TOWARD A TYPOLOGY OF THE ELECTRONIC TEXT

Dr Toby Burrows, Scholars' Centre, The University of Western Australia Library, Nedlands, W.A. 6907

Paper presented at the Conference of the Bibliographical Society of Australia and New Zealand, Perth, October 1997.

In a spirit of bibliographical analysis rather than exhortatory polemic, I aim to develop a preliminary typology of scholarly electronic texts, through an analysis of a sample of current electronic texts and editions. The characteristics of these texts will be considered under five main headings:

1. The markup scheme employed

   An electronic text may be marked up in several different ways, ranging from the so-called "markup-free" ASCII to full-scale SGML encoding. HTML markup is commonly used. Even proprietary word-processing software contains its own form of markup. An emerging possibility is XML.

2. The extent to which the edition is dependent on specific software

   Some electronic texts require specific software to interpret them while others may be handled by a variety of different software. In some cases, while the text is intrinsically independent, it is nevertheless distributed in a package with specific software.

3. The method of distribution or publication

   Electronic editions were originally likely to be distributed on computer diskettes. CD-ROM subsequently became a popular means of publication, but the most common form of distribution is now the Internet, either by file transfer protocol (FTP) or - more usually - through the World Wide Web.

4. The overall structure or architecture of the edition.

   Scholarly electronic texts may be presented as simple linear text. More complex approaches include a segmented linear text, with tables of contents; a collection of images; and a structure which is essentially a database. Combinations of some or all of these are also possible.

5. The type of edition involved.

   There are various alternatives here, mirroring those which are possible with printed editions: diplomatic, facsimile, critical, single version (whether copy text or eclectic), multiple versions (with or without a critical text, annotations and cross-references). Electronic editions may well combine two or more of these approaches.

Toward a Typology of the Electronic Text

The debate over the social and cultural implications of electronic texts has been raging for years already, pursued with polemical fervour on both sides. I don't intend to take part in this debate, however. Instead I'd like to attempt something different: a preliminary typology of the electronic text, from a formal, analytical point of view rather than a cultural or polemical one. Perhaps a suitable analogy might be to an archaeologist examining the fragmentary remains of artefacts from a distant time, and trying to construct a typology of the artefact concerned.

A possible model might be the approach adopted by the late Sir Eric Turner in his book The Typology of the Early Codex. Turner was an eminent papyrologist, Professor at University College London, who wrote the definitive book on Greek manuscripts. After his death in the early 1980s, his library was purchased by our own University of Western Australia Library, but was not kept as a separate collection, unfortunately. Turner analysed the physical and structural features of early Greek codices from the period between the second and sixth centuries A.D. He examined such things as their page dimensions, the material of which they were made (papyrus or parchment), how they were assembled into books, and the scribal techniques used for writing them. He finished with an inventory of more than eleven hundred codices and their physical characteristics.

It's a truism to say that the invention of the codex was a watershed in the history of texts. As Turner remarks, we should place the inventor of the codex among the greatest benefactors of mankind, along with the inventor of the wheel and the deviser of the alphabet (Turner 1975:1). Does the electronic text mark a similar watershed? I think it does - though I'm talking here strictly in terms of its formal, structural properties, not its cultural and social significance. In a formal sense, the electronic text is the next major watershed after the emergence of the codex. The invention of printing, in contrast, simply perpetuated the form of the codex. Its immense cultural and social significance arose from the way in which it mechanized the production of the codex, without changing its essential form.

We are living in the early days of the electronic text - its "incunabula" period. But sufficient evidence has emerged in the last three or four years on which to base at least a preliminary and tentative typology. I propose to look at five main characteristics of electronic texts as they exist today:

• The markup scheme employed
• The extent to which the edition is dependent on specific software
• The method of distribution or publication
• The overall structure or architecture of the edition, and,
• The type of edition involved.

Before I move on to these features, a few general comments. I shall be focusing on electronic texts which offer a new version of an already-existing printed or manuscript text. There are numerous examples of texts composed entirely in electronic form, with no corresponding codex version. These include novels and poetry as well as more hybrid forms - including multimedia. But as far as I know there has been no attempt to subject these to textual scholarship. The new MLA Guidelines on Electronic Scholarly Editions reflect this - they focus on electronic editions of non-electronic works, and contain no mention of electronic editions of electronic works. It's hard to imagine what form such editions might actually take - a subject for speculation at another time, perhaps. But I am emphatically NOT intending here to undertake a typology of the entire range of electronic writing and publishing. My concern is to develop a typology of electronic texts which have previously existed in printed or manuscript form.

I should also quickly deconstruct the term "electronic text". Strictly speaking there's no such thing - there is only a representation of a text on a computer screen. The electronic text actually exists as strings of binary digits deep in the machine's memory, shuffled through layers of software of ascending complexity into the dots or pixels which appear to combine to form a text on the screen. Yet another layer of illusion and artifice is added to the already-existing abstraction of formed letters. Still, let's accept the illusion of an electronic text and analyse it as if it had an objective existence!

MARKUP SCHEMES

I said that there were five key characteristics of electronic texts. I'll begin with the way in which the text is marked up. This is probably the single most important characteristic and I intend to spend rather longer on it than on each of the other four areas.

Markup, in the broadest sense, can be defined as "all the information in a document other than the 'contents' of the document itself, viewed as a stream of characters" (Sperberg-McQueen 1991:35). In other words, anything other than the actual letters which make up the text should be regarded as markup. Most obviously, perhaps, markup includes all information about the format of the text, such as margins, fonts, and page breaks. It also covers all information about the structure of the text, such as chapter divisions, paragraphs, headings, footnotes, endnotes, sentences, and even word-division.

Markup is independent of the form in which the text is distributed. Both printed books and manuscripts contain substantial markup, though eprhaps only members of the Bibliographical Society are likely to notice it!

For the representation of texts in electronic form, markup is essential. Without it, a computer cannot display a text to a reader in a meaningful way. Also crucial is the type of markup scheme chosen, since this determines the eventual appearance of the text and the ways in which it can be used. Two main types of markup can be identified: procedural, and descriptive. Procedural markup refers to the punctuation, formatting, and appearance of the text.

Descriptive markup, on the other hand, focuses on the content and structure of the text. Particular elements within the text are identified using starting and ending tags, which are distinguished from the text itself by being enclosed in angled brackets. Instead of formatting a chapter heading as "Times 18-point bold", it can be labelled with the tags or something similar. No formatting is presupposed, but the function of the heading within the text is recorded and specified.

With descriptive markup, a text can be easily transported between different types of software, without the need for reformatting or rekeying. Files can also be reused in a variety of different formats - print, CD-ROM, networked distribution - without any need to change the markup.

In analytical terms, electronic texts can be characterized by the form of markup they employ. Some texts use the procedural markup inherent in word-processing software like Word or Word Perfect, or in type-setting systems like TeX and LaTeX. These apply largely to earlier electronic texts or to those intended for publication only in printed form. Neither approach is now regarded as an acceptable means for distributing electronic texts.

Most texts use descriptive markup instead. The simplest version is the so-called plain ASCII text. ASCII is the American Standard Code for Information Interchange. An ASCII text generally means a file which contains the raw text with only the most basic markup: spaces, carriage returns, and punctuation. There is no indication of formatting, or of the structure and content of the text.

There are various services which provide ASCII texts over the Internet. The best-known are Project Gutenberg, the Online Book Initiative, and Wiretap. They contain dozens of electronic texts ranging from the works of Plato to the speeches of President Clinton. But these ASCII texts are basic raw material only. Their lack of markup means that they are of little scholarly value or usefulness, except as a somewhat unreliable reading text.

The next level of complexity in descriptive markup is the HyperText Markup Language, or HTML. This is the standard format used for documents published on the Internet through the World Wide Web. HTML files can embed within themselves the so-called "hypertext" links, enabling the reader to jump to other HTML files stored anywhere in the world. They can also link to images and other kinds of "multimedia" files. HTML files can be designed to include some structural information, in the form of tables of contents, and can be searched for specific combinations of letters and other characters.

There are a considerable number of texts available in HTML form. A good example is the collection produced by James O'Donnell at the University of Pennsylvania. It includes classical and early medieval texts in both Latin and English, such as Augustine's Confessions, and The Consolation of Philosophy of Boethius.

HTML only makes possible a comparatively low level of encoding for the structure and content of a literary text. HTML files contain little more than some typographical markup and a basic structure of paragraph division. In O'Donnell's case, the files are not even segmented, though some are very long - an English translation of Aristotle's De anima, for instance, fills over a hundred consecutive screens.

The most sophisticated level of descriptive markup is provided by the Standard Generalized Markup Language, or SGML for short. This is a highly elaborate and detailed framework for indicating the structure and content of a text. SGML is a generic approach only, and must be applied to a specific text through the use of a Document Type Definition, or DTD. This is a reference file which sets out rules for marking up a document or class of documents. For electronic texts in the humanities, a DTD has been developed which is known as the Text Encoding Initiative, or TEI. The TEI describes about four hundred possible markup tags, which can be used in various combinations for different genres.

Most of the more ambitious electronic texts are encoded in accordance with the TEI or another type of SGML markup. The University of Virginia Library, for instance, offers a corpus of numerous Middle English texts which use the TEI type of markup. Chadwyck-Healey Ltd publishes a range of large databases with SGML encoding, including the Patrologia Latina Database, English Poetry, and the Bible in English. The Oxford Text Archive is a large repository of public-domain texts, many of which follow the TEI encoding scheme. There are numerous specific projects which are based on SGML or the TEI, among them the Canterbury Tales Project, the Electronic Piers Plowman, and the Dante Gabriel Rossetti Archive.

SGML markup is generally regarded as the definitive way of encoding electronic texts. It makes possible a sophisticated approach to the text. With the right software, complex questions can be put to the text, which draw on the analytical information contained in the markup. It is possible, for example, to search for instances of word usage in very specific contexts in a large text or corpus of texts. SGML encoding can also be used to derive tables of contents automatically, and to build in internal and external hypertext and hypermedia links.

The major drawback of SGML is its complexity. It requires considerable expertise to understand and make full use of the information contained in Document Type Definitions and in marked-up texts. This disadvantage is being addressed by a new generic approach to markup, known as the Extensible Markup Language, or XML. It promises to provide less complexity than SGML but more sophistication than HTML. As yet, there are no electronic texts which use XML, but they are likely to start appearing within the next twelve months or so.

Image files

Image files are a minimalist approach to electronic texts. They do not involve markup at all, and do not permit any searching or analysis of the text. They only offer the ability to browse the text as if one were turning over the pages of a printed book or journal.

SOFTWARE

The second major characteristic of an electronic text is the extent to which it is dependent on, or independent from, specific software. An electronic text must have software to interpret and display it, but this does not mean that the text must only be used with a single type of software. The most flexible and versatile electronic texts are those which can be used with various different types of software. In computer jargon, this is called "portability" between software platforms. Publishing a text which requires a specific brand of software to read it and use it might be likened to publishing a book which can only be read with prescription spectacles from a specific firm of opticians!

The software used for electronic texts falls into three main groups. Word-processing software is the original means of creating and viewing electronic texts on personal computers. There are many available varieties, most of which offer a wide range of helpful formatting devices. They work best with texts which are in their own proprietary format or in basic ASCII, but some are able to handle files in formats created by their competitors. Some brands of word-processing software also have the ability to read and produce files in HTML format.

A second group of software is designed specifically for the creation and manipulation of electronic texts. Among the better-known varieties are: TACT, OCP, WordCruncher, DynaText, Open Text, and Folio Views. In some cases, such as DynaText, the software works with SGML-based files which can be transferred to and reused by other kinds of software. In other cases, the files are in a format specific to the particular brand of software involved. Also in this category are several varieties of SGML editing software, which are designed specifically for the creation and viewing of SGML-based texts.

Texts published commercially on CD-ROM usually have this kind of software included with them. Chadwyck-Healey Ltd sells its texts on CD-ROM with a version of the DynaText software. The American company Intelex publishes a CD-ROM series of philosophical texts (known as Past Masters), which come with software called Folio Views.

The third main group is Web browsing software like Netscape and Internet Explorer. These are probably the most common means of viewing electronic texts in HTML or ASCII formats, though they do not have the capacity to create texts. There is also SGML browser software called Panorama, as well as various kinds of Gopher browsing software, popular in 1993 but now largely obsolete.

Some texts on CD-ROM require Web browsing software like Netscape but do not include a copy of it. Texts published over the Internet are distributed independently of any specific software.

Image files require a range of different types of viewing software, depending on the format of the file. The simplest approach is to store the images in a format called GIF and to use Web browsing software like Netscape to view them. The images provided over the Web by the Electronic Beowulf Project use this approach. But if an image of higher resolution is required, a more specialized viewer must be used. Many electronic journals use a type of image file called PDF, which requires Adobe Acrobat viewing software. Acrobat, like Netscape, is widely available free of charge. There is also an image format called TIFF which offers high resolution, but can only be viewed with specialized commercial software.

DISTRIBUTION AND PUBLICATION

The third variable affecting an electronic text is the way in which it is distributed and published.

In earlier years, electronic texts were likely to be distributed on a floppy disk or diskette. Our library has a copy of the works of David Hume, published in 1990 on eight floppy disks for IBM personal computers. This kind of approach now looks extremely antiquated. Diskettes have only a small capacity, and their format is likely to be a source of difficulties. Newer computers generally do not accept five and a quarter inch disks, and a disk created with a early version of MS-DOS is unlikely to run successfully under Windows 95.

Diskettes were quickly superseded by CD-ROMs, which have been used to distribute electronic texts since the late 1980s, and are still used extensively for commercially-produced texts. In the last eighteen months, Cambridge University Press has published several major CD-ROM texts, for instance. These include Samuel Johnson's Dictionary of the English Language (the first and third editions), and the Collected Works of John Ruskin (originally published in 39 large printed volumes between 1903 and 1912), as well as Peter Robinson's remarkable edition of the Prologue to the Wife of Bath's Tale. CD-ROMs have a considerable storage capacity - equivalent to about 450 diskettes or 300,000 printed pages. The entire second edition of the Oxford English Dictionary fits on to a single CD, and so does the Encyclopaedia Britannica.

Using a CD-ROM requires a personal computer with a suitable CD-ROM drive. Though these are now a standard component of most new computers, they must be bought separately for attaching to older machines. The speed of the CD-ROM drive is an important consideration. Twelve-speed drives are now commonplace, but our Scholars' Centre machine is battling on with its two-speed drive from three years ago. The operating system of the local machine is another major issue. Some CD-ROM texts will only run on a Macintosh, while others require older versions of Windows or even DOS. It is possible to make CD-ROMs available over a local network, but generally not to Macintoshes from a Windows machine and vice-versa.

Another possibility is the distribution of electronic texts on magnetic tape. This has been extensively used by university libraries for making bibliographic databases and indexes available over a campus network, but its use for electronic texts has been fairly limited. Most larger North American universities have a networked electronic text service of this kind, but in Australia only the University of Sydney and the University of Western Australia currently offer such a service. It needs a comparatively large computer known as a server, to which the texts on magnetic tape can be loaded for access from across the campus network. A substantial degree of local computing expertise is necessary. The operating system for such a service is usually UNIX, and a variety of software can be used to access the texts over the network. The best approach is probably to use what has been dubbed an "intranet" - a kind of internal Internet which relies on Web browsing software like Netscape or Internet Explorer.

Distributing electronic texts over the Internet has been around for quite a few years, but only began to become widely used in the early 1990s. The earliest method of distribution was through the File Transfer Protocol, or FTP. This involved connecting to a remote computer, browsing its list of available files, and copying one to the local computer, where it could be read with word-processing software. FTP was then joined by a method called Gopher, which allowed files to be read at a distance without being transferred. Both of these are still used to some extent. The Dartmouth Dante database is a Gopher service, and most of the collections of ASCII texts - such as Project Gutenberg - rely on FTP. Even some SGML-based collections, like the Oxford Text Archive, use FTP.

Both FTP and Gopher have now been incorporated into the World Wide Web and are usually accessed through it. The Web has become, in little more than three years, the most popular method by far for distributing non-commercial electronic texts. There are literally thousands of texts available through the Web, such as James O'Donnell's collection which I mentioned earlier. Many of these texts are in HTML format, but the Web doesn't necessarily mean HTML. The Virginia corpus of Middle English texts, for example, stores texts in an SGML format and translates them automatically into an HTML display.

Commercial publishers are also beginning to move towards direct publication and distribution over the Web. At the end of 1996, Chadwyck-Healey Ltd launched Literature Online, advertised as "the home of English and American literature on the World Wide Web". Literature Online offers thousands of electronic texts of poetry, drama and fiction in English. A parallel service from the same company offers the Patrologia Latina Database over the Web. Access to both these services is by annual subscription, rather than by an outright purchase. This is a clear endorsement of the Web as a delivery mechanism, and is likely to mark the beginning of the demise of CD-ROMs as a means for publishing electronic texts.

There are major advantages in Internet-based services, at least for the local user. There is no need for local storage or maintenance of large files or specialized programmes. As long as pre-existing local network structures can be employed, no additional local computing expertise is necessary. But such services rely heavily on the speed and quality of Internet access, and are dependent on generic Web browsing software for their use. This tends to result in a lower level of sophistication in searching and browsing than is possible with more specialized software.

STRUCTURE AND ARCHITECTURE

The fourth main characteristic by which electronic texts can be categorized is their structure or architecture. By this I mean the way in which the text is arranged for presentation to the reader.

At the lowest level, this means a simple linear presentation. The reader scrolls through the text from beginning to end, usually on successive screens. There is no alternative way of reading the text. Most ASCII texts, like those of Project Gutenberg, are presented in this way. So too are some HTML texts, such as those made available by James O'Donnell. This kind of structure is strongly reminiscent of the manuscript scroll of the classical world, and lacks any of the advantages of the codex.

A higher level of sophistication is offered when a text is divided into segments which can be browsed separately, and in any order. The text usually comes with a table of contents which is hyper-linked to the individual segments. This kind of presentation is common in HTML texts distributed on the Web - especially for large texts like William Schipper's HTML version of a large encyclopaedia composed by Raban Maur in the early ninth century - known as De naturis rerum. With this kind of text, the reader can choose to go directly to a particular chapter or section and read from there, and can read the complete text in any sequence. This has some similarity to a printed book, inasmuch as a book can be opened at any chapter and can be read out of its linear sequence.

Also rather similar to a codex is an electronic text which consists of a linked series of image files. These, too, usually have a table of contents and can be viewed by taking the files in a variety of possible orders. The reproductions of Australian colonial newspapers provided by the Australian Co-operative Digitisation Project work in this way. So too do most electronic journals, such as those from Johns Hopkins University Press, distributed on the Web under the name Project Muse.

These types of textual structures are still comparatively close to their non-electronic predecessors. A distinctively electronic approach only begins to emerge when a text is presented in a hypertext structure. By the term "hypertext", I mean a text which contains within itself numerous embedded direct links and jumps. These links can be to related pieces of text, footnotes, or associated images and other kinds of media.

Even the most complex hypertext requires some kind of entry point, which is the rough equivalent of a table of contents. But the underlying structure is more likely to resemble a network or web of textual fragments or components, which can be navigated by numerous different routes.

The World Wide Web itself, from one point of view, can be considered as a gigantic hypertext. So too can sections of the Web; the Labyrinth Library of Medieval Studies, maintained at Georgetown University, is an example of a kind of meta-text composed of numerous individual medieval texts arranged in a loose hypertext network.

While the web-like hypertext is undoubtedly the most typical and distinctive form of electronic text, there is one more type of structure to be considered. In this kind of architecture, there is no browsable text at all. Instead, there is only a database which can be searched for particular combinations of letters, words or characters. The Dartmouth Dante works like this. The only entry point is a search screen, and the only way of reading the text is to retrieve a small fragment which contains one or more search terms.

This approach is rarely presented as the ONLY approach to a text, however. It is usually found in parallel with another kind of structure, especially with a hypertext one. The text as a database, and nothing more than a database, violates our sense of reading. Most of us need to be able to browse a text in some way, instead of just searching it at random.

Ultimately, there may be only two basic architectures for electronic texts: the browsable text and the searchable text. Browsing can encompass simple linear scrolling, parallel browsing of multiple texts and versions of the same text, browsing from a table of contents, and browsing through a hypertext web. Searching can cover the text as a searchable database, and the text as a component within a searchable corpus.

The most sophisticated electronic texts contain a variety of different structures simultaneously. A good example is Peter Robinson's CD-ROM edition of the Prologue to the Wife of Bath's Tale. It contains a table of contents, a text segmented into browsable sections, searchable databases, and hypertext links at numerous points in the text, especially to variant readings and transcriptions. There are also images linked to the text which can be browsed separately. Among its distinctive features are dozens of multiple versions of the text presented in parallel, as well as spelling databases covering all morphological forms. The text can be read in a linear order, browsed by almost innumerable different routes, or searched for occurrences of specific letters or words.

TYPE OF EDITION

The fifth and final characteristic by which electronic texts can be analysed is the type of edition which they embody.

Electronic editions have a close connection with postmodernist approaches to editing. Critics like Peter Shillingsburg and David Greetham have praised the ability of the electronic medium to provide an eclectic, open-ended and "reader-driven" edition. "Most scholars now recognize," says Greetham, "that there is a natural affinity between the computer and the variable discourses of contemporary textual scholarship" (Greetham 1994:345).

It does not follow, however, that an electronic text must necessarily take a postmodernist form. Electronic texts can in fact take any of a variety of editorial approaches, and most are exhibited in the texts available at present.

Many electronic texts simply reproduce an earlier printed edition. The archives of ASCII texts, like Project Gutenberg, tend to rely on copy texts derived from a printed version chosen only for its ready availability and lack of copyright ownership. In many cases, the source of the copy text is not even documented.

Commercial publishers generally also reproduce printed editions which are out of copyright. Chadwyck-Healey follow this approach consistently, using Migne's Patrologia Latina for their medieval and patristic texts, and pre-twentieth century editions for their databases of English literature. The copy texts are carefully documented, however, and their physical features are recorded in the electronic version.

The philosophy texts distributed by Intelex under the title Past Masters are, in contrast, generally taken from twentieth-century critical editions.

Collections of non-commercial electronic texts, like those at the University of Virginia or the Oxford Text Archive, also use a single printed version of the text as their base. This may be either a critical edition or just a convenient printed text, but the source is usually carefully documented.

A growing number of electronic texts are more original and ambitious than this, however.

The Electronic Beowulf Project, directed by Kevin Kiernan, is a good example of a facsimile edition in electronic form. It contains hundreds of colour images of the Beowulf manuscript, as well as of transcripts of the manuscript, related manuscripts, and printed editions. The images are far more numerous and accurate than would be possible with a printed facsimile edition, but the underlying approach is the same.

Hoyt Duggan's electronic archive of Piers Plowman combines several types of edition. It contains documentary editions of the eight main manuscript witnesses to the B version of Langland's poem. But there is also a reconstruction of the archetype of the B version, and a critical text derived from this archetype. Facsimiles of the manuscripts are thrown in for good measure!

The electronic edition of the Prologue to the Wife of Bath's Tale also combines several different editorial approaches. It includes documentary transcriptions of all fifty-eight witnesses to this text which are earlier than 1500. Each is also accompanied by a facsimile. Instead of a critical text, though, there is a copy text - the Hengwrt manuscript, lightly amended, which forms the base to which all the transcriptions are linked.

While the electronic text can reflect any of the major approaches to editing texts, it certainly offers the editor far more flexibility in presenting an edition than is possible within the confines of the printed page. And it seems to be true that open-ended, postmodernist editions need to use the electronic form to achieve the flexibility they require. Variant texts can be stacked for separate and simultaneous viewing. The multiple intentions of the author, and of subsequent readers, can be presented in parallel. The reader can construct a personal critical edition from the range of materials presented. In David Greetham's apt simile, these "fragmented, spliced, versioned, polysemous texts" are the equivalent of the Pompidou Centre in Paris, with all its structural elements exposed, or like a giant Meccano set (Greetham 1993: 14-15).

Conclusion

To a future textual scholar looking back on these early electronic texts, the dominant theme is likely to be variation and experimentation. Our electronic incunabula are distinguished by their eclecticism in every area: the type of markup used, the kind of software required, the method of distribution employed, the structure and architecture of the text, and the kind of editorial approach used. It is possible that, over time, a more limited range of standard varieties may emerge, as happened with the manuscript codex and the printed book. But this seems unlikely. Eclecticism is inherent in the electronic format and is likely to persist for some considerable time.

There is no single best or right way to present an electronic text. Some of the features I've discussed are undoubtedly more suited to particular purposes. SGML-based markup is more appropriate for a scholarly edition than plain ASCII, for example. But each type of electronic text serves a purpose, just as popular paperbacks and glossy magazines have their place in the history of print culture. We undoubtedly need the enthusiasts and the Luddites to debate the cultural and social significance of electronic texts. But there is also room for the archaeologist to investigate and document the different types of texts, and to analyse their properties.

[ References - printed and Web - to be added in due course! ]