Daejeon, July 27–31
To date, the technologies available for the study and analysis of manuscript books are both numerous and rapidly evolving. These tools have not only facilitated the examination of original medieval documents but have also generated new research questions. The widespread adoption of the IIIF protocol has made high-resolution, interoperable, and easily manipulable images accessible through an increasingly large number of institutional portals, thereby supporting scholarly research and the exchange of data among preservation institutions (Salarelli 2017, Manoni 2020, Martín Contreras and Rincón Narros 2025). At the same time, non-invasive imaging techniques are enabling remarkable progress in the study of the material composition of manuscripts: they can reveal hidden texts preserved in bookbindings, visually separate script layers in palimpsests, and virtually reunite dispersed fragments held in archives worldwide (Kessel 2022, Sargan et al. 2022, Perino et al. 2024). AI-based software for Handwritten Text Recognition (Transkribus, eScriptorium) and scribal-hand identification is likewise continuously improving – recently with a growing emphasis not only on producing results but also on offering user-friendly explanations through the integration of XAI models (De Cesaris et al. 2025).
Although research has benefitted substantially from technological innovation, teaching has not advanced at the same pace. Currently, no user-friendly applications exist that provide a comprehensive introduction to manuscript study, even though palaeographical and codicological skills remain essential for understanding, using, and further developing digital tools related to medieval sources. Moreover, as manuscripts are increasingly studied first – and sometimes exclusively – through digital copies, there is a growing need for pedagogical tools designed for digital learning environments, where high-resolution images enable forms of analysis, such as precise measurement, that are impossible in physical libraries. Addressing this gap, the project eManuSkript (Thompson et al. 2025, eManuSkript 2026, Dorofeeva et al. 2026), funded by the Stiftung Innovation in der Hochschullehre, is creating an integrated suite of browser-based applications designed to support learners and researchers in the study of medieval Latin manuscripts. Each application is accompanied by tutorials – including text, videos, and images – that introduce traditional palaeography while also modelling best practices for digital methods. Some tools concentrate on material features, while others focus on textual or scribal characteristics. By lowering the technical and epistemic barriers to manuscript analysis, the project contributes to a more inclusive engagement with cultural memory, extending access beyond a narrow circle of specialist.
Proteus is a post-processing tool developed to enhance the readability of difficult manuscript texts, such as palimpsests (erased and reused parchment) or folios affected by chemical damage. It employs multispectral imaging (MSI) techniques derived from HOKU (Knox 2022), including pseudocolour, principal component analysis (PCA), and visible-light enhancements such as sharpening, noise reduction, and contrast adjustment (Davies and Zawacki 2020). All image operations are automatically logged with metadata to ensure reproducibility. Moreover, user-supplied information about the manuscript and the imaging process is stored alongside the enhanced files, making Proteus particularly valuable for work on complex or layered manuscript materials.
Fenius addresses three interlinked areas of codicology: it documents pricking and ruling patterns; it visualises sewing on book spine; and it models the construction of a bookbinding. These three interactive and customisable interfaces have been developed in collaboration with a professional illustrator and the manuscript conservation department of the Göttingen State and University Library.
Mergen is an AI-based layout segmentation model for detecting key manuscript elements such as text zones, lines, and graphic components. Built on the YOLOv10 architecture (Wang et al. 2024), it follows a three-layer detection strategy: zone detection, instance segmentation, and line detection. Its training data combine harmonised open-source datasets such as CATMuS Medieval (Clérice et al. 2024), YALTAi (Clérice 2022), and MedieYOLOv1 (MedieYOLOv1 Dataset 2025), supplemented by custom and synthetic annotations to improve accuracy. Once completed, Mergen will enable large-scale structural analysis and support downstream tasks such as script comparison and content extraction.
Seshat is an interactive tool for script analysis and annotation designed to examine the micro-features of handwriting in digitized manuscripts. Users can upload images or IIIF links to interact directly with the script, trace pen strokes, highlight specific features, and take measurements. Seshat enables users to measure numerous handwriting features – such as stroke angles, stroke thickness, and line height – through user-defined pixel coordinates. These measurements add a structured, quantifiable layer of data to manuscript analysis, enriching both traditional palaeographic approaches and computational methods. To ensure transparency and error tracking, all annotations are time-stamped and stored in a viewable log. Users will also be able to generate statistical similarity comparisons of scribal stints to identify individual scribes.
The applications will be accompanied by tutorials designed to help users learn quickly and effectively. In addition to technical documentation for each tool, theoretical palaeographical tutorials are being developed to provide non-specialists with the competencies needed to interpret manuscript evidence correctly. They follow a learning-by-doing approach: each chapter closes with playful activities that encourage the immediate application of concepts. Non-specialist users have been involved in the design process from the outset, ensuring that the applications are accessible and appealing to a broad audience. For this reason, the interfaces adopt a simple and intuitive design aimed at producing tools genuinely usable by all.
By the end of the project, all applications will be made available through a dedicated website, while datasets and backend code will be openly archived on GitHub and Zenodo. The suite will be further supported by an integrated and searchable Zotero bibliography.
Beyond providing practical support for learners and researchers, eManuSkript contributes to the evolving field of digital palaeography by fostering interoperability, transparency, and open access. In particular, the tools open new perspectives for quantitative research in a discipline that has traditionally relied on largely subjective assessment. Through the integration of measurement-based analysis and computational methods, it becomes possible to describe specific palaeographical phenomena in more systematic and reproducible terms. By combining open digital infrastructures, transparent AI-based approaches, and pedagogically informed design, eManuSkript engages with current debates on the curation, reuse, and transmission of documentary heritage, and contributes to broader reflections on the social role of cultural memory in the digital age.