Let us first recall how a print-based editorial project conceived in pre-computing times was organised, since even today, such projects, often large-scale and multi-volume editions that were begun perhaps a generation or two ago, are still in operation. Project-specific editorial conventions and rules would have been developed for them. These were likely to be codified in editorial guidelines explicitly geared to traditional book production: the preparation of a comprehensive edition script as printer's copy, the reliance on publishers editors and professional book designers, machine (or even manual) typesetting, proof-reading, and conventional machine printing.

By contrast, let us imagine a print-based editorial project, begun after the introduction of the computer into the scholarly editor's study. Under these circumstances, several of the preparatory operations for an edition will be performed electronically. We can rely upon the computer to assemble and store the edition script, that is, the text we are preparing, as well as the apparatus, the notes, the commentary, the introduction, and the indexes that go with it. Electronic tools have begun to modify the sequences, and perhaps even the internal logic, of the editorial task. For example, they can be used to ascertain the edition's accuracy. Verifying the data input already constitutes the principal round of proof-reading; proof-reading at the stage between typesetting and printing, on the other hand, will no longer be required, since that stage of the work process has itself become obsolete. Altogether, repeated proof-readings of the text of an edition should be significantly reduced, if not rendered altogether superfluous. Similarly, the mark-up necessary for data input has particular consequences. On the one hand, it relates the input data to the source or sources from which they derive. On the other hand, it pre-codes the formatting of the book envisaged. Above all, the use of the computer alters the production process of an edition. The edition script will not be re-keyed in the printing-house, as used to be the case in traditional book production. In fact, the high standards of accuracy appropriate to scholarly editing demand that, once established electronically, it must not be re-keyed. But this means that, if and when the goal of the project is a printed book, its editor is now already herself required to conceptualise the appearance of the edition in terms of typography and book design. As a corollary, this can mean that she may find herself in an easier position to produce the edition she ideally imagines. Thus the dividing line between the respective domains of the scholarly editor and the professional publisher and printer as they previously existed have been fundamentally redrawn. Practically speaking, they must agree on the format in which the edition in all its parts is to leave the editor's hands. Either the edition will be submitted as electronic data, not yet transformed into print via electronic typesetting; or else the publisher will ask for ‘camera-ready copy’.

Not only the older type of pre-computer project, but also the more recent type, already computer-assisted in its preparation, must be defined as essentially print-based. To transfer the older type of print-based editorial project into electronic form, involves the primary consideration of how, given its particular conventions and rules, to make it amenable to electronic processing and thus, how to rework its editorial guidelines and practices in terms of computer aid. Transferring the more recent, already computer-assisted type of edition into electronic form amounts to divorcing it from the former goal of the printed page, and shaping, or reshaping, its electronically stored data in such a way as to enable its consultation and study via the computer screen. Given the appropriate access software, the structuring and organisation of the data of the edition should allow computer-based, question-and-answer interaction with the edition. In other words, the electronic medium, instead of merely being deployed as an aid in the preparation of a print edition, should become and be recognised and established as the proper site and natural medium for a scholarly edition.

It is useful to rehearse how a print-based editorial project might be transferred from a pre-computer environment to a computer-based editorial practice, before focusing on what might be involved in transforming a computer-assisted edition into a fully computerised edition. This is a necessary step because, before considering the electronic medium as the site most appropriate for a scholarly edition, through which it communicates and is communicated, we must account for the role of the computer as aid and tool in the making of an edition.

Designing a computer environment for an editorial project involves acquiring and co-ordinating computer hardware and software; inputting the data for the edition; structuring the editing work-flow; and securing the stability of the results of the editing.

As regards hardware, there is a wide choice of standard, commercially-available equipment. In addition to the computer unit itself, a flat screen would be an advantage, for ergonomic reasons as well as to provide comfortable resolution, e.g., to facilitate work with manuscript scans. A printer is obviously needed, and a scanner of superior capacity; a CD-ROM burner might also come in handy.

Software requirements are highly variable, and only a few basic considerations can be suggested here. For the input, these should include scanning software, preferably teachable to cope with varieties of fonts and other typographical features; and word processing software whose proprietary characteristics can either be circumvented, or which can be converted down without loss to standard ASCII, or converted up to standard mark-up norms (such as html, sgml, xml).

In order to verify the input and to carry out the subsequent editing activities, text-processing software will be required with a capacity distinctly above that of the run-of-the-mill word processor. In particular, it should be strong in the areas of collation, text formatting and re-formatting, and in administering and keeping log automatically of a reference grid (for example, reference numbering by page / [paragraph] / line / word; or act / scene / speech / line / word; or chapter / paragraph / line / word; etc.). This stable reference grid should be trackable and recoverable through all editorial formatting and reformatting stages; and at the final disposition of the editorial result into reference-linked blocks of text, notes, apparatus and introduction, every reference and reference connection should be automatically generated and distributed.

For individual editing tasks, pre-designed and custom-packaged software may be found useful and sufficient (e.g., the 'Classical Text Editor'), but compatibilities should be carefully tested. Between systems as well as within systems, the option of interactive work on the screen should always be provided. At the same time, however, an editor should never find herself confined to screen-and-keyboard, item-by-item surface progression through an edition (that is, she should not see, or be forced to use, the computer merely as a substitute for the typewriter). Every key-stroke carries the fifty-fifty chance of hit or miss. Statistically speaking, therefore, keyboard interactivity is just as prone to introduce, as it is likely to avoid error, and only half as likely to correct it (since every correcting stroke carries that fifty-fifty chance again). Consequently, text processing software will prove its value by possessing powerful batch functions to automate batch-definable editorial operations and carry them out consistently.

What these operations might be, is already part of the definition of the editorial work-flow structure. The editorial process, in so far as it provides the intellectual solution to the given editorial demands, is independent of the computer. Yet deploying the computer as the editor's work-tool is likely to modify the work-flow of that process. This, as we have said, applies to the proofing built into the verification of input. Often, automatic collation recommends itself for this purpose. The automatic collation of two parallel inputs ensures a high degree of accuracy in the input data. Ideally, such a double input should be produced by separate agents, and/or by different processes (e.g., one scanned in, the other keyed in). Where the making of an edition involves two (or more) versions, the single input of each version will result in the same verification effect: the differences revealed by automatic collation will either be the result of errors in input, requiring to be corrected; or, they will be genuine differences between the versions that will thus be accurately recorded, once any input errors have been corrected.

It is, thereafter, the collation of the verified input materials (the second collation in the work-flow) that provides the textual variation on which editorial judgement and decision-making must be carried out. The crucial gain of automatic collation is that it relieves the editor of the task of wading through the multiple records of text identity and allows her to concentrate her mind on adjudicating between textual variation. Furthermore, the use of the computer permits searches, to be built into the editorial work-flow at will, for accuracy and consistency of text assessment and editing. These are valuable auxiliary operations that were either impossible or prohibitively arduous before the computer was used in editing. And in another respect, usually involving a third round of program activation, automatic collation, together with co-ordinated formatting and referencing routines, should be capable of extracting, as well as formatting and correlating, the edition's apparatus and index sections.

The tasks of scholarly editing call for these kinds of computer assistance regardless of how the editorial results will eventually be presented; that is, whether they are made available as a book, or through the electronic medium itself. In order to transfer a print-based editorial project into electronic form, that project must first be thoroughly cast to draw upon computer assistance in the production process. The scholarly procedures for preparing an edition must be systematically designed for the computer, comprehensively applied and rigorously carried out. This needs emphasising since, in terms of application and performance, the common-sense answer to particular local difficulties and 'special-case' situations is still often, and even in front of the computer screen, the intelligent shortcut. But in terms of data-processing, intelligence can only be used so long as it does not countermand the non-intelligent, as well as radically counter-intuitive, procedural logic of the electronic medium. The essential pre-requisite for realising an edition in electronic form is that the computer operations that assist in the preparation of the edition, and as a consequence, the electronic formatting and recording that result from them, are fully consistent in themselves, compatible with each other, and comprehensive.

Granted all this, the fundamental challenge offered by the transference of an edition into electronic form is that of rethinking the whole editing enterprise. The traditional emphasis of editorial endeavour has always been on production. The material nature of the printing process has encouraged us to conceive of the aim of editing in terms of that consummate artefact of cultural techniques, the book. And, since we are thoroughly acculturated to the book, we perceive it as a commodity product which we unreflectingly believe we know how to use, rather than reflecting on the modes of use implied in its sophisticated design and artful crafting. Specifically with regard to scholarly editions in book form, it seems safe to say that they are assessed according to whether and how they fulfil the formal conventions of well-made editions, rather than for how aware they are of their inherent user potential.

Preparing editions to be realised, presented, and engaged with in the electronic medium requires that we define them from the user end and design them from the beginning with the widest possible range of uses in mind. That this must be so is suggested by the lack of this awareness revealed by a number of enterprises from the current pioneer years of electronic text projects. These often (rightly) refer to themselves as 'archives,' or even 'libraries,' and they tend to engage in (sometimes defensive) theorising of the concepts of 'edition' and 'editing'. In so doing, they reflect a general need to reconceptualise the principles of the scholarly edition and the modes of scholarly editing. The electronic medium, then, specifically suggests that we reconsider the notions and practices of editing in relation to what purposes editions serve, as well as how they are used. For it is only by asking pragmatic questions as to how users would deploy, or should be encouraged and guided to deploy, the computer in exploring a scholarly edition that we shall arrive at design solutions for the 'user interface,' prerequisite as the port of entry into the edition's data and their structuring.

The user of a scholarly edition in electronic form needs to be clear that what is being supplied on a given CD-ROM or website is indeed an edition: it should announce explicitly what the edited materials are, what principles and methods have been brought to bear on them, and what the edition claims to have achieved. This will require a discursive exposition along the lines of the introductions customary in print-based editions.

The edition in electronic form will also be expected to provide an edited text. That expectation is carried over from print-based scholarly editing, and it is on the basis of this assumption that the essentials for computer assistance in the preparation of editions were outlined above. Much of the theorising that has surrounded the 'archive'- and 'library'-type electronic text projects of recent years has questioned whether we need to edit for the electronic medium in the same ways that we edited in the traditional medium of print, engaging the editor's choice, critical judgement, and decision as the means of establishing the given edition's edited text. It is precisely on this issue that the various types of text projects in electronic form divide. It would be fair to say that it is the editor's responsible establishment and provision of an edited text within a comprehensive editorial enterprise that defines an electronic edition, just as it defined every traditional edition in print. It is, therefore, the incorporation of an edited text that distinguishes the electronic, or computer, edition from 'archives', 'libraries' or similar electronic text and document repositories.

At the same time, both the introduction and the edited text are parts of an edition that a user would expect, and be expected, to read as continuous texts. But the electronic medium is not a particularly comfortable site for sustained sequential reading. This might suggest the need for a double provision when an edition goes electronic. There have been a number of successful experiments with so-called hybrid editions in recent years. Such electronic editions also comprise a book component, in which their continuous texts remain presented in print form. But the double provision is important. The printed sections must be included in the electronic form of the edition as well. There they are not incorporated for the purpose of reading, but for the exploratory use to which they are capable of being put, in correlation with the whole range of discourses that the edition comprises.

In addition to the edited text and the introduction, a scholarly edition's discourses should include, first, the textual apparatus and notes. These complement the edited text and, by way of reporting the variation in readings and versions, make of the edition in question an edition of the given work. The range of the editorial discourses, furthermore, will commonly include additional documentation relevant to the text and work, possibly incorporating images (facsimiles and digital scans) of the surviving witnesses to the processes of its composition and transmission. Further discourses will enlarge upon content and interpretation through commentary and annotation. Lastly, to link their several discourses, editions are generally supplied with indices, and sometimes glossaries.

For the traditional print-based edition, this discursive conglomerate is structured in a descending hierarchical order. The book, through which the edition's set of discourses is communicated, expects them to be organised on linear principles. We recognise that these discourses are correlated, but in the print-based edition, their correlation, except through the connections made by the index and glossary, remains latent, supported, at most, by cross-referencing -- and the reader's and user's memory. It would surpass the capabilities of the book as medium to realise (except by stimulating recall, and offering marginal space for readers' notes) that networking of its discourses that the scholarly edition, as an intellectual construct, might ideally demand.

As will be evident, the opportunities provided by the electronic form to enhance a traditionally print-based edition do not so much arise from the electronic medium as an alternative to the book for presenting and representing continuous text (since experience shows that it is generally a poor, and always a cumbersome alternative). Such opportunities lie, rather, in the potential that the electronic medium offers for networking the different elements of an edition, that is, for correlating and linking its entire range of tributary discourses. How such a network might be shaped and structured remains to be explored, a field of virtual resources yet to be prospected. And what it takes to transfer a print-based edition into electronic form, can at present only be expressed negatively: a print-based project that only aimed to translate its print-oriented procedures one by one into computer-based routines would clearly be failing to utilise the full innovative potential of the electronic medium. Transferring a print-based edition into the electronic medium involves rethinking the conceptions on which the traditional scholarly edition is based. From a linear and hierarchically structured product of scholarly effort, it needs to be reconceptualised as a use- and user-oriented, relationally networked text data base and electronic site for the exploration of knowledge. Such a fundamental re-conceptualisation will alone generate in its wake the required electronic techniques that we only rudimentarily possess at present.

---