Subscribe now for full access and no adverts
For the last quarter of a century, geophysical surveys over deserted Roman sites have become almost commonplace, revealing entire cities and forts beneath now-rural landscapes. But what about those that did not fail, but continued to thrive until the present day, and now lie beneath the streets of their modern successors? Can geophysics untangle millennia of construction and see metres beneath the soil – and, more challengingly, beneath tarmac and concrete – to reveal underlying archaeology? Several projects are currently trying to see how far we can push these boundaries; here, we will share the results of a recent attempt in York.
Since the 1980s, heritage management has fiercely protected the city’s archaeology, meaning that few large-scale, deep excavations have happened in the centre since the underpinning of York Minster (see CA 17) and the discoveries of Viking Coppergate (CA 58) in the 1970s. Our AHRC-funded project, ‘Roman York beneath the streets’, sought to investigate this settlement’s origins, looking afresh at antiquarian material alongside the findings of more recent (albeit small and fragmented) excavations and evaluations. We also undertook some new fieldwork, carrying out ground-penetrating radar (GPR) surveys across the city, which really brought home the potential and limitations of this technology.
Towns are not obvious places for large-scale GPR surveys. It is easy to think of urban environments as dense and impenetrable, but it is remarkable how much space is still available there. In the centre of York, only 40% of land is covered with buildings; the rest includes rather-more accessible roads and pavements, parkland and parking spaces, and private gardens and yards. For our project, we chose to investigate three different settings: the streets, the Museum Gardens, and the Minster Precinct. All three presented their own research questions about the city’s Roman archaeology – and their own logistical challenges – and we knew that we were inevitably going to encounter medieval and early modern features, too, during our work. So, what did we find?
The streets of York
Using GPR in city centres is not a new idea. Commercial companies increasingly provide these services to survey the structural integrity of roads and bridges, and to map buried utilities, and many of them deploy top-of-the-range equipment, operated by teams well-versed in the regulatory and logistical hurdles of working in urban spaces. One issue with GPR antennae is that different frequencies are effective at observing features at different depths, but step-frequency systems also exist, which vary the signal, potentially enhancing visibility across a much greater range. This kind of data has rarely been examined for its archaeological potential in city environments, so we thought we would give it a try.
In the first of our three settings, the streets, our contractor (Catsurveys) covered 11km (6.8 miles) of roads at around 8kph (5mph), often working at night when the pedestrian areas were accessible and not too crowded. The warren of streets in the medieval centre of York has largely remained unchanged over the last millennium, which gave us hope that underlying Roman structures – which were on a very different layout – would be easy to recognise. Our challenge was going to be depth. In many places, we knew that features would be 2-3m (6.6-9.8ft) below the surface, but in places by the river, deposits could be as deep as 12m (39.4ft). We started by examining data from places where we already knew massive walls lay beneath the roads, beginning with the Queen’s Hotel site off Micklegate. There, excavations in 1988-1989 had revealed part of a late Roman baths complex with substantial walls (2.2m/7.2ft thick and surviving to a height of 3.5m/11.5ft) heading straight under the road, at a depth of around 2.5m (8.2ft; see CA 200). Unfortunately, nothing of these remains revealed itself in the GPR – and indeed we could see little if anything in the signal below about 1m (3.3ft). Among other locations, we also looked for the legionary fortress wall that is known to run under Feasegate – but, while its remains had been observed in cellars and excavations on either side, again, nothing was visible there.
There was clearly a problem. Services and sewers were plentiful in the results, but the signal got weaker very quickly below a depth of 1m. In the other two areas that we surveyed (which we will discuss below), however, we were able to see Roman features 1.8m (5.9ft) below the surface, even beneath stone buildings and modern tarmac. Clearly, it was specifically the foundation of the roads that was defeating us, absorbing and scattering the signal passing through the clays and aggregate – something called attenuation – and, on reflection, if we were to try again it would be more productive to focus on pavements and backyards, with their thinner makeup.
Museum gardens
There is a happier story to tell from our other two investigations, which were carried out around York Minster and the Museum Gardens, whose open grass, paths, and hardstanding with shallow foundations provided no such difficulties. Project member Dr Lieven Verdonck optimised our survey by testing five different antennae arrays next to the Minster Library, where we knew that the remains of Roman barracks lay 1.5-2m (4.9-6.6ft) below the surface. The best results came from 250 MHz antennae, and so these were used to survey each area multiple times at close intervals, to get high-quality, research-grade results.
The Museum Gardens was the simpler of the two areas. This area lies just outside the legionary fortress walls, and the gardens themselves are located within the precinct of what was St Mary’s Abbey (established in 1088). After the Dissolution of the Monasteries, the precinct hosted Henry VIII’s Council of the North, which met in the former abbot’s residence, today known as King’s Manor. In the 1820s, the Abbey was excavated and its surroundings landscaped prior to the construction of the Yorkshire Museum – and 200 years later we set out to learn more about what lay within its grounds. Our key Roman research question was to establish if a road, hypothesised in the 1960s by the Royal Commission and enshrined on maps of Roman York ever since, actually passed through the area. Roman remains had also been discovered just east of the Abbey nave in the 1950s, during excavations by the museum’s curator, and it was thought that buildings might have once stood on the edge of the terrace overlooking the river. Would any of these show up in our survey?
Pre-existing evidence for the road was weak, being conjectured on the basis of gravel surfaces seen 550m (0.3 miles) to the north-west, so we were unsurprised to find no trace of it in our results. Meanwhile, although some other GPR features did show under the Abbey’s west range on the same alignment as the 1950s findings, these were not enough to make out a plan of a specific building.
Medieval features were much more prevalent, with St Mary’s Abbey dominating our results. Our findings generally matched nicely with the plan of its remains created in 1829 (while highlighting a few inaccuracies) and these, combined with digitised plans from more recent riverside excavations linked to flood defence works in 1985 and 2021, allowed us to create a revised map of the precinct. We have also been able to refine the plan produced by Christopher Norton in 1994, which used a 1545 survey of the dilapidated buildings to identify the purpose of each structure.
York Minster
Our third area, the precinct of York Minster, was far more complicated to interpret, with the GPR revealing a palimpsest of structures spanning the Roman period to the present day. The precinct lies over the northern quarter of the legionary fortress, with York Minster itself located on top of the principia (headquarters building) and some of the First Cohort’s barracks. Excavations in 1967-1973, conducted during the underpinning of the Minster’s tower, together with further explorations in 1980 and 1997, had revealed elements of barrack blocks, so we had a fairly clear idea of what to expect. Sure enough, this time, Roman remains could be seen in the GPR.
We managed to detect a number of barrack buildings in the northern corner of the precinct, around 1.5-1.8m (4.9-5.9ft) underground. This was especially pleasing considering that in one area we were able to see through and between the walls of a major medieval structure by the Minster Library, which had itself been hitherto unknown. In the Dean’s Garden – the lowest-lying portion of this area – the barracks were only 0.8-1.4m (2.6-4.6ft) from the surface, allowing us to establish the arrangement and spacing of individual blocks. Unfortunately, to the south and east the modern land surface rose again, and where the remains dipped below c.2m (6.6ft) they ceased to be visible. These differences highlight how important the choice of antenna frequency is in GPR surveys – what is optimal in one area may not work across the entire site.
Just as in the Museum Gardens, however, our major revelations here related to the medieval period – though to understand them we had to carry out a detailed examination of historic mapping, the antiquarian record, 19th-century paintings, and the results of modern excavations. Previous work by Stefania Merlo Perring (Historic Towns Atlas: York, 2015), Kate Giles, Stuart Harrison, and others proved invaluable in this respect, and we are now able to propose some new interpretations of the Archiepiscopal Palace and its post-Dissolution conversion into Ingram Mansion. Much can be understood from OS maps back to the 1850s, but beyond that is harder. So, like good archaeologists, we must peel back the layers one by one.
Dean’s Park was created in the 1820s, but for the preceding two centuries it had been the location of a Jacobean mansion built by Charles I’s Secretary of the Council of the North, Arthur Ingram – a somewhat acquisitive property developer, who took out leases on the former Prebendary Mansion of South Cave, the old Archiepiscopal Palace grounds, and finally the subdeanery, joining them all together with new wings and rooms to create a single opulent residence. Plans of the building exist from various court cases, and thanks to our cartographic research and GPR surveys we now know that the old palace precinct wall ran down the middle of Ingram’s new mansion. This insight also helped us to determine a more precise form for the gatehouse and western range of the preceding Archiepiscopal Palace.
The palace was founded when Thomas of Bayeux was appointed Archbishop of York by William the Conqueror and began work on the Norman Minster in the 1070s. Almost nothing of this structure survives today, although some later additions can still be seen, including the archbishop’s private chapel (constructed by Walter de Grey in 1231-1234), and a 12th-century arcade running along the edge of the park. The location of other important features, though, like the Great Hall, guestrooms for the king, and various service rooms, were unknown. Our GPR survey has done a lot to remedy this, and to flesh out the plan.
For example, the west range of the palace complex, later developed into Ingram Mansion, can now be seen with the church of St Mary and the Holy Angels (built c.1177-1181) at the southern end, as well as the gatehouse separating it from the Archbishop’s Prison and an array of other service buildings. We also observed that the space between the precinct wall and the start of the arcade was precisely the length of the missing Great Hall (recorded in 1486), allowing us to position it between the service range to the west and the archbishop’s private quarters and chapel to the east.
It was the remains of these private quarters that produced our biggest discovery. We were able to locate them between the 12th-century arcade and the 13th-century private chapel, with the GPR survey picking out a complex arrangement of wall foundations, obviously representing several phases, which can in part be related to features that were excavated when the Minster Library was extended in 1997. This previous investigation had revealed the end walls and a square plinth from a 13th-century building with a vaulted undercroft and hall above, and its lines can be seen continuing in the GPR, clearly forming a hall with three bays. The rest of the foundations, however, were entirely new and on a large scale, possibly representing a single build or, more likely, a long-lived construction that had evolved over the 12th-14th century, if not beginning earlier. Only further excavation can resolve the question of the dating and function of these intriguing structures – but we have at least added some detail to their footprints.
Another interesting insight related to the neighbouring chapel. Bishops often had a private entrance from their chapel into the cathedral, but at York the construction (in 1829) and demolition (in 1940) of a former deanery had destroyed any evidence close to the chapel itself. Further south, though, two parallel linear features in the GPR suggest a structure, perhaps a covered walkway, leading from the chapel into the 13th-century door by the chapter house. Taken together, this collection completes for York all the expected characteristics of a bishop’s palace next to the cathedral church.
Searching for York’s Anglo-Saxon Minster
All of the discoveries mentioned in the previous section relate to the Norman minster – but one of the greatest unknowns in the archaeology of York remains the location of the original Anglo-Saxon minster church, dedicated to St Peter the Apostle, first attested in 626/627, and rebuilt in stone by 641. Its precinct is known to have included two other churches (St Mary’s and Alma Sophia), as well as a library and other accommodation, but its layout is similarly obscure.
There has been much discussion about precisely where St Peter’s Church once stood, and we have considered five possibilities in light of our GPR surveys. A strong case can be made for placing the church within the ruins of the Roman fortress’s principia, but – although we surveyed the paved area south of the present minster, including a transect across the suggested location (Option 1) – nothing was found. That said, estimates from the excavations in 1966-1971 suggest that the late Roman surface lies c.2.9-3.4m (9.5-11.2ft) beneath the modern ground level, which would put it beyond the reach of our 250 MHz antenna. We were therefore unable to confirm or disprove the idea that the Anglo-Saxon church was located here.
An alternative theory looks to the north of the extant minster, where Christopher Norton suggested in 1998 that the church might lie in the middle of a 91m by 91m (300ft by 300ft) enclosure, framed and respected by the later Archiepiscopal Palace, with the Norman minster built to the south. The situation proposed as Option 2 is akin to that at Winchester, where the New Minster was constructed alongside, instead of over, the Old Minster in the 10th century. Again, though, no obvious evidence of an early building was visible in the GPR data in the middle of the square, though its northern side could have been obscured by the Static Water Tanks that were built in 1941 to protect the minster from fire during Second World War bombing raids.
There is one possibility in the data just to the south of this spot, though. Option 3 focuses on a stretch of thick rubble, or possibly a foundation, measuring c.45m by 4.5m (148ft by 14.8ft), with a curved end, which could be envisaged as one wall and part of the apsidal end of a building. This feature (visible 0.6-1.4m/ 2-4.6ft below the surface) is relatively thick and is aligned at a slight angle to Thomas of Bayeux’s minster – but a single wall (if that is what it is) does not make a church. It is also true that the feature is suspiciously aligned with a mid-18th-century street that passed through the park on the same line as the current (narrower) footpath. However, it is deeper than the line of the street, and only visible along a limited stretch of it, suggesting that it is earlier and not part of the road construction. We cannot resolve this without excavation, but if there is any future ground disturbance in this area, investigating this feature should be a priority.
Another suggestion (Option 4) is that the newly discovered medieval complex that our surveys identified by the Minster Library was not just a Norman development, but had been constructed on top of something earlier. The scale and potential symmetry of some of the architectural remains are not entirely dissimilar to some of the early plans of 8th- and 9th-century churches in Britain – though, of course, the GPR discovery is currently entirely undated.
The final possibility (Option 5) is one that has been hiding in plain sight. During his earlier research, Christopher Norton was curious about why the 13th-century Chapter House at York was so late. He wondered if it might have been constructed on the site of the ‘lost’ church of Alma Sophia (built 767-780 and known from poetry by the 8th-century ecclesiast Alcuin) – but could this predecessor have been the minster church itself? The Anglo-Saxon minster, where Edwin of Northumbria was baptised in 627, had developed from an early baptismal church, with a circular form surrounded by a square or rectangle. It could be that the chapter house was memorialising its location and mimicking its form.
Archaeology always seems to deliver as many questions as answers, but hopefully some of the ideas discussed above will help inform interventions into the ground around one of Britain’s greatest minsters. This research has also shown that GPR can and does have a role to play in the investigation of complex urban environments, especially when combined with analysis of historic maps. Our surveys have enhanced the understanding of the Minster Precinct and St Mary’s Abbey, and has revealed Roman remains beneath both. While we were not able to see what lies beneath York’s streets, its pavements, yards, and gardens remain full of potential. What is clear, though, is that GPR has a future in understanding the archaeology of contemporary bustling cities, and not just surveys of greenfield Roman settlements.
Source:
Dr John Creighton is an independent researcher focusing on the Iron Age and Roman Europe.
Dr Thomas Matthews Boehmer is a Leverhulme Trust Early Career Fellow at the University of Leicester working on graves from Rome’s northern provinces.
Professor Martin Millett is Emeritus Laurence Professor of Classical Archaeology at the University of Cambridge.
Dr Lieven Verdonck is a post-doctoral researcher at the University of Freiburg (Germany), working on geophysical prospection in archaeology, as well as use of AI and machine-learning for the interpretation of geophysical survey data.
• John Creighton, Lieven Verdonck, Martin Millett, and Thomas Matthews Boehmer (2025a) Ground Penetrating Radar Surveys within the Precinct of York Minster; Roman York: beneath the streets GPR Report 1 (January 2025), https://doi.org/10.17863/CAM.115022.
• John Creighton, Lieven Verdonck, Martin Millett, and Thomas Matthews Boehmer (2025b) Ground Penetrating Radar Survey within the Precinct of St Mary’s Abbey (York Museum Gardens); Roman York: beneath the streets GPR Report 2 (January 2025), https://doi.org/10.17863/CAM.115023.
• John Creighton, Lieven Verdonck, Martin Millett, and Thomas Matthews Boehmer (2025c) Ground Penetrating Radar Survey along the Streets of York City Centre; Roman York: beneath the streets GPR Report 3 (January 2025), https://doi.org/10.17863/CAM.115024.
All images: courtesy of the authors, unless otherwise stated
