Over the last decade, many humanities scholars have been persuaded by the promise and the power of encoding schemes for electronic texts to create texts, sometimes very large and complex, encoded using these schemes. This is specially true of SGML/XML based encodings, with the implementation of the Text Encoding Initiative being particularly influential in the community. However, scholars who have made such texts have typically discovered that software to publish them is too expensive for their limited budgets, or too difficult to use, or lacking essential facilities, or all three of these. The Anastasia electronic publishing system, developed in the last five years in a partnership between the Centre for Technology and the Arts, De Montfort University, and a new electronic publishing company, Scholarly Digital Editions (SDE), attempts to supply this deficiency. Anastasia stands for ‘Analytic system tools and SGML integration application’. As this implies, it is able to handle all valid SGML and XML documents, with no limits on their complexity. Particularly, Anastasia has been designed to meet the needs of humanists, and especially textual scholars. It is a common complaint of humanists that SGML/XML systems constrain a single hierarchical view of a document, while humanities texts can be seen as containing many overlapping and competing hierarchies. SGML/XML publishing systems usually cannot support facilities which cut across the primary document hierarchy, and so cannot satisify even such simple needs as display of a single page of transcription, or display of a tabular list of key word in context search results with formatting of all returned search strings according to the embedded encoding. Anastasia seeks to escape these limitations by adopting a document processing model that sees the document as made up of a series of events which are defined not only by their hierarchical relation, but also by their left to right relation in the document stream. As a result, Anastasia provides tools which allow the document to be manipulated according to alternative hierarchies implicit in the element relations. Thus, one can very easily extract views of the text by column or page, or indeed start a display at any point in any element and continue to any point in any other element. A KWIC display, for example, requires that we display an arbitrary number of characters before a hit, then display the characters in the hit themselves, and then display an arbitrary number of characters after the hit, all with complete awareness of the document encoding within those spans of characters: Anastasia can do this. Then, one should be able to click on a link from the KWIC display to the document itself, and see the hits highlight in the full-text context: once more, Anastasia has been designed to make this easy. One can also manufacture virtual texts by extracting and combining multiple and even overlapping segments. Anastasia is also designed to fill another need: for a mode of publication which is identical on both CD-ROM and the internet, on the major Windows and Macintosh systems. Typically, the scholar will prepare a body of SGML/XML documents for publication using the Anastasia GroveMaker application, which compiles the documents into a binary database. The Anastasia Reader then serves the documents to an internet browser, either over a network or from a CD-ROM. Control of all aspects of the publication's display and behaviour (including fully SGML/XML aware searching) is achieved through a series of Tcl script files. A key factor in the development of Anastasia has been the desire to achieve publication without compromise. That is: if it is possible to achieve a certain kind of computer display effect, then Anastasia will allow this. For example, we might want to use some of the advanced dynamic HTML features permitted by Javascript: pop-up menus, text which changes colour as the mouse passes over some other part of the document (for instance, to show that a word or phrase in one window is a translation of, or is otherwise related to, a word or phrase another window), synchronous scrolling or separate windows, and more. Practically, this means that we should be able to generate streams for display in any format whatever, directly from the XML: in pdf, SVG, rtf, any variant of HTML and XML, and send it directly to the display engine. We have concentrated on using Anastasia to generate HTML with Javascript: an example of the effects possible through this can be seen in the work on the digital 28th edition of the Nestle-Aland Greek New Testament, accessible through nestlealand.uni-muester.de. Other instances can be seen from the SDE website, http://www.sd-editions.com/anastasia. Anastasia is designed to work as a Apache webserver module. It also requires C-language support, and the Tcl (Tool Control Language) libraries. In theory at least, this means Anastasia can operate whereever Apache operates: our main development is on Macintosh OS X and Windows machines; there is also a Linux port. The search systems in Anastasia are based on SGREP, written by Jani Jaakkola and Pekka Kilpelainen of the University of Helsinki: we have heavily customized the SGREP code to improve its performance with large texts. Perhaps one of the most distinctive (if not controversial) features of Anastasia is that the style sheets we use to control exactly how the source XML is sent to the browser are written in Tcl, and not in any of the various XML-based systems which have appeared in the last years (such as XSLT, XPATH, and others). In part this is historical: the roots of Anastasia lie some distance back, as far as the first work done by myself on the Canterbury Tales Project with Elizabeth Solopova and Norman Blake) in 1993, long before even XML made an appearance. In part, it is because those systems themselves remain in a start of flux. But it is also because there is room for argument about the efficiency of such schemes. There is no doubt that XML is superb at representing textual structure. But this does not mean it is suitable for use as a programming language, requiring ease of use, rapid development, efficient maintenance, and widespread support across many different computer systems. Tcl does offer all these. Anastasia is not intended to be the tool of choice for everyone who works with XML. It is designed for situations where the very best possible presentation is required of highly complex XML. A single screen of the digital Nestle-Aland, for example, may draw XML from hundreds of different places within the source, reformat into HTML interwoven with Javascript commands, and spread this across a series of frames nested within the browser display Ð all in a fraction of a second, in response to a request from the reader. It is also designed to run identically on CD-ROM and over the internet. Reports of the death of CD-ROM appear rather exaggerated: indeed, the availability of cheaply priced publication tools such as Anastasia may make it possible for high-quality CD-ROMs to be made available at much lower prices than hitherto, and so create a market which has been previously elusive. Finally, my hope when designing Anastasia was that a single scholar, with reasonable dedication, good knowledge of XML and with no more computer resources and support than are commonly available within university departments, would be able to use it to make high-quality XML based publications. There have been some encouraging signs that Anastasia can indeed be used in this manner. In the same context, it should also be appropriate for use by smaller academic publishers. This is the first conference presentation of Anastasia as a mature publication system. There has been one previous conference presentation of the system, at the DRRH conference in Sydney in September 2001, when only a preliminary version of the software was available.