Instead of describing all these developments or reviewing the extensive cross-fertilisation between contending critical theories, this paper focuses on a single overarching problematic—that of the relation between knowledge and its object—in order to indicate the epistemological shift which gets underway with Saussure's linguistic theory. It is a shift which is all the more emphatic because of the split between structural and functional (socio-linguistic) views of language. Though the early phase of structuralism has been challenged and even superseded by post-structuralism, the epistemological shift remains basic to both and has ideological implications for both marxist and non-marxist critics. Given the continuity of these developments, it is now possible to place the historical locus of this shift within the modernism which flourished in the late nineteenth and early twentieth centuries, and to connect it with contemporary post-structural theory. Finally, in answer to the inevitable question of why this webbed terrain of critical theory needs to be negotiated at all—all I can say is that if the problem of meaning and representation is being grappled with yet again in Europe, then our immediate need is to examine the premises of this enquiry before we are caught up in its methods, since our cultural logic is different and our responsibilities local. Roughly speaking, in the Platonic and Christian world views, reality (whether divine or earthly) was, in theory, knowable. If it could not in practice be fully known this was because of the inadequancy or fallibility of the human mind. The modern moment coinddes with the recognition that reality is in itself indeterminate, that perception is relative and subjective. Within earlier transcendental schema even if the truth about the world was finally unknowable (and desired as unity between subject and object) it was essentially held to exist. With modern theories (starting from about the early nineteenth century) the truth about reality is doomed to be partial, evanescent, bound by empirical constraints, but still experienced to be there. So even as the individual subject increasingly gathers power for himself as the creator of his own reality under the aegis of subject oriented theories of perception, yet the problem of knowing retains some of the contours of traditional epistemology. Either the subject or the object is sought to be illuminated. In Ferdinand de Saussure's (1857-1913) linguistic theory, the basis of structuralist projects, truth or reality do not verifiably exist, only systems for knowing them exist. There is an epistemological shift in emphasis from subject and object, observer and observed, to the relationship between them. This relationship becomes 'the only thing thaican be observed. It becomes the stuff of reality itself.1 The relation between word and meaning, signifier and signified is arbitrary. Since there is no given or rational correspondence between word and meaning, language does not'represent reality. The continuum of sound and the spectrum of possible meaning are carved up by language in a purely arbitrary way.2 Thus arbitrariness comes to define the very nature of both signifier and signified. Reality is only notional, or at best a 'formless chaos.3, to which there is no direct access. Language and its codes relentlessly structure, mediate, even 62 July-September 1984