IDENTIFYING THE PEASANT CLASSES-IN-THEMSELVES IN RURAL INDIA 97 posit a hierarchy of resource endowments relative to the production capacity and consumption needs of the peasant households, at a given point of time. This generates relations of labour-hiring and land-leasing in the process of production, and determines the terms on which the different groups of producers participate in the markets for commodities, land lease, credit, or labour hire. It is not 'markets9 which are 'interlinked9: the interlinking is between different classes of persons involved in the production process, and the nature of this exploitative interlinking is determined by relative resource endowments, in short by the distribution of property. The social relations of labour-hire, land-lease, usurious credit relations and commercial profit extraction, appear to the neoclassical economists in the 'fantastic form' of an interlinking of "markets', because the theoretical framework adopted precludes an appreciation of class relations and emphasises exchange to the exclusion of production. Given a hierarchy of resource endowments generating exploitative relations, it follows that at a given point of time there will be a hierarchy of different situations with regard to satisfaction of subsistence needs and appropriation of surplus : some 'peasants9 will fail to get enough to eat, other 'peasants9 will break even, yet other 'peasants9 will generate large enough surpluses not only to maintain higher-than-average consumption levels but also be able to invest in improving agricultural production (if it is profitable for them to do so). In short, the peasantry may be expected to be highly differentiated economically into more or less distinct classes. Since we know that investment in new 'green revolution9 technology has been profitable, the existence of a class hierarchy within the peasantry implies that there will be a hierarchy of technical-organisational production method-complexes in use. At one pole we may expect on the part of capitalist producers maximising profit, very intensive cultivation methods with a high level of application of capital per unit of area (including in 'capital9, wage-advances to hired labour) and high land and labour productivity ; at the other pole on the part of poor peasants struggling for subsistence, we may expect extensive cultivation methods with very low application of modern inputs and machinery relative to labour input, consequently low land and labour productivity. In short, there will not be one, but a range of 'production functions9 attributable to a cross-sectional sample of households. It follows from the above that all attempts to estimate 'production functions' whether for individual crops or in the aggregate, by using farm economics data from a cross-sectional sample of holdings, would represent quite meaningless attempts in every situation where class differentiation exists. Similarly the question of the 'efficiency9 of the typical 'tenant9 compared to the typical 'owner9, would represent a meaningless way of posing the question where 'tenants9 run the gamut from large-scale capitalist producers to poor peasants hiring out their labour, and 'owners'