Roman Scotland: Conquest, Frontiers and the Limits of Empire

1. The Flavian invasion: Agricola and Mons Graupius

The Roman governor Gnaeus Julius Agricola pushed Roman power deep into what is now Scotland during the early AD 80s. Our only detailed narrative comes from Tacitus's Agricola — written by Agricola's own son-in-law, and therefore a source with privileged access but also a clear literary and political agenda. Tacitus dates events by campaigning season rather than calendar year, so even the year of Mons Graupius, conventionally placed around AD 83–84, should be treated as approximate.

By his third season Agricola's forces had reached the Tay; forts followed the advance, and the Forth–Clyde isthmus — the narrowest crossing of Britain — was fortified as a defensible line, a strategic logic the Romans would return to sixty years later. The campaign built toward a set-piece battle, Mons Graupius, where Tacitus has Agricola defeat a coalition of northern Britons led by a chieftain named Calgacus. No battlefield has been securely identified despite many proposed locations, and Tacitus's casualty figures — heavy Caledonian losses, minimal Roman ones — conveniently complete his portrait of Agricola as the ideal commander, so they should not be repeated as verified fact.

The most quoted line from this whole episode comes from Agricola 30, within the speech Tacitus attributes to Calgacus before the battle: "auferre trucidare rapere falsis nominibus imperium, atque ubi solitudinem faciunt, pacem appellant" — "to plunder, slaughter and steal they give the lying name of empire; they make a desert and call it peace." This is a landmark of ancient political rhetoric — but it is Tacitus's own composition, a convention of Graeco-Roman historiography in which historians wrote speeches appropriate to the occasion. As standard scholarship puts it, here "we are hearing the words and thoughts of Tacitus, not Calgacus" — evidence for Roman elite thinking about empire, not a transcript of Caledonian political philosophy.

2. Inchtuthil and organised withdrawal

The clearest evidence that Rome intended more than a raid is the legionary fortress built at Inchtuthil, Perthshire — a complete base of some 21.5 hectares, associated with the Twentieth Legion (Legio XX Valeria Victrix), capable of housing an entire legion of around 5,000 men plus workshops and stores. Alongside it, the Gask Ridge system — a line of forts, fortlets and closely spaced watchtowers through Strathearn, sometimes called Rome's earliest constructed land frontier — extended monitoring and control toward the Tay.

Yet within a few years, this entire network was dismantled. The excavations that revealed this, conducted 1952–65 under the direction of Sir Ian Richmond (with J.K. St Joseph), and definitively published by L.F. Pitts and J.K. St Joseph in their 1985 Britannia Monograph, produced the single most evocative piece of evidence for what happened next. During the 1960 season, excavators uncovered a huge hoard of unused iron nails, carefully buried in a pit before the fortress was abandoned — a documented, counted total of at least 875,428 nails, weighing around 7 tonnes, with Richmond himself estimating the original deposit at close to or over a million. The nails appear to have been buried deliberately so the valuable metal could not be recovered by anyone else after the army left.

The larger point is what this tells us about how Rome left: the evidence points to a planned, organised evacuation, not defeat. There is no sign of violent destruction at Inchtuthil. The old story of undefeated Highland resistance simply driving the Romans out is not supported by the archaeology. The likelier explanation is that Rome needed the legions elsewhere in an empire under pressure on other frontiers — a useful distinction for this article: Agricola won battles in Scotland; Rome did not achieve a durable conquest of Scotland.

By around AD 105, the far-northern network had contracted to the Tyne–Solway line, formalised from AD 122 as Hadrian's Wall.

3. The Antonine Wall: Rome's second attempt

Roman policy reversed under Antoninus Pius. From about AD 139–142, governor Quintus Lollius Urbicus led forces back into southern Scotland, and by 1 August 142 a military diploma confirms Antoninus Pius had been acclaimed Imperator for the second time — the formal Roman recognition of a significant victory, and the epigraphic anchor for dating the whole campaign.

What followed was the Antonine Wall — a turf-topped rampart on a stone base running from Old Kilpatrick on the Clyde to the Forth (the precise eastern end, Bridgeness or Carriden, remains debated), with forts, fortlets, a road (the Military Way) and an accompanying ditch. Its length was long given as roughly 59 km (37 miles) in older accounts, but a detailed 2017 LiDAR survey of the Wall's distance-slabs re-measured it at approximately 63 km, or 42 Roman miles — the figure now standard in current scholarship, though heritage and tourism material sometimes still uses the older rounded numbers.

This was not simply a barrier. Its gates, closely spaced forts and supporting road network suggest a system built for monitoring and channelling movement, patrol and rapid deployment, as much as static defence. Building inscriptions show detachments from all three of Britain's legions contributed construction labour, while auxiliary cohorts — units like the First Cohort of Hamians and the Sixth Cohort of Nervii, both attested epigraphically at named forts — provided the actual garrisons.

The Wall's abandonment date is one of the liveliest disputes in Roman British archaeology, and it is worth stating precisely because the answer has genuinely changed. Older accounts, built on pottery and the latest stratified coins, placed withdrawal around AD 162–164, sometimes positing two separate periods of Antonine occupation. Research by Nick Hodgson has since made a strong case for a single occupation ending around AD 158 — the date epigraphically attested (RIB 1389) for renewed rebuilding work back on Hadrian's Wall, which only makes sense if Scotland had already been given up. Hanson and Breeze's 2020 synthesis of the Wall's archaeology treats this earlier date as now widely accepted, connecting the abandonment to a wider manpower shortage across the empire — legionary detachments were pulled to Germany that same year. What is clearer than the exact date is the manner of the withdrawal: wells were deliberately filled and installations decommissioned in an organised fashion, echoing the pattern already seen at Inchtuthil sixty years earlier.

4. Severus's last campaign, AD 208–211

The final major Roman intervention came under the emperor Septimius Severus, who arrived in Britain in 208 with his sons Caracalla and Geta, making York his base. Roman forces reoccupied Cramond on the Forth and built a major new base at Carpow on the Tay, using naval and river transport to push supply lines far beyond the old frontier — this was a combined-arms, logistically sophisticated campaign, not simply an army marching north along a road.

Our narrative source here is Cassius Dio, a Roman senator writing in Greek in the early third century (his account survives in Book 77, numbered 76 in some editions). Dio names the opposing peoples precisely: "there are two principal races of the Britons, the Caledonians and the Maeatae... the Maeatae live next to the cross-wall which cuts the island in half, and the Caledonians are beyond them." He describes gruelling, unconventional warfare — cutting through forests, bridging rivers, an enemy that avoided pitched battle and harassed exposed troops, so that Severus, in Dio's words, "fought no battle and beheld no enemy in battle array." His claim that fifty thousand Romans died — "unable to walk, they would be slain by their own men, in order to avoid capture, so that a full fifty thousand died" — is treated with real suspicion by historians: it plausibly exceeds the entire invasion force, and it is hard to square with the absence of any archaeological sign of catastrophic losses.

Severus forced a settlement; Dio then reports the Maeatae rebelled and the Caledonians joined them, prompting orders for harsh reprisals — but Severus died at York on 4 February 211 before he could complete a renewed campaign. His son Caracalla made peace and withdrew. No further Roman attempt to conquer the north followed.

5. Why "Caledonians and Maeatae," not "Picts"

This is a precision worth insisting on. Tacitus's opponents in the 80s are "Britons" and a "Caledonian" coalition; Dio's opponents in 209–211 are specifically the Caledonians and Maeatae — two named groupings, likely shifting political confederacies rather than fixed ethnic nations. The term "Pict" (Picti) enters Roman usage only later, in the following century. There is likely real population continuity between these communities and the later Pictish polities — but continuity of population is not the same as continuity of political identity, and collapsing two centuries of change into one label misrepresents the history. This distinction is also the natural bridge into Card #3 of this series, on the Picts themselves: Rome did not withdraw and leave "the Picts" waiting in the wings. It withdrew, and the communities beyond its reach continued changing and consolidating on their own terms, eventually producing the polities later called Pictish.

6. Beyond the battlefield: forts, supply and contact

A Roman garrison required far more than soldiers: granaries, workshops, wells, baths and roads all appear in fort plans across the frontier, and bulk supply moved through a mix of long-distance imperial networks (fine pottery from Gaul, specialist military equipment) and local sourcing (grain, livestock, timber, stone). Coastal bases such as Cramond and Carpow made sense precisely because water transport carried supply far more efficiently than pack animals and wagons.

Contact between Rome and native communities was never simply military. Roman goods circulated well beyond occupied forts — finds at major native centres such as Traprain Law show Roman material moving through indigenous exchange networks — but the meaning of any individual object is hard to pin down: it could reflect trade, diplomatic gift, military subsidy, loot, or recycling long after manufacture. Excavation at Bar Hill even preserved dozens of children's and women's shoes in waterlogged deposits, a small but important corrective to any picture of the frontier as an all-male military world.

7. A find still changing the story

Roman Scotland's archive is not closed. In 1906, excavators at Newstead (Trimontium) under James Curle uncovered more than one hundred fragments of brass armour, initially interpreted as body or thigh armour. Later reassessment in the 1990s identified the object instead as a flexible sword-arm guard — one of only three such guards known from the entire Roman Empire, and by far the most complete. It was fully reconstructed from its surviving fragments in early 2024 for the British Museum's Legion: Life in the Roman Army exhibition, and went on permanent display in the Early People gallery of the National Museum of Scotland on 16 October 2024. It is a useful reminder that even century-old excavations can still change what we know about Roman Scotland.