What the Ramparts Were For

ABetween about 800 BC and the Roman invasion of AD 43, communities across Britain raised banks and ditches around some three thousand three hundred hilltops, spurs and promontories. Antiquarians of the eighteenth century named these enclosures hillforts, and the name has done a great deal of quiet work ever since, because it settles in advance the question of what they were for. Some are vast: the circuit at Cadwr Ring in the Welsh Marches encloses nineteen hectares, and its outer bank still stands six metres above the floor of its ditch. Others enclose less than half a hectare and would have held two households. What unites them is the labour. One recent estimate puts the earthmoving at a single medium-sized enclosure at forty thousand person-days, a commitment no community makes absent-mindedly. Whether that effort was spent out of fear, or on something else entirely, has divided later prehistorians for four decades.

BMarged Trevalyn, who studies settlement economics at the Institute for Later Prehistory in Aberystwyth, holds that the name has misled the field and that no quantity of fresh fieldwork will help until it is set aside. Between 2009 and 2019 her group re-examined 214 enclosures in the Welsh Marches, measuring every excavated interior and classifying every structure standing within it. Storage accounted for 68 per cent of roofed floor area — grain pits, raised granaries on four posts, stock pens — against 21 per cent at contemporary open farmsteads in the same region. Fewer than one enclosure in ten preserved a burnt gate or any other trace of assault. Over the same decade the number of enclosures recorded by aerial photography and lidar survey nearly doubled, an expansion that reads in synthesis chapters as a swelling landscape of war. What did not grow was the proportion showing destruction. The question, on her account, is what a rampart could physically hold rather than what it could repel.

CThe account yields a prediction, and Trevalyn has tested it. If storage rather than defence set the scale of the earthworks, then rampart volume ought to track the extent of workable farmland within an hour's walk, not the distance to the nearest comparable neighbour. Across 76 excavated circuits her group found a correlation of 0.58 with arable catchment and of 0.11 with distance to a rival. The result was confined to univallate enclosures; where the circuit was multivallate, rampart scale tracked intervisibility between neighbouring forts instead, and the farmland signal vanished altogether. Trevalyn therefore treats the earthwork as a form whose purpose shifts with its plan, not as a single institution that one label could cover. Where later ploughing has stripped away the ancient soils — the case at perhaps a quarter of sites — she accepts that the catchment cannot be reconstructed and that no test can be run at all.

DAnton Reuss, who works on conflict archaeology at the Institute of Field Archaeology in Durham, answers that a granary is not an alternative to a stronghold but a reason for one. Weapons cluster where the earthworks are heaviest: excavation of the eastern gate at Cadwr Ring in 1974 produced a single cache of 11,400 slingstones, against 260 from every excavated lowland farmstead in Britain combined. Human remains tell the same story. In the cemetery outside that gate, 17 of 63 adults carried unhealed blade injuries, against 2 of 90 in the lowland series of the same century. Reuss concedes two things. His evidence comes overwhelmingly from places chosen for excavation in the 1930s because their ramparts looked dramatic, so the record is filtered before anybody counts it. And the chronology once claimed for the burning has dissolved: radiocarbon redating since 2016 places most destruction between 400 and 200 BC, generations before the Roman armies its excavators blamed.

EThe argument left the seminar room some time ago. Since 2018 several planning authorities have obliged developers to fund full excavation of any enclosure interior threatened by building, on the grounds that the interiors, and not the banks, hold what survives of the evidence. Reuss points out, uncomfortably for his own position, that the sites he leans on are the ones antiquarians dug earliest and recorded worst, so the evidence that most favours his case is the evidence least likely to satisfy a modern assessment. Trevalyn answers that a taste for the dramatic created both problems at once, since the appetite that emptied the great ramparts in the 1930s also fills the literature with places that can no longer be re-examined. Both regret one habit above all: that the field still reports enclosed area rather than storage volume or labour cost, a figure that flatters every party to it.

FA settlement of sorts is emerging in which the two accounts are assigned to different questions rather than ranked against each other. For anything about the way a prehistoric community organised labour, the earthworks remain unmatched and the quarrel about warfare is beside the point, since forty thousand person-days of digging is a social fact whether or not anyone was ever besieged. For anything about how often people fought, the enclosures must be corrected against the skeletal series before they can be used at all, and where no cemetery has been found beside them, which is the majority of cases, they cannot be used for that purpose at all. The unresolved question is whether the meaning of a rampart itself altered across the period as scattered farms gave way to nucleated settlements, which would make the old label wrong in a second way; no excavated sequence yet spans enough of the interval to settle it. What both researchers reject is the assumption behind the earliest surveys, namely that a bank and a ditch can only mean fear.