Setting out this account of multiple structures risks a new kind of reductionism, implying a coherence and completeness in each which earlier was denied to gender relations as a whole. Certainly the exercise of separating structures is pointless unless there is some gain in coherence. But it is important that none of the three structures is or can be independent of the others. The structure of cathexis in some respects reflects inequalities of power; the division of labour partly reflects patterns of cathexis, and so on. In none is there an ultimate determinant, a ‘generative nucleus’ to use Henri Lefebvre’s term, from which the rest of the pattern of gender relations springs.

There is, however, a unity in the field, an orderliness, which needs to be understood. People do not walk around all day with an ache in their shoulders from the sociological cross-pressures. A gay man can be beaten up in the streets of Sydney in a way that somehow fits with the experience of being beaten up in New York. There is some coherence between the division of labour in childcare, the psychodynamics of femininity and the possibilities of women’s liberation.

My argument, briefly, is that this unity is not the unity of a system, as functionalist analysis would imply. Nor is it the expressive unity that would be provided by the existence of a generative nucleus. It is a unity — always imperfect and under construction – of historical composition. I mean ‘composition’ as in music: a tangible, active and often difficult process of bringing elements into connection with each other and thrashing out their relationships. It is a matter of the real historical process of interaction and group formation. The difference from music is not that the process lacks a composer, but that it has a whole stack of them; and that all are inside the piece being composed, since what is being composed is their own lives. The product of the process is not a logical unity but an empirical unification. It happens on particular terms in particular circumstances. At the level of a whole society it produces the gender order to be discussed in the next chapter.

The idea of ‘composition’ implies that a structure may be less than perfectly formed, and a field of practice less than completely governed by a particular structure. In short, the level of systematic – ity in gender relations varies. It is possible for the processes I have called ‘empirical unification’ to work powerfully and achieve a high degree of order, as seems to be the case for the core complex within the structure of power. But if this is achieved, it is not the consequence of an inherent or categorical logic, and never justifies functionalist analysis. It is the outcome of strategy in the historical process of group formation and interaction.

A high degree of systematicity is likely to reflect the dominance of a group whose interests are served by a particular gender order. For instance, the extent to which housing, finance, education and other spheres of life are all organized around the model of the heterosexual couple reflects the dominance of heterosexual interests and the subordination of homosexual people. The extent to which homosexual experience itself has been organized in heterosexual terms (‘butch’ and ‘femme’ for instance) is a striking proof of subordination. A major task of gay-liberation politics has logically been to contest this way of organizing homosexual experience – to ‘desystematize’ or ‘de-compose’ gender, so to speak.

It is historically possible to have a low degree of systematicity, a good deal of incoherence and contestation. And it is possible to have the combination of structured conflict of interest and potential for de-composition that can be recognized as a crisis tendency. This will be explored in chapter 7.