and the prices and costs of inputs; the former is associated with the local environment and the latter with lacal market conditions, either of which can dominate his decision his decision, at least for a time