on which later editors will construct them, even at the expense of the material witnesses.37
With medieval texts these imaginary communities often bear strong nationalistic overtones. This is the case with two of the manuscripts I consider. The first, Digby 23, contains what is often considered the foundational text of the Old French literary canon, the poem known by its modern title as La Chanson de Roland, yet Digby 23 is an Anglo-Norman manuscript, copied and read in communities located in what is now called England. Marie de France, too, has been claimed for French literature as a writer who “gazes from the elegant window of a truly French castle upon a legendary and mythical landscape.”38 But this French castle is no less mythical, hovering somewhere between the modern French nation and the medieval diaspora where some version of romans was the language of cultural prestige but not of national union. With works like the Song of Roland or Marie’s lais, it is the perdurability of the nation and its literary heritage that is at stake, and this can only be undermined by a serious consideration of the manuscripts’ earlier social milieux.
To these varying explanations for the occlusion of the material, we must add one other: the condition of modern mechanical printing has so fused with our understanding of what constitutes a text that it has become difficult to untangle one from the other. The force of mechanical reproduction has allowed the printed text to approach both the stability of the transcendental text and the plenitude of human discourse. When I read a modern novel, I do so with an assumption so confident that I do not normally recognize it that the version I see before me is the one the author signed off on and the one that readers across the world will share. The physical accidents that distinguish one copy from another (such as whether the book is in paperback or hardback, the currency it is sold in, the nature of the cover illustrations, or whether it is second-hand) all appear trivial. The assumption of total stability can be seen clearly in the conventions of academic referencing. A footnote assumes that the essential text will be the same for all readers and distinguishes between the essential information required (which edition is being used, for example) and inessential information (such as whether the book is in hardback), which it simply omits. In this world, the materiality of the books fades before the order of print.
In most cases involving works published during the last century, the assumption of textual stability is not wildly wrong. It would be harder for McGann, McKenzie, Chartier, or de Grazia and Stallybrass to make the case for the attention to textual materials of recent authors who appear only in modern printed editions. Although, to use one of McGann’s examples, W. H. Auden’s decision to revoke “September 1, 1939” makes the editorial history of this one poem of critical moment, the majority of Auden’s poems reappear in various editions with wording, spelling, capitalization, and even punctuation that are almost identical. This relative stability is surely part of the reason those who live in a world of modern printed editions find the arguments for textual criticism first trivial and then deeply frustrating. Referring to one sanguine critic of Emily Dickinson and his assurance that the words on the page before him were the absolute poem, McGann notes that “he could not even see such problems.”39
As one moves away from the relative stability of modern printing, the challenge of textual variance becomes more pressing. The apparent stability of a mechanically printed text may on occasion be illusory (as McGann demonstrates, a reader who thinks there is no textual problem with regard to Dickinson or Auden is living in a fool’s paradise), but only a few novels or poems of the last two centuries would approach the degree of fluidity, the continual mouvance, that is the norm for vernacular texts in the Middle Ages. The stability of the modern printed text and its apparent existence as a self-contained object have set the limits of our understanding of what a text is. If we are to see the problem, we must try to understand how this has happened.
This development did not happen quickly or easily. Printed books were not inherently reliable or stable and only became so within an elaborate system of regulation. As Adrian Johns has shown, it took several centuries to move from the slippery world of pirated editions and clandestine volumes of the early book trade and establish something approaching modern copyright, in which author and text have clear and stable identities.40 Into the eighteenth century, “Unauthorized translations, epitomes, imitations, and other varieties of ‘impropriety’ were … routine hazards.”41 The tribulations of John Flamsteed (1646–1719), Astronomer Royal, at the hands of Grub Street pirates offer a case in point. Johns notes that an early modern reader “could not necessarily take for granted that something calling itself John Flamsteed’s Historia Cœlestis would be owned by Flamsteed himself as the product of his authorship.”42 In England, under the direction of the powerful Stationers’ Hall, a combination of commercial organization and government licensing gradually curtailed illicit copies and ensured reliable transmission. Only under these conditions could printed material inspire general trust, a precondition for the widespread dissemination of the new experimental philosophy to which Flamsteed was contributing.43 Once those conditions were established, however, it became very difficult to think outside them. As Johns notes, “We ourselves routinely rely on stable communications in our making and maintenance of knowledge.… That stability helps to underpin the confidence we feel in our impressions and beliefs.… Even the brisk skepticism we may express about certain printed materials—tabloid newspapers, say—rests on it, inasmuch as we feel confident that we can readily and consistently identify what it is we are scorning.”44 Reliable print was not just a prerequisite for modern ways of knowing but became inseparable from them. An author was someone whose writings had been accredited by being printed; knowledge was what could be expressed in a printed book.45 To this day the term “publication” in academic parlance essentially means printing. As print became knowledge, all that was not print ceased to be knowledge, so that both handwritten documents and speech fell increasingly into a nebulous realm of untrustworthy ephemera.
Not that print became more exclusive. On the contrary, one of the reasons it is difficult to think outside the norms of print is that print covers so much. While more prestigious texts would eventually circulate freely across much of the world in uniform and well-identified editions, cheaper forms of printed material would cover an ever wider range of social discourse. In Europe, steam-driven printing gave rise to a spate of cheap publications: posters, broadsheets, advertisements, political and religious pamphlets, billboards, journals, and newspapers, as well as popular novels, and these ventured into colloquial, erotic, and quotidian areas that had previously only been expressed in speech or private writings. By the time mechanical printing reaches its full force in the later nineteenth century, print almost seems coextensive with human discourse, as Marc Angenot demonstrates in his monumental study of the state of social discourse in Paris in a single year, 1889, a study based entirely on printed texts.46 The sheer volume of printed material, combined with its expansion into almost every area of human activity, reinforces the impression that print covers all that can be known.
In the Western tradition, the printed book sets the limits of our understanding of what a book is, and it is the printed book’s apparent self-sufficiency that may be the most difficult limit to think beyond.47 As a commodity that circulates in a marketplace of strangers, a printed book appeals to a social contract that tells the reader either exactly what the book is and who wrote it or alternatively, as in the case of generic fiction such as mysteries, westerns, or romances, what kind of a book it is. A book can indeed be judged by its cover; readers demand this much predictability. This implicit social contract is contained in the paratextual material—the title, colophon, prologue, jacket blurbs, and the like—and in the design and typography of the text itself. From the single object, one can therefore potentially reconstruct much of the book’s social status, as McKenzie demonstrates in the case of Congreve. These implicit contracts can then be assessed against the evidence of actual reading practice, the kinds of poaching and tinkering, or braconage and bricolage, that specific readers have inflicted upon their books.48 To express the problem in the terms used by Chartier, we might say that while the material support of the text can never be limited solely to its physical support in a concrete object, in the case of a printed book, the concrete object
