CONTEMPORARY governments appear to J have embraced too easily three incompatible goals; full employment, prices, and an inflexible exchange rate. More sensible policies could be pursued if one or possibly two of these goals were abandoned. Clearly, the least fundamental of the three is the foreign exchange rate, and if it could be shown that a flexible rate is not all that damaging, the step would be made much easier. The central arguments against a variable rate are that it would discourage trade and lending, and that it would encourage noxious speculation. The force of both these objections is lessened if the market can be satisfactorily stabilized. A simple procedure for the market will be discussed in this note. Since this problem of flexible rates was first discussed in the thirties, notably by Professor Harris, a great deal of experience has accumulated to show how extraordinarily effective feed-back control methods are for a great variety of mechanisms. If it could be shown that there exist reasonably simple and successful exchange rate stabilization policies, then a more rational discussion of policy alternatives should be possible. I shall assume a free foreign exchange market with complete convertibility, the exchange rate being set and maintained by a monetary authority which makes up any deficit and absorbs any surplus in foreign exchange at the announced rate. The aim is to stabilize this market in the sense that the rate may bend but will not break. Yet there is a fundamental ambiguity in the meaning of which we encounter in the question of stabilizing the market. Technically, by degree of we mean the rapidity of approach to equilibrium (defined here as equality of supply and demand). Thus with a given amount of disturbance, the more a market is the closer it will tend to be to its equilibrium. However, for a market with shifting supply and demand curves, the equilibrium price will hop about, so that the more stable, in this sense, a market is the more agitated its price will be. This contradicts our commonsense notion of stability and, what is much more important, is normally undesirable. In fact we want the market, as dominated by the authorities, to be markedly sluggish, to be stable in the popular sense and not very stable in the technical sense. More specifically, I assume that ideal behavior for the exchange rate is to be insensitive to short-run fluctuations in supply and demand. Thus the authorities are to try to equate supply and demand but only in the long run, not in the short run. They are to use the behavior of the market itself as a guide to altering the rate to accomplish this purpose. Consequently they will be acting on sound feed-back principles; technically they will be simulating a zeroing servo, that is one that aims to reduce to zero the difference between supply and demand. They will never quite accomplish their object, of course, since new disturbances are always disrupting their course. Given these aims the authorities are to alter the exchange rate continually in the light of two criteria, (1) the difference between actual and desired holdings of foreign exchange, and (2) the current payments gap. They must so weight their reactions that the former outweighs the latter, which it will naturally tend to do since it is the cumulated current payments gap. If they always alter the rate so as to push actual foreign exchange back towards the desired level, they will, in the long run, tend to maintain the level of foreign exchange constant and hence equate supply and demand. A more sophisticated approach could take the desired level of foreign exchange, m*, as variable, say, with the average domestic demand for foreign exchange. Purely for simplicity I shall treat m* as a constant. Our basic control routine for the exchange rate p, is then