Most ancient documents come in fragments: A view of the Iliad papyri as a Rosetta stone (Image: University of Michigan Library Digital Collections)

Solving Ancient Puzzles

How technology can help to reconstruct age-old manuscripts

The Papyrus Collection of the University of Basel holds a small fragment that appears insignificant at first sight: It fits the palm of a hand and only bears the first or two words of a dozen lines. The initial editor of the collection, Professor Ernst Rabel, did not consider it important enough to include it in the volume he published in 1917 and, thus, the fragment remained unnoticed — until very recently.

Ever since digital technologies have been introduced to the Humanities, the work of papyrologists has changed quite a bit. Take for instance the international database, which contains the texts of all papyri that have already been published. In the case of the Basel fragment, it took only a few keyword searches to find out that it is part of a much larger papyrus housed in the collection of the British Library in London.

And just as in a jigsaw puzzle, once the fragment was put back in its place, the text was legible again. After all, the document is a land lease — but not the type so commonly known from Late Antique Egypt. Here, the tenant is the father of the landlord, which is quite unusual; and both men are known to us from other sources, a dozen of papyri that have survived from this family of soldiers who lived in the city of Hermopolis in the 6th century CE — a family and a society that the Basel fragment now helps to better understand.

A new approach to search ancient sources

Like the Basel artifact, most of the ancient papyri reach out to us by fragments. There are tens of thousands scattered in collections around the world. To piece them back together by means of a word search is only possible when the text contains rare proper names or uses recognized terms like the lease mentioned above. Other than that, the field of papyrology has been relying on experts with a sharp “paleographic eye,” combining an immense visual memory with an ability to recognize unique patterns in handwriting. And yet, even the most recognized experts will admit that sometimes a little bit of luck is required!

The author viewing an ancient document in Strasbourg’s collection

But what if, instead of keywords, we could systematically look for shapes of letters, for similar handwriting patterns in all the papyri around the world? Imagine, a machine could be trained to develop a “paleographic eye.” In fact, that`s what new computational methods allow us to do and, in a nutshell, the goal of my Ambizione project called D-scribes, which is funded by the Swiss National Science Foundation.

Together with a computer scientist and several research assistants, I am building a program that recognizes fragments with very similar handwriting, potentially written by the same person. The challenges in developing such a program are many. For example, how to isolate the handwriting from a yellow-brownish background with holes, stains, and breaks? This is task much more complicated than even analyzing medieval parchment. Or, how to prove that the same person has written two texts, which, even in nowadays forensics, is far from being straightforward!

Digital Paleography

Left: An Iliad papyrus in quest of its missing pieces (Photo: University of Michigan Library Digital Collections). Right: The word χαλκός (copper, bronze) first as on the original, then manually binarized, and lastly reconstructed and encoded according to the local orientation of the shape outline using REX.

To overcome these difficulties, our first step was to build two data sets: one with ten papyri bearing passages of Homer’s Iliad of various dates that we manually binarized. We included this material in an international competition. Fifteen teams of computer scientists around the world submitted twenty-four algorithms. Yet, due to the difficulty of papyri, even state-of-the- art methods failed to reach high levels of performance and our research is still ongoing. Meanwhile, we have started to measure handwriting by using REX. This program uses the orientation at each pixel location to classify samples by characteristics such as slant, writing contour, density and other perceptual features.

How to classify scripts? By describing their main features by means of REX.

Writer Identification

The second data set relates to the task of writer identification. Thanks to a village archive from the 6th century CE, we could select ten notaries who have penned four to seven documents, gathering a total of fifty texts to date. We are currently testing several computerized methods of writer identification on this challenging data set.

Which one of the three notaries on the right penned also the piece on the left? Specialists can tell, can you? Can the computer program? (Images: GRK-Papyri)

The Future: More Open Access to Cultural Heritage

By the end of this four-year project, the collected data will be accessible online, providing a collection of samples by writers of various social status including notaries, merchants, monks, soldiers and others as well as a panorama of the evolution of Greek handwriting styles. The program developed in the course of this project will be available as open source tool in order to further advance the field of papyrology.

In the meantime, a key success factor will obviously be the collaboration with other teams working on ancient handwriting to pool our findings and share our expertise. To this end, an international conference entitled Neo-Paleography will take place in Basel in early 2020, where traditional paleographers and computer scientists will meet and exchange their approaches.

Besides, a mailing list has been created to bring together scholars with common interest in digital paleography, i.e., the application of computational methods to the study any ancient handwriting. Don`t hesitate to subscribe, if you would like to be informed of the latest news of this project and other exciting achievements in the field. Ancient manuscript puzzles are about to be solved!

To show your support for this post and recommend it to your followers, click on the clap icon 👏 below.

The University of Basel has an international reputation of outstanding achievements in research and teaching. Founded in 1460, the University of Basel is the oldest university in Switzerland and has a history of success going back over 550 years. Learn more

sci five | University of Basel

Inspiring projects and impact stories from the University of Basel since 2016.