Skip to main content


What are the important dates?

I would like to contribute something, but I don’t have the time to compete for the Grand Prize, what should I do?

  • Consider competing in the Ink Detection Progress Prize on Kaggle. It is a lot smaller in scope, and will mark an important milestone towards the Grand Prize. There are 10 prizes available, for a total of $100,000.
  • The First Letters Prize is similar to the Grand Prize, but much smaller in scope.
  • You can make smaller open source contributions, which would benefit the whole community. Everyone in the community will be grateful for your work, and you might even be able to win a Segmentation Tooling Prize!

Can I share my progress on social media?

Yes, in fact we encourage you to share your progress. Be sure to also post in our Discord, to get feedback from the community.

The only exception is that per the data agreement, you’re not allowed to share material revelation of text (e.g. entire words) without our permission (including the associated code), or share the raw data.

I’m outside the United States, can I participate and win prizes?

Absolutely! As long as we can legally pay you (no US sanctions) you can win prizes.

Do I have to pay taxes on my prize earnings?

This depends on the jurisdiction you live in, but generally yes, you do have to pay taxes. Consult your tax advisor.

I’m a researcher or student, and I would like work on this. Can I publish my results in journals?

Generally yes, with the conditions that are specified in the Data Agreement:
  • Any publications and presentations must cite the EduceLab-Scrolls Dataset.
  • You won’t publish any revelation of hidden text (or associated code) without the written approval of Vesuvius Challenge.
  • If you find enough hidden text, you’ll win the Grand Prize ($700,000), and we'll work with you to put the texts in historical context, and co-publish them in academic venues. The winning code will be made public under a permissive open source license, so that others can reproduce and build on your work.

We very much encourage researchers and students to work on this! Be sure to reach out to us on Discord or by email.

I have made some progress, who do I inform about this?

If you want to share your work privately with the contest organizers, please email us at [email protected] We will keep it completely confidential. We really appreciate you keeping us in the loop!

If you're open to sharing your improvements publicly (and be eligible for the Segmentation Tooling Prize), you can post in Discord.

What are the key academic papers I should read to understand the work done so far to read the Herculaneum Papyri?

For a comprehensive overview of the field, see this list by Dr. Seales' lab.

What are the best talks that have been given on this work?

The Getty Villa in California, a reproduction of the Villa of the Papyri

What are some good books that I should read to learn more?

The best book we have found is David Sider’s The Library of the Villa dei Papiri at Herculaneum

Here are some other excellent books we recommend:

Herculaneum scroll reconstruction from a book by Marzia D’Angelo (source)

What are some good YouTube videos I should watch?

Do we really need 8µm resolution? These data files are huge!

We are releasing about 5.5TB worth of 3D volumetric data of the scrolls (8µm), and 1.8TB worth of data of the fragments (4µm). We don't know what the minimum resolution necessary to detect ink is, but this paper suggests that it may be 8µm: From invisibility to readability: Recovering the ink of Herculaneum.

If an algorithm can read ink from a fragment X-ray, is it likely to work on a scroll?

We hope so, but we suspect this will be a key challenge.

We’re organizing the Ink Detection Progress Prize, to find the best possible approach for detecting ink on fragments. With this solid foundation we hope it will be easier to apply these algorithms to the scrolls.

Can machine learning models hallucinate letters that aren't there?

This is a risk for models that are trained on letterforms. We strongly recommend that participants guard against the risk of hallucination in their models and, will review all submissions with this in mind.

What is papyrus and how is it made?

Papyrus is a grassy reed that grows along the banks of the Nile in Egypt. It can grow up to 4.5 meters tall and 7.5cm thick. The tough outer rind is peeled away. The green inner pith is peeled or sliced into strips.

The strips are laid out in two layers in a grid pattern. They are pressed together until the layers merge like velcro. And then left out in the sun to dry, where they turn light brown. The sheets – called kollemata – are smoothed out with an ivory or seashell ruler. The kollemata are then glued together with paste made of flour and water. Then the areas where they are joined are hammered smooth. This forms a long piece of papyrus, usually 15-30 feet, comprised of up to 20 kollemata.

The Papyrus is rolled up around a dowel called an umbilicus. Portions of it are unrolled for writing. The first section, called the protokollon, is usually left blank. Text is written in columns with a quill and inkwell. Inks are made of varied substances.

There are two ways you can write on papyrus: horizontally (“volumen”) or vertically (“rotulus”). All of the Herculaneum papyri are horizontal scrolls.

Horizontal and vertical scrolls (source)
For some good videos about how papyrus is made see:

How big are the letters, and where can we expect to find text?

Letter sizes vary, and of course we don’t know what’s inside the unopened scrolls, but we expect the opened fragments to be fairly representative. You can measure how big the letters are by looking at the aligned surface images, which have a voxel resolution of approximately 4µm, like the original CT data (though there can be some local variation due to the registration / flattening process). So you could open, for example, fragments/Frag1.volpkg/working/54keV_exposed_surface/ir.png, measure a letter size in pixels, and multiply by 4µm.

There are also some measurements in this paper by Richard Janko, though it’s a little hard to infer actual letter sizing from it. If someone wants to do a more thorough review of the range of letter sizes found in all the Herculaneum papyri, we’d happily include your results here!

From our first Q&A with the Papyrology Team, we learned (summary courtesey of Santiago Pelufo on Discord):

  • 2 orthogonal layers of fibers in a sheet.
  • ~100um sheet thickness
  • Scroll outer layer of a sheet = back of a sheet = vertical fibers.
  • Scroll inner layer of a sheet = front of a sheet with writing = horizontal fibers (text written here).
  • Sheet layers unlikely to delaminate with carbonization
  • Carbonization likely fuses multiple sheets (big issue IMO)
  • 10-20cm blank at start and end of scroll.
  • 4.5-6cm columns, 1/6 of column space between columns
  • ~1/8 height paddings top and bottom
  • Text typically written on the inside (to protect against damage), and on the side with horizontal fibers (easier to write on).

How can we get more ground truth data? Can I make my own carbonized scrolls?

Yes! Just buy this papyrus on Amazon, roll it up, and put it inside a Dutch oven at 500F+ (260C+) for a few hours.

This is very instructive and we highly recommend doing it! You will see how fragile the charred scroll is, how it blisters in the heat, how the layers can separate, how it turns to dust as you handle it.

Of course, for it to be useful as ground truth data, you will need to find someone to let you image it in their CT scanner.

What software is available currently that might help me?

The two main pieces of software developed by Dr. Seales’ lab are Volume Cartographer and ink-id. Both are open source and available on Github.

In the tutorials we also show you how to use generic software to work with 3D volumes (Fiji) and meshes (MeshLab).

Where can I find collaborators?

What would the papyrus scrolls look like when unrolled?

Something like this:

Reconstructed scroll (source)

Why are there no spaces in the text?

In ancient Latin and Greek, they didn’t use spaces! Spaces were added later to aid foreign language learners.

How does CT-scanning work exactly?

We take X-ray photographs of the object from different angles. Typically this is done by having an X-ray source on one side of the object, and an X-ray camera on the other side, and rotating the object on a platform. If the object doesn’t fully fit in the frame of the camera, it can be moved around as well.

Just like with any digital camera, there are a lot of settings and parameters. The most important for you to know are:
  • Resolution: the dimensions of each pixel in an X-ray photo, typically denoted in µm (micrometers or “microns”). Lower is better. We scanned the scrolls at 8µm, which we think should be enough to detect ink patterns, but we scanned the fragments at 4µm just in case. Renting beam time on a particle accelerator is expensive, but if we need to we can go back and scan objects at even lower resolutions.
  • Energy level: the energy of the X-ray electrons, typically expressed in keV (kiloelectronvolts). For particle accelerators this is one precise number, whereas for bench top scanners this is more of a range. We think lower is better, since carbon responds better to lower energy levels. We scanned everything twice, at 54keV and 88keV (though for the scrolls we only had time for a smaller slice at 88keV).

At high resolutions the field of view of the camera is too small to capture the object in its entirety, so multiple passes have to be made. Typically these are stitched together as part of the scanning process.

Raw X-ray photos
A fragment rotating (source)

From the X-ray photos from different angles we can reconstruct a 3D volume, using a clever algorithm called tomographic reconstruction (which is where “CT scanner” gets its name; ”computed tomography”). This is typically done by software that comes with the scanner.

Volumetric representation of a fragment, showing multiple layers of papyrus
Mesh representation of the same fragment

The resulting 3D volume is like a 3D image. Each unit is called a “voxel” (instead of “pixel”), and has a particular brightness (it’s greyscale). This 3D volume is typically represented as a “.tif image stack”. This is just a bunch of .tif images where each image (called a “slice”) represents a different layer the z-direction, typically starting at the bottom and moving upwards.

What signals might be present in the 3D X-ray scans for ink detection?

We don’t really know, but we suspect that ink might be filling in between the grid pattern of papyrus, kind of like syrup filling in gaps in a waffle.

Syrup filling in gaps in a waffle (source)

Ink might also be sitting on top of the papyrus, causing a slight bump on the surface. In Tutorial 4 we should several examples of where the ink is directly visible in slices of 3D X-ray scans, which is promising. The talks at the top of this page also go into some details.

There might be some effect of indentation of the writing instrument, but it’s probably not very significant. The thought has generally been that any indentation effect would be even smaller than ink w.r.t. the scan resolution and maybe not significant when compared against the natural relief of the papyrus fibers. However, this has not been explored in detail on this type of material (look at the paper "Revisiting the Jerash Silver Scroll" for work on an etched metal scroll), so we don’t know for sure.

It could be worthwhile to try to reverse engineer what machine learning models are seeing, so that perhaps we can see it more directly. Perhaps this could influence other ink detection methods or future scanning efforts.

Does segmenting and flattening need to happen before ink detection?

This ordering is largely historical and due to the way we’ve constructed label sets, which relies on doing the segmentation and flattening first. But this can’t be the only way to do it, and we’d love to see the pipeline get shaken up.

For example, the model input of ink detection could be sampled directly from the original 3D X-ray volume, instead of using a “surface volume” as an intermediate step. This could avoid loss of resolution during the sampling process into a differently oriented volume, which happens when constructing a surface volume.

The downside of such an approach is that a lot more data needs to be accessible on disk, since the original 3D X-ray volumes are much bigger than the surface volumes (37GB vs 1.6TB in total for all fragments). This can be problematic for cloud training, which might not have enough available hard drive space. However, since we only need to access the voxels around the mesh, the data size could be reduced (creating something like a surface volume, but retaining the original coordinate space, and avoiding any resampling).

Fiji/ImageJ crashes, what can I do about that?

Fiji/ImageJ doesn’t work well with extremely large datasets such as our scrolls or fragment volumes, though downsampling might help. If you’re experiencing problems even with the dataset, then try to increase the memory limit: “Edit > Options > Memory and Threads”. It might also help to run the software in a different operating system, such as in a Linux VM. For example, on Windows the following setup seems to work well: WSL2, Ubuntu 20, Windows 11, using the default WSL X server setup.

A great contribution to the community would be to build an open source 3D volume viewer that is tailored to this problem. If you are interested in building something like that, do let us know in Discord!

How can I run non-Python/R code in Kaggle notebooks?

Will from Kaggle says: if the code can be run on Linux, you can upload it as a dataset and call it from within Python, or using Juptyer's magic command !.

What are the triangle artifacts in the surface volumes?

There are triangle artifacts in the surface volumes, from the way the original volume is sampled using the mesh to create the surface volume. The triangles likely do correspond to mesh triangles. They don’t typically show up so distinctly, so we guess the mess geometry is “interesting” in this area.

I would like to read the works that have been recovered from the scrolls so far, where I can I find them?

Most are by Philodemus. This is a list of English translations we have found so far:

What happened to the people when Mount Vesuvius erupted? 😢

We recommend starting with the only surviving eyewitness account: Pliny the Younger, Letters 6.16 and 6.20.

The story of the eruption of Mount Vesuvius has captured imaginations for centuries. The cities of Pompeii and Herculaneum are unique in how well they were preserved. A great introduction to this story is A Timeline of Pompeii.

Map showing Mt. Vesuvius in relation to Herculaneum, Pompeii, and Misenum (where Pliny the Younger witnessed the event) (source)
Some more resources:

Why did you decide to start this project?

Nat read 24 Hours in Ancient Rome during the 2020 COVID lockdown. He fell into an internet rabbit hole that ended up with him reaching out to Dr. Seales two years later to see how he could help speed up the reading of the Herculaneum Papyri. They came up with the idea of the Vesuvius Challenge. Daniel was intrigued by this idea and decided to co-sponsor it with Nat.

Is this going to work?

We think so! Based on the results that Dr. Seales and his team have been able to produce so far, we believe that it is possible to read the Herculnaeum scrolls using the scans that we already have. We don’t think it’s easy, and we’re not certain, but we believe it’s possible.

I have a lot of money! Can I help sponsor this?

Vesuvius Challenge Inc. is a 501c3 non-profit organization that was formed solely to solve the puzzle of the Herculaneum Papyri. It is currently funded by the sponsors listed on the homepage, and by many hours of volunteer contributions.

If you want to contribute money to support our operational costs or to increase the prize amounts, please get in touch!

Has the mainstream media covered this work in the past?

I’m a journalist and I would like to interview someone from the Vesuvius Challenge!

Ok! Please email [email protected].

Do you have a scroll that looks like the Nintendo logo from GoldenEye N64?

Of course (🔊 sound on).

Do you have a super-cringe, over-the-top, and factually questionable trailer video for the competition?