Hadrian's Wall

by Adrian Goldsworthy

  Between the inner edge of the ditch and the Wall itself is a flat patch of ground known as the berm—once again a term lifted from siege warfare in the age of cannon, but it is convenient to use because we do not know what the Romans called this feature. Along the Stone Wall, it was usually some twenty feet (6 m) wide, whereas along the Turf Wall, it was narrower, on average some eight feet (2.44 m). First discovered a few decades ago, it is now certain that on the eastern section of the Wall, the berm was covered in obstacles. Excavation has revealed a checkerboard pattern of three rows of pits, in which were set substantial posts, which in turn had sharpened stakes fixed into them (cippi). This presented a hedge of spikes—an ancient equivalent of barbed wire. At several sites, a low mound was found on the southern edge of the ditch, which would have made it difficult for anyone to crawl under the spikes or to cut at the base of their supports. Some sites show that the system was repaired and renewed during the third century. Like the ditch itself, these wooden obstacles were not expected to stop any attack, but they would slow the attackers down. Any attempt to hack through them was likely to be noisy, making them hard to get past stealthily, even at night.

  The northern gateway at Milecastle 37 just west of Housesteads is the best preserved of any on the Wall, showing the arc of the arch. It leads nowhere—within a few yards is a virtual cliff—but in the initial design it was included regardless. Later, like many milecastle gateways, it was closed up apart from a narrow doorway (closed by the modern gate) that permitted access to the north face of the Wall. All this suggests that providing routes through the Wall was not the sole or even the most important function of milecastles. On the horizon can be seen part of the modern Kielder Forest. In the Roman period, there would have been far fewer trees.

  The Stone Wall shows several variations in building style, including variants where foundations or lower courses of Broad Wall were completed in Narrow Wall. Foundations were usually of flagstones set in clay. The facing of squared rubble—or occasionally less finished stones—was usually bonded with limestone mortar and rarely with clay, whereas the core of stones, rubble, and earth might be dry stone bonded with clay or more often with mortar. In some cases an original dry-stone core was later replaced with mortar during one of the major repairs or rebuilds.

  In the eighth century, Bede wrote that the Wall ‘is eight feet wide and twelve high, running in a straight line from east to west, as is plain for all to see even to this day.’ Twelve feet (3.6 m) is higher than any sections surviving into the modern era, so this figure is taken as a bare minimum and estimates usually range between twelve feet and fifteen feet (4.5 m). In Milecastle 48 (Poltross Burn) there are stone steps that when intact would have reached to a similar height. It should be noted that this only tells us about the likely height of the wall-walk inside this milecastle. The Wall itself is unlikely to have been built to a uniform height along its entire length, especially in the central sector where it crosses the crags and successive narrow and deep valleys or gullies with very steep slopes. These may have required the Wall to run in a series of steps up or down, or perhaps to be higher to bridge the sudden dips in the landscape. It has also been suggested that it might have been higher in sections where the ditch was shallower than usual.1

  With current evidence, it is impossible to prove that the Wall itself had a walkway along its top like the ones in milecastles and forts. There is ample evidence for a string course of stones levelling the top of the Wall. This would be essential if the walkway existed, but in itself it does not prove that there was one. One argument against a walkway is that the timber stockades constructed on other Roman frontiers did not have them and were simple barriers, and that this was also sometimes true of some of the dry-stone features built in sections of desert frontiers, such as in Africa. Against this we must set the width of the Turf Wall, which looks far more like a standard Roman army turf wall with walkway and parapet on top. In addition, even the Narrow Wall—and the few sections of extra Narrow Wall—would be excessively wide for a simple barrier not intended to carry a patrol walk, which would at every point allow one man to pass another. In the initial design, the lack of a road close behind the Wall would have made a walkway the most convenient way for small parties to move along the line, and the footbridges across the rivers make most sense if they allowed this communication to continue. The nature of any parapet is conjectural, given that its very existence cannot be proven, but judging from evidence from Roman forts, a battlement—perhaps with wider gaps than was common in Medieval castles—is most likely.

  If the Wall was some fifteen feet high, then most estimates of the height of turrets would place them at double this, around thirty feet (9.2 m). They did not project in front of the curtain. Trajan’s Column and evidence from elsewhere suggests that many free-standing towers were built with tiled or shingle roofs and apparently a balcony running round all four sides. There is no direct evidence for either feature on Hadrian’s Wall, and although either or both are possible, the turrets may have been simple towers with open tops and parapets, giving good all-round visibility. Soldiers not on guard duty lived in the shelter of the room below, which was reached from ground level by a ladder sometimes mounted on a low platform. As with milecastles, there was some flexibility in the positioning of turrets. There is also one known instance of an extra tower at Peel Gap, making three between Milecastles 39 and 40. This additional tower stands at the entrance of a pass that would have been out of sight of the other towers—and out of sight of men on a wall-walk unless they were at that moment at the pass itself. Discovered fairly recently, this extra turret raises the possibility that there were other exceptional features added to deal with local conditions elsewhere.

  Artefacts found in turrets suggest relatively short-term occupation, with detachments probably spending only a few days and nights rather than weeks or months at these outposts, but there was time for soldiers to cook and gamble—the latter shown by the presence of dice and gaming boards. Weapons, especially the iron heads from a range of spears and javelins, are fairly common. So far, just one bolt head of the type shot by a light ballista—a twin-armed catapult looking like a large crossbow—has been found in a turret. This raises the possibility that some or all turrets were equipped with one of these machines, often known as ‘scorpions’, which were highly accurate and propelled their missiles with great force. One theory based on the limited available data suggests that the berm immediately in front of a turret was made deliberately narrower so that the ditch almost abuts the Wall. This would have permitted men in the tower to see—and shoot—along the line of the ditch in each direction, something that would have been more difficult with the normal berm width. However, this phenomenon is not seen on most of the surviving length of the ditch, and we can only hope that future excavation will reveal whether this was the standard layout or a local peculiarity.

  This painting of Housesteads in the third century AD should be compared with the aerial photograph of the site today (see illustration here). The Vallum is the line running across the top of the picture, beyond the civilian buildings. This reconstruction was done by the late Peter Connolly several decades ago. Recent survey work shows that the civilian housing should be extended from the buildings shown at bottom right to join with those in the top centre and also should extend farther down the slopes. In other respects, this reconstruction reflects our best understanding of the site. Towers are shown with tiled roofs, which is likely but as yet unproven. Unlike a Medieval castle, a Roman fort’s dominant features are its internal buildings. This was where the army lived rather than where it planned to fight. Even so, the circuit walls and towers presented a formidable obstacle to any attackers from the local tribes because these lacked siege engines. Note the long barrack blocks and the central range of (from left to right) granaries, principia, and praetorium, in this case built on a pronounced slope.

  Reconstructions of several features of the Wall painted by Peter Connolly. In the bottom right is
Milecastle 37 (see also illustration here). Height of the tower above the northern gateway is a guess, but the usual assumption is that these at least doubled the height of the Wall. Note that the parapet of the milecastle is shown with crenellations, whereas that of the Wall is a simple parapet. A single building is shown on the right on the basis of excavations at the site. At other milecastles, structures stood on each side of the courtyard, as shown in the plan on the left. The walls are rendered as painted to suggest construction from larger, neatly cut stones. This was a known practice elsewhere but is not clearly attested for Hadrian’s Wall.

  On the left we have a cutaway view of the turret. Although this is shown with a tiled roof, it is unknown whether these existed or whether the tops of towers were open to the elements. Another possibility is that turrets had balconies all around the top, something depicted on Trajan’s Column. Men appear to have spent longer periods of time in milecastles than in turrets.

  At the top we have a cross section of the Wall and Vallum. From right to left (north to south) is the ditch, berm, Wall, the road or Military Way, north mound, Vallum ditch, and south mound.

  The photograph shows how stones were laid only in horizontal courses on the steeper slopes.

  Milecastles were considerably smaller than the independent outposts or fortlets usually built by the Roman army. Their side walls join the Wall itself at a right angle, but the southern corners are rounded on the outside like those of larger forts. Most are around 50 feet by 60 feet (c. 15 by 18 m) in size, covering an area of some 2,981 square feet (277 sq. m). A few are larger, with Milecastles 47 and 48 nearer 70 feet by 60 feet (c. 21 by 18 m) and covering an area of 4,252 square feet (395 sq. m). The largest of all is the stone version that replaced the original turf Milecastle 52 and covers some 6,932 square feet (644 sq. m). Interestingly, its turf predecessor was also significantly larger than all other known milecastles on the Turf Wall, and clearly the location was significant and felt to require a larger than normal presence. Little is known of the internal buildings in Milecastle 52, but 47 and 48 were provided with a range of barracks on either side of the fort, as well as a stone oven built into the inside corner of the walls, and the famous steps in the case of Milecastle 48, which have not been found at any other site. In spite of poor-quality stone used in the construction, these buildings are more substantial than those seen in other milecastles and appear to have been well furnished, even having glass in the windows. It seems likely that these larger milecastles predated the decision to add forts to the Wall.

  There is no trace of a standard plan to the internal structures in other milecastles, but these buildings usually appear cruder and suggest less accommodation for fewer soldiers. Garrisons of a dozen or so men may have been typical, with anywhere up to thirty or forty in the largest ones. A few finds of horse equipment hint at the presence of horses, and one or more cavalrymen in a garrison would make considerable sense, as they would permit messages to be sent more quickly to one of the forts. On the whole, the finds from milecastles suggest that soldiers spent more time there than at turrets, with a lot of evidence for food preparation and some for repair or manufacture of equipment. On other frontiers, we hear of soldiers being sent to distant outposts for many months, but on the Wall, a fort was never farther than four miles away, so tours of duty in one of these little bases may well have been a lot shorter.

  The reconstructed gateway at South Shields shows towers with tiled roofs. Soon after work was completed, evidence was discovered in North Africa that suggested most army gate towers had four rather than three storeys. This would significantly add to their height and dominating appearance, especially for tribesmen accustomed to thatched round houses. At an early stage, the decision was made to add forts to the line of the Wall itself. This provided additional large double gateways leading to the north as well as better communication across another new addition, the Vallum. Several forts have three out of their four main gateways north of the Wall, but later this was viewed as unnecessary and just one was deemed sufficient.

  With one possible exception, every milecastle had a large entrance in the north and south walls. Solid construction suggests that a tower over the north gate was normal, and it is possible that there was another over the south gate, although this appears less useful, as it would not have significantly increased all-round visibility. The towers in the north wall would have meant that combined with the turrets, there was a raised viewing platform every one-third of a Roman mile. The milecastle gateways were substantial, the twin gates mounted on pivots, and they provided a wide road through the Wall. In some cases the route was scarcely practical—for instance, at Milecastle 42, where it opens onto a 45-degree slope, which would have been difficult for men, very difficult for horses, and impossible for ox- or mule-drawn vehicles. Milecastle 35 lies just south of a cliff dropping some one hundred feet (thirty m) and thus would have been useless as a crossing place. This milecastle is the possible exception to the rule that all milecastles had two gates, for there is no clear trace of the northern one ever having been built. If so, then the planners of the Wall once again demonstrated a willingness to adapt the design to the topography.

  ALL THE ELEMENTS DESCRIBED SO far were part of the original template for Hadrian’s Wall, before the decision was made to add forts and the subsequent addition of the Vallum. The latter was a huge obstacle to unauthorised crossing, most of all by cattle, sheep, or horses, whether legitimately owned or stolen by raiders. The Vallum is difficult to understand, mainly because only a little excavation work has been done along it, but also because there is no close parallel to it from any other Roman frontier. Although essentially an earthwork, considerable care was taken in its construction. At the stretch running behind Limestone Corner, the auxiliaries digging the ditch of the Vallum cleared it thoroughly, unlike the legionary work party who had left the ditch in front of the Wall unfinished.

  Recent excavation has discovered no trace of the Vallum being re-cut or even routinely cleaned out, but it is hard to tell whether this was the case throughout its length. Opinion is also divided over whether the marginal mound—on the south side of the Vallum’s ditch—was part of the original design, but more likely it was raised after the return from the Antonine Wall to block the causeways crossing the Vallum ditch. This was probably a simpler method than digging all the causeways out or re-filling the gaps in the north and south mounds. Some have argued for an additional marginal mound on the north side of the ditch, but the evidence is confusing and uncertain.

  The addition of forts to the line of Hadrian’s Wall may well have rendered large milecastles unnecessary and led to a reduction in the size and garrison of those completed after this decision was made. The forts also rendered the crossings through the Wall at all milecastles far less important, especially once the Vallum was dug and blocked access to and from the south. Where gaps through the northern mound of the Vallum were later cut to allow access from milecastles, there is no clear evidence for a route across the ditch and through the southern mound. Even so, it was decades before the north gates on many milecastles were reduced to simple doorways. The continued occupation of these outposts makes it clear that from then onwards, they performed a useful function not connected with offering a path across the line of the Wall. If it is correct that Milecastle 35 never had a north gateway, then this function had from the start been important. The demolition of many turrets suggests that these were no longer considered useful. This was especially common in the central section of the Wall, which lends further weight to the belief in a walkway along the top of the wall. On such high ground, a sentry or patrol looking over the parapet of the Wall would have had only a slightly less good observation point than a man on top of a tower.

  One of the most recent reconstructions of a stretch of Hadrian’s Wall was built near the site at Wallsend. On the left we see the reconstructed section of Stone Wall, shown with a walkway and a crenellated parapet. To the right are the foundations of the Wall itself, in
this case built to the width of Narrow Wall. In front of it timbers mark the spots where excavators discovered rows of post holes used to mount a hedge of obstacles between the Wall and the ditch. Such additional obstacles are attested at several sites on the eastern stretch of the Wall, and it is highly likely that they existed along most or all of the Wall’s length, at least where there was a ditch.

  Six

  FORTS AND TOWNS: SOLDIERS AND CIVILIANS

  THE FORTS ON THE WALL offered the best routes across the Wall and Vallum, the latter as big an obstacle as the Wall itself. In addition, as already mentioned, there was definitely another major crossing at the Portgate, allowing Dere Street to continue to the north, and almost certainly one near Carlisle for the other major north-south road. Close to Housesteads, another gate was added at Knag Burn, which appears to date to the late second or early third century. This was surely under the control of the Housesteads fort’s garrison and may have compensated for the inconvenient position of the fort’s own northern gateway.