Reconstruction
Boxgrove lies at the foot of the South Downs in West Sussex, on what was a coastal plain below a chalk cliff at the end of the Marine Isotope Stage 13 interglacial, about 480,000 years ago. Quarrying at Eartham Pit exposed a sequence of marine sands, intertidal silts and freshwater deposits. Within it, a test pit designated GTP17 cut a single buried land surface.
On that surface lay the dismembered skeleton of a large adult female horse, Equus ferus, and around it tight clusters of flint. The clusters are knapping episodes: individual handaxes made on the spot, each scatter representing one person kneeling and working, with the finished biface removed and the debitage left behind. The bones carry cut marks made before any carnivore touched them, so the butchers had the carcass first, not as scavengers of a lion kill. Long bones were split for marrow. Fragments of horse bone were themselves used as hammers and retouchers — the earliest bone tools known from Europe.
The resolution is what makes the site extraordinary. Fine intertidal silt sealed the surface almost immediately, and analysis of the scatters is consistent with a single episode, plausibly a single day's work by a group large enough to keep scavengers off a horse for as long as it took.
One find from GTP17 drove the site's public reputation and has now been reinterpreted. A horse scapula preserved a rounded, semicircular fracture read as the entry wound of a thrown or thrust spear, and it circulated for three decades as the oldest direct trace of projectile hunting. A 2025 replication study drove wooden spears into scapulae and also struck them with hammerstones, then compared fracture geometry. The curvilinear break on the Boxgrove specimen matches percussion damage from a hammerstone driven into the bone for grease and marrow, and falls outside the range produced by spear impacts. The hunting argument for Boxgrove survives on other grounds — early carcass access, a prime adult animal, an assembled group — but not on this piece of evidence.
The people are known from very little: a partial left tibia, robust and thick-walled, and two lower incisors from a different individual found lower in the sequence. Both are usually filed under Homo heidelbergensis. A direct comparison with the large Sima de los Huesos sample from northern Spain placed the tibia outside the range of variation there while finding the incisors a close match, which raises the awkward possibility that the two Boxgrove finds do not sample the same population and leaves the species label doing more work than the fossils can support.
What is not in doubt is the meal: a horse, taken in the open, opened by people who knew the sequence of cuts, worked down to the marrow and then to the bone itself.
Historical context
Britain was a peninsula of Europe, reachable on foot across the ground where the Channel now runs, and occupied only during warm intervals. Hominins had been here far earlier — flint artefacts at Happisburgh and Pakefield in East Anglia predate Boxgrove by several hundred thousand years — but the record is a sequence of arrivals and abandonments rather than continuous residence. The coastal plain below the Boxgrove cliff carried rhinoceros, red and giant deer, bison, bear and lion; the mare butchered at GTP17 came from a herd sharing that plain with all of them. Elsewhere in the same broad interval, Acheulean toolmakers were working the Kenya Rift at Olorgesailie and Homo erectus occupied the Zhoukoudian fissure in northern China, while in northern Spain bodies were accumulating in the shaft at Sima de los Huesos. Habitual fire use and the wooden spears of Schöningen both lie roughly 180,000 years in the future.
Evidence
Archaeological
StrongGTP17 preserves an undisturbed single land surface with in-situ knapping scatters, refittable debitage and spatially discrete working areas, sealed by fine intertidal silt — among the highest-resolution Lower Palaeolithic contexts in Europe.
Zooarchaeological
StrongThe Equus ferus carcass shows systematic disarticulation, filleting cut marks stratified beneath carnivore tooth marks, and deliberate long-bone breakage for marrow. Hominin-first access is directly demonstrated.
Preparation method
StrongDefleshing, disarticulation, marrow extraction by hammerstone percussion and the secondary use of horse bone as hammers and retouchers are all documented on the bones themselves.
Dating
ModerateThe Boxgrove sequence is assigned to the end of MIS 13, about 480 ka, on combined biostratigraphic, sedimentary and amino-acid grounds. There is no direct radiometric date on the land surface, so the age is an interval assignment rather than a measured one.
Food identification
StrongThe species is securely identified as wild horse, Equus ferus, a single large adult female; the meal is not reconstructed from proxies.
Historical interpretation
ModerateHunting rather than scavenging remains the better-supported reading, resting on early carcass access, prime-adult prey and the group size implied by the knapping clusters. The specific killing method is unknown now that the scapula perforation has been reattributed.
Visual reconstruction
LimitedAny depiction of a spear wound, a hearth or cooked horse meat at Boxgrove goes beyond the evidence: no fire is recorded at the site and the spear wound has been reinterpreted as percussion damage.
Sources
- 1.Milks, A.; Carr, D.J.; Godhania, et al. (2025). A reanalysis of the ‘perforated’ scapula from the Boxgrove Horse Butchery Site. Quaternary International 749: 109995. doi:10.1016/j.quaint.2025.109995Scientific literature
- 2.Roberts, M.B.; Stringer, C.B.; Parfitt, S.A. (1994). A hominid tibia from Middle Pleistocene sediments at Boxgrove, UK. Nature 369(6478): 311–313. doi:10.1038/369311a0Scientific literature
- 3.Hillson, S.W.; Parfitt, S.A.; Bello, et al. (2010). Two hominin incisor teeth from the middle Pleistocene site of Boxgrove, Sussex, England. Journal of Human Evolution 59(5): 493–503. doi:10.1016/j.jhevol.2010.06.004Scientific literature
- 4.Lockey, A.-L.; Rodríguez, L.; Martín-Francés, et al. (2022). Comparing the Boxgrove and Atapuerca (Sima de los Huesos) human fossils: Do they represent distinct paleodemes?. Journal of Human Evolution 172: 103253. doi:10.1016/j.jhevol.2022.103253Scientific literature
- 5.Hosfield, R. (2021). Review of Pope, M., Parfitt, S. & Roberts, M. (2020) The horse butchery site: a high resolution record of Lower Palaeolithic hominin behaviour at Boxgrove, UK. Antiquity 95(383): 1348–1350. doi:10.15184/aqy.2021.104Scientific literature
- 6.Pope, M.; Parfitt, S.; Roberts, M. (2020). The Horse Butchery Site: A high-resolution record of Lower Palaeolithic hominin behavioural organisation at Boxgrove, UK. SpoilHeap Publications, Woking (SpoilHeap Monograph). discovery.ucl.ac.uk/id/eprint/10114053/Scientific literature
- 7.Roebroeks, W.; Villa, P. (2011). On the earliest evidence for habitual use of fire in Europe. Proceedings of the National Academy of Sciences 108(13): 5209–5214. doi:10.1073/pnas.1018116108Scientific literature
Limitations
- majorthe site's most-cited find is no longer read as a spear wound
The semicircular perforation in the horse scapula from GTP17 was interpreted for three decades as a projectile impact and is still the standard illustration of the earliest direct evidence for spear hunting. Milks et al. (2025, doi:10.1016/j.quaint.2025.109995) replicated both wooden-spear impacts and hammerstone percussion on scapulae and found the Boxgrove curvilinear fracture metrically matches percussion damage from breaking into the bone for grease and marrow, outside the range of the spear impacts. Any text, image or hotspot depicting a spear wound on this animal is out of date.
- moderate'a hominin tibia' invites a species attribution the fossils cannot carry
The tibia (Roberts, Stringer & Parfitt 1994, doi:10.1038/369311a0) and two incisors (Hillson et al. 2010, doi:10.1016/j.jhevol.2010.06.004) are referred to Homo cf. heidelbergensis, a taxon whose validity is itself disputed. Lockey et al. (2022, doi:10.1016/j.jhevol.2022.103253) found the tibia falls outside the Sima de los Huesos range while the incisors closely resemble that sample, and the two Boxgrove finds come from different levels — so they may not even represent the same paleodeme. Naming a species here overstates the evidence.
- moderatethe framing omits the fact that makes the site exceptional
Describing GTP17 as 'an unusually detailed snapshot' understates it. The surface was sealed by intertidal silt almost immediately, the flint scatters resolve to individual knappers, cut marks underlie carnivore tooth marks, and the episode may represent a single day. Horse bone from the same carcass was reused as hammers and retouchers — the earliest bone tools known in Europe. The temporal resolution and the bone tools, not the horse itself, are the headline results.
- minorthe date range implies a duration the deposit does not have
A 500,000–480,000 BCE window suggests twenty millennia of activity, but the butchery surface is a single-episode context. Boxgrove is assigned to the end of MIS 13 at about 480 ka on biostratigraphic and sedimentary grounds; '500,000' is a rounded upper bracket for the wider Slindon and Eartham sequence, not a dated start for the butchery event, and no direct radiometric date exists for the land surface.
- minorno fire is recorded at Boxgrove despite the epoch label
The site sits in an epoch called The Fire Revolution but has produced no hearths, ash or burnt bone. Roebroeks & Villa (2011, doi:10.1073/pnas.1018116108) argue hominins colonised northern latitudes without habitual fire, with regular fire management appearing only after about 400,000–300,000 years ago. The horse was processed raw as far as the evidence goes.
- minorhunting is well supported but the kill method is unknown
Early carcass access, a prime adult mare and the group size implied by multiple simultaneous knapping stations still favour hunting over scavenging, and Milks et al. (2025) say so explicitly. But with the scapula perforation reattributed, no evidence at Boxgrove indicates how the animal was killed. Depicting a specific weapon or technique is reconstruction, not record.
