hypothesis, perhaps permissible in the very short run, that it is uniquely determined by national income. The latter’s current value is by definition identically equal to current consumption plus current investment, all three quantities being expressed in wage-units. 9 And, with all the ‘givens’ implied, current value of national income may be said to be ‘determined’ by three functions or schedules that Keynes dignified with the title of ‘psychological laws’: 10 the consumption function, the investment function, and the liquidity-preference function, the three great simplifiers, which are to implement Keynes’s vision of the economic process, in particular the intention to prove the existence of underemployment equilibria and, to put it with perhaps inadmissible emphasis, his conviction that saving (or, alternatively, the rate of interest) holds the role of villain of the piece that impoverishes nations. 11 In a sense similar to that in which the Marshallian demand curve descends from Cournot (and, objectively, also from Verri), the Keynesian consumption function descends from Malthus and Wicksell 12 but received added pre- assuming the quantity of means of payment to be externally given, i.e. of being freely malleable by governments and central banks. This assumption brings us, all protests notwithstanding, dangerously near to a crude quantity theory of which an externally given quantity of money is, as we have seen, an outstanding feature. It would be un bearably unrealistic, even for modern England, unless, as Arthur Smithies has pointed out, we define the quantity to mean legal tender outside of banks plus the maximum of deposits the law and the ‘authorities’ permit banks to create. 9 Since savings are defined as the difference between income and consumption, this identity yields the familiar identity between current savings and current (rate of) investment. But for the latter identity to be valid, current investment must not in turn be identified, as has been done by Keynes, with the rate of production of new capital equipment. See on this P.A.Samuelson, ‘The Rate of Interest under Ideal Conditions,’ Quarterly Journal of Economics (February 1939), pp. 292–5. 10 They have of course no claim to this title, not even in the sense in which, at a push, it is admissible so to call Gossen’s law of satiable wants. 11 Those who are of the opinion that economic facts and Keynes’s analysis do not bear out this conviction and that the element of truth in it reduces to so much of it as had been recognized by J.S.Mill, W.Roscher, and A.Marshall (see Keynes’s and even Hobson’s grudging admission of this, op. cit. p. 19n.), will naturally look for a non-analytic explanation. They may find it, first, in England’s situation in which many difficulties were capable of being solved by the expropriation of ‘rentiers’ which, for political reasons, were practically insurmountable in any other way; and, second, in the kind of man Keynes was—he, the unattached intellectual, who abhorred bourgeois virtues but was much too civilized to like violent measures, had a not unnatural preference for the ‘euthanasia’ of the creditor interest. 12 There is, however, this difference between the cases of Marshall and Keynes. They would be exactly similar if the historian could say that Marshall saw his demand curve in the less rigorous presentation of Mill and added rigor and ‘edge’ to it. But Marshall must have found all the rigor and edge that anyone could desire in Cournot. Keynes, had he even been inspired by Malthus (Wicksell he then hardly knew), would have had still to do all that remained for a disciple of Mill to do in the case of the Marshallian demand curve. However, the (objective) affinity between Keynes and Malthus stands out with particular clearness at the beginning of the General Theory (p. 25), History of economic analysis 1142 cision at the hands of Keynes. As everybody knows, it represents current total national consumption (total expenditure on ‘consumption’ in terms of wage-units) as a function of current national income (in wage-units) and expresses the arbitrary postulate that any increase in the latter is always attended by an increase in the former but by a smaller one. 13 The investment function is less easy to convey in a few words because of its connection with the very important dynamical considerations of Keynes’s chapters 11 and 12, which do not enter into its explicit statement. It relates the rate of aggregate investment to the marginal efficiency of (physical) ‘capital in general which that rate of investment will establish’ (op. cit. p. 136), the marginal efficiency of capital being defined as the relation between the expected yield of one more unit (properly chosen) of any capital good and the cost of producing this unit. 14 This, as Keynes pointed out, is the same as Fisher’s ‘marginal rate of return where Keynes worked out his concepts of Aggregate Supply and Aggregate Demand Functions and of effective demand. Whatever weight we may attach to Keynes’s warnings, we are inevitably driven to considering these functions, which are capable of intersecting once (or several times) but are in any case not identical, as generalizations of the genuine concepts of supply of and demand for individual commodities. Keynes, aware of the pitfall, makes little use of this notion later on. But it is the Malthusian notion. And if valid it would of itself suffice to establish the possibility of an equilibrium of which full employment would not be a property. I repeat that the arguments that Keynes set forth against what he conceived to be the classical theory (in his sense) are entirely irrelevant against any correct statement of the full-employment equilibrium theory and that his indictment that the classical theory knows no unemployment except a frictional one is true only if the term frictional is defined so widely as to rob the indictment of all significance. 13 This ‘psychological law’ of the propensity to consume must of course refer to individuals. But the postulate in question refers to social aggregates. I have called it arbitrary merely to emphasize the fact that it formulates only one of several possibilities. We can express the consumption function by writing C=f(Y), then the postulate reads that the marginal propensity to consume, dC/dY, is always smaller than unity. But let us note at once that, since instead of this consumption function, we might just as well write a savings function, S=φ(Y). It has become usual with many Keynesians to insert into both functions a second variable, namely the rate of interest, i, the importance of this concession being minimized by postulating that the influence of i is negligible. 14 The investment function is usually written I=F(Y, i), which expresses the marginal efficiency of capital by the form of the function F. The marginal propensity to invest at a given rate of interest is then δF/δY. But we may also leave out the i in order to emphasize the cases where investment is ‘autonomous,’ i.e. either imposed upon the system by an external factor, such as government, or else entered into without any regard to current conditions. Or we may, on the contrary, consider investment as wholly ‘induced’ by consumers’ buying and then write the investment function I=φ(C, i) as has been done by Lange (see note 4 above). These and other expressions suggest themselves for the purpose of underlining this or that possibility but none of them, taken by itself, does full justice to the thought of Keynes, who wisely refrained from presenting any of them himself. Keynes and modern macroeconomics 1143 over cost.’ 15 But there is this difference between the two: whereas with Fishėr this marginal rate of return over cost—which implies a discounting process of the series of expected yields—constitutes the basic fact about the interest phenomenon, Keynes broke away at this point from what I have termed the Barbon tradition and, in intent at least, established a monetary theory of interest, according to which interest is not derived from, or expressive of, anything that has, in whatever form, to do with the net return from capital goods. 16 This brings us to the third of Keynes’s basic functions or schedules, the liquidity- preference function. In Chapter 13 of the General Theory Keynes seemed to accept the theory that makes the rate of interest ‘depend on the interaction of the schedule of the marginal efficiency of capital with the psycho- 15 Theory of Interest (1930), p. 168. I can, however, testify to the fact that Keynes, whose knowledge of economic literature and particularly of contemporaneous and non-English literature was not of the first order, arrived at his concept quite independently and that he inserted the acknowledgment in question upon his attention’s having been drawn to Fisher’s formulation. When he received the information, Keynes possibly acknowledged too much. Such, at least, is Professor Lerner’s opinion. On the other hand, it may be argued that both concepts are indeed improvements upon the concept of marginal productivity of capital as developed by Marshall and especially Wicksell—and this again points back to Böhm-Bawerk—but not more than that. The ‘prospectiveness’ of marginal productivity of capital and its relation to its replacement costs, few if any authors who used it would have denied. 16 Since this is, speaking from the standpoint of theoretical analysis alone, perhaps the most important original contribution of the General Theory, a few comments are in order. First, Keynes’s monetary theory of interest was subjectively original qua monetary theory of interest but not objectively so. From the scholastics through their Protestant successors to several pre- Keynesian modern writers, explanations of the interest phenomenon have been offered that link it to money and whose authors would all agree to the divorce that Keynes pronounced between the yield of non-monetary capital and interest. The objective importance of Keynes’s work, so far as this goes, was the success that his teaching met with: he actually converted a large number of fellow economists who twenty years or so before considered a monetary theory of interest hardly worth serious attention. Second, we must again recall the fact that Wicksell, without adopting a monetary theory, made such an important stride toward it as to inspire Swedish followers of his to do so. This Wicksellian line of advance is most easily accessible, for the English reader, in Professor Erik Lindahl’s Studies in the Theory of Money and Capital (1939) and Professor Bertil Ohlin’s two articles ‘Some Notes on the Stockholm Theory of Savings and Investment,’ Economic Journal, March and June 1937. The articles gave rise to a discussion between Keynes, Ohlin, Robertson, and Hawtrey that was followed by a number of articles by other economists. But we must be content to add, third, that the particular form that Keynes gave to his monetary theory was original both subjectively and objectively. It may differ much or little from the Swedish one or from the one sponsored by Professor Hicks (Value and Capital, ch. 12), which perhaps comes nearest to holding the field in that part of Anglo-American literature that accepts the monetary theory of interest at all; but it differs a great deal from other forms—so much so that it is a mere question of temperament whether one wishes to see in them any affinity at all with the Keynesian form. History of economic analysis 1144 logical propensity to save’ (time preference). For he stated as his only objection that it is impossible to deduce the rate of interest merely from these two factors bcause it will also depend on the form in which the saver wishes to hold whatever he saves. Having decided how much he will ‘reserve in some form of command over future consumption’ (p. 166; note the classical ring of this phrase), he has still to decide whether and to what extent he will part with immediate command for a specified or indefinite period, that is, on his liquidity preference. 17 On the face of it this clearly amounts to not more than an amendment. Later on, however, even in the General Theory, Keynes himself and still more some orthodox followers of his, especially Professor Lerner, 18 went much further than that in the direction of the propositions that interest is nothing but a payment for overcoming one’s reluctance to part with the one ideally liquid asset in existence (own- rate theory of interest) and that the quantity of money, considered relatively to the amount of it that is absorbed by transactions, is the sole directly governing factor in its determination. 19 Current saving and current investment, being identically equal, cannot determine anything. Planned (ex ante) saving and planned (ex ante) investment determine income (total net output) but not interest. And a number of paradoxes follow for which some verification can be found in the freakish situations of deep depression. 20 17 In the exact formulation, this liquidity preference is usually introduced in an equation of the form , which compares the available amount of money (see note 8 above)—I bar the M in order to indicate that it is given—with a ‘demand’ for money that is partly determined by the volume of transactions, represented by Y, and partly by people’s expectations about the future behavior of the various interest rates (the ‘speculative motive’), which is represented by i. 18 On Professor Lerner’s argument see Franco Modigliani, ‘Liquidity Preference and the Theory of Interest and Money,’ Econometrica, January 1944, p. 79. I take this opportunity to recommend this paper as a general commentary on this whole range of questions. 19 Observe that, among many other things, interest must indeed also equalize the advantages of holding cash and other assets. This is another instance of Keynes’s Ricardian way of reasoning: the fact that the rate of interest must be such as to compensate savers for their marginal ‘abstinence’— if it did not, it could not be what it actually is—is obviously insufficient to establish the abstinence theory of interest. Observe further that, like everything else, both propositions—the own-rate and the abstinence theory—can be made formally true by a sufficient number of ‘givens,’ with the added advantage that the tables can be turned upon the objector with the utmost ease on the ground that he does not understand the assumptions of the argument. See in this connection W.Fellner and H.M.Somers, ‘Alternative Monetary Approaches to Interest Theory,’ Review of Economic Statistics, February 1941. 20 To mention one example: the Keynesian theory, taken literally, yields the conclusion that an increase in the inducement or the propensity to invest, or in the propensity to consume, will only increase employment but have no tendency to raise the rate of interest. The opposite in any normal situation is evident and has been stated as a theorem by Professor Samuelson, not as an objection to Keynesian doctrine but as part of it (see Foundations, p. 279, and compare J.R.Hicks, ‘Mr. Keynes and the “Classics”; A Suggested Interpretation,’ Econometrica, April 1937, pp. 152–3). Keynes and modern macroeconomics 1145 [3. THE IMPACT OF THE KEYNESIAN MESSAGE] By means of those three basic functions or schedules a system of three equilibrium conditions (equations) and one identity can be written that will, with the quantity of money as an externally imposed datum, and under proper assumptions, uniquely determine interest, investment, and either savings or consumption and can be extended to include also other variables such as Keynes’s wage rates. 21 But it was not this exact and crippled rendering of Keynes’s message which fascinated, but the resplendent whole of it. Particularly in its bearings upon saving, interest, and underemployment, this message seemed to reveal a novel view of the capitalist process not only, as we saw before, to the public and ‘writers on the fringes’ but also to many of the best minds in the sphere of professional analysis—a novel view that was as attractive to some as it was repellent to others. 22 This created almost immediately an atmosphere 21 The identity is either Y≡C+I or S≡I. In the first case we may use the equations C=f(Y, i) and I=Φ(C, i), in the latter case the equations S=φ(Y, i) and I=F(Y, i). For an extension see Modigliani, op. cit. p. 46. But the system, especially any extended system, is not so simple a matter as it might seem to the layman. This is the reason why, though I cannot go into the problems involved, I have tried to save my conscience by inserting the words, ‘under proper assumptions.’ It is not difficult to draw up a system that will display inconsistencies and fail to define an equilibrium or even multiple equilibria. This is important to observe because such disequilibrium systems play a role in the Keynesian discussion: for the Keynesian they may be a means for showing that, without government expenditure (‘fiscal policy’), the economy may be incapable of hitting upon an equilibrium state, in particular a full-employment equilibrium state. Let us note, in passing, another important point. If we speak not of current but of planned consumption and investment, then a condition of stability of the system is that the sum of the marginal propensity to consume and the marginal propensity to invest—the marginal propensity to spend—be smaller than unity. If it is equal to or greater than unity, the system will still be determined but it will ‘explode’ instead of converging toward equilibrium when displaced. Now several writers are prone to argue as if this stability condition could be used to ‘prove’ that the propensity to spend is actually smaller than unity because, so they hold, in capitalist reality the economic system does not explode. This argument is quite inadmissible on logical grounds of which I mention only one: a short-run theoretical system may be explosive while the corresponding long-run system is not; and a long-run theoretical system may be explosive while the corresponding reality is not. 22 The division of professional opinion cannot be described in the same terms for every country: in some it amounted to not more than a ripple on the surface. But in England and the United States it went deep, and here a phenomenon asserted itself unmistakably that deserves passing notice. Keynesianism appealed primarily to young theorists whereas a majority of the old stagers were, more or less strongly, anti-Keynesian. One aspect of this fact is too obvious to detain us and has, in addition, often been emphasized: of course it is true that part of the resistance which every novel doctrine meets is simply the resistance of arteriosclerosis. But there is another. The old or even mature scholar may be not only the victim but also the beneficiary of habits of thought formed by his past work. I am not referring now to that deeper understanding History of economic analysis 1146 that was ideally suited for a struggle full of zest—as much so, in principle, as was the atmosphere created by Ricardo in 1817, but more so, in fact, owing to the temperature produced by the vastly increased number of professional economists. All that can be done in this sketch is to list the three types of tasks that were undertaken and together account for the torrent of more or less Keynesian literature that is so characteristic of the decade after 1936. The first task of course proceeded from the need felt by almost every economist to find out and to tell how he stood in relation to a message that nobody could ignore. The bulk of the profession’s work went on as usual and was but little affected by that message. But for all theorists, general economists, and workers in the fields of money, banking, and business cycles, that need could be satisfied only by laborious analysis, criticism, development. Since we cannot survey the literature of this type satisfactorily, 23 we merely note two facts. The one is that to have created such a response is, in and by itself, an achievement, frank recognition of which is the greatest and most deserved of the compliments that may be paid justifiably to the memory of Lord Keynes. It was not the analytic performance which did it; nor was it the attraction of the practical issues raised. As in the case of Ricardo, it was the intellectual performance spiced by the—real or putative—relevance to burning questions of the time which achieved what, in our field, neither could have achieved by itself. The very blemishes of the intellectual performance and the very objections that may be raised against Keynes’s practical answer were instrumental in bringing about spectacular success and in extending controversy over the whole field that lies between, and includes, recommendations and purely logical questions of method. The other fact is the cumulative property of success of this kind, which can be best conveyed with reference to teaching. Any successful work of scientific standing must be mentioned in courses on the subject to which it is relevant. But a teacher, as soon as he discovers that students will take to a work independently of his teaching or can be trusted to have become acquainted with it before they entered his course, will also discover the pedagogical advantages to be reaped from referring to, and building upon, such previous knowledge; and he will, whatever his own opinions, deal much more intensively with such a work than he would merely on its merits. Thus as in banking or insurance, growth induces further growth merely by increasing reserves, success engenders success. Literature produces further literature. The second task that the General Theory presented was the development, critical or constructive, theoretical or factual, of a large number of individual of things that can hardly be acquired except by the labor of decades: apart from this and the difference in attitude to ‘policy’ that results from this, there is such a thing as analytic experience. And in a field like economics, where training is often defective and where the young scholar very often simply does not know enough, this element in the case counts much more heavily than it does in physics where teaching, even though possibly uninspiring, is always competent. 23 A sample of it, though heavily weighted ‘in favor,’ the reader will find in The New Economics, edited and introduced by S.E.Harris (1947). Keynes and modern macroeconomics 1147 points. 24 There were the questions of Keynesian underemployment equilibrium, of the ‘own-rate’ versus the ‘loan-fund’ theory of interest, of the principle of aggregative (macroeconomic) theory, of the relation between money and real wages, and many others, all of which produced ‘special literatures’ of their own. But one example must suffice, the work that has been and is being done on the consumption function. No theorist worthy of the name can accept as an exact statement the postulate that links expenditure on consumption (in terms of wage-units) with income (in terms of wage- units) alone. Still less is it possible to accept the Keynesian property of this function (dC/dY, see note 13 above) as universally valid. We have therefore an approximation before us. But how close is this approximation and, in particular, precisely how imperative is it to add a term to allow for shifts of the function in time? And how seriously are we sinning if we decree that the function be linear? Or must we take in sails and admit independent variables other than income—for instance, the amount of assets or at least liquid assets that individuals happen to have already? All these questions are theoretical questions in the first instance to be answered with reference to the autonomy of the function 25 and to its consistency with other relations that we mean to accept on the same plane of argument. But evidently they have also a most important factual aspect. 24 It should be observed that, so far as effects upon the content of a theory are concerned, criticism or elaboration and even apologetics come to much the same thing: irrespective of the worker’s intention, his work gradually changes, and in the end annihilates, original meanings. But this is not so as regards the renown of a work and the position it will ultimately occupy in the history of a science. Here, the worker’s attitude and value judgments are much more important, even for the opinion of future theorists but, of course, still more so for the future opinion of the profession and the reading public. For instance, it would be easy to compile a list of arguments (all of them valid) from the writings of, say, Hicks, Lange, Modigliani, and Samuelson that in hands less friendly than theirs would sum up to a very damaging criticism. But they had no intention to damage. In Keynes’s case, merit and luck combined to blunt the edges of the criticism of some of those who were most competent to inflict injury—compare Marshall’s attitude to Ricardo. 25 The concept of the autonomy of a function or equation is due to Professor Frisch. In a system of relations (mathematical or not) that are supposed to hold simultaneously in a given framework of data, there may be some that hold individually only if the others do and perhaps also if the given framework of data remains unchanged, and others that retain individual validity even if some do not hold (and in another framework of data). The latter we call autonomous, though we use this term also in other senses (as in ‘autonomous investment’). The property is not absolute: a relation may be more or less affected by a failure of the others. Therefore we had better speak of higher or lower degrees of autonomy. Frisch’s paper on the subject (not published to my knowledge) is one of the most interesting contributions of our time to the pure logic of modern theory. [In reply to a query by the editor, Professor Frisch stated that the idea of autonomy of a function or equation is explained at great length in several of his mimeographed lectures in Norwegian and that in printed form it is mentioned only briefly in a note ‘Repercussion Studies at Oslo,’ American Economic Review, June 1948.] History of economic analysis 1148 And it is not surprising but a matter for congratulation that a dozen or so econometricians have devoted, and are devoting, attention to it. The third task springs from the necessity of ‘dynamizing’ the Keynesian system either on the lines suggested by Keynes himself or on others. This necessity became obvious as soon as people began to ‘work’ the Keynesian system seriously, for, as we know, even the mere question of the stability of a static system quickly leads into dynamic considerations. But in addition many Keynesians set about introducing into their models the usual ‘dynamizers,’ especially lags. As examples I mention Professor Smithies’ model 26 and then again the Hansen-Samuelson equation which we have already met. Thus, Keynesian equilibrium analysis gradually gave way to Keynesian ‘process analysis,’ and at present this Keynesian process analysis tends to merge with the older and broader macrodynamics, the development of which we have glanced at before. Here, at long last, we are at the point from which it is possible to define and locate the historical importance of Keynes’s purely analytic contribution to economics. This being important and, owing to the brevity of our exposition, not easy to grasp, the reader’s attention is requested for the following résumé. So far as the exact core is concerned, Keynes’s system is essentially static. This static theory sufficed for the purposes he had most at heart, particularly for his doctrine of underemployment equilibrium. However, partly because it was inevitable that he should have had to add dynamic considerations to that core, partly because his work impinged on a situation in the field of pure theory that was dominated (independently of him) by the novel interest in macrodynamics, this macrodynamics absorbed his work. But, owing to the position Keynes’s work conquered in the thought of the profession, it was not simply swamped by macrodynamics but in turn helped to mold and to propel the latter— for which Keynes’s model was particularly qualified by virtue of its simplicity. Professor Hicks was obviously right in saying that ‘the General Theory of Employment…is neither the beginning nor the end of Dy- 26 Arthur Smithies, ‘Process Analysis and Equilibrium Analysis,’ Econometrica, January 1942. It has been pointed out already that most Keynesians (or writers who use the Keynesian or a similar apparatus) introduced planned (or ex ante) savings and investments so as to be able to make equality of savings and investments an equilibrium condition instead of the identity it is in the case of savings and investments actually performed. This is in agreement with Keynes’s position for he surely emphasized the gulf that exists between saving and investment decisions strongly enough. Nor does it, in itself, involve leaving the precincts of Keynesian statics. But it does so as soon as we connect savings and investments explicitly with some quantity of the past, e.g. with yesterday’s income. And then we are led away easily, though not by logical necessity, not only from Keynesian statics but also from the Keynesian structure as a whole. Take for instance the concept of idle savings. Laymen sometimes believe that the Keynesian argument implies that there must, somewhere in the economy, exist savings that are idle in the sense that they are not being invested. But this notion is meaningless within the Keynesian argument. However, it immediately acquires meaning if we introduce lags. Nor is it difficult to tell where the introduction of lags tends to lead us, if it leads us away from Keynes: it leads us toward Robertson and Lundberg. Keynes and modern macroeconomics 1149 namic Economics.’ 27 But it is also true that, unintentionally and perhaps even against his will, 28 Keynes gave a mighty impulse to it—almost all work in macrodynamics now starts from a ‘dynamized’ form of his model. In a history of analysis this is the point to stress. 29 In a history of economic thought Keynes’s policy recommendations—time- bound as they were—and certain characteristically Keynes’s doctrines—which are losing their hold already—may be much more important. [The manuscript breaks off at this point; there are brief notes, partly in shorthand: ‘Other points to be added… Macroeconomics will need a new conceptual apparatus …new general objects…multiplier…accelerator…’] 27 ‘Mr. Keynes and the “Classics,”’ Econometrica, April 1937, p. 159. 28 ‘Forget all about periods,’ he once said to a pupil. 29 An interesting instance of a macrodynamic model of inventory cycles that makes use of Keynes’s consumption function and thus illustrates well what I was trying to convey is Lloyd A.Metzler, ‘The Nature and Stability of Inventory Cycles,’ Review of Economic Statistics, August 1941. History of economic analysis 1150 Editor’s Appendix THIS APPENDIX is written for the specialist who is interested in the order in which the various parts of the History were written and to what extent they were completed. I have already touched briefly upon these problems in the Editor’s Introduction and in editorial notes (in square brackets) throughout the book. The ordinary academic reader will find everything he needs to know in the introduction. As I stated there, everything had been written out in long-hand originally; some of the chapters had been written early and rewritten later; most of them had been typed and corrected in pencil by J.A.S.; a few sections had been typed toward the end and the typescript had been read very hastily or not at all; and, finally, there was some material still in manuscript. There were even alternative versions of some of the manuscript in the chapter on Equilibrium Analysis (ch. 7 of Part IV). The reader is reminded that the original manuscript, the alternative versions, many bits of discarded manuscript, the notes (some of them on those little pieces of yellow paper with which every student and close associate was so familiar), and the first typescript with corrections and suggested revisions in the hand of J.A.S., will all be deposited in the Houghton Library at Harvard University, where they may be consulted by the interested scholar. Although I attempted to present as complete and accurate a version as possible of what was actually written, there are undoubtedly some places where a different interpretation would be possible. It is, therefore, a great source of satisfaction to me that the original manuscript and the notes for revision will be available in the Houghton Library. In my anxiety to show that the author would not have sent his History to the publisher without further work upon it, it may be that I have overemphasized the degree of its incompleteness. In reality, the History was substantially finished. The three main Parts (II, III, and IV) needed a little polishing here and there, a few additional pages to complete some of the sections, a few titles and subtitles, and, of course, the references needed to be checked. The chapters on The ‘Mercantilist’ Literature (Part II) and on Sozialpolitik and the Historical Method (Part IV), which date from the early period, would have been revised. This was also true of the section on Senior’s Four Postulates in Chapter 6 of Part III. A little more work remained to be done on some of the sections in the chapter on Equilibrium Analysis. On the whole, however, Parts II, III, and IV were all but completed. As for the rest, the introductory Part I and the concluding Part V were being written at the very end and were somewhat less complete. These two parts, however, were distinctly subsidiary and not absolutely necessary to the main plan, the divisions of which were based on those of the Epochen der Dogmen- und Methodengeschichte of 1914. Part I was to be a brief exposition of methodological problems of which all but the last two sections of the final chapter had been written; Part V was to be an equally brief treatment, relating the present state of economics to the work of the past as described in the three main parts. The photostats of the plans of Part V (which appear below in this appendix) indicate that perhaps two-thirds of the conclusion . beneficiary of habits of thought formed by his past work. I am not referring now to that deeper understanding History of economic analysis 1146 that was ideally suited for a struggle full of zest—as. were the questions of Keynesian underemployment equilibrium, of the ‘own-rate’ versus the ‘loan-fund’ theory of interest, of the principle of aggregative (macroeconomic) theory, of the relation. in macrodynamics now starts from a ‘dynamized’ form of his model. In a history of analysis this is the point to stress. 29 In a history of economic thought Keynes’s policy recommendations—time- bound