The small human skeleton uncovered at Liang Bua cave on Flores, Indonesia, in 2003 came from an adult barely a metre tall. Its exceptionally small skull attracted immediate attention, but the rest of the skeleton was equally puzzling: its bones looked unlike those of a modern human child. Researchers announced a new species, Homo floresiensis, in 2004. The popular name “hobbit” reflected both its height and an obvious comparison with Tolkien’s characters. Researchers still had to explain how this human relative reached Flores and survived there until the Late Pleistocene.
The best-preserved individual, LB1, is now dated to about 60,000 years ago under the cave’s revised chronology. She combined traits familiar from the human lineage with much older-looking features: a very small braincase, short legs and feet unusually long in proportion to the body. Excavators also found animal remains and stone tools in the wider sequence. The excavators had to separate traces left by these hominins from later modern human activity, animal remains and material moved by the cave sediments.
Liang Bua is only part of the story. At Mata Menge in central Flores, fossils place remarkably small hominins on the island around 700,000 years ago, while stone artefacts indicate that hominins arrived about one million years ago. Studies published in 2025 and 2026 have prompted further reassessment of how the species obtained food, whether it controlled fire and why it disappeared from the cave. These discoveries extend the history of the island’s hominins far beyond Liang Bua, although the record still has long gaps.
The discovery at Liang Bua
In western Flores, limestone hills and river valleys surround Liang Bua, a cave already known to researchers before the excavations of the early 2000s. A joint Indonesian-Australian project led by Michael Morwood, with Indonesian archaeologists including R. P. Soejono and Thomas Sutikna, began revealing unexpected hominin material there. The team recovered the partial skeleton designated LB1 in September 2003. Peter Brown and colleagues formally described the species in 2004, naming it Homo floresiensis. The name places the fossils in the genus Homo and identifies Flores as the place of discovery.
LB1 includes much of the skull and portions of the pelvis, legs, arms, hands and feet, allowing investigators to compare far more than an isolated tooth or jaw. Dental and skeletal development show that the individual was an adult. The original estimate gave a height of approximately 1.06 metres and a cranial capacity of about 380 cubic centimetres; later scans yielded a figure closer to 426 cubic centimetres. The 380 measurement remains widely quoted, although it is not agreed to that precision by every subsequent study. Either estimate is far below the living human average. The 2004 species description therefore described an anatomical combination whose significance went well beyond stature.
Other bones from Liang Bua represent additional individuals, even though none approaches LB1 in completeness. Those additional individuals mattered when critics proposed that LB1 was a modern human with a disorder. Illness and developmental variation can certainly affect one skeleton. Finding related anatomical patterns in more than one individual, and much older small-bodied hominins elsewhere on Flores, makes a purely individual explanation less convincing. That does not remove the remaining uncertainty about which branch of the human family these fossils represent.
What the bones tell us about the hobbit’s body
LB1’s skull has a low vault, prominent brow ridges and no chin of the modern human kind. The face and teeth show a mixture of characteristics. Certain features recall Homo erectus, an earlier human species known across parts of Asia, while others are unusual even within Homo. It would be misleading to imagine an ordinary modern human reduced in size. Shrinking the dimensions of a body cannot, by itself, account for every distinctive feature of its anatomy.
The same difficulty appears below the head. LB1’s pelvis was broad, its collarbones comparatively short and its shoulders differently placed from those of living people. Three wrist bones were particularly informative. Matthew Tocheri and colleagues showed in a 2007 study that their arrangement resembled structures found earlier in hominin evolution, rather than the derived configuration shared by Neanderthals and modern humans. Their findings challenged the identification of LB1 as a diseased Homo sapiens. In anatomical discussions, “primitive” means resembling an earlier condition; it is not a judgment about the person’s abilities.
The foot reconstructed by William Jungers and colleagues was long in relation to the lower leg. LB1’s big toe lined up with its neighbours, as it does in a human adapted to walking rather than an ape with grasping feet. The overall proportions were nonetheless substantially different from ours. In the 2009 anatomical study, the researchers described a biped whose gait probably differed from that of living humans. Nothing in the bones establishes that those feet evolved for climbing cliffs or evading Komodo dragons. Such stories remain conjecture without independent evidence.
A small braincase does not offer a direct measure of intellect. An endocast, made from the inner shape of the skull, can suggest brain shape and the placement of some surface structures, but it cannot tell us what LB1 thought, whether she spoke or how her community was organised. Early work suggested that aspects of the brain might have been reorganised in ways compatible with complex behaviour. That interpretation remains debated. Securely associated archaeological traces offer better evidence of the hominins’ activities than brain volume alone.
Was Homo floresiensis a separate species?
The suggestion that LB1 was a modern human with a medical condition emerged soon after its discovery. Microcephaly, a disorder involving unusually small brain development, and other growth disorders were proposed. The objections arose partly because the combination of an adult’s tiny brain and such a late date upset established expectations about human evolution. But the proposed diagnosis had to explain the whole skeleton, including its feet, wrists and pelvis. A diagnosis explaining the skull but not the rest of the anatomy would remain incomplete.
The wrist configuration, teeth, feet and pelvis resisted explanation as one modern human disorder. The discovery of small-bodied fossils at the much older Mata Menge site added another problem for that hypothesis: the unusual proportions were present on Flores long before LB1 lived. Most palaeoanthropologists now accept Homo floresiensis as a distinct species. Its recognition as a species does not settle its ancestry, which remains disputed.
There are two principal explanations for its ancestry. An early population related to Asian Homo erectus may have settled Flores, subsequently becoming much smaller over many generations. Alternatively, the ancestors may have belonged to an older, already small-bodied lineage closer to early Homo. A 2017 phylogenetic analysis supported the latter, more primitive relationship, while subsequent Flores fossils have made an Asian erectus-related origin more plausible. Both hypotheses rest on anatomical comparisons among incomplete fossils. No ancient DNA from the species has yet established its genealogy.
How the dates changed from 18,000 to 60,000 years ago
When LB1 was first announced, it was said to be about 18,000 years old, with other evidence initially suggesting survival to approximately 12,000 years ago. These dates raised the possibility of an extremely late-surviving human relative in a world already occupied by modern humans across much of Southeast Asia. The apparent youth of the remains helped make the discovery famous. Later excavation revealed that the dates relied on a mistaken reconstruction of deposits that had been eroded and rearranged within the cave.
Cave floors are not always orderly sequences in which each layer lies neatly above an older one. Erosion at Liang Bua stripped away deposits in some areas, bringing sediments of different ages into misleading proximity to the hominin remains. A date obtained beside a bone is useful only if researchers can establish that both belong to the same depositional sequence. More excavation, combined with several dating methods, allowed the team to trace the fossil-bearing layers and distinguish them from surrounding, younger sediments.
In their revised chronology published in 2016, Thomas Sutikna and colleagues placed the sediments containing H. floresiensis skeletons at roughly 100,000–60,000 years old. Tools attributed to the species belonged to the wider span of about 190,000–50,000 years ago. Those two ranges describe different categories of evidence: the most recent stone tools are not the youngest skeleton, and a site’s last recovered tool does not fix a species’ final day. It tells researchers only how late the presently excavated evidence reaches.
Another study of Liang Bua’s deposits found a shift in the stone used for tools at around 46,000 years ago, including a greater reliance on chert. Sutikna and colleagues interpreted this change as a possible sign of the arrival of Homo sapiens at the cave. The proposal remains cautious, since the kind of stone selected is not a reliable species label. Later modern human activity and remains demonstrate a subsequent change in occupation, but there is still no excavated evidence showing the two human populations meeting at Liang Bua.
Mata Menge: small hominins 700,000 years before LB1
To learn how far back this small-bodied lineage extended, researchers turned to Mata Menge in the So’a Basin of central Flores. In deposits about 700,000 years old, excavators found part of a lower jaw and six separate teeth belonging to at least three individuals. Several dimensions are comparable with, or smaller than, those of LB1 and her contemporaries. The 2016 description of these fossils also identified anatomical similarities suggesting connections to early Asian Homo erectus. By then, exceptionally small hominins had already existed on Flores for hundreds of thousands of years before the Liang Bua individuals.
Mata Menge’s setting 700,000 years ago was an open landscape rather than a limestone cave. A small stream deposited the sediments that eventually covered the fossils, and volcanic material helped researchers establish their age through radiometric dating. The sediments and animal remains indicate relatively dry grassland with wetlands nearby. A linked geological and archaeological study documented stone artefacts and the geological context. These findings support a long hominin history on Flores, although they cannot demonstrate uninterrupted occupation by one community or an unchanged culture between Mata Menge and Liang Bua.
A further discovery, reported in 2024 by Yousuke Kaifu and colleagues, included an exceptionally small fragment of adult upper-arm bone and two additional teeth from Mata Menge. Microscopic examination established that the arm fragment belonged to an adult, despite its size. The researchers estimated that the complete bone would have been about 9–16 per cent shorter than LB1’s, implying a height near one metre. In their open-access 2024 paper, they argued that body-size reduction was already pronounced 700,000 years ago and favoured descent from early Asian Homo erectus.
Even with the newer fossils, a substantial interval separates the earliest island tools from the Mata Menge skeletons. The teeth, jaw fragments and single piece of upper arm cannot reveal the size of the original settlers, the rate at which bodies became smaller or the number of distinct populations involved. The fossils establish that small-bodied adults existed 700,000 years ago, but the evolutionary process behind that size remains uncertain. That limitation should remain clear whenever island dwarfism is offered as an explanation.
Why Flores was an unusual place for human evolution
Flores belongs to Wallacea, the islands lying between continental Asia and the former Australia-New Guinea landmass. At times of low sea level, broad areas around Java and other western islands were exposed, connecting stretches of land that are separated today. Flores, however, remained across deep-water passages. Whatever the species of the earliest settlers, they or their predecessors had to cross the sea. Their archaeological remains give no direct indication of whether a crossing was intentional or the result of drifting.
On islands, the pressures of limited resources and unfamiliar combinations of predators and competitors can favour changes in body size. Large mainland animals sometimes evolve smaller island forms, a tendency called insular dwarfism. It is neither inevitable nor caused by a single factor. Diet, climate, reproduction, predators and the number of founders may all affect the outcome. Lower energy requirements would be one possible advantage of a smaller hominin body, but it is unjustified to assume that every unusual LB1 bone was a tailored solution to life on Flores.
Flores had its own unusual fauna: the extinct elephant relative Stegodon, giant rats, Komodo dragons and the immense marabou-like stork Leptoptilos robustus. The animal species occur in deposits of different ages, so they were not necessarily all present together. The small Stegodon florensis insularis shared the later Liang Bua environment with hominins, whereas other stegodont forms belonged to earlier periods. Chronology matters here because describing all these animals as one timeless island community would produce a misleading picture of the ecology.
The giant stork approached two metres in height. It was once reconstructed as flightless, until additional wing fossils forced that picture to change. As reported by the Natural History Museum in 2022, the bird could fly and may have joined other scavengers at carcasses. Its presence offers no evidence that it habitually hunted small hominins. Komodo dragons likewise leave no proof of special cliff-climbing feet, permanent cave guards or coordinated defences among the hominins. Such interpretations would require archaeological support beyond sharing the island with dangerous animals.
Stone tools and what they reveal about behaviour
The Flores stone industry consisted largely of flakes struck from stone cores. A core is the stone from which useful sharp pieces are removed, and a flake may cut flesh or scrape material without being shaped into a formal handaxe. Archaeologists reconstruct production by examining the striking scars, material choices and locations where objects were left. When the same kinds of removals recur, they may indicate an acquired skill, even if the resulting objects seem simple compared with tools made by later humans.
The 2010 study by Adam Brumm and colleagues dated artefacts at Wolo Sege in the So’a Basin to about one million years ago. Those artefacts establish that hominins reached Flores long before the Mata Menge individuals lived, although they do not identify the toolmakers. The tools from Mata Menge, about 700,000 years old, and the much younger Liang Bua assemblages have broad similarities in how flakes were produced. A straightforward technique may remain useful over immense stretches of time; similar flakes alone cannot demonstrate an unchanged, continuously transmitted tradition.
Liang Bua’s fossil-bearing deposits contain stone tools together with animal bones. The association establishes that hominins were working stone in this setting, but tells us little on its own about specific daily activities. People may visit caves without living in them continuously; animals can die there or carry carcasses inside, and water may move material between layers. The presence of tools beside large animal bones cannot, without examination of those bones, prove deliberate hunting, cooking or permanent settlement.
Did the hobbits hunt Stegodon or use fire?
For years, Homo floresiensis was frequently presented as a hunter of dwarf elephants that cooked its prey. Stone flakes, Stegodon remains and apparently blackened bones at Liang Bua appeared compatible with that scenario. Testing that picture required identifying what marked the bones and what made some of them dark. Teeth and stone cutting edges leave different kinds of damage, while minerals in cave sediments can blacken bone without fire.
In a study published in Science Advances in July 2026, E. Grace Veatch and colleagues tested those claims. They compared thousands of bones from the cave with experimentally produced Komodo-dragon feeding marks and examined which parts of animals were present. Dragon tooth marks occurred especially on the meat-rich portions of Stegodon carcasses. Cuts attributed to hominins were much less frequent and were more often found on parts of lower food value. Scavenging after dragons had fed fits the evidence better than primary large-animal hunting, though it cannot specify who reached every carcass first.
The July 2026 team also assessed the supposed signs of fire. None of the 4,240 rodent bones sampled from Unit 1, associated with H. floresiensis, showed fire damage. In a later unit associated with modern humans, roughly one fifth of the sampled bones were burned. Manganese deposits offered a better explanation for some earlier descriptions of charred bone. The researchers therefore found no convincing evidence for controlled fire in the Liang Bua deposits associated with H. floresiensis. The finding is restricted to the excavated context; it cannot show that this species never used or encountered fire anywhere.
The stone flakes and animal remains support hominin toolmaking and access to carcasses. Evidence for organised dwarf-elephant hunts or cooking is much weaker. The evidence also discourages broad conclusions about ancestry from supposed behavioural sophistication. A population might obtain calories by scavenging, gathering plants or catching smaller animals without routinely killing large prey. Nor can the absence of confirmed big-game hunting measure cooperation or intelligence. Giant rats are common in the Liang Bua deposits, but their precise mode of acquisition remains uncertain.
Could a small-brained hominin cross the sea?
The arrival of any hominin population on Flores required passage across the sea. During lower sea levels, Java and mainland Asia could be linked by land, while deep channels still separated Flores. Floating vegetation carried by storms offers one possible explanation for an accidental arrival. Deliberate water crossings have also been proposed, particularly because settlement requires a population large enough to reproduce. There is no surviving boat or raft, and tools found on Flores do not reveal whether their makers planned a voyage or were carried there inadvertently.
Wolo Sege establishes a hominin presence by roughly one million years ago, and the fossils at Mata Menge place very small-bodied hominins there 700,000 years ago. These discoveries do not show that the first arrivals were as small as the later individuals. They also cannot tell us whether all subsequent hominins descended from one founding event or whether several groups crossed at different times. Either prolonged isolation after a single colonisation or multiple arrivals remains compatible with the evidence.
Claims about early seafaring often treat crossing water as certain proof of advanced foresight. A group could survive an accidental drift, and deliberate crossings would not have required modern sailing equipment. The fossils and artefacts establish dispersal across a marine barrier in island Southeast Asia, but not the decisions behind it. Judging the possibilities solely by the hominins’ estimated brain size would obscure the very behaviour archaeologists are trying to understand.
When and why did Homo floresiensis disappear?
At Liang Bua, the most recent identified skeletal remains of H. floresiensis are around 60,000 years old. Tools attributed to the species extend to about 50,000 years ago. Neither date is a directly recorded extinction. Modern humans had reached other parts of the region in the Late Pleistocene, and evidence of later Homo sapiens activity survives in the cave. But no securely dated layer or fossil establishes that both human populations lived there together. Stories of warfare, competition or direct encounters therefore go beyond the excavated record.
Changes in climate may help explain why the hominins ceased using Liang Bua. In a 2025 study in Communications Earth & Environment, Michael Gagan and colleagues analysed a nearby Liang Luar stalagmite and isotopic evidence in Stegodon teeth. Their reconstruction suggests that average annual rainfall declined from about 1,560 to 990 millimetres between 76,000 and 61,000 years ago. Summer rainfall became especially scarce around 61,000–55,000 years ago, while Stegodon also became less common in the Liang Bua sequence. Reduced water and changes in the movement of prey could have driven hominins out of the cave area.
A drying climate is a possible cause of local departure, though it does not demonstrate that drought killed the species. People might have moved to better-watered parts of Flores and persisted beyond the last known occupation at Liang Bua. Vegetation change, fewer prey animals, volcanic eruptions or contact with other human populations could have mattered at different times. There is too little evidence to rank those suggestions definitively. The giant stork and small Stegodon also disappear within the wider interval, pointing to ecological disruption without establishing a single, island-wide catastrophe.
The 2025 rainfall reconstruction and 2026 study of animal bones address different questions. One investigates changes to the availability of water and prey; the other examines access to carcasses and possible fire use. Their results are compatible with a stone-tool-using population whose fortunes depended partly on the local environment, but neither reconstructs every aspect of daily life or identifies the last members of the species on Flores. The end of occupation at Liang Bua cannot therefore be equated with extinction across Flores.
What about the people living on Flores today?
People living on Flores today are Homo sapiens. Some communities around Liang Bua are relatively short on average, which led to suggestions that they might be related to the ancient small-bodied hominins. A 2018 genomic study by Serena Tucci and colleagues detected no evidence in the modern individuals sampled for ancestry from an otherwise unidentified archaic population. It did identify genetic signals consistent with short stature evolving among modern humans, together with the established presence of Neanderthal and Denisovan ancestry. Similar height alone cannot settle an ancient ancestry question.
No usable genome has yet been recovered from H. floresiensis, leaving researchers without a direct ancient DNA sequence against which to test living populations. The findings from 2018 therefore do not support claims of demonstrable hobbit ancestry, although they cannot exclude every possible interaction in all places and periods. Two groups separated by tens of thousands of years can independently evolve similar body size, especially under island conditions. Similar stature does not establish descent from the ancient island hominins.
What Homo floresiensis changed about human evolution
Before the Flores fossils came to light, the later record of humans in Asia was frequently pictured through the spread of Homo erectus, followed eventually by the expansion of Homo sapiens. LB1 added a small-brained, small-bodied lineage with a distinctive mixture of old and newer anatomical characteristics that lasted into the Late Pleistocene. Mata Menge then showed comparable small body size hundreds of thousands of years earlier, and the oldest tools established occupation of Flores around one million years ago. Human evolution in the region had followed more than one course.
Neither brain volume nor the apparent simplicity of flaked stone should determine what researchers think the species could do. The remains show that small-brained hominins made tools and survived for a long time; they do not justify imagining all the behaviours familiar from modern humans. Re-examining carcass damage and possible burning in 2026 altered once-popular interpretations, while the 2025 rainfall study drew evidence from a cave environment beyond the fossil deposit itself. Separating hominin activity from animal feeding and geological processes has changed the interpretation of the site alongside the fossil discoveries.
The identity of the first population to cross the water remains uncertain. It might have belonged to an early Homo erectus lineage or a more ancient branch of the human family. We also cannot yet establish how rapidly body size declined, how many populations reached the island or whether the species survived elsewhere after its last appearance at Liang Bua. More fossils and securely dated deposits may narrow those gaps. The history of Homo on Flores already establishes that very different human bodies, and distinct evolutionary paths, persisted for hundreds of thousands of years.
Discussion
Comments