Northern China was under Japanese occupation in 1933, and the man building the bridge over the Songhua River knew exactly what would happen if he handed the skull to the authorities.
He had turned it up while digging foundations for the Dongjiang Bridge in Harbin, in the far northeast of the country. It was heavy, unusually thick, more complete than any modern human skull he had ever seen. He took it home. He put it in the bottom of an abandoned water well near his house. Then he covered the well and told no one.
For the next eighty-five years, the skull sat in the well. The bridge above got completed. The Japanese occupation ended. Wars came and went. Regimes changed. The man grew old and had children. In 2018, on his deathbed, he told his family what was in the well. They retrieved the skull, and after some hesitation, they donated it to a university.
What the family handed over
The skull was passed to Qiang Ji, a paleoanthropologist at Hebei GEO University who had spent decades working on Chinese hominin fossils. When Ji examined it, he immediately recognised that it did not fit neatly into any known human species.
It was large — one of the biggest hominin skulls ever recorded, with an oversized braincase, thick brow ridges, and a wide low face with square eye sockets. The dating, based on uranium series analysis of the specimen’s own bone, placed it at least 146,000 years old.
In 2021, Ji and his colleagues published a paper describing the skull as a new species, which they named Homo longi — “Dragon Man,” a reference to the Long Jiang, or Dragon River, where it had been found. The naming was contested from the start. Many paleoanthropologists suspected the skull was actually a Denisovan — the mysterious archaic humans first identified in 2010 from DNA in a Siberian finger bone. But Denisovans, at that point, had never been represented by a complete skull. All previous evidence for the group came from a few teeth, some finger and toe bones, a jaw fragment from Tibet, and other small pieces from Laos and Taiwan. Nobody knew what a Denisovan face looked like.
The Harbin skull was, if the identification was right, the first complete answer to a question that had persisted for fifteen years.
The problem was getting hard confirmation. The skull’s DNA had degraded. Initial attempts to extract usable genetic material from the bone and teeth failed. The Denisovan identification remained a strong suspicion — supported by the age, the region, and the general skull morphology — but without molecular evidence, it could not be settled.
The 0.3 milligrams that solved it
A team led by Qiaomei Fu at the Chinese Academy of Sciences tried a different approach.
Fu’s specialty is ancient DNA extraction from unusual sources. Rather than continuing to work with the skull’s bones, she and her team turned to the small amount of fossilised dental plaque — dental calculus — still clinging to the skull’s single remaining tooth. Calculus contains bacterial DNA, mostly, but it can occasionally trap and preserve genetic material from its host as well, protected inside the mineralised crystal structure of the plaque itself.
The sample size the team worked with was 0.3 milligrams. That is roughly the weight of a poppy seed.
From that fragment, they recovered mitochondrial DNA sequences that matched other Denisovan specimens, particularly early Denisovan samples from Siberia dating between 187,000 and 217,000 years ago. A parallel proteomic analysis of the skull’s petrous bone — the inner ear bone that preserves ancient proteins better than most bone — recovered protein sequences with Denisovan-specific variants.
Two independent lines of molecular evidence, from two different tissue types, pointed the same way. In June 2025, Fu’s team published back-to-back papers in Cell and Science confirming that the Harbin skull was Denisovan. “After 15 years,” Fu told National Geographic at the announcement, “we give the Denisovan a face.”
The Harvard archaeogeneticist Christina Warinner, quoted in Science, summarised the technical significance: “In cases like this, calculus can be our only hope of getting DNA from really old individuals.” A skull too degraded for conventional DNA work had yielded its identity through a technique that would have been impossible a decade ago.
What is still uncertain
The specific details of the 1933 discovery cannot be fully verified.
Some scientists have quietly noted that the “grandfather in 1933” story may be partly invented — that the actual finder may have been the man who eventually turned the skull over in 2018, who could have been trying to avoid trouble for not reporting the artifact when he found it. Chinese authorities take a strong interest in undocumented antiquities, and giving a specimen a longer provenance can make its handover less legally awkward. The 1933 date is plausible, and the political context fits — Harbin was under Japanese occupation, and confiscation of significant finds was a real risk — but the specific chain of custody rests on family testimony that cannot be independently verified.
The scientific identification is not similarly uncertain. The DNA evidence, the protein evidence, and the morphological analysis all converge on Denisovan status. Whether Denisovans should be called Homo longi as a separate species, or simply treated as a variant population within the broader Denisovan lineage, is a live debate among paleoanthropologists — Svante Pääbo of the Max Planck Institute has argued against species-level naming for either Neanderthals or Denisovans, on the grounds that they interbred with modern humans and each other freely enough that separate species designations obscure more than they clarify.
The one thing everyone agrees on is that we now know what a Denisovan looked like. A large, heavy-boned individual with thick brows and a wide low face, who lived somewhere in northern China at least 146,000 years ago, and whose skull spent 85 years at the bottom of a well before anyone could tell them.
That, and 0.3 milligrams of plaque, was enough.
