Article

Bringing an Ancient Egyptian Capital to the Digital World

Authors

Abstract

The Hut-Sekhem project is producing a database of artefacts in relation to the ancient Egyptian city. While there is a strong textual component, as many artefacts are text-bearing objects, the corpus also shares a similarity to a museum collection. The article discusses the decision-making process that led us to choose a static website over other technical infrastructures available when preparing a digital edition, such as TEI Publisher in conjunction with XML-encoded texts, or a platform such as Omeka meant for displaying museum or other scholarly collections. As we are not annotating texts but rather displaying them, a complex XML-based solution is not required. A static website turned out to be a perfect solution for us, with easy maintenance on the long term and a very low level of technical knowledge required to update the website.

Keywords:

How to Cite: Jerome, E. & Nury, E. (2026) “Bringing an Ancient Egyptian Capital to the Digital World”, Amherst demo. 2027(1).

Abstract

The Hut-Sekhem project 1 is producing a database of artefacts in relation to the ancient Egyptian city. While there is a strong textual component, as many artefacts are text-bearing objects, the corpus also shares a similarity to a museum collection. The article discusses the decision-making process that led us to choose a static website over other technical infrastructures available when preparing a digital edition, such as TEI Publisher in conjunction with XML-encoded texts, or a platform such as Omeka meant for displaying museum or other scholarly collections. As we are not annotating texts but rather displaying them, a complex XML-based solution is not required. A static website turned out to be a perfect solution for us, with easy maintenance on the long term and a very low level of technical knowledge required to update the website.

Background

Hut-Sekhem – in Ancient Greek Diospolis Mikra, or modern Hiu – was the capital of an administrative region in Ancient Egypt from the Middle Kingdom to the Greco-Roman period (approximately 1990 BCE to 565 CE) and is situated between the two prestigious religious centers of Dendera and Abydos. Unlike its neighbours, Hiu does not contain well-preserved archeological remains and has been largely ignored by archeologists and travellers 2.

To bring this ancient city back to life, a database of artefacts and texts will help us paint a better picture of the city and its history. The database will include objects either excavated in Hiu itself, or objects from Egypt attesting the city. Each item is accompanied by metadata, images and spatial coordinates when available, bibliographic references as well as links to related items.

The hieroglyphic texts inscribed on the artefacts will be available alongside their transliteration and translations both in English and French, and Arabic translations are planned in the future. In addition, the database is completed with more recent accounts from Egyptologists or travellers mentioning the city and the eventual archaeological remains they may have observed during their journey.

Choosing a Static Website

For the two-year duration of the project, digital humanities expertise is funded. However, the excavations and research work are expected to continue well in the future, and the Egyptologists must be able to keep updating the database with new findings as long as possible. A part-time engineer is handling most DH projects in the Arts and Humanities faculty at the University of Geneva, but it means that the amount of time devoted to individual projects is highly restricted. With long term in mind, the data should also be in a format flexible enough to be easily transferred to a better suited tool in the future if needed, or to ensure compatibility with the tools already used for recording excavation findings 3.

Our choice of digital infrastructure was therefore influenced by these three constraints of long term usability, flexibility, and minimal amount of technical knowledge. We needed a solution that would be accessible to non DH scholars, easy to maintain and update with limited support.

This excluded for instance @teipublisher. It could become complicated to maintain XML data without support, but that is not the only difficulty. Without regular maintenance and updates, TEI Publisher websites can partly stop working properly. The IT engineer explained that a database built on eXist could be turned off in case of a security vulnerability, while a static website could be maintained virtually forever.

We also decided against a CMS such as @omeka and @heurist, although they are both supported by the University of Geneva. Omeka, for instance, is not so easily customized to precise requirements if the existing modules are not satisfying out-of-the-box 4. Our preference was to gather data in a spreadsheet format, as it is well mastered, easily shared and versioned. It would also allow for an easier change in technology, should the need arise to transfer the database from one tool to the other after funding is ended.

Static sites were also a good option among the ones we were examining, as their advantages have already been noted: “Static site generation, unlike CMS or database-driven projects, allows websites to be created with significantly fewer dependencies, and reduces maintenance overhead, as well as minimizes the risk of technological obsolescence” [@smith2020] 5.

Eleventy

Among the available static site generators, we quickly settled for @eleventy. Its strengths are its flexibility with various data formats and sources, and its use of JavaScript which allows for easy adoption and a large user community [@delannay2024][p.:59-60], including users in the Scholarly Editing community 6.

In addition to these advantages, Eleventy provides a collection feature 7, a particularly useful mechanism to group items contained in the database by date, material, type, etc. Collections allow us to display together all the items found in a larger structure, e.g., all the blocks that were found in the Ptolemaic temple of Hiu. In the bibliography, we also use the collection feature to create a list of all the items of the database that are mentioned in a bibliographic source.

Fig. 1: example of a bibliographic reference with links to items in our database.

The bibliography is a good example of how easily we can include external data from a private Zotero group. Eleventy creates a local copy of the bibliography in JSON through Zotero’s API, and we can choose how often to refresh the data.

The absence of a server-side database and indexes makes search on a static website less straightforward than in a dynamic one, but there are alternatives relying on JavaScript 8. A search function is necessary to search the full contents of the website, including introductory materials, blog posts, and complete database. While we explore the options and decide which solution will be best suited for our needs, our database items are presented with a selection of metadata in a searchable table format using @datatables. This JavaScript library is already used to display Humanities data: it is the solution adopted for instance to list texts contained in a manuscript in the de Heresi digital edition [@rehr2026], as well as in the Coptos database [@desclaux2019], a similar project that is gathering archeological artefacts from the city of Coptos in Egypt, less than 100km away from Hiu.

Fig. 2: the database in table format.

The table does not contain all the metadata, rather we have included the ones that were considered most relevant for understanding and filtering the items: title, description, type, material, date, findspot and modern location. We have not included modern inventory numbers, or general remarks and textual commentary. All the transliterations from hieroglyphic, original texts, and translations are included in a hidden column, which can be displayed when clicking on the left-side arrow next to the title and are therefore searchable. In addition, columns can be searched individually with filters 9. For the datation we have specifically created a range filter, where users can add a lower or upper date limit, or both. The “from” and “to” limits are added as “data-” attributes in the HTML table, and accessed through JavaScript. This is the main difference with the Coptos database, where dates are filtered and ordered alphabetically on the datation column.

Since the Egyptologists are planning on filling the database with excavations findings in the next few years, it is relevant to ask what would happen with this infrastructure when scaling up. The DataTables FAQ lists the three processing modes and their limits with regard to speed: there are two options for client-side processing, DOM sourced data (our current setting, which can deal with up to roughly 5.000 rows) and Ajax sourced data (up to about 50.000 rows). We tried inflating the database up to 6.500 items, which was considered a rather optimistic number of findings. At that stage the table takes 1 or 2 seconds to appear on the page, the column filters are slightly slower while not preventing searches, and the full-text search is instantaneous. We do not expect to reach such a high number of items, however the table would still be usable as it is, and it could be switched to Ajax sourced data instead.

Editorial Workflow

The table presentation of the database is mirroring how the data is entered in a tabular format into a Microsoft Excel spreadsheet. The workflow is set up so that when the spreadsheet is updated in GitLab, it triggers the script to convert each row into an item’s file in Markdown format, which means that the team’s members can update the website with a few clicks. A sample of our dataset can be found on Zenodo with description of its metadata 10. For this paper I would like to focus more closely on the two types of texts in our database, which are both entered directly into the spreadsheet.

Accounts from scholars and travellers are precious for documenting buildings or artefacts that were visible in the past, but have disappeared. Such texts range from a couple of sentences to at most a few paragraphs. They are entered in the spreadsheet as Markdown, which lets us add light styling with bold and italics. We plan to include links to items in our database with Markdown, for instance [temple-fort](Hiw_108) in item Hiw_141, where an item’s identifier is sufficient to create a link to its page.

Hieroglyphic Texts

Although unicode version 12.0 published in September 2019 has improved the display of hieroglyphs with additional control characters [@glass2021], scholars still need to use a specialized software to write hieroglyphs using a dedicated common syntax that is converted into unicode or an image format. Our texts are written with JSesh [@rosmorduc2025] and stored as plain text with the “.gly” extension 11. These files are our primary format for hieroglyphic texts, as they will be required by publishers for any book or article produced during the course of the project.

To display the texts online, we convert the plain text syntax to Unicode first in a JSON format, and then HTML format through the suite of tools called Hierojax [@nederhof2026]. Ultimately, we produce three files for the hieroglyphs, with three different extensions: .gly, .json and .html. They are linked by their identical filenames, which is included into the spreadsheet under the field “hieroglyphs”.

The texts are usually short, since in case of a very long text we will transcribe only the portion relevant to the city of Hiu, and refer readers to the full publication of the text. However, the texts often need to be separated into parts indicating where to find them. For example, in the “Stela of Irethoreru” (Hiw_58), the text is divided into “first line of the text” and “Column of text on each side”. In this case we are using a JSON notation for the transliteration and translations structured as follows:

{"First line of the text" : "\[… wʿb n (?)\] sḫm.t n ḥw.t-nṯr n ḥw.t ḥʿ.w nfry (?) ỉr.t-ḥr-r=w mn \[wȝḥ (?)\]",  
"Column of text on each side" : "ḏd-mdw ỉn wsỉr ỉr.t-ḥr-r=w mȝʿ-ḫrw (?) \[sȝ pȝ-\]šrỉ-(n)-? mȝʿ-ḫrw"}

Final Remarks

Storing textual data in a spreadsheet, instead of the recognized standard XML TEI, is maybe unusual, but it seems to be a convenient way to work with static websites, as a spreadsheet is also used by the Recetas project [@tew2024] or spreadsheets and Word documents for the Van Dyck Catalogue [@carius2024]. It is feasible because all our texts are short. This setting does not allow a transcription as fine-grained as what would be possible with TEI, although we are able to include light markup in the form of Markdown. Since our primary goal is to make the texts accessible online, and not to annotate them with morpho-syntactic or philological aspects, this lack of fine-grained markup is not problematic. In fact, converting the texts to XML — even if automatically from the spreadsheet — would add another layer of material to maintain which is not necessary. Especially in the case of hieroglyphs, where we are already dealing with three different formats, adding TEI XML seems of marginal benefit. The current organisation of textual data does suit our needs for flexibility and minimal technical knowledge, and the static website offers a guarantee of long-term usability.

Notes


  1. Project no 10003966 funded by the Swiss National Science Foundation. https://data.snf.ch/grants/grant/10003966.↩︎

  2. The site was superficially excavated by Sir W. M. Flinders Petrie in 1898-1899 and by Jean Capart in 1927, and more extensive excavation work was started again in 2024 by the Egyptology and Coptic department of the University of Geneva. See here a brief summary of excavations and discoveries in Hiu: https://web.archive.org/web/20251208070031/https://www.unige.ch/Hout-Sekhem/campagnes/historique-des-fouilles.↩︎

  3. The tools used by the project’s team are @qgis and @archeobase.↩︎

  4. Omeka relies on themes and plugins for customisation, which may cause maintenance issues in the long term [@leblanc2019][§:35-40]. The necessity for customisation is also cited as an argument by the authors of the Van Dyck Catalogue, who decided against using the more generic @quire developed by Getty and chose to develop a solution tailored to their needs [@carius2024].↩︎

  5. See also @meier2022 and @carius2024 for arguments in favour of static websites, and @delannay2024 for a detailed description of three editing modes for digital scholarly publications: XML-based solutions, CMS, and static sites.↩︎

  6. Eleventy was chosen as the static website generator for converting a TEI Publisher dynamic edition to a static website for long-term preservation [@meier2022]. The publishing tool Quire for digital scholarly books created by Getty switched from @hugo to Eleventy as its site generator, citing among other arguments a better development support (see discussion #314 on Github: https://github.com/thegetty/quire/discussions/314). On the other hand, see @tew2024 for arguments in favour of Hugo, in particular for multilingual websites.↩︎

  7. https://www.11ty.dev/docs/collections/. In Hugo, taxonomies offer a similar function.↩︎

  8. Alternatives such as the @pagefind library, Publican’s @staticsearch-publican or the @staticsearch-endings codebase developed for the Endings Project.↩︎

  9. Datatables filters are illustrated in the documentation, for text input (https://datatables.net/examples/api/multi_filter.html) search and dropdown selection (https://datatables.net/examples/api/multi_filter_select.html).↩︎

  10. https://doi.org/10.5281/zenodo.20327590.↩︎

  11. Examples of texts encoded with JSesh: https://github.com/rosmord/MDC-texts.↩︎