Thus all the means of production are separated from the national income as 'wealth', and this income is divided into surplus value and labour power, or better, its equivalent, the variable capital. This, then, though still far too vague, is our division into constant capital, variable capital and surplus value. But 'national income', it soon transpires, means for Sismondi the annual aggregate product of society: 'Similarly, annual production, or the result of all the nation's work in the course of a year, is composed of two parts: one we have just discussed--the profit resulting from wealth; the other is the capacity for work, which is assumed to equal the part of wealth for which it is exchanged, or the subsistence of the workers.'[171] The aggregate social product is thus resolved, in terms of value, into two parts: variable capital and surplus value--constant capital has disappeared. We have arrived at Smith's dogma that the commodity price is resolved into _v + s_ (or is composed of _v + s_)--in other words, the aggregate product consists solely of consumer goods for workers and capitalists. Sismondi then goes on to the problem of realising the aggregate product. On the one hand, the sum total of incomes in a society consists of wages, capital profits and rents, and is thus represented by _v + s_; on the other hand, the aggregate social product, in terms of value, is equally resolved into _v + s_ 'so that national income and annual production balance each other (and appear as equal quantities)', i.e. so that they must be equal in value. 'Annual production is consumed altogether during the year, but in part by the workers who, by exchanging their labour for it, convert it into capital and reproduce it; in part by the capitalists who, exchanging their income for it, annihilate it. The whole of the annual income is destined to be exchanged for the whole of annual production.'[172] This is the basis on which, in the sixth chapter of book ii, 'On Reciprocal Determination of Production and Consumption', Sismondi finally sets up the following precise law of reproduction: 'It is the income of the past year which must pay for the production of the present year.'[173] If this is true, how can there be any accumulation of capital? If the aggregate product must be completely consumed by the workers and capitalists, we obviously remain within the bounds of simple reproduction, and there can be no solution to the problem of accumulation. Sismondi's theory in fact amounts to a denial of the possibility of accumulation. The aggregate social demand being the bulk of wages given to the workers and the previous consumption of the capitalists, who will be left to buy the surplus product if reproduction expands? On this count, Sismondi argues that accumulation is objectively impossible, as follows: