texts continue to merit discussion and debate, which ensures their continued place in our scholarly firmament, not simply conservatism nor unreflective claims of enduring or intrinsic merit. This conversational understanding of canonical persistence has been formulated in terms of 'interpretative communities', 6 perhaps most eloquently by Kermode who concludes that 'the need to go on talking is paramount,' seeing this in quasi-Wittgensteinian terms: 'the only rule common to all interpretation games, the sole family resemblance between them, is that the canonical work, so endlessly discussed, must be assumed to have permanent value and, which is really the same thing, perpetual modernity'. 7 Kermode also opened up another line of argument, highlighting the role of chance and error in the construction of canons, that it is not solely intention-political, conversational or aesthetic-which canonises texts, but that historical good fortune is also part of the process. 8 The canon wars, then, are ramified and complex. One further element of this debate, less touted but equally important, should be mentioned as it opens the second stand to which this essay will attend from the notion of 'kanonikos': pedagogy. 'The canon is an imaginary totality of works' which only becomes concrete through being taught in a real but inevitably selective syllabus. 9 And whilst there may be an ideological/political component to the process which converts an imagined canonical totality into a real syllabus, it is not the only set of forces at work in that transformation. Reflecting on his editorship of American Intellectual History: A Sourcebook, for example, Hollinger is clear that some texts are removed because they do not 'work' in undergraduate teaching, where some which less well represent their historical moment are retained because students can access their import more readily. 10 Readers and teachers, then, impact on the creation and evolution of actual canonical syllabi. Likewise, publishers, permission rights, profitable page limits and the mechanics of publication can also determine the crystallisation of a canon into a teaching tool. While all this reinforces Herrnstein Smith's point that literary 'value', contrary to universalistic arguments about a transcendental canon, is in fact entwined with economic 'value,' 11 it also makes the point that canonicity, the memorialisation of texts and authors, is driven by pedagogic pragmatics as well