San Juan ante Portam Latinam and the Arrowhead Problem

A Basque rock shelter holds 338 Neolithic dead and a few dozen flint arrowheads. How many of them were killed by arrows, and why the answer is contested.

5 November 2026 · Part of The Stone Age — a guide to 12 articles →

High on a ridge above the Ebro valley in the Basque Country, a rock shelter used for burial about five thousand years ago holds the remains of 338 people and a scatter of flint arrowheads. Whether those two facts describe a war is one of the sharpest arguments in the archaeology of Neolithic Europe.

San Juan ante Portam Latinam, near Laguardia in Álava, is not a massacre site in the way Talheim is. Nobody claims that its dead were killed on one day. The question is what a burial place used over several centuries can tell us about conflict, and how much weight a handful of stone points can carry.

A shelter used for the dead

The shelter was used as a collective burial place in the Late Neolithic and Chalcolithic, dated by radiocarbon on the human remains to roughly 3380–3000 BC. The bones were not found as tidy skeletons. As in many collective tombs of the period, the dead were placed over time, earlier remains were disturbed by later ones, and the material came out of the ground commingled. Specialists therefore count individuals from the bones that survive, here principally from skulls, rather than from articulated bodies.

The count that now stands is 338 individuals, of whom 107 could be assigned a sex. Behind those numbers is a difficult judgement: a minimum number of individuals is a lower bound, and the real population using the shelter may have been larger.

What the restudy found

In 2023 a team led by Teresa Fernández-Crespo published a systematic re-examination of the skeletal material. They reported 65 unhealed and 89 healed traumas, 77 of which had not been recorded before, affecting 23.1 per cent of the 338 individuals. The share of individuals with unhealed trauma, the injuries around the time of death, was 10.1 per cent.

The pattern by sex and age is the heart of the case. Adolescent and adult males were the group most affected, and they account for 97.6 per cent of the unhealed trauma and 81.7 per cent of the healed trauma. That is not what accidents or domestic violence would be expected to produce, and the authors read it as evidence that men were exposed to violence and sometimes killed in battle or raids.

The arrowhead problem

Here is the difficulty that gives this article its title. Flint arrowheads were found in the shelter, and some were embedded in bone or lay close to the dead. The authors cite earlier work that counted 52 arrowheads, of which 36 showed minor damage consistent with impact. Arrows are the weapon of choice for a historian who wants to argue for killing at a distance, and they are also the hardest to prove from a skeleton.

The reason is anatomical. An arrow that strikes the torso or the soft tissue may leave no mark on bone at all. A point can be lost in a body that is later buried, and a point that arrives in a grave may have been placed there as a grave good, the property of the dead, not the cause of death. So the authors argue that the proportion of casualties was probably greater than the 10.1 per cent with unhealed bone trauma. Critics can fairly reply that this is an inference from absence, and that an argument which grows in strength precisely where the bone is silent should be handled with care.

This is where the dispute lives. On one reading, the arrowheads and the bone injuries together show repeated conflict: a community burying the casualties of raids over generations. On the other, a long-used burial place accumulates every kind of death, including accident, and individual points need not be weapons. The evidence is disputed, and the restudy does not close the question so much as raise the cost of the sceptical position.

Single event or long conflict?

The distinction matters because it separates this site from the best-known Neolithic massacres. At Schöneck-Kilianstädten in Hesse, 26 people were deposited together after a single episode of killing. The same logic applies to Talheim and to the unburied dead in the ditch at Asparn-Schletz. In each, the dead were deposited at one time, so the violence is a single act.

San Juan ante Portam Latinam is the opposite kind of evidence. Its healed injuries show survivors, and its unhealed ones show deaths, spread across many generations. A restudy of Jebel Sahaba in Sudan made a similar argument for a much older cemetery, where individuals with both healed and unhealed trauma suggest repeated episodes. The Spanish authors set their site beside it for exactly that reason.

For the same reason, our own register of prehistoric collective violence treats this shelter as a stated exception to its single-deposition rule, rates it with medium certainty, and records both readings in the row. A reader who applies the rule strictly should discount it.

Chronology

Why the reading is not settled

Three things make the case hard to close. The first is the commingled bone: assigning an injury to a particular body, or a point to a particular wound, is often impossible. The second is the long span of use. Some four centuries is a long time, and a rate of violence that looks high when summed can be modest when divided across the years. The third is the lack of a local control: there is no comparable burial place nearby against which to measure what an ordinary rate of injury would look like. The later Bell Beaker pottery tradition belongs to the period that follows, so it offers context for the era, not a yardstick.

Compare this with the ambiguous cases elsewhere on the site. At Ofnet and Herxheim the argument is whether the treatment of the dead was violence or ritual. San Juan ante Portam Latinam is different: the injuries themselves are largely agreed, and the argument is about how to add them up.

Conclusion

The shelter shows that violence reached a large share of one Late Neolithic community, and that it fell overwhelmingly on young and adult men. What it cannot show, on its own, is whether this was organised warfare, recurrent raiding, or the slow build-up of a long-used cemetery. The arrowheads make the stronger reading plausible without making it certain. That is a modest conclusion, and it is the honest one.

Frequently asked questions

How many people are buried at San Juan ante Portam Latinam? The 2023 restudy counts 338 individuals, based mainly on skulls. This is a minimum number, so the real total may be higher.

Was it a single massacre? No one argues that it was. The dead were buried over several centuries, c. 3380–3000 BC, and the question is whether they include the victims of repeated conflict.

Why do the arrowheads matter? Embedded and associated flint points suggest killing by arrow, but arrows often leave no mark on bone, and points may also have been placed in graves. The authors argue the casualty rate was higher than the bone evidence alone shows; sceptics note that this rests on missing evidence.

Who were most of the injured? Adolescent and adult males, who account for 97.6 per cent of the unhealed trauma in the restudy.

As an Amazon Associate I earn from qualifying purchases. Affiliate links in Recommended reading — and in Further reading on older articles — may earn Writesides of History a commission at no extra cost to you.