Analysis of skulls finds more Denisovans – and unknown human species (2026)

And then there were three.

The world has been aching for skulls from the enigmatic human species known as Denisovans; one has finally been identified; but we may have had three of them in hand for a century, according to a new paper published on Tuesday in PNAS by an Israeli team led by David Gokhman of the Weizmann Institute and Liran Carmel of the Hebrew University.

The Denisovans were discovered only in 2010 following the casual genetic analysis of a finger bone found in a cave in Siberia. It was assumed to belong to a Neanderthal but shockingly, DNA extracted from the bone showed the species was unknown. Based on the name of the cave where the finger was found, they were called the Denisovans.

• Where was the Garden of Eden? Not in the story of human evolution
• ‘Ghost’ Ancestor Detected in DNA of Today’s West Africans
• Exit from Africa: Hominins could have been leaving for 8 million years

But what they looked like remained a mystery because no burial of theirs was ever found, no articulated skeleton; just that finger and a few teeth in Denisova Cave; and years later, a jaw and rib on the Tibetan plateau; and another jaw dredged from the seafloor between China and Taiwan. Not much from which to paint a portrait.

Genetics showed then to be a contemporaneous sister species to Neanderthals. Archaeology showed their teeth were double the size of ours and Neanderthals', but little else. Science yearned for something more indicative.

Then in mid-2025, the holy grail of paleo-humanity was announced: the skull of a Denisovan! The world went wild.

Actually, said skull had been found a century ago in Harbin, China, where it was classified as an unknown species named Homo longi, or Dragon Man. Prof. Israel Hershkovitz of Tel Aviv University shrugged that Dragon Man resembled an early Neanderthal, very much like one found in the Nesher cement works in Ramla, Israel. Anyway, a century after its discovery, ancient DNA was retrieved from Dragon and its identity was outed; and the world hailed its first real look into the face of a Denisovan.

But Dragon Man's classification as a Denisovan was no surprise to the team led by David Gokhman of the Weizmann Institute's Department of Molecular Genetics with Nadav Mishol, advised by Yoel Rak of Tel Aviv University, Gadi Herzlinger and Uzi Smilansky. It was vindication, because six years before the Harbin skull's genetic analysis, they had predicted what Denisovans looked like. Harbin fit the bill in almost all characteristics.

How did they do that? By "reading" the DNA extracted from the fingertip and teeth. They even produced a portrait of what they thought a Denisovan girl would look like.

What do we mean by reading their DNA? Genes in our DNA can be turned on and off, and their expression can be regulated by a process called methylation. Without delving into how, the team checked the state of genes associated with appearance to see which are on and off in modern humans, Neanderthals and Denisovans.

Necessarily, they made some underlying assumptions. For instance, if genes affecting facial width were expressed more in Denisovans than in We, it meant they had a wider face, Gokhman explains. If our genes affecting forehead height are more expressed than theirs, we will have a higher forehead, and so on. In 2019 they published the seminal paper "Reconstructing Denisovan Anatomy using DNA Methylation Maps." In short, they mapped each genetic trait: greater or diminished in the Denisovans compared with Neanderthals and modern humans and painted a portrait of what they might look like based on genetics.

Their huge teeth may have suggested that Denisovans were giants but their genetics indicate they were squat, like Neanderthals, while we Homo sapiens are relatively tall and gracile. They had barrel bodies, we are elven. We and Neanderthals also have relatively dainty teeth compared with the Denisovans though they weren't particularly prognathous, Gokhman adds.

In 2019, the team tested their methodology for predicting appearance by applying it to Neanderthal and chimpanzee datasets. We know what Neanderthals looked like because we have a few hundred skeletons (versus zero Denisovans). We know what chimps look like. The team's prediction of what a chimp would look like based on DNA methylation was accurate: It looked like a chimp. This is good.

In their new paper, "Candidate Denisovan fossils identified through gene regulatory phenotyping," they team studied 10 fossil skulls found in Eurasia and Africa to see whether they match their image of Denisovans. Harbin was the best fit, and close behind were two more candidates, sitting for decades on museum shelves but similarly misidentified: Dali Man and Broken Hill Man, aka Kabwe Man or Homo rhodesiensis, whose skull was found by copper miners in 1921.

Being from China, Harbin Person and Dali Man fit perfectly with the theory of where Denisovans ranged: Asia.

Being from central Africa, Broken Hill Being or Kabwe Man is quite the outlier.

Neanderthals and Denisovans emerged in Eurasia and never inhabited Africa, according to classic theory. Kabwe is a city about 80 kilometers from Lusaka, the capital city of Zambia, and is about 13,700 kilometers from Siberia, the most western point of the Denisovan range that we know. What would a Denisovan have been doing there 300,000 years ago?

Neanderthals nyet

Let's start with the seven skulls the team studied that they don't think fit the Denisovan profile. "Some are definitely not suitable," Gokhman explains by telephone: if anything, the opposite. But they don't fit the profile of any other lineage either.

A specimen found in Steinheim, Germany in 1933 particularly hints at being an unknown species, he adds – he isn't Neanderthal or Homo erectus. Definitely a hominin though, not some monkey.

The specimens from Ndutu and Eliye Springs in East Africa sport characteristics of both Neanderthals and Homo sapiens. Nu. Narmada from India – not much to work with there, but atypical; then Jinniushan, China has a specimen in exactly the Denisovan stamping ground but he groups near Neanderthals. A candidate from Spain called Arago looks like erectus.

There is also a partial skull from Xuchang, China that the team didn't study, because there just isn't enough of it to work on, but they suspect it too may be Denisovan. So maybe there are four.

What about Nesher Ramla Homo – maybe he's Denisovan-type too? It's possible, but the skull is too fragmented to have its characteristics measured, Gokhman says.

What have we? One certainty, Harbin, which fit their predictions; two powerful candidates, Dali and Kabwe; species with characteristics that don't fit any mold; and a lot of questions about what this all means.
One thing it means is that human evolution wasn't linear as once thought but a huge mess. "Species left Africa, went back, split, intermixed, went to Asia, to Europe, replaced one another – there was nothing linear about it," Gokhman says.

Fine. But if the base theory is that Neanderthals and Denisovans evolved in Europe, what was a candidate Denisovan doing in Zambia?

Kulanu Kabwe?

Kabwe Man may represent a species ancestral to Neanderthals and Denisovans that lived in Africa. Perhaps about half a million years ago, a band of this species roamed out of Africa and reached Eurasia, and split again; the ones that turned left would become the Neanderthals, and the ones that turned right would become Denisovans. But maybe, Gokhman suggests, a branch stayed behind and would produce Kabwe.

Or, Kabwe was an emerged Denisovan who returned from Eurasia. The Neanderthals and Denisovans both got about, spreading over all the habitable areas of Europe and Asia respectively. A new theory posits how Neanderthals specifically might have crossed brutal terrain in their migrations.

If there's a new prediction here, it's that Steinheim and others could represent more unknown species. Fact is that in recent years, several new (extinct) human species were identified, including miniature ones in the Philippines and Indonesia: Homo luzoniensis and Homo floresiensis, now thought to have descended from Homo erectus.

Our species, Homo sapiens, also split about 50,000 years ago. One branch stayed in Africa, begetting Africans. The other branch reached Eurasia and survived and interbred with both Neanderthals and Denisovans and begat everybody else alive today.

Today all non-Africans have roughly 2 percent Neanderthal DNA, and Denisovan DNA to a variable degree. The presently known highest levels of Denisovan ancestry are in the Negrito population of Ayta Magbukon in the Philippines, which is proposed to stem from a distinct and intense admixture event with Negritos. And now we know what the attraction looked like.

No Denisovan fossils have been found in the areas of Southeast Asia with the highest levels of Denisovan ancestry. But they have to have been there.

But a question. Sub-Saharan Africans have no Denisovan DNA signal and gained a smidgen of Neanderthal DNA believed to have originated from hybrid humans who returned to Africa. But if Kabwe Man was of the Neanderthal/Denisovan ancestral lineage that stayed in Africa, wouldn't Africans have genes from them?

They may well do. "If Kabwe or its relatives admixed with Africans 300,000 years ago, then all modern humans [Africans and non-Africans] would have Kabwe-like DNA, and we wouldn't be able to distinguish it because it's shared by everybody, and because we don't have Kabwe's DNA," Gohkman answers.

"However," he adds, "if Kabwe's descendants admixed with Africans after the ancestors of non-Africans had left Africa, then Africans would have Kabwe-like DNA but non-Africans wouldn't, and we might be able to identify it. In fact – there was a paper that showed that Africans have between 2 percent to 18 percent archaic admixture from a 'ghost lineage.' Some speculated it to be Homo naledi, some thought it might be Homo erectus, but it could also be Kabwe's lineage!"