I reject Fisher's notion that the average of prices is an independent entity. But I do not consider that the idea lying behind this untenable doctrine is absurd. Cost of production, supply and demand, and the other price theories _do_ presuppose something more fundamental. They do presuppose _money_, and the _value_ of money, as has been shown at length in Part I. The trouble with Fisher's notion comes in his definition of the value of money in purely relative terms as the _reciprocal of the price-level_, and his contention that the study of the value of money is identical with the study of price-levels.[353] Value is not a mere exchange relation.[354] Rather, every exchange relation involves _two_ values, the values of the two objects exchanged. These two values _causally_ determine that exchange relation. In the case of particular prices, then, we must consider not only the value of goods, but also the value of money. And the causes determining the general price-level will therefore include not alone the values of goods, but also the value of money. In the foregoing arguments by which I have shown that the price-level can vary independently of the other factors in the quantity theory scheme, I have been concerned only with changes in the values of goods, measured by a constant unit of value. If the value of money should also be varying, the concrete results on the price-level would have been different. On the face of things, there was nothing in the cases I discussed to require us to suppose that the value of money would also vary. The argument ran on the assumption of a fixed value of money. I have shown, in earlier chapters, that the assumption of a fixed value of money is fundamental to the laws of supply and demand, cost of production, and the capitalization theory. In point of fact, this assumption is rarely true--never strictly true. For causes which are in considerable degree independent of the causes governing the values of goods (as the causes governing their values are in considerable degree independent of one another), the value of money varies, now in the same direction as the values of goods in general, now in an opposite direction. Further, money itself does not escape the general laws of concatenation of values. The value of money has causes which are bound up with the values of other goods. Thus, when prices are rising and trade expanding, there is a tendency--commonly a minor tendency--for money also to rise in value, and so prices do not go quite as high as they would have gone had money remained constant. This tendency arises from the fact that there is more work for money to do in a period of active trade and rising prices. Gold also tends to rise in value in the arts, with prosperity. The reverse tendency manifests itself when prices are falling: money tends, in some measure, to fall in value with the goods,[355] and so prices do not fall as far as they would fall if money remained constant. But in general, the causes governing the values of goods, and the causes governing the value of money, are sufficiently independent to justify us in studying each separately, in abstraction, on the assumption that the other is unchanged. Hence, supply and demand, cost of production, and the other price theories, which assume a fixed value of money, are proper tools of thought for the study of the prices of goods.