Derek Sayer (1987)

The Violence of Abstraction
The Analytical Foundations of Historical Materialism

3 Relations of Production

Source: The Violence of Abstraction (1987) publ. Basil Blackwell, 1987. Just two Chapters out of six reproduced here.


If productive forces are not the things they are so often taken to be, then definitions of production relations, and hence Marx’s ‘economic structure of society’, in terms of ownership of these forces are clearly in some jeopardy. But there is far more to be said about the concepts of production relations and economic structure than this. We may begin with a common criticism of historical materialism, to which I briefly alluded in chapter 1. In the 1859 Preface, as we know, Marx says that property relations are ‘but a legal expression for’ relations of production, and his writings frequently assimilate the two terms. H. B. Acton, amongst others, has argued that this renders Marxism incoherent. It amounts, he says, to defining production relations – the alleged ‘economic base’ of society – by legal forms which belong in the very ‘superstructure’ which that base, within historical materialist theory, is supposed to determine. The theory is thus viciously circular. Acton goes on to argue, more generally, that ‘the “material or economic basis of society” is not ... something that can be clearly conceived, still less observed, apart from the legal, moral and political relationships of men’ (quoted Cohen, 1978: 235).

This charge is obviously a serious one for any Marxism which wishes to retain a standard base/superstructure model, and it has not gone unchallenged. Cohen, in particular, has replied at length. On the face of it, he has little trouble in disposing of Acton’s argument. He simply develops a general ‘rechtsfrei’ definition of ‘ownership’ in terms of real powers over, as distinct from legal rights in, the objects owned – productive forces, as he conceives them. Other Marxists, like Balibar (in Althusser, 1970: 226 ff.), take a similar tack. Within such views, Marx’s ‘production relations’ are ‘ownership relations’ in some distinctively non-juridical sense. Legally formalised property relations can then coherently be conceived as both distinct from and expressive of production relations thus defined, exactly as the 1859 Preface seems to require. This resolves Acton’s problem of the independent conceivability of economic structure and superstructure. Cohen further argues, not unreasonably, that the fact that economic variables may not be empirically observable independently of the non-economic variables they are held to determine does not of itself invalidate the claim of determination. Such situations, he maintains, are commonplace in any developed science.

Cohen’s solution is a nice, and characteristically elegant, piece of philosophical reasoning. But whether it satisfactorily resolves the substantial problem which lies at the heart of Acton’s comments is another issue. Consider, by way of example, the following satirical observation of E. P. Thompson’s on the supposed ‘relative autonomy’ of law. His ire is directed at Althusser’s ‘structural Marxism’ rather than Cohen’s, but the point is pertinent none the less:

I have, as it happens, been interested in this myself, in my historical practice: not, of course, in any grand way – for the whole of history, nor for the capitalist mode of production everywhere, but in a very petty conjuncture: in an island on the edge of the Atlantic, very well supplied with lawyers, at a moment in the eighteenth century. So my evidence is highly marginal, as well as being seriously contaminated by empirical content. But what I discovered there would make La Structure ā Dominante boggle. For I found that law did not keep politely to a ‘level’ but was at every bloody level; it was imbricated within the mode of production and production relations themselves (as property-rights, definitions of agrarian practice) and it was simultaneously present in the philosophy of Locke; it intruded brusquely within alien categories, re-appearing bewigged and gowned in the guise of ideology; it danced a cotillion with religion, moralising over the theatre at Tyburn; it was an arm of politics and politics was one of its arms; it was an academic discipline, subjected to the rigours of its own autonomous logic; it contributed to the definition of the self-identity both of rulers and ruled; above all, it afforded an arena for class struggle, within which alternative notions of law were fought out. (1978a: 288).

‘La Structure ā Dominante’ is Althusser’s ‘structural causality’ model, referred to above (p. 9). He, like Cohen – and many other Marxists supposes the separability of the ‘economic’ from other levels of social structure. It is this separation Thompson rejects, going on to ridicule the very notion of “levels” motoring around history at different speeds and on different schedules as ‘an academic fiction’ (ibid.: 289). I could have cited other Marxist historians like Christopher Hill on Coke’s Institutes, or Rodney Hilton on jurisdiction in feudal Europe, to similar effect.

Cohen’s rebuttal of Acton is, significantly, carried out entirely within theory, at the level of a general, transhistorical concept of property. It is undoubtedly successful, as a piece of philosophical argument. To develop a rechtsfret definition of property raises no particular difficulties of a theoretical sort. But what the passage from Thompson scathingly brings out is that so soon as we try to apply such a definition, and specify, historically, what real power over objects – property, as Cohen conceives it – empirically consists in, then those things excluded from its general rechtsfrei definition tend ineluctably to creep back in.

For we rapidly discover that ‘power over objects’ (including, here, people’s labouring capacities) only exists, empirically, in forms which include things which according to the standard dogma ought to belong in the supposed ‘superstructure’. In eighteenth-century England, for instance, we may not be able to say, in any empirically meaningful way, what ‘property’ or production relations were, without talking about game laws, the ‘bloody code’, Parliamentary Acts of enclosure, strict settlement and entail, magistrates’ regulation of hours of work and rates of pay, Combination Acts, the theatrical rituals of the assizes, or laws prohibiting the shearing of sheep within five miles of the coast. Historical inquiry shows what Adam Smith called ‘the orderly oppression of the law’ (quoted D. Winch, 1978: 88-9) to be an integral part of – internal to, constitutive of – the complex web of social relationships we summate as ‘property’ in this particular time and place, not a ‘relatively autonomous instance’ at all.

In the real world, then, ‘power’ over objects turns out to be neither the abstraction, nor the simple relationship, of Cohen’s impeccably rechtsfrei definition. It exists only in a multiplicity of often rechtsvoll empirical forms, to whose analysis, if Thompson is correct, a categorical framework built on prior and exclusive definitions of supposed social ‘levels’ is remarkably ill-suited. In certain cases, like England in the eighteenth century, law will emerge as inextricably ‘imbricated within’ – indeed constitutive of – any property relations we might want to consider relations of production, ‘part of the same nexus of relationship’ (Thompson, 1965: 84). So too might other supposedly superstructural ‘instances’, like moral codes, political institutions, or ‘forms of social consciousness’ (all of which are fairly evidently entailed in Smith’s ‘orderly oppression’). In which case, to seek to expunge these from the concept of property or production relations a priori, for the sake of theoretical coherence or elegance, would seem to be a gross artificiality which does considerable violence to the very facts Marx’s concepts are meant to help us understand: a species of what he himself castigated as ‘violent abstraction’ (see Sayer, 1983: 121-2).


Cohen, however, tells us that he does ‘not need to be advised that history is “always richer in content, more varied, more many-sided, more lively and ‘subtle’” than any theory will represent it as being’ (1978: ix, quoting Lenin). Balibar’s concern is likewise with outlining only what he calls the ‘pertinent differences’ on whose basis the abstract concepts of modes of production can be theoretically developed (in Althusser, 1970: 209 ff.). Either would therefore presumably dismiss the foregoing line of argument as beside the point: a reminder, perhaps, of the real complexity of history, but not something which should be allowed to affect the basic formulation of Marx’s theoretical concepts as such. To forestall this objection, a short philosophical aside is in order. For I do not cite Thompson here simply to counterpose the real complexity of history against the arrant simplicities of theory, worthwhile and necessary as that enterprise sometimes is.

I want rather to make a theoretical point concerning the character of general concepts like property and their relation to empirical particulars of the sort invoked by Thompson. This follows on the argument I developed in section II of the last chapter, and relates to the difference between Marx’s transhistorical and historical categories. Only if this difference is ignored, I shall argue, can Thompson’s – or indeed Acton’s – arguments be dismissed as a merely ‘empiricist’ quibble which is irrelevant to considerations of high theory, and the relation between abstract theory and concrete fact be construed in the way it is by Cohen or Balibar. The empiricism, in fact, is if anything theirs, and resides in the tacit assumption – one shared by positivist philosophies of science generally – that fact and theory are wholly separate domains of discourse.

In The Holy Family there is a celebrated discussion of some of the more absurd practices of idealist philosophy. One of these is the fallacy of reification, or misplaced concreteness: mistaking abstract concepts for real entities. Marx dubs this ‘the mystery of speculative construction’. By way of illustration, he satirises philosophers – his specific target here being the Young Hegelians – as taking the general concept ‘fruit’ to refer to some real essence of fruitiness which is distinct from the empirical forms in which fruit alone exists – apples, pears, and so on (1844b: 5 7 ff.). Wittgenstein was later similarly to chide those who imagined that beauty was a real essence inhering in and defining all beautiful objects. Nobody, we might think, would be so silly.

But like fruit, or beauty, ‘property’ too – at least as Marx conceived it – does not exist except in particular, empirical forms. As with production, it has no existence as a generality. We might indeed be able to abstract from these particulars conceptually, in order to construct a general notion of property. But such an abstraction rechtsfrei or otherwise – does not refer to some real essence of property independent of its empirical forms, any more than the generic ‘fruit’ refers to anything other than real apples and pears. In more technical terms, generic concepts refer to what defines classes of phenomena, not, or at any rate not directly, to any empirical particulars – real objects – as such at all. And as Wittgenstein, again, argues, what makes phenomena members of the same class may be something as loose as ‘family resemblances’ rather than an unvarying list of empirical characteristics which they all necessarily have in common.

Now such class concepts, precisely because they are generic categories, are not, in themselves, complete descriptions of any of the individual phenomena which can be subsumed under them. Indeed and crucially – they do not even adequately comprehend the essential features of those phenomena, if by that we mean those features which make such phenomena what they individually are, and are thus essential to their definition. They only designate those aspects of a given set of phenomena which make them members of the relevant class, the resemblances by virtue of which they can be conceived as members of the same family. What defines phenomena in their individual particularity – in other words, in their real existence – is precisely those features they do not share with other members of the class to which they belong, and are therefore not contained in their class concept. We have not, for instance, adequately defined a lemon, in any empirical sense, when we say it is a fruit. Part of any adequate empirical description of a lemon would include, for example, its colour. But yellowness is not a characteristic of fruits as such or in general, and cannot be derived from the general concept of fruitiness. Yellowness is, however, essential to what makes a lemon a lemon, and is thus part of its concept.

Similar arguments apply to property. Certainly we may choose to define the class ‘property’, in general and abstractly, in ways which exclude ‘superstructural’ terms. There may even be a point in doing so, in so far as such a definition enables us to classify certain legal and non-legal forms of relationship as kindred phenomena, which we certainly could not do if we included legal criteria in property’s general concept. Thus far we may go along with Cohen against Acton. To do so is also consistent with Marx, who undoubtedly did take pains to insist that relations which were substantially property relations could exist without being legally expressed as such, as in passages as this:

With [Wagner] there is, first, the law, and then commerce; in reality it’s the other way round: at first there is commerce, and then a legal order develops out of it. In the analysis of the circulation of commodities [in Capital] I have demonstrated that in a developed trade the exchangers tacitly recognise each other as equal persons and owners of the goods to be exchanged respectively by them; they do this while they offer the goods to one another and agree to trade with one another. This practical relation, arising through and in exchange itself, only later attains a legal form in contracts etc.; but this form produces neither its content, the exchange, nor the relationship, existing in it, of persons to one another, but vice versa. (1880a: 210)

If our generic concept of property includes legal terms, the ‘practical relation’ Marx recognises here as a property relation could not be acknowledged as such, which would be plainly contrary to his intent.

But we should not then confuse such a general definition of property with its substantial reality, or take our generic concept adequately or immediately to designate real empirical objects. Nor, in particular, can we infer that what our concept excludes as a general characteristic of the class ‘property’ can a fortiori be excluded from the concepts of the individual members of that class, property’s particular forms – the only forms in which it empirically exists. To do so would be exactly like denying that yellowness is an essential characteristic of lemons, because it is not a feature of the generality ‘fruit’ – an evident non sequitur.

What my quote from Edward Thompson suggested, on empirical grounds, was that precisely as we move from the general concept of the class ‘property’ to the concepts of its members – as soon as, in other words, we try to specify empirically those historical forms of property which are Marx’s particular concern – we cannot any longer always exclude ‘superstructural’ terms. Like the yellowness of the lemon, they might be essential, and hence defining features of the property relations empirically at issue. At first sight, this appeared simply as an ill-bred ‘empiricist’ counter to Cohen’s or Balibar’s attempts rigorously to demarcate economic and legal at the level of a general theory. But it is in fact testimony to a lapse in their own logic. Even had Thompson not been burying his head in the archives, we would have had no licence to exclude law – or anything else – a priori from counting, should the evidence warrant it, as internal to a given property form.

For recognition of the internality of ‘superstructural’ relations to a given empirical form of property in no way conflicts with adherence to a rechtsfrei definition of the concept of property in general. No general concept of property can or ever could be exhaustive of the empirical characteristics which define property’s particular forms. It is impossible, in logic, to infer the concepts of members of a class, their differentia specifica, from the concept of the class itself. But for exactly the same reason, rechtsvoll specifications of these forms cannot be ruled out by virtue of the general rechtsfrei definition alone either. Cohen can only evade this conclusion by implicitly reifying his general concept of property – taking it as immediately specifying an empirical particular. He treats rechtsfrei property as if it were an entity in itself, a substance really distinct from the legal and other forms in which property empirically exists, rather than – like ‘fruit’ a mere class concept, specifying some characteristics (or family resemblances) real empirical forms of property have. A predicate, to use Marx’s own vocabulary, is thus surreptitiously transformed into an independent subject.

If we refuse the reification, however, and properly distinguish the general concept of property – however defined – from the concepts of its particular forms, Acton’s contention may still make eminent sense, when applied to the latter. Cohen’s refutation of Acton rests on his tacit, and wholly illicit, conflation of these distinct levels of conceptualisation: transhistorical and historical, to employ my earlier terminology. But one cannot argue thus from the general to the particular. Certainly we may, if we choose, define property in general in rechtsfrei terms. There are some obvious advantages to doing so. But in any given empirical instance ‘the economic structure of society’ may still turn out to be neither conceivable, nor observable, ‘apart from the legal, moral and political relationships of men’. The burden of this chapter is to show that Marx never for a moment thought otherwise, and his concepts of production relations and economic structure must be comprehended accordingly.

Similar arguments, we might note, apply to another of Cohen’s attempted restrictions on the category of production relations, his exclusion from Marx’s ‘economic structure’ of so-called ‘work relations’ – relations of production which Cohen terms ‘material’ as distinct from ‘social’ relations between producers. ‘Work relations’ include the forms of co-operation in the labour process, and division of labour ‘in the workshop’, discussed (also) as productive forces in the last chapter. By a specifically social relation between producers, Cohen means one which ‘entails an ascription to persons – specified or unspecified – of rights or powers vis-a-vis other men’ (1978: 94). This is another way of delineating his ownership relations. Not all relations between people in production are of this kind, and only those which are, in his view, enter Marx’s ‘economic structure’.

The foundation of this distinction between social and material relations of production is Cohen’s perfectly correct observation that Marx himself systematically distinguishes between material and social properties of the phenomena he analyses. We saw this, and its relevance to Marx’s theory of fetishism, in the last chapter. But Cohen commits a similar reification here as he does with property. For the point is that in Marx, the relevant distinction lies between material and social attributes – properties, qualities, characteristics, features – of productive phenomena, not between different kinds of phenomena in production as such. These too are in other words class concepts. Use as raw materials or instruments in the labour process, for instance, is a material characteristic which defines the class means of production, while embodying value and commanding surplus-value are social characteristics which define the class capital. But the empirical form, constant capital, exists only as a phenomenon which conjoins both these sets of attributes. Accordingly, as argued above for class or attribute concepts generally, no purely ‘material’ description (or, come to that, no purely social description) can be a complete empirical account of any real entity in this realm of analysis. We have not sufficiently described constant capital when we enumerate its material characteristics qua means of production. This is to refer only ‘to the simple matter of capital, without regard to the formal character without which it is not capital’ (1858a: 267). Remember the lemon again. Conversely, means of production only ever exist in some social form or another, just as there is no fruit that is not an apple, a pear, and so on.

Cohen is aware of this; it is basic to his own account of fetishism. What he apparently fails to realise, however, is that this totally invalidates his attempted work relations/social relations contrast. If ‘material’ and ‘social’ in Marx refer to attributes of phenomena, not phenomena as such, we cannot use his distinction to differentiate two different kinds of production relation, material and social. Nor, therefore, can we on these grounds expunge ‘work relations’ from the ‘economic structure of society’. These supposedly separate ‘material relations’ are but social relations of production regarded – abstractly from their ‘material side’. Marx’s distinction lies between the material and social dimensions of one and the same set of activities – activities, we might recall, which he explicitly conceives from the start as ‘double’, simultaneously material and social – not between substantial kinds of relations as such.


Having, I hope, cleared some ground, let us now turn more directly to Marx. I have already quoted The Poverty of Philosophy on the impossibility of defining property as ‘an independent relation, a category apart, an abstract and eternal idea’ (above, p. 21). ‘Moralising Criticism and Critical Morality’, a text dating from the same year, agrees: ‘private property, for instance, is not a simple relation or even an abstract concept, but consists in the totality of the bourgeois relations of production’ (1847c: 337). In both cases Marx defines property in terms of production relations, rather than the other way about. The production relations in question are always historically specific – the relations of a given mode of production – and property therefore emphatically a historical category. Marx’s approach here contrasts sharply with Cohen’s or Balibar’s, both of whom, as we have seen, seek to ground the concept of production relations on a transhistorical, albeit rechtsfrei, concept of property.

Indeed property, as Marx uses the term in these texts, far from being susceptible to independent definition, is in fact synonymous with the ‘economic structure of society’ as delineated in the 1859 Preface. ‘Moralising criticism’ says that property consists of ‘the totality of ... relations of production’, which is exactly the Preface’s description of the ‘economic structure’. We might, for a change, take quite literally the Preface’s claim that property and production relations are but different expressions for ‘the same thing’. As usual, however, Marx offers no general definition, of a substantive kind, of what these production relations, whether construed as ‘property’ or ‘economic structure’, actually are. But given the apparent synonymity of the referents of these terms, exploration of what he had to say, in various historical contexts, about particular forms of property might take us some way forward in the search for what he understood by an economic structure.

We may begin with some critical remarks on Hegel in Capital III. Hegel, according to Marx, seeks to explain private property in land from the (transhistorical) premise that ‘man as an individual must endow his will with reality as the soul of external nature, and must therefore take possession of this nature and make it his private property’. For Marx this will to possession is ‘comical’:

Free private ownership of land, a very recent product, is according to Hegel, not a definite social relation, but a relation of man as an individual to ‘Nature’, an absolute right of man to appropriate all things ... [Hegel] makes the blunder at the outset of regarding as absolute a very definite legal view of landed property – belonging to bourgeois society. (1865a: 615-16n)

Basic to Marx’s argument here is the idea that property is a social relation, not – or at least not primarily, but only consequentially – a relation of individuals to things. He says the same thing in Grundrisse, drawing a linguistic parallel we have encountered previously in his discussion of production. ‘Language as the product of an individual is an impossibility. But the same holds for property’ (1858a: 490). Possession, Hegel’s starting-point, is indeed ‘the subject’s simplest juridical relation’, but ‘there is no possession preceding the family or master-servant relations’; ‘the concrete substratum [like family or master-servant relations] of which possession is a relation is always presupposed’ (1857: 102, my interpolation). The appearance of property as an unmediated relation of possession between individual owners and the things they own is an illusion arising out of the phenomenal form property takes within specifically capitalist relations. It is an historical product.

In previous forms of society, neither individuals as owners, nor their property, had their modern exclusivity or simplicity. Property did not even appear as a simple relation of person and thing. Who owned what, or even what it meant to be an owner, were by no means clear-cut; the very terms at issue are anachronistic. This is well brought out in the historian Marc Bloch’s discussion of the inapplicability of modern concepts of property to medieval Europe. There, he points out:

the word ownership, as applied to landed property, would have been almost meaningless ... The tenant who – from father to son, as a rule, ploughs the land and gathers in the crop; his immediate lord, to whom he pays dues, and who, in certain circumstances, can resume possession of the land; the lord of the lord, and so on, right up the feudal scale – how many persons are there who can say, each with as much justification as the other, ‘That is my field!’. Even this is an understatement. For the ramifications extended horizontally as well as vertically and account should be taken of the village community, which normally recovered the use of the whole of its agricultural land as soon as it was cleared of crops; of the tenant’s family, without whose consent the property could not be alienated; and of the families of the successive lords. (1967:115-16)

Marx himself argues that landed property receives ‘its purely economic form’ – i.e. its modern capitalist form, of a relationship of exclusive, individual possession, wrongly theorised as ‘an eternal idea’ by Hegel – only ‘by discarding all its former political and social embellishments and associations’ (1865a: 618), to become divorced from ‘relations of dominion and servitude’ (ibid.: 617). Only then can property appear as a simple relation of possession between individuals and objects. As any reader of Capital I (or anyone with any familiarity with modern history) will know, the making of this ‘purely economic form’ of property was a protracted, bitter, and often bloody struggle – one Marx first encountered, incidentally, as early as 1842, with the transformation, in the Rhineland, of erstwhile common rights to gather wood into a crime (1842; cf. 1859a). We will see later that the independent ‘natural individual’ who appears to Hegel as the self-evident possessing subject is equally a modern phenomenon.

One might perhaps maintain that the complexity to which Bloch points can be analytically encompassed within some notion of degrees of control of persons over things. Cohen’s argument moves in this direction, when he seeks to define the serf as a part-owner of both means of production and labour-power (1978: 65). It is questionable whether such a merely quantitative index of control is sufficiently subtle to allow us to differentiate the qualitative variety of forms which anything we might want to call property has taken historically. But this is, I would suggest, in any case to miss the main thrust of Marx’s argument. For what he does is precisely to shift the focus of the concept of property away from relations between people and things – from ostensibly simple possession – to the social relations between people which make such appearances possible in the first place. In the process – just as with the fetishised conceptions of productive forces considered in the last chapter – Marx criticises everyday ideas of property, of the sort systematised by Hegel, as an ideological expression of specifically capitalist conditions, whose extension elsewhere is profoundly anachronistic.

These points are reinforced when we turn to Marx’s (extensive) discussions of property in his early writings. Thus in the Paris Manuscripts of 1844 he writes, for instance, that:

Private property is ... the product, the result, the necessary consequence, of alienated labour, of the external relation of the worker to nature and to himself ... True, it is as a result of the movement of private property that we have obtained the concept of alienated labour (of alienated life) in political economy. But analysis of this concept shows that though private property appears to be the reason, the cause of alienated labour, it is rather its consequence ... Later this relationship becomes reciprocal ... estranged labour is the direct cause of private property. (1844a: 279-80)

Marx goes on to say that ‘the character of private property is expressed by labour, capital, and the relations between these two’ (ibid.: 289; cf. 1845a: 278). This strikingly anticipates later and more famous formulations, like Capital’s observation that ‘capital is not a thing, but a social relation between persons established by the instrumentality of things’ (1867a: 776). The overall argument of the 1844 Manuscripts on private property is succinctly captured in two sentences: ‘material, immediately perceptible private property is the material perceptible expression of estranged human life’ (1844a: 297). Conversely, ‘only when labour is grasped as the essence of private property, can the economic process as such be analysed in its real concreteness’ (ibid.: 317).

In The German Ideology Marx concretises this analysis somewhat but without losing its essentials. He argues there that ‘different forms [of society] are just so many forms of the organisation of labour, and hence of property’ (1846a: 78). Specifically, ‘division of labour and private property are ... identical expressions: in the one the same thing is affirmed with reference to activity as is affirmed in the other with reference to the product of the activity’ (ibid.: 44). For,

the various stages of development in the division of labour are just so many different forms of ownership, i.e., the existing stage of the division of labour determines also the relations of individuals to one another with reference to the material, instrument and product of labour. (ibid.: 33)

Marx elaborates:

with the division of labour ... is given simultaneously the distribution, and indeed the unequal distribution, both quantitative and qualitative, of labour and its products, hence property; the nucleus, the first form, of which lies in the family, where wives and children are the slaves of the husband. This latent slavery in the family, though still very crude, is the first property, but even at this early stage it corresponds perfectly to the definition of modern economists who call it the power of disposing of the labour-power of others. (ibid.)

The equation of property with ‘power of disposing of the labour-power of others’ is one Marx employs repeatedly in texts of this period. That power, in more developed modes of production like capitalism, is certainly materialised in (and effected in part through) ownership of physical means of production. But it is worth underlining the fact that this is not the case with the family, whose relations Marx none the less claims here to be property relations. Power over people’s labour-power derives in this instance from direct, personal, patriarchal relations between individuals themselves. modern Marxist anthropologists, not to mention feminist writers, concur with this insight.


I want now to consider exactly what relations between people would, for Marx, qualify as property or production relations. I shall show, from his own texts, that he indeed did (where historically relevant) include within this category relations which are, for traditional historical materialism, eminently ‘superstructural’. The ‘economic structure of society’ was therefore for him a very different – and a very much broader – totality of social relations than is normally recognised. The examples I shall consider are drawn from Grundrisse and Capital, both undeniably ‘mature’ works. They are therefore not open to the sort of objections Althusserians might raise to my use of the 1844 Manuscripts and The German Ideology immediately above. They are, however, perfectly consistent with these early texts. Since both my examples concern the ‘economic structure’ in non-capitalist societies, they also have the additional merit of suggesting, if only by way of contrast, the rootedness of conventional, narrow definitions of the ‘economic sphere’ in specifically capitalist forms of appearance.

The section of the Grundrisse entitled ‘Forms which precede capitalist production’ is a long, and largely self-contained disquisition on various historical forms of communal property and production, their gradual disintegration, and the eventual ‘freeing’ of the elements of capitalist production. Such communal forms were to remain an abiding interest for Marx, increasingly so in his later years. A large part of his copious, and still largely unpublished, notes and manuscripts of the 1870s and 1880s are given over to this theme (1879a, b; Shanin, 1984). In Grundrisse, he distinguishes three main communal types: primitive, ancient and Germanic. All have variants. I am not concerned here with the historical or anthropological accuracy of Marx’s characterisation of these forms of society, which by modern standards undoubtedly leaves a lot to be desired (Godelier, 1973: part II, discusses this), nor with how his views on the character and development of communal property might have altered in later writings. I invoke these analyses simply to illustrate how he used his concepts.

Marx’s ‘first form of landed property’ (1858a: 472-4), the primitive commune, is characteristic of initially pastoral, nomadic societies. Here individuals relate to land as ‘property of the community, of the community producing and reproducing itself in living labour’. They are proprietors of the land in so far as, and only in so far as, they are members of the community. Indeed, Marx says, strictly speaking individuals as such are here merely possessors of the soil, for land is the property of the community as a whole (ibid.: 476). Marx is clear that the ‘initial, naturally arisen spontaneous community’ – which is the ‘family, and the family extended as a clan’ – ‘appears as first presupposition’ for this form of property:

the clan community, the natural community, appears not as a result of, but as a presupposition for the communal appropriation (temporary) and utilisation of the land. When they finally do settle down, the extent to which this original community is modified will depend upon various external, climactic, geographic, physical etc. conditions as well as on their particular natural predisposition – their clan character. This naturally arisen clan community, or, if one will, pastoral society, is the first presupposition – the communality of blood, language, customs – for the appropriation of the objective conditions of their life, and of their life’s reproducing and objectifying activity (activity as herdsmen, hunters, tillers, etc.) ... The real appropriation through the labour process happens under these presuppositions, which are not themselves the product of labour, but appear as its natural or divine presuppositions. (ibid.: 472)

Here, then, particular social relations – specifically, familial relations, extended into the ‘clan community’ – are presupposed to appropriation through the labour process, and integral to these relations are ‘communality of blood, language, customs’. These latter are of course ‘superstructural’ phenomena par excellence for most adherents to base/superstructure models, and things which belong neither in the ‘economic structure’ nor the superstructure (and thus lie outside the scope of historical materialist theory altogether) for Cohen.

The so-called ‘Asiatic form’ of property (ibid.: 472-4) is a variant of this primitive form, though, Marx says, it at first sight appears not to be because of the individual’s apparent propertylessness. Here:

the comprehensive unity standing above all these little communities appears as the higher proprietor or as the sole proprietor; the real communities hence [exist] only as hereditary possessors. Because the unity is the real proprietor and the real presupposition of communal property, it follows that this unity can appear as a particular entity above the many real particular communities. (ibid.: 473)

Marx’s language here is dense. But he appears to be saying that in the Asiatic form, the ‘unity’ formed in the co-operation of several communities becomes a ‘presupposition of communal property’, in exactly the same way as within any single community the existence of the commune itself is such a presupposition. He instances water control as an important form of inter-commune co-operation. Clearly where artificial irrigation is a condition of production, and it requires inter-communal co-operation, such co-operation also becomes a condition of production, and hence of property. Karl Wittfogel (1957) was to develop this line of argument (and apply it to the modern USSR), though in ways uncongenial to most Marxists; modern ‘bureaucratic collectivist’ theorists reason similarly, drawing on Marx’s analyses of the Asiatic mode of production, vis-à-vis the ‘socialist’ state in modern planned economies (Melotti, 1977).

In such cases, this trans-communal ‘unity’ may be personified in the form of a ‘despot’; property ‘appears mediated for [the individual] through a cession [sic] by the total unity – a unity realised in the form of the despot, the father of many communities – to the individual, through the mediation of the particular commune’(1858a: 473). Such a personification of social relationships is analogous to fetishism in the capitalist mode of production (though Marx himself does not draw the explicit parallel here). Surplus then goes to the personification of this ‘unity’ – oriental potentate, clan patriarch, state, or whatever. Marx sees varieties of this ‘Asiatic’ form of society as being very widespread, embracing Mexico, Peru and early Celts, as well as India. He also suggests that among Slavs this may have been the origin of villeinage. What is of interest for the present argument, however, is that once again relations internal to a given mode of production and property form embrace evidently ‘superstructural’ phenomena: quasi-political relations (of a tributary sort) and, in so far as a specific mystification of the ‘unity’ is involved, ideology too. Marx writes, for instance, of ‘common labour for the exaltation of the unity, partly of the real despot, partly of the imagined clan-being, the god’ (ibid.).

Marx’s second form of landed property, the ancient (ibid.: 474-6) Calso assumes the community as its first presupposition’ (ibid.: 474). The ancient Greek or Roman community, however, differs from the primitive or Asiatic archetype in several respects. Its basis is urban rather than rural: ‘the cultivated field here appears as a tenitorium belonging to the town’. We also see here the beginnings of private property: ‘communal property – as state property, ager publicus – [is] here separate from private property. The property of the individual is here not, unlike in the first case, itself directly communal property’. The great problem, in ancient society, is no longer the conquest of nature, but maintenance of the territorium against other communes:

War is therefore the great comprehensive task, the great communal labour which is required either to occupy the objective conditions of being there alive, or to protect and perpetuate the occupation. Hence the commune consisting of families initially organised in a warlike way – as a system of war and army, and this is one of the conditions of its being there as proprietor. (ibid.: 474)

The ancient commune, then, unlike the primitive, consists of individual private peasant proprietors. But at the same time, the condition of their property remains their communal organisation, particularly for war. In this connection Marx offers some interesting parentheses on antiquity’s view of agriculture as the sole ‘proper occupation of a free man, the soldier’s school’, and the consequent exclusion of craftsmen from citizenship (ibid.: 477).

Importantly, however, the emergence here of private property does not obviate the communal character of the ancient form as a whole. Rather, ‘membership in the commune remains the presupposition for the appropriation of land and soil, but, as a member of the commune, the individual is a private proprietor’; ‘the commune ... is the presupposition of property in land and soil’ still, ‘belonging [is] mediated by [the individual’s] being a member of the state, by the being of the state’ (ibid.). The individual is in other words only a private proprietor in so far as he remains a member of the commune.

The German Ideology similarly observes that in antiquity, ‘the citizens hold power over their labouring slaves only in their community, and on this account alone ... are bound to the form of communal ownership’ (1846a: 33). Certainly, ‘the survival of the commune is the reproduction of all its members as self-sustaining peasants’. But equally, their ‘surplus time belongs precisely to the commune, the work of war etc.’. For:

The property in one’s own labour is mediated by property in the condition of labour – the hide of land, guaranteed in its turn by the existence of the commune, and that in turn by surplus labour in the form of military service etc. by the commune members. It is not co-operation in wealth-producing labour [unlike in the primitive form] by means of which the commune member reproduces himself, but rather co-operation in labour for the communal interests (imaginary and real), for the upholding of the association outwardly and inwardly. Property is quiritorium [i.e. the property of the Romans or quirites], of the Roman variety; the private proprietor of land as such only as a Roman, but as a Roman he is a private proprietor of land. (1858a: 476, my interpolations)

Property here is therefore integrally bound up with citizenship; it is only as a civis that an individual can be a proprietor. In this instance, then, it is membership specifically of a polis – and participation in its military and other obligations – which Marx sees as presupposed to property and the material labour-processes in which it is realised. That polis is understood, moreover, as a variant of the clan system which also underpinned Marx’s first communal form, the primitive. The ancient polis remains, in essence, an association of families grounded in the needs of war, even if the later Roman gens are not strict blood-kin.’ Marx approvingly quotes Niebuhr to the effect that ‘there was in the world of antiquity no more general institution than that of kin groups’ (ibid.: 478).

Finally to Marx’s third form of communal property, the ‘Germanic’. He sees this form as typical of the medieval period in Europe. It is in many ways a transitional form, representing a further development towards modern private property. Settlement here is sparse, and society predominantly rural rather than urban. The commune now exists only as a periodic gathering together of its members ‘to pledge each other’s allegiance in war, religion, adjudication etc.’ (ibid.: 484). It thus ‘appears as a coming-together, not as a being-together; as a unification of independent subjects, landed proprietors, and not as a unity. The commune does not therefore in fact exist as a state or political body, as in classical antiquity’ (ibid.: 483). Marx still, however, considers that the ‘unity-in-itself of commune members is none the less ‘posited in their ancestry, language, common past and history, etc.’ (ibid.); and he continues to insist that the commune ‘is presupposed in-itself prior to the individual proprietors as a communality of language, blood, etc.’, even if it only becomes ‘a real assembly for communal purposes’ (ibid.: 484-5).

There remains undivided communal land – for grazing, hunting and so on – but this does not take the separate form of state property, the ager publicus, as in antiquity. Rather, it is ‘really the common property of the individual proprietors, not of the union of these proprietors endowed with an existence separate from themselves’ (ibid.: 485). ‘Individual property does not appear mediated by the commune; rather, the existence of the commune and of communal property appears as mediated by, i.e. as a relation of, the independent subjects to one another’ (ibid.: 484). This characterises the Germanic form more generally: ‘the commune exists only in the interrelations between these individual landed proprietors as such’ (ibid.).

Following these sketches, Marx draws some general conclusions. ‘In all these forms’, he writes, the ‘relation to land and soil, to the earth, as the property of the labouring individual ... is instantly mediated by the naturally arisen, spontaneous, more or less historically developed and modified presence of the individual as member of a commune – his naturally arisen presence as member of a tribe etc.’ (ibid.: 485). Hence:

If the objective conditions of his labour are presupposed as belonging to him, then he himself is subjectively presupposed as member of a commune, through which his relation to land and soil is mediated. (ibid.: 486)

Marx goes on explicitly to discuss what ‘property’ means in this context. He argues that, ‘property ... originally means no more than a human being’s relation to his natural conditions of production as belonging to him, as his’. He then elaborates on the character of these ‘natural conditions’ (making it clear, once again, that ‘natural’ and ‘social’ were not for him by any means exclusive categories):

The forms of these natural conditions of production are double: (1) his existence as a member of a community; hence the existence of this community, which in its original form is a clan system, a more or less modified clan system; (2) the relation to land and soil mediated by the community. (ibid.: 491-2).

Hence, ‘a natural condition of production for the living individual is his belonging to a naturally arisen, spontaneous society, clan etc. This is e.g. already a condition for his language, etc. His own productive existence is possible only on this condition’ (ibid.). This of course powerfully echoes Marx’s characterisations of production as always a ‘double relationship’ in the supposedly immature German Ideology. Marx concludes – and it is the only logical conclusion – that in these forms of society, ‘Property, therefore, means belonging to a clan’ (ibid., emphasis added).

Now I argued earlier that for Marx, property and production relations were substantially synonymous concepts. He says much the same here, for instance in this passage, three pages later:

The original unity between a particular form of community (clan) and the corresponding property in nature ... which appears in one respect as the particular form of property – has its living reality in a specific mode of production itself, a mode which appears both as a relation between the individuals, and as their specific active relation to inorganic nature, a specific mode of working (which is always family labour, often communal labour) ... This relation as proprietor – not as a result but as a presupposition of labour, i.e. of production – presupposes the individual defined as a member of a clan or community (ibid.: 495)

If property means ‘belonging to a clan’, we must conclude that production relations, in all these forms of society, include all those relations by virtue of which individuals are clan or commune members. Not only do these include kinship relations. They embrace ‘communality of blood, language, customs’, and in the Germanic case a common historical experience or tradition. They extend to the entire polis of antiquity, membership of which made the civis into a Romanus. They encompass, in the Asiatic case, ideology. There is simply no way, for any of these pre-capitalist socioeconomic formations, that we can even begin to exclude ‘superstructural’ terms from the very definition of ‘economic structures’. To do so would make nonsense of Marx’s entire analysis. It would also render the societies at issue incomprehensible.

It will do no harm in passing here to move beyond Marx, to look, for a minute, at modern historical materialist anthropology. Mefflassoux (1972; cf. his 1975) argues that ‘agricultural self-sustaining communities ... rely less on the control of the means of material production than on the means of human reproduction: subsistence and women. Their end is reproduction of life as a precondition to production’. He uses this to explain why social power in the cases he examines tends to be distributed on the basis of age and gender, rather than property as traditionally conceived. Godelier goes still further, arguing of such societies that ‘here relations of kinship serve as relations of production, and this from within’ (1978). Elsewhere he develops a similar case for the polis of ancient Athens (1984). On this basis he rejects any general distinction between basis and superstructure as “levels” of social reality, as distinctions within social reality which are in some sense substantive ... institutional divides in its substance’ (1978). He maintains, as I do, that it is specifically capitalism which ‘for the first time separated economics, politics, religion, kinship, art, etc., as so many distinct institutions’. This is of course why restrictive conceptions of production relations, and the hard and fast base/superstructure distinctions they sustain, do have some phenomenal purchase in bourgeois society (and cease to as soon as we move beyond its historical boundaries). Such work is immensely valuable precisely because it is not founded in merely theoretical argument, and recognises criteria of adequacy other than coherence alone. If Marxist anthropologists have been forced (as was Marx himself, a century earlier) to take a broad view of what comprises an ‘economic structure’ it is because the facts have resisted characterisation otherwise.


To return to the Grundrisse. A little later, Marx significantly extends the argument I have traced thus far, telling us that ‘slavery and serfdom are ... only further developments of the form of property resting on the clan system’ (1858a: 493). The analysis of property developed here, then, would seem to apply to all the non-capitalist ‘epochs in the economic formation of society’ – Asiatic, ancient, feudal – delineated in the 1859 Preface: a very wide historical compass indeed. With that in mind, we may turn to the second substantive analysis of property/production relations in Marx I wish to discuss, that of feudalism. His fullest treatment of the topic is in Capital III, where he raises it in the context of an inquiry into the origins of capitalist ground-rent.

As with the section of the Grundrisse we have just been examining, Marx’s underlying concern in this text is with what distinguishes capitalist forms of landed property, so he focuses on feudal forms mainly for their contrasts with capitalism rather than for their own sake. The major contrast he draws is this. Unlike in capitalism, where labourers are separated from the means of production, in feudal society ‘the direct producer ... is to be found ... in possession of his own means of production, the necessary material labour conditions required for the realisation of his labour and the production of his means of subsistence’ (1865a: 790). This difference has an important implication for the mechanism of exploitation – the transfer of surplus labour or its products to the ruling class – in the two societies. We should remember, in discussing this, that this mechanism is absolutely crucial to Marx in characterising a mode of production:

The specific economic form, in which unpaid surplus-labour is pumped out of direct producers, determines the relationship of rulers and ruled, as it grows directly out of production itself and, in turn, reacts upon it as a determining element. Upon this, however, is founded the entire formation of the economic community which grows up out of the production relations themselves, thereby simultaneously its specific political form. It is always the direct relationship of the owners of the conditions of production to the direct producers ... which reveals the innermost secret, the hidden basis of the entire social structure. (ibid.: 791)

In capitalism, ‘economic’ pressures alone suffices to ensure that the worker produces surplus for the capitalist. Given the labourer’s lack of means of production, the only conditions under which he can produce his subsistence at all – selling his labour-power for a wage are ones which entail the performance of unpaid surplus-labour. But in feudalism, things are different. From a purely ‘economic’ viewpoint the peasant can produce his subsistence, on the land, and with the animals, tools, and so on, which he possesses, without having in the very process to produce surplus for his lord. Since feudal lords none the less manifestly do extract surplus-labour, and feudal society would not be what it was if they did not, Marx is led to draw the following conclusion:

Under such conditions the surplus-labour for the nominal owner of the land can only be extorted from them by other than economic pressure, whatever the form may be ... Thus, conditions of personal dependence are requisite, a lack of personal freedom, no matter to what extent, and being tied to the soil as its accessory, bondage in the true sense of the word. (ibid.: 791)

It is a conclusion which he generalises. ‘In all forms in which the direct producer remains the “possessor” of the means of production’, he argues, ‘the property relationship must simultaneously appear as a direct relationship of lordship and servitude’ (ibid.: 790, emphasis added). Herrschaft, we must conclude, is here the ‘specific economic form’ in which surplus is pumped, and hence constitutes ‘the innermost secret, the hidden basis of the entire social structure’. Other texts make this explicit. For Capital I, ‘personal dependence here characterises the social relations of production just as much as it does the other spheres of life organised on the basis of that production ... personal dependence forms the groundwork of society’ (1867a: 77). We could hardly ask for a clearer statement of the internality of Herrschaftsverhältnisse to society’s ‘economic structure’. For the Grundrisse, likewise, ‘the master-servant relation is an essential element of appropriation’ (1858a: 500). For Theories of Surplus Value, the latter’s ‘basis is the forcible domination of one section of society over another’ (1863c: 400). In feudal society, relations of personal dependence – relations, moreover, not founded upon ‘effective control’ of the means of production, which Marx assumes the direct producer to enjoy – are essential relations of production, and hence, consistently with the stipulations of the 1859 Preface, form the ‘groundwork of society’.

To reduce Herrschaft, as Cohen does, simply to part-ownership of labour-power is in my view a travesty of what Marx is getting at in these analyses: the essence of the relationship, he makes clear, is ‘appropriation of an alien will’ (ibid.: 500-1, emphasis added), personal subjection, unfreedom. It also begs the (empirical) question of how – in other words, through what set of relations – such alien control over labour-power is actually established. At one point in the Grundrisse Marx goes so far as to assert that feudal landed proprietorship developed ‘out of purely military relations of subordination’ (1858a: 165); given the amount of land administered by the medieval church, one might also want to give due weight to the power of the word. It is ironic, considering the labours of Cohen and others to produce a rechtsfrei definition of property, that in European feudal society by the high middle ages Herrschaft was in fact largely organised through jurisdiction. In Perry Anderson’s words, ‘justice was the ordinary name of power’ (1974a: 153). This has led Rodney Hilton recently to count jurisdiction explicitly as a key social relation of feudal production (1985); unsurprisingly, perhaps, since his own earlier writings estimate that by the twelfth century, more than half of all feudal ruling class income took the form of profits of jurisdiction or taxes, rather than ground-rent as such (1976).

Anderson himself argues similarly, and like Marx generalises the point beyond feudalism alone:

All modes of production in class societies prior to capitalism extract surplus labour from the immediate producers by means of extra-economic coercion. Capitalism is the first mode of production in history in which the means whereby the surplus is pumped out of the direct producers is ‘purely’ economic in form ... All other previous modes of exploitation operate through extra-economic sanctions – kin, customary, religious, legal or political. It is therefore impossible to read them off from economic relations as such. The ‘superstructures’ of kinship, religion, law or the state necessarily enter into the constitutive structure of the mode of production in pre-capitalist social formations. They intervene directly in the ‘internal’ nexus of surplus-extraction, where in capitalist social formations, the first in history to separate the economy as a formally self-contained order, they provide by contrast its ‘external’ preconditions. In consequence, pre-capitalist modes of production cannot be defined except via their political, legal, and ideological superstructures, since these are what determine the type of extra-economic coercion that specifies them. (1974b: 403-4)

I would want to express the point more forcefully, and say that what Anderson is getting at renders the very terms in which he (like Marx himself) formulates his argument – ‘economic’ versus ‘non-economic’ – anachronistic. ‘Superstructures’ cannot reasonably be said to ‘intervene’ in a structure of which they are ‘constitutive’. A far better way of putting it would be to say that in these societies, a distinction between ‘economic relations as such’ and ‘superstructures’ – at least as conventionally drawn – simply does not obtain. Indeed if we persist in using these concepts in the usual way, Anderson’s observation comes very close to denying the applicability of Marx’s claims for the primacy of the ‘mode of production of material life’ outside capitalism. ‘Superstructures’, Anderson appears to be saying, determine the core social relations of these modes, their specific modes of exploitation, upon which, for Marx, ‘is founded the entire formation’ of the economic community’. Anderson’s frequent recourse to quotation marks is perhaps indicative of some unease; I doubt he intends thus to curtail the scope of historical materialism. None the less he goes on to make a very important point.

Since, he says, feudal societies are defined by their ‘non-economic’ mechanisms of surplus-extraction, to specify them in rechtsfrei terms, abstracting from ‘superstructures’, leaves us with no basis for differentiating between them. This has a paradoxical historiographic consequence. ‘If, in effect, the feudal mode of production can be defined independently of the variant juridical and political superstructures which accompany it, such that its presence can be registered throughout the globe wherever primitive and tribal social formations were superseded, the problem arises: how is the unique dynamism of the European theatre of international feudalism to be explained?’ (ibid.: 402). A ‘colour-blind materialism’ (ibid.) eventuates in historical idealism, for the differentiating factors can then, logically, only be sought outside the sphere of ‘material life’.

Cohen does discuss Marx’s analyses of feudalism; he could hardly fail to, given their evident awkwardness for ‘traditional historical materialism’. What he says is to my mind implicitly subversive of his overall conception of production relations. He cannot but recognise that for Marx ‘the production relations of slavery or serfdom include the authority of the superior over the producer’s labour-power, and he exploits it by exercising that authority’ (1978: 83). Violence and ideology are thus admitted to be internal to feudal exploitation, or constitutive of the specifically feudal property form. On the face of it, this must collapse the base/superstructure distinction as Cohen conceives it, to yield a broader conception of economic structure. Cohen tries to evade the difficulty – in so far as he recognises it at all – by arguing that the serf is not in fact the owner, but merely the possessor, of his plot, so that ‘the rights he enjoys over it are tied to the performance of his duties’ (ibid.: 84). To the extent that we can speak of ‘ownership’ without anachronism in this context, this is undoubtedly true, but it hardly helps Cohen. He remains hoist by his own petard. For Marx’s point, well taken by Anderson, is that ‘economically’ the serf precisely is in a position to produce his own subsistence, and any rechtsfrei definition of ownership would have to acknowledge this fact. The only thing which curtails his ‘effective control’ of either means of production or his labour-power is the eminently rechtsvoll relationship of Herrschaft. It is only through the latter – Marx’s ‘non-economic’ coercion – that we can in fact meaningfully characterise the lord’s ‘ownership’, as distinct from the peasant’s material possession, at all. As with Marx’s three forms of communal property, the ‘political and legal superstructure’ turns out once more to be internal to the ‘economic structure of society’, a defining element of ‘property’ in the feudal context.


In the early sections of this chapter I sought to establish, on both empirical and theoretical grounds, that we had no good reason for excluding any kind of social relation from being a possible relation of production, or for arbitrarily assigning some social relations to the ‘base’ and others to the ‘superstructure’ of society, a priori. These questions could only be resolved for particular historical forms of society, on empirical criteria. I have now shown that in analysing concrete historical formations Marx himself was no respecter of neat base/superstructure models. Indeed, I have suggested that the apparent separability of the ‘economic sphere’ from Acton’s ‘legal, political and moral relationships of men’ is a phenomenon of capitalism alone, and orthodox conceptions of production relations are therefore methodologically akin, in their immediate generalisation of capitalist appearances, to the fetishised views of productive forces discussed in chapter 2 – a theme I shall take further below.

The clear implication of this argument is that, for Marx, production relations are, very simply, any and all social relations which are demonstrably entailed in a given mode of production, or ‘way in which [people] produce their means of subsistence’ (1846a: 31). To put it the other way around, the production relations of a given mode are all those relations between people, in whose absence they would not be producing in that particular way. Such production relations and the ‘economic structure’ they define, may include Cohen’s ‘material relations’, ‘superstructural’ relations, or relations he deems beyond the purview of historical materialism entirely, like familial forms. Examples of all of these being treated as internal to an economic structure’ can be found in Marx’s work. What is, or is not, a production relation in any given instance can only be determined a posteriori, on the facts of the case. We have no good theoretical grounds for defining the category more restrictively, and Marx’s own practice suggests he did not intend that we should. Since property is for him merely another appellation for production relations, the same holds for what comprises ‘ownership’.

Greg McLennan (1981: 17 ff.), criticising an earlier (and less elaborated) formulation of this argument, has expressed the worry that so open a definitional criterion of production relations might be satisfied by random empirical evidence. His concern is that everything will be collapsed into a vague concept of ‘social relations’, and historical materialism lose its distinctiveness. I do not see the danger. If it is claimed that social relation x is essential to, and therefore a production relation of, mode of production y, the case needs to be empirically argued, and counter-claims can be empirically evaluated. Certainly there may be disputes, but I fail to see why their resolution should be considered ‘arbitrary’. Arbitrariness is only a risk so long as such claims are not subjected to any process of empirical adjudication. What I think worries McLennan (and other critics) is my refusal of a determinate concept of production relations at the level of general theory, which is where Marxists habitually seek it; my point, however, is that in Marx’s analytic framework conceptual determinateness can only be provided a posteriori, for specific historical formations. That is not to say, however, that it cannot be provided at all.

Two final quotations might appropriately be cited by way of conclusion to this long argument. When, in The German Ideology, Marx describes production as a ‘double relationship’, both material and social, he adds: ‘by social we understand the co-operation of several individuals, no matter under what conditions, in what manner, and to what end’ (1846a: 41). We might take this openness, for once, at face value. Similarly with this remark, against Proudhon, in the Grundrisse: ‘human life has from time immemorial rested on production, and, in one way or another, on social production, whose relations we call, precisely, economic relations’ (1858a: 489). What Marx is doing here is so devastatingly simple, and at the same time so genuinely revolutionary, that generations of commentators have somehow managed to overlook it.

He is not, as with ‘traditional historical materialism’, reducing social relations to economic relations as conventionally conceived, or explaining the former in terms of the latter. He is precisely redefining ‘economic’ relations – and thus the ‘economic sphere’, or ‘economic structure’, or ‘economic base’ of society – as comprising the totality of social relations, whatever these may be, which make particular forms of production, and thus of property, possible. These social relations are simultaneously forms of material relation of human beings to nature. This totality is Marx’s ‘groundwork of society’, and its extensiveness indicates why he could plausibly treat material production as being synonymous with production of ‘the society itself, i.e. the human being itself in its social relations’ (1858a: 712), or assert that ‘the relations of production in their totality constitute what are called the social relations, society’ (1847b: 212). Such propositions are grossly reductionist on any other interpretation. As he wrote in 1846, introducing the notion of a ‘mode of production’, this mode ‘must not be considered simply as being the reproduction of the physical existence of ... individuals. Rather it is a definite form of activity of these individuals, a definite form of expressing their life, a definite mode of life on their part’ (1846a: 32). What we have here is less an economic (and still less a ‘technological’) theory of history or society, in any standard sense of the word ‘economic’, than an agenda for an historical sociology of economic forms and phenomena.


The final issue I wish briefly to air in this chapter – and I do not pretend to do any more – concerns what I take to be a serious lacuna in Marx’s work, and the historical materialist tradition generally. This is the question of human reproduction, and its connection with ‘material production’ as Marx conceived it. The context in which I raise this (and what has made it visible as a theoretical problem) is, of course, the renaissance of feminist scholarship in recent years, which questions the paradigms of all classical sociologies, including Marxism. I have argued that Marx’s conception of the ‘material groundwork’ (1867a: 80) of society was far broader than that normally ascribed to him. The question I want to raise here is: was that conception sufficiently broad – or could it be made sufficiently broad – to accommodate (without subordinating) the undoubted sociological insights of feminist analysis? I shall suggest that it was not, but could be made so; and that Marx’s own writings provide some grounds for undertaking the task with his posthumous blessing. Thus to revise historical materialism, however, would leave few familiar Marxist theoretical landmarks intact.

Conventionally, by production Marxists understand production of material goods, and modes of production are defined and differentiated according to how this is effected. Marx speaks of a mode of production as a way in which people produce their means of subsistence (1846a: 31). It is true, and important, that he recognised such production as simultaneously being the production of social relations. But the relations he had in mind were specifically those entailed in the production of goods, such as the kinship relations of the primitive commune, the ancient polis, feudal Herrschaft, or the labour/capital class relation. It is production of goods, therefore, which demarcates the field, and whose social forms constitute the specific object, of historical materialist inquiry. But in certain of Marx’s writings, notably The German Ideology, the notion of ‘production and reproduction of real life’ has a significantly wider compass than this.

Near the beginning of that text Marx and Engels list a number of ‘material’ factors in human life routinely overlooked by idealist historiography, which they say are ‘premises’ of their own viewpoint. The first and second of these are the need to produce (goods) to live, and the development of new human needs and capacities in the process. They continue: ‘the third circumstance which, from the very outset, enters into historical development, is that men, who daily remake their own life, begin to make other men, to propagate their own kind: the relation between man and woman, parents and children, the family’ (1846a: 40). These three ‘aspects of social activity’, they go on to make clear:

are not of course to be taken as three different stages, but just as three aspects or, to make it clear to the Germans, three ‘moments’, which have existed simultaneously since the dawn of history and the first men, and which still assert themselves in history today. (ibid.: 41)

The next sentence – the one I have severally quoted on production as a ‘double relationship’ – includes within the ‘production of life’ both production ‘of one’s own [life] in labour’ and production ‘of fresh life in procreation’. Production of goods and production of people are thus given equal status here, conjointly defining ‘the production of life’ and equally integral to people’s ‘materialistic connection’ (ibid.), Marx’s declared analytic starting-point. A page or two later, in a passage I have already quoted more fully (above, p. 62), Marx refers to ‘the nucleus, the first form’ of both property and division of labour as being ‘the family, where wife and children are slaves of the husband’ (ibid.: 44), while later in the same manuscript he writes of marriage, property, the family’ as ‘the practical basis on which the bourgeoisie has erected its domination’ (ibid.: 195).

It must be said, however, that Marx did not take these fragmentary observations much further in his subsequent work. Engels did, in his Origins of the Family, Private Property and the State (1891), but in ways that are problematic. Marx recognises the importance of familial and clan relations to pre-capitalist property forms, as we have seen, in the Grundrisse and elsewhere – but only in so far as these are relations of production of material goods. The probable reason for this failure to return to family forms in later writings – also stated in The German Ideology – is that Marx considered the family to be ‘a subordinate relationship’ in more developed forms of society (1846a: 40). Thus The Communist Manifesto roundly declares that capitalism has ‘put an end to all feudal, patriarchal, idyllic relations ... and has left remaining no other nexus between man and man than naked self-interest, than callous “cash-payment”’ (1848a: 486-7). This sweeping, and undoubtedly inaccurate, judgment reads somewhat ironically in the light of recent work on the multiple ways in which capitalist production continues to be organised through patriarchal forms of relationship both within and outwith the ‘workplace’ (narrowly considered) and the labour-market. It explains, however, why Marx should have given his third material ‘circumstance’ so little explanatory weight in later writings by comparison with the modes of production of goods.

None the less, it is clear that in principle Marx did consider production of human beings – and thus, presumably, the relations through which such production is organised – to be as essential to ‘production of life’ as is the production of goods. It would have been surprising (and inconsistent) had he not, since he recognised human labour as common to all labour processes, and repeatedly stressed, moreover, that individuals were always social – and therefore specifically and differentially socialised – human beings. His exclusion of ‘the relation between man and woman, parents and children, the family’, therefore, from the ways in which modes (and thus forces and relations) of production are conceptualised, can be defended if, and only if, his claim that this is a ‘subordinate relationship’ stands. He must show that such relations are wholly a corollary or a consequence of ways in which people produce their means of subsistence. Now this is a substantive judgment, open to empirical inquiry. If it turns out to be wrong, then – on Marx’s own, self-declared premises – the notion of ‘production of life’, and the concepts employed to analyse it, must perforce be extended to encompass modes, forces and relations of human reproduction as well as of ‘material production’ more narrowly considered.

The historical record, I suggest, powerfully argues for just such a revision. The relations, generally of a patriarchal sort, within which social individuals are produced cannot wholly be reduced to aspects or epiphenomena of modes of production as historical materialism has traditionally conceived them. Indeed if anything patriarchal relations have independently informed and influenced the ways the production of goods has been socially organised, to a massive (and still largely unrecognised) extent. Relations between capital and labour, for instance, were frequently structured through patriarchal, master-servant norms, well into the nineteenth century (and understanding this might make the ‘transition’ to capitalism that much easier to grasp: not as a total rupture, but as a subtler gradation of forms). Outside capitalism’s metropolitan heartlands, they often still are. Developed capitalism rests on a separation of household and enterprise, Max Weber tells us (1978: 375ff.); how can we begin to comprehend that separation empirically without reference to divisions of labour of an eminently patriarchal kind? English law and polity in the nineteenth century – that heyday of supposedly individualist Victorian capitalism – were permeated through and through with the social power of the ‘household head’, the paterfamilias, who alone, as Diana Barker (1978) reminds us, could actually be a ‘freely contracting individual’. That patriarchal household had been an organising matrix of productive activities, political authority, taxation and administration, law and franchise, and religious worship, as well as ‘domestic ideology’, in England, since time out of mind.” Indeed its unspoken assumption underpins even Marx’s own determination of the value of labour-power in Capital I (1867a: 172). Here, a man of his own very patriarchal times, Marx did not seek to unravel ‘natural, self-understood forms of social life’. He took their naturalness” very much for granted. Such is the enduring power of patriarchy.

In fact, these forms of patriarchy would emerge as essential relations of the relevant stages of capitalist production on the criterion advocated above, since they are clearly social relations in whose absence production of goods would not have taken the empirical forms it did. This is undoubtedly an advance on ‘traditional historical materialism’. But that is not the point, or at least not the whole point. To acknowledge the patriarchal dimension of a given set of production relations, conceived as those relations necessary to a mode of production of material goods, is not to explain patriarchy itself. I reject the view – advocated by some Marxist-feminists – that patriarchal relations can be explained with reference to their economic functionality, on both theoretical and empirical grounds. The burden of modern feminist argument is rather to suggest an independent (if very material) basis for age and gender relations which is the particular concern of feminist theory. Now the specific terrain of that theory is precisely ‘the relation of man and woman, parents and children, the family’, or in other words the mode of production (conception, birth, nurturing, socialisation) of human beings – exactly the dimension missing, because considered a ‘subordinate relationship’, from Marx’s analysis of ‘production of life’ after 1846. Explicitly to revise Marx’s concept of production to include – as he first intended – production of human individuals within familial and other relations, and what might follow from that for human society as a whole, might allow the possibility of reconciling Marxist and feminist perspectives without subordinating the specific concerns of the latter. Retention of Marx’s own conception of ‘mode of production’ as the starting-point for sociological analysis, notwithstanding its capacity to recognise the economic import of relations of human reproduction, does not.

Had Marx developed his broader German Ideology view of ‘the production of life’, the conceptual apparatus of historical materialism might have looked very different. Mode, forces and relations of production would be very much wider notions even than those I have advocated here – if indeed those concepts, or only those concepts. remained the fundamental categories of historical materialism at all. Class relations would remain a central dimension, but would not necessarily be seen as the central – let alone the exclusive – dimension of social structure. Age and gender relations would be as integral an analytic concern. One way forward, again mooted in The German Ideology, might be to approach all these social relationships as forms of division of social labour – labour, however, extending well beyond just those activities which produce material goods. And to think the unthinkable, we might not even be differentiating social formations, or periodising history, in the traditional Marxist ways at all. What feudalism and capitalism have in common, for instance – a certain patriarchal infrastructure – might become as significant as what distinguishes them, forms of the sexual division of labour (both within and outwith ‘production’ narrowly considered) being a common presupposition of either set of class relations. These are my views, not Marx’s. To attempt to develop historical materialism along these lines, however, seems to me not inconsistent with the intellectual and emancipatory spirit of Marx’s own enterprise. He was, after all, engaged in a ‘critique of the economic categories’ (1858b), and there is no good reason to see that critique as terminating in the nineteenth century, or bound forever in its ‘natural. self-understood forms of social life’.


Further Reading:
The Forces of Production, (previous chapter) | Marx's 1859 Preface
The Concepts of Capital, Geoff Pilling | Marx's Grundrisse and Hegel's Logic, by Hiroshi Uchida