This paper combines simplified Harrod-Domar and Mahalanobis-type models, first to show that the internal consistency of a development plan may well depend on supply relationships that are usually ignored in aggregate planning, and second, to suggest the value of a more general version of the “two-gap” model . An empirical application of this analysis, to be published separately, will look at these supply relationships and their influence on Pakistan’s development. The argument of the paper is simple and not unfamiliar. To the extent that installed capital is specific in what it produces and the products themselves are specific and non-substitutable between investment and consumption uses, the pattern of investment in one period may determine the allocation of subse¬quent income to saving or consumption and, therefore, the rate of growth of national income. As this constraint of insufficient physical capital is softened by trade, a foreign exchange constraint may enter. Part I sets the context by broadly contrasting the Harrod-Domar and Mahalanobis views of growth. Part II develops a formal but quite simple model of an economy whose growth is limited by either saving or physical output constraints. Part III introduces trade as an alternative source of capital, relaxing the physical capital constraint. Part IV deals with geographical specificity of capital sources in a ‘four-gap model’.