Forts and fortresses still provide the most conspicuous structural remnants of Roman Britain. This is partly due to the sheer scale of the remains, especially in the north. Even in the south, the coastal forts like Pevensey and Portchester have survived in far more dramatic form than the villas and towns. But in the north, stone was more readily available, and the remoteness of these areas made them less susceptible to robbing. The developing interest in Roman antiquities in the eighteenth and nineteenth centuries led to a great deal of excavation, uncovering buildings which were often left exposed. This is particularly evident on Hadrian’s Wall, where the central sector has some of the best-preserved forts in Britain, including Housesteads. Paradoxically, the much less well-preserved forts at Wallsend and South Shields have now become home to the more imaginative efforts to reconstruct Roman buildings.
Military bases absorbed a huge amount of resources, ranging from deforestation to generate massive timber stocks, to large-scale stone quarrying. According to one estimate, Caerleon is thought to have used up 150 ha (380 acres) of woodland just in being built, quite apart from the requirements of long-term maintenance. A writing tablet from the fortress, dated to c. 75–85, refers to the organized collection of wood by troops. The army was a hive of industrial activity, and remains of massive works-depots have been found, such as the mid-first-century vexillation fortress and depot at Longthorpe, near Peterborough, and the works-depot of the XX legion at Holt, near the legionary fortress at Chester. At Caerleon, large quantities of roofing and other tiles made for the II legion have been found, all stamped with the legion’s titles and manufactured in the fortress’s works-depot on an industrial scale.
Southern Britain was only briefly occupied by the army during the first few decades of the conquest. A variety of relatively short-lived forts and fortresses, including the legionary fortress at Colchester and the vexillation fortress at Longthorpe, were used for variable periods from the late 40s on into the 70s. By the 70s, the bulk of the army’s activity was in the west and north. This was where it established its hold by building permanent bases in timber and turf, later consolidated in stone. The only known permanent forts in the south until the third century were at London, built to house the governor’s garrison from the early second century, and Dover.
Some forts were in use only for a year or two, while others were built, occupied, abandoned, reoccupied and rebuilt before being permanently given up. This particularly applies to the south in the conquest phase, where major forts rapidly gave way to towns, for example at Cirencester, and where campaign forts might be used for as little as one or two seasons before being dismantled and levelled. During the Agricolan campaigns in Scotland, many forts were built that became redundant when Roman forces were withdrawn by the late 80s. The timber fortress at Inchtuthil, and the timber auxiliary forts at Strageath, Elginhaugh and Fendoch, were amongst those carefully dismantled shortly after they had been built.
Some of these Scottish forts were rebuilt in the 90s, usually to different plans and configurations to accommodate different garrisons in different circumstances. Newstead had already had four Agricolan phases, and would go on to have two Antonine stone periods before being finally given up. The excavation of many of these bases has shown how difficult it is to be certain about the garrisons, since the numbers of barracks and arrangements almost always varies from ‘textbook’ plans. The excavators of the Agricolan and Antonine fort phases at Strageath struggled to make sense of what they had found, and their published discussion is a classic example of the problems faced. The plans of the various timber phases at Strageath show that the basic layout survived through several phases, but also show that some of the fort’s principal buildings were laid out on surprisingly irregular plans.
Even where new forts were built in stone, construction work was prone to suspension, sometimes for very long periods. This has been observed in the archaeology at Birdoswald, and also in an inscription from Netherby. Occupation was episodic in some cases, while inscriptions show (or at least claim) that major structures in the fort had been allowed to decay to the point of being ruinous before being repaired. At long-term forts, this often results in extremely complicated archaeology. Birrens, in southwest Scotland, started life as a 0.5-ha (1.3-acre) Flavian fort. It was replaced under Hadrian with a 1.7-ha (4.2-acre) fort, as an outpost for Hadrian’s Wall. This version was timber, but the main central buildings were made of stone. The Hadrianic fort was demolished when the Antonine Wall was built, and replaced with a 2-ha (4.9-acre) stone fort with turf ramparts, later destroyed by fire. Birrens was rebuilt again in around 158, but the new structures were in some cases built right onto the remains of the earlier buildings.
Buildings from a fort’s earliest phases were often altered over time, if not completely demolished. Structural components can turn up reused in later phases of entirely different structures. At Birdoswald, a section of stone screen, probably from the headquarters buildings, was found in one of the granaries, reused as a threshold. Inscriptions usually only survive because they were reused in later building work. An inscription of c. 297–305 from Birdoswald, recording repairs to the headquarters, the commandant’s house and possibly the baths, had been reused as a paving slab in a late fourth-century barracks. At South Shields in the fourth century, an impressive new commandant’s house replaced third-century barracks.
The Marching Camp
Marching camps were overnight bivouacs for the Roman army, and were modelled on the same principles as permanent forts. In the second century BC, Polybius provides the earliest detailed account of a marching camp, designed to accommodate two legions and around the same number of auxiliaries. It was square, measuring 2,017 Roman feet (596 m, or 1,955 ft) on each side creating an area about 36 ha (88 acres) in size. Once the spot for the commanding officer’s tent had been chosen, the fort was laid out and was based on a regular grid with specific areas allocated for each unit, its officers and troops. The fort’s borders were marked with a ditch, and a rampart made from the spoil with a stockade along the top. An internal intervallum created a buffer zone between the fortifications and the internal accommodation.
The idea was that when the camp was built, every man knew what his job was. The process was therefore efficient, and could be executed under duress. Internal communications and systems could be repeated nightly, regardless of whether the army had moved on or not. In the third century AD, Hyginus described a marching camp for a large army of around 40,000 men, but the basic principles remained the same, though at 33 ha (81 acres) it was obviously rather more congested. Here, the fortifications included a ditch measuring 1.5 m (5 ft) wide by 0.9 m (3 ft) deep, and a rampart about 1.9 m (6 ft) high and 2.4 m (7 ft 8 in) wide. The ditches were of two types: the fossa fastigata, which was V-shaped, and the fossa Punica, which had a steep outer slope and gentle inner slope. Some ditches had square-cut trenches along the bottom, now believed to have been caused by using shovels to clear silt from the bottom of the trenches. Special arrangements were used at fort gateways. The ditch and rampart could curve in or out from the fort to cover an entrance (clavicula), a type mainly first century in date, or there could be a small section built outside the fort entrance (titulum), a type used from the first to the early third centuries.
Only the defences of a marching camp leave traces, but these are normally only detectable from the air as the army levelled the defences when it moved on. Dozens of camps are now known, with new ones being discovered almost annually. There were endless variations in dimensions and shapes of the defences. The successive camps at Y Pygwn (Powys) show that a 15.2-ha (37.6-acre) camp was followed, on a slightly different alignment, by a 10.3-ha (25.5-acre) camp. On the Hyginus model, these were theoretically big enough to accommodate up to 20,000 men, suggesting that at the very least one legion and several auxiliary units could have used the bigger one.
Marching camps are almost impossible to date, but those in north Wales probably belong to the campaigns of the first century. Either the II or XX legions could have been responsible, but it is no less probable that vexillations from both, and even from XIV, were involved. Just how difficult it is to attribute these camps to a particular period, let alone a campaign, is evident in Scotland, where two main marches north are known to have taken place: under Agricola (77/78–83/84) and Septimius Severus (208–11). Many marching camps have been detected up the east coast. Testing by excavation has produced the theory that camps with clavicula are late first century, those with titula are either late first century or Antonine, while the large irregular camps around 48 ha (118.6 acres) in size are thought to be Severan.
The Permanent Fort
The permanent fort was a consolidated form of marching camp. Buildings and troops were distributed within the fort in similar positions, but no two forts or fortresses are the same. Before the mid-first century, forts were often polygonal or irregular rectangles. The ‘playing-card’ shape then became normal and lasted until the mid-third century. These forts had regular street grids dividing the fort into square and rectangular plots, within parallel ramparts built in a playing-card shape with curved corners and at least one gate in each side. Forts often had annexes, a kind of fortified appendix to the main plan that could be used for a variety of ad-hoc purposes.
Every Roman fort represented a unique combination of factors: location, available materials, the intended garrison, and the preferences of individual fort surveyors and architects. Hod Hill, an early conquest-period fort, was built into the corner of an Iron Age hillfort in Dorset and utilized some of the old ramparts. The Period 2 fort at the Lunt, built in the mid-60s, had a very peculiar shape dictated by the reduction of the previous fort’s plan, while still accommodating a circular gyrus from the earlier fort built c. 60.
When permanent forts in stone were built along the northern frontier there were fewer major variants. It is usually possible to predict the main features of fort plans once the basic dimensions have been established. But forts could still buck the trend. South Shields, in its third-century guise, retained a stereotypical outline, but was extended and had many of its internal buildings replaced or moved about to accommodate an exceptional number of granaries. South Shields is a conspicuous reminder of the fact that forts rarely (if ever) remained in the form in which they were built. Forts that remained occupied were almost invariably modified in some way, with structures being rebuilt or falling out of use, and ramparts extended or reduced.
Several surveyors and architects are attested in Britain, all in a military context, though in no case can we associate an individual with his work. An inscription from Piercebridge records Attonius Quintianus, a military surveyor (mensor), who would have laid out the ramparts and street grid of a fort, and allocated plots for each building. The building work was probably planned by the architectus, though in a military context this is better translated as ‘engineer’, rather than ‘architect’. At Birrens, the arcitectus [sic] Amandus left a dedication to Brigantia , and another, Gamidiahus, left one to Harimella. The only other architect we know about is Quintus, at Carrawburgh.