Papyri in Pixels: A Research Visit to UC Berkeley, California

By Prianka Schmidt

In June 2026, Marcel Schaeben and I were guests at the Center for the Tebtunis Papyri at the Bancroft Library (University of California, Berkeley), where we set up and commissioned two workstations for digitising the papyrus collection held there. For me, it was my first research visit of this kind and an opportunity to test a project that had been months in the planning under real-world conditions.

Where the papyri come from and what they tell us

The collection comprises several thousand texts written in Ancient Greek, dating from between the 3rd and the 1st century BC. The papyri were extracted from mummy cartonnage, the material used to wrap people and crocodiles for burial, and are consequently highly fragmented. While there, we learnt that many of the fragments come from the interior of the crocodile mummies. They are particularly fragile, as they were exposed to a great deal of moisture during mummification. The mummies were discovered in the winter of 1899/1900 during excavations at Tebtynis, present-day Tell Umm el-Baragat in the Fayum. The ancient town was an important cult centre of the crocodile god Sobek, which also explains the large number of mummified crocodiles.

In terms of content, the texts range from official and private correspondence, through contracts, registers and receipts, to petitions, reports and notes. They are particularly revealing about the living conditions of the population that had been migrating to Egypt from various parts of the Greek world since the 4th century BC. The fragments thus offer a unique insight into everyday life at that time.

What the project aims to achieve

The Academy project “The Crocodile and Human Mummies of Tebtunis” combines traditional methods of textual editing with digital approaches. As all the papyri in the collection originate from the same excavation context, there is the prospect of using information technology to reunite numerous previously unstudied fragments and to make the collection’s value as a historical source more fully accessible. The digital component has two main focuses: first, the digital capture of the fragments in accordance with established standards and in the highest possible quality, and making them available via a research and publication platform; second, supporting the reuniting of related fragments, in particular by testing and adapting AI-assisted methods for “fragment matching”.

Consistent, comparable colour values are essential for this process. Only under reproducible imaging conditions can fragments be reliably compared by machine on the basis of fibre orientation, broken edges, script and colour. Reproducibility was therefore a fundamental prerequisite.

Preparing without a precedent

There was no direct precedent for a project of this kind. A project in Pisa has already worked with various types of digital images; we are going a step further in that AI-assisted fragment matching was considered from the outset and the imaging conditions were consistently tailored to it.

We had previously worked with reproduction photography in other contexts, but never with the aim of training an AI model. This objective changes the requirements considerably. A substantial part of the preparation was therefore devoted to planning the set-up, as well as to questions of material selection and the handling of the delicate fragments. This also involved an in-depth engagement with the fundamentals of photography, for which we were able to draw on the expertise of several colleagues.

Initial tests took place at the Cologne Papyrus Collection. The necessary equipment was hired and tested extensively in advance so that the numerous variables of the set-up could be controlled. The different ways in which the fragments are stored played a particular role: some are kept between Vinylite, some between glass and some are loose, which particularly affects reflections. In addition, the formats vary widely.

How the images are captured

At the two stations, the fragments are captured under different lighting conditions. The first uses visible light and produces colour-accurate images. The second uses infrared light, which I found particularly fascinating: ink that has faded and is barely legible to the naked eye often reappears clearly in the infrared image, revealing passages of text thought to be lost. Together, the two stations provide a standardised basis for further work.

Both stations are controlled by the software “Crocodile Capture”, which Marcel Schaeben developed specifically for the imaging. The name is a tongue-in-cheek nod to Tebtynis and its crocodile god Sobek. The software ensures that every fragment is captured under the same controlled conditions. Exposure, colour profile and file naming all follow a fixed scheme.

What we found on site

We transported all the equipment to Berkeley ourselves, which required early planning of both the transport and the customs formalities. Once on site, our first task was to test the previously developed set-up under real-world conditions. A particular circumstance arises from the fact that technical support will in future be provided from Cologne, with a nine-hour time difference to the collection in Berkeley. Quick questions therefore cannot easily be resolved, which makes reliable documentation all the more important. Alongside the set-up itself, our task was therefore to produce a detailed guide describing the entire workflow at both stations step by step. In addition, we recorded every step in videos and photographs, so that details of the set-up and handling can also be followed.

Even the most carefully planned concept has to prove itself on site. In Berkeley, requirements emerged that could not be fully anticipated in advance: the lighting conditions in the room differed from our assumptions, and the available power supply was insufficient for the lamps, so a transformer had to be obtained first. After calibrating the stations and carrying out several test runs, the workflow is now established and the first papyri are already being digitised. When everything finally worked, the colour values were right and the results were reliably reproducible, it was a special moment after such a long preparation. This lays the foundation for the gradual digital cataloguing of the collection.

At the same time, it is clear that the processes will continue to be developed. Special cases, such as fragments with unusual formats, arose earlier than expected, and the workflow is being adapted accordingly. For me personally, the visit was an instructive introduction to practical digitisation work in an international collection, and a first impression of how a research project actually takes shape on site.

Picture credit

Prianka Schmidt; unknown (photography of the excavation under the direction of British papyrologists Bernard Pyne Grenfell and Arthur Surridge Hunt in the winter of 1899/1900); Marcel Schaeben (bottom left).