The concept of‘social structure’, though fundamental to the social sciences, is ambiguous. Its usages range from the tight and sophisticated models of Piaget, Levi-Strauss and Althusser to the much more numerous cases where anything that shows a detectable pattern at all is called a ‘structure’. Most writing about gender is well towards the latter end of the spectrum. Authors commonly get by with a rough idea that there is some large-scale order in gender relations, but little clarity beyond that.

To short-circuit what could be a very long discussion about definitions, I will assume that the concept of structure is more than another term for ‘pattern’ and refers to the intractability of the social world. It reflects the experience of being up against something, of limits on freedom; and also the experience of being able to operate by proxy, to produce results one’s own capacities would not allow. The concept of social structure expresses the constraints that lie in a given form of social organization (rather than, say, physical facts about the world). ‘Constraints’ may be as crude as the presence of an occupying army. But in most cases the constraints on social practice operate through a more complex interplay of powers and through an array of social institutions. Accordingly, attempts to decode a social structure generally begin by analyzing institutions.

The most sophisticated accounts of gender relations as a social structure, those offered by Juliet Mitchell and Gayle Rubin, focus on the institution of kinship as the cross-cultural basis of sex inequality. Their account of the structure underlying kinship is derived from Claude Levi-Strauss’s classic The Elementary Structures of Kinship, which boiled down the enormous variety collected by ethnographers and historians into variants of a universal basic system of exchange. Levi-Strauss described this as the exchange of women among groups of men and took it to be what constituted society itself. To Mitchell and Rubin this exchange founded the subjection of women.

The notion of ‘structure’ as a fundamental relationship that is not immediately present in social life but underlies the surface complexity of interactions and institutions, is shared by all varieties of structuralism in the social sciences. It is a major advance over simple descriptive notions of structure. But it has considerable problems, in forms like Levi-Strauss’s theory of kinship. The central difficulty that has emerged from two decades of criticism is that structuralism is based on a logic that is incompatible with the concept of practice as the substance of social process, and hence with a thoroughgoing historicity in social analysis. (For this concept see chapter 7 below.) Without historicity, a politics of transformation becomes irrational.

Mitchell tries to reintroduce practice and history and thus save the rationality of feminist politics. She does this by arguing that the underlying structure (the exchange of women), and the patriarchal social order it gives rise to, have been cultural universals up to the era of capitalism but no longer need be. Mitchell’s argument was important in the mid-1970s in providing a theoreti­cal rationale for an autonomous women’s movement. But it also implies that struggle against patriarchy was irrational in all previous periods of history. This seems arbitrary, to put it mildly. To avoid drawing such a line across history, we need to overcome structuralism’s sharp separation between underlying structure and surface practice.

An illustration of how a more active connection can be made is found in another classic on kinship. Michael Young and Peter Willmott’s Family and Kinship in East London describes the matrifocal kinship structure of the working-class community of Bethnal Green, with Mum as the core figure and mother-daughter relations as the axis of the family. This structure is shown in its process of constitution, constantly being made and remade in a very active social practice. Daughters and mothers pop in and out of each others’ houses up to twelve times a day; they exchange services like care in sickness, and negotiate about other family relationships – including the daughter’s marriage. The notion of ‘structure’ here is not abstracted from practice, though it is also not given in experience. The Bethnal Greeners have no concept of ‘matrifocal – ity’. Being present in everyday practice, structure is vulnerable to major changes of practice. This is shown in Young and Willmott’s famous description of the migration to an outer-suburban housing estate, which produced an unwanted and unintended nuclear family pattern. As one of the migrants remarked, ‘It’s like being in a box to die out here’.

The idea of an active presence of structure in practice, and an active constitution of structure by practice, has now been formalized theoretically. It is particularly clear in the ‘dualist’ accounts of structure developed by Pierre Bourdieu and Anthony Giddens. Bourdieu’s Outline of a Theory of Practice ties structure and practice together mainly by an ironic emphasis on the unintended conse­quences of the strategies social actors pursue. The pursuit of individual or family strategies results in the reproduction of the social order from which those strategies took off. Bourdieu’s approach has the great merit of recognizing the inventiveness and energy with which people pursue their lives – an unusual feat in theoretical sociology. But his image of social structure depends so heavily on the idea of social ‘reproduction’ that it is difficult to reconcile with any idea of a historical dynamic, except something that happens behind the backs of the actors. History does happen in Bourdieu’s world, but it is not produced.

Giddens’s ‘theory of structuration’ locks structure and practice even more tightly together. Human practice always presupposes social structure, in the sense that practice necessarily calls into play social rules or resources. Structure is always emergent from practice and is constituted by it. Neither is conceivable without the other.

The balance formulated by Giddens as the ‘duality of structure’ is, of all current frameworks for social theory, the closest to the requirements of a theory of gender. Yet it leaves two outstanding problems. By making the link of structure and practice a logical matter, a requirement of social analysis in general, Giddens closes off the possibility that its form might change in history. This is the possibility raised implicitly by Mitchell and explicitly by the practical politics of liberation movements; its significance for the analysis of gender is evident. And in attempting to convey the impact of a structure as a whole, Giddens moves sharply back towards classical structuralism. His paradigmatic example here is the structure of language, and this is deeply misleading for structures like gender or class. Stressing the ‘virtuality’ of structure, an argument that sees the context of an event as the alternatives allowed by given structural principles, rather than seeing its context as being its history, Giddens is back with the logic of reversible transformations that characterizes structuralism.

Dualist models, then, need an opening towards history. The crucial point is that practice, while presupposing structure in much the sense Bourdieu and Giddens explain, is always responding to a situation. Practice is the transformation of that situation in a particular direction. To describe structure is to specify what it is in the situation that constrains the play of practice. Since the consequence of practice is a transformed situation which is the object of new practice, ‘structure’ specifies the way practice (over time) constrains practice.

Since human action involves free invention (if ‘invention within limits’, to use Bourdieu’s phrase) and human knowledge is reflexive, practice can be turned against what constrains it; so structure can be deliberately the object of practice. But practice cannot escape structure, cannot float free from its circumstances (any more than social actors are simply ‘bearers’ of the structure). It is always obliged to reckon with the constraints that are the precipitate of history. For example, Victorian women rejecting marriage were not free to adopt any other sexual life they pleased. Often the only practicable alternative was chastity.

Most treatments of the social relations of gender make no subdivision of structure. Texts like Michele Barrett’s Women’s Oppression Today certainly distinguish topics such as ideology, education, production and the state. But the strong tendency in feminist thought has been to see all such fields as manifestations of a single structure, the subordination of women and superordi­nation of men. Sex role theory, for once, is in agreement.

That there might be some problem here is suggested by the remarkable proliferation of ‘ultimate causes’ proposed for that single structure. Evolutionary imperatives, hormonal aggression advantage, the physical strength of men, the demands of childbear­ing, the universality of the family, the functional imperatives of capitalism, the sexual division of labour in childcare, and others, jostle each other as explanations. Since none can be decisively established, the rival root causes cancel each other out. So in the feminist theorizing of the later 1970s and the 1980s there is a tendency for subordination/superordination to float free as pure phenomenon with no roots at all.

But this impasse could also suggest that the treatment of structure is too simplified. A fundamentally different approach was proposed by Juliet Mitchell in her first book, Woman’s Estate; indeed it was outlined in her famous article ‘Women: The Longest Revolution’ as early as 1966. Following the method though not the letter of Althusser’s revision of Marxism, Mitchell divided gender relations into four ‘structures’: production, reproduction, socialization and sexuality. Each of them, she argued, generates its own form of the oppression of women. Each has its own historical trajectory, and at different times may be changing at a faster or slower pace than the others. Though Mitchell does not emphasize the point, it is implicit in her argument that the patterns of relationship in the different structures may come into conflict with each other. That is, the structure of gender relations may be internally contradictory.

I would argue that the concepts of internal differentiation, historical unevenness and internal contradiction are essential for understanding the structure of gender relations. This aspect of Mitchell’s thought has been remarkably neglected by later theorists, though the fourfold model has been used to good effect in social history by Kerreen Reiger and Michael Gilding.

One reason why the model has not been taken up in theory is that Mitchell’s version of it was inconsistent. Production, reproduction, etc. are not, strictly speaking, structures at all. They are types of practice, and they overlap each other. Sexuality is implicated in reproduction, clearly enough. Socialization — if we accept feminist analyses of childcare as work – is a form of production. Certainly there are structures to be discovered in these practices, but there is nothing in Mitchell’s argument to imply that the structures active in these domains of practice are separate ones. So Mitchell has pointed towards a new form of structural analysis, but has not produced it.

The detailed research on women’s subordination since Woman’s Estate has nevertheless fulfilled its promise. Studies over the past decade have traced the outline of two substantially different structures of relationship between women and men. One has to do with the division of labour: the organization of housework and childcare, the division between unpaid and paid work, the segregation of labour markets and creation of ‘men’s jobs’ and ‘women’s jobs’, discrimination in training and promotion, unequal wages and unequal exchange. The second has to do with authority, control and coercion: the hierarchies of the state and business, institutional and interpersonal violence, sexual regulation and

surveillance, domestic authority and its contestation.

To say these structures are substantially different is not to imply they are generally separate. Indeed they interweave all the time. It is to point to some basic differences in the ordering of the social relations involved. As a first approximation one might say that the major organizational principle of the former is separation or division, and of the latter, unequal integration. The accumulation of wealth through the production of goods and services follows a different historical trajectory from the institutionalization of power, and has different effects in the shaping of femininity and mascu­linity.

Many of the institutional and psychological issues about gender can be understood in terms of the structures of labour and power, but not all. The ways people create emotional links between each other, and the daily conduct of emotional relationships, seem to follow a different, though undoubtedly social, logic. The issues raised by gay liberation, by psychoanalysis and by feminist arguments on sexuality, are not accounted for by labour and power alone. In short, there seems to be a third major structure. It has to do with the patterning of object-choice, desire and desirability; with the production of heterosexuality and homosexuality and the relationship between them; with the socially structured antagon­isms of gender (woman-hating, man-hating, self-hatred); with trust and distrust, jealousy and solidarity in marriages and other relationships; and with the emotional relationships involved in rearing children.

The argument of this chapter now proceeds on the assumption that these three structures are empirically the major structures of the field of gender relations. That is to say they (a) are discoverable in current gender research and sexual politics, and (b) account for most of the structural dynamics currently understood. The argument does not assume that they are the only discoverable structures, that they exhaust the field. Nor does it claim that they are necessary structures (a claim which in any case might land us back with the metaphysics of ultimate causes). The argument rests on the gentler, more pragmatic but perhaps more demonstrable claim that with a framework like this we can come to a serviceable understanding of current history.

The general conception of social structure as the pattern of constraint on practice inherent in a set of social relations can be made specific in a number of different ways. The structures just outlined – the division of labour, the structure of power and the structure of cathexis – are examples of what I will call ‘structural models’. They operate at a particular level of logical complexity and implicitly make comparisons across historical situations at this level. They lead to questions like what changed in the sexual division of labour as a result of capitalist industrialization; or whether there is a difference in sexual power structure between communist and non-communist states.

In the long debate about structuralism, structural modelling became the be-all and end-all of structural analysis. There was an understandable intoxication with structural models based on virtual transformation, such as Levi-Strauss’s theories of kinship and myth, Piaget’s theory of intelligence, Chomsky’s theory of syntax. They are intellectually powerful, and they gave order and precision to social science at a time when it was dominated by a mixture of functionalism and muddy empiricism. Yet they drew attention away from other possibilities for structural analysis; and especially from the connection with history.

These possibilities have been indicated, in a rather programmatic way, by Lucien Goldmann. His theory of ‘genetic structuralism’ never got beyond a methodological sketch, though he gave classic practical demonstrations of the structural analysis of culture. Goldmann showed that it is possible to study the transformations of a complex structure over real time as opposed to the ‘virtual’ time of structuralism – an approach adopted in the rest of this chapter. His method in The Hidden God also indicates the importance of a second kind of structural analysis, the compiling of what might be called an inventory of the structural features of a given situation.

Where structural models push towards comparisons across situations at a given logical level, structural inventories push towards a more complete exploration of a given situation, address­ing all its levels and dimensions. There is nothing arcane about this. Any historian reviewing the background to a particular event, any politician scrutinizing the current state of play or balance of forces, is compiling a structural inventory. Any attempt to grasp the current moment in sexual politics, to define where we have got to, any attempt to characterize the gender relations of another culture or another time, likewise involves a structural inventory.

Two useful concepts have been developed for this procedure. Jill Matthews speaks of the ‘gender order’, a historically constructed pattern of power relations between men and women and definitions of femininity and masculinity. Following Matthews I will use this term for the structural inventory of an entire society. The concept of the ‘gender regime’ developed in our educational research to describe the state of play in sexual politics within a school involves the same kind of logic on a smaller stage. I will use this term for the structural inventory of a particular institution.

Recognizing the procedure of structural inventory does not introduce a separate set of issues or topics. The division of labour, the structure of power, the structure of cathexis are the major elements of any gender order or gender regime. Structural models and structural inventories are in principle complementary ways of looking at the same facts. In practice they are constantly done together, with shifting emphases. This chapter and the next discuss them separately to emphasize the logic of analysis, but the separation is far from absolute.