Scholarly Editing in Perspective

Summary

Scholarly Editing in Perspective offers a critical reflection on the theory and methods of textual editing, as a contribution to a wider, comparative understanding of editorial practice. The analysis, written in a cogent, concise and accessible manner, offers an insight into the textual-philosophical principles and foundations of scholarly editing from the beginning of the twentieth century to the new opportunities offered by digital technologies in the twenty-first. Scholarly editing is presented as a process that makes an intervention in the text whereby the editor mediates between competing versions of textuality, authorship, and authority. In analysing the assumptions, beliefs, and critical underpinnings of scholarly editing, this Element provides a new perspective on the standard editorial models within the English tradition, how they have evolved, and how they are adapted for the digital age.

1 An Orientation

As a practice, scholarly editing has a long and venerable history. Its origins probably date back to the third century BCE, when scholars at the Library of Alexandria collected, preserved, catalogued, and edited learned works from the ancient world. Aristarchus of Samothrace’s commentary on the textual transmission of the Homeric writings is without doubt the earliest of these first contributions to textual scholarship.

A modern concern with the ‘correct’ state of texts, however, did not emerge in Western culture until the Renaissance, when humanists such as Erasmus studied the textual transmission of biblical texts. From there, concern with the ‘true’ word of God spread to the ‘study of human words’, as Giambattista Vico called philology (1948, 125). By the nineteenth century philology had entered a golden age when the establishment of reliable texts in the vernacular provided the means to study the development of national languages.

In the English-speaking world, concerns about the notoriously imperfect printing of Shakespeare’s plays in the early eighteenth century led to efforts by ‘gentlemen-scholars’ to ‘improve’ the bard’s texts. In the absence of a disciplined method, however, this editorial work often consisted of purging the Shakespearean text of blemishes and infelicities that could not possibly be authorial: the Bard was deemed incapable of writing an imperfect line (McKerrow 1939, 24–35).

Scholars in the early decades of the twentieth century had pushed back against the subjective and sometimes capricious approach of these early editors by introducing stricter rules and principles. But these rules did not always serve editors well either. The pendulum had swung the other way. A. E. Housman (1921) complained that editing had become too ‘scientific’ and he issued a strong warning that editing should not amount to a blind application of arbitrary rules. Instead, the imponderable problems with Shakespeare texts demanded a solid but flexible methodological and theoretical basis that would strike a balance between the blind application of a set of a priori rules and editing according to taste without any rules at all. In his famous ‘The Rationale of Copy-Text’ (1950), W. W. Greg harmonised the strictly procedural approach to editing with the use of evidence-based critical judgement in the resolution of textual errors. For the first time, a theory of scholarly editing emerged.

Scholarly Editing in Perspective looks back at how theories of scholarly editing developed over the last eighty years. Although this book is not intended as a guide, it is written with non-expert readers in mind.¹ The idea is to help newcomers navigate the vast hinterland of scholarship on the subject, explaining in a cogent, concise, and accessible manner the principles and foundations that have underpinned editorial practice. For this, I have gone back to the sources. Rather than providing my own definitions and clarifications, I have anchored my explanations of key terms and concepts in the words of the original thinkers and commentators as they reflected on and argued about their methods, approaches, and perspectives on what editing is and should do. The main impetus behind this Element, however, is a critical unravelling of scholarly editing: what it is, what it does, how it does it, and what for? It undertakes this analysis in light of the wider perspective that the ‘Editing and Editors’ strand in the ‘Publishing and Book Culture’ Elements series invites. Considering editing as a form of making and mediation, it directs our efforts to a new understanding of the co-creative aspects and conditions that are common across different forms of editorial practice. As such, this strand redresses the fact that, as Susan Greenberg observes, little attention is devoted to a critical analysis of editing across time, medium, and technology – with the exception of scholarly editing (2010, 8; 2018, 3, 232). Notwithstanding this preoccupation with theory, it remains relevant to look at scholarly editing anew from the new comparative perspective that this strand offers: in short, how the critical triangulation of editor, author, and reader works in literary and documentary editing.

As such, this book differs significantly from other monographs that offer a theoretical and critical analysis of scholarly editing, such as those by Van Hulle (2004), Shillingsburg (2017), and Eggert (2019), which make the case for analysing the material processes of production and dissemination as a way of understanding the literary work. I wholeheartedly subscribe to their perspective. However, whereas these scholars offer an ontological take on the theories and methodologies of scholarly editing, my book follows an epistemological approach. Scholarly Editing in Perspective is a self-reflexive endeavour to unfold what happens in and during the editorial process. Specifically, I engage in an examination of scholarly editing’s methodologies, exploring their critical foundations, justifications, and ramifications. I do this in three stages: (1) an elucidation of the main paradigms and models that have occupied the field from the middle of the twentieth century; (2) an inquiry into the editor’s position mediating between author and reader centred on questions of authority and authorisation; and (3) an exploration of how the digital medium has shaped – and continues to reshape – the methods, ambitions, and perspectives on editing. By taking stock of these changes, the third stage moreover probes how they have transformed our understanding of the role of the editor and the editorial process and how this has generated new pressing concerns for questions around identity, inclusivity, and representation. Who is ‘deserving’ of a scholarly edition? Who can access it?

To begin with, let us consider what scholarly editing actually is and how it can be fruitfully informed by the ‘poetics’ of editing. Simply defined, scholarly editing is an activity by which an editor establishes a reliable text of a literary work or document from the past. There are two reasons why such reliable texts need to be created. The first reason is that a text is faulty because of errors that have occurred in its transmission; the editor ‘restores’ the text to its ‘intended’ form. The second involves a text existing in a form that is difficult to access, for instance a manuscript or rare printed pamphlet, and make it available to a wider readership.

What makes the edited text ‘reliable’ is the crux of the matter. Correcting textual errors is a straightforward matter if we know where and when the error occurs and what the correct reading is. The humorous compositor’s lapse (was it Virginia Woolf?) in the Hogarth Press edition of The Waste Land that turned ‘A crowd flower over London Bridge’ into ‘flowed under London Bridge’ (Eliot 2015, I, 57 and II, 376) is simple enough to fix. But not all errors are this obvious, nor is their resolution clear-cut. Exacting levels of textual analysis and critical judgement are needed; even so correcting the text is often a matter of conjecture. Such knotty problems – or textual ‘cruxes’ – are rampant in Shakespeare’s texts, so much so that they have kept editors busy for centuries. Falstaff’s nose, for instance, is described as being ‘as sharpe as a Pen, and Table of green field’.² One of the more famous cruxes, the second clause is obviously nonsensical: garbled by the compositors of the First Folio. Since the clause is missing from the Quarto text, it is practically impossible to reconstruct what the original might have been.

Creating a reliable text from historical documents involves a different set of procedures. Because their physical form is different from modern texts, documentary editing, which emerged in the eighteenth century with the rise of Antiquarian studies (Kline and Holbrook Perdue 2008, 4; Hunter 2009, 4–5), is not just about duplicating what is found in the sources. The translation of handwriting or old typography into new print requires a series of decisions and changes that are less straightforward than might appear (Kline and Holbrook Perdue 2008, 21). Along with wider issues involving selection, arrangement, and general presentation, documentary editing is no less critical than literary editing.

All these actions and procedures have rhyme and reason, but while to some they may seem mechanical, or even intrusive, editing is not merely an auxiliary instrument. It is a critical inquiry in its own right, not literary criticism’s handmaiden.

Critically editing a text indeed constitutes an intervention in that text and hence also in the meanings produced by that text. Such intervention, therefore, cannot happen nilly-willy or on a whim according to the editor’s preferences, beliefs, and tastes; it must respect the authority of both the author and the text. Editorial practice thus requires careful thought and consideration. The editor must be guided by, on the one hand, the textual evidence and, on the other, a set of well-designed principles that make the process reasonable and reasoned. These guiding principles – or editorial rationales – must be logical, practical, and compatible with the textual and bibliographical conditions of the work to which they are applied; furthermore, they must be consistently applied across the edition. It is the combination of all of the above that make the edition scholarly.³

In analysing the critical basis of the methods applied in scholarly editing, this book provides a new perspective on the standard editorial models found predominantly within the English-speaking tradition. While tracing how they evolved and were adapted for the digital age, these models will be situated, moreover, within wider concerns about authorship, authenticity, and competing versions of textuality that the editor must negotiate in making the interventions.

All of these aspects belong to a ‘poetics of editing’, as defined by Susan Greenberg, who is the first scholar to date to examine all forms of editorial practice (from the ‘curatorial’ role of the commissioning editor in a publishing house to the ‘mechanical’ work of the line- or copy-editor) with a view to uncovering the common principles they share in the way they shape the texts in front of them (2018, 11). The importance of Greenberg’s thinking lies in the way she reorients editing from an act of interference to one of co-creation. This insight is applicable, also, to scholarly editing, because it asks a pressing question about the place of the editor working behind the scenes. Commonly, the editor’s hand remains invisible in the text. Yet the editor sits in between author and reader as a ‘midwife’ or ‘janitor’ (2018, 3), exerting significant leverage over the making of the text. Both the ‘scholarly’ and the ‘practical’ editor perform an act of quality control: to do what is right for the text concerning its accuracy and meaning (Greenberg 2018, 154).

In terms of accuracy, the scholarly editor’s work differs from that of the ‘practical’ editor in that they have an unequivocal responsibility towards the historicity of the text (Shillingsburg 1996, 3). Insofar as the scholarly editor is a participant in the making of the text, one might liken their involvement to that of the conservator, restoring the texts from the past. Still, as Paul Eggert pointedly argues, any act of restoration is by necessity an act of reconstruction, reorienting and reinterpreting the object being restored (2009, 45, 61–62). The scholarly editor, negotiating between past works and present readers, faces a similar dilemma. Accuracy and authenticity may well be incompatible ideals.

For Greenberg, the key point is the correlation between ‘meaning’ and ‘imagination’, between getting the text right and being able to see what is needed to get the text right: ‘to imagine what the text might become’ (2018, 239). Thus, too, the processes of scholarly editing require ‘an act of imagination’ (Bordalejo 2013, 64). This insight is vital to Scholarly Editing in Perspective. Indeed, it is with regard to the imagination that the word perspective does its heavy lifting. To make a scholarly edition that aspires to an adequate representation of the work, one needs to run through a gamut of critical and creative questions regarding the work’s textual condition and the agencies involved in its production and transmission. At one end of the spectrum is textual analysis. What is the available evidence? What does it mean? How does it fit together? At the other end is design. How do we meaningfully present the links and connections? How do we convey the edition’s argument?

In raising these questions, and the issue of reliability tout court, this book concerns itself with authorship, authority, and authorisation, as it shifts the discussion about scholarly editing from what it is to what it does for the literary work and its readers. Although it was not always thus previously, the notion that the scholarly editor mediates between works from the past and readers in the present is now an accepted tenet in textual scholarship. Even if the task of the scholarly editor is oriented towards restoration rather than augmentation, the outcome is nonetheless a form of co-creation between author and editor across space and time.

What these considerations indicate is the extent to which the making of a scholarly edition is by default a speculative, critical-creative undertaking: an exploration of the theoretical and practical opportunities that make the edition possible (McGann 2001, 80). Once again, these ‘concrete acts of imagining’, of envisioning ‘what you don’t know’, belong to the very substance of the poetics of editing (McGann 2001, 83; Greenberg 2018, 239) at the crossroads where the critical and the creative come together (Nabugodi and Ohge 2022, 2–3). In this regard, Scholarly Editing in Perspective links up with the ‘Editing and Editors’ strand in the ‘Publishing and Book Culture’ Element series, enriching our understanding of all forms of editorial practice.

Let us next turn to the topics covered in this book. The section following this introduction treats the main editorial paradigms and models that have been in circulation since the middle of the twentieth century. Generally speaking, we find only two models: literary (or ‘critical’) and documentary editing. Literary editing is concerned with removing the faults of textual transmission (printing errors and other such mistakes) to create an accurate text. The second relates to the editing of historical (i.e. non-literary) texts to make them available and accessible for a wider readership. Naturally, there can be some fluctuation between these two models.

Within literary editing, several paradigms occur that vary from one another according to how they choose an appropriate editorial rationale suited to deal with the ‘textual condition’: the circumstances of the work’s creation and transmission. This section traces the evolution of these paradigms in a roughly chronological order, identifying where they intersect and contradict one another.

The dominant method in the editing of English literature is eclectic editing, also sometimes referred to as copy-text editing after Greg’s foundational essay, ‘The Rationale of Copy-Text’ (1950/51). Eclectic editing permits the editor to choose on critical grounds the best readings from among the variants in those texts of the work that are closest to the author. What to do in cases of extreme doubt when the editor cannot decide between equally valid variants? As a pragmatic solution, the editor should defer to the copy-text, the text considered textually superior to all other texts and follow its reading. The resulting edition thus combines readings from two or more texts, resulting in a ‘patchwork’ text that did not exist before but which nonetheless better captures the work as the author had intended it.

Developed for the early modern period, Greg’s ‘Rationale’ proved adaptable for works from later periods too. Both the author’s final manuscript or the first edition, properly proofread, served equally well as copy-text. The choice between them did not lie solely in the authority of the document, but involved the editor’s preference for original or final intentions. Greg’s ‘Rationale’ was less suited, however, for authors who substantially revised their work, in effect creating widely different versions that had equal authority. The ‘versioning’ school therefore sought to edit these multiple versions in a way that would enable comparison sometimes in parallel-text editions.

The ‘social’ school of editing, finally, recognised that, besides the text’s ‘linguistic codes’, its ‘bibliographical codes’ hold meaning as well (McGann 1991). In effect, the ‘social’ school was no longer preoccupied with textual errors per se but instead placed new importance on the ‘primacy of the document’ (Gabler 2007), legitimising texts in the form that original readers had encountered them. In that regard, the ‘social’ school was initially seen as advocating a form of un-editing, until the computer offered the digital means to capture both the text’s textual and visual forms. The advent of digital archives and digital genetic editions completed a rapprochement with documentary editing, which since the mid-1980s had itself come under the influence of the theories of McGann and others.

Predominantly the domain of the historians, documentary editing experienced more of gradual evolution in perspectives with fewer clearly-delineated paradigm shifts. The main driver was to make historical texts and documents accessible for study. Although the task of editing single witnesses seemed simpler, there were calls in the 1950s for greater rigour in execution. In the United States, the work of Julian Boyd and Lyman Butterfield on the papers of Thomas Jefferson set a new standard in reliability (Kline and Perdue Holbrook 2008, 6–7). Documentary editors, however, did not take a theoretical interest until some decades later. Despite new standards, Tanselle (1978) had found documentary editions deficient in the way they articulated and applied a coherent editorial rationale. Tanselle’s criticisms prompted new concerns with standardisation of method as well as the sharing of best practice. It also led to the first real paradigm shift when, with the arrival of computer technologies, historical documents came into purview, not just as sources of information but also as documents whose history and materiality were of interest as well. The editor was no longer there to make the documents available for analysis, but to intervene in their unlocking, interpretation, and representation.

The next section, ‘Mediation and Authority’, examines in detail the effects and consequences of the paradigms and models. Although changes in perspective arose from the efforts to best deal with the textual condition of certain authors, this did not lead to a state of chaos in the field with editors increasingly inventing their own ad hoc methods. What emerged instead was an appreciation for textual instability, not as something that needed to be done away with but celebrated. The objective of scholarly editing is not to produce a definitive edition but one that is as good as it can be. Editors increasingly understood that editing involved the making of choices in a way that was ‘critical’ rather than ‘scientific’. The making of a scholarly edition can never attain to the exactitude and certainty of the sciences. As the product of human endeavour, texts are ‘fluid and variable’, not ‘rigid and constant, like lines and numbers’ (Housman 1921, 68); judicious interpretation is therefore key, mediating oftentimes conflicting and incomplete evidence.

This process of mediation comes as a result of different levels of authority that the editor must negotiate. This is where the epistemological approach, flagged earlier, comes into its own. First, there is the authority of the author, but this authority is often precisely the unknown factor with which the editor has to contend. Second, there is the authority of the documents in which the texts are contained. As Greg recognised, the authority of a document is never absolute, but only relative. The reasons are varied and go beyond what Greg had in mind. A fair-copy manuscript, for instance, as the document closest to the author, best reflects authorial intention in so far as the author wrote it. However, the publication of this original text involves interlocutors besides the author who impose their own authority over the text. Magazine editors and trade publishers often impose house styling or editorial cuts, sometimes without the author’s knowledge. Finally, and most problematically, there is the authority of the editor. Their authority, however, rests on the three basic principles of scholarly editing, which are: (1) to respect the textual evidence; (2) to be transparent about the editorial process, that is, to record and account for any editorial intervention; and (3) to use an editorial rationale that is robust, justifiable, and best suited to the nature of the work and its textual condition, and that is consistently applied. The models and paradigms discussed in the previous section provide a solid basis for appropriate scholarly rigour.

The corollary of the editor’s authority is, however, that their presence in the edition is more noticeable than before. When the traditional imperative was to be objective and neutral, the editor was barely visible in the edition. The editor-as-mediator occupying a position in between the author and the work, on the one hand, and the reader, on the other, not only makes the editor more present but also gives rise to various new perspectives. Some of these perspectives are conceptual, while others include a new ethical awareness within editorial practices. The transition to digital editing is largely responsible for these changes and insights. In effect, the critical apperceptions around mediation and authority would not have been possible if it had not been for the way the affordances of digital production directed editors to think again about the fundamental concepts and process that were taken for granted in the age of print.

The final section in this Element title, ‘Interfacing and Interaction’, follows the methodological and theoretical shifts brought about by the use of digital technologies in scholarly editing. By creating new modes of interaction and engagement, digital technologies had a profound impact on the field in the final decade of the twentieth century. The changes they wrought were surprising and unexpected. The entry of computers was at first entirely pragmatic. On the one hand, they made the processing and collating of large numbers of textual variants much easier; while on the other, they freed the edition from the constraints of the printed page.

One immediate outcome of this liberation was the creation of digital archives rather than editions. As the critical apparatus gave way to the more complete digital representation of the textual condition, the idea was that, with easy access to all the evidence, the reader could make their own decisions about which readings were right and which were not. The reader-as-editor philosophy was not without issues, but it did signal a significant shift at the turn of the century away from eclectic editing and towards new forms of documentary editing.

Changes in practice also elicited changes in orientation. The first of these was a new understanding of the user: who is the scholarly edition really for? For decades, the tacit belief existed that the scholarly edition served not just the specialist but readers of all capabilities. This was a consequence of editorial objective being narrowly defined as the attempt to construct the text of the work as the author had intended it (Bowers 1966 123–24; Tanselle 1989, 76). The commercial interests of academic publishers, who sometimes reissue the text stripped of its editorial paraphernalia for the student and general reader market, reinforced the idea. In truth, scholarly editions are not really made for all readers at once but for readers wanting to gain a deeper understanding of the work and its history beyond close reading (Shillingsburg 1996, 37). The digital scholarly edition drives home this point. As complex resources that work like relational databases, they are not so much for reading as for querying the data that they contain. The type of reading they facilitate is a relational reading that thrives on user interaction (Gabler 2018, 129–41). Innovative ideas about interfacing, paradigmatic editions, and ergodic design are gaining attention as a way of enriching the user experience and empowering users.

User engagement and useability, however, are just one area of interest. Another concerns the critical aspects of access, representation, and identity, particularly considering inclusivity and diversity. The design of digital scholarly editions has hitherto performed poorly in its consideration of users with disabilities, while the editorial field as a whole has suffered from a historical lack of diversity, showing a dearth of representation of marginalised voices. Calls for editorial inclusiveness have long been made for editions by and about women; more recently, these calls have been extended to black and queer writers. Recently there has been an uptick in digital editions of writers from the Global South and elsewhere through the emergence of the new documentary editing, which is supported by Scholarly Editing, the journal of the Association for Documentary Editing. This new model accommodates a social definition of authorship. Shifting the focus away from the single author, or the great when men of traditional documentary editing, it situates the author within their wider network of social interactions. This model seems particularly conducive to recovering authors and movements outside of the traditional canon.

This emergent practice also leads to the conclusion that, while there must be space for innovation, we should also continue to devote attention to the opportunities and standards of existing technologies. Alongside envisioning a future for the digital edition, we can embrace the tools that we already have and put them to use with a new goal: to realise a more inclusive editorial practice that explores new avenues of exploration and creative engagement with texts, works, and authors in the digital landscape.

Finally, a word about nomenclature. Throughout this book I use the term scholarly editing to refer to all aspects of preparing and making a scholarly edition. Other terms that are commonly used are textual editing and textual criticism. Although these two terms are largely synonymous, and often used interchangeably by scholars, there is a technical difference. Textual criticism belongs, strictly speaking, to the work of textual analysis: collating variants, establishing stemmata, identifying copy-texts, and everything else that goes with assessing and interpreting the textual evidence. Textual editing belongs to the next stage in which the editors, using the results from their critical analysis, apply their reason and judgement in emending the text. Occasionally, I also refer to textual scholarship. This is an umbrella term that encompasses all forms of textual and bibliographical study and analysis, including, but not limited to, aspects of scholarly editing.

2 Models and Paradigms

Broadly speaking, there are two models of scholarly editing: critical (or literary) and documentary editing. The first is associated with the editing of literature, the second with the editing of historical texts. For a long time, these models existed in a parallel universe, kept apart by disciplinary boundaries that separated literary studies from history. If the two ever spoke about one another, it was with an element of disdain. Thankfully, this is now less the case than it used to be. Alongside these models exist a variety of editorial ‘schools’ which developed a variety of rationales suited to different kinds of text. They emerged from notable paradigm shifts in practice when editors realised that the textual condition of a particular author or their work defied the application of accepted methods. In some cases, it was sufficient to adjust the method. Greg’s theory of copy-text, for instance, developed for the editing of early modern play texts, was adjusted by Fredson Bowers and G. Thomas Tanselle to fit the work of authors from later periods. In other cases, a simple expansion of the original method was not possible, requiring not just a different approach but an entirely new understanding of key concepts.

This section surveys the dominant editorial models and paradigms that exist in the field of scholarly editing. My intent is to trace their evolution – how they emerged, where they intersect, and why they contradict one another. What drives this evolution are the different ways an editor understands the very nature of text, work, and authorship, and the relationship between them. This is called the textual condition, a term that refers to the circumstances of textual creation, transmission, and reception, and to the different versions – or textual and bibliographical states – that result from these processes.⁴ Equally instrumental are the editor’s varying ideas about the end goal: the reasons why editing is desirable or necessary, what purpose the edition serves, what form it takes, and who the edition is for. The answers to these questions have shifted over time. Variations in editorial practice occurred when established methods had reached their limits: what worked for early modern playtexts was by no means suited for modernist poetry. Other changes were precipitated by pressing cultural and political issues, like gender and race, that motivated editors to think differently about their responsibilities. At present, the debate is not only about how to edit canonical works accurately and rigorously but also about how to mediate marginalised voices fairly and inclusively (Butler 1995).

Overall, the direction of travel is one of an increasingly historicist manner of thinking and, to some degree, a rapprochement between critical and documentary editing.

Eclectic Editing

The dominant paradigm in critical editing is eclectic editing (sometimes also referred to as copy-text editing). It is the one that is still used for most major editorial projects published by university presses. Based on Greg’s theories, and later adaptations by Bowers and Tanselle, eclectic editing permits the editor to choose, on critical grounds, the best readings from among the available textual variants in any of the ‘authoritative’ witnesses, for example, in the fair-copy manuscript, the first edition, or a revised edition. What constitutes the ‘best’ reading is a matter of scrutinising the evidence for the forms that are (most likely) authorial. The edited text that ensues is one that is ‘upcycled’. Like an old cabinet whose broken parts are replaced with knobs and panels of another piece, the eclectic text is purged of printing errors with readings borrowed from other sources in the textual record or, when they fail, conjectured by the editor.

But this patching together does not happen willy-nilly. What should the editor do if the record is insubstantial or incomplete or when the editor encounters two variant readings that are both equally valid? For cases like this, editors needed some kind of control text that would help them decide: the copy-text. This is the text that the editor identifies as being generally more reliable than other texts of the work. Although not necessarily free from error, its accuracy is higher: for instance, because the author is known to have personally corrected page proofs. In cases where the correct reading cannot be ascertained on normal critical grounds, the editor defers to the reading in the copy-text as the one that is most likely correct.

As a concept, the copy-text was thus a pragmatic solution for when variants appear equal in authority. R. B. McKerrow coined the term in his edition of The Works of Thomas Nashe as the English answer to the ‘best text’ tradition promulgated in France by Joseph Bédier (1928). The copy-text provided McKerrow with the ‘base’ text of his edition, which he would follow in all respects. Only ‘as regards evident misprints’ would he deviate from this base (1904, xi). The advantage was that, when faced with no obvious choice, rather than guessing, the editor would trust the copy-text. But it also raised a serious question: which misprints are ‘evident’ and which are not?

In ‘The Rationale of Copy-Text’, a polemic response to McKerrow that became one of the most influential and distinctive essays on editorial theory, Greg argued that the editor should never abdicate responsibility in choosing between variants, except when it is absolutely necessary to do so. Logically it does not hold that because a misprint is not obvious, the copy-text always contains the best reading. Instead of McKerrow’s blanket acceptance, the copy-text supplies authority only where the variants are truly indifferent. This is mainly the case with the so-called accidentals (punctuation, spelling, capitalisation, etc.), but not with the substantives (the words of the text). To choose between them is usually a matter of applying critical judgement. By contrast, although they cause differences in meaning, accidentals are arbitrary elements in the text. An editor is on much less solid ground, in that it is much harder to explain whether an accidental is authorial (1950/51, 22).

To understand the distinction between substantives and accidentals, it is important to know that Greg was writing about the early modern period. Matters of punctuation and spelling were not only more fluid than they are now, they were also subject to the ‘habits or inclination’ of the copyist or compositor (Greg 1950/51, 22). In other words, it is practically impossible to tell whether a particular comma or capital is the work of the author or that of another hand involved in the text’s transmission. By contrast, substantives are subject to ‘significant variants’ whose ‘preponderant authority’ is intrinsic (Greg 1950/51, 20, 29). Critical analysis of these variants and the bibliographical context in which the variants appear can reveal the correct reading.

McKerrow’s distinction between obvious and inconspicuous misprints was simply not valid for Greg. Just because misprints are not immediately apparent does not mean they are less significant. Greg consequently raised the bar for scholarly editing. By ensuring that editing became a truly critical process, instead of a mechanical one, he gave new importance to the critical apparatus within the edition, for it is the collation of textual variants that, on the one hand, reveals the inconspicuous misprints and, on the other, provides the evidence and the justification for the editor’s textual emendations (see Eggert 2013, 107).

Most importantly, he created the conditions for a new understanding of the relationship between text and work. By shifting the emphasis away from the copy-text as a whole to individual variants, he introduced a new textual eclecticism that signalled a radical departure from previous editorial work. The copy-text was no longer the substantive edition (McKerrow 1939, 8–13) that furnished the editor with a base text but rather a control text which the editor relied on in case of doubt. The outcome was a composite text that was drawn from multiple editions with ‘comparable authority’ (Greg 1950/51, 28). As such, this new text that the editor created differed from any text the author had known; undoing the accidents of textual transmission, it nonetheless had a higher degree of accuracy.

For Greg, the copy-text was in other words an instrument in the hands of the editor. The textual idealism that he introduced stemmed from the textual condition that he worked with. Because few autograph manuscripts survive for the early modern period, and authors – least of all Shakespeare – did not routinely see their writing through the press, textual transmission was still similar to that in the Middle Ages. Yet the inadvertent paradox – that the reconstruction of a lost original entails the making of a text that did not exist before – was not something that Greg was overly concerned about. The issue was epistemological rather than ontological. Notably, even if the newly established text better reflected the work as the author might have wanted it, Greg did not refer to authorial intention, neither as a guiding concept nor as an editorial goal.

As time went by, resistance to textual idealism grew. At first this happened subtly, then more programmatically, when editors advocated greater respect for the historical particularity of texts they were editing. But before we get to this point, we must follow the trajectory of scholarly editing’s conceptual development. In doing so, I will especially draw attention to the paradoxical tension between historicism and idealism that induces this development. By its very nature, editing is historicist. But the desire to create an editorial method suited to works from different periods and genres necessitates generalisation, which leads in turn to a drift in meaning and mission. Where McKerrow and Greg, for instance, believed that the editor’s task was to reconstruct a text that was closest to the author’s lost original, a later generation of editors spoke about expurgating the errors of transmission in the text in their pursuit of textual purity. It matters, Bowers wrote, that literary critics have access to the words of the authors they study ‘in as close a form to their intention as we can contrive’ (1966, 8).

This new ideal underscored the edited text as an ideational construct in a way that Greg had not foreseen. For Bowers, the eclectic text was an inevitable but necessary outcome. He wanted to represent ‘the words of an author in their most authoritative form’. That the form could not be found in the ‘most authoritative original documents’ was the result of historical accident (1966, 125). The texts of these documents are imperfect because of the inherent flaws in their production and transmission. Tanselle, in this respect, pushes back against the notion that the eclectic text is ahistorical simply because that text did not exist before. All historical work, he argues, attempts to reconstruct past events from artefacts that are incomplete or defective; no archive is an absolutely faithful record of what happened in the past. So, likewise, the ‘mental texts’ that authors have in their mind can be reconstructed from the tangible texts they produce with the help of their compositors and publishers (Tanselle 1996, 57–59).

Both Bowers and Tanselle thus put authorial intention at the heart of the editorial enterprise where Greg, notably, had not. What Bowers and Tanselle recognised, however, is that intention is not the unitary, immanent force despised by the New Critics but rather a series of creative ideas, inventions, and turning-points that drive the making of the work. Logically, it follows that if the editor creates the ‘most authoritative form’ of the work, the edited text is concomitant with the author’s intention. For Bowers, therefore, the purpose of critical emendation is precisely ‘the recovery of the author’s intentions – on the available evidence – from their transmitted form’ (1966, 147).

Of course, recovering this intention is far from straightforward. Tanselle does not explain the method by which the reconstruction of the ‘mental texts’ should proceed.⁵ Nor does Bowers, for that matter. Other textual scholars (Thorpe 1972, 79; Shillingsburg 1996, 34–37; Gabler 2018, 126, 192) have therefore rightly questioned whether intention works at all as a criterion for critical emendation. They turn the problem on its head again when arguing that it is not intention but textual evidence, available through bibliographical research and collation, that provides the key to the best reading.

To cite a curious example from James Joyce’s Ulysses, we know that Joyce intended to write ‘I do not like that other world’ (and not ‘I do not like that other word’ as The Little Review (January 1919) and the first American edition print it), because world is in the fair copy manuscript and appears unaltered until publication in 1922 and all subsequent lifetime editions (Joyce 1986, 154–55 and 1763). Rather than prosaically revealing the sexual fantasies of his main character, Leopold Bloom, Joyce introduces a deliberate slip in Martha Clifford’s letter – a woman with whom Bloom is having an illicit epistolary liaison – to suggest how Bloom and she are worlds apart, so to speak. She does know the ‘real meaning’ of the word Bloom used, but presumably she cottoned on to the suggestion of mild BDSM eroticism that lay behind it. Without the incontrovertible textual evidence, no appeal to intention could solve the problem. This is often the situation in Shakespeare. In King Lear, the choice between the sycophantic rogues who ‘Reneag, affirme, and turne their halcion beakes | With euery gale and varie of their maisters’, as the first Quarto has it, or ‘Revenge, affirme’ in the Folio is a matter of judicious weighing of the semantic differences between reneging and revenging.⁶ This brings us back to the principles set down by Greg. In choosing between variants, the editor applies critical judgement as to what makes sense in context and what is a reasonable explanation for what caused the variation (Greg 1928, 3). It stands to reason that arriving at the ‘most authoritative form’ correlates to the author’s intention. Yet one can never be absolutely certain about this.

Moreover, the question of intention raises itself more profoundly when authors from the eighteenth century onwards become directly involved in seeing their work in print. Having recourse to manuscripts, or knowing that the author has read proofs, should simplify the work of the editor. The opposite happened when the ‘most authoritative form’ was in fact found in competing texts. Making its entry into the literature, the term ‘final authorial intention’ did little to bypass the concomitant conundrum: when is intention final? Is final intention best represented by the manuscript that the author sent to the printer? Or by the first edition that the author was able to read in proof? (McGann 1983, 28–35). Or perhaps, as the case may be, by that last edition that the author revised?

Interestingly, these questions also raise the corollary of ‘final’ intention, namely the concept of ‘first’ or ‘early’ intention. In some cases, the first edition does not represent the author’s true intention because of far-reaching intervention in the text by the publisher, usually because of censorship. The manuscript of Theodore Dreiser’s debut Sister Carrie, for example, passed through the wringer several times. Before the book was submitted to Doubleday, Page & Co, Dreiser’s wife Sara and his friend Arthur Henry made numerous stylistic changes and cuts, and admonished Dreiser to rewrite the ending to make it more palatable to readers. The publisher subsequently insisted that Dreiser remove all profanities and alter any mention of real people, businesses, and books (Shillingsburg 1996, 72–73; Eggert 2009, 192–93). The case is strong for saying that the first edition (1900) was a different work from the one that Dreiser had written. A scholarly editor would therefore reject the first edition as copy-text in favour of Dreiser’s final manuscript, filtering out all the changes introduced since he completed the manuscript. The argument is equally compelling in the case of D. H. Lawrence, who typically made significant compromises by first revising passages that his publishers found objectionable and then begrudgingly allowing them to make any cuts that they deemed necessary. Tanselle certainly believed that, where there is ‘a conflict between author and publisher’, the author’s ‘uninfluenced intentions’ should prevail in the choice of copy-text (1986, 127).

Although Tanselle is correct, the choice between early and late intention involves other factors too, such as historical and social context. Henry Fielding’s The Life and Death of the Late Jonathan Wild, the Great was published in two forms. This satirical novel about the life of a thief and miscreant, whose identity (although based on a historical criminal) was, in the popular imagination, associated with Sir Robert Walpole, first appeared as the third volume of Fielding’s Miscellanies (1743), a collection of poetry, essays, plays, and fiction. Its second appearance was as a separate volume published in 1754. Not only did Fielding revise Jonathan Wild after the successes of Tom Jones (1749) and Amelia (1751), he also removed most of the specific allusions to Walpole, to reflect changed political allegiances (although the general identification was still clear [Miller 1961, 54]) and refashioned himself from ‘Grub-Street hack’ to respectable ‘gentleman, magistrate, and novelist’ (Goldgar 1997, 216). Furthermore, although Fielding’s revisions are, textually speaking, not deep and extensive, bibliographically each text has its own integrity that prevents them from being coalesced in an eclectic edition (Goldgar 1997, 218).

Similarly, the way the Oxford Complete Works of Evelyn Waugh treats copy-text is strongly influenced by its editorial orientation. The decision to go with the first British edition of Waugh’s novels, notwithstanding his penchant for post-publication revision, follows from those texts being ‘intimately bound up with the material and social conditions that prevailed in the period of their genesis’ (Cooke and Milthorpe 2024, 35). When Waugh changed his mind, these later intentions affected the integrity of the work. Therefore, the Oxford Waugh resolutely opted for Waugh’s early intentions, aligning his writings with the historical moment from which they sprang. The matter that affects Richardson and Waugh is not whether the revised text is better or worse than the original, or has in some way become adulterated, but that the author has created different versions of the work, each with their own aesthetic intent (see further Thorpe 1972, 187–91).

The bottom line is that ‘first’ and ‘final’ intentions are used as a factor in selecting the copy-text. The editor’s ability to select readings eclectically from any authoritative text is unchanged. Nonetheless, the differentiation between first and final intention created a new understanding of what ‘authoritative’ entailed. Back in 1939 the choice of copy-text for McKerrow, although a matter ‘of our own judgment’, was not extraordinarily difficult because ‘fortunately’ most editions (for instance Shakespeare’s Bad Quartos) were obviously ‘much inferior to the other or others’ (1939, 13). Greg and Bowers added bibliographical rigour to the process, but still very much focused on a single authority that stood above the others. Only the editing of authors for whom one can identify early and late intentions adds an important qualifying factor as to whose authority is at stake. A different conception of authorship comes into play.

The Cambridge edition of the works of Joseph Conrad illustrates this beautifully. Its origins date back to the 1980s when the edition adopted the classic eclectic model, in which the first English edition was selected as copy-text. In the 2000s, however, a new team of general editors took a different view, choosing as their copy-text the last text from Conrad’s hand, either the typescript or manuscript. Conrad’s writings had been routinely subjected to heavy house-styling, first by the magazines that serialised his novels and then by his publisher, leading to some very substantial differences in punctuation. Conrad’s drafts are usually only lightly punctuated; he added more punctuation during revision, which his wife Jessie would further supplement when she typed out his manuscripts. Even so, the resulting punctuation was more fluid than the exacting and heavy-handed system imposed by the magazines and publishers. The new Cambridge editors argued that Conrad endured rather than accepted his publishers’ punctuation; they favoured the accidentals of the typescripts or, when absent, the final manuscripts. Whereas a good argument can be made for choosing either first edition or typescript/manuscript as copy-text, the editors’ preferences ensue from a different appreciation of the work’s authenticity: the work as private creation or the work as public good. With the first, the author creates for himself; with the second, the author wants his work to be made public. On that score, Tanselle distinguishes between the author’s intentions, which are best represented in manuscript or typescript, and the author’s expectations, which are reflected in the first edition (1989, 76–79). The transformation the work undergoes when prepared for publication – becoming the best it can be – is a crucial stage in its making. Jerome McGann likewise maintains that ‘an author’s intentions toward his manuscript may be quite different – have special aims and reflect special circumstances – from his intentions toward his published text’ (1983, 42). An experienced writer like Conrad expected such transformation (see Eggert 2022).

Ultimately, neither option is editorially better. What we have instead are competing authorities, each with their own characteristics and authenticity. Textual plurality – which is the subject of the next section – is not meant to be overcome, as it was for eclectic editors. It is part of the ineluctable condition of textuality.

Versioning

The idea that both early and final intention can have equal validity brought to the fore, for the first time, the idea of textual plurality. Hitherto, plurality had been an obstacle in the editorial tradition – something that needed resolving. But in the 1980s scholars editing the Romantics did not see how they could overcome the multiplicity of intentions they encountered. The idea of having one authoritative text was incompatible with works that existed in different versions because of authorial revision.

The classic case is that of S. T. Coleridge’s ‘Rhyme of the Ancient Mariner’. The poem’s first outing was in the first edition of The Lyrical Ballads, which he and William Wordsworth published anonymously in 1798. Here the poem was called ‘The Rime of the Ancyent Marinere, in Seven Parts’. In the second edition, Coleridge changed the title to ‘The Ancient Mariner. A Poet’s Reverie’. He also eliminated some of poem’s archaisms and, besides other minor revisions, cut several quartets. Finally, after further intervening revisions, the version in Sibylline Leaves (1817) added narrative sub-headings as well as a raft of new lines (Coleridge 2001, II, 1, 504–39). Altogether, Coleridge produced eighteen different versions (Stillinger 1994, 60–67). The extent of the changes is such that it is practically impossible to edit the poem eclectically. Because any type of amalgamating these versions would be completely unsatisfactory, if not impossible, the only way to do justice to the poem is to treat all versions as having equal authority: no ‘single version … can be equated with the work to the exclusion of all others’ (Stillinger 1994, 134). Versioning thus provided a new historical and documentary perspective on the literary work in the way it shifted the emphasis away from textual states that were considered imperfect to versions that enjoyed independent status. For Stillinger, this independence was tied to their physical embodiment in the manuscripts or printed books from which they were ‘inseparable’ (1994, 133).

The incontestable merits of this textual plurality notwithstanding, editorially there are considerable challenges. How does one represent eighteen versions of ‘The Ancient Mariner’? (Stillinger 1994, 139–40). Parallel texts provided at least some scope to print two versions on facing pages. Yet comparatively few parallel editions saw the light of the day. Of these the Norton edition of William Wordsworth’s The Prelude, edited by Jonathan Wordsworth, M. H. Abrams and Stephen Gill (1979), which offers a version of the 1799 version as well as a parallel presentation of the 1805 and 1850 texts; the Longman King Lear: A Parallel Text, edited by René Weis (1993); and Langland’s Piers Plowman: A Parallel Text Edition of the A, B, C and Z Versions, edited by A.V.C. Schmidt (1995/2008) are probably the best known. The reasons for this low uptake are fairly obvious. Parallel texts take up twice as much space and therefore increase the cost of production for publishers. The advent of digital editions, as Stillinger anticipated (1994, 140), would bypass this obstacle.

Although the parallel text edition signalled an important paradigm shift in editorial theory and practice, its format was not new. The Loeb Classical Library had been using a parallel format since 1912 for their editions of Greek and Roman classics accompanied by English translation. More venerable still is the Complutense Polyglot Bible, printed in Alcalá de Henares (1514–17), which placed the Hebrew, Greek, and Latin texts side by side. The parallel text edition also had its precursor in the variorum edition whose purpose it was to provide a synopsis of textual states anchored to the base-text from which the reader might reconstruct the original versions.

In terms of placing the evidence in readers’ hands, the parallel text edition is easier to use. But that was not the only reason why the parallel was attractive. More so than the variorum, it was particularly suitable for works extensively revised by their authors. On the one hand, this raised the profile of the autograph manuscript as an object of interest for textual study. On the other, as we will see, it also sparked a debate about textual ‘primitivism’, which was at least in part a continuation of the debate surrounding early and late intentions.

A landmark edition that demonstrated the critical potential of autograph manuscripts was Bowers’s Whitman’s Manuscripts: Leaves of Grass (1860): A Parallel Text (1955). The edition reveals an important nexus in the expansion of Leaves of Grass. It places side by side the text of the 1860 edition with the manuscript text of the poems that Whitman composed c. 1875–79, allowing readers to trace the volume’s ceaseless growth. Although Bowers calls his manuscript text a ‘diplomatic’ edition, the edited text only captures features like line breaks, while removing all the dynamic traces of writing.

Expanding Bowers’s model, the ‘Cornell Wordsworth’ series (1975–2007) supplied facsimile images with corresponding transcription of the most important states of Wordsworth’s poems. Minor versions were represented in the form of an apparatus. The volumes are closer to being an ‘archive’ than a parallel text. The three versions of The Prelude for instance each have their own volume. Nonetheless, the series’ origins lie in the parallel-text tradition by including clear-reading texts of Wordsworth’s poems in a ‘contingent state of completion’ (Bushell 2009, 78). These reading texts represent the manuscript not as writing, with all its hesitations and cancellations, but as a finalised text as if all of Wordsworth’s changes and revisions were fulfilled.⁷

These reading texts received the disapprobation of Jack Stillinger in a debate about textual primitivism that implicated the Cornell series and other parallel-text editions. With a certain degree of drama, Stillinger noted that ‘an unintended result’ of these editorial projects was that they were ‘burying, possibly forever, some of Wordsworth’s most admired writing’, forcing ‘the later Wordsworth … out of the picture’ (1989, 4). No edition truly does that, of course: they complement, rather than replace, earlier versions. More to the point, Stillinger resisted the legitimation of the ‘early’ Wordsworth on the grounds of ‘aesthetic politics’ (1989, 26). The critics’ preference for ‘early’ Wordsworth, he indicated, was the result of subjective taste combined with a predilection for the ‘radical’ Wordsworth as opposed to the Conservative poet of later years. In his view, no one version should be considered better than the other; each manifestation has its unique characteristics and its own merits and demerits (1989, 26–27). Stillinger took special umbrage to the prevailing romantic ideology that revising authors make their writing worse by tinkering; any alteration was considered ruinous to the beauty and integrity of the original poem. Yet, despite defining poetry as ‘the spontaneous overflow of powerful feelings’ (1965, 456), Wordsworth himself did not see any contradiction between his theory of creativity and his persistent self-revising. In recognising this, Stillinger vociferously resisted the spurious connection between spontaneity and authenticity.

Still, Stillinger allowed space for a ‘genetic’ perspective that deconstructs the very notion of the final text as well-wrought urn.⁸ As soon as the attention shifts to the process of composition, the outcome of that process looks itself somewhat less stable and permanent in its authority (1989, 27). J.C.C. Mays’s monumental edition of Coleridge’s Poetical Works (2001) was developed with this perspective in mind. The edition treats the poems both as versions that co-exist independently in their respective editions and as versions that overlay and succeed one another genetically (Mays 2001, cxiii).

While this genetic purview emphasises the importance of manuscripts as documents, ‘versioning’ as a method remained text-oriented. This is already apparent from Bowers’s edition of the Whitman manuscripts. But also some of the more extreme proponents of textual primitivism found ‘texts’ lurking in the manuscripts. Wordsworth supplies again the example. It is generally understood that Wordsworth completed at least two versions of The Prelude: the ‘Thirteen Book’ Prelude (which Wordsworth arrived at in 1805) and the ‘Fourteen Book’ Prelude (the version he completed in 1835 and which furnished the text for the first posthumous edition of 1850). There also exists an incomplete early version: the 1799 ‘Two Book’ Prelude. Jonathan Wordsworth has made the case that, in addition to these three versions, Wordsworth also completed a ‘Five Book’ Prelude around 1804 (see Wordsworth 1977). The manuscripts of this ‘version’, however, differ from the others in that for the ‘Two’, ‘Three’, and ‘Four Book’ Prelude at least one scribal document exists that contains the text of The Prelude in a contingent state of completion.⁹ This is not the case with the Five Book Prelude, whose ‘text’ is dispersed across multiple manuscripts and which can, at best, only be reconstructed from them. In other words, Jonathan Wordsworth makes the materiality of the drafts, revealing the inchoate nature of the poem’s gestation, subservient to its textual unity.

Still, while ‘versioning’ remained text-oriented, it acknowledged that a work can exist in multiple forms, whether in printed editions or autograph documents. As such, it was an important step in the trajectory towards privileging the documentary perspective.

Social Editing

Stillinger maintained that the later versions of Wordsworth’s poems should not be summarily set aside, for they are the texts that Wordsworth worked towards and that put him ‘among the English poets’ (1989, 28). The ‘versioning’ theory thus neatly links with the final paradigm in Anglo-American editing associated with the work of Jerome McGann and D. F. McKenzie. Despite formulating their ideas independently of one another, the work of the two textual scholars converged on the single idea that the form in which texts appeared was not something that should be disregarded.

Before articulating his position, McGann expanded the notion of textual authority: ‘The fully authoritative text is therefore always one which has been socially produced; as a result, the critical standard for what constitutes authoritativeness cannot rest with the author and his intentions alone’ (McGann 1983, 75). In the modern period, authors were more closely involved in seeing their work through the press than had been the case in early modern times. Where the intervention of printers and compositors was dismissed by traditional eclectic editors as contamination of the author’s intention, the work done by publishers and copy-editors was to an extent collaborative. Their role in making the author’s work public involved ensuring the best possible quality for the publication. This process necessitated close, direct contact with the author. Ordinarily this means the author proofreading their text, but this collaboration can extend to other areas as well, when author and publisher consult each other on format, typeface, illustrations, and so on (McGann 1983, 34–35, 52).

The upshot is that literary production is not the work of the author alone but the result of ‘specific acts of production’ of which the writing is only the first (McGann 1983, 52). Matters of form, presentation, and layout had hitherto not been of concern to the textual critic. What drew McGann’s attention was the existence of different ‘versions’. His case study is Byron’s The Giaour, a Fragment of a Turkish Tale. In 1813 alone eight editions appeared, each with new additions that in total tripled the poem’s original length. The transformation of Byron’s poem from one version to the next, McGann argues, exposes a situation ‘where intentions are plainly shifting and changing under the pressure of various people and circumstances’ (1983, 62). This realisation shifted attention away from bibliographical analysis as a tool for textual criticism – the means for choosing the copy-text – towards textual history an object of study in its own right.

Arriving at this new understanding of literary production, McGann introduced a theoretical distinction between the ‘linguistic codes’ of a work and its ‘bibliographical codes’. The linguistic codes are the printed signs on the page – what editors are used to calling the text. The bibliographical codes are those elements which give form and shape to that text: the typeface, type size, leading, margins, running headers, and any other feature of mise-en-page. These are features that of old were of no interest to textual critics but that have since gained attention. With his distinction, McGann postulates that even if the ‘linguistic text’ (McGann 1991, 58) is the same, the bibliographical differences between editions create a divergence in meaning. His example is again from Byron. ‘Fare Thee Well!’ (1816) is a poem about Byron’s acrimonious separation from his wife. Privately printed and circulated among his friends, the poem is basically Byron’s saying ‘good riddance’ to Annabella. Soon after The Champion, a Liberal paper, published an unauthorised version framed by an editorial attack accusing Byron of hanging his dirty linen to dry in public. McGann argues that this attack turned the tables on Byron: what was meant as revenge on his wife now was used ‘as a weapon against him’ (1991, 59). Byron’s inclusion of the verse in Poems (1816) was an attempt to regain control over the poem’s message, but only with partial success. What this example demonstrates is that circumstances not in the author’s control have an impact on the way the work is received.

Independently of McGann, D. F. McKenzie advanced a similar idea about the role of bibliographical variance, advocating that ‘bibliographers should be concerned to show that forms effect meaning’. He did so because he believed that, outside the text itself, books in their material form possess ‘an expressive function in conveying meaning’ (1999, 13, 17). Both McGann and McKenzie recognised ‘the intimate relation between book production and textual meaning’ (McGann 1991, 154). Any individual or institution involved in the production and dissemination of the author’s work takes part in the meaning-making process. Key to this social model is that when the physical form changes, readers interpret the text differently – thus also for the first time bringing the reader into the bibliographical domain. Possibly the social model may seem overly deterministic. In the case of ‘Fare Thee Well!’, one might object that it is the paratext, rather than the bibliographical code, that alters the way the readership received this poem. That is not the case with the other authors McGann analyses – William Blake and Ezra Pound – whose mise-en-page is highly iconic; any flattening of the formal features would create a different reading experience. McKenzie, by comparison, is more subtle in his exposition in that textual production affects the reading of that text, but it is the task of the textual scholar to elucidate if, when, and how this happens.¹⁰ He also turns this task back from hermeneutics to bibliography and to understanding the decisions made by author, printer, and publisher in the shaping of the text (McKenzie 1999, 18).

What is important about the social theory of textual production is that investigating the textual condition is no longer a preparatory stage in the work of the editor but now holds a preeminent position in the representation of the work. In other words, the choice between early and late intention was too simple. Instead, each bibliographical iteration of the work possessed its own significance, for it is in these versions that readers encountered them. The argument works most powerfully in the case of mediaeval texts. For conventional editors, the manuscript tradition served only as a means for reconstructing of the Ur-Text; they could only see textual transmission as an ineluctable corruption of the lost original. The ‘new philology’, by contrast, respected the cultural value of each manuscript as a unique historical product representing a distinct moment in the work’s reception (Nichols 1990, 7–8).

This radical move towards the material side of the work posed an important obstacle for the critics of social editing who were uncertain what to make of this new form of multiple authority. Indeed, the only option seemed, as McGann achieved in The Rossetti Archive, to represent these versions in a documentary edition.¹¹ Consequently, the detractors called this kind of activity ‘un-editing’, arguing that this ‘straight’ representation of the text’s materiality signified an abdication of editorial duty (Tanselle 1991, 140–42; see also Shillingsburg 1996, 165–66).

Others saw this abdication as a virtue. Especially some Shakespeare editors were of the view that it was time to pause the proliferation of new editions, and of new theories and perspectives, that provide ‘no end, no closure, and no purpose’. Far better than this ‘editorial encrustation’ would be a return, by means of facsimiles, to the unmediated texts from the Quartos and Folios for a reawakening to the social and historical conditions in which Shakespeare’s works circulated (Taylor 1993, 123, 125). With this new primacy of the document in editing, we have landed in the territory of the other type of editorial activity that evolved parallel to literary editing.

Documentary Editing

Documentary editing concerns practices in the editing of historical (or non-literary) texts. These texts usually exist in only one document (e.g. a letter, diary or will). There are of course also texts that exist in multiple textual states (Hunter 2009, 60–64) and that therefore are not immune from textual instability that arises from textual transmission. Thomas Hobbes’s Leviathan, for example, appeared in three editions, all purporting to be printed in 1651; of these three, the ‘true’ first edition had a complex production history involving several printers and proofs corrections added at different times. These corrections compete with a scribal manuscript with corrections in Hobbes’s hand whose readings differ from the printed editions. Other important works, such as Robert Burton’s Anatomy of Melancholy or Charles Darwin’s On the Origin of Species, were published in successive editions that were revised and expanded by their author (Faulkner et al. 1989; Shillingsburg 2006a). The textual nature of these philosophical and scientific writings is no different from literary works; their editing poses the same challenges for which eclectic editing and versioning are ready solutions.

That is not the case for the vast array of historical papers and records that survive in the archives. Each document is basically unique and therefore its textual accuracy is not in question (although some do exist in multiple versions, e.g. scribal copies made from the original). In essence, the documentary editor represents the text in these documents ‘as is’. But this is a simplified view, for document and text are not identical. The lexical text is obviously crucial, but as a ‘communicative event’ the document also contains non-lexical elements (Shillingsburg 2006b, 65). The question is not only what is on the page but also how it came onto the page. Especially in hand-written documents, the moment and method of inscription are not without significance. It is the editor’s task to be aware of this process and, where relevant capture it. As Stevens and Burg put it, ‘[a]t its heart, documentary editing is a form of translation, converting original documents into readable text’ (Stevens and Burg 1997, 12). This conversion of manuscript to print is a critical endeavour involving a series of editorial decisions about rendering the dynamics of handwritten text into the homogeneity of print. The crux is to interpret what is on the page to begin with.

Like critical editing, documentary editing spans a range of approaches. Such is already apparent from the ambiguity of its name. The word ‘documentary’ can simply mean the editing of documents, but it can also mean representing the document as document.

The first case is the more common one: editing documents from the past for use by historians. The main driver is convenience of access to the information contained therein. However, depending on the nature of the document, there are other use functions as well. A student of mediaeval diplomatics or social conventions in early modern letter writing will be interested in the form and function of the document as well as the text. A historical linguist needs a faithful rendering of language features such as spelling and punctuation (Beltrami 2012). The basic principle is that every feature of the document – from page breaks to unusual signs – should ideally be faithfully preserved or noted. But this is as far as the ‘theory’ of documentary editing goes. Indeed, much writing on documentary editing limited itself to offering practical guidance – which is not to say this type of editing is not ‘critical’. As such, it did not produce the same variation in schools and methodologies as critical editing did.¹²

In the United States, the tradition is the so-called Boyd-Butterfield model, named after the editors of who started editing the Papers of Thomas Jefferson at Princeton University in 1950 (Kline and Holbrook Perdue 2008, 6–7). An important moment in the professionalisation of editing historical documents, this model was held up as a standard owing to the systematic collation of all surviving documents from which the most authoritative version could be derived. Earlier editors had not always been as rigorous; their texts were frequently inaccurate or incomplete, and sometimes sanitised (Riter 2011, 10). While the Boyd-Butterfield model focused on ‘documenting’ the past through authentic documents, these documents were relevant insofar as they provided historical evidence but not as documents in their own right. Standard practice was to normalise the idiosyncrasies of writing and present a legible text. In Britain, too, the ‘Report on Editing Modern Historical Documents’ (1925) was not at all stringent in its application of strictly dutiful transcription, opening the door to modernising spelling, punctuation, and so on, privileging linguistic content over matter and form.

While standardisation benefits legibility, it does create a methodological issue: how far can the editor intervene in the text (Stevens and Burg 1997, 21)? For David Nordloh, the answer is: quite far. Because unpublished texts are rarely polished, complete, and final, he maintained that the documentary editor should enact the intentions that the author themselves did not. This would thereby create a clean text out of the ‘fallibilities and possibilities implicit in the original creation’ (Nordloh 1980, 2). In creating the text that the author ‘would have wanted’, Nordloh’s documentary editor resembles the copy-editor in a publishing house whose role it is to improve the text for consistency and grammar. His position is extreme and therefore easy to contest.

That said, documentary editions tend to favour clean texts over authorial process so as not to do a disservice to readers (Kline and Holbrook Perdue 2008, 174). Hunter (2009, 88) resists ‘hyper-critical’ editions that only serve specialists. At the same time, he is conscious that modernising a text is much more complicated than one might think. Private documents are not normally intended for publication; therefore, their ‘imperfect’ state is inherent to their condition.

One of the important consequences of cleaning up a text is that it creates a sense of order that may be absent in the document, suggesting greater clarity of thought and expression than was the case. The vituperative nature of Ezra Pound’s letters, for example, is elided when his all-caps words, multiple underscorings, and handwritten additions are regularised or omitted. Other commentators – most notably the literary editor Tanselle – have come out in defence of the integrity of the document that must be reproduced, warts and all. Crucially, the writer’s indecisions and hesitations, the cancellations and revisions, tell us something about the process, the state of mind in which the text was made. These features are part of its meaning; removing them is tantamount to falsifying the nature of the document (Tanselle 1978, 47).

Aside from these problematic issues of standardisation, Tanselle’s ‘The Editing of Historical Documents’ amounted to an extraordinary and blistering attack on current practice. He criticised documentary editors for inconsistencies in normalisation, deficiencies in the editorial rationale, and for not (or inadequately) recording textual data and editorial decisions. This lack of procedure was tantamount to a theoretically ‘weak foundation’ upon which documentary editors were uncritically building their editions (Tanselle 1978, 23). The essay sent shockwaves through the field, instigating a paradigm shift that began with the creation of the Association for Documentary Editing (ADE) in 1978. The ADE provided a new forum for discussing editorial practice. In 1987, it published the first comprehensive handbook for documentary editors, A Guide to Documentary Editing (Kline and Holbrook Perdue 2008, 22).

The new organisation differed from the stalwart National Historical Publications and Records Commission and its predecessor. On the one hand, the ADE looked beyond the papers of persons of national importance; on the other, it adopted an interdisciplinary purview, engaging with the ideas and approaches of literary editors (Cook 1981, 2). The result was a change in perspective on the documentary edition’s role. Editions no longer served only as a container for historical information. The form and function of the document came into view as well. As Tanselle had stated, editing is ‘more than a matter of technique’ (1978, 56); it entails a critical process of decision-making. Who to edit? What to include? Who is the edition for? And what choices does the editor make when transcribing and representing documents (Oberg 1982)?

These questions introduced for the first time a theoretical awareness into debates around documentary editing. In particular, they led to a rising concern with the need for greater diversity in gender and race. The final decades of the twentieth century saw an increase in editions devoted to women and African-Americans. In turn, these editions provoked questions as to how the writing of these figures should be edited. As is so often the case with writing from the margins, these editorial projects are about the recovery of lost or silenced voices that need to be framed differently from the editions of ‘Great White Men’. This has an effect, for instance, on the way editors contextualise and elucidate the texts and documents in the annotations (Kline and Holbrook Perdue 2008, 233–34). But the specific modes of production of these documents deserve attention as well. A well-known example is the ‘literary’ work of Emily Dickinson: the furtive conditions in which she composed her poems, the unique material features of her ‘fascicles’, and the unsettled dashes that punctuate her writing can only be truly represented through ‘documentary’ means (Werner 2011).

The new ideas and perspectives that emerged in the final decades of the twentieth century culminated in the position expressed by Carol DeBoer-Langworthy in her 2012 Presidential address to the ADE:

We are the people who figure out that narrative, how the documents fit together to reflect at least one version (sometimes more) of what may have happened or what someone thought or wrote. In doing that, we select and organize the materials for presentation and dissemination. We stand as intermediaries between the original and the surrogate that is being presented for use. All of us in this room know that most documentary editors know more about our subjects than anyone else – even biographers – because we have looked at every leaf and every image in the process of deciding whether that is a comma or not. We have become keepers of the context of these documents, and increasingly, with digital editions, ongoing curators of the corpus as well.

(DeBoer-Langworthy 2013, 5)

The digital revolution, then, also belongs to the paradigm of mediation. Digital editions are of course also better at preserving the material features of documents than print.

Documentary editors first became interested in the digital technology for its storage potential in the 1980s; in effect, CD-ROMs could provide a more efficient means of storing and accessing surrogates compared to microfilm and facsimile editions. (The second were only suited to relatively small archives anyway.) But even in those early days, at least one commentator noted that the digital medium could inspire ‘a more intimate sense of involvement in the documents’ (Schulz 1988, 347). With the foundation of the Model Editions Partnership in 1994, an investigation into the feasibility of creating standard documentary editions began in earnest, with efficiency, reliability, and access as the main objectives (Chesnutt 1995). For one, digital publication on the web was particularly advantageous for accessing archives of individuals from a minority background and enriching the national consciousness (Stevens 1998, 84; Gailey 2011). This aspect is an important motivation behind the ‘New’ Documentary Editing, which I discuss later in this Element. What is essentially new about this type is its emphasis on (re)covering marginalised voices, presenting them within the cultural frameworks of their time, and embedding them within our cultural memory. This is editing as cultural studies.

3 Mediation and Authority

In this section, I examine the effects and outcomes from the models and paradigms treated in the previous section. The changes described there testify to versatility in the way the field self-examines and responds to variations in the textual condition. While these shifts in methods were frequently defended polemically, by the mid-1990s the outcome was a very catholic view of the field in which no single approach or solution was preferred. What emerged, too, was an appreciation for textual instability as something to be celebrated, not done away with. The objective of scholarly editing is no longer to produce a ‘definitive edition’ but one that is as good as it can be. A good edition is one that is both reliable and authoritative.

From the start, scholarly editors have understood that editing involves the making of choices in a process that is critical rather than scientific or empirical. The making of a scholarly edition, in other words, involves mediating between conflicting or incomplete evidence to create a text that is nonetheless better than any edition produced during the author’s lifetime. It also means interrogating the textual condition, no matter how complex or unstable, and translating that information to the reader by means of the edition.

In recent decades, many new questions about the function and functionality of the scholarly edition, and about its audiences, have been prompted by the advent of editing by digital means and the exciting new possibilities that the medium offers. It has also inspired a rethinking of the form, function, and design of the printed edition whose ‘interface’ represents the edition’s argument (see Eggert 2013; Gabler 2018, 315–62). This section takes these discussions as a starting point. In addition to considering what or who they are for, I also ask: what do scholarly editions do? With this question, I aim to change the traditional perspective by seeing the edition not just as a product of, and a tool for, scholarship, but as something performative. I believe this will help us understand better where the ‘authority’ of the edition comes from as well as clarify how editors mediate the work to its readers.

One of the insights of the new poetics of editing is that all forms of editing are performative (Greenberg 2018, 49) and that performance implies agency. The editor takes a text and does something with it: improves its quality, enhances its accuracy, curates its content, or creates a framework for it. Although the specific tasks the scholarly editor undertakes vary from those in other types of editing, the convergences are there in essence. With that in mind, there is much to be gained by using the language of ‘editing’ generally to reflect on the nature of making scholarly editions. All editors – not just the scholarly type – re-make the texts they work with in a cooperative way, whether this is through direct intervention in the text (correcting, rewriting, etc.) or framing its meaning and significance for readers (in the preface or introduction, for instance). The main difference is that the scholarly editor works with texts of authors who are usually dead and whose work has achieved special distinction. Indeed, the making of a scholarly edition is a consolidation of their canonical status. These conditions create additional responsibilities. Hence, again, the importance of mediation and authority.

This process of mediation comes about because of the different levels – or epistemologies – of authority that the editor must negotiate. It almost goes without saying that the editor is, first of all, confronted by the authority of the author. Admittedly, this authority often remains unfathomable. We cannot know exactly what transpired in the author’s mind. Still, insofar as composition is the execution of the author’s intentions, the effects of those intentions are traceable in the textual record. To give a simple example, when Charles Tansey becomes ‘an insufferable bore’ instead of ‘an intolerable’ bore in Mrs Ramsay’s mind in the page proofs of To the Lighthouse (1927), Virginia Woolf makes a deliberate and intentional change (Woolf 2013); the adjectives may be nearly synonymous, yet they have an alternate inflection.¹³ The revision is in need of interpretation, whereby we make sense not only of the text but also of the changes to the text (Shillingsburg 2006b, 70).

Second, the authority of the documents that contain the text supplements the authority of the author. As Greg recognised in 1949, the authority of a document is never absolute, but only relative. The reasons for this are themselves varied (and go beyond what Greg had in mind for early modern texts). As the document closest to an author, a fair-copy manuscript best reflects the author’s intention, at least in so far as they wrote it. But the process of publication entails steps and changes involving other interlocutors besides the author that impose their own authority over the text. These changes can range from the imposition of a house style to editorial cuts. Although many authors were meticulous about matters of punctuation, capitalisation, and italics, many others tacitly accepted the changes effected by their printers and editors as part and parcel of making their text public (Thorpe 1972, 16–17, 141–51). Bigger changes usually prompted some discussion between author and publisher. But even if an author like D. H. Lawrence complained bitterly about having to remove certain passages because of sexual impropriety, he was also aware that his publishers were liable to prosecution for obscenity. Finally, there is the authority of the editor who intervenes in the author’s text. This authority derives from a process of authentication that can be expressed in the three basic principles of scholarly editing, which are: (1) respecting the textual evidence; (2) being transparent about the editorial process, by recording and accounting for any editorial intervention; and (3) applying an editorial rationale that is robust, justifiable, consistently applied, and best suited to the nature of the work’s textual condition. In the following pages, I will expand on the three principles from which the scholarly edition’s authority derives.

Author – Authority – Authorisation

The authority of the editor is a vexed question. What right does the editor have to meddle with the author’s text? The question is without doubt important. However, it surfaces mostly in reviews that express resistance to the work of the editor, stemming from a lack of awareness about the nature and function of the scholarly edition (something I hope this book redresses). Furthermore, there has been no shortage of controversies surrounding scholarly editions: Gabler’s Ulysses (1984/1986) and Gary Taylor’s New Oxford Shakespeare (2016) are among the most notorious. These controversies, some fought very publicly, have created distrust, even if the vast majority of editions are methodologically robust. What causes this hesitancy is an overriding but ultimately ill-informed belief in the integrity of the text as a well-wrought urn that stems from the legacies of formalism. As a matter of fact, there is no such thing as the text. The textual forms in which the work manifests itself are always plural, unstable, and indefinite, and therefore malleable.

Those that invest authority in a single text often do so from a sense of that text’s authenticity. Scholarly editions, consequently, are sometimes treated with suspicion out of fear that they somehow replace the textus receptus – or received text – that readers and critics are familiar with. But familiarity is not an indication of quality. An unwillingness to accept the scholarly edition as accurate, authoritative, or in any way valuable sometimes runs deep among literary critics.

A case that illustrates this unwillingness time and again is James Joyce’s Ulysses. In 1984, the Critical and Synoptic Edition of Ulysses, edited by Hans Walter Gabler, caused quite a stir. The edition was met with both high acclaim and deep suspicion. Among its critics was Bernard Benstock, who asked the very question which opened this section: what right does anyone have ‘to determine the fixed form of Joyce’s Ulysses other than James Joyce’ (1986, 2)? At least in part, the question was rhetorical. Benstock was aware of the imperfections in the available editions. The production of Ulysses was extremely convoluted. Joyce’s extensive and successive revisions had made the work of typists and compositors very difficult. The text of the first edition was riddled with printing errors as a result: words garbled, passages omitted or misplaced, and deliberate word play inadvertently ‘corrected’.

Joyce was irritated. Later attempts to resolve these misprints were unsuccessful, however. First, Joyce undertook a half-hearted attempt to compile a list of errata, abandoning it halfway. Second, while each resetting of the text offered a chance for corrections, they also led to new corruptions. The one serious endeavour to improve the text happened in 1932, when the Odyssey Press was bringing out a limited edition in two volumes. Even then, Joyce delegated the task of correction to Stuart Gilbert, who in turn left it to a freelance editor at the publisher’s (McCleery et al. 2008, 50).

Benstock finds no fault with corrections that bring sense to otherwise incomprehensible passages. What he objected to was the restoration of words and phrases that had inadvertently dropped out: ‘When Gabler restores phrases and passages to Ulysses which were never part of any printed text that passed through Joyce’s hands, he is providing us with marginalia rather than with the entelechy, the form of forms’ (1986, 4). By excluding from the work any text that was not made public, Benstock accords legitimacy only to ‘any printed text’ in an argument seemingly about the limits of the work. As Michel Foucault adumbrates in ‘What Is an Author?’, drawing a boundary around the work is not a self-evident matter. If the work, he asks, does not include the manuscripts, plans, discarded segments, or laundry lists, why not? (1991, 103).

In fact, Benstock’s argument is more contradictory than that and exposes the pitfalls of seeing the text and the work as coterminous. On the one hand, the text of Ulysses is unstable, ‘bits and pieces occasionally fall out’; on the other hand, ‘[t]he text – such as it was – closed in upon itself (for better or worse) to form its own form’ (Benstock 1986, 4–5). The notion of the ‘form of forms’ reminds us of Vladimir Nabokov, who once proclaimed that even if printing errors were found in his texts, they should be left to stand for, now being part of the text’s fabric, such errors legitimately belong to the work and may take on a meaning of their own as a result (Nabokov 1990, 16). But this position is tantamount to crediting the author with passive authorisation, which is a problematic concept (not least because the absence of something cannot be evidence of something).

In the debates over the Gabler Ulysses, critics frequently cite Joyce’s fascination with happy coincidences, making the case for ‘passive authorisation’ – a term used for an author who neglects the opportunity to correct a typographical error. Whether this happens inadvertently or intentionally is hard to say. Inaction does not leave a trace. A good illustration is a passage from Ulysses in which Stephen Dedalus ponders, ‘what is the word known to all men?’ The text of the first edition does not provide an answer; in the Rosenbach fair copy, however, the answer is unequivocally ‘love’ (Joyce 1986, 418). Due to an eye-skip, the typist who made a copy of the manuscript likely omitted the answer by accident. An editor would therefore feel compelled to restore the answer. Because the passage is so symbolically important, it is indeed unlikely that Joyce had not noticed the omission. (He later weaves the question in again elsewhere in the text, this time without providing ‘love’ as the answer.) Therefore, it is said that he enjoyed the existential mystery more than the expository answer and accepted the omission (Ellmann 1986, 30).

Whatever the solution, passive authorisation is based on a misunderstanding of the bon à tirer. Proofreading is never flawless. Authors rarely read proofs against their final manuscript; Joyce certainly did not. So as authors pass their text for publication, the authorisation they give to their text is not absolute. They approve their text globally, but without necessarily approving every individual word or punctuation mark.¹⁴ This too is an important aspect of textual plurality.

The belief in the sacrosanct immutability of the immanent text has not disappeared from literary studies since 1986. More recently, reviewing The Poems of T. S. Eliot, edited by Christopher Ricks and Jim McCue (2015), Steve Ellis welcomed the immense scholarship that went into the making of the edition, but nonetheless hoped that it would not ‘be received as the sole or “authoritative” means of access to Eliot’ (2017, 467). Ellis’s real concern lay with the fear that the edition might replace Collected Poems, 1909–1962 (1963). (Of course, no edition literally replaces a prior text, even if it gains wide acceptance.) In ascribing a definitiveness to the textus receptus, Ellis disavows any fluidity in that edition’s own making. As Shillingsburg puts it, only when one takes no notice of how a text came to be can one ‘maintain the fiction that work of art is monolithic, the product of a single, consistent, ever-improving intention culminating in a perfected end product’ (Shillingsburg 1996, 38). Without doubt Collected Poems embodies a form of Eliot’s work that the poet himself had established. But this form was achieved in stages. The first Collected Poems, 1909–1935, published in 1936, included ‘Burnt Norton’, but not the other poems from Four Quartets (1943), nor the five Occasional poems with which Collected Poems, 1909–1962 finishes. That early edition was itself an expansion of Poems, 1909–1925 (1926). Furthermore, the text of Collected Poems itself, which went through successive editions, was neither stable nor final. Corrections made in Selected Poems (1948) were not transferred to Collected Poems and vice versa, making it not entirely the expression of Eliot’s intentions (Ricks and McCue 2015, xii).¹⁵ Admittedly, the variants are minor. Yet the point is that Eliot was not able to tightly control his text, despite his relatively small output and the extreme care he took in getting it right. The Ricks–McCue edition notes several inconsistencies, exteriorising the fluidity that lies hidden beneath Collected Poems. This is the kind of historicising argument scholarly editions present.

No text can simply be, let alone be definitive. Unlike the scholarly edition, which is open and transparent about how it is constituted, the textus receptus does not normally foreground the process of its own making. These texts therefore seem beguiling in their immediacy. For decades formalist literary theories have given credence to the notion of self-authorising texts, but this is an illusion. Texts are made, not begotten. Even the New Critic Cleanth Brooks realised this: ‘Poems do not grow like cabbages … They are written by human beings’ (Brooks 1968, x).

Ultimately, imbuing a single text with paramount authority is critically unsound. Of course, Greg’s classic theory states that primacy is an important condition in selecting the copy-text. But by no means does this exclusively mean the first edition. A scholarly editor choosing the 1922 Ulysses as copy-text would do so only as a last resort. According to Greg’s theory, the editor should select as their copy-text the edition that best preserves the author’s wishes. Owing to the fraught production process of Ulysses, that choice cannot be the first edition, nor any of the subsequent editions issued during Joyce’s lifetime that introduced almost as many new errors as they corrected old ones. The next logical choice would be to select a final, complete manuscript or typescript. The fact that no such document exists prompted Hans Walter Gabler to construct a ‘continuous manuscript text’ from the fair copies, typescripts, and page proofs that do survive.

Gabler’s controversial method entailed a daring new conceptualisation of copy-text that brought together Anglo-American editorial principles, promulgated by Greg and Bowers, with the ‘genetic’ tradition developed in German editorial theory (see Van Hulle 2004, 28; Gabler 2018, 218–19). Gabler, in effect, adhered to the primacy of the authorial document, using the manuscripts, typescripts, and page proofs to build up the text of Ulysses again, eliminating all instances of ‘transmissional departure’.¹⁶ Retracing Joyce’s composition step by step, he was able to determine precisely where typists and printers had introduced errors and remove them from the text. That the authority of the document – whether that is a handwritten manuscript or a printed text – is paramount is one of the central pillars of scholarly editing.

Moving on from Joyce, I want to delve more deeply into the editorial process and how a scholarly edition can lay claim to authority. Leaving nothing to chance, editors are expected to ensure that their work is rigorous, complete, and accurate to the last detail, even if it makes the apparatus both exhaustive and exhausting. Never tiring of making the point that ‘opinion’ has no place in editing, Bowers reminds us of the lengths an editor can go to support their hypotheses with reliable evidence. Illustrating the importance of analytical bibliography, he demonstrates how for the early modern period in particular a detailed understanding of print-shop practices is fundamental to evaluating textual variants. The variability in printing processes is especially crucial (Bowers 1964, 52). Did one or more compositors set the text? Were one or more presses used for printing? In what order did the formes go through the press? How were proofs corrected, and so on? A simple example is a line from Hamlet (V.1), which in the second Quarto (1604/5) reads, ‘An hour of quiet thereby shall we see’ and in the First Folio (1623) ‘An hour of quiet shortly shall we see’.¹⁷ The line is spoken by the King shortly after Hamlet jumps into Ophelia’s grave; seeing that his son is not in a right state of mind, he bids Laertes to be patient, for Hamlet (he believes) will bring about his own undoing. Which is the correct variant? Because Q2 derives from an authorial manuscript, the so-called ‘foul’ papers, while F1 uses a later scribal copy that is further removed from what Shakespeare wrote, the Quarto would appear to supply the correct reading.¹⁸ This is not the case. The reading ‘thereby’ is in fact a stop-press correction of a clear misprint – ‘An hour of thirtie shall we see’ – that is found in four of the six surviving Q2 copies. (In early modern printshops, printing did not stop while proofreading was going on, resulting in press variants within one print run.) Since ‘thirtie’ is clearly nonsensical, Bowers (1964, 44–45) argues that ‘thereby’ was inserted by the proofreader who was trying to make sense of the passage without recourse to the exemplar. By means of elimination, Bowers concludes that F1’s ‘shortly’ is likely what Shakespeare had written.

This example illustrates the paramount importance of evidence. As a general principle, rigorously respecting the evidence and the authority of the document are what differentiates scholarly editing from other types of editing. The scholarly editor must make sure not to stray beyond the bounds of the evidence, lest they be accused of ‘meddling’ with the text. Therefore, any element which is not documented has no place in the scholarly edition. But there is an important caveat: not all evidence – be it textual, bibliographical, or palaeographical – completely speaks for itself. As the Hamlet example shows, the evidence does not necessarily point in one direction; it needs to be analysed and interpreted to make sense of what it is telling us about how the printed signs on the page came to be. The archival and textual record is often ambiguous and equivocal. Often it is also incomplete, with crucial evidence lacking when vital manuscripts are lost or the precise genealogical relationship between printed editions cannot be ascertained with certainty. These are called textual cruxes. Using critical judgement, the editor determines which readings are preferred over others. This is called conjecture. While this forces the editor sometimes to make difficult choices, as they move out of the territory of ‘the demonstrable’ into that of ‘the probable’ and ‘the possible’ (Bowers 1964, 77), transparency is key. The editor must meticulously record and explain the critical interventions they have made in the text. The parts of the edition where this happens are the critical apparatus and textual notes.

Because textual fluidity and uncertainty are the norm, not the exception, the question of how to resolve textual cruxes is not always straightforward. To make an informed decision, the editor should be directed as far as possible by the ‘general guidance’ of the textual and bibliographical evidence (Bowers 1964, 57) as well as their editorial rationale. This rationale supplies the ‘standard’ that supports the editor in their critical judgement. Later in this section, I will explain the formation and function of the editorial rationale more fully, but first I will look at some examples that elucidate the range and complexity of editorial conundrums.

A common occurrence with authors from the late nineteenth and early twentieth centuries is the overlapping authority of the British and American edition. This situation arose as a result of the 1891 Chase Act, which for the first time gave copyright protection to foreign works in the United States. To protect the labour of American printers, the Act stipulated that the books had to be manufactured in the United States. In practice, this meant that British authors would enter corrections and final revisions on a duplicate set of the English proofs that were then sent to the United States. From this duplicate the local printers would set and print their text. These circumstances make the choice of copy-text somewhat arbitrary. In principle, both the UK and the US edition are ‘final’, especially if the author consistently entered corrections and revisions on both sets of proof. Therefore, as a rule of thumb, editors give priority to the British proofs, since authors rarely saw US proofs. Unfortunately, not all authors were consistent.

The changes Woolf introduced between her UK and US editions sometimes significantly alter the meaning of the work. The ending of ‘The Window’, the first part of To the Lighthouse, furnishes a poignant example. The family tussle over whether the children can sail their boat across the bay to the lighthouse on the following day finishes like this in the British edition:

‘Yes, you were right. It’s going to be wet to-morrow.’ She had not said it, but he knew it. And she looked at him smiling. For she had triumphed again.

(Woolf 1927a, 191)

In the American edition, however, the final paragraph reads:

‘Yes, you were right. It’s going to be wet to-morrow. You won’t be able to go.’ And she looked at him smiling. For she had triumphed again. She had not said it: yet he knew.

(Woolf 1927b, 186)

Mrs. Ramsay’s triumph is that she had proved her husband wrong, who throughout the day had maintained that he would definitely take children to the lighthouse the next morning despite obvious signs that bad weather was coming. The passage speaks of the silent tension in the Ramsays’ marriage. After years together, they each know what the other is thinking. But because Mr Ramsay is stubborn and overbearing in his behaviour, making Mrs Ramsay feel ill at ease, she finds it difficult to express her ideas: ‘So naturally it was always he that said the things’ (Woolf 1927a, 190). Before long he reproaches her for not saying them herself.

The variation in reading between the British and American edition places a different inflection on the passage. Whereas the British text emphasises Mrs Ramsay’s unspoken triumph over her husband, the American text accentuates their silent agreement. Aesthetically, the British text might sound like the stronger version; it also seems more in line with what Woolf wrote in her manuscript, which reads more simply: ‘he knew what she felt: & she need only say to him: “Yes: its [sic] going to be wet tomorrow”’ (NYPL/Berg, Notebook III, f. 25/148). Woolf, first of all, probably thought that ‘You won’t be able to go’ was too explicit. She cancelled this on the UK proofs, but not on those for the US. On these, conversely, she added in ink: ‘She had not said it; yet he knew’ (Mortimer Rare Book Room, Smith College, MS–0000/5/187–89, p. 191). The addition appears at the bottom of the paragraph, without indicating the exact point of insertion, so the American compositors took this to be a new final sentence. Although the British proofs do not survive, it is evident from the UK edition that she had added the new sentence here as well, albeit using different wording. (The grammatically repetitive structure in the UK version seems less terse than the fragmented clauses divided by a semicolon in the US text.) It is hard to imagine that Woolf was unaware she was creating these textual differences.¹⁹

Which of the readings the editor chooses is almost arbitrary. But if in Woolf’s case the variants are equally authoritative, this is not so with my next example: W. B. Yeats’s ‘The Song of the Happy Shepherd’. Choosing between variant readings from the poem’s textual history is really a matter of whose authority one accepts. One line in particular is textually tricky. In its first outing in Poems (1895), the text runs as follows:

Go gather by the humming sea

Some twisted, echo-harbouring shell,

And to its lips thy story tell,

And they thy comforters will be,

Rewording in melodious guile

Thy fretful words a little while

(Yeats 1987, 66)

In 1904, the fifth line in the passage above suddenly appears as ‘Rewarding in melodious guile’ (Yeats 1987, 66), causing a significant change in meaning. Whereas in the first edition the shell’s echo was melodically changing the addressee’s words until they drive away his dejection, the later text rewards the speaker’s words with a beguiling melody. An inveterate reviser, Yeats frequently made alterations from one edition to the next. Whenever these changes were major, this called for a resetting of the text – which in turn could lead to new printing errors. Whether the change from ‘rewording’ to ‘rewarding’ is a revision or a misprint is impossible to say with certainty. One can argue that the new wording appears rather non-sensical and therefore must be erroneous. How can a ‘melodious guile’, a melody more luring than alluring, be a reward? Yet this is the wording that, with one exception, survives in all subsequent editions of Poems (a further ten editions in total) as well as in Early Stories and Poems (1925) and Collected Poems (1933).²⁰ Richard Finneran postulates that, because Yeats did not spot the alleged error, ‘Rewarding’ must be the intended reading. But this is not guaranteed. ‘Rewarding’ is meaningful enough for Yeats to have overlooked the change. Aside from accidentals, the replacement of ‘Rewording’ with ‘Rewarding’ is also the only substantive change in the poem throughout its entire publication history. If it had not been for an aborted Edition de luxe, the poem’s text would have reverted to ‘Rewording’. Left unfinished at the time of Yeats’s death in 1939, the edition was supervised by Yeats’s widow, George, who spotted the anomaly in ‘The Song of the Happy Shepherd’ and declared it an obvious mistake (Finneran 1990, 45–46). Most likely she was right. It is possible that Yeats himself had been ‘beguiled’ by the restorative power of the ethereal melodies produced by the shell, but the comfort provided by this magical intervention goes against the meaning of the poem as a whole, which admonishes us not to seek too ‘fiercely after truth’, for ‘there is no truth | Saving in thine own heart. Seek, then, | No learning from the starry men’ (Yeats 1987, 65–66). The scholarly editor can simply follow the textual evidence and accept the last printed state that emanated from the poet or opt to follow the authority of the poet’s collaborator who more than anyone else knew the workings of his poetic mind.

As the preceding discussion indicates, the choices an editor makes depend on how they define their author’s ‘authorship’. Indeed, Tanselle identifies such definition as a pre-requisite for all types of scholarly editing (1986, 14). In the traditional eclectic edition, the aim is to remove from the text all elements that are external to it. But as we have seen, there are models that recognise the social and processual nature of literary creativity. Some editions respect the wishes of collaborators or the circumstances of production. Yeats’s Poems, edited by A. Norman Jeffares (1989), takes account of George Yeats’s involvement and that of Thomas Mark, Yeats’s editor at Macmillan. D. F. McKenzie’s The Works of William Congreve acknowledges the typographical innovations that Congreve and his printer, Jacob Tonson, effected for his Works (1710). Tonson’s typography was deliberately designed to present Congreve’s work as an oeuvre and therefore as a text for reading, not for the theatre. Yet these editions only contain the vestiges of social editing. McKenzie, the originator of the sociology of the text, requested certain layout features (e.g. the placement of speech prefixes) that diverged from OUP’s house style, yet these were quite minor interventions in what is otherwise a standard eclectic edition (Congreve 2011, I, xi). Only the digital medium can do justice to editions that resolutely orient themselves to a social perspective. Ultimately, authorship is less an objective criterion than an ideological construct. It encompasses the critical interpretation of the creative process combined with the author’s self-fashioning of their public persona and the public perceptions that arise from it.

What then makes the authentication of a scholarly edition robust in the face of what might still seem a subjective position?

Authentication

A good point to start the discussion of this so-called subjective position is Susan Greenberg’s contention that all forms of editing tend to be understood via ‘metaphysical’ binary concepts, such as ‘truth and falsity, original and copy, productive and reproductive, creative and critical, idealism and materialism’ (2018, 143). I have already alluded to some of these binaries. On the one hand, the editor’s authority to edit is a reproductive authority; this authority is vested in the availability of the appropriate textual evidence. Under no condition must the editor deviate from or bypass this evidence. This is the first principle of scholarly editing. Furthermore, to guarantee neutrality and objectivity, the evidence must be carefully laid out and any editorial decision meticulously recorded so that readers can ascertain and verify where and how the editor has intervened in the text. This is the transparency principle that is the second principle of scholarly editing. On the other hand, the textual evidence does not speak for itself; the textual record is almost always ambiguous, contradictory, or incomplete, and therefore the evidence must be critically analysed and interpreted. Because making a literary work public is, as Shillingsburg puts it, ‘fraught with all the difficulties of execution, entailing writing, typing, copying, typesetting, proofing, revising, and correcting’, words like original or authentic do not easily apply to any published text. The exigencies of literary production are such that authors are not in complete control of their work; their wishes are thwarted, inadvertently or deliberately, by mechanical failure and the intervention of others. No work, therefore, can be ‘adequately represented by a single or simple authenticated text’ (Shillingsburg 1996, 35). One might posit that, ideally, the work of the editor is to fulfil these wishes on the author’s behalf. In practice this is not possible. At best, the editor reaches nearer to what the author intended or, in the case of documentary editing, captures a specific historical moment. They do this by following a process of authentication provided by their chosen editorial rationale, which the editor is expected to articulate in their edition. This rationale permits the editor to mediate between evidence and authorship: the third principle of scholarly editing. To this we can add a sub-principle – that the rationale must be applied consistently.

Today editorial rationales come in many shapes and forms. This was not always the case. For most of the twentieth century, eclectic editing based on Greg’s theory of copy-text, which he formulated in 1950, was the only rationale in use. Before Greg other conceptions of copy-text were in circulation that adhered to the idea of ‘best text’ editing. This method stipulated that one text should be selected from all available texts that best represented the author’s intentions and that this text should be corrected removing any obvious errors. Going back even further still, to the nineteenth century and before, we find no presence of any systematic principles by which to edit a text. Indeed, the main benefit of Greg’s theory was that it identified the middle ground between editing as an art and a science, between the application of critical judgement and faithfulness to the evidence (see Housman 1921;Greg 1950, 26). The ingenuity of Greg’s reasoning was such that his theory, developed for conditions in the early modern period, was successfully adapted by Fredson Bowers, among others, for the work of authors from the eighteenth and nineteenth centuries. After its foundation in the 1960s, the Center for Editions of American Authors (CEAA) of the Modern Language Association elevated eclectic editing to a dogma; the Committee awarded their seal of approval only to those editions that had rigorously followed the Greg–Bowers method. This would prove to be an unwise decision.

In scholarly editing, theory and practice are in fact strange bedfellows. At the time there were those who were not swayed by theoretical principles. Morse Peckham (1971), James Thorpe (1972, 180–202), and Philip Gaskell (1978, vii) argued that practice trumps theory in scholarly editing, because of the differences in authorial practices. The specific circumstances of the textual condition therefore make generalisation difficult about how to handle textual variance. Copy-text theory was such a generalisation. Gaskell and Thorpe considered it of little value in emending the text. What complicated matters for the CEAA was that they, perhaps blinded by their own methodological intransigence, had affixed their seal to editions that proved to be less than satisfactory. The problems were twofold: editorial principles that become rules of thumb were blindly followed by editors whose critical understanding and resolution of textual cruxes was simply inept (Freehafer 1975, 376; Shillingsburg 2017, 104). Something of a crisis of faith erupted. In 1976, the CEAA was dismantled and replaced with the Committee on Scholarly Editions (CSE). The CSE took a more catholic view of what constitutes a good edition. It stopped short of prescribing one accepted method and even shied away from providing ‘a detailed step-by-step editorial procedure’ (‘Center’ 1977, 584), rendering their Guidelines less specific and practical as the CEAA’s had been.²¹ The pendulum had swung the other way.

While the variation in possible methods is too wide to capture, it is important to remember that the validity of any editorial rationale does not reside in tried and tested methods only, but rather in ensuring that the rationale is (1) fit for purpose for the textual condition it is dealing with and that (2) the rationale is consistently and correctly applied to the situation at hand.

The Gabler Ulysses is a case in point of an edition that deviated from the standard methods. However, its editorial rationale was developed in response to the specific conditions in which Joyce’s work had come into being. It provided a reasoned and workable template that, when correctly applied, served its purpose well in resolving the thousands of misprints that had marred the first edition. The real test is always on whose authority is the text established. The Cambridge University Press editors of Joseph Conrad’s Tales of Unrest (1898) were faced with a particular difficulty regarding punctuation and copy-text (see Simmons and Stape 2012). As I mentioned above (pp. 21–22), because the Cambridge Conrad is a traditional eclectic edition aiming to capture Conrad’s final intentions before the introduction of intrusive house styling by magazines and trade publishers. For that reason, the edition usually chooses the final typescript or, when non-extant, the fair-copy manuscript as copy-text. During composition, Conrad was usually very sparing with his pointing; punctuation was added during revision and subsequently by his wife Jessie, who produced the typed copies, which were then corrected and augmented by Conrad. Returning to these early versions would restore the fluency and modernity of Conrad’s style. But not only did the stories in Tales of Unrest each have their own publication history before they were gathered in book form, the textual record is very incomplete. For three of the five stories, only the final manuscript or typescript survives.²² For the two other stories the earliest available witness is the text from serial publication. Furthermore, the typescripts are not entirely reliable. Jessie Conrad’s typewriter was defective, which sometimes affected the legibility of punctuation marks. In the face of these obstacles, the editors circumvented the issue of the copy-text altogether and undertook a detailed analysis of Conrad’s pointing habits by comparing how his punctuation in the manuscripts and typescripts differed from the house-styled pointing in the serial text. From this, they in effect retro-engineered Conrad’s punctuation as it might have existed in the missing documents.

As an editorial rationale, this is an ingenuous and by no means impossible solution, but ultimately, without documentary basis, it is highly speculative, raising the question as to how it authenticates the text of the work. A robust editorial rationale must guard against ‘editing to taste’ – a form of editing in which editors make ad hoc decisions about what they believe to be right. As a procedure, it provides the means by which the editor treats the textual evidence, ensuring the editing is done systematically and rigorously.

Mediation – Remediation – Transmediation

In the previous section, I explained that the editorial rationale is effectively a tool in the hands of the editor with which they negotiate the complexities of the surviving evidence. This is what makes editing a critical activity, a way of understanding the work and its textual condition. The literary work is not a concrete, discernible given but a dynamic system of textual variance. Through its composition and publication history, the work regenerates itself within its various manifest forms. The scholarly editor adds another layer to this process of regeneration. The editor’s participation in that process is, as all forms of editing are, collaborative, but the limitations that are placed on scholarly editing in terms of authority make this a collaboration of an unusual kind. In older conceptions of the editorial process, the editor must stand aside as a neutral arbiter on the work’s textual constitution. The editor, however, is an ‘other’ who cannot but stand in the ‘middle’, negotiating between the textual evidence and the conception of the work that the edition wants to present.²³

As a critical term, mediation has gained a lot of traction in textual scholarship. The point is most obvious when we realise that each edition remakes the work, whether through active emendation of the text or discursive reframing. Students of Shakespeare might appreciate this best, when they consider that each new edition is not unlike a new production (Whitworth 1991, 132), except that editions are constrained by the editor’s chosen rationale. Mediation is really about historicity and communication, however: how does the scholarly edition make past literary works speak to audiences in the present? As Shillingsburg puts it (2006b, 12), what editing does is re-presenting the texts: literally making them present again in new forms for new audiences. Editions create new texts, for instance by emending the work using the techniques of textual criticism or by modernising the spelling and punctuation. Editions also result in new bibliographical forms. In the era of print, the scholarly edition is a special kind of book; its bulk, uniformity, and paratexts give the edition a certain cachet; but also, as a book using contemporary mise-en-page and typography, the modern edition stands visually apart from, say, its early modern ancestor. Consequently, the edition offers a different textual and readerly experience. (This point becomes abundantly clear when we consider the digital edition.) Finally, the edition offers its reader a new discursive form that did not hitherto exist. The edition’s paratexts – introduction, history of the text, textual commentary, and annotations – provide a discursive frame that was not part of the historical iterations of the work. At the same time, the edition’s very core – the critically established text and the textual apparatus – furnishes an ‘argument’ about the text: the instantiation of its own making.

If the gap between the work’s ‘pastness’ and the present needs to be bridged, so does the gap between the original and the new text. First, we need to interpret what the text says in the first place. Printed texts are usually not very problematic in that respect. Handwritten documents are a different matter altogether. The heavy use of abbreviations – not all standard, not all unambiguous – in mediaeval manuscripts, the use of special scripts like Secretary hand in early modern documents, or the heavy cancellations and revisions in modern manuscripts are all impediments. Second, text cannot exist without a medium: clay inscriptions, paper and ink, digital bytes on a computer disk. The remediation of the text to its new form requires decisions on the part of the editor. These remedial acts, therefore, are more than a coming in between (Eggert 2019, 82); they change the nature of the document by ‘flattening’ its scriptural dynamic, even if this increases legibility.

The idea of the scholarly edition that mediates between past works and present readers is the corollary of the linguistic turn in other historicist disciplines. Like other historic objects, the meaning and significance of textual artefacts are not self-evident. The scholarly editor’s task is to make sense of these artefacts and find the best way to represent them. ‘Best’ is of course a relative term. No edition can ever really be definitive. It is only as good as it can be, until such time new information or a new perspective makes re-editing the work desirable (Shillingsburg 2006b, 76, 105).

The third domain – that of the cultural and historical context – is everything else in the edition that helps the reader gain a better appreciation of the specific time and place in which the work was made. Much of this appears in the introduction, in particular the work’s reception and publication history. The first provides an account of how original audiences read and reacted to the work. The second deals with aspects and circumstances of the work’s production, i.e. how it reached its early audiences from its early conception and composition to its printing and publication. These activities provide the biographical and historical context for the textual condition. Another context is the work’s cultural fabric: the world that the author puts into his work through references, allusions, or mentions of actual events, places, people, and ideas. Annotations and explanatory notes (which, after the text, is the most used component of any scholarly edition) provide detailed elucidations of these references.

The above components are discursive rather than textual. Consequently, they frame the edition more directly than the textual and documentary features of the edition. Explanatory notes in particular are therefore sometimes seen as intruding on the reading experience. While common sense dictates that annotations should be informative, not interpretive, hermeneutically speaking, it is difficult to draw a firm boundary around annotations that clarify without closing down interpretive possibilities (Small 1991; Van Mierlo 2020). In fact, their virtue lies precisely in framing the cultural and historical dimensions of the work.

Curiously, we expect Shakespeare and Aphra Behn to be fully annotated, because their world is very different from ours. Not everyone believes the work of Eliot or Joyce should too. Whether the encyclopaedism of modernist writers is an imposition or an opportunity is a matter for debate. Should readers relish the abstruseness as part of the modernist aesthetic? That question is not one for the scholarly editor to answer. As a work of scholarship, the duty of the scholarly edition is towards the literary work from the past: ‘Offering readers a historical text without also offering the fruits of scholarship into the contexts of origination is not good enough’ (Shillingsburg 2010, 179). Working as a textual archaeologist, the editor should engage with all aspects of the history of textual production whose recovery and presentation is an essential critical activity.

Nonetheless, the agency of the reader remains in all this an unresolved issue. Who is the edition for? What place (if any) does the reader have in the edition? Because scholarly editing is still traditionally inclined towards old ideals of objectivity, neutrality, and authority propagated by the likes of Bowers and Tanselle, the reader has not been overly present in editorial debates. However, scholarly editing in a digital environment is raising all kinds of new, fascinating questions not just about technological prospects and possibilities but also the epistemologies of the edition itself and its impact on users, usage, identity, and representation. How technology is reshaping the scholarly edition and redefining our relationship with it is the subject of the next and final section.

4 Interfacing and Interacting

The advent of digital technologies has created a disruptive force in scholarly editing. Their impact on theory and practice is noticeable in almost every aspect. Digital technologies pushed scholarly editing more firmly in the direction of documentary editions. The digital edition’s greater storage capacity and screen interfacing made it highly suited to presenting digital surrogates of original documents in full. At the same time, the capability to link together ‘historical moments’ in the history of texts and works enabled new critical perspectives (Tanselle 1996, 54). The emergence of the digital edition in the 1990s also challenged editorial orthodoxies. Traditional models and paradigms now found their rival in radically new forms of editorial representation that also effected a shift in the discourses surrounding scholarly editing. The notions of authority and authorisation, representation and mediation discussed in the previous section are all, to a greater or lesser extent, a consequence of the intellectual shifts brought about by the confrontation with digital technology. The exigencies of technology necessitated a back-to-basics exploration of the fundamentals of scholarly editing. Suddenly it was necessary to define even basic concepts such as ‘text’ so that a computer could understand it (Eggert 2010, 185–88; Pierazzo 2015, 37–64). Gone was an intuitive grasping of every aspect that the scholarly editor works with, resulting in a new theoretical awareness about the nature of text and editing.

Changes in thinking were in many ways surprising and unexpected. At first the use of computers was entirely pragmatic. They made the processing and collating of large numbers of textual variants much easier and freed the edition from the constraints of the printed page. One immediate outcome of this liberation was the creation of digital archives rather than editions. As the critical apparatus gave way to more dynamic ways of capturing the textual condition, readers were offered easy access to the textual evidence, enabling them to make their own decisions about which variants were errors and which were not. The outcome was that both the editor’s and the reader’s position vis-à-vis the edition was put into relief.

Let’s first look at the editor, whose traditionally central role changed when the process of editing digitally created new tasks and workflows. Producing a digital scholarly edition is very much a collaborative undertaking that requires expertise in textual scholarship, computer programming and encoding, digital infrastructure, and interface design. One unintentional effect of this new way of working is that the editor has become more visible within the editing process. Of course, the editor never really occupied a neutral position within the edition, yet, as I discussed in the previous section, discourses of twentieth-century editing were largely blind to the editor as mediator. By comparison, the subject position of the editor is more clearly articulated in the digital edition because of the increase in their level of involvement in all aspects of planning and execution. In the print era, the editor had so to speak more off-the-peg choices. The format and presentation of the printed edition, which in any case were a concern of the publisher, had been remarkably stable (Gabler 2018, 315–62). Editorial rationales were standardised and editors only needed to adapt these rationales to the specific textual condition of the work they were editing. The digital edition does not offer the benefit of uniformity. As a result, the digital editor not only needs to make more decisions as to how the edition presents its argument, the very process of decision-making itself needs to be inscribed in the edition’s data model. (The data model is the structured representation that defines how information is organised, stored, and related within a digital environment.)

If the editor is granted greater agency, and therefore stands more openly between the author and the work, this alters the editor’s – and the edition’s – relationship with the work’s audience. In a sense, the printed edition took the reader for granted. As a product of scholarship, the scholarly edition served the scholarly community in the first instance. But there was also the expectation that the edition serve a general readership. These expectations were (and still are) in line with the commercial expectations of the publishers of scholarly editions who want to sell their product to the widest possible readership. Sometimes this included issuing the critically established text without apparatus for the student market.

The belief, however, that the edition’s scholarly rigour and quality naturally exists for the benefit of the reader rests on formalist conceptions of text and work (and the coalescence of the two). As a self-contained, immanent entity, the work is the object of the reader-interpreter in literary criticism. This is commonly understood. Less well understood is that the ‘reader’ is no more than a construct of the literary-critical act: an implied reader rather a historical being. The way text and context are segregated in the traditional ‘eclectic’ edition in particular, privileging the first over the second, leaves the impression that scholarly editions simply exist to be read. This is not the case. The critical apparatus is an integral part of the edition; by tabulating the variants between the texts of the work, it relays the history of textual transmission, reminding us that any text is not coterminous with the work. Serious critical engagement with the work should not bypass the history of its own making, meaning that in reading the apparatus has to be brought to bear on the text. By further deconstructing the text-context binary, the digital scholarly edition invites a different kind of use: a form of ‘relational reading’ that is akin to consulting a reference work or a database. In essence, this describes a digital edition: it is a relational database. Digital scholarly editions are, in other words, complex resources designed mostly for specialist users.

As I review the changes brought about by the digital turn, my aim in this section is to analyse its effects on scholarly editing. This analysis centres once again on how the digital edition’s new affordances mediate the text and the work. Among the various topics to consider is the move towards the new documentary editing and the inclusion of authentic primary and archival materials in the edition that has become a predominant feature. Related to this is a discussion of the changes in form and functionality that gave digital editions the edge over print. Where printed editions were characterised by a degree of uniformity, digital editions show a greater level of variety in the way they operate, which affects their use and usability. Although there are agreed standards and protocols on the technological side, there is as yet no emerging or accepted model where form and functionality are concerned. No doubt this is a consequence of the medium itself, which ought to be seen as an opportunity rather than a shortcoming (Vanhoutte 2010, 120). Nonetheless, from the user’s point of view this disparity in form does pose several challenges for which, ultimately, we need to define an ethics of editorial mediation, especially around inclusive access.

Without doubt, the development of the World Wide Web has made access to scholarly editions more convenient for scholars worldwide. Yet the potential for editions to reach, or indeed originate in, the Global South is of enormous importance. The Bichitra: Online Tagore Variorum, created at Jadavpur University, is a good example.²⁴ Because the revolution in digital editing was led by higher education institutions, not academic publishers, the majority of digital editions and archives are available for free. By contrast, usability has been less of a priority for developers and creators. While poor interface design is a nuisance, the lack of consideration for users with disabilities can form a real impediment. Whether digital interfaces meet inclusivity criteria is a question that textual scholars are only just beginning to ask (Martinez et al. 2019).

Aside from access and usability, the ethics of mediation also extend to questions of representation. Since those deemed deserving of scholarly editions tended to be ‘dead white men’, editing has been imbricated in the processes of canonisation. Owing to its democratising ‘do-it-yourself’ approach, digital scholarship has again created opportunities for ‘recovery’ projects that have potential to redress prior ethnocentrism. Especially in North America, the Black Lives Matter movement has been a catalyst in drawing attention to the need for greater diversity in scholarly editing.

As I go on to take up these points in more detail, the argument about mediation in this section shifts to one about the presence of editors, authors, and users. While scholarly editing still subscribes to the objectivist-historicist aims of before – as it must, in order to be deemed scholarly – there is also a new urgency to understand the editor’s subject-position within the critical enterprise. From an intersectional perspective, the editor can no longer be invisible. Within a poetics of editing, editorial agency is a crucial component.

Looking Back at Looking Forward

Like anything else in the digital sphere, digital editing was captivated from the start by visions of the future. Taking my cue from Edward Bellamy’s famous utopian novel, Looking Backward (1888), I look back in this section at how scholarly editors from the 1990s looked towards that future. They imagined what capabilities the new technologies could herald as much as they talked about what was technically possible at the time. Within these discussions lies the history of how technology shaped existing editorial models and paradigms into new ways of representing the texts, works, and authors from the past.

Scholarly editors first ventured into ‘electronic editing’ (as it was at first called) to take some of the drudgery out of the editorial process. Collating variants was a laborious, time-consuming task involving a manual checking of every word, comma, and line break across multiple editions. What if a computer could be programmed to automate this? This led to the development of systems such TUSTEP and COLLATE that not only allowed for the speeding up of collation and the processing of a greater number of witnesses but also eliminated human error (Robinson 1994). For an editor like Peter Robinson, confronted by The Canterbury Tales’ 84 manuscripts, this was a major step forward. The other advantage was that collations could be performed ‘on the fly’ within the edition itself.²⁵

The first electronic editions were delivered on CD-Rom: The Wife of Bath’s Tale on CD-Rom, edited by Peter Robinson (1996), and The Piers Plowman Electronic Archive, edited by Robert Adams (2000). However, obsolescence and the rapid expansion of the World Wide Web during the 2000s as a way of delivering content to a multitude of users introduced an irreversible modal shift. From now on a digital edition was ineluctably a web-based edition. It quickly became clear that the capabilities of the World Wide Web and the ease of access were far greater than any other format.

Initially, the web had its limitations. HTML, the markup language that web browsers use, was still crude: it was suitable for rendering text and images, but the scripts needed for dynamic interactivity were not yet available. After 2006, and the full deployment of Web 2.0, these functionalities were radically upgraded. JavaScript, CSS, and better cross-browser compatibility led to a better user experience and graphic user interfaces that were better suited for rendering the complexities of the textual condition of texts, works, and documents (Dumont 2018, 111). Within scholarly editing, improved technical standards and the introduction of TEI/xml, the work of the Text Encoding Initiative (TEI), were a game-changer in terms of the modelling and development of editions.²⁶

Remarkably, practically all digital editions – then as now – are grassroots intellectual projects. The digital scholarly edition was born in institutions of higher learning. Some editions saw the involvement of academic presses. Robinson’s Canterbury Tales project, for instance, was published by Cambridge University Press (CUP). But after two CD-ROMs, CUP bowed out.²⁷ The production and distribution of digital outputs did not fit easily with publishers’ workflows (see Ohge 2021, 110–14). This do-it-yourself ethos had important practical consequences in the drive towards innovation, collaboration, and the sharing of best practice. But together with rapid changes in technology, it also created a lack of uniformity across editorial projects. While the TEI provided a standard for encoding the edition, this standard exists ‘under the hood’, while graphic user interfaces differ from edition to edition.

In the run-up to the new millennium, digital editing was the new promised land. Breaking the mould of the printed codex, the digital editor could boldly go where no editor had gone before. In the spirit of the times, digital editing was steeped in utopian language. The concept of hypertext and hypermedia occupied a key position in these discussions. As McGann put it, ‘the “open” structure of digital hypertexts’ stood in stark relief to the ‘systems of textual closure built into most traditional paper-based scholarly editions’ (2001, 25). The very idea of hypertext resists closure in the sense that any external boundaries are always artificial and easily breached; the hypertext invites expansion. Furthermore, the data – the texts within the hypertext – are arranged in a non-hierarchical manner. This means that hypertext exists without a centre, enabling an unlimited expansion of the nodes within its system (McGann 2001, 71).

Hypertext’s utopian vision, however, eventually came under attack for being no more than a theoretical possibility. Practically, hypertext is impossible to build and precarious to use. True hypertextual openness is well-nigh unachievable, for the reality is that the hyperlinks that facilitate non-linear reading require a great deal of planning and effort; the fact that they are manually coded, resulting in an extraordinary amount of ‘editorial intervention’ (McGarry 2018, 77), runs counter to hypertextual freedom. By the same token, too much openness is not conducive to a good reading experience. Without structure and guidance, readers are bound to lose purpose in navigating the decentred hypertext (Vandendorpe 2009, 111, 131).

Spurred on by technological innovation, visions of the future sometimes moved faster than what technology could offer. In his Foreword to McGann’s A Critique of Modern Textual Criticism, D. C. Greetham was rightly sceptical as to whether ‘electronic hypertext or hologram’ would ever be able to convey the materiality of the book (1983, xix). In 2001, McGann admitted that hypermedia had not really moved ‘beyond elementary connecting, sorting, and gathering operations’; although hypertext had brought a new dynamic (in the double sense of the word) to the scholarly edition, a ‘fully networked hypermedia archive’ remained the ‘optimal’, but as yet unrealised, ‘goal’ (2001, 16, 58). What was – and often still is – missing was the ability for the user to ‘order, filter and arrange’ the textual information (Robinson 2010, 160).

It is no surprise, then, that utopian visions of hypertext live on when new technologies emerge. Robinson (2007) has advocated for the use of Linked Data to connect information in the edition seamlessly and automatically with information in other data repositories. Shillingsburg has similarly proposed creating digital ‘knowledge sites’ that function like a library carousel in which a reader keeps to hand all relevant scholarship (2006b, 88, 100–102). Peter Boot and Joris van Zundert have conceptualised a multimodal edition that integrates video and sound files, geo-spatial data and maps, or data worksheets within the edition. In a prototype for their genetic edition of ‘Funerall Ode II’ by the Greek Romantic poet Dionysos Solomòs, Eleni Petridou and Katerina Tiktopoulou (2022) have included an audio guide that offers commentary about the writing process linked to specific places in the manuscript.

The networked edition, meanwhile, is part of a wider idea advocating for interoperability through distributed scholarly editions that are open and decentralised. Existing outside of the silos of individual projects, these editions would provide digital services through sharing software, functionality, and other digital components as and when they are needed (Boot and Van Zundert 2011, 141–45). Following on from this, Christopher Ohge looks at editions as data rather than edited text; he advocates that editions should become better suited to text mining and other computational tasks, whether by building new tools within an edition or opening up the data to external applications (Ohge 2021, 60–71). Finally, I too recently committed to an editorial future for editions that provide users with greater ownership through a curatorial functionality that enables them to manipulate, extract, and freely arrange the textual information in new configurations (Van Mierlo 2022).

Although few of these ideas of what the edition should be have come to fruition, the concepts have without doubt driven innovation.²⁸ The state-of-the-art digital editions and archives of today have achieved remarkable results in the way they have produced a richer, more dynamic user experience. The Beckett Digital Manuscripts Project is the best in the business in terms of its integration of the documents from Beckett’s archive, and the marginalia from his personal library, with an extensive, dynamic representation of the writing process. With innovation plainly in view, we need to ask what benefits the digital edition affords. This question is best considered from the point of view of the user.

The User Perspective

In an early comment from 1995, Tanselle admired the digital edition’s potential. He saw that, in essence, the traditional apparatus with its list of variants was no different from the dynamic versioning that digital editions perform, but the latter freed the reader from the ‘logistical burden’ of unpicking the hyper-condensed lemma in the apparatus and mentally reconstructing the textual states that lie underneath (1995, 591). Aside from convenience, he also recognised that the digital edition could deliver much more, not just for the primary materials in toto but in presenting several critical editions in one. If editing ‘brings a documentary text more in line with what someone intended at some point in the past’, the digital edition ‘can present multiple critical texts, each one attempting to show what was intended by one or more persons at a particular stage in the work’s history’ (Tanselle 1995, 592). In effect, Tanselle foresaw the ‘generous’ or ‘paradigmatic’ edition in which different editions are present in potentio in the source encoding, each of which the user can activate according to their needs (Pierazzo 2015, 29).

McGann, by contrast, realised that the dialectical relation between the text and the physical book or manuscript was more immediately discernible through digital images (2001, 63–68). In The Rossetti Archive (2000–2007), he aimed to exploit this possibility, envisioning not just choice for users but active participation. Instead of serving up the editorial work in the edition, the digital archive presented the raw, primary materials directly to the user who was to make sense of these materials by undertaking the text-critical analysis themselves.

This liberation from the auspices of the editor was a radical departure from the print paradigm. Facilitating active participation changed the edition’s entire relationship with authority and authorisation. First of all, the digital archive put into action the precepts of social editing. The archive no longer privileged a single copy-text but considered all texts as historical manifestations of the work, thus placing them on an equal footing with each other, regardless of how well they represented the author’s wishes. At the same time, the digital archive was able to do better justice to the bibliographical codes of the books in which these texts had appeared, even if it captured those codes in an intangible, two-dimensional form. The documentary perspective was the prime motivation. But because the archive was offered in its raw form, as it were, readers were expected to do their own textual and editorial work. Not everyone agreed with this reasoning, however, because the critical skills and expertise needed for this type of engagement, not to mention the time to negotiate the archival and bibliographical record, are obstacles that stand in the way of ordinary users (Shillingsburg 2006b, 82–83; Sutherland 2009, 22–23).

Despite the push-back against the idea that every reader can be their own editor, digital editions were conceived as something that could empower the user. An important driver behind essentially documentary projects like The Rossetti Archive was ease of access. Books and documents that are normally available only in the reading rooms of university and state libraries can be accessed by a click of the mouse. Archives of high cultural importance that are dispersed across institutions – like Jane Austen’s manuscripts in the Jane Austen Fiction Manuscripts – are brought together within a single digital environment.²⁹ But digital editions, regardless of form or model, were always about much more than providing easy access to texts, works, and documents; they also offered new possibilities in gathering, sorting, and connecting material, and in user interaction.

In one sense, this user interaction was not new. In his early comments on digital editions, Tanselle observed that the user had always been a participant: ‘serious reading has never been limited to straightforward linear movement’ (1995, 591). The user’s role was to make sense of the connections suggested through the apparatus. But where the printed edition largely envisaged a passive participant, its digital counterpart openly facilitates interaction. It does this by building this interaction into its design and thereby inviting new perspectives that were difficult to detect before. Using the concept of the paradigmatic edition, for instance, an edition that is oriented towards the process of the work’s coming into being can contain two different modes: a documentary presentation of the manuscripts and a genetic presentation of the manuscripts’ fourth dimension, time.³⁰ In the latter case, it can further present the growth of the text in the order of inscription by indicating the sequence in which the author wrote the words onto the page or the order in which the text is to be read (Pierazzo 2009).³¹

The true power of the digital scholarly edition, therefore, lies not only in its ability to represent the rich and complex information from the textual edition but also in the extent to which the user can explore these data networks. As McGann articulated it in respect of hypertext, ‘[o]ne is encouraged not so much to find as to make order – and then to make it again and again, as established orderings expose their limits’ (McGann 2001, 71). In the era of print, McGann gave the name of ‘radial reading’ for this type of participatory reading ‘in which the activity of reading … transcends its own ocular physical bases’ (1991, 116); in the digital medium, it is referred to as relational reading (Gabler 2018, 322–23; see also McGarry 2018, 64–76). The antithesis of linear close reading, relational reading is a dynamic and iterative process that explores the symbiotic connections between the text, the text’s textual history, and the text’s cultural genetics: the biographical, bibliographical, historical, and cultural contexts that contributed to the making of the text. The reader, one could say, is less a reader than a consulter; the edition mimics more closely the function of a reference work or database than a book.

To enable this kind of relational reading, the design of ‘ergodic’ editions is needed whose interface offers an ‘aggregate of means by which the user can interact with the text, commentary, and ancillary material’ (Vanhoutte 2010, 120). Theoretically, this is a very attractive prospect, not least because it better draws together the textual and discursive components of the edition and prompts us to consider the inclusion of those non-textual elements in our data models, such as meaning and intention (Ohge 2021, 25), the user (Van Mierlo 2022), and all material, ‘architextual’ components (Treharne 2009). In practice, however, an edition that is open to active user content creation, or that permits users to craft their own path through the data, is difficult to realise. To build these kinds of dynamic systems requires resources that are not always available. User engagement remains as a result an area where there is still a lot to do.

The Post-Editorial Perspective

User engagement and usability go hand in hand. But functionality and user-friendliness are only one side of the coin when it comes to access. The question of who can access the digital scholarly edition is equally crucial as the question who is in the edition. These are questions that lead us to issues around representation, identity, and inclusivity which go beyond the mechanics of dealing with copy-texts, versioning, and textual fluidities, but which still manifestly belong to the epistemologies of editing.

The fact that digital editions can be accessed from anywhere in the world was seen as an important benefit. But so much more is at stake than the advantages of remote access. In a recent study, for instance, Martinez et al. (2019) found that few editions are designed for inclusivity and diversity. Users with disabilities are rarely, if at all, considered in the development of digital editions. About two-thirds of respondents surveyed indicated they had made ‘efforts’ to make their projects compliant or partially compliant with web accessibility standards; the remaining third was not at all aware of these standards (Martinez et al. 2019, 27). In addition to a lack of awareness, the authors found that a shortage of resources and training; deficiencies in the planning for accessible design, especially during the early stages of a project; and disregard for the fact that inclusive design benefits all users, not just those with disabilities, lay at the heart of the systemic failures in catering for user needs.

Likewise, the study identified issues with a notable lack of diversity. The vast majority of scholarly editions in the English-speaking world are devoted to the work of white, male authors. Women are no longer as underrepresented as they used to be. The early modern period in particular has done well in editing the work of female authors, among others through The Perdita Manuscripts project, which recovers the work of writers that survives only in manuscript form.³² But for other periods there is still catching up to be done. Likewise, there is real dearth of in the editing of texts, works, and authors from the Global South and other marginalised voices (Martinez et al. 2019, 34).

The same observations apply to editors’ identities, who, until 1990s, were predominantly white and male (Thompson 1997, 85–86). This dominance has had a far-reaching impact on the theory of scholarly editing. Kate Ozment sees the lack of diversity in the field as concomitant with discourses (around textual authority and historicity) and practices (applied to white male authors) that reflected the values of ‘a certain kind of scholar and a certain kind of labor’ that normalised editorial and bibliographical work as ‘genderless and raceless’ (2020, 166, 161; see also Bennett 1993).

As noted at the beginning of this section, the transition to digital editing encouraged a new awareness of the editor’s presence in the edition, which in turn sparked interest in questions about identity, gender, and race. Darwin Online, for instance, is an amalgam of documentary and textual material of Darwin’s work, including among others a variorum edition of On the Origin of Species (edited by Barbara Bordalejo). Crucially, the project also contains the letters and diaries of Emma Darwin, the biologist’s wife, which expands the edition beyond the textual into the social sphere (Browne 2007). Likewise, the Cambridge Edition of the Works of Aphra Behn, under the general editorship of Elaine Hobby, connects Behn’s texts with the social conditions of the theatre and the book trade (Behn 2021, xv, xvii).

Where the edition makes its biggest intervention, however, is in the domain of attribution. Her play The Younger Brother was attributed posthumously to Behn by Charles Gildon, a literary hack who sought to profit from her reputation when he staged and then printed the play in 1696. How Gildon came in the possession of Behn’s play is not known. But he admitted to making alterations to the text, raising questions as to how much the play, which Behn left unfinished, is in fact Gildon’s. Using stylometrics, the Cambridge edition establishes with a high degree of certainty that most, but not all, of the writing is Behn’s (Behn 2021, 714–19; see also Hogarth and Evans 2020) and thus makes an important contribution to redefining the canon of one of the most significant female authors from the seventeenth century.

These and other similar projects, part of a new gender-aware approach to scholarly editing, are concerned with female authorship rather than with texts or works that are subjected to new editorial method. Indeed, the difference lies in their treatment of what lies beyond the purely textual to the contextual.³³ In her review of R. W. Franklin’s variorum edition of The Poems of Emily Dickinson, Marta Werner makes this very point. Given the nature of Dickinson’s writing in the so-called fascicles, her poems are both text and performance. By segregating the text from its material support, the poems lose their creative performance and therefore their ‘full intentionality’. This is because Dickinson’s work only exists within, not separate from, its archive (Werner 1999, 260; see also Werner 2011). Although Werner is sympathetic to the potential of hypertext, the dynamic elements in Dickinson’s work are such that no edition – not even a digital edition – can adequately capture them: the work is too ‘unruly’, its data ‘unclassifiable’ (1999, 263). This ineffability of the work is not in itself intrinsically gendered but it is the feminist awareness and critique of closed, binary, and hierarchical structures within the work that should make the editor conscious of its ‘openness to the unforeseen’ (Werner 2006, 298).

Werner’s call for an intellectual and editorial suppleness is a call for editorial inclusiveness. The beauty of being open to the unforeseen is not so much that we reimagine the scholarly edition but that we reimagine editorial practice and the way we mediate our texts and works from the past for the present. What is needed, especially when we think about decolonising scholarly editing, is a new form of perspectival thinking. When it comes to the question of how we edit women, black, or queer writers, the new representations that we seek can exist in new editorial forms. But some of the necessary reframing can also be achieved in the edition’s discursive elements – the introduction, commentary, and annotation (Andrews 1997, 51; Thompson 1997, 90–101).

This is where the new documentary editing comes into play. Unlike literary editing, documentary editing is more about preserving, reclaiming, and rehabilitating texts and authors not otherwise accessible or available. New documentary editing is different, first, because it puts specific emphasis on the ‘thick description’ of historical and cultural contexts, which makes it particularly suitable for editing texts and authors at the fringes of literary history. Second, new documentary editing favours authorial networks over the single author, allowing the focus to fall on networks of textual production in which there is no discernible authorial centre. The motivation for this networked approach sits somewhere in between the microfilm projects of a bygone age and big-data projects such as Mapping the Republic of Letters that unlocks the political, social, and creative networks in the European Renaissance and Enlightenment.³⁴ Because it includes data about genre and social status, the project brings to the fore voices that can otherwise remain easily hidden. Third, the new documentary editing also shifts attention to the editing and inclusion of non-narrative and non-linear sources, such as account books, marginalia, or personal libraries (see Van Hulle 2016, 37–38; Scheltjens 2022).

An excellent recent edition that combines a rich layering of historical and cultural components and the network is The Letters of Charles Harpur and His Circle, edited by Paul Eggert and Chris Vening (2023). Harpur (1813–1868) was a poet, journalist, and farmer from New South Wales who lived and wrote during Australia’s colonial period. His work, though not wholly unfamiliar, is not yet widely studied and understood, not least the role that its romantic tenets played in bringing about a new Republican spirit desirous of breaking ties with England. The edition, however, does not sidestep the vexed issues in Harpur’s work and those around him. The colonial spirit that pervades his writings ‘conditioned a sense of newness and awe’ for the Australian ‘landscapes and fauna’ that all but hid a darker side in the creation of a national identity: the repression and displacement of the Aboriginal peoples and anxieties about racial degeneration (Eggert and Vening 2023, xix). The other area in which this edition is analytically innovative is in making manifest Harpur’s participation in Australia’s emergent print culture. Its day-to-day features, familiar to contemporaries but unknown for modern readers, are meticulously captured in the explanatory notes. The edition’s e-book version contains working URLs that link directly to historical newspapers and magazines in Trove, the National Library of Australia’s digital archive of cultural collections, as well as Harpur’s manuscripts in The Charles Harpur Critical Archive.³⁵

The journal of the Association for Documentary Editing, Scholarly Editing, has been at the forefront of this new form of documentary editing, regularly publishing standalone, open-access micro-editions, privileging writing from minority authors who bear witness to important moments within the cultural heritage.³⁶ At the time of writing, the latest issue contains two such editions.

The first, ‘Marm the Doctor, Mill Children, and the American Dream in “One Way to Get an Education”: A Recovered Short Story by Elizabeth Stuart Phelps’, edited by Roxanne Harde (2023), is a short story that had remained untraced in the bibliographies of this radical feminist and children’s writer until now. The story, which appeared in May 1871 in The Youth’s Companion, a popular children’s magazine published by Nathaniel Willis and Asa Rand in Boston, relates how a young boy named Jake deliberately harms himself to escape the plight of factory work in order to go to school. In the hands of its creator, the incident is a pretext for a moral tale about the evils of child labour in the late nineteenth century. But Phelps’s story reveals out that Jake is treated by a female doctor who looks forward to the day ‘when there is a woman physician in every town’ (Harde 2023), turning this simple moral tale into an argument about gender equality.

The second micro-edition in this issue is ‘Selections from the Revue des Colonies (July 1834 and July 1835): From the Prospectus to the Bill for Immediate Abolition’, edited by Maria Beliaeva Solomon. The Revue des Colonies, founded by the abolitionist Cyrille Bissette, was France’s first magazine for black readers that campaigned for immediate abolition, while exposing incidents of colonial abuse and the influence slaveowners enjoyed on government (Beliaeva Solomon 2023). The selection of articles offers an important snapshot of the kinds of interventions and critiques made by people of colour of the French colonial system and its support of slavery. Both editions are richly annotated and are accompanied by relatively short introductions that offer historical and scholarly context; they also offer the reader full access to the TEI source files.

The value of these editions lies not in their technological innovation but in the way familiar technologies can harness ‘generative modes of analysis’ (Beliaeva Solomon 2023), sparking new questions and avenues of exploration. As such we have come full circle in that the digital scholarly edition brings us back to the type of edition imagined by W. W. Greg and his contemporaries. To be sure, the generative edition fits very well with the post-editorial paradigm in that they assume a new ‘conventional’ form in their application of tried and tested standards and practices in scholarly digital editing.

This seems like a good point on which to draw this section, and this Element as a whole, to a conclusion. The disruptive force that the entry of the digital brought about in scholarly editing has surely come to pass. Are we as digital editors too obsessed with imagining the digital edition of the future while overlooking the opportunities and standards that have already been created?

The new documentary editing, which has a sure eye on inclusivity and diversity, suggests as much. So too does the lesson we have learned from the digital humanities about creative practice being a valid critical method of exploration. As Mathelinda Nabugodi and Christopher Ohge remind us, scholarly editing is not just about producing reliable texts; it is a critical process: ‘To approach a text as an editor means to question it in a different way than when the same text is approached by a critic.’ In recognising this, they advocate the pursuit of ‘an editorial practice that does justice to an aesthetic, experimental, and embodied notion of literature experience’ (Nabugodi and Ohge 2022, 3–4). Although Nabugodi and Ohge are not exclusively thinking about the digital edition, this is nonetheless where the future is headed. Editing entails a critically and creative exploration of how best to convey the textual condition of our historical texts, works, and authors with the aim of making scholarly editions that function as the incubators of knowledge. We need to heed the call to think imaginatively about the forms or our editions so that they can become genuinely inclusive in every sense of the word.

5 An End Perspective

This Element set out to put scholarly editing in perspective in the English-speaking world. Its primary aim was to show what scholarly editing does to the literary work and how the scholarly edition captures the ‘textual condition’, the historical and material circumstances that together characterise the ‘making’ and transmission of a work in manuscript and print. Thus, the book articulates an argument about the epistemologies of scholarly editing: not only what editing is but also what it does. The focus on editing as an activity, which aligns with the poetics of editing proposed by Susan Greenberg (2018), situates the arguments somewhere in between editorial theory and practice.

In Section 1, ‘Models and Paradigms’, I disentangled the various models and paradigms that have dominated the field since the middle of the twentieth century. The two main models stem from the disciplinary boundaries between English Studies and History: literary (or critical) editing and documentary editing. It was primarily within the first that different paradigms emerged. Charting as far as possible the chronological development of these schools of thought, I explained how and why paradigms became paradigm shifts. The outcome of these shifts was that over time, literary and documentary approaches moved closer together, both in practice and in a new shared awareness of each other’s philosophy of editing. This rapprochement was effected not least as a result of the evolution towards ‘the digital’. What computer technologies enabled, among other things, was a better way of representing documentary features alongside the ‘text’ and the ‘work’.

In broad terms, the difference between literary and documentary editing was one of the critical establishment of texts versus the representation of the material form of books and manuscripts. While documentary editing grew in critical sophistication in the way it saw the task of representing these materials forms, its theoretical and methodological approaches did not show as much as variation and divergence as literary editing. By comparison, literary editing underwent at least three major paradigm shifts. Often still considered the ‘dominant’ paradigm, the eclectic editing approach informed by the work of Greg, Bowers, and Tanselle proposed a method of selecting the best readings from among the available textual variants that would correspond to the author’s intentions. The method hinges on the selection of a ‘copy-text’, a text that is on bibliographical grounds the best text as a control text. The copy-text may not be perfect in its own right, but because it stands closest to the author, it can serve as the arbiter when all other considerations fail to provide the correct reading. This happens often with spelling, punctuation, and so forth – the so-called ‘accidentals’ – whose authorial origin, unlike that of ‘substantives’, is difficult to verify.

Eclectic editing was developed for works from the early modern period, during which accidentals and other matters relating to the printing were particularly unstable. Notoriously, Shakespeare did not have a hand in seeing his plays through the press. While eclectic editing was successfully applied to authors from later periods, the method nonetheless showed some limitations. Not all works easily fitted a model where one text was clearly superior to another or where variants could not be readily coalesced into a single text. The approach proved difficult with authors who substantially and repeatedly revised their own work, creating radically different versions that nonetheless had equal authority to one another. The best possible way of representing these works was to give each version equal representation in a parallel text.

By recognising the distinctiveness of each text, this second paradigm, often referred to as ‘versioning’, also drew attention to the specific material and bibliographical conditions in which the texts were produced. The third model in the evolution of scholarly editing in the twentieth century, finally, shifted the focus to the precise conditions of production and the book as medium that gave the text and its meanings a distinct shape. In the sociological school of editing, both McGann and McKenzie turned to the ‘expressive function’ (McKenzie 1999, 17) which they saw as part and parcel of the work’s bibliographical identity. Because each textual manifestation of the work has its own history and reception, editors had better capture these material forms in their editions, something that McGann argued could only be done in the digital medium.

The second section, ‘Mediation and Authority’, delved into the nuanced effects and consequences of evolving editorial models and paradigms. It highlighted, among others, a shift towards acknowledging textual instability not as a flaw to be eradicated but as a characteristic to be celebrated. If scholarly editing is about making editions that are reliable and authoritative, how exactly do they achieve this? The answer is predicated on three essential principles that give editing rigour: respecting textual evidence, transparency in the editorial process, and applying a coherent editorial rationale. It also involved acknowledging the inherently ‘critical’ nature of editing that relies on judgement and decision-making. Characterised by ambiguity and incompleteness, textual evidence is by its very nature fluid and in need of interpretation. At the same time, the process of establishing and re-presenting the text – whether this is removing accidents of the printing press or faithfully capturing the text from a document – cannot be done without an approach, a perspective, and a plan. Furthermore, making texts from the past ‘present’ again involves mediating between various levels of authority: the author’s agency, the textual, bibliographical, or archival evidence, and the editor’s own interpretive stance which derives from the editorial rationale. With this new perspective in mind, it is also important to understand how the editor’s role in scholarly editing transformed from an invisible, ostensibly neutral intermediary to a more conspicuous mediator between the author, the text, and its readers. This shift was not purely conceptual but also entailed a new ethical dimension within editorial practices, largely propelled by the advent of digital editing. It was natural that the transition towards the digital prompted a reappraisal of fundamental concepts and stages in the editorial workflow and thereby profoundly changed the field.

The final section in this Element, ‘Interfacing and Interacting’, turned to the emergence of digital scholarly editing, discussing the origins and effects of these changes. Initially, computers were used for the processing of textual data, freeing editors from some of the manual work of collation. Unrestrained by the restriction of the printed page, the medium also inspired innovation in how way editions looked and functioned. Hypertext provided a richer means of navigating content, effectively turning the edition into a relational database rather than a book. Other innovative interfaces and designs explored new ways of interacting with the digital edition’s content through paradigmatic editions and ergodic design that sought to enrich the user experience. This empowering of users in the way they could interact with the data resulted in a re-evaluation of the relationship between edition and user. Instead of a top-down approach in which the editor fixed the form and function of the edition, the bottom-up perspective allowed the user greater creative freedom in the way they could interrogate, manipulate, and repurpose the data. While digital editions continued to evolve in line with technological developments, often these innovations were conceptual only, a vision of what could be. Practical applications did not always keep pace with the speed at which ideas about the digital edition of the future were articulated. It is therefore not surprising that there was a significant gap when it came to accessibility for users with disabilities.

The new possibilities that digital technologies afforded, and the editorial reorientations that came with them, also created space for new forms of democratisation. As a field, scholarly editing was dominated by white men, in terms of both the editors who undertook editorial projects and the authors deemed worthy of scholarly treatment. This lack of diversity is now thankfully being addressed. While significant progress was made in editing works by women, this is not yet the case for writers of colour and other figures from marginalised groups. Interestingly, the projects that do exist have taken their cue from what one may call the new documentary editing. This work still focuses on the individual author but places that author in context by foregrounding social and historical connections within a networked perspective. This new orientation is eminently suited to writers whose identities are often eclipsed by the forces of canon formation.

The trajectory of this Element relates each step of the way to the poetics of editing. Before closing, I want to recap some of these connections so that we can see how scholarly editing fits within wider editorial practices. In my view, such a reorientation is important to underscore the relevance of scholarly editing for the study of literature with new pragmatism and insight. The aim is to take us out of the old binary that pits textual work against literary criticism.

Time and again, textual scholars have argued that editing is critical work and that it is therefore also a form of literary criticism. Literary scholars, by comparison, have not always reciprocated with equal awareness of editing as an undertaking. On the surface, the difference stems from the fact that all editors are also literary critics, but not all literary critics are editors. At the very core is a difference in perspective: literary criticism deals with text as a linguistic entity, whereas textual scholars and editors deal with text as a ‘material’ entity. In that respect, scholarly editing shares with all other forms of editorial enterprise the basic premise that published texts are the product of many individuals, social, cultural, and mechanical forces.

Scholarly editing bridges the gap between past and present in three domains: the text, the textual condition, and culture and history. Foremost in its process therefore is the decision-making process targeted at selecting, shaping, and linking the text to other contexts (Greenberg 2018, 14). At every step of the way, the editor faces choices about editorial selection, approach, and execution. The deep bibliographical analysis that Bowers advocated is a primary untangling of the textual, bibliographical, and archival evidence so that it can be reconstituted both as a formal presentation and an argument within the edition. This formal presentation is bound by the affordances of the edition which entail a choice, for instance, between traditional collation in the print apparatus and genetic encoding in a digital environment. The chosen editorial rationale supplies the set of principles that binds the two together, logically and reliably.

The scholarly editor’s intervention in the text’s verbal and aesthetic meaning is at once subtle and complex, for it is here that the editor must negotiate between language, aesthetics, intention, and authenticity on the one hand and the ambiguities within the textual record on the other. The editor clearly does not have a free hand. At the same time, they cannot but rely on their judgement and interpretation of the available evidence. Precisely for this reason, I have emphasised in this Element the epistemological nature of editing rather than the editor’s confrontation with the ontological or phenomenological status of the work, as other commentators have done (see Eggert 2019, 31–34).

In the scholarly edition, the editor, author, and text come together, as they do in all forms of editorial activity (Greenberg 2018, 25). The difference is that in most forms of editorial practice, the editor actively participates in creating the text before it is published, striving to make it ‘as good as it can be’ (Greenberg 2018, 19). In journalism the editor performs a variety of checks and balances to ensure that the article is accurate, consistent in tone and style, and that it meets with editorial policy and professional and ethical standards. An editor in a commercial publishing house will work with the author on successive copy-edits, moulding every aspect of a book from the minutiae of grammar to the overall structure, shape, style, and voice.

In scholarly editing, the editor too is concerned with quality control. They are no less an active agent, albeit that they have a responsibility for the work as a historical entity. From this derives their role as mediator, connecting new audiences with works from the past. The aim is not just to create a new experience; any new reprint could achieve this. What is imperative, rather, is situating this experience within a new knowledge framework. That the scholarly edition presents an argument about the textual edition is a valid but somewhat abstract point. To say instead that the editor engages in a decision- and sense-making process is far more meaningful because it brings to the fore the essence of the edition’s function: to facilitate the discovery of new knowledge and new meaning. At the same time, it highlights just how transformational the role of the editor is.

Key to this book is the assertion that the scholarly editor is not a neutral orchestrator performing a service. They are rather a direct participant in the remaking and re-presenting of the literary work or historical text. Within the poetics of editing, the editor is an ‘embodied reader’, the third operator in the triangulation between author and text (Greenberg 2018, 19). This means that in ‘practical’ forms of editing the editor is both a reader – the text’s first real reader – and an ‘author’ or co-creator who ventriloquises the original author’s voice and intentions.

For me, Greenberg’s reorientation of the editorial enterprise is acutely germane not just as a perspective on what scholarly editing is but on what it does, and should do. In this way, the scholarly edition remains an intellectually relevant, reliable, responsive, representative, and engaging tool for scholarship. As an intelligent reconstitution of the work, the scholarly edition serves a unique purpose that is distinct from other, so-called ‘uncritical’ editions. Project Gutenberg and the Oxford World’s Classics have without doubt a place in the landscape, but, being typically less self-conscious about their editorial perspective, their relationship to the work is not of the same order.

Because the scholarly edition foregrounds the work as a human textual construct, it concerns itself with the forces that went into its production, distribution, and reception as well the actors (the author, printer, publisher, and so on) that set these forces in motion. Equally, the scholarly edition is concerned with representing the work’s textual condition in a new dynamic framework to make these forces intelligible. The scholarly edition is the result of meticulous research that captures a wealth of new knowledge and insight about the work. Executed properly, it also drives forward knowledge and insight, generating new experiences and perspectives for and by its users.