Lava letters: Inside the race to virtually unwrap 2,000-year-old papyri

First, a bit on the papyri. This was originally a set of over 1,000 burned scrolls and fragments found in an underground library in the ancient Roman city of Herculaneum (near present-day Naples).

It is the only library to have survived fundamentally intact from Antiquity (the period between the 8th century BCE and 5th centuries CE, considered particularly culturally significant because it marked the rise of direct democracies, complex Western philosophy, theatre, poetry and writing, around the Mediterranean).

The scrolls were both burned and preserved when Mount Vesuvius erupted in 79 CE. The heat from the eruption sealed the papyri into the ruins of what is believed to have been the mansion of a Roman senator.

They lay there undisturbed for over 1,700 years, and were found accidentally in 1752, by well-diggers. Some scrolls are charred but retain their original structure; others survive as burned fragments.

In the late 18th century, an Italian monk spent decades unravelling more than 200 scrolls, using a machine he devised that could slowly unroll them, by 1 cm a day, using small weights attached to strings.

As late as the 1980s, researchers were using a chemical solvent in attempts to make the writing more legible, but to do this they had to first pick each scroll apart. Between these various attempts, some details emerged, but the cost was too high to continue. Several were damaged, some destroyed. Today, over 250 of the intact carbonised scrolls remain to be studied.

The details that did emerge from some of the attempts to read them suggest that the texts were found in the personal library of the Epicurean philosopher Philodemus (110 BCE to 30 BCE), at the Villa dei Papyri, a sprawling mansion owned by the Roman senator Lucius Calpurnius Piso Caesoninus (101 BCE -43 BCE), about a century before the eruption. (It might interest history buffs to know that Caesoninus’s daughter was Calpurnia, wife of the Roman emperor Julius Caesar, who lived from 100 BCE to 44 BCE.)

The texts likely belonged to the philosopher and the senator, who are believed to have been friends, with the philosopher also acting as the senator’s guide.

“Philodemus was collecting a whole series of books, most relating to the Epicurean school of thought, many written by Epicurus (the 3rd century BCE Greek philosopher). A lot of the books are by Philodemus himself, including drafts in the author’s own handwriting,” says Richard Janko, a professor of classical studies at the University of Michigan who has been studying these scrolls since 1985. “The library also contains some books from contemporary life, a few books in Latin, some of poetry. There was a speech by Seneca the Elder, of which we didn’t have anything before. There were fragments of comedies in Latin, so there were literary papyri as well as philosophical ones.”

If the remaining scrolls can be “read”, “we can sort of reach back and pull something up from that period that we otherwise don’t have,” Janko says.

Scorched bright

Now to the contest. It invites machine-learning and computer-vision experts to participate in a race to virtually unwrap two scrolls and four fragments. There are prizes up for grabs worth a total of $1million.

The Vesuvius Challenge, as it is called, was launched in March 2023, by former GitHub CEO Nat Friedman, entrepreneur Daniel Gross, and Brent Seales, a computer science professor at the University of Kentucky.

Seales is a pioneer of virtual unwrapping. For over 20 years, he’s been developing technologies to restore and read ancient historical and cultural artefacts that have been damaged with time.

In 2015, he used X-ray tomography and computer vision to “unwrap” and help scholars read the badly charred En-Gedi scroll that was found in the Dead Sea region of Israel. Dating to the 3rd or 4th century CE, it was found to contain texts from the Biblical book of Leviticus. In 2008, he was the first to discover the presence of lead in the ink used in the Herculaneum texts.

This year, Seales and his team had a breakthrough when they found that their machine-learning model was able to detect ink and some letters from the data acquired through the various scans. The model revealed Greek lettering, but reading the texts on the scrolls is a much larger challenge, given how the burned layers of papyrus have merged, morphed and frayed over time.

All the data collected over the years by Seales and his team has now been opened up to the public, in the hopes that others will be able to design digital tools that can help scholars read and study these texts.

Almost 8 TB of data, including infrared images and 3D X-ray scans of the two scrolls and four fragments, as well as detailed tutorials on the existing virtual unwrapping tech (pioneered by Seales and his team), is available on scrollprize.org.

The four winners of the first round of prizes ($2,500 each; for designing tools that can help participants virtually unravel the scrolls) were announced in April. The winners were picked from among 300 entries.

Software engineer Luke Farritor won for his 3D scroll viewer; Moshe Levy created segmentation tools to help map scrolls in segments. Brett Olsen, a computational biologist, was awarded for creating meticulous notebooks elaborating on the many tools available. Ryosuke Tanaka won for his starter notebook on ink detection.

The grand prize of $700,000 ( ₹8.26 crore; to be announced in December) will be awarded to the first team to read four passages of text from inside the scrolls.

Among the main challenges for participants is detecting black ink, which often looks exactly like carbonised papyrus in scans. That’s why the presence of lead is so important; it’s a clue to where the ink was laid. The overarching challenge, of course, will be to digitally unpeel the layers that have become enmeshed.

“What interesting about these two scrolls is that we have no idea what they are,” says Janko, who is on the Vesuvius Challenge’s panel of papyrology experts.

If the software and the scientists can rise to the challenge, there might be a lot more to unwrap. Archaeologists believe there are more collections of books waiting to be unearthed at Villa dei Papyri.

“It always seemed odd that we’ve only found Philodemus’s books. It’s inconceivable that such a cultivated and educated man as Piso, with so much wealth, would have not had more,” says Janko. “In the 18th century when early discoverers were tunnelling through 30 metres of debris, they entered a room shaped like a lecture theatre… There have to be more books.”