Perry Anderson

  • A Treatise on Social Theory. Vol. II: Substantive Social Theory by W.G. Runciman
    Cambridge, 493 pp, £35.00, February 1989, ISBN 0 521 24959 7

Under a flat, anonymous title and in serial guise one of the most exotic – even flamboyant – intellectual projects of recent years is coming to fruition. The first volume of W.G. Runciman’s Treatise on Social Theory, devoted to the dry topic of methodology, set out in reasonable and moderate tones an agenda for social understanding combining – in so many words – ambitions of a Ranke, a Comte, a Proust and a Hart: to report accurately, to explain scientifically, to re-create imaginatively, and to judge impartially and benevolently. Perhaps the most striking feature of this programme was its association of two aims normally reckoned antithetical: an explanatory structure continuous with the natural sciences and an imaginative recapture modelled on literary fictions. Few practitioners of the social sciences have the confidence to invoke the ideals of Herbert Spencer and Henry James simultaneously.

In his second volume, Runciman proceeds to his own substantive social theory. He remarks at the outset that he will here be concerned only with the explanatory segment of his overall prescriptions. But this involves no shrinkage of horizon – on the contrary. For the two questions he seeks to explain are nothing less than these: what kinds of society are possible at any given stage of human development, and why any given society became the kind it did, in the course of it. The answers he arrives at, Runciman goes on to argue, have profound intellectual consequences, offering ‘a very great deal’ more for our general understanding of history than can be found in the contributions of Marx or Weber, or anyone else. For Volume II of A Treatise on Social Theory proposes a novel account of social structure and an original theory of social evolution. It is the integration of these two conceptions, he contends, that marks a fundamental advance in our ability to grasp the pattern of the past.

What is the import of each? For Runciman, power is as basic a concept of sociology as energy is in physics. Societies are to be conceptualised as so many different ways of allocating power. Such power comes in three and only three sorts: economic, ideological and coercive. These are always interdependent, but none is ever reducible to the others. Runciman’s principal early debt was to Weber, from whom this axiom derives. He reformulates it, however, in terms designed to effect a bridge to Marx. Societies are modes of distribution of power, founded on differential control of the means of production, persuasion and coercion. It is the variations in these three dimensions of power that furnish the key to a scientific classification of societal forms. At the same time each of them is composed of a specific set of social practices; and these practices are perpetually subject to selective pressures, generated either within or between the societies concerned – above all, from the competition of alternative practices. The process of this selection in turn constitutes the essential mechanism of social evolution.

Runciman’s calm, considered commitment to an evolutionary theory of history is a gauge of his independence of mind. Nothing could be less in fashion today. For some time now evolutionism has been a disreputable term for many sociologists: one of the cardinal, egregious errors of earlier generations, since generally repudiated. The last major thinker to attempt a systematic theory of social evolution was Talcott Parsons, in his final years. It is significant that Runciman, amidst a vast bibliography, never mentions him. Moreover, the discredit into which evolutionism has fallen has attached principally to variants of Marxism which assert no more than some directionality to historical development. Runciman’s evolutionism, however, is far more specific and stringent. It is a sociology literally sculpted after modern biology. Natural selection serves throughout as the model of social selection. Runciman is well aware of the fate of previous versions of social Darwinism. But he is undeterred by the failure of every previous bid to map The Origin of Species onto the genealogy of social forms, because he believes he has corrected the root mistake common to them. All such theories took as their unit of selection – the basic materials out of which only the fittest survived – either whole societies or social groups. Runciman substitutes particular practices. These, he maintains, are the true counterpart of genes. Like DNA, their mutations are random, in the sense that their origins are extraneous and irrelevant to the process of selection itself. The bodies on which they confer advantages in the competition for power are roles, that are in turn attributes of social groups or societies, which form the equivalent of species.

Armed with this analytic equipment, Runciman then directly broaches the historical record. His purpose is twofold: to establish a Linnaean taxonomy of all societies known to us, and to demonstrate the Darwinian mechanisms of their speciation – how and why successive forms evolved from their predecessors. The result is a dazzling display of erudition. On this showing, Runciman’s command of the comparative historical record has few, if any, rivals. Old Babylonia, tribal Africa, archaic Greece, Pre-Columban America, Stone Age Melanesia, Classical Rome, Dark Age Lombardy, Medieval Japan, Imperial China, feudal Poland, republican Venice, caliphal Islam, absolutist France, industrial Britain, revolutionary Mexico, Stalinist Russia, populist Argentina, social-democratic Sweden, racist South Africa – all these and many more parade across what astonishingly remains a compact, middle-sized book, each deftly and economically captured for the purposes at hand. There are a few gaps: Egypt or Assyria from the ancient world, the Third Reich from the modern; and the sources drawn upon could be thought too consistently Anglophone (about 90 per cent of the modern citations – certainly more than the balance of scholarship). But these are trivial limitations. Runciman handles the enormous range of his evidence with a precision and assurance that are deeply impressive. Vaulting in global ambition, his survey is unfailingly sober and careful in local execution. It is difficult to think of a single obvious lapse or extravagance of historical judgment in the whole account, where stray instances could readily be pardoned. An even, scrupulous, dispassionate voice controls what might otherwise appear an inordinate enterprise. The tone is one of imperturbable suavity.

How should the completed theory be assessed? The comparative testing of Runciman’s hypotheses yields a historical inventory of great richness and fascination. But there are a number of difficulties with the theoretical framework itself. The first of these concerns Runciman’s preliminary move – the definition he adopts of society itself. Can societies be satisfactorily envisaged simply as so many intertwined networks of power? Runciman is not alone in thinking they can. The same assumption can be found in the work of Michael Mann, of which Runciman has been a severe critic, but whose scale and focus invite comparison. The common source of this bias is Weber – the dominant influence on this cohort of British sociologists. Fixation with power has, of course, gone much further elsewhere, in France and the United States, under the spell of Foucault. The excesses and absurdities of the metaphysic now current in much literary theory, and even cultural history, are generally foreign to the more prosaic world of English social science. But they have been warned. Whether in its Weberian or its Foucauldian version, engrossment with power appears hard-headed: in fact, it is naive. (Weber’s delusions over Ludendorff, Foucault’s over Khomeini, are the fitting emblems of each.) Societies are not just power-stuff. Three very large domains of collective life resist such ingenuous reduction. These are the production of persons, of goods and of meanings. Demographic, economic and cultural systems are never mere transcriptions of power relations between human actors: for they always involve transactions with nature that surcharge or overflow these. The sociology of power characteristically seeks to avert this objection by extending its central concept, in a direction originally indicated by Parsons, towards positive-sum usages. Mann, for example, distinguishes between ‘distributive’ and ‘collective’ power – the former exercised by some agents over others, the latter shared between agents as a common enhancement of their capacities. For Runciman, similarly, the notion of power encompasses both ‘domination’ and ‘co-operation’. This doubling of the term is not a mere artifice. Technological invention or economic progress can legitimately be entered under it, since these directly involve an increased social power over the natural world. But this is not true, for example, of most recorded demographic regimes, of major world religions, or significant art forms. It is no accident that they virtually disappear from Runciman’s conspectus. Even economic activity, in principle more amenable to his treatment, gets short shrift. Indeed, it is noticeable that he makes much less of his ‘co-operative’ than Mann does of his ‘collective’ forms of power. After a perfunctory initial mention, they are all but forgotten thereafter. Symptomatically, Runciman – himself a leading industrialist, who must know more about modern economic realities than most of his peers put together – can nevertheless write that ‘the fundamental economic practice is the exchange of goods or services’: as if production did not exist. The ellipse is dictated by the parti pris of the theory as a whole, where to all intents and purposes power is ‘distributive’ domination.

The same inflection recurs in Runciman’s account of the ideological dimension of social structure. Once societies are conceptualised as so many practices of power, competition between these must be anchored in the dispositions of the individuals who compose them. Runciman postulates universally effective, and equivalent, strivings for the possession of wealth, force and prestige. The pursuit of ideological power is a quest for the latter, and the ‘fundamental practice’ of ideology is deference. Noting that some might doubt whether the wish for prestige is really a historical force on a par with the desire for command or riches, he replies that honour is a profound and ubiquitous value, Locke even holding ‘credit and reputation’ to be the most important of all springs of human action. But this is a non-sequitur. For reputation, which can just as well be moral, intellectual or aesthetic, need have – normally does have – nothing to do with power: of Locke’s contemporaries, what ideological sway was exercised by Aubrey or Spinoza or Vermeer? Even where real ideological systems are at work, moreover, in such massive structures as the great religions, is their principal function just to instil social deference? Plainly not: Christianity as a cultural meaning-complex cannot be reduced to the mere authority of priests. Its world-historical role, across successive forms of social order, is not to be squeezed into any such straitjacket. Runciman is right to distinguish between economic, ideological and coercive forms of power, and to insist that the variable relations between them need to be studied empirically, in their different historical settings. But he is wrong to suggest that history itself consists simply of their interplay. Neither labour nor faith, to say nothing of birth or death, are ever just adjuncts of power; and the changes in their regimes – affecting technology, reproduction, belief – cannot simply be assumed to have less causal weight for social evolution than the practices of domination. That has to be shown.

The strains that Runciman’s restrictive optic impose on his own enquiry become apparent as soon as he moves to develop a concrete taxonomy of societal forms. Since these are identified with modes of distribution of power, each a specific combination of separate modes of production, persuasion and coercion, their number should be determined by the range of variation of the latter. Runciman proceeds to enumerate these: eight forms of economic power (serfdom, tenancy, small-holding, debt nexus, corvée, caste, wage-labour, slavery); eight of ideological power (purity-pollution, hereditary status, sacred rank, ethnicity, age-set, genealogy, occupation, charisma); and seven of coercive power (conscript army, warrior aristocracy, civilian militia, magnate levies, servile troops, volunteer professionals, alien mercenaries). That yields a Linnaean grid of some 450 possible kinds of society, as he points out – without considering hunter-gatherer bands or communities prior to the emergence of a state. He then immediately adds that empirical overlaps are possible in each column, and that not all combinations are conceivable, so that the historical gamut of realised forms will necessarily be less. The proviso is reasonable enough. But nothing in it could prepare one for the eventual figure Runciman arrives at: no more than ten distinct kinds of society in all, from the dawn of civilisation to the present. (Another four, prior to the emergence of the state, are later subjoined.) In other words, less than 3 per cent of those theoretically called are empirically chosen. The discrepancy between the potential and actual typology is so large that it cries out for explanation. In effect, either the initial classification of variants is misguided and redundant, or there is some intermediate mechanism capable of a drastic transformation of profusion into parsimony. Runciman, however, does not appear to register the anomaly of his own conclusions. Here, if anywhere, the issue he announces at the outset – what kinds of society are historically possible? – is unambiguously posed. But the promise of an answer is not kept, the question itself seemingly forgotten.

There is a further anomaly within the taxonomic outcome itself. For the 14 separate forms ultimately classified as discrete modes of distribution of power are themselves aligned in a developmental sequence which commands their order yet remains external to their identity. Runciman, in effect, lists them as follows: limited-power; dissipated-power, shared-power, obstructed-power; patrimonial; citizen, warrior, bureaucratic, feudal, bourgeois; liberal-democratic capitalist, authoritarian, state socialist, racist. Each of these kinds of society is defined in terms of the configuration of power within it. But at the same time each is de facto situated within a larger set organised on a very different principle.

Limited-power obtains in hunter-gatherer societies. Dissipated, shared or obstructed power is characteristic of horticultural, nomadic pastoral or primitive agricultural societies. Citizen, warrior, bureaucratic and feudal forms belong to agrarian civilisations proper – bourgeois to early commercial capitalism. Liberal-democratic, authoritarian, state socialist and apartheid societies are specific to industrial nation-states. These correlattions are not merely deducible from Runciman’s account: they are conceded by him, in so many terms. In effect, behind the apparently static classification of power lies a surreptitious dynamic based on an economic periodisation. There is no reason to quarrel with the latter. But its discreet presence inevitably throws the rationale for the taxonomy itself into question. Can human societies really be conceived just as patterns of power, if their overall development answers to another – presumably a more fundamental – logic? Should not a theory of social evolution address itself in the first instance to the series of basic historical modes of livelihood, rather than to the various crenellations of power supported by them? Another way of saying this is that while Runciman’s taxonomy purports to classify societies, its operative units are for the most part actually states. This is the level at which his concentration on power is in fact appropriate.

Here, however, the denomination of his distinct ‘botanical’ species is not wholly consistent. Having argued that any given mode of distribution of power is always composed of separate modes of production, persuasion or coercion, which can vary independently of each other, Runciman goes on to say that a classification of societies along these lines must therefore take a Linnaean ‘polynomial’ form – i.e. spell out the particular combination of economic, ideological and coercive power in each case, as in ‘liberal-democratic-capitalist’ to describe Western societies today. But this rule is waived in the larger part of his enumeration. No attempt is made to apply it to early proto-states or semi-states. The five pre-industrial state forms are captured not by polynomial categories, but essentially by designation of their dominant strata (Runciman calls these ‘systacts’ to avoid pre-emption of their character): citizens, warriors, lords, bureaucrats, bourgeois. This is certainly one reason why the number of forms drops so precipitously – the criteria for them have been tacitly reduced. It is only with the industrial species that polynomials come into play, yielding such alternatives as ‘authoritarian-nationalist-capitalist’ and ‘liberal-democratic-socialist’ (nowhere yet realised, but in Runciman’s view a possible future for Sweden).

A certain casual ease with his own formal scheme is in another sense an attractive quality of Runciman’s writing. There is no trace of the manic attachment to pet minutiae so often characteristic of those with a taste for taxonomy. On the contrary, what is striking is the largeness of mind with which Runciman puts his categories to work – depth of learning, one might say, precluding rigidity of construction. His actual depictions of the many societies that he discusses to illustrate his argument are a chief instruction and pleasure of the book. If they stimulate disagreement on occasion, it is nearly always because of their placing within the logic of classification, rather than any error in their representation. How important are the demurrers indicated on this score?

If we start with the transition to statehood itself, there is one curious spot of uncertainty within Runciman’s account. He describes the earliest proto-states as ‘patrimonial’ – rudimentary forms in which those who hold civil or military office stand in a purely personal relationship to the ruler. As such, patrimonialism does not figure in his five-fold typology of pre-industrial constellations of power, species he locates on a plane ‘over and above’ it. Categorised as a simple beginner, it would appear to be structurally – in general also chronologically – superseded by the advent of these higher forms. But later on patrimonialism re-emerges as a state form proper, of notable range and potential sophistication, including such diverse structures as the Greek tyrannies and the Carolingian Empire, not to speak of the Tudor monarchy or the Portuguese kingdom in the Age of Discoveries. These contradictory emphases are never really reconciled. This may have something to do with the fact that some of the Hochkulturen often regarded as archetypes of the patrimonial state – Pharaonic Egypt or Shang China – are among the few major historical experiences Runciman bypasses in his survey. The exception is Hammurapi’s Babylon, to which he devotes special attention. But the purpose of this is to argue that Old Babylonia was a kind of sport: an unclassifiable hybrid of different forms of the distribution of power, whose only counterpart was – Anglo-Saxon England. The two societies are judged sui generis on the grounds that ‘they cannot be fitted into any of the five modes just listed’. But why should they not be assigned to the patrimonial mode, to which Runciman elsewhere allocates Sumerian, Lombard and Merovingian societies, neighbours in time and space? The only reason, discernible from a single stray aside, seems to be that for Runciman the royal administrations of Hammurapi and of Edward the Confessor exceeded those of a patrimonial ruler. This is scarcely persuasive. Was the staff at the disposal of Henry VII, or the House of Avis, really less? Patrimonialism, one is forced to conclude, the least anchored of modes in the subjacent economic narrative, remains something of a black hole in Runciman’s spectrum of forms.

There is, of course, another obvious objection to be made to the way this first of state structures features in the survey as a whole. This is simply the enormous historical spread between the cases Runciman adduces. Can the societies of Peisistratus, Ashoka, Chilperic I and Henry the Navigator be in any regard sensibly grouped together under a common label? The difficulty is rendered particularly acute in this case by the brevity and sketchiness of Runciman’s initial definition of patrimonialism itself. But the same problem recurs within his more carefully worked-out forms as well. Three examples will suffice. In his very interesting discussion of the ‘citizen mode’, about which he writes with the authority of an accomplished Classicist, Runciman applies the concept alike to Greek cities and the Roman Republic, Swiss cantons, North Italian towns and the Icelandic commonwealth. Here the institutional resemblances, as well as differences, do indeed constitute a productive field for comparison. But however cognate the political fact of a free citizenry, is it plausible to claim that Rome in the epoch of Cicero and Iceland in the time of the Njalasaga were historically the same kind of society?

One was a wealthy multi-continental empire, controlling a population of some fifty million from a capital city of 750,000, with massive social stratification, large standing armies, heavy taxation, high levels of monetisation, advanced literary culture; the other a desperately poor, tiny pocket of perhaps thirty thousand peasants, without mint, militia, trade, administration or written culture. The decisive structural difference, commanding all others, was that the principle of Roman civilisation was urban from the outset, whereas Icelandic society was more completely rural than any other sovereign polity on record – there were scarcely even villages, let alone towns, in the heroic age of the commonwealth. A socio-economic contrast of this magnitude surely breaks any bracketing apart. Runciman is able to minimise its import only by deciding that ‘for the purposes of sociological theory’ towns can be regarded simply as ‘institutional expressions of power’ (a view close to that of another Cambridge sociologist, Anthony Giddens, who terms them ‘power-containers’). The limitations of this way of looking at the historical significance of cities needs no labouring.

Feudalism as a form of society provides a set of similar paradoxes. Here there is a considerable existing literature, going back at least to Bloch, which has tried to grapple with the taxonomic problems its poses. Opinion has generally divided into two camps: those who restrict the term to areas where the institution of the fief, as conditional landed property deriving from the nexus between benefice and vassalage, involves jurisdiction over a dependent peasantry, and those who extend it to any pre-capitalist regime of large landowners extracting customary rents from the direct producers. For the first, Medieval Europe and Japan tend to exhaust the field; for the second, India, Africa, China, the Near East, Mesoamerica all had their examples of feudalism. Runciman takes up an intermediate position, by defining as feudal any regime of decentralised magnate power where landlords extort a surplus from dependent cultivators. This looks like a reasonable compromise. Nevertheless, its application leads to some curious results. For, on the one hand, he excludes Medieval England from the range of feudal societies, on the grounds that the Angevin and Plantaganet monarchies were too centralised. On the other hand, he includes the Latin American states of the 19th century. The resistance this proposal arouses is not just to do with an erratic use of the criterion of ‘decentralisation’ itself – though that is certainly relevant: by no standard could governments like those of Roca in Argentina, Guzman Blanco in Venezuela, or for that matter the Empire in Brazil, be deemed weaker than Lancastrian rulers, even at their height. More important, however, is the sheer developmental distance between the sword and lance world of Capetian or Ashikaga knights, to take two original cases of feudalism accepted by Runciman, and the railway and carbine universe of Latin American presidents.

Something of the same difficulty is visible in the discussion of ‘absolutism’ as well. Runciman identifies this with the bureaucratic mode of distribution of power, in which there is a strong central administration composed of salaried state servants without patrimonial ties to the ruler. He then takes as two paradigm examples the Roman Principate and the Bourbon monarchy. But so much had changed, materially and culturally, between the two that the pairing loses analytic force. Without considering the wider society, the two state structures themselves were qualitatively distinct in ways that Mann – contrasting the ‘territorial empire’ with the ‘organic monarchy’ – has graphically shown. Louis XIV could raise an army larger than Trajan’s from a realm an eighth of the size. In Asia, Runciman treats among his major instances of absolutism both Imperial China, from Sung times onwards, and Tokugawa Japan. But the dissimilarities between the two societies and their respective states are so large as to form a virtual topos in Far Eastern scholarship – not least because of the dramatically divergent outcomes of their common collision with the West. Mandarin and Samurai – to refer only to the structural profiles of the holders of power – differed in recruitment to office, focus of allegiance, relation to land, role in war, function in learning. Nor was the machinery of power at all alike. The Heavenly Kingdom commanded, in principle and in normal times in practice, an effective monopoly of armed force. The Shogunate did not even control a majority of the military levies in Japan, where the combined troop strength of the outlying han always out-numbered its own – a weakness brought home in the hour of its overthrow. Can a state of this kind be usefully termed ‘absolutist’?

A final quirk of the discussion here, it may be noted, concerns England. Having contested that Angevin or Plantaganet rule were ever feudal, he goes on to deny that Stuart government was ever absolutist. ‘England alone evolved from a patrimonial to a bourgeois monarchy,’ he writes: ‘there was no genuinely feudal stage any more than there was ever a genuine absolutism.’ This seems categorical enough. Yet elsewhere, in a rare moment of apparent inadvertence, he paradoxically classifies the Norman monarchy after the Conquest as absolutist. There have been so many perverse claims for English exceptionalism in this period that it would be churlish to make too much of this contradiction. But it points to a more general problem.

The tacit modulation from society to state in the typology as a whole unbalances the triad that was intended to govern it. For virtually all pre-modern states devoted the bulk of their resources to the mode of coercion, in Runciman’s sense: the administration of force was nearly always much the most costly and most vital of their functions. Concentration of focus on states thus necessarily tends to relegate modes of production and persuasion to the margins of this field of vision. The result is twofold. On the one hand, economic disparities between the states assembled under a common rubric are largely disregarded, once their type has been positioned in the understated background sequence. But the sheer developmental distance between many of Runciman’s cases strains to breakingpoint the attempt to yoke them conceptually together. Technology in its widest sense, which is always in part specific to ecological setting, cannot be factored out so easily from historical comparison. France circa 860 and Argentina or Brazil circa 1860 were separated not just by a millennium, and a hemisphere, but by an industrial revolution capable of material transformations across the world. Even on Runciman’s own criteria they should be allocated to different modes of distribution of power, since serfdom predominated in the first, wage-labour in the second, and slavery in the third. Closer settings, or epochs, should put less burden on common definitions. But even here acute problems arise. The England of Edward I and of Elizabeth I was one country: but was it really the same kind of society, within the terms of the taxonomy itself, when villeinage had disappeared between the two? Sung China was a contemporary of Norman Sicily, and Runciman judges them absolutist alike. But where were the free peasant tenants, and high commercial agriculture, in the latter – to say nothing of industrial skills and teeming urban radiance?

In such classifications, culture fares no better than economy as a marker of substantive difference. The same double attrition occurs. In effect, Runciman first confines his ‘mode of production’ to what Marxists would call relations of production, excluding forces of production from it; and then often sets aside these relations themselves in constructing his typology. Similarly he restricts his ‘mode of persuasion’ to the formal figures of deference, excluding all larger meaning-systems; and then gives only intermittent attention to these figures themselves. By his own criteria, for example, the form of legitimation of power was hereditary in the Poland of the 16th-century szlachta, but charismatic in the Bolivia of the 19th-century caudillo – each of which, however, he qualifies simply as feudal. Despite their role in the theory of social variation, distinctions of this kind are absent from the taxonomy of pre-industrial forms which issues from it. Yet they constitute only one particular sort of ‘persuasion’ – the justification of rule. In other words, differentiation here is ideological rather than cultural. Runciman’s solution to the problem of material productivity is to displace it into a generic chronicle outside the topology. His solution to the problem of cultural diversity is to trim culture itself to the typology, by treating it as no more than the content of the institutional rules of power, whose pattern composes a social structure. This definition then allows him to abstract structure from culture for the purposes of historical classification. This is the operation which yields such startling juxtapositions as the twinning of Old Babylonia and Anglo-Saxon England, which depend on the elimination from the picture of central empirical features such as cuneiform script or Christian belief. Given the eloquence of Runciman’s own case for the capture of cultural interiority, in Volume I of his Treatise, there is something wry in this procedure. Runciman himself remarks that in practice structure and culture never vary independently of each other, but this is a maxim whose consequences he never follows through. He observes at one point, for example, that Muslim societies were generally either warrior or patrimonial in structure, in part because Islam was inimical to feudalism. But the reasons why, and extent to which, religious culture may have contributed to delimit the range of political structures in the Middle East – that is, potentially select their form, rather than merely supply their content – is left quite unexplained. Questions like this fall beyond the ambit of the enquiry.

One result is that Runciman can repeatedly assert that the diversification of human societies has greatly increased over time, a thesis which is remotely defensible only if societies are identified with structures in his sense (although even then the argument seems curious, since no more than four modes are held to exist today). In fact, it is the opposite process that is surely striking: the enormous reduction in the number of discrete forms of society under the impact of modern capitalism – above all, from the extraordinary variegation of primitive cultures, to the lesser number of religious civilisations, down to the global uniformities of industrial living today; but also from the formidable diversity of pre-capitalist relationships of production and power, to the relatively stark simplicity of the institutional map of the contemporary world.

In the spirit of Economy and Society, Runciman’s typology is resolutely oriented to formal comparison – in diametric contrast to Mann’s single-track narration. Each contends that the one excludes the other, Mann denying the possibility of a comparative macro-sociology, Runciman scouting the prospect of an inter-societal history. But although the Treatise is faithful to Weber in its taxonomic drive, pursued with a lucid rigour that leaves the sprawling mass of Economy and Society well behind, Runciman has no time for the teleology of rationalisation which is the other side of Weber’s inheritance. His theory of social evolution is designed to supersede it and all its counterparts, including Marx’s dialectic of class or any other. This second moment of the enterprise occupies the long concluding part of the book, where Runciman tests the theory against a gamut of empirical cases. The originality of the evolutionary account put forward here lies, as has been seen, in its focus on practices rather than societies or groups. But Runciman’s evolutionism also differs from its most recent predecessor in another way. For Parsons, the principal mechanism determining evolutionary change was adaptation – of societies to their environment. For Runciman, this mechanism is selection – or competition between practices. This is a sharp change of emphasis, which brings his sociology much closer to the thrust of neo-Darwinian biology today, that the dynamic element in evolution is not the adjustment of species to an environment but the warfare of species with one another – Dawkins’s ‘arms race’; and also, of course, marks a return to the earlier tradition of evolutionary social thought which, in the 19th century, gave pride of place to the ‘survival of the fittest’.

The central objection to the Parsonian notion of adaptation – lately levelled most scathingly by Anthony Giddens – was always its vacuity: for it never defined what it was that societies were adapting to, as they evolved, or what adaptation itself, taken as a nebulous ‘reduction of uncertainty’, actually meant. Does Runciman’s alternative avoid these pitfalls? Certainly the idea of social selection is much more precise than that of adaptation, and specific practices more plausible as units for it than whole societies. But there is still a difficulty. The process of selection, in Runciman’s account, turns on competition. How does he envisage the workings of this competition? The answer, if one looks at the test-cases he presents, proves to be oddly elusive. One of the most extended treatments is his discussion of the origins of European feudalism in France, Northern Italy and Germany, the purpose of which is to show the ‘selective pressures’ that picked each end-result in the three regions – why the monarchy eventually achieved dominance in the first, towns in the second, magnates in the third; whereas in the different context of the Byzantine state there was a stand-off between monarchy and magnates. Runciman’s account of these variants is succinct and sensible. But is it more than an intelligent inspection of the respective balance of forces in each zone, of a kind which historians without evolutionary equipment make all the time – indeed to which he later appeals? The explanatory bonus yielded by the idea of competitive practices is unclear. Its effect seems more like a redescription.

The same could be said of Runciman’s analysis of the ending of Japanese feudalism, as he sees it – the emergence of the bakuhan order under the first Tokugawa rulers. Here he advances the unusually strong claim that the isolation of Japan from exogenous pressures made its evolution from a feudal to an absolutist mode ‘as near to inevitable as any such shift can be’. The crucial practices that permitted it were those, he suggests, which involved the transformation of bushi from armed retainers into salaried officials. But how these were competitively selected remains unstated: he contents himself with saying that this was the trend of the time. So far as the larger thesis goes, it is surely more plausible to argue the reverse: that the Shogunate never developed into an absolutism precisely because there was no external danger to impose true military (or administrative) centralisation. This was why, in part, it proved so easy to overthrow from within once a threat did materialise from without. What an example like this indicates is that for the notion of competitive selection to have force, there must be some unambiguous medium or Kampfplatz in which rival practices can conflict, for one to triumph over the other.

Revolution would seem to form one major kind of arena for such a shock. Runciman contests the use of the term for the armed movement that created the Meiji state, which he views rather as the ideal-typical model of a (successful) reform. His own discussions of revolution, rebellion and reform – to which a substantial excursus is devoted – is full of ingenuity and interest. But it is Linnaean rather than Darwinian in emphasis, an illuminating typology more than an integral moment in the account of evolution. In fact, Runciman shows a consistent scepticism about the importance of revolutionary change for social selection. The two examples he analyses at length are the Mexican and the French. He concludes that the first merely restored the Porfirian status quo ante in more centralised and secular guise; while the second was accidental in origin and irrelevant in outcome, leaving the evolution of French society unaffected. The Russian Revolution, by contrast, was truly transformative, but it, too, was historically an ‘improbable’ occurrence, let loose only by the hazards of the Great War.

For classical theories of social evolution, the prime arena of social selection was, of course, precisely war. It was the battlefield that provided the supreme medium of decision between alternative forms of collective life. Paradoxically, however, Runciman also shows the utmost reticence towards this kind of test, which acquires little or no salience in his survey. The reason he gives, revealingly, is one that recurs in his treatment of revolutions. Wars are dramatic events, which may indeed ‘divert the whole course of human history’. But their ‘outcome is so often a matter of chance’, the product of ‘accidents of fortune’, that they frequently ‘cannot be grounded in an evolutionary theory concerned only with the sociological causes of differences in modes and sub-types of the distribution of the means of production, persuasion and coercion’. Empirically, the claim that the issue of military conflicts tends to be especially fortuitous is quite unsupported here. Runciman’s only illustration is the Second World War – in many ways a particularly unhappy one, given the huge disparity in material resources between the victorious Allies and the Axis powers. Theoretically, however, what is significant is the apparent downplaying of war, as the regular terrain of competitive selection that might be expected within a quasi-Darwinian framework. Here it looks as if Runciman may be over-compensating for Spencer’s notorious stress on it, in a wish to put as much distance as he can between their two accounts of evolution.

There is, finally, one other medium of competition which does not feature as prominently as might have been imagined in Runciman’s theory of social selection. This is simply the market itself. The most widespread version of an evolutionary outlook today is the belief that the Cold War is coming to an end as the superior economic practices of the West visibly prevail over those of the East, portending the rapid elimination of an alternative societal form that has been proved historically uncompetitive. This has, of course, long been the prediction of Von Hayek, a thinker for whom the truths of economics have always been continuous with those of biology, in a neo-Darwinian synthesis of his own. It is noticeable that Runciman’s reserve extends to this diagnosis as well. His treatment of Communist societies, cool and judicious throughout – it contains a remarkable shaft of observation on Stalin’s purges – gives no hint of such a prospect. Perhaps he is just suspending judgment.

These uncertainties all concern the mechanisms of social selection. But biological evolution, of course, has another and prior side to it – the genetic variation that furnishes the raw materials for natural selection. It is here that the real, insurmountable weakness of any theory of social, or for that matter cultural, evolution lies. For all analogies break down at this crux. The mutation of genes is a strictly random element within the process of natural evolution. That is, the causes of misprints in DNA have no relation whatever to the filter that selects these: they belong to another order of determination altogether. The origin of mutations is in this sense completely irrelevant to the origin of species. They belong to two separate planes of intelligibility. But does this hold in social or cultural life? The formal counterpart of genes in Runciman’s explanation of social evolution are practices of power. His biological model then commits him to the claim that the origin of these practices – the mutations which are selected or eliminated – is also random. Innovations consequently become ‘intrusions from another causal sequence’, which can be treated simply as ‘acts of God’ for the purposes of social explanation. The structure of argument is the same in Franco Moretti’s splendid Signs taken for wonders (with Gould rather than Dawkins in the background), which proposes a theory of literary evolution where genres play the role of species, and individual works furnish the random mutations.

Human history, however, exhibits no such Darwinian dualism. For in the process of its unfolding, innovation belongs to the same plane as selection. There is no split-level causation here. For both the genesis and the generalisation of social practices always involve the common material of conscious human agency. The origins of institutional (or aesthetic) forms are thus never divided by an epistemological abyss from their stabilisation. The two compose one continuous flow of intentional action. They are therefore, contrary to Runciman’s suggestion, amenable to the same kind of social explanation. There is no sense in which major social innovations can be treated as ‘random inputs’ for a plausible theory of historical development. Some of the finest Marxist work of recent years has, in fact, been devoted to elucidating the origins of such momentous new practices as the emergence of the Greek polis from the debris of the Mycenaean palace-kingdom (Ellen Wood), of feudal relations from the involution of late-imperial Roman society (Chris Wickham), of capitalist farming from the unravelling of English manorialism (Robert Brenner) – all innovations of central importance to Runciman’s own account of social evolution. But there is nothing specific to historical materialism in such enquiries. Perhaps the greatest modern example of one is Peter Brown’s masterpiece The Body and Society, which explores the emergence of sexual ascesis in Early Christianity, a far-reaching practice of the kind Runciman’s theory leaves aside.

Beginnings such as these can all be reconstructed as intelligible patterns of intentional action, typically yielding unintentional consequences – in this no different from their ongoing sequels. But there are, of course, historical episodes which bear the impress of a much more deliberate kind of will. The practice of social insurance, with all its long-term significance for the construction of the contemporary welfare state, was inaugurated by Bismarck for specific political ends. Can the initiative of the Iron Chancellor really be treated as an act of God? Were the reforms by the Meiji oligarchs of rank-order or land tenure mere ‘random inputs’ into the evolution of modern Japanese society? Should the inception of international monetary co-ordination by Keynes and White at Bretton Woods be regarded as belonging to another cognitive world from the workings of the gold-dollar standard? Social engineering of this kind is a central, pervasive fact of the modern world. Runciman compares it to stock-breeding – no disproof, he insists, of natural selection from random genetic mutation. But no rancher or pigeon-fancier has ever produced a new species. What, in Runciman’s own typology, was the birth of the Soviet state other than a true speciation – and who would care to deny the prior purpose of Lenin and the Bolsheviks in bringing it about? In all these cases, the skein of historical causality is one. The reasons why Bismarck, Kido, Keynes or Lenin made their innovations – responses to the rise of the socialist movement, the threat of foreign take-over, the memory of pre-war slump, the opportunity of revolution – are no different in kind or level from the reasons why these practices ‘took’ as institutions thereafter, more or less as their authors intended: say, moderation of popular insecurity, contribution to national growth, stabilisation of foreign exchanges, social promotion of labour. Just because that is so, historical development shows not the slightest similarity to natural evolution in the tempo of its innovations. The whole force of Darwinian biology lies in its stress on the implacable slowness with which natural selection necessarily works, since the only variations offered to it are infinitesimal modifications, whose gradual acumulation alone, in an unimaginable series of tiny steps, produces evolution. The social innovations that are Runciman’s topic are utterly different both in the scale of the variation they represent and the speed of the changes they unleash. It is a far cry from the reign of Monera, lasting three billion years, to the rise of Islam in fifty.

Undeterred by these disparities, Runciman concludes his work with a grand restatement of his major themes, and claims. ‘However much further refinement and elaboration is called for, the fundamental idea that the evolution of human societies and their constituent roles and institutions proceeds through a continuing struggle for power whose outcome is determined by the competitive selection of practices in the three mutually irreducible dimensions of social structure is no less demonstrably superior to its rivals – or so I believe – than Darwin’s was.’ Social evolution, like natural evolution, is directional in the sense that both reveal a cumulative increase in complexity. Yet each buries every teleology beyond hope of resurrection: the history of humanity, as of life itself, has no pattern, goal or meaning. The one significant difference between the two is that in society ‘power, not survival, is the criterion of success.’ That was the shift made by Nietzsche, critic of the generalised notion of a ‘struggle for life’ in The Twilight of the Idols: ‘where there is a struggle, it is a struggle for power.’

Nietzsche himself is a reminder, however, that to repudiate teleology is not necessarily to renounce metaphysics. His will to power was resolved out of a – fabulously – speculative instinctual psychology. For all the distance in temperament and outlook, there is a trace of the same temptation in Runciman. He looks forward to the day when advances in psychology will provide a more complete underpinning for his sociology. Meanwhile, it can be said with assurance that in social evolution there are only people ‘competing with one another for economic, ideological and coercive power, as they will continue to do as long as the species survives’. The agonistics of this vision are close to fortune-telling. Should we say, with Nietzsche, that competition for power will be perpetual between the sexes too? A doctrine of Eternal Rivalry belongs with that of Return, outside any rational social science. This gesture aside, the seriousness of Runciman’s commitment to such a science is never in doubt. He explains that it was his conviction of the essential unity of scientific method, eloquently defended in Volume I, that helped him re-appropriate evolution as the idée maîtresse of Volume II of his treatise. The result is a theory of history which, by any standards, amounts to a large intellectual sum.