it is impossible to know anything very definite about the comparative total labor costs of different articles. But beyond this, we do know that the existence of differential rents destroys the possibility of proportionality between total labor costs and exchange values. The concept of the _marginal cost_ of a good appears in the Ricardian theory of rent, and has been involved more or less clearly in the reasonings of most modern economists, but it is almost entirely to J. B. Clark that we owe the consistent development of this idea. The marginal subjective cost of a good may consist either of labor or of abstinence, but not of both combined. In this essay we will arbitrarily set aside the problem of abstinence cost. The marginal labor cost of a good is, of course, determined by ascertaining the marginal product of labor in producing this good. To illustrate in the simplest manner possible, we will follow the time-honored procedure of eliminating capital for the moment, and suppose successive doses of labor to be applied to a given piece of land.[252] Let the labor force applied stand at a certain amount, and suppose the dose to consist of a labor day. Then, if experimentation reveals the fact that the addition of one more dose will increase the whole product by the amount of two bushels, we define these two bushels to be the marginal product of a labor day. Professor Clark frequently refers to this same quantum as the _specific_ product of labor. The land in this case may have consisted of a 100-acre field and the total labor applied may have amounted to, say, 300 labor days. The total produce may have been 3,000 bushels of grain. By hypothesis, capital being eliminated, the total labor cost of these 3,000 bushels is 300 labor days, or ten bushels cost one labor day, or the total cost of a bushel is one-tenth of a day. On the other hand, the marginal cost of a bushel is one-half of a labor day, since two bushels are the marginal product of a day.[253] Ricardo, who so explicitly defined total labor cost as consisting of the labor both directly and indirectly applied to a commodity, also assumed that in one respect value-determining cost is _marginal_, though he never used the word “marginal.” It was for this reason that not far back we described his method of defining cost as hybrid. His doctrine that rent does not enter into cost was but one way of stating that _on land_, it is only marginal cost (as he expressed it, the cost of the most costly portion of the supply) which determines value. In real life, products are the result of combining not land and labor alone, but land, labor and capital (in the sense excluding land—our usage at present). When Ricardo was expounding and illustrating the theory of rent which bears his name, he was forced to suppose that the successive doses added to land were composed of capital and labor jointly,[254] which left his marginal quantum the product of both of these agents.