Do Wiltshire hillforts avoid long barrows?

What the map is showing

The visual effect is real and measurable, but it is not avoidance. Hillforts and long barrows follow different rules for where to sit on the chalk: hillforts want a hill that stands up — a scarp with height and prominence — while long barrows are spread much more evenly across the chalk uplands, sitting mainly on the plateaux rather than favouring any particular spot within them. Different site preferences, applied across the same landscape, produce the appearance of mutual avoidance without either tradition ever reacting to the other.

The separation itself is measurable at short range. On a 2 km grid, only eight squares contain both hillfort and long barrow, against roughly 35 expected if the two were distributed at random within the areas they already occupy. Zoom out to 10–15 km, though, and they share the same downs: a typical hillfort lies just 3.1 km from the nearest long barrow — not far in a county this size, and if anything slightly closer than chance would predict for forts scattered across Wiltshire.

Geography explains most of that pattern. Both traditions favour chalk — 93% of barrows, 71% of hillforts — but they choose different seats on it. Barrows sit back on plateaux and gentle rises, typically 147 m up and only about 7 m more prominent than the surrounding land: a broad, even spread across the high chalk. Hillforts, by contrast, cluster on the scarps: 177 m up and about 41 m more prominent, a preference for standing ground rather than open plateau. Woodland explains little of the residual pattern: barrows sit away from surviving ancient woodland simply because they sit on open chalk, not because they were avoiding trees, and they show no avoidance of modern woodland either.

That the two traditions never actually reacted to each other is confirmed by their timing and by the exceptions. Early Neolithic tombs and Iron Age hillforts were not choosing around each other in a single planning meeting — roughly three thousand years separate them. Several forts were in fact built on or beside a long barrow: Bratton Castle (88 m away), Oldbury Castle (245 m), Vespasian's Camp and Old Sarum (both around 600 m). That is the opposite of a taboo.

The six causewayed enclosures offer a useful check, since they belong to the same centuries as the long barrows rather than the hillforts. They sit close to the barrows (typically 0.9 km away) rather than in the empty-looking hillfort gaps (typically 3.2 km). Windmill Hill, Robin Hood's Ball and Larkhill fall within the Avebury and Stonehenge clusters; Knap Hill is 900 m from Adam's Grave. Two later acquired a hillfort on the same hill — Rybury, under Rybury Camp, and Whitesheet Hill, 650 m from White Sheet Castle. Six sites are too few to support the same randomisation tests as the main dataset, but the pattern is consistent with Neolithic gathering rather than burial.

Randomisation confirms the geographic reading. Control for each tradition's own clustering — sliding the hillforts around the county, or redrawing them within their usual districts across hundreds of randomised maps — and 3.1 km is an ordinary gap. Drop the hillforts into the barrow clusters instead, and they would sit roughly twice as close. Long barrows are tombs, not villages; hillforts are not a map of every Iron Age home. Overlaying the two is the wrong test for settlement continuity — but a fair picture of two different uses of the Wiltshire chalk.

Same chalk, different seats. Hillforts hug the scarps; long barrows sit on the plateaux. The pattern is real at walking distance. It is not a rule of keeping apart.