improvement made in their manufacture. Has the value of the coat, under these circumstances, fallen or risen? It is impossible to say: all that can be said is, that it has fallen in relation to one thing, and risen in respect to another. But there is another case, in which no one would have any hesitation in saying what sort of change had taken place in the value of the coat: namely, if the cause in which the disturbance of exchange values originated, was something directly affecting the coat itself, and not the bread or the glass. Suppose, for example, that an invention had been made in machinery, by which broadcloth could be woven at half the former cost. The effect of this would be to lower the value of a coat, and if lowered by this cause, it would be lowered not in relation to bread only or to glass only, but to all purchaseable things, except such as happened to be affected at the very time by a similar depressing cause. We should therefore say, that there had been a fall in the exchange value or general purchasing power of a coat. The idea of general exchange value originates in the fact, that there really are causes which tend to alter the value of a thing in exchange for things generally, that is, for all things which are not themselves acted upon by causes of similar tendency.
In considering exchange value scientifically, it is expedient to abstract from it all causes except those which originate in the very commodity under consideration. Those which originate in the commodities with which we compare it, affect its value in relation to those commodities; but those which originate in itself, affect its value in relation to all commodities. In order the more completely to confine our attention to these last, it is convenient to assume that all commodities but the one in question remain invariable in their relative values. When we are considering the causes which raise or lower the value of corn, we suppose that woollens, silks, cutlery, sugar, timber, &c., while varying in their power of purchasing corn, remain constant in the pro-
