In 1752, workmen digging in a well shaft at a site near Naples pulled up something that looked like a lump of charcoal. It was a scroll. There were hundreds more down there.
The site was Herculaneum, a Roman town buried by the eruption of Vesuvius in 79 AD. The well was inside a large private house now known as the Villa of the Papyri, generally identified as belonging to Lucius Calpurnius Piso Caesoninus, the father-in-law of Julius Caesar. When the eruption struck, a surge of hot gas and ash moved through the town at a temperature high enough to carbonise wood and paper. The scrolls in that library did not burn away. They were turned into black, brittle cylinders, each roughly the size of a fist.
Nobody could open them. The carbonised papyrus had the consistency of a burnt match. Any attempt to unroll one produced fragments. Early excavators still tried. In the 1750s a Vatican priest named Antonio Piaggio built a machine that suspended a scroll and used fine strings and small weights to peel the layers apart over weeks or months, sometimes succeeding, often destroying the scroll in the process. Some scrolls were opened, copied by hand, and then crumbled. Others were abandoned as unopenable and stored. That was the situation for the next two hundred and fifty years: roughly 1,800 scrolls, sitting in museums, known to contain text, and unreadable.
Most of the library has still never been read. It is the only large library to survive from the ancient world, and the reason is that it was destroyed.
The change came from a direction nobody in the eighteenth century could have imagined. Brent Seales, a computer scientist at the University of Kentucky, developed a method he calls virtual unwrapping: scan a rolled or folded document with high-resolution X-ray tomography, build a three-dimensional model of its internal surface, and then mathematically flatten the surface and look for ink. In 2015 his team used it to read a charred scroll from a synagogue at En-Gedi, written in Hebrew, without opening it.
In 2019, Seales took Herculaneum scrolls to Diamond Light Source in the United Kingdom, a particle accelerator that produces an extremely intense X-ray beam, and scanned them at a resolution fine enough to record the texture of the papyrus surface.
In 2023, Seales launched the Vesuvius Challenge with the technology investors Nat Friedman and Daniel Gross, offering a prize pool of about one million dollars to anyone who could extract text from the scans. The organisers published the data and let strangers work on it.
In October 2023, a 21-year-old computer science student named Luke Farritor, then interning at SpaceX, trained a machine learning model to detect a subtle "crackle" pattern left by ink on the papyrus surface. He read one word: porphyras, Greek for purple. He was awarded a 40,000 dollar progress prize. Two weeks later, Youssef Nader, a biorobotics doctoral student in Berlin, independently found the same word using a different approach.
On 5 February 2024, the grand prize of 700,000 dollars went to Farritor, Nader, and Julian Schilliger, a Swiss robotics student. Their entry recovered more than 2,000 characters from about fifteen columns of a scroll catalogued as PHerc. Paris 4 and held by the Institut de France — roughly five percent of the text. The challenge had required four passages of at least 140 characters each, with 85 percent of the characters recoverable. They cleared the bar and then kept going.
The text is Greek philosophy. The author is generally identified as Philodemus, an Epicurean philosopher in residence at the villa. It discusses pleasure: music, food, how to enjoy life, and whether scarce things give more pleasure than abundant ones. At one point it appears to mock the Stoics for having nothing to say about pleasure.
There is one more thing about the villa. It has only been partly excavated, because the modern town of Ercolano sits directly on top of it. Digging out the rest would mean moving a residential neighbourhood. So the largest unread library in the world is, for now, under a town.
What the scrolls teach has less to do with whether we are clever than with what it takes for a message to survive. These documents lasted two thousand years because they were destroyed — fully carbonised, sealed against air and water. The scrolls that merely got wet or scattered are gone. What survived was the most violently damaged version.
And they became readable only because a small number of people decided to record the object before anyone could extract meaning from it. The X-ray scans are, in effect, a physical description of a scroll stored as data. The text came later, interpreted by strangers, on machines that did not exist when the scans were made. Nobody could read them at the time the record was created. The people who made the record were storing a possibility.
That is a message surviving a very long time without a reader. Writing something down does not require the reader to exist yet. The scrolls waited two thousand years for someone to arrive, and the arrival was not guaranteed. It was made possible by an earlier decision that the object was worth recording even though nothing could be read from it.
https://3026alive.com