This paper proposes an XML-based Edition Publication Model (EPM). While there now exists a number of mature tools for publishing TEI-encoded texts, such as TEI Publisher, Edition Visualisation Technology and CETEIcean, the paper will argue there is room for a software-independent specification of the functionality of digital editions. Just as the TEI itself was among other things an attempt to move away from software-specific encoding of textual features (Ide & Sperberg-McQueen 1995), an Edition Publication Model should be able to describe how the edition data (TEI XML files) should be displayed and what interactional features this display should offer, without assuming a certain software context. The paper will discuss (1) the reasons for this proposal, (2) the relation to similar initiatives and existing tools, (3) some aims that an EPM should fulfil and (4) a first schema and implementation of an EPM. The aim of the presentation is not just to inform the public, but also to attract potential collaborators. Ad 1. Briefly: all software is temporary and eventually dies. Data that conforms to a public specification is likely to outlive the software that handles it. If an edition’s interface dies but the definition of the interface is still available, it is easier to recreate that interface in another tool. Besides, if there is a public specification of edition functionality, it becomes easier to write software that can display multiple editions out of the box, rather than the custom-written software that we often see today. Ad 2. While there is a longer history of related initiatives, the ones that the paper will highlight are the Ediarum manifest (Fechner 2018) and the TEI Processing Model (Turska, Cummings & Rahtz 2016). Ediarum manifests provide a high-level definition of some aspects of the edition but leave other aspects to the implementation and the details of processing to XSLT stylesheets. The TEI Processing Model describes a set of edition behaviours and provides tools to map XML structures to these behaviours. The paper will argue an extension of the TEI Processing Model is an essential ingredient for an EPM (Boot 2024). Ad 3. The EPM should facilitate a complete specification of the content and behaviour of an edition. As new edition functionalities continue to appear (think of Pure3D[1] or editions that integrate network displays)[2], EPM should be extensible. It should also be modular: it should be possible to model the several components that an edition can contain (say charters, personographies, introductions) as well as the interaction between these components. Finally it should be chainable: it should be possible to override behaviour mappings defined at a general level with other mappings useful in specific situations. Ad 4. Finally, the paper will describe a first iteration of a schema and an XSLT-based implementation for an EPM.[3] This first iteration only aims to describe data display, and is not concerned with interaction or searching. The discussion will present a number of design issues for an EPM. Literature Boot, P. (2024). The TEI Processing Model: Introduction, limitations and potential extensions. In Declarative Amsterdam 2024. https://doi.org/10.1075/da.2024.boot.tei-processing-model. Fechner, M. A (2018). Standardized Interface for Digital Scholarly Editions. DHd 2018. https://edoc.bbaw.de/frontdoor/index/index/docId/3327. Ide, N. M., & Sperberg-McQueen, C. M. (1995). The TEI: History, goals, and future. Computers and the Humanities, 29, 5-15. Turska, M., Cummings, J., & Rahtz, S. (2016). Challenging the myth of presentation in digital editions. Journal of the Text Encoding Initiative, (9). [1] See e.g. https://editions.pure3d.eu/project/14/edition/1/. [2] See e.g. https://db.innovatingknowledge.nl/edition/. [3] See https://gitlab.huc.knaw.nl/edition-publication-model/edition-publication-model.