UK Computer Science Professor Working To Recover Writing From Ancient Carbonized Scrolls
In case you missed it, NPR recently did a story on an effort to recover the text contained on 2,000 year old carbonized scrolls unearthed from Herculaneum which was buried under tons of ash following the eruption of Mount Vesuvius in 79 A.D. After working more than 10 years on the project, University of Kentucky Department of Computer Science Chair Brent Seales’s next task is to build software that will visualize the writings as they would be if unrolled. He talked with WUKY's Alan Lytle to explain a bit more fully what that will entail.
From UK P.R: A breakthrough not only in digital imaging techniques, the first-of-its-kind software could also have profound impacts on history and literature. Seales says that each scroll may well be the only remaining copy as of yet unknown literature from the Classical era. Each scroll is 20 to 30 feet long, and Seales estimates each to contain at least 3,000 words.
"The sheer volume of words available for discovery is probably larger than the entire works of Shakespeare," said Seales.
The scrolls aren't your typical 2,000-year old papyri manuscripts; they were carbonized in the Mount Vesuvius volcanic eruption of A.D. 79, and later discovered as charred clumps in the Villa of the Papyri in the ancient Italian city of Herculaneum beginning in 1752. When attempting to open, the artifacts would often shatter beyond repair.
To reveal the works inside the remaining intact scrolls, Seales and his research collaborator from the Institut de France, Daniel Delattre, knew that "virtual unrolling" was the only way.
After successfully creating 2-D images of two Herculaneum scrolls in 2009 but not being able to detect the ink in them, Seales' colleagues believe they have recently identified ink in the scrolls after applying an x-ray method often used in the medical and archeology communities.
The method, called "propagation-based phase contrast imaging," was recently featured in a Nature Communications article, "Revealing letters in rolled Herculaneum papyri by X-ray phase-contrast imaging," by authors Vito Mocella, Claudio Ferrero, Emmanuel Brun and Delattre, citing Seales' work on the scrolls. Seales says the researchers claim to "see letters and, in a few instances, whole words."
Now that he and his team can see the writings, the next step in unveiling the writings to the world is to organize them. Without unrolling the scrolls, Seales' software will run extremely high-resolution images from the tangled surfaces, making sense of the jumbled letters into words, and words into passages.
"The software will combine novel methods for finding the scroll surfaces together with a user-guided interface for correcting mistakes and improving the automatic first-guess," he said.
In other words, it will pull out a page that displays writing from the data they currently have, and then identify where that page is inside the scrolls that now resemble charcoal. Because of this, Seales, his team, partners and physicists will be able to optimize the scanning process on site, allowing them to see an entire page "unwrapped" without ever leaving the facility. Eventually, the outcome will be as complete of a manuscript as possible of the remaining Herculaneum scrolls.
"We have a ton of data from all of our preliminary work, and from the 2009-2010 work. We're using that data to build software so that we can pull out large sections and flatten them," said Seales. "To date, no tool exists that can accomplish that. The software we're building will be the first to visualize data in that way, and it's crucial to uncovering the works inside the Herculaneum scrolls."
Supported by a three-year, $500,000 National Science Foundation grant and by Google, where Seales spent his sabbatical in 2012-2013, the computer science professor has begun working to develop the software. Seales' sabbatical at Google was crucial to the new imaging method, and he credits Google as the "impetus for being unstuck" in the project.
UK students are also driving the progress. The computer science professor is working on the software with a team of UK undergraduate and graduate students including:
- Melissa Shankle, a pre-computer science sophomore and member of the Honors Program from Mayfield, Kentucky;
- David Pennington, a computer science senior from Union, Kentucky;
- Michael Roup, a computer science and mathematics senior from Crestwood, Kentucky;
- Nickolas Graczyk, a computer science senior with a minor in mathematics from Lexington;
- Anastasia Kazadi, a computer science senior from Lexington;
- Abigail Coleman, a computer science graduate student from Princeton, Kentucky;
- Sean Karlage, a computer science and computer engineering graduate student from Edgewood, Kentucky; and
- Chao Du, a computer science graduate student from Beijing, China.
In addition to UK students, Seales is working with Seth Parker, video editor at the UK Center for Visualization and Virtual Environments, and collaborating with Delattre in France, as well as Roger Macfarlane, a researcher at Brigham Young University. They hope to travel back to Grenoble, France, in the spring to conduct major scans on the two scrolls scanned in 2009. The scans will utilize Seales' software, as well as the new x-ray technique.
Seales said the project plan is to release working software and datasets as soon as possible for scholars to examine.
"By project's end, the team hopes to have created a software tool and a set of scans of scrolls that together will transform the hopelessly damaged Herculaneum collection into new literary discoveries," he said.
Unmasking the Herculaneum writings is only the beginning. Seales hopes the work to uncover and decipher these ancient scrolls will propel other efforts forward, leading to an even greater impact on our understanding of classical history and literature, and revolutionary digital imaging.
"I dream of seeing renewed excavation at the Villa of the Papyri. Many believe that a treasure trove of undiscovered scrolls are waiting there to be unearthed. If more are found, these methods could be used to read them," said Seales.