The Hobbit ate the leftovers. A Komodo dragon had already finished dinner.
On Flores, roughly 100,000 years ago, a Komodo dragon ate first. Homo floresiensis, the small-bodied hominin known as the Hobbit, standing about 3 feet 6 inches tall and weighing between 55 and 71 pounds (approximately 25–32.5 kg), got what was left. A taphonomic analysis published July 3, 2026 in Science Advances by Veatch et al. found 54 stone-tool cut marks attributable to H. floresiensis on fossil Stegodon florensis insularis bones from Liang Bua cave, and nearly twice as many Komodo dragon (Varanus komodoensis) tooth marks on the same bones. The dragon marks cluster on meat-rich areas where large muscles attach. The Hobbit’s cut marks are confined to low-meat margins: ribs, phalanges, skull fragments, vertebrae. The sequence is unambiguous: reptile first, hominin second, always.
How you read a 100,000-year-old meal from the scars on a bone
A primary consumer targets zones with the densest muscle attachment. A secondary scavenger inherits the stripped margins and low-value outlying bones. The spatial pattern of marks on a skeleton encodes a feeding sequence as reliably as a timestamp.
Distinguishing Komodo dragon tooth marks, characteristic pits, notches, and furrows, from stone-tool cut marks required a reference library that did not yet exist. Veatch et al. built one. A captive Komodo dragon at Zoo Atlanta was given a domestic goat carcass; after it fed, the team catalogued every resulting mark by type and anatomical location. That reference set was then applied to 3,155 Stegodon florensis insularis bone fragments from Liang Bua, making a bone-by-bone comparison possible for the first time.
No fire, no hunt: what that means for where the Hobbit sits in the family tree
Not one analyzed Stegodon bone showed signs of cooking. Veatch et al. also examined 4,240 or more rodent bones from the same Liang Bua deposits, and none showed burning either, fire, had it been used, would almost certainly have left its mark somewhere across that sample. What had previously been interpreted as charred bone is manganese oxide staining, a mineral process producing black dendritic and spotting patterns visually nearly identical to heat-related carbonization.
The absence of hunting behavior and fire places H. floresiensis outside two capacities otherwise considered foundational to the Homo lineage. Either the Hobbit branched from that lineage before hunting and fire control were established traits, or it descended from a more behaviorally advanced ancestor, with Homo erectus as the leading candidate, and lost those capacities under island isolation on Flores. Veatch et al. raise the question. The bones do not answer it.
The sophisticated Hobbit, skilled hunter, fire-keeper, master butcher, never existed
The prior interpretation rested on three interlocking pieces of evidence. Stone tools confirmed deliberate behavior. Cut marks on a small number of Stegodon bones suggested active butchery of large prey. Apparently charred remains were read as evidence of fire management. Together, they built a picture of a miniaturized but fully Homo-like behavioral package: hunt, butcher, cook.
Veatch et al. contradicts each of the three. Stone tools are confirmed, but the cut marks they produced are outnumbered nearly two to one by Komodo dragon tooth marks and fall on precisely the anatomical zones a secondary scavenger would access. Prior studies had identified cut marks on only three Stegodon bone elements without systematic taphonomic analysis; Veatch et al. examined 3,155 fragments. The charred-bone interpretation is replaced by manganese staining. What remains is a species with stone tools, no fire evidence, and a secondary place at a carnivore’s table.
The one question the bones genuinely cannot answer
The ancestry of H. floresiensis was contested before July 3, 2026, and remains so. Two hypotheses compete. Under the first, H. floresiensis was always a small-bodied, behaviorally limited hominin that diverged from the Homo lineage before hunting and fire were established traits. Under the second, it descended from H. erectus, documented on neighboring islands and understood to have been behaviorally more complex, and underwent profound regression in isolation on Flores. Adam Brumm of Griffith University, an external expert not involved in the study, has noted that island isolation on Flores may have enabled the loss of behaviors otherwise deeply conserved across hominin evolution, but characterizes that conclusion as speculative. The behavioral record from Liang Bua is now clearer. The evolutionary lineage that produced it is not.