Reading the Invisible Library: Rescuing the Hidden Texts of Herculaneum

Using authentic materials from the national libraries in Naples and Paris, this project will apply proven computerized techniques and innovate new technical approaches to reveal the writing hidden in the most iconic collection of damaged humanities manuscripts--the scrolls from Herculaneum. By applying our virtual unwrapping software and advanced machine learning tools to non-invasive computed tomography scans of the materials, we plan to accomplish four primary goals: (1) we will refine our computational pipeline, via a machine learning component, that will adapt it to the unique materials found in Herculaneum manuscripts--carbon ink on severely compressed papyrus; (2) we will employ this enhanced technique to create a born-digital version of the text buried underneath layers of opened Herculaneum papyri; (3) we will disseminate our results and make the software open source so that our tools can be applied to any and all types of damaged cultural objects containing hidden text; and (4) we will develop new metadata and other standards to address the unique provenance and scholarly usage issues that these types of born-digital images create. This work will set the stage for the virtual unwrapping and reading of the entire Herculaneum library, including intact scrolls.