Philippa Kaur

Joe Wilson
Deep within a particle accelerator, theoretical physicist Giorgio Angelotti is hard at work. He sets a black cylinder on a mount, bolts it down, then runs through some safety checks before retreating from the chamber, known as “the hatch”. “You have to be sure there’s no one in the hatch before you close the door,” he says. “So no one dies.”
That’s because he is about to blast the sample with a super-powerful beam of X-rays. You might expect the target to be some advanced new material or delicate crystal. But, at its heart, this isn’t really a physics experiment – and the object protected inside the cylinder is far from pristine. You could easily mistake it for a misshapen lump of old charcoal.
It is in fact a priceless relic, a 2000-year-old papyrus scroll, scorched beyond recognition in the cataclysmic eruption of Vesuvius in AD 79. It is just one of the Herculaneum papyri, a cache of hundreds of scrolls that are too fragile to be opened by hand, meaning their contents have long remained a mystery. But with the help of particle accelerators, artificial intelligence and a crack team of coders assembled online, Angelotti and his team are starting to make these charred lumps talk. They could soon be uncovering entire lost works of Greek philosophy, or texts written by the earliest Christians.

Discovered near Angelotti’s home city of Naples, Italy, in the 1750s, the scrolls come from the library of a partly excavated, 1st-century-BC villa in Herculaneum. The town, a smaller neighbour of Pompeii, was once a seaside holiday destination for rich Romans. The luxurious villa is thought to have been owned by Roman senator Lucius Calpurnius Piso Caesoninus – none other than Julius Caesar’s father-in-law.
At least some of the 900 scrolls originally discovered were authored by the philosopher Philodemus of Gadara, one of those credited with bringing Epicurean philosophy from Greece to Italy. Classicist David Blank at the University of California, Los Angeles, explains that Philodemus had joined Piso’s entourage, a cohort whose intellectual prowess publicly signalled the senator’s importance. In turn, Piso became a patron of Philodemus’s work, ensuring that a lot of his philosophical writings, including unique early drafts, ended up in Piso’s personal collection.

The Herculaneum papyri
Piso and Philodemus had been dead for decades when Mount Vesuvius blew, but the library remained. As hot mud and ash engulfed Herculaneum, heat dehydrated the scrolls, not burning them, but turning them to charcoal. “The fact they are carbonised is the only reason we have them,” says papyrologist Federica Nicolardi at the University of Naples Federico II. Papyrus normally survives only in very dry climates. Other European examples rotted away centuries ago.
The Piso collection has since dwindled, however. The papyrus layers are tightly stuck together and early attempts to unwrap them resulted in a great many being mashed, sliced, peeled and otherwise processed in ways papyrologists would rather save for potatoes. Starting in the 1750s, the scrolls’ first curator, a man named Camillo Paderni, bashed out their insides to leave just the exterior layers. “He would take the roll, cut through it… then take the butt end of his knife and pound the middle of the roll into dust,” says Blank.
The Herculaneum papyri were turned to charcoal in the AD79 eruption of Mount Vesuvius. This one is known simply as “scroll 2”The Digital Restoration Initiative, The University of Kentucky
A little later, Antonio Piaggio, a manuscript restorer from the Vatican Library, subjected some of the scrolls to a homemade machine. By mounting each scroll and sticking the end of the papyrus to a sheet of animal guts using glue made from fish, he was able to carefully unroll about 18 of them. These early abuses did yield several volumes’ worth of readable texts. This is how we know that at least some of the scrolls were authored by Philodemus. But most of the charcoal lumps languished unread in the National Library in Naples.
And that was how things stood for centuries, until Brent Seales at the University of Kentucky entered the frame. Seales had lived through the early wave of digitisation, when the internet was becoming a repository for knowledge of all kinds. He wasn’t much interested in the mass scanning of ordinary books, but he became gripped by the notion that parts of this global library might be left out due to damage to the physical works. “The idea that technology could create a representation of, or even extract new information from, the damaged stuff – that really appealed to me,” he says.

In 2000, Seales used 3D scanning and computer software to digitally uncrumple and flatten pages from fire-damaged medieval documents amassed by Sir Robert Cotton, part of the founding collection of the British Library. Some books in the trove, however, were too fragile to be opened, so couldn’t be restored using standard imaging techniques, which are based on visible light. Seales began to wonder whether the same methods we use to see inside bodies could be used to see inside books.
The first time he fired X-rays at a book from the Cotton collection, the ink showed up much like bones do in the black and white images, he says. Immediately, he wanted to get his hands on other collections containing unopened texts, and his thoughts turned to the most famous example he knew of: the Dead Sea Scrolls. But when Seales described his plan to conservators, he was met with a “hell no”. Meanwhile, the Herculaneum scrolls entered his radar, courtesy of a tip-off from classicist Richard Janko at the University of Michigan, who had studied the contents of some of the physically opened scrolls.
These particular papyri, though, presented some special challenges. For one thing, unlike medieval writers, who used metallic inks, Philodemus and his contemporaries often wrote in soot-based ink. That meant the challenge was to discern an ink made mostly of carbon from a scroll that was also now mostly carbon. It wasn’t exactly easy. Sure enough, Seales failed to find any ink in initial attempts with a small CT scanner in 2009.
Herculaneum was once a holiday destination for wealthy Roman citizensCCinar/Shutterstock
Many Hebrew and Egyptian scribes used easier-to-image metallic inks. By 2015, Seales was able to read unseen text inside a charred 4th-century-AD Hebrew scroll. And not long after, a European team including Verena Lepper at Berlin’s Egyptian Museum and Papyrus Collection used X-ray-based scans to read the words “oh Lord” inside an ancient papyrus package from the island of Elephantine on the Nile river. But scans from inside the Herculaneum scrolls still hadn’t revealed a single word.
The digital unwrapping process wasn’t straightforward, either. The papyrus layers are so jammed together that it is tricky to peel them apart, even virtually. If the software doesn’t know the difference between one layer and the next, Nicolardi explains, “you produce something that’s actually very similar to what happens with the mechanically opened scrolls”. Pieces of text get spliced between layers, mangling the narrative.
By then, though, AI was on the rise and machines were starting to pick out features that human couldn’t. It turned out that scans of the Herculaneum papyri were, in fact, picking up ink, but it was visible only to properly configured AI. Seales and his colleagues finally demonstrated this on unrolled Herculaneum fragments and fake scrolls inscribed with carbon ink in 2019. That was enough to help secure them use of the particle accelerator at Diamond Light Source near Oxford, UK. He used it as a supercharged CT scanner and obtained images of the insides of rolled-up, intact papyri. But still the scrolls taunted them. Seales’s student Stephen Parsons taught AI software to spot ink on these high-resolution scans, but it struggled to see anything beyond mere traces.
That was when things changed decisively. Seales had connected with tech investor Nat Friedman, previously CEO of Github, hoping to pitch for more research funding. But Friedman had a different idea: put out a public challenge to see if anyone could write a program that could read the scrolls. Seales initially struggled with the proposal. This kind of cash-for-code challenge might be commonplace in the tech world, but for academic researchers it was unfamiliar territory – and it meant opening the scan data and Parsons’s algorithms to a wider community. “It wasn’t an obvious right move for me,” says Seales. “But we realised the only reason we were balking at the idea is that we might not get all the credit, and that was a really bad reason.”
The Vesuvius Challenge
And so, in March 2023, the Vesuvius Challenge was born. Any prize-winning solutions would become public, the code released for the team or others to build on, in the hope that this would speed things up a bit. And so it proved: by Christmas, the challenge’s Discord channel had more than 1000 users.
Angelotti was one of them. Fresh from a doctorate in AI, he had barely heard of the Herculaneum scrolls, despite being born and bred in Naples. But the more he learned about them, the more they intrigued him. Between consultancy work and founding an AI start-up, he poured over digitised papyrus sheets online. As he knew nothing about papyrology, it was a steep learning curve, but it turned out to be time well spent, resulting in cash prizes including $20,000 for work to speed up image processing – and a job offer. Now the research project lead for the Vesuvius Challenge, Angelotti says reading the scrolls has become “a sort of quest to restore the cultural heritage of my homeland”.

Meanwhile, students began to steal the limelight. In December 2023, ink-detection algorithms developed by Youssef Nader and Luke Farritor helped reveal around 2000 Greek characters. Nader taught AI to see ink by carefully training it on broken-off scroll fragments where the papyrus surface was already exposed. At the same time, Farritor was picking out the first word, porphyras (purple), from inside an unopened scroll by using a separate AI model trained on sections where a faint, but just visible, “crackle” pattern seemed to be associated with the inked parts.
By pooling their code and working with Julian Schilliger, a student at ETH Zürich in Switzerland who had been successfully stitching digital papyrus sheets together from pixels, they were able to get better results, not to mention a nod in a peer-reviewed papyrology paper. The translated text uncovered ancient musings on food, music and pleasure, in which the author seemed to ponder the timeless question of what makes life worth living.
Their efforts won them the Vesuvius Challenge’s $700,000 grand prize in 2023 – and, for Nader, a Mount Vesuvius cake (complete with scroll) baked by his family in Egypt. He, too, has since joined the challenge team, continuing to work on ink detection. This is far from a fully solved problem, because the ink varies from one scroll to another. In the long term, the team aims to build a fast, general ink-detection software that works for everything. “So that we can, at some point, just upload a scan of a scroll and download the text,” says Nader.
Students Youssef Nader, Luke Farritor and Julian Schilliger produced this prize-winning image of the text inside one of the scrollsVesuvius Challenge
The unrolling problem hasn’t been completely solved yet, either. Initially, the inked surfaces of the papyrus layers were painstakingly mapped to flattened sections of digital papyrus by humans. But, with help from community members like Schilliger, the team is now increasingly able to get AI to do the task, which should yield faster results.
Could solutions to these problems help researchers read other ancient papyri too? “I don’t think there’s one solution and there doesn’t need to be,” says Lepper, whose work on the Elephantine papyri used more traditional, non-AI software. Each collection has its quirks, she explains. Elephantine papyri, for example, aren’t charred, but many are folded instead of rolled, which can make unwrapping them more complex.
Revealing hidden text in ancient manuscripts is no trivial task. But for the Vesuvius Challenge, at least, progress continues to accelerate “as a direct result of the contest”, says Seales, his initial reservations now seemingly forgotten. Both Seales and Angelotti are optimistic that there will come a time when it is as easy as pressing a button and letting the software do the rest. Right now, though, there are still plenty of scrolls left to scan, meaning more time spent kicking around in the control rooms of particle accelerators.
When New Scientist spoke to Angelotti in mid-July, he had just finished scanning more than 30 Herculaneum scrolls at Diamond Light Source and the European Synchrotron Radiation Facility, the particle accelerator in Grenoble, France, with “the hatch”. He had also been carrying out crucial experimental work, the early results of which suggest that scanning at a higher resolution may help AI see features common to ink across all the scrolls. If so, the whole collection could become imminently readable. The only problem, Angelotti groans, is that it would mean the scans take about six times longer than usual – so more hours to kill in a control room.

Meanwhile, the Vesuvius Challenge team has been preparing to release more data to its community of coders, and successes have continued to mount up. In May 2025, computer science graduates Marcel Roth and Micha Nowak at the University of Würzburg in Germany adapted medical-imaging software to read the first-ever title from within the scrolls, winning themselves $60,000. Roth says the pair got hooked on the contest, at one point skipping university for nearly three months.
And the title? Philodemus, On Vices. “We were all very happy to see it was really Philodemus,” says Angelotti, because it confirmed the AI wasn’t hallucinating. It is unlikely to be the last we hear from Philodemus, either, because most of the scrolls read so far seem to come from the philosophy section of Piso’s vast library.
Back in the Bay of Naples, there could be many more scrolls still to excavate. After all, part of the villa remains unexplored, obstructed by 20 metres of volcano spew and messy local politics. The New Testament puts Paul the Apostle on the scene around AD 50, before his execution about a decade and a half later. Could his movements have been recorded before Vesuvius’s eruption? Perhaps, “if the Herculaneum library had a current events section,” quips Seales. Until recently, of course, there wouldn’t have been much point in looking for such long-lost treasures, since we couldn’t unlock their contents. But now that we can, there’s a good argument for getting out the shovels.

Topics: More

The skeleton of a woman cradling a baby in her left arm, buried at an Anglo-Saxon cemetery in Scremby, UKDr Hugh Willmott, University of Sheffield
Scientists are homing in on a pregnancy test for women who lived hundreds or even thousands of years ago.
For the first time, researchers have detected levels of oestrogen, progesterone and testosterone in the skeletal remains of women from the 1st to the 19th century AD – some of whom were buried with fetuses. The findings show that ancient bones and teeth preserve clear traces of certain sex hormones, which could help identify which individuals in archaeological sites were pregnant or had just given birth at their time of death, says Aimée Barlow at the University of Sheffield in the UK.

“The physiological and emotional experience of pregnancy and pregnancy loss and childbirth are very profound for women, but so far, they’ve largely remained invisible in the archaeological record,” she says. “This method has the potential to revolutionise the way we study reproductive histories of past populations. I’m thrilled, to be honest.”
Pregnancy is difficult to see in ancient individuals, especially if the fetus didn’t have a visible skeleton yet. Even fetuses in the second and third trimester can be overlooked since their bones can resemble those of the mother’s hands – which are often placed over their abdomens for burial.
Modern pregnancy tests measure levels of hormones like hCG in blood or urine. But hCG quickly breaks down, leaving little trace of its presence in the body.

Progesterone, oestrogen and testosterone, however, can linger in tissues longer. Recent research shows that these steroid hormones can be found in people’s blood, saliva and hair – even in long-buried strands from Egyptian mummies.
To assess the potential for detecting ancient pregnancies, Barlow and her colleagues sampled rib fragments and one neck bone from two men and seven women buried in four English cemeteries. They also sampled the people’s teeth, along with those of a third man.
Two of the women had confirmed fetal remains in their abdomens, and two others were buried with newborn babies. The sexes of the other people had been determined by DNA analysis.
The team ground each sample into a powder and used chemicals and other techniques to isolate any steroid hormones. Laboratory testing then determined how much oestrogen, progesterone and testosterone each of the 74 samples contained.
Oestrogen only showed up in four samples, with no clear pattern – possibly because it breaks down quicker than progesterone and testosterone, and might not store well in tissues.
Progesterone, however, showed up especially high in the vertebra of a young woman who died carrying a full-term fetus between the 11th and 14th centuries. The other third-trimester woman, buried in the 18th or 19th century, had elevated progesterone in her rib. Moderate progesterone levels also appeared in the dental plaque of the two women buried with babies in the 5th or 6th century.

Notably, these four women had no traces of testosterone whatsoever in their bones, nor in any part of their teeth – although one buried with a premature baby had a small amount in her plaque. By contrast, the three women not associated with fetuses or infants, who were buried in an 8th-to-12th century cemetery and a Roman-era grave, had testosterone in their ribs and in all layers of their teeth.
Testosterone at low levels plays important roles in women’s health, so its presence in those samples isn’t surprising, says Barlow. “But perhaps the absence of testosterone indicates a recent or current pregnancy at the time of death,” she says.
“This is an exciting and unexpected intersection of archaeology with hormone science,” says Alexander Comninos at Imperial College London. “These techniques could be used to detect pregnancy in skeletal remains more reliably and so give us more accurate insights into ancient pregnancy.”
Even so, while the results are promising, further research must iron out the details, says Barlow. Men’s bones and inner teeth often showed moderate levels of progesterone, for example, for reasons yet to be understood, she says. “The interpretations are very cautious at the moment.”

Walking Hadrian’s Wall and Roman innovation: England

Follow in the footsteps of the Romans on this immersive walking tour along Hadrian’s Wall, one of Britain’s most iconic, ancient landmarks and a UNESCO World Heritage Site.

Topics:archaeology/pregnancy More

A model of Homo heidelbergensis, which might have been the direct ancestor of Homo sapiensWHPics / Alamy
A timeline of genetic changes in millions of years of human evolution shows that variants linked to higher intelligence appeared most rapidly around 500,000 years ago, and were closely followed by mutations that made us more prone to mental illness.
The findings suggest a “trade-off” in brain evolution between intelligence and psychiatric issues, says Ilan Libedinsky at the Center for Neurogenomics and Cognitive Research in Amsterdam, the Netherlands.

“Mutations related to psychiatric disorders apparently involve part of the genome that also involves intelligence. So there’s an overlap there,” says Libedinsky. “[The advances in cognition] may have come at the price of making our brains more vulnerable to mental disorders.”
Humans split from our closest living relatives – chimpanzees and bonobos – more than 5 million years ago, and our brains have tripled in size since then, with the fastest growth over the past 2 million years.
While fossils allow scientists to study such changes in brain size and shape, they can’t reveal much about what those brains were capable of doing.

Recently, however, genome-wide association studies have examined many people’s DNA to determine which mutations are correlated with traits like intelligence, brain size, height and various kinds of illnesses. Meanwhile, other teams have been analysing specific aspects of mutations that hint at their age, providing estimates of when those variants first appeared.
Libedinsky and his colleagues pulled both methods together for the first time, to create an evolutionary timeline of humans’ brain-related genetics.
“We don’t have any trace of the cognition of our ancestors with regard to their behaviour and their mental issues – you can’t find those in the palaeontological records,” he says. “We wanted to see if we could build some sort of ‘time machine’ with our genome to figure this out.”
The team investigated the evolutionary origins of 33,000 genetic variants found in modern humans that have been linked to a wide variety of traits, including brain structure and various measures of cognition and psychiatric conditions, as well as physical and health-related features like eye shape and cancer. Most of these genetic mutations only show weak associations with a trait, says Libedinsky. “The links can be useful starting points, but they’re far from deterministic.”
They found that most of these genetic variants emerged between about 3 million and 4000 years ago, with an explosion of new ones in the past 60,000 years — around the time Homo sapiens made a major migration out of Africa.

Variants linked to more advanced cognitive abilities evolved relatively recently compared with those for other traits, says Libedinsky. For example, those related to fluid intelligence – essentially logical problem-solving in new situations – appeared about 500,000 years ago on average. That’s about 90,000 years later than variants associated with cancer, and nearly 300,000 years after those related to metabolic functions and disorders. Those intelligence-linked variants were closely followed by variants linked to psychiatric problems, around 475,000 years ago on average.
That trend repeated itself starting around 300,000 years ago, when many of the variants influencing the shape of the cortex – the brain’s outer layer responsible for higher-order cognition – appeared. In the past 50,000 years, numerous variants tied to language evolved, and these were closely followed by variants linked to alcohol addiction and depression.
“Mutations related to the very basic structure of the nervous system come a little bit before the mutations for cognition or intelligence, which makes sense, since you have to develop your brain first for higher intelligence to emerge,” says Libedinsky. “And then the mutation for intelligence comes before psychiatric disorders, which also makes sense. First you need to be intelligent and have language before you can have dysfunctions on these capabilities.”

The dates also line up with evidence suggesting that Homo sapiens acquired some of the variants linked to alcohol consumption and mood disorders from interbreeding events with Neanderthals, he adds.
Why evolution hasn’t weeded out the variants that predispose for psychiatric conditions isn’t clear, but it might be because the effects are modest and may confer advantages in some contexts, says Libedinsky.
“This kind of work is exciting because it allows scientists to revisit longstanding questions in human evolution, testing hypotheses in a concrete way using real-world data gleaned from our genomes,” says Simon Fisher at the Max Planck Institute for Psycholinguistics in Nijmegen, the Netherlands.
Even so, this kind of study can only examine genetic sites that still vary among living humans – meaning it misses older, now-universal changes that may have been key to our evolution, Fisher adds. Developing tools to probe “fixed” regions could offer deeper insight into what truly makes us human, he says.

Topics:genetics/human evolution More

The skull of King Richard IIICarl Vivian/University of Leicester
The oral microbiome of King Richard III of England has been assembled by investigating the plaque on his teeth, and suggests he had a disease that can destroy the jaw.
In 2012, skeletal remains were discovered beneath a car park in Leicester, UK, on the grounds of the former Greyfriars church. The remains were suspected to belong to Richard III – who was killed in the battle of Bosworth Field in 1485 and brought to lie in state in Leicester – due to the head wounds matching accounts of his death and spine curvature matching that of his stance. Genetic analysis later confirmed it was him.

Although he reigned for just two years during the wars of the roses, Richard left a strong mark in English history, with rumours of him plotting the murder of his nephews after putting them in the Tower of London, and William Shakespeare portraying him as a ruthless villain in his eponymous play.
However, little is known about the king’s day-to-day life. To learn more, Turi King at the University of Bath, UK, and her colleagues scraped off samples of the dental calculus, or hardened plaque, on three of his well-preserved teeth.
They did this because plaque can work like a time capsule, preserving the DNA of microbes or food. “The amount of DNA recovered from the calculus of King Richard III is among the highest we have ever measured from an archaeological context,” they wrote in a paper where they reported detecting more than 400 million DNA sequences.

“No one has sequenced ancient DNA to 400 million sequences, that’s just astronomical,” says Laura Weyrich at Pennsylvania State University. “It shows to us that we can probably do things with ancient DNA that we didn’t think we could do before.”
King and her colleagues identified nearly 400 microbial species from the DNA. The number and types of species were similar to those detected in well-preserved dental calculus samples from England, Ireland, Germany and the Netherlands from the past 7000 years, spanning the Neolithic Period to the present. “It’s telling us that elite people have the sort of same microbial strains [as everyone else], despite this extravagant lifestyle, despite the travelling he would have done and the wars,” says Weyrich.
The team couldn’t recover enough plant or animal DNA to investigate Richard’s diet. However, a previous analysis of his bones revealed in the last two years of his life he consumed non-local wine and many game animals, fish and birds such as swans, herons and egrets.
However, Weyrich says the microbiome results might be different if the team were to zoom in on a sample from one part of one tooth and compare that against samples from the equivalent tooth in other populations, like those in Germany or the Netherlands. We also have different bacteria in the front of our mouth versus the back, and inside of teeth versus outside, so the team’s limited samples can’t tell us too much about Richard’s oral microbiome as a whole, she says.
The king’s well-preserved teeth meant they could be analysed to gauge his oral microbiomeCarl Vivian/University of Leicester
Nevertheless, one abundant bacterium was Tannerella forsythia. This has been linked to peridontal disease, a serious gum infection that destroys the bone that supports teeth. Oral hygiene in the 15th century was poor, and Richard had cavities when he died aged 32, but this doesn’t mean he necessarily had peridontal disease.
“One person can be colonised by potentially pathogenic bacteria and they will never cause any disease, whereas other people may have an infection,” says Pierre Stallforth at the Leibniz Institute for Natural Product Research and Infection Biology in Germany. Weyrich says an analysis that looks for bone loss in his jaw would be needed to tell whether Richard had periodontal disease.
“What I really like about this field is that it creates a link between social sciences, history and genetics,” says Stallforth. “Just having access to dental calculus from historical figures is amazing in terms of understanding more about their lives.”

Topics:microbiome More

A life-sized camel engraving at Jebel Misma, Saudi ArabiaSahout Rock Art and Archaeology Project
Ancient inhabitants of the Arabian desert created monumental works of rock art on cliff faces, including life-sized images of camels, perhaps as a way to mark sources of water.
Michael Petraglia, at Griffith University in Brisbane, Australia, and his colleagues discovered 176 engravings on 62 panels in the Nefud desert in Saudi Arabia in 2023. There are 90 life-sized images of camels, another 15 smaller camel engravings and two camel footprints.

One of the rock art sites, featuring a 3-metre-tall dromedary, was more than 40 metres up the cliff and impossible for members of the team to safely reach and survey without deploying a drone.
“It would have been dangerous to make these engravings,” says Petraglia. “There’s no way I would go up there.”
Alongside the camels, and highlighting how much more benign the climate must have been, are other large animals including ibex, horses, gazelles and aurochs. The team also found engraved human figures and face masks.

“It’s not just doodling or marking the landscape,” says Petraglia. “These are engravings of things that would have been iconic for them culturally.”
The researchers say the images were possibly carved to warn any outsiders that the land was already occupied or to act as a signpost for ephemeral water sources. The new discoveries add to evidence of extensive past occupation of Saudi Arabia in prehistoric times.
Indicating the antiquity of the images, a natural varnish had formed over the engravings, a process that researchers know would have taken around 8000 years. However, it wasn’t possible to directly date the artwork, so the team excavated in the sediments under the rock art panels.
Excavation of a trench directly beneath a rock art panel at Jebel Arnaan, where engraving tools were discoveredSahout Rock Art and Archaeology Project
There, they found stone points, beads and ochres indicating links with Late Neolithic people in the Levant, as well as tools that would have been used to make the engravings. These objects were able to be dated and ranged in age from 12,800 to 11,400 years old.
Excavations were also undertaken in the small temporary lakes, called playas, near the engravings, which the ancient people would have relied on. Sediments and pollen records confirmed that the region would have been much wetter and greener.
But, even so, the environment was challenging and unlikely to be a place where people could settle and stay for long periods of time, says Petraglia.
“These were likely very mobile people and highly innovative,” he says. “These are sophisticated hunter-gatherers and definitely not people just sort of figuring it out.”

Scientific pioneers of the ancient world, Cairo and Alexandria: Egypt

Embark on an unforgettable journey through Egypt’s two most iconic cities, Cairo and Alexandria, where ancient history meets modern charm.

Find out more

Topics:archaeology More

The Yunxian 2 skull was squashed, but has now been reconstructed and appears to be an early DenisovanGary Todd (CC0)
The origins of our species may lie much further back in time than we thought, and the same may be true of our extinct Neanderthal and Denisovan cousins. According to a new analysis of fossil remains, the shared ancestor of the three groups lived over a million years ago – more than twice as old as previously believed.
“It does mean that we are missing a huge bit of the early story of these lineages, if we’re correct about these ancient branching points,” says Chris Stringer at the Natural History Museum in London.

The results could potentially help settle the search for Ancestor X: the population that gave rise to modern humans, Neanderthals and Denisovans. They could also mean that the Denisovans were our closest relatives – even closer than the Neanderthals, though not everyone is convinced on this last point.
Stringer and his colleagues, including Xijun Ni at the Institute of Vertebrate Paleontology and Paleoanthropology in Beijing, China, re-examined a fossil hominin from Yunxian in central China.
Two partial skulls were uncovered in a terrace above the Han River in 1989 and 1990, and described in 1992. Both skulls had been squashed during their time in the ground. However, the second, Yunxian 2, was less badly damaged.

Stringer, Ni and their colleagues used the latest methods to reconstruct Yunxian 2, including a technique that can use CT scans to digitally separate individual fragments of bone from the surrounding “matrix” of rock and sediment. “It’s long and low, with a big brow ridge,” says Stringer. It also has “a bit of a beaky nose”, and while the teeth are large, the third molars are small.
The Yunxian 2 cranium is 940,000 to 1.1 million years old. Hominins of that age are often thought to belong to Homo erectus, which emerged in Africa around 2 million years ago before spreading to southern Asia, including Indonesia, where it survived until perhaps 108,000 years ago. However, Stringer says it doesn’t fit the profile. Many of its features are typical of later groups, like the Neanderthals.
To figure out what Yunxian 2 is, the team compared it to 56 other hominin fossils. Based on the shapes of the remains, they drew a family tree, with similar fossils being closely related. On this basis, they identified three major groups, which include most of the fossils from the past million years.
The first was modern humans (Homo sapiens). The second was Neanderthals (Homo neanderthalensis), who lived in Europe and Asia during the past few hundred thousand years, vanishing around 40,000 years ago. The third was Denisovans, from eastern Asia.

Denisovans were originally discovered in 2010 using DNA from a bone fragment, and it has taken 15 years to identify larger fossils. Stringer was involved in the description of a skull from Harbin in China, dubbed Homo longi, which was identified as a Denisovan using molecular evidence in June. Yunxian 2 appears to be an early Denisovan, as do several other Asian fossils.
It is helpful to tie all these fossils into the Denisovan lineage, says geneticist Aylwyn Scally at the University of Cambridge. “We can get a better idea about where the Denisovans were, how they lived, and what kind of species they were.”
The finding that Yunxian 2 is Denisovan rewrites the story of recent human evolution in two ways. First, it appears to change how the three populations emerged. The conventional story, as revealed by genetics, is that an ancestral population – Ancestor X – split into two: one half became modern humans, and the other half became Neanderthals and Denisovans, who split from each other a bit later. However, in this reconstruction, it was the Neanderthals who broke away first, 1.38 million years ago, with modern humans and the Denisovans separating 1.32 million years ago.
If that is correct, Denisovans were our closest relatives, instead of being equally close to us as Neanderthals were, as the genetics indicates. However, Scally is dubious. That is partly because the history of these populations seems to be complex. “It isn’t actually well described by a simple tree,” which is the model the researchers used, but rather by a “tangled network”, he says. Furthermore, Scally says genetics is a better guide to such relationships than morphology – especially when you have only partial skeletons – and the genetics tells a clear story.
The second change is bigger: all three groups are much older than we thought. Genetics has suggested that the ancestors of modern humans split from the progenitors of Neanderthals and Denisovans around 500,000-700,000 years ago. But Yunxian 2 indicates that the Denisovan group was already separate over a million years ago.

It may be that there isn’t a single date for any of these splits, says Scally. They could have been protracted, with groups sometimes separating and sometimes coming together. In that case, Stringer and his colleagues could be correct that the divergence began over a million years ago, but it took hundreds of thousands of years to play out.
This longer timeline opens new questions. The oldest known fossils of our species are about 300,000 years old. So, where are all the older ancestors dating back a million years ago? “Either we don’t have them, or they’re there and they haven’t been recognised,” says Stringer.
We also don’t know what Ancestor X was like, or where it lived. “Ten years ago, I would have said probably Africa is the ancestry of most of these groups,” says Stringer. “It looks more likely that the ancestor was outside of Africa, perhaps in Western Asia,” he says. “That would imply that the ancient sapiens ancestor must have gone into Africa and then evolved in Africa for most of the rest of that 1 million years.”
Stringer points out that there are few known fossils from western Asia dating to a million years ago, and India has yielded only one hominin fossil. “There’s many, many places where we really don’t have the evidence,” he says.
One source of data may be the Yunxian site. In 2022, a third skull was found there, apparently in better condition, but it hasn’t yet been described.

Neanderthals, ancient humans and cave art: France

Embark on a captivating journey through time as you explore key Neanderthal and Upper Palaeolithic sites of southern France, from Bordeaux to Montpellier, with New Scientist’s Kate Douglas.

Find out more

Topics: More