Surprises from the Roman frontier: excavating Hadrian’s Wall in urban Tyneside

As Hadrian’s Wall is a protected World Heritage Site, opportunities to excavate across the line of the Roman fortifications are rare. In the summer of 2021, however, planned development works in the Ouseburn area of Newcastle upon Tyne offered the possibility to do just that – and Pre-Construct Archaeology’s investigations revealed not only the first turret to be uncovered along the Wall in over 40 years, but the largest example identified to date, together with other defensive features. Project Officer Scott Vance discusses the background to the discovery and its implications for the study of Hadrian’s Wall within urban Tyneside.


Many CA readers will be familiar with the dramatically craggy landscape where the central portion of Hadrian’s Wall runs through rural Northumberland. The fortifications’ rather less picturesque eastern end, which is located in urban Tyneside, is often shunned by tourists, however – perhaps unsurprisingly, as only brief glimpses of upstanding remains can be seen in this area, notably at Segedunum (Wallsend) fort; a turret on the West Road (A186) out of Newcastle; and short sections of the Wall along the eastern part of the A69. Even the Hadrian’s Wall Path National Trail deviates from its true route at Wallsend, preferring to follow the more scenic course of the River Tyne before finally rejoining the monument at Heddon-on-the-Wall, c.17km west of Segedunum at roughly the location of Milecastle 12. However, recent excavations in the eastern sector have dramatically demonstrated that significant Wall remains can survive in these more built-up areas (even if they are less likely to appear on postcards).

Excavations by Pre-Construct Archaeology have uncovered the remains of the largest Roman turret yet to be identified along Hadrian’s Wall. Here the team are shown recording the Wall ditch.

This is important because the area around Newcastle includes the largest stretch of Hadrian’s Wall where archaeological remains have not yet been located. Indeed, until our investigations, no reliable evidence for any of the turrets associated with the fortifications (see box on p.22) had been uncovered further east than Turret 7b. This is located on the West Road, about 5.4km north-west of the site of Pons Aelius Roman fort in Newcastle city centre (which now lies under the 12th-century castle keep). Remains described as ‘like a cellar’ or ‘a square tower’ were also uncovered, in 1886, on the corner of Eastfield Avenue and Stott’s Road in Wallsend; over the years, these have been attributed to both Turret 0b and Milecastle 1, although today most archaeologists agree that the description better fits the remains of a turret. Otherwise, though, finds of this kind have been distinctly scarce within urban Tyneside – and along the whole course of the Wall, no new turrets had been identified in over 40 years – but in 2021, that was all set to change.

These plans show the site’s location in urban Tyneside, as well as (in the more detailed image) St Dominic’s Church, a short distance to the west, where other Wall foundations were found in previous investigations.

PCA’s excavation was commissioned ahead of development at Norris House, within Newcastle’s Ouseburn area and located at the top of the western bank of the valley of the same name. The site lies approximately 1.1km east of Pons Aelius fort and 4.3km west of Segedunum, and we knew that the path of Hadrian’s Wall ran through the area that we were to investigate. As the site had seen intense development over the course of the 19th and 20th centuries, however, we had no idea how well preserved any remains might be – if they had survived at all. Nonetheless, having the opportunity to excavate across the line of Hadrian’s Wall was an exciting prospect; such opportunities do not come often, as the monument and its surrounding ‘buffer’ zone is protected as a World Heritage Site.

While the site was known to lie within the frontier system, the exact positioning of milecastles and turrets within the Newcastle to Wallsend section has always been unclear. In this area, these installations do not appear to follow the assumed spacing (a milecastle at approximately every Roman mile, with two turrets spaced equidistantly between), and although its precise location was unknown, it was thought that the nearest turret to our site, Turret 3a, lay c.300m to the south-west of where we would be working. Imagine our shock, then, when our investigation revealed the remains of this very structure.

Tracing turrets

There were at least 160 turrets along the length of Hadrian’s Wall, but to date only 57 (including the recent discovery discussed in this article) have been investigated. These explorations attest that, on average, turrets measured 5.79m across externally, 3.67-3.93m wide internally, and that they would have functioned as observation towers, with room to house perhaps four to six Roman soldiers. As for their height, this is harder to ascertain, as none have survived standing to any significant degree – however, it is thought that their viewing platform would have been somewhere between 7.5m and 9m above ground level. The platform itself would have comprised either an upper balcony or a window providing a vantage point to monitor activity around the Wall (contemporary towers shown on Trajan’s Column in Rome show similar features rather than observations being carried out from a flat roof). There are many interpretations of what these turrets looked like, and Michael J Moore has produced three possible reconstructions, shown here.

Three possible interpretations of how a Hadrian’s Wall turret may have looked. Above shows a turret with a flat roof and wall walk. Image: Michael J Moore
A turret with a shingle roof and observation platform. Image: Michael J Moore
A turret with a tiled roof and observation platform but no wall walk. Image: Michael J Moore

A developing story

Before we describe the turret’s remains, it is worth exploring how they came to survive to the present day. Historical maps show that the site remained undeveloped throughout much of the post-medieval period, and even with the advent of industrialisation in the 19th century, when the surrounding Ouseburn valley blossomed with new premises, this spot stayed as an open field. By the first edition Ordnance Survey (OS) map of 1861, though, a number of animal stalls set in enclosures – perhaps pig sties – had been built, and an even more dramatic change came at the end of the century, with the area being cleared for the construction of Red Barns Saw Mills. The mills only stood for 30-odd years, however, and when they were demolished in 1928, this provided the opportunity for a (rather limited) archaeological investigation of the site, just over a century before our own works. Unfortunately, our predecessors failed to locate any trace of the Wall, although they did uncover part of the northern ditch associated with these frontier defences; lying adjacent to a natural spring, it yielded significant waterlogged environmental samples. Then, in the 1930s, everything was covered over once more when Mawson, Swann, and Carter Ltd (general provisions merchants) constructed warehouses on the site. These would ultimately become Norris House, and no further activity of note occurred on the site until 2015, when planning permission was sought to demolish the structure and build student accommodation in its place.

The remains of the 19th-century Red Barns Saw Mills, which once stood on the site.

Modern development-led archaeological investigation started with a desk-based assessment, followed by trial trenching. So far, so straightforward – but it was this initial evaluation that uncovered robbed-out and disturbed, but unmistakable, remains of Hadrian’s Wall, as well as a section of its northern ditch. Within the berm (the area between Wall and ditch) a possible obstacle pit known as a cippi pit was also identified; it would have been one of a cluster holding sharpened stakes to ward off any would-be attackers. These were highly significant finds of international importance, and a planning condition was duly placed on the site requiring an archaeological excavation to expose any Wall remains that would be affected by the development, so that the footprint of the proposed accommodation block could be designed to avoid them.

With these measures in place, our work began in earnest, and the remains of the turret itself were discovered at the north-east end of the excavation area. There, we exposed a 12m-long stretch of substantial foundations representing the northern side of the turret and the adjacent curtain wall, as well as parts of the east and west walls. The interior was less well preserved; the 19th- and 20th-century developments described above had disturbed the remains, sweeping away any internal floor surfaces, and finds associated with the turret were similarly sparse. Indeed, the foundations produced only a single fragment of tegula (Roman roof tile). Might this provide a clue to how the top of the turret had looked? Tegula fragments are known from other sites along the Wall, as are stone slates, though both materials have always been found in low quantities. Wooden shingles and thatch represent another possibility; these do not tend to survive so well in the archaeological record, but the towers depicted on Trajan’s Column are crowned with pyramidal thatched roofs.

The substantial foundations of Turret 3a.

As we examined the exposed portion of the turret’s footprint, it quickly became apparent that T3a was unusually big. Prior to our investigation, the largest-known turret was T40b at Melkridge, which measures 8.21m externally in width, compared to the average of 5.79m. T3a, however, measures an imposing 10.26m across. Moreover, while T40b’s remains are a shade wider (by 0.03m) internally, it is possible that T3a was larger in this respect too, depending on where its walls (which in turrets are always narrower than their foundations) had been positioned on the underlying masonry. Why did this construction deviate so dramatically from the standard design?

Erik Graafstal’s viewshed analysis revealed what could have been seen from the top of the turret.

Taking a view

T40b’s unusual dimensions have been explained as reflecting a desire to see as wide an area from the structure as possible – from there, on a clear day, both Milecastle 30 and Milecastle 50 would have been visible, and it is possible that T3a had a similar panorama. To test this, Erik Graafstal kindly undertook a viewshed analysis within a 6km radius of the newly found turret. For readers who are not familiar with geographic information system (GIS) applications, a viewshed analysis enables archaeologists to reconstruct what can be seen from predefined locations. This may not appear too exciting to the casual observer but, in this case, it enabled Erik to set the ‘eyeball’ height to 7.6m, enabling him to see which other parts of the Wall would have been visible from the turret’s observation platform. He found that the forts at Wallsend and Benwell were not in view, but coverage of the River Tyne and the Ouseburn valley was excellent.

A detailed plan of the excavated remains, together with an extrapolation of the rest of the turret’s footprint (which continued outside the trench).

To the east, T3a would have had clear sightlines towards Milecastle 3 on the other side of the valley, and Turrets 2a and 2b beyond that. To the west, you would have been able to see Newcastle’s fort and all of the frontier installations up to, perhaps, Milecastle 6. The image that the viewshed presents clearly shows that the siting of T3a provided an ideal strategic vantage point, with the structure not only overlooking the Ouseburn valley but also two navigable reaches of the Tyne – something that might explain why it had been built away from the expected route of the frontier defences.

We have already discussed the unusually large space enclosed by T3a’s foundations, but what of the remains themselves? Unfortunately, these had been reduced to their lowest levels, with all the facing stones of the upper walls having been robbed out during later periods (although several small areas of wall core did survive along the northern edge). Nevertheless, we could still see that the foundations were substantial in construction – more substantial than expected when compared to other known turrets. These other sites have foundations ranging from c.0.91m (T24b’s east wall and T50b’s east and west walls) to 2.23m (T27a’s north wall, which included the footing of the curtain wall), with an average width of about 1.21m. Turret 3a’s foundations, however, were recorded at 2.46m wide (north wall), over 2.4m wide (east wall), and 2.36m wide (west wall); they are closer to the width of the curtain wall foundations that have been documented east of Newcastle (measuring 2.3-2.65m) rather than those of other turrets.

Cleaning the foundations of the turret’s north wall.

This similarity to the Wall foundations is intriguing. When Hadrian’s Wall was initially built to fortify what was then the northern frontier in Britain, the original plan comprised a system of milecastles and turrets, built to standardised dimensions and linked by a turf or stone wall. Where the Wall was built in stone, its early sections took a ‘broad’ form, though some portions were later amended to a narrower design, albeit still retaining their originally intended broad foundations in some places. Within the Newcastle to Wallsend section, the Wall was built using only this later, narrower gauge – might the evident variation in size of turrets that we see along the Wall, especially within the central sector, represent another amendment to the original standardised design?

Six cippi obstacle pits were identified within the berm (the area between the Wall and its northern ditch).

From the ground up

Local conditions such as geological deposits and the topography of the site would have also influenced the size and siting of turrets. T3a’s unusually large construction would have required well-built and solid foundations to remain standing over the years, and they were found to have been dug into deposits of natural clay. Slightly to the north and south-west, though, were deposits of sand, and unsurprisingly the Wall does not survive where it crossed these, due to subsidence issues. The Roman construction teams could have become aware of these potential issues, either while quarrying raw materials or when excavating the northern ditch, and mitigated the risk of collapse by digging wider and deeper foundations than is typical for a turret.

The nature of the local landscape also presented an additional challenge to the stability of the turret. T3a is located on the western edge of the valley, with the land sloping down towards the Ouse Burn to the east. On the downhill slope side of the turret, the foundations of the east wall were over 2.4m wide (they continued beyond the scope of our trench) and were exposed to a depth of over 0.64m where they met the north wall. Due to the limitations of the excavation, we were not able to investigate the depth of the west wall, but its width was noted at 2.36m.

An overview of all the phases and features uncovered during the excavation.

We might compare these dimensions to sections of foundations that were excavated a short distance to the west, at St Dominic’s Church, in 1928 and 1981 by George Redesdale Brooker Spain and Julian Bennett, respectively. These measured a comparable 2.43m and 2.3-2.65m in width, but were much shallower, set in a trench just 0.1m deep. Two sites on Shields Road to the east also produced foundation trenches that, at up to 0.42m deep, were not quite so insubstantial, but still lacked the depth of those associated with T3a. The wider and deeper foundations on the eastern side of T3a were conceivably required to compensate for the slope of the valley and to mitigate the risk of collapse.

Our excavation did not locate any Wall remains surviving to the west of T3a, though we did identify some other aspects of the frontier defences. It is difficult to give an accurate width of the berm in this location, as the Wall did not survive to the south-west of the turret, and the northern ditch continues outside the limits of the site. In other areas it measures c.6m wide, but when the course of the Wall is extrapolated to the south-west of T3a it implies a berm potentially measuring 10m wide. How do we reconcile this? It is worth mentioning that variations have been observed in the width of the berm alongside the Turf Wall, averaging c.2.44m but ranging to a maximum of 9-12m. There is no reason why the Stone Wall sections of the berm could not also vary in this way.

Rather clearer within the archaeological record, though, was a cluster of pits within the berm area, directly adjacent to the Wall. As mentioned at the start of this article, initial investigation had uncovered traces of one possible pit; as the excavation expanded, this total was increased to six. They were all oval in plan, had maximum dimensions of 0.92m by 0.84m and up to 0.17m deep (although they had most certainly been truncated during ploughing and development over the centuries), and they represent a feature of the frontier that is familiar from a number of other sites.

While the waterlogged nature of the Wall ditch made it a challenge to excavate, it also provided ideal conditions for the preservation of environmental evidence.

Pits and ditches

There are several kinds of defensive pits known from the various frontiers of the Roman Empire, with slightly different functions. They include entanglements (an intertwined array of sharpened branches known as cippi or cervi/cervoli pits); entrapments (sharpened stakes covered with brushwood and leaves to act as traps known as lilia); large open pits (designed to slow down attackers); and irregular patterns of postholes representing discontinuous lengths of fencing known as Flechtwerkzaun on the Raetian Limes in Germany. The pits uncovered to the north of T3a are thought to be cippi pits, although they had been dug unusually close to the face of the Wall.

This might again be due to a quirk of the local landscape; in some other areas of the Wall, the northern ditch can be seen changing course and heading directly towards the nearest turret, thereby reducing the width of this portion of the berm. This can be seen at T11b (Throckley), T26b (Brunton), and several turrets in the eastern sector of the Turf Wall. At T3a, though, the course of the ditch did not converge towards the turret, and we suggest that this could have been due to the presence of a natural spring to the west of the site. If spring water was running down the ditch, then having it pass close to the Wall would have risked undercutting the foundations and causing localised areas of collapse, especially where the geological deposits of sand were noted. Without the increased protection of the ditch being brought closer, perhaps the obstacle pits were deliberately located immediately adjacent to the structure to impede attacks on the observation tower whilst not risking structural problems.

The northern ditch was located some distance away from the turret remains, possibly to avoid the foundations being undercut by water from a natural spring.

As for the northern ditch itself, this was exposed in the north-west corner of the site, running for 9m east-west. It was a sizeable defensive feature measuring over 8m wide and c.2m deep, and, sure enough, the presence of the nearby spring resulted in distinctly waterlogged conditions at its base – which made excavating this feature a challenge. Nevertheless, what makes for unpleasant working conditions for an archaeologist created excellent conditions for the preservation of environmental evidence. Analysis of samples taken from the ditch help us to paint a picture of how the landscape looked at the time of the turret’s operation; the results evoke open grassland, with weed varieties indicative of disturbed damp ground. It is no surprise that, if the turret was to serve as an effective observation tower, the area within its vicinity must have been kept clear of trees and shrubs to ensure clear sightlines so that movement could be monitored north of the Wall.

Urban Tyneside sounds like the last place to uncover surviving elements of Hadrian’s Wall, but what it lacks in picturesque beauty, it more than makes up for in opportunities for archaeological investigations prior to modern development. The discovery of T3a at the top of the Ouseburn valley has provided new insights into the construction of Hadrian’s Wall and its installations. It indicates that local factors influenced the positioning of structures along the Wall, and that strategic interests outweighed the original standardised spacing scheme. Our discoveries also demonstrate that significant remains relating to the Wall can and do survive within the more built-up areas of urban Tyneside. This makes the extreme eastern sector of the Wall arguably one of the most dynamic sections along the whole frontier, and will most certainly provide future archaeologists with a rewarding area to research.

Further reading

David J Breeze (2006) J Collingwood Bruce’s Handbook to the Roman Wall, 14th edition (Society of Antiquaries of Newcastle upon Tyne, ISBN 090-1082651).

David J Breeze (2019) Hadrian’s Wall: A study in archaeological exploration and interpretation (Archaeopress, ISBN 978-1789691672).

Erik P Graafstal (2021) ‘The original plan for Hadrian’s Wall: a new purpose for Pons Aelius?’, Archaeological Journal, 178:1, pp.107-145. (This covers in detail the spacing problems of milecastles and turrets within the first seven Wall miles and the original eastern terminus of the Wall).

All images: Pre-Construct Archaeology, unless otherwise stated.

More on the Wall: