The Layers Were Read in the Wrong Order

Liang Bua is a limestone chamber in the hills about fourteen kilometres north of Ruteng, in the Wae Racang valley, with a mouth wide enough that daylight reaches most of the floor. A Dutch missionary dug in it in the 1960s. Indonesian archaeologists worked it through the 1970s and 1980s and found what you would expect: pottery, burials, the ordinary residue of the last few thousand years. In 2001 a joint Indonesian–Australian project went deeper. In September 2003, roughly six metres down, they uncovered a skull, and then most of a skeleton, of an adult who had stood a little over a metre tall.

It was published in Nature in October 2004 as Homo floresiensis and renamed by the press within the week as the hobbit. The skeleton, catalogued LB1, had a braincase first measured at about 380 cubic centimetres and later remeasured at around 426 — chimpanzee territory — on a body with very long feet, no chin, and a shoulder joint rotated forward relative to ours. None of that is why the find detonated. The reason was the date. The deposit the bones came out of was read as roughly eighteen thousand years old, which put a separate human species alive on an Indonesian island long after modern humans had reached Australia.

What came out of the trench

Digging carried on and the cave produced fragments of at least nine individuals, none of them large, along with thousands of stone flakes and the bones of the animals that shared the island: Stegodon florensis, a dwarfed relative of the elephant, giant rats, vultures, and Komodo dragons, which are the only member of that fauna still walking around. Islands do this. Large mammals arriving on them tend to get smaller over generations and small ones tend to get larger, and Flores had been running that experiment on everything that washed up for a very long time.

The counter-argument was that LB1 was not a species at all but a sick modern human — a pygmy with microcephaly, or later, in other versions, Down syndrome or congenital hypothyroidism. What eroded that case was anatomy that pathology does not produce. The wrist bones, described in 2007, are shaped like those of apes and early hominins and unlike those of any modern human or Neanderthal; the wrist is laid down early in development and is not remodelled by disease. The foot, described in 2009, is long, flat and archless. And there was never a satisfying answer to why a cave would contain a series of individuals all conveniently afflicted with the same rare condition.

The bones go to Yogyakarta

In late 2004 Teuku Jacob, the senior palaeoanthropologist at Gadjah Mada University in Java and the loudest advocate of the microcephaly reading, had the fossils moved to his laboratory in Yogyakarta. They came back in February 2005 damaged — the pelvis crushed, jaw fragments broken, casts taken along the way. The excavation permit was then withheld and the original team did not dig at Liang Bua again until 2007. Two years of the most consequential site in the field sat idle over a dispute about who was entitled to hold the material, which is a category of loss that never appears in a paper.

For ten years the fight was about the anatomy, and the anatomy is the part that held. What gave way was the arithmetic nobody was arguing about.

The layers were read in the wrong order

When work resumed, the team kept opening ground around the original find and the sequence stopped making sense. The answer, published in Nature in March 2016, was an unconformity: the deposits holding LB1 had been cut away by erosion at some point after they were laid down, and much younger sediment had banked up against the surviving remnant. Dates taken a short distance sideways, at what looked like the same level, were therefore dating material that post-dated the skeleton by tens of thousands of years. The eighteen-thousand-year figure was never a laboratory error. It was a correct measurement of the wrong dirt.

Redating the pedestal itself with uranium series on the bones, luminescence on the sediment, argon on volcanic ash and radiocarbon on charcoal put the skeletal remains between about one hundred thousand and sixty thousand years old, and the stone tools attributable to the species from roughly 190,000 to about 50,000 years ago. That deleted the headline. Homo floresiensis had not survived to the edge of the Holocene, had not plausibly inspired the Nage stories about small hairy people in the forest, and had not spent twelve thousand years sharing an island with us. Evidence of modern humans appears in the cave sequence not long after the last sign of the small hominin — close enough to be suggestive, nowhere near close enough to be proof of anything.

Older, smaller, and further inland

The second 2016 paper, in June, came from the So'a Basin in the middle of the island, seventy-five kilometres east of the cave and a short drive from Bajawa. At Mata Menge, a jaw fragment and six teeth from at least three individuals dated to around 700,000 years ago proved to be slightly smaller than the Liang Bua material. In 2024 a fragment of upper arm bone from the same site indicated an adult of about a metre, shorter still than LB1. Stone tools elsewhere in the basin have been dated to around a million years. Whatever this lineage was, it was on Flores, and it was already small, long before the cave in the west had anything to say about it.

What a visitor gets from either place is deliberately modest. Liang Bua is an open cave with a caretaker, a single room of display boards and backfilled trenches marked with numbers; Mata Menge is a fenced slope of grazing land with a field building on it. Neither is a museum and neither pretends to be. The thing worth carrying away is the correction rather than the discovery. Somebody went back and re-dug a trench that had already made everyone's career, on the suspicion that the numbers rather than the bones were the weak part, and the answer cost the find the single fact it was most famous for. That is what the process is supposed to look like, and it very rarely gets to happen in public.