Hidden Library: How Science Is Virtually Unwrapping the Charred Scrolls of Herculaneum

University of Kentucky/Brent Seales
University of Kentucky/Brent Seales

Brent Seales called them Fat Bastard and Banana Boy. They were two charred, highly fragile relics that had survived the Mount Vesuvius volcanic eruption of 79 CE, which doused residents of Pompeii and neighboring Herculaneum in a searing blast of destructive gas and volcanic matter. Herculaneum was buried under 80 feet of ash that eventually became solid rock.

Entombed for centuries, the city was rediscovered in the mid-1700s. Incredibly, the library of Herculaneum (known as the Villa dei Papiri) was still filled with over 1800 scrolls, solidified into dark husks. The words inside—religious text, scientific observation, poetry—could provide unprecedented insight into human history. Yet unraveling them has proved difficult. The papyri are so damaged and rigid from lack of moisture that they suffer from a kind of archaeological rigor mortis. And unlike the paralysis that seizes the body upon death, this condition is permanent. Delicate attempts to open the scrolls by hand have been destructive. For a long time, it seemed as if the secrets of the texts would remain locked away for good.

But as Seales stared at the two hardened masses in front of him in 2009, he didn’t share that pessimism. A professor of computer science at the University of Kentucky, he believed that the manual unwrapping that had long failed could be replaced by virtual unwrapping—the digital opening of the texts using computer tomography (CT) scanning and software to penetrate inside the rolled-up scrolls, revealing layers once thought invisible to the eye.

“It’s the only library from antiquity that we have,” Seales tells Mental Floss. “All the knowledge that seems lost, your imagination can run wild.”


Seales first grew curious about the role of digital manipulation in 1995, when he was invited to assist the British Library in London in scanning and preserving Beowulf. Its 1000-year-old pages had been damaged by fire and warped by the passage of time, imperfections that 2D scans left intact. The use of special software and a 3D visualization, Seales realized, could make it possible to actually flatten the pages and restore smeared copy.

The idea of capturing and manipulating visual data came from Seales's experience in medical imaging, where CT scans can peer inside the body in a noninvasive manner. What if, Seales wondered, the same principle could be applied to the study of fragile documents? What if a relic could be examined in the way a radiologist can visualize, say, the lungs? "That was the eureka moment," he says.

A digital CT scan of a damaged scroll that is being reconstructed
A CT scan of a damaged scroll, with layers visible (L). The red outline is digitally reconstructed in a process called "segmentation" (R).

Seales believed he could use these diagnostic tools to virtually rebuild manuscripts, and returned to the British Library in 2000 to examine other warped documents. After taking images using a prototype of a machine that achieved 3D scans without physical contact, he wrote software that smoothed out the buckled and bunched pages. He likens it to a computer mimicking the tug of gravity, or reversing the direction of a billowing flag. The technique worked—he was able to achieve realistic, flat versions of centuries-old damaged pages.

But Seales believed he could set his ambitions higher: to not only virtually repair a damaged page, but peer inside the Herculaneum scrolls without the risk of causing additional harm. Like many scholars before him, the allure of Herculaneum's vast repository of knowledge had captured his curiosity.

However, the idea of subjecting the scrolls to even minimal handling was something few would consider. Only the Institut de France—one of four major holders of the scrolls—would entertain the idea, and it took four long years to convince them of the possibilities. In 2009, they finally granted permission to Seales's team to scan two Herculaneum scrolls they had in their possession. Officially, the scrolls were categorized as P.Herc.Paris 3 and P.Herc.Paris 4. Seales nicknamed them Fat Bastard and Banana Boy.

The easiest way to imagine the first part of his process is to visualize a sheet of dough that is covered with small red letters and then rolled up. Seen from its edges, the wrap displays its layers and colored pieces, though no observer could possibly identify sentences from that perspective. By slicing the roll into cross-sections as small as 14 microns thick (human hairs are around 75 microns) in a process known as volumetric scanning, Seales can then use geometric "mesh" to reassemble them into a readable surface, depicting the paper so it appears to be as flat as the day it was first written on.

In 2009, the technique allowed Seales to peer inside a closed Herculaneum scroll for the first time, revealing a fibrous labyrinth of data that initially looked like coiled string.

“We saw this amazing structure,” Seales says. But that's where things went wrong.

Seales had believed that trace metals commonly found in the ink of the period could be isolated by the imaging, separating them from the page once the scroll was unraveled and rendering the script legible. But so little of the metals were present that it didn't allow him to identify letters. Nor could Seales distinguish the carbon in the papyrus from the carbon in the ink, which rendered them indistinguishable from one another. The software also wasn't prepared to process the terabytes of data from the scan. While he technically had been able to look inside the scrolls, there was no functional way to determine what he was seeing.

Over the next several years, “Seales Stymied” became something of a headline in academic circles. That ignored the larger point: Seales had proven it was possible to retrieve images from inside the Herculaneum scrolls. It was now a matter of how best to visualize and process it.


The Herculaneum scans pushed Seales and his team to renovate their software, an act made easier by Seales’s sabbatical work as a visiting scientist at Google’s Cultural Institute in 2012 and 2013. “The interns helped me with the algorithms,” he says, which was a major perk of working for one of the world’s most concentrated and talented assembly of programmers.

His software was vastly improved by the time Seales was approached in 2014 by Pnina Shor, the curator of the Dead Sea Scrolls Project at the Israel Antiquities Authority. Shor had heard of Seales’s work and wanted to know if he could take a look at some CT scan data she had gathered from a 3-inch stick of parchment found in En-Gedi, Israel, in 1970. There was probably ink, but it was obscured by the folds and twists of the parchment.

A CT scan of the En-Gedi scroll, along with a virtual example of how it might look unfolded
The En-Gedi scroll's layers are tightly wound (L). Special software is able to isolate one layer to look for text (R).

Seales looked at the scans and applied his process for virtual unwrapping. He used a step he called "texturing," which identifies density differences and other data on the paper that indicate where ink has been applied and assigns a value to that point. Logging the information on individual voxels—the 3D equivalent of pixels—he's able to reassemble them so they appear as a familiar letter shape. The data is then flattened so it resembles an unrolled sheet.

The En-Gedi scroll was made from animal skin, which Seales says is better for contrast against the ink than papyrus, and also benefited from resolution that was twice as good as what he used in 2009. He sent his findings to Shore in 2015; she wrote him back an email humming with excitement. Seales didn’t know what he had uncovered—he doesn’t read Hebrew—but Shor did: It was the first two chapters of the Book of Leviticus, the earliest example of Bible text after the Dead Sea Scrolls themselves.

“When we saw the results we almost fainted,” Shor told reporters. “We had been certain it was just a shot in the dark.”

The fully unwrapped En-Gedi scroll with writing visible
The fully unwrapped En-Gedi scroll revealed writing that had not been seen in centuries.

Shor’s willingness to embrace new technology helped reveal text locked away for centuries. Conservators are notoriously cautious when it comes to handling such delicate relics—even though Seales never touches one personally, since curators are responsible for getting scrolls in and out of CT scanners. Only recently has Seales been able have more productive conversations at the Officina dei Papiri at the National Library of Naples in Italy, where the bulk of the Herculaneum scrolls are kept, and the University of Oxford. (The Institut de France and the British Library also hold Herculaneum scrolls.)

He remains optimistic that the method used for the En-Gedi material will work for the Herculaneum collection. At a conference this past March, he and members of his team presented new findings showing success in determining the column structure of one text (17 characters per line), as well as reading specific letters—and even entire names. Part of the breakthrough comes from high-powered x-ray beams like the one housed at Diamond Light Source in the UK, which are proving potent enough to isolate the trace amounts of lead in the ink.


The progress can seem glacial, but Seales has nonetheless gone from imaging a wrapped papyrus to isolating a clearly defined letter. Next, he hopes, will come sentences, possibly isolated by artificial intelligence software he's currently writing.

But even with permission, Seales’s pursuit of a viewable Herculaneum fragment is still dependent on funding. “I sometimes cringe when I see people say, ‘Seales has been working on this for two decades, unable to figure out the problem,’” he says. “Funding comes and goes.” Commercial applications for his software and methodology—like bone scanning or even virtual colonoscopy—could one day underwrite the academic work.

With access, cooperation, and a little luck, he remains optimistic we’ll eventually be able to uncover the knowledge long buried by Mount Vesuvius—time capsules that are slowly revealing their secrets, one micron at a time.

All images courtesy of University of Kentucky/Brent Seales.

The Best Place to Park at the Mall, According to Science

Diy13/iStock via Getty Images Plus
Diy13/iStock via Getty Images Plus

It’s Black Friday, and you are entering the battlefield: a mall parking lot. You’re determined to nail that doorbuster deal, and quantities are limited. The field is already full of other combatants. You must find the perfect parking spot.

Do you grab the first one you see, or drive as close to the mall as you can and hover? Or, do you choose a tactic that lies somewhere between?

Parking at the mall has long frustrated drivers and taxed the minds of traffic engineers—but after working on the problem for three years, physicists Sidney Redner of the Santa Fe Institute and Paul Krapivsky of Boston University have gotten closer to a winning strategy. “There are lots of studies of parking lots, but it’s just that they’re so complicated, you don’t get any insight into what’s actually happening,” Redner tells Mental Floss.

Redner and Krapivsky, whose work employs statistical physics to make sense of large systems, simplified the messy dynamics of a parking lot by modeling it with a one-dimensional grid of cells, each representing a parking space. They tested three simple, yet realistic, parking strategies using basic probability theory. Their model tested the following strategies to see which one resulted in least time spent walking and driving in the parking lot:

Meek Strategy: Meek drivers park in the first open space they see, however distant it is from the mall. As a result, they often spend the most time walking to and from the mall.

Prudent Strategy: Prudent drivers look for the first open spot but then keep driving toward the mall. They continue to drive until they see a parked car and then park in the best open spot between that first open spot and that first parked car. There may be a block of open spaces between the first open space and the first parked car. From that block of open spaces, they choose the one closest to the mall.

Optimistic Strategy: Optimistic drivers drive as close to the mall as possible and look for a parking space close to the entrance. If they see one, they grab it. If there are none, they backtrack and choose the first open space they see. Optimistic drivers probably spend the most time driving and the least time walking. In the worst-case scenario, they end up parking back where a meek driver would have parked.

Naturally cautious drivers are more likely to default to the meek mode, while aggressive drivers often use the optimistic strategy, well, aggressively. And most drivers have tried something like the prudent method.

So, which is your best bet in a crowded mall parking lot this holiday season?

In the experiments, the prudent strategy fared best, followed closely by the optimistic strategy. The meek strategy finished a distant third (“It’s hard to comprehend just how bad it is,” says Krapivsky, a self-described meek driver).

And even better: The more crowded the lot, the better the prudent strategy works, he adds.

One clear takeaway from the study is that meek drivers may want to ramp up their parking skills before going to the mall. “You don't want to park on the very outskirts of the lot, like a mile away from the stores. You want to go to the first place there’s an open spot and park somewhere in that first open area,” Redner says. They published their findings in the Journal of Statistical Mechanics [PDF].

The researchers say this is the best of the strategies they tested, but it has its limitations. It does not take into consideration competition among a sea of drivers all looking for parking spaces at the same time, and it doesn’t include (perhaps optimistically) the psychological aspects of operating a vehicle. “We are not rational when we are driving,” Krapivsky tells Mental Floss.

The researchers’ one-dimensional grid model also assumed that there would be one car at a time entering the lot through one entrance, unlike messier lots in the real world, where many cars enter from a multitude of entrances.

The optimal parking strategy, one that would best all others every time, has yet to be found. In their research, though, Redner and Krapivsky are homing in on one that integrates the more complicated aspects of parking.

For now, science says prudence is a virtue in the parking lot. And while the meek might inherit the Earth, they certainly won’t find the best parking space at the mall.

'Lost Species' of Tiny, Rabbit-Sized Deer Photographed in Vietnam for the First Time in 30 Years

Global Wildlife Conservation
Global Wildlife Conservation

The silver-backed chevrotain, also called the Vietnamese mouse-deer, is elusive. It's so elusive that scientists had feared it was extinct after none had been photographed for decades. But as The Washington Post reports, the first images taken of the mammal in nearly 30 years prove that the species is still alive in the woods of Vietnam.

No larger than small dogs, chevrotains are the tiniest ungulates, or hoofed animals, on Earth. They have vampire-like fangs and skinny legs that support their bodies. Silver-backed chevrotains are characterized by the silver sheen of their tawny coat.

The tiny population native to Vietnam has been devastated by poachers in recent decades. That, and the animal's natural shyness, make it incredibly difficult to study. Before this most recent sighting, the last time scientists had recorded one was in 1990.

Global Wildlife Conservation, the Southern Institute of Ecology, and the Leibniz Institute for Zoo and Wildlife Research teamed up in hopes of documenting the lost species. Researchers interviewed residents and government forest rangers in the Vietnamese city of Nha Trang about the silver-backed chevrotain, looking for tips on where to find one. Residents said that while populations had been hit hard by hunting, the animals were still around.

Based on this local ecological knowledge, scientists set up three camera traps in the Vietnamese woods. In just five months, they captured 275 photographs of the little mouse-deer. They then installed 29 additional cameras and snapped 1881 new images in that same length of time.

“For so long this species has seemingly only existed as part of our imagination," Global Wildlife Conservation associate conservation scientist An Nguyen said in a statement. "Discovering that it is, indeed, still out there, is the first step in ensuring we don’t lose it again, and we’re moving quickly now to figure out how best to protect it.”

Now that a silver-backed chevrotain population has been located, researchers plan to conduct the first-ever comprehensive survey of the species. Once the data is collected, it will be used to build a plan for the species' survival.

[h/t The Washington Post]