Category: Arts and Culture / Science / Artificial Intelligence
Introduction: A Book Nobody Could Open
Imagine finding a library whose books have survived nearly two thousand years, only to discover that opening them would destroy the very words you hope to read. The books are there. Their pages are there. Somewhere within the tightly rolled material are arguments, observations, jokes, and questions written by people who lived before the Roman Empire reached its height. Yet their contents might as well be locked behind a wall. This is not the plot of a historical thriller. It is the real problem posed by the carbonized papyrus scrolls of Herculaneum, an ancient town overwhelmed by the eruption of Mount Vesuvius in AD 79.
In June 2026, researchers working through the Vesuvius Challenge announced a remarkable milestone. They had digitally unwrapped and read the entire surviving length of a particularly difficult scroll, cataloged as PHerc. 1667, without physically opening it. The achievement did not restore every word of the original ancient book: earlier attempts to unroll the object had already destroyed its outer sections, and damaged passages remain incomplete. But the team recovered a continuous sequence of writing from the surviving core. Other scrolls yielded newly identifiable titles and long stretches of Greek text. A library that had been present but effectively silent was beginning to speak again.
The breakthrough brings together a surprisingly wide cast of specialists: archaeologists, papyrologists, physicists, computer scientists, conservators, and volunteers who compete to solve technical problems. It also asks an unusually human question. When machines help us recover words that history seemed to erase, what do those words change about our understanding of the past? The answer begins not with artificial intelligence, but with a disaster that accidentally preserved what it buried.
1. The Day Vesuvius Closed the Library
Herculaneum lay on the Bay of Naples, near Pompeii, within reach of the wealth, trade, and intellectual life of the Roman world. In AD 79, Vesuvius erupted and buried nearby communities under volcanic material. The destruction killed people and devastated buildings, but it also sealed extraordinary evidence of everyday life. Walls, wooden objects, and household spaces survived in conditions that would otherwise have caused them to decay. In one luxurious coastal residence, now known as the Villa of the Papyri, archaeologists eventually found something even more unusual: a collection of ancient books.
A papyrus scroll was made from strips of the papyrus plant pressed into sheets and joined into a long writing surface. Readers moved through the text by unwinding one side and winding the other. Unlike a modern bound book, a scroll could hold columns of writing across a continuous length. Its physical shape mattered. A sentence might start in one column and continue into the next, while the author developed an argument over many turns of the roll. To understand an ancient work, scholars need more than scattered letters. They need the order, context, and continuity of its pages.
Heat from the eruption carbonized many of the Herculaneum rolls. Their organic material became blackened, brittle, and extraordinarily difficult to distinguish from ordinary pieces of charcoal. Carbonization helped some of the material persist, but it created a cruel preservation problem. Unrolling the scrolls mechanically could tear their fragile layers apart. The library survived in a form that resisted being read. For centuries, the balance between saving the object and exposing its contents seemed impossible to maintain.
2. The Long History of Trying to Read Ash
The scrolls were rediscovered during excavations in the eighteenth century. Early investigators recognized that the black, cylindrical objects were not merely burned wood. They were rolled manuscripts. That realization offered an almost unbearable temptation: important works of classical literature might still exist inside. Methods developed to open them included painstaking mechanical techniques, some of which exposed readable surfaces but also risked serious damage. The name of Antonio Piaggio became associated with a machine designed to unroll the fragile objects with greater control, an impressive intervention for its time.
The resulting texts transformed scholarship, even though only part of the collection could be deciphered. Many works were associated with Philodemus, an Epicurean philosopher active during the first century BC. His writing treated subjects such as ethics, theology, rhetoric, and the art of living. But the collection was not a tidy bookshelf arranged for modern historians. Some papyri were broken, some were badly damaged, and some remained tightly closed. The easiest pieces to read naturally shaped what scholars thought the library contained. Unopened rolls could still harbor unfamiliar authors and unfamiliar arguments.
This is an important reminder about historical evidence. What survives is not necessarily a representative sample of what people once wrote. Fires, floods, wars, reuse of writing materials, and changing cultural preferences select which documents reach us. Even when a manuscript physically survives, its condition may prevent its words from joining the historical record. Herculaneum is exceptional precisely because it preserves a large body of material that has not yet been fully accessible. It offers the possibility of revising the archive without excavating an entirely new site.
For a long time, noninvasive reading remained more hope than reality. Ordinary photographs reveal only the surface. Conventional X-rays can show internal structure, but they do not automatically provide clear writing. The ink and the papyrus can be frustratingly similar in composition after carbonization. Researchers needed a way to see the interior, reconstruct a sheet buried inside a complex roll, and identify marks so subtle that a human observer might miss them. The solution would require several technologies working together rather than a single miraculous machine.
3. The 2026 Breakthrough: What Was Actually Read?
On June 25, 2026, the Vesuvius Challenge reported that the surviving core of PHerc. 1667 had been virtually unwrapped and read from end to end. The team described nearly 1.5 meters of continuous text spanning about twenty columns, with the surviving lower portions of twenty-two columns examined in the detailed account. The small core is only part of a once larger scroll. Historical attempts to open it had destroyed outer layers, so the announcement must not be confused with recovering a complete original book exactly as its ancient owner held it. It is a complete reading of what remains of the rolled object.
The recovered work concerns ethics, human nature, impulses, and moral development. Its themes and an apparent reference to Aristocreon, a relative and student of the Stoic philosopher Chrysippus, suggest a Stoic intellectual setting. Researchers have proposed a date in the second century BC, potentially placing it among the older texts in the Herculaneum collection. The title and author have not been securely established. Those distinctions matter because exciting headlines can make scholarship sound more certain than it is. A readable sequence does not erase missing passages or settle every historical attribution.
Other results show why this was more than the successful treatment of one artifact. In PHerc. 139, researchers identified an attribution to Philodemus and the title On Gods, Book 8. That matters because the designation points to a larger multi-book work than the surviving evidence had previously confirmed. At Oxford, work on PHerc. 172 produced more than seventy columns of recoverable text, including material identified as On Vices, Book 1. The team also demonstrated clearer direct detection of ink in scan data from another scroll, offering an important independent check on earlier machine-assisted readings.
The new material is not a glamorous lost novel with a perfectly preserved beginning and ending. Much of it is philosophical argument, reconstructed through gaps and damaged lines. Yet that is precisely its value. Ancient thinkers debated the shape of a good life, the force of desire, the nature of divinity, and how humans might become wiser. Recovering those conversations in sustained passages gives historians evidence for ideas they previously knew only through quotations, summaries, or scattered fragments. It makes the ancient world less like a collection of famous names and more like an ongoing conversation.
4. Seeing Through a Scroll Without Touching Its Pages
The technical story begins with tomography, a method for reconstructing the interior of an object using measurements taken from many angles. In a medical CT scan, X-rays pass through a body and detectors record information that software turns into cross-sectional images. The Herculaneum project uses related principles at much finer scales. Researchers have collected high-resolution X-ray microtomography data at facilities including the European Synchrotron Radiation Facility in France and Diamond Light Source in the United Kingdom. Their powerful beams and sophisticated detectors can reveal minute structures inside the carbonized rolls.
Think of a sealed scroll as an enormous, tightly packed maze made from a single crumpled ribbon. The task is not simply to photograph the maze. It is to determine where one surface ends and the next begins, to follow a specific layer around folds and distortions, and to reconstruct what that same surface would look like if it were laid flat. A small error can jump between adjacent sheets and scramble the sequence of letters. The computer must respect the scroll's three-dimensional geometry before its image can resemble a readable page.
This operation is called virtual unwrapping. Specialized software traces the internal papyrus layers through a three-dimensional scan, produces a model of their surfaces, and maps those surfaces into flatter two-dimensional images. That mapping is not equivalent to pulling open a physical roll. The artifact remains in its protected state while its digital representation is explored. Researchers can revise the segmentation, inspect ambiguous regions, and return to the original volume when a proposed reading looks suspicious. The process preserves both the object and the possibility of correcting the analysis.
The geometry alone, however, is not enough. On the reconstructed surface, the letters may still be almost invisible. In many papyri, carbon-based ink and carbonized papyrus have similar X-ray absorption properties. The problem resembles trying to spot a charcoal drawing made on a sheet of charcoal. Advanced imaging captures subtle differences in texture and material structure; computer models then help isolate signals that may correspond to ink. The result is an intricate chain of measurement, reconstruction, detection, and human interpretation rather than an optical photograph of a hidden page.
5. What Artificial Intelligence Contributes
Artificial intelligence enters after scientists have gathered the scan data and begun reconstructing the surfaces. Machine-learning systems can learn statistical patterns associated with ink or with features that help experts trace the papyrus. They are especially useful when the difference between a mark and its background is too faint or complicated for simple image processing. A model can examine tiny patches of surface and estimate where writing is likely to appear, providing a map that researchers can inspect and refine.
But the phrase 'AI read an ancient book' can conceal essential details. Software does not independently know which interpretation of an ancient Greek sentence is historically justified. It can confuse cracks, fibers, mineral deposits, and scanning artifacts with letters. An algorithm that appears impressive on familiar data may fail on a scroll with different damage. And a language model could produce plausible Greek where the image provides inadequate evidence. This is why the most responsible workflow separates visual evidence from linguistic expectations and makes proposed readings available for scholarly verification.
Papyrologists remain central. They identify letter shapes, compare them across damaged areas, examine spelling and handwriting, and decide whether a reading fits the surviving traces. They also know when to leave a gap instead of inventing a word. The best result is not the smoothest translation. It is the most defensible reconstruction of marks that can actually be observed. In the 2026 announcements, direct visibility of ink in improved X-ray data was especially significant because it provided another way to test what an earlier reading had proposed.
This partnership changes how we should think about AI in cultural research. The machine excels at processing enormous visual datasets and exposing patterns that human attention might overlook. The historian provides contextual judgment and an understanding of what counts as evidence. The conservator protects the original object. The physicist knows what the instrument measured and what it could not measure. No single participant can solve the entire problem. The breakthrough succeeds because these forms of expertise constrain and strengthen one another.
6. Why Reading an Ancient Scroll Is More Than Decoding Letters
To someone outside classical studies, the distinction between twenty lines and twenty columns might sound technical. It is actually the difference between overhearing a few words of a conversation and following the speaker's reasoning. A fragment may show that an author mentioned happiness. A sustained passage can reveal whether the writer considered happiness a goal, a consequence, a mistake, or a disputed definition. Context lets scholars distinguish a position an author endorses from an objection the author is trying to refute.
Continuous text also helps establish structure. An argument develops through premises, examples, rebuttals, and conclusions. The language of ancient philosophy is full of specialized vocabulary whose meaning depends on the school, author, and passage. When more of the surrounding text becomes available, experts can compare terminology, follow transitions, and notice changes in emphasis. They can begin to understand not only what a writer said, but how the writer tried to persuade an audience. That richer evidence can revise modern textbooks and translations in ways an isolated word never could.
The Stoic themes suggested by PHerc. 1667 make this especially intriguing. Stoicism remains popular today, often reduced to slogans about resilience and controlling emotions. Ancient Stoic philosophy was a much more elaborate intellectual tradition, addressing logic, physics, ethics, community, and the relationship between human impulses and reason. A new witness to this tradition might clarify debates among thinkers rather than simply repeat familiar advice. It may show how ideas circulated, how students adapted their teachers' positions, or how ethical vocabulary shifted between generations.
The Epicurean material offers a complementary perspective. Epicurean philosophy is sometimes mistakenly treated as a program of luxurious pleasure, but its ancient practitioners also valued tranquility, friendship, and freedom from unnecessary fear. Texts by Philodemus preserve detailed discussions about moral behavior and intellectual life. Recovering more of his work can illuminate disagreements among schools, the educational interests of Roman elites, and the way Greek philosophy moved through the Roman world. These scrolls do not merely add quotations. They expand the evidence from which an entire intellectual landscape is reconstructed.
7. An Ancient Library as a Scientific Dataset
Perhaps the most surprising part of the story is that a library can now be studied as a three-dimensional dataset. A scroll is simultaneously a historical document, a fragile material object, and a complex arrangement of surfaces. Its physical condition contains information about how it was rolled, damaged, heated, compressed, handled, and stored. Computer scientists model that structure; materials researchers examine how papyrus and ink respond to imaging; historians connect the recovered marks to actual human language.
The Vesuvius Challenge has emphasized open data, accessible software, and collaboration beyond a single institution. Public releases of scan volumes and related processing results let different groups test techniques on the same problems. That is valuable because digital reconstruction involves many choices. A researcher must decide how to trace a surface, which distortions to correct, and which features constitute meaningful evidence. Independent teams may disagree, and those disagreements can expose weaknesses that would be missed within one laboratory.
Openness does not mean the task is easy for anyone with a laptop. High-resolution scans can be extremely large, and specialized computation may be needed to process them. Expert interpretation is still scarce. Open resources instead lower the barrier for new algorithms, new checks, and creative collaborations. Students in computer vision might devise better segmentation tools. Mathematicians might improve surface flattening. Scholars of ancient Greek might examine new readings. A formerly inaccessible manuscript becomes a common problem whose pieces can be studied across disciplinary borders.
This model has implications far beyond one buried villa. Museums and archives hold folded papers, sealed correspondence, damaged manuscripts, and books too fragile to open. Every object presents different materials and different imaging challenges, so success at Herculaneum is not a promise that all hidden writing can now be recovered. Nevertheless, the principle is powerful: digitize the object carefully, reconstruct the structure, verify the marks, and share enough evidence that other researchers can challenge the result. Preservation and access no longer have to be treated as permanent opposites.
8. The Difference Between Recovery and Reconstruction
The story invites excitement, but it also deserves precision. Digital unwrapping cannot restore physical material that no longer exists. If an ancient layer was pulverized during an earlier excavation, no algorithm can truthfully read its absent ink. If two layers are fused or distorted beyond the scan's resolution, the resulting geometry may remain uncertain. Better equipment may recover more evidence later, but an image-processing system cannot turn a missing surface into an authenticated original. Responsible scholarship keeps that boundary visible.
Even surviving traces do not always lead to unique interpretations. Ancient handwriting varies. Ink may be faint. A broken letter can resemble several others, while a damaged word may support multiple grammatical readings. Scholars therefore use editorial conventions to distinguish clearly visible letters, probable restorations, and unresolved gaps. These symbols can make a transcription look less elegant than a polished book passage, but they are a record of intellectual honesty. They tell future readers what the artifact actually supports.
There is also a danger of mistaking improved image quality for certainty about historical meaning. A clear word can still be translated badly. A plausible author attribution can remain tentative. A reference to a known philosopher may reflect admiration, disagreement, or a later commentary rather than direct authorship. A text written in Greek and preserved in a Roman villa does not automatically tell us what every resident believed. The recovered words are evidence to be interpreted, not a time machine that erases historical distance.
The researchers' more careful descriptions acknowledge these limits. The term 'complete' refers to the surviving rolled core of PHerc. 1667, not to the lost original extent of the book. Other texts have been recovered in long passages but require continued editing and critical study. Public enthusiasm is justified by the scale of the technical achievement, yet the work will become most valuable when patient specialists test individual readings, compare alternative interpretations, and publish editions that explain the uncertainties instead of hiding them.
9. Why the Discovery Matters Outside Ancient History
The first wider lesson concerns how technology can make previously unreachable evidence accessible without damaging its source. This is a different vision of innovation from replacing an old object with a digital substitute. The papyrus remains irreplaceable, and the digital copy is useful because it respects that fact. Similar thinking can guide conservation of paintings with hidden sketches, fossils trapped inside rock, mechanical devices too fragile to dismantle, and archival materials that cannot withstand repeated handling. The proper question becomes what can be learned while disturbing the original as little as possible.
The second lesson concerns human judgment in the age of generative AI. It is easy to be impressed by software that outputs fluent paragraphs. Herculaneum reminds us that fluency is not evidence. In a project where a single false letter could distort an ancient argument, teams must trace claims back to measurements, images, and scholarly reasoning. That approach has broad value for medicine, scientific research, journalism, and education. A convincing answer is not necessarily a supported one; confidence should follow verification rather than appearance.
A third lesson concerns cultural memory. Much of what modern societies consider the classical canon reflects accidents of copying and survival. Some works were preserved because medieval scribes valued them. Others vanished because manuscripts decayed or were never reproduced. The Herculaneum collection offers a rare opportunity to recover voices beyond the narrow pathways through which ancient literature usually reaches us. It might confirm familiar ideas, complicate them, or introduce concepts that deserve a place in the story of philosophy.
The discovery also changes the emotional relationship between the public and archaeology. A museum object that looks like a burned lump can contain an argument made by a person who lived before the common era. Scientific images reveal not only structure but the possibility of a sentence, a thought, and a debate. That transformation encourages visitors to see artifacts as unfinished sources of knowledge, not simply beautiful displays. The past is not always complete when an excavation ends. Sometimes a new instrument gives an old discovery an entirely new beginning.
10. What Happens Next?
Hundreds of Herculaneum rolls remain unopened. The University of Kentucky's 2026 overview put the number at approximately six hundred. That does not imply six hundred perfectly preserved books waiting to be scanned and translated next week. The objects differ in condition, dimensions, access restrictions, and the complexity of their internal structure. Every successful reading may require costly scanning time, meticulous conservation, specialized processing, and long scholarly review. The next phase will likely consist of gradual improvement and careful publication rather than one dramatic release of an entire lost library.
Researchers will continue refining methods that trace layers accurately and distinguish ink from background signals. Improved imaging might reduce ambiguity in difficult regions, while better computational techniques could make previously expensive reconstructions more manageable. At the same time, papyrologists face a growing interpretive task: establishing reliable Greek texts, comparing them with known works, translating them, and making the evidence accessible to broader audiences. A readable image is a beginning, not the end, of publication.
Some of the most interesting discoveries may come from identifying what a scroll is before reading all of it. A title, an author's name, or a distinctive passage can connect a damaged roll to a lost work mentioned elsewhere. Other results may show that a text belongs to an unfamiliar tradition, has been misclassified, or preserves a variant of something already known. The collection may also reveal how a wealthy Roman household assembled books and which philosophical debates were circulating in its intellectual environment.
There is a public role here as well. Schools can use the project to teach the difference between an image, an inference, and a historical conclusion. Museums can explain why conservation sometimes requires not opening an object. Computer science students can study how uncertainty propagates from a scan into a final transcription. General readers can follow the open releases and scientific papers, looking for findings that survive expert checking. The most valuable outcome will not simply be more text on screens, but a better understanding of how knowledge is recovered.
Conclusion: The Future of a Very Old Conversation
The Herculaneum scrolls embody a strange reversal. The eruption of Vesuvius destroyed the world in which these books were read, yet helped preserve physical traces of their words. For centuries, those traces remained trapped inside blackened rolls. Now high-resolution X-rays, computational geometry, machine learning, and classical scholarship have made it possible to recover substantial writing without breaking the remaining pages apart. The 2026 reading of PHerc. 1667 marks a major threshold because it shows that a continuous surviving text, not merely isolated characters, can emerge from an object once treated as unreadable.
What comes next will involve less spectacle and more sustained intellectual work. Scholars must determine what the recovered arguments mean, where they belong in the history of ideas, and how confidently particular words can be established. Machines may locate signals that humans cannot easily see, but people still have to ask good questions about the evidence. That partnership is what turns an image into a document, and a document into history.
The most compelling possibility is not that artificial intelligence has conquered the past. It is that modern tools have given ancient writers another opportunity to participate in a conversation that never truly ended. Somewhere inside the next tightly rolled fragment may be a clarification of a familiar philosopher, an unexpected disagreement, or a name that has not been spoken for centuries. The books do not need to move. We are finally learning how to listen.
Sources
Vesuvius Challenge — We Read an Entire Scroll Without Ever Opening It — June 25, 2026. University of Kentucky — The Day the Herculaneum Scrolls Began Speaking Again — June 25, 2026. Diamond Light Source — New Secrets Revealed from the Herculaneum Scrolls — June 25, 2026. European Synchrotron Radiation Facility — The Day the Herculaneum Scrolls Began Speaking Again — June 25, 2026. Bodleian Libraries — Inside of Herculaneum Scroll Seen for the First Time in Almost 2,000 Years — February 5, 2025.
The conversation
Post a Comment