pith. sign in

arxiv: 2606.29085 · v1 · pith:4HIQR7OVnew · submitted 2026-06-27 · 📡 eess.IV · cs.CV· cs.LG· cs.MM· physics.ins-det

Complete virtual unwrapping and reading of a rolled Herculaneum papyrus

Giorgio Angelotti , Stephen Parsons , Federica Nicolardi , Youssef Nader , Sean Johnson , David Josey , Paul Henderson , Hendrik Schilling
show 19 more authors
Johannes Rudolph Forrest McDonald Elian Rafael Dal Pr\'a Paul Tafforeau Alessandro Mirone Clifford Seth Parker Jan Paul Posma Benjamin Kyles Claudio Vergara Alessia Lavorante Rossella Villa Maria Chiara Robustelli Marzia D'Angelo Gianluca Del Mastro Michael McOsker Kilian Fleischer Christy Chapman Nat Friedman William Brent Seales
This is my paper

Pith reviewed 2026-06-30 08:02 UTC · model grok-4.3

classification 📡 eess.IV cs.CVcs.LGcs.MMphysics.ins-det
keywords Herculaneum papyrivirtual unwrappingX-ray microtomographyancient textscomputational imagingPhilodemus
0
0 comments X

The pith

High-resolution X-ray scans and computational unrolling fully recover the text from one sealed Herculaneum papyrus scroll.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper shows that phase-contrast X-ray microtomography, paired with refined unrolling algorithms and machine learning, can extract the entire text from a rolled and carbonized ancient scroll without opening it. Previous efforts had only produced partial or surface readings from similar materials. A reader would care because the Herculaneum collection is the sole large surviving library from classical antiquity, and physical unrolling has long been too risky. The work applies the pipeline to PHerc. 1667 under explicit coverage criteria and validates ink recovery on two other scrolls, one of which yields a title and author attribution. This shifts the approach from isolated tests toward routine recovery of the unread rolls.

Core claim

Using high-resolution phase-contrast μCT acquired at the ESRF together with improved computational unrolling and machine learning, the authors achieve the complete virtual unwrapping and reading of PHerc. 1667 under explicit coverage and papyrological-review criteria, making it the first Herculaneum papyrus fully digitally unrolled and read for extended scholarly study without physical opening.

What carries the argument

The improved computational unrolling and machine learning pipeline that segments and flattens the layered tomographic volume to recover ink text from the carbonized papyrus.

If this is right

  • The same scan protocol and processing make ink directly visible in PHerc. Paris 4, enabling three-dimensional ink segmentation as an independent check on surface-based recovery.
  • The pipeline recovers title and author-attribution evidence in PHerc. 139, identifying the scroll as Philodemus, On Gods, Book 8.
  • The results establish a pathway from localized demonstrations to a scalable framework for systematic recovery of the remaining unopened Herculaneum scrolls.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If the method scales, previously inaccessible philosophical and literary texts from the Herculaneum library could become available for study without any physical risk to the originals.
  • The same layered-imaging and flattening approach might be tested on other classes of damaged rolled documents, such as certain medieval or non-Western manuscripts.
  • Automation of the coverage-criteria review step could further reduce the human time needed per scroll.

Load-bearing premise

The unrolling and machine-learning steps recover the original text layers without significant distortions or missing content, as judged by papyrological review under the stated coverage criteria.

What would settle it

A side-by-side comparison in which recovered passages from PHerc. 1667 show systematic omissions or layer-order errors when checked against any independent physical fragments or against expected textual content known from other sources.

read the original abstract

The carbonized papyri from Herculaneum preserve the only large-scale library to survive from classical antiquity, but many unopened rolls remain unread because physical opening risks irreversible damage. X-ray computed microtomography ($\mu$CT) and virtual unwrapping offer a non-invasive route to their texts, yet previous work on sealed Herculaneum scrolls has recovered only localized readings or limited surface regions. Here, using high-resolution phase-contrast $\mu$CT acquired on the BM18 beamline at the European Synchrotron Radiation Facility (ESRF), together with improved computational unrolling and machine learning, we achieve the complete virtual unwrapping and reading of PHerc. 1667 under explicit coverage and papyrological-review criteria. This makes PHerc. 1667 the first Herculaneum papyrus to be fully digitally unrolled and read for extended scholarly study without physical opening. In PHerc. Paris 4, the optimized scan protocol makes ink directly visible in the tomographic volume, allowing three-dimensional ink segmentation and independent validation of surface-conditioned ink recovery. In PHerc. 139, we recover title and author-attribution evidence identifying the scroll as Philodemus, On Gods, Book 8. These results move virtual unwrapping of the Herculaneum scrolls beyond isolated demonstrations towards a scalable framework for systematic recovery of the still-unopened library.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 2 minor

Summary. The manuscript reports the complete virtual unwrapping and reading of the rolled Herculaneum papyrus PHerc. 1667 via high-resolution phase-contrast μCT acquired at ESRF BM18, combined with improved computational unrolling and machine-learning segmentation. It claims this is the first Herculaneum scroll to be fully digitally unrolled and read for extended scholarly study without physical opening, under explicit coverage and papyrological-review criteria. Supporting demonstrations include direct ink visibility and 3D segmentation in PHerc. Paris 4 plus title/author recovery identifying PHerc. 139 as Philodemus, On Gods, Book 8.

Significance. If the accuracy and completeness claims hold under the stated criteria, the work constitutes a meaningful advance toward systematic, non-destructive recovery of the Herculaneum library. The shift from localized readings to a full-scroll result, together with the explicit coverage criteria and cross-validation via ink visibility in a second scroll, strengthens the case for scalability. The synchrotron protocol and ML pipeline are presented as reusable components.

major comments (1)
  1. [Results / Methods] Results and Methods sections: The central claim of 'complete' reading under explicit coverage and papyrological-review criteria is load-bearing, yet the manuscript provides only high-level statements of validation without reporting the numerical coverage statistics, the precise definition of the coverage criteria, or quantitative agreement metrics between the recovered text and the papyrological review. This prevents independent assessment of whether distortions or missing layers remain below the threshold for 'extended scholarly study'.
minor comments (2)
  1. [Figures] Figure captions and supplementary material: Several figures showing unrolled surfaces and segmented ink would benefit from scale bars, explicit voxel resolution values, and side-by-side comparison with any ground-truth fragments to improve clarity for readers outside the immediate imaging community.
  2. [Throughout] Notation: The distinction between 'surface-conditioned ink recovery' and the new 3D ink segmentation pipeline is not always terminologically consistent across sections; a short glossary or explicit cross-reference would reduce ambiguity.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their thoughtful review and for recognizing the significance of achieving a complete virtual unwrapping of PHerc. 1667. We address the single major comment below by committing to expanded quantitative reporting in the revised manuscript.

read point-by-point responses

  1. Referee: [Results / Methods] Results and Methods sections: The central claim of 'complete' reading under explicit coverage and papyrological-review criteria is load-bearing, yet the manuscript provides only high-level statements of validation without reporting the numerical coverage statistics, the precise definition of the coverage criteria, or quantitative agreement metrics between the recovered text and the papyrological review. This prevents independent assessment of whether distortions or missing layers remain below the threshold for 'extended scholarly study'.

    Authors: We agree that additional quantitative detail is required to support the 'complete' claim under the stated criteria. In the revised manuscript we will insert a new subsection in Results that (i) states the precise definition of coverage (fraction of total surface area for which both layers and text are recovered without prohibitive distortion), (ii) reports the numerical coverage statistics achieved for PHerc. 1667, and (iii) provides the quantitative agreement metrics (e.g., percentage of recovered text segments independently confirmed by the papyrological reviewers together with any noted discrepancies). These additions will allow readers to evaluate whether residual distortions or missing regions fall below the threshold for extended scholarly use. The Methods section will be updated with the corresponding computational definitions and validation protocol. revision: yes

Circularity Check

0 steps flagged

No significant circularity

full rationale

The paper reports an empirical achievement: high-resolution phase-contrast μCT scans of PHerc. 1667 processed with improved computational unrolling and machine learning, followed by papyrological review under explicit coverage criteria. No derivation chain, equations, fitted parameters, or self-citations are presented that reduce the central result to its own inputs by construction. The work is scoped to demonstrated recovery on specific scrolls with external validation, remaining self-contained against physical data.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

As an experimental and computational imaging paper, the abstract introduces no free parameters, mathematical axioms, or new invented entities.

pith-pipeline@v0.9.1-grok · 5905 in / 1206 out tokens · 51516 ms · 2026-06-30T08:02:16.899958+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

36 extracted references · 5 canonical work pages · 2 internal anchors

  1. [1]

    Paul Getty Museum, Los Angeles, 2005)

    Sider, D.The Library of the Villa dei Papiri at Herculaneum (J. Paul Getty Museum, Los Angeles, 2005). 9 Complete virtual unwrapping and reading of a rolled Herculaneum papyrus

    2005

  2. [2]

    Seales, W. B.et al. From damage to discovery via virtual unwrapping: reading the scroll from En-Gedi. Sci. Adv. 2, e1601247 (2016)

    2016

  3. [3]

    Parker, C. S.et al. From invisibility to readability: recovering the ink of Herculaneum. PLoS One 14, e0215775 (2019)

    2019

  4. [4]

    H., Larsen, J

    Barfod, G. H., Larsen, J. M., Lichtenberger, A. & Raja, R. Revealing text in a complexly rolled silver scroll from Jerash with computed tomography and advanced imaging software.Sci. Rep. 5, 17765 (2015)

    2015

  5. [5]

    L., Lai, Y.-K

    Liu, C., Rosin, P. L., Lai, Y.-K. & Hu, W. Robust virtual unrolling of historical parchment XMT images. IEEE Trans. Image Process. 27, 1914–1926 (2018)

    1914

  6. [6]

    Rosin, P. L. et al. Virtual recovery of content from X-ray micro-tomography scans of damaged historic scrolls.Sci. Rep. 8, 11901 (2018)

    2018

  7. [7]

    Unlocking history through automated virtual unfolding of sealed documents imaged by X-ray microtomography.Nat

    Dambrogio, J.et al. Unlocking history through automated virtual unfolding of sealed documents imaged by X-ray microtomography.Nat. Commun. 12, 1184 (2021)

    2021

  8. [8]

    & Arlazarov, V

    Kulagin, P., Polevoy, D., Chukalina, M., Nikolaev, D. & Arlazarov, V. V. Fully automatic virtual unwrapping method for documents imaged by X-ray tomography.Document Analysis and Recognition – ICDAR 2024, Lecture Notes in Computer Science 14806, 233– 250 (2024)

    2024

  9. [9]

    & Delattre, D

    Mocella, V., Brun, E., Ferrero, C. & Delattre, D. Revealing letters in rolled Herculaneum papyri by X-ray phase-contrast imaging.Nat. Commun. 6, 5895 (2015)

    2015

  10. [10]

    Virtual unrolling and deciphering of Herculaneum papyri by X-ray phase-contrast tomography.Sci

    Bukreeva, I.et al. Virtual unrolling and deciphering of Herculaneum papyri by X-ray phase-contrast tomography.Sci. Rep. 6, 27227 (2016)

    2016

  11. [11]

    Revealing metallic ink in Herculaneum papyri.Proc

    Brun, E.et al. Revealing metallic ink in Herculaneum papyri.Proc. Natl Acad. Sci. USA 113, 3751–3754 (2016)

    2016

  12. [12]

    Ancient Greek text concealed on the back of unrolled papyrus revealed through shortwave-infrared hyperspectral imaging.Sci

    Tournié, A.et al. Ancient Greek text concealed on the back of unrolled papyrus revealed through shortwave-infrared hyperspectral imaging.Sci. Adv. 5, eaav8936 (2019)

    2019

  13. [13]

    & Seales, W

    Angelotti, G., Nicolardi, F., Henderson, P. & Seales, W. B. Ink detection from surface topography of the Herculaneum papyri.Sci. Rep. 58467 (2026)

    2026

  14. [14]

    S., Chapman, C., Hayashida, M

    Parsons, S., Parker, C. S., Chapman, C., Hayashida, M. & Seales, W. B. EduceLab-Scrolls: verifiable recovery of text from Herculaneum papyri using X-ray CT.https://arxiv.org/ abs/2304.02084 (2023). Preprint athttps://doi.org/10.48550/arXiv.2304.02084

    work page doi:10.48550/arxiv.2304.02084 2023

  15. [15]

    Revealing text from a still-rolled Herculaneum papyrus scroll (PHerc

    Nicolardi, F.et al. Revealing text from a still-rolled Herculaneum papyrus scroll (PHerc. Paris. 4). Z. Papyrol. Epigr. 229, 1–13 (2024). 10 Angelotti et al

    2024

  16. [16]

    & Fowler, R

    Nicolardi, F., Delattre, D., Del Mastro, G., Janko, R. & Fowler, R. The final columns of PHerc. Paris. 4 revealed through virtual unwrapping.Cronache Ercolanesi 54, 9–27 (2024)

    2024

  17. [17]

    Vesuvius Challenge 2023 Grand Prize awarded: we can read the first scroll!https://scrollprize.org/grandprize(2024)

    Vesuvius Challenge. Vesuvius Challenge 2023 Grand Prize awarded: we can read the first scroll!https://scrollprize.org/grandprize(2024). Accessed 19 June 2026

    2023

  18. [18]

    Vesuvius Challenge – Ink Detection

    Vesuvius Challenge. Vesuvius Challenge – Ink Detection. Kaggle,https://www.kaggle. com/competitions/vesuvius-challenge-ink-detection(2023). Accessed 19 June 2026

    2023

  19. [19]

    Vesuvius Challenge – Surface Detection

    Vesuvius Challenge. Vesuvius Challenge – Surface Detection. Kaggle,https://www. kaggle.com/competitions/vesuvius-challenge-surface-detection(2025). Accessed 19 June 2026

    2025

  20. [20]

    Vesuvius Challenge. $60,000 first title prize awarded! Title of a still-rolled Hercu- laneum scroll has been read for first time ever.https://scrollprize.substack.com/p/ 60000-first-title-prize-awarded(2025). Accessed 19 June 2026

    2025

  21. [21]

    ~70% of PHerc

    Vesuvius Challenge. ~70% of PHerc. 172 is now digitally unwrapped.https://scrollprize. substack.com/p/70-of-pherc-172-is-now-digitally(2026). Accessed 19 June 2026

    2026

  22. [22]

    Eikonal phase retrieval: unleashing the potential of fourth-generation sources for enhanced propagation-based tomography on biological samples.J

    Mirone, A.et al. Eikonal phase retrieval: unleashing the potential of fourth-generation sources for enhanced propagation-based tomography on biological samples.J. Syn- chrotron Radiat. 32, 1291–1301 (2025)

    2025

  23. [23]

    & Sorkine-Hornung, O

    Rabinovich, M., Poranne, R., Panozzo, D. & Sorkine-Hornung, O. Scalable locally injective mappings.ACM Trans. Graph. 36, 16:1–16:16 (2017)

    2017

  24. [24]

    Oquab, M. et al. DINOv2: learning robust visual features without supervision.https:// arxiv.org/abs/2304.07193(2023). Preprint athttps://doi.org/10.48550/arXiv.2304.07193; published in Transactions on Machine Learning Research (2024)

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2304.07193 2023

  25. [25]

    & Marlet, R

    Moulon, P., Monasse, P., Perrot, R. & Marlet, R. OpenMVG: open multiple view geome- try.Reproducible Research in Pattern Recognition – RRPR 2016, Lecture Notes in Computer Science 10214, 60–74 (2017)

    2016

  26. [26]

    OpenMVS: open multi-view stereo reconstruction library

    Cernea, D. OpenMVS: open multi-view stereo reconstruction library. GitHub,https: //github.com/cdcseacave/openMVS. Accessed 19 June 2026

    2026

  27. [27]

    Garrido-Jurado, S., Muñoz-Salinas, R., Madrid-Cuevas, F. J. & Marín-Jiménez, M. J. Au- tomatic generation and detection of highly reliable fiducial markers under occlusion. Pattern Recognit. 47, 2280–2292 (2014)

    2014

  28. [28]

    Ravi, N. et al. SAM 2: segment anything in images and videos.https://arxiv.org/abs/ 2408.00714 (2024). Preprint athttps://doi.org/10.48550/arXiv.2408.00714; published at ICLR 2025. 11 Complete virtual unwrapping and reading of a rolled Herculaneum papyrus

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2408.00714 2024

  29. [29]

    pgs-recon: the EduceLab photogrammetry reconstruction pipeline

    EduceLab. pgs-recon: the EduceLab photogrammetry reconstruction pipeline. GitHub, https://github.com/educelab/pgs-recon. Accessed 19 June 2026

    2026

  30. [30]

    M., Gureyev, T

    Paganin, D. M., Gureyev, T. E., Pavlov, K. M., Lewis, R. A. & Kitchen, M. J. Phase retrieval using coherent imaging systems with linear transfer functions.Opt. Commun. 234, 87–105 (2004)

    2004

  31. [31]

    Vesuvius Challenge – CT scans of Herculaneum papyri

    Angelotti, G.et al. Vesuvius Challenge – CT scans of Herculaneum papyri. Vesuvius Challenge,https://scrollprize.org/data(2026). Accessed 19 June 2026

    2026

  32. [32]

    Isensee, F. et al. nnu-Net revisited: a call for rigorous validation in 3D medical im- age segmentation. https://arxiv.org/abs/2404.09556(2024). Preprint at https://doi. org/10.48550/arXiv.2404.09556; published in Medical Image Computing and Computer Assisted Intervention – MICCAI 2024, doi:10.1007/978-3-031-72114-4_47

    work page doi:10.48550/arxiv.2404.09556 2024

  33. [33]

    Skeleton Recall Loss for connectivity conserving and resource effi- cient segmentation of thin tubular structures.https://arxiv.org/abs/2404.03010(2024)

    Kirchhoff, Y.et al. Skeleton Recall Loss for connectivity conserving and resource effi- cient segmentation of thin tubular structures.https://arxiv.org/abs/2404.03010(2024). Preprint athttps://doi.org/10.48550/arXiv.2404.03010; published in Computer Vision – ECCV 2024, doi:10.1007/978-3-031-72980-5_13

    work page doi:10.48550/arxiv.2404.03010 2024

  34. [34]

    Recenti tentativi di svolgimento dei papiri ercolanesi.Cronache Ercolanesi 1, 23–39 (1971)

    Sbordone, F. Recenti tentativi di svolgimento dei papiri ercolanesi.Cronache Ercolanesi 1, 23–39 (1971)

    1971

  35. [35]

    & Angeli, A

    Capasso, M. & Angeli, A. Papiri aperti col metodo osloense (1983-1989): descrizione e classificazione.Cronache Ercolanesi 19, 265–270 (1989)

    1983

  36. [36]

    Vittorio Emanuele III

    Hara, K., Kataoka, H. & Satoh, Y. Can spatiotemporal 3D CNNs retrace the history of 2D CNNs and ImageNet? Proc. IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 6546–6555 (2018). Methods Provenance, custody and permissions The objects reported here are Herculaneum papyri presently held at the Biblioteca Nazionale di Napoli, Museo Archeolo...

    2018