AI helps read papyrus scroll burnt to crisp during Vesuvius eruption

An ancient papyrus scroll, charred beyond recognition during the catastrophic Vesuvius eruption nearly two millennia ago, has been virtually unwrapped and deciphered using advanced AI technology.

Researchers successfully revealed twenty columns of hidden text spanning over one metre of carbonised papyrus, all without physically unrolling the extraordinarily fragile document.

The breakthrough represents a landmark achievement in the application of machine learning to classical archaeology and manuscript preservation.

The scroll, designated PHerc 1667, contains a discourse on Stoic philosophy addressing ethics, art, and human behaviour, dating to the second or late third century BC.

Were it not for this innovative technology, such philosophical insights would have remained permanently inaccessible within the blackened papyrus.

The document is among the oldest in a collection of hundreds recovered from the private library of a luxurious Roman villa in Herculaneum.

The villa's library was engulfed by searing heat and subsequently buried under volcanic ash when Vesuvius destroyed the neighbouring city of Pompeii in AD 79.

Centuries of deterioration and misguided handling had taken a devastating toll on the scroll, which at some point broke in half.

Previous attempts to unwrap the document physically had caused its outer layers to flake off or disintegrate entirely, leaving a fragment merely eight centimetres tall and two centimetres wide.

Doctor Federica Nicolardi, a papyrologist at the University of Naples Federico the Second, emphasised the significance of the accomplishment.

She argued that although the full scroll no longer exists, the fact that researchers were able to unwrap the surviving object completely demonstrates the viability of this technique.

Not only does it validate the methodology, but it also opens the door to deciphering other carbonised scrolls in the collection.

The findings are set to be formally announced at a conference in Naples and constitute the latest milestone from the Vesuvius Challenge.

Launched in 2023 as a global competition, the initiative has galvanised researchers worldwide to harness machine learning in recovering texts from these seemingly irretrievable artefacts.

Should the technique prove equally effective on additional scrolls, scholars could gain unprecedented access to lost works of ancient philosophy and literature.