Every edition published today is, on a certain level, a new witness to that text, with a status no higher than the manuscripts from which it was produced. The way in which we publish ancient texts should reflect this. Gregory Crane has written a lucid piece on the challenges of salvaging scholarly editions produced in recent decades for use with digital research methods. He reflects on efforts to secure the use of new editions for future scholarship without the inhibitions of copyright, calling the problem ‘a purely political one’. I would suggest that these complexities are so pervasive that they produce practical issues, preventing even those already convinced of the need to publish their work in an open format from accomplishing this. Crane asks, in essence, whether we are digitizing the past in the right way by focusing exclusively on the edited texts of printed editions, and wonders whether it would be worth the effort to fund the creation of open, digital versions of newer critical editions with their apparatus. This is a matter of viability: digitizing an apparatus is cost-prohibitive both in terms of the amount of time it requires and because it has a restricted copyright status. Crane notes that an edited text of a public-domain work is free of copyright after no more than thirty years in the European Union (or immediately, in countries such as Canada); but its apparatus is not, as argued by Margoni and Perry. The absence of a critical apparatus makes is difficult to determine the relationship of the edition to its source material, meaning that scholars have long distrusted digital texts. Editors have long seen its inclusion as key to making digital editions intellectually acceptable.

Everyone can likely agree that the goals of digitizing an ancient document are to preserve it, make it more accessible, and allow us to analyse it more thoroughly. But what is the best way to fulfil these over the long term? Rather than pouring more resources into the detritus of the last century, we should invest first in ensuring that our libraries have the resources to scan and conserve primary sources whose loss would be irreversible. Recent events have reminded us that artefacts in libraries and museums are not immune to disappearance and damage. If Shackleton Bailey’s edition of Horace is not fully digitized for another century, it will only be a minor inconvenience; but scholars a century from now will have nothing but condemnation for us if inscriptions and manuscripts are lost that could once have been easily photographed.

Rather than trying to replicate the printed apparatus, we should instead put our resources into creating new documentary editions of manuscripts that can be examined in parallel with modern editions. Computers can be used to show textual differences and reuse far more accurately than the typical apparatus, but they cannot explain variants, giving a purpose to the textual commentaries Crane envisions. Relatively bare edited texts – what, in other words, is most straightforward to digitize, and has few copyright restrictions – is what we most need for future research, along with photographs of our source material. Commentaries and translations, as Chris Francese points out in his own response to Crane, are likely to have the most perceptible effect on making premodern works approachable to the public. The solution to the problem of making digital texts centred around primary sources is to put long-held editorial theories into practice, and to reimagine scholarly editions as networks of texts.

Reinterpreting Texts with Documentary Editing

In a printed context, the creation of a critical edition was the first step towards making a text accessible; it was only the most important works that received a facsimile edition or published diplomatic transcription. The Internet has turned this assumption on its head, meaning that it is the critical edition that has become a luxury. In my work with medieval Latin, I often need to read a book that has never been published in a modern form – an irritating situation, but one that allows the luxury of considering how to best publish the work for the first time. In such cases, the quickest way to improve this situation is to create a documentary edition from a single manuscript that can be displayed online either in a diplomatic or normalized format. In the past, this could not be published until one was certain that the text was reasonably free from error, and it is just as likely that the edition would remain in a drawer, only to have another scholar repeat the work. But in a culture where it is acceptable to make preliminary work available, with means of correcting a text after making it available, a text can be made public even it is abandoned, and even an inaccurate edition is still far better than nothing at all. Someone else might later transcribe another manuscript for more historical context, leading in time to a better understanding of the text; only after several transcriptions were available would it then be of help to use the methods of textual criticism to produce a modern reader’s version that conjectures an ideal form of the text. If our goal is to have texts that are as accurate as possible and will allow the broadest range of future uses, this is best brought about by working with the manuscripts themselves, and enabling intelligent comparison of their texts, each of which reproduces a historical witness.

Even if we digitize every printed critical edition, we will continue to define ourselves by the short-sighted modern age; but one can only imagine what we might be able to bring to fruition in an ideal universe where from a single point we can consult every document of the premodern world. In print, editors are simply unable to provide all the information that might be useful to researchers, and typically provide a single rendition of a text with a carefully orchestrated apparatus. When dealing with editions that stem from many manuscripts, this is not a record of every known textual variant, but an abbreviated narrative to justify the particular edition presented. This model has never met the needs of historians, since the editor’s focus is almost always to account for the reconstruction of the archetype rather than explain its later reception. Nor does it always meet the needs of those only interested in this reconstruction: the laconic format of the critical apparatus is entirely defined by the constraints of the print medium, and freed of space considerations, its aim would be better fulfilled though a commentary explaining the editor’s decisions.

Where a printed version of a medieval text does not allow me to answer my research questions, my solution has been to create a new edition of a significant manuscript, giving credit to the intellect of earlier editors by citing their corrections. This model is particularly suited to the texts I happen to have worked on, which are often represented in only a handful of manuscripts, and at a short remove from the archetype; in some cases, medievalists even have the benefit of an autograph. Its application to works with hundreds of witnesses is hardly a new idea: in 1993, for instance, it was proposed as a solution for editing the complicated letter collections of Peter of Blois. The work is facilitated by having an edited text from which to start rather than transcribing a manuscript anew; similarly, M.L. West’s Textual Criticism and Editorial Technique (pp. 101–2) notes that editors in his day commonly submitted new editions to printers by modifying a copy of an older edition rather than submitting a new typescript, though he also cautions that this practice has allowed inaccuracies to survive undetected between editions. Because my research aims to understand the reading of a work in the historical context of its manuscripts, I typically use their punctuation rather than take a past editor’s version, making note of obvious errors; this is often much more interesting than what was printed. The result is something that is not only free of copyright restrictions, but is also an advancement of knowledge by any measure, as this process often reveals errors in the printed edition, and produces a text that is more historically grounded than its predecessor. The critical text is still left with something to contribute in its own right, whether that is to show a hypothesis of what the archetype might have looked like, or to provide a simplified reader’s version. Neither of these aims necessarily requires the encoding of its apparatus, if it is only one more version of the text available in parallel with its source material.

Critics may protest that this model is far too laborious to bring it about; but the old ideal of critical editing is far too ambitious to serve our needs. Higher scholarly standards have had the perverse effect of allowing less scrupulous editors of past centuries to retain a stranglehold on classical and medieval works. This is evident in the many cases for which we are entirely reliant on nineteenth-century collections such as the Patrologia Latina, most of which reprints early-modern editions made from the random manuscripts that happened to be available to their printers. It is also conspicuous in the many announced editions on which scholars laboured for decades, only for their work to be lost at their death. A published text with a documented relationship to a single good manuscript is more accurate than many of the printed editions we rely on, and just as importantly, is achievable within funding models favouring scholarship that can deliver results quickly.

Practical Barriers to Publishing Digital Editions

Crane is accurate in saying that there are no practical reasons not to publish new scholarship in an open format, but that does not mean that there are no practical barriers. This is why we look with such anticipation towards the Digital Latin Library and the Perseus Open Publications Series: more than anything, the lack of a credible publisher for open-access editions and translations of premodern texts is what drives scholars away from taking advantage of digital publication methods. Self-publication is not a serious choice for most scholars; the brave souls who have tried it (such as Jeffrey Witt, with his work on Peter Plaoul) have lacked the benefit of an editorial board overseeing a stringent peer-review process, and have had to spend a great deal of time reinventing publishing infrastructure.

The cause of digital editing also needs to justify its promotion of the Text Encoding Initiative guidelines. Assumptions in the humanities are still governed by a print mindset, where words such as ‘interoperable’ do not address the true needs of scholars. I myself was a skeptic of TEI until I saw a demonstration of a working example last year that showed how I could use it to answer my questions more quickly and accurately. This is what scholars care about. We are also, though few of us care to admit it, deeply influenced by beauty and usability. We must ensure that open-access editions are presented in a user-friendly manner, especially from a typographic perspective, and take advantage of responsive design techniques, allowing them to work equally well on a desktop computer or tablet. A few examples of this, such as Roman Inscriptions of Britain, are finally beginning to appear.

The lack of guidance provided for the use of TEI has sustained concerns about its suitability for exchanging information. It is to be hoped that the forthcoming series of digital texts will release guidelines for submission as soon as possible, given how many years editions can take to produce: it was helpful for Perseus to announce their intention to standardize on EpiDoc, which arbitrates between some of the most otiose encoding options available in TEI – such as the several different methods of indicating an abbreviation in a primary source, or showing divisions in textual structures – in a way that makes sense for working with ancient sources. Still, there are many variables to be considered, leading to many small projects that have been obliged to invent idiosyncratic encoding methods. The problem here is not primarily TEI itself, as Desmond Schmidt has suggested, but the reality that we have yet to form clear expectations of what needs to be included in online editions. Without clear direction for the use of TEI, it is difficult to have confidence that one’s encoding will be usable by future publishers without significant alteration.

We also lack enjoyable tools for working with TEI. This is largely due to the antipathy felt toward XML in many technical circles, arguing that it is too over-engineered for direct use. Anyone would be mad to write TEI without software that validates the markup as one works; but the advanced schema languages it requires (RELAX NG and Schematron) are not widely supported. The Oxygen editor is the only remotely usable product that offers this, and even it falls short of the expectations set by progressive user interfaces. Its commercial licence is a significant barrier to starting with TEI, and some users simply refuse to consider it due to its dependence on Java, whose reputation has been marred by security issues and Oracle’s bundling of adware. Web developers eager to expand their trade have sustained the creation of highly usable open-source editors are available for the languages they rely upon, such as Atom and Brackets, as well as document conversion tools such as Pandoc. If we do not think there are any workable replacements to XML as the basis for TEI, we need to advocate for its support in such software to prevent the entrenchment of a single set of tools.

It should hardly be a surprise that few people are publishing digital scholarly editions when such contributions are neither actively solicited nor supported. Academics continue to work they way they do because it seems necessary to self-publish if one is going to produce a text electronically, and they see this as the loss of an opportunity to have others examine their work and a waste of time that they could devote to research. This should change once someone with authority begins to publish digital texts, and we make it easier to work with TEI. If we can allow ourselves not to slavishly reproduce the modern era, but to learn from it, and reimagine how to make primary sources most accessible to today’s audiences by taking advantage of the best techniques available to us, scholarship of the future will be open indeed.