INTRODUCTION
King Arthur’s Round Table is one of three henge monuments at the confluence of the Rivers Eamont and Lowther near Penrith, Cumbria (fig 1). They were first recorded in the seventeenth century, but all of them have been damaged since then. The largest is the embanked enclosure of Mayburgh, which was built of glacial cobbles. A single standing stone – originally one of four – survives in its centre. A second monument is Little Round Table, a circular earthwork of which virtually no trace remains on the surface, although its position has been established by geophysical survey (fig 2). The third is the monument first excavated in 1937 and 1939 (fig 3).Footnote 1
Fig 1. Air photo showing King Arthur’s Round Table towards the bottom right, and Mayburgh towards the top left. Photograph: Bob Bewley.
Fig 2. The locations of King Arthur’s Round Table, Little Round Table and Mayburgh in between the Rivers Eamont and Lowther. Information from Topping Reference Topping1992.
Fig 3. Aerial view of the surviving part of King Arthur’s Round Table at the beginning of excavation in 2023. Photograph: Bob Bewley.
King Arthur’s Round Table is a circular henge defined by an external bank and an internal ditch. There is a low circular mound inside it (fig 4). It originally had two entrances: one to the north and the other to the south. The northern entrance was flanked by a pair of monoliths, but this part of the site has been destroyed and today it is occupied by buildings and a road. The earthwork perimeter was altered when the monument was converted into a tea garden in the early nineteenth century. Its ditch was partly recut, the profile of the bank was changed and the central area was buried beneath a layer of rubble.Footnote 2
Fig 4. The surviving earthwork of King Arthur’s Round Table. Information from Topping Reference Topping1992.
THE 1937 EXCAVATION
The only intrusive fieldwork was undertaken in the 1930s, when King Arthur’s Round Table was excavated on a significant scale and its neighbour, Little Round Table, was located by three trenches. These excavations are best known for their place in the history of field archaeology. Their contribution to prehistoric studies remains controversial even now.Footnote 3
The excavation of King Arthur’s Round Table took place under two different directors, in 1937 and 1939 respectively. The contrasts between their findings have dominated subsequent accounts of the site. Work in the first season was directed by the famous philosopher and scholar of Roman Britain, Robin Collingwood, whose thinking influenced many of his contemporaries and remains important today. He was then the president of the Cumberland and Westmorland Antiquarian and Archaeological Society. He published an interim report on the first season of what he had planned as a larger project, but was then taken ill.Footnote 4 Two years later, a second season was directed by Gerhard Bersu, Hon FSA, who was in exile in Britain after he was removed from his post in German archaeology.Footnote 5 He published a detailed report of his findings,Footnote 6 but in Britain its impact has been eclipsed by his work at the Iron Age settlement of Little Woodbury.Footnote 7
The investigation of such an early monument was a new venture for Collingwood. Although he had worked with his father at sites in Cumbria, his own projects concerned the northern frontier of Roman Britain, where at different times he had collaborated with his mentor Francis Haverfield, FSA and with Gerald Simpson. King Arthur’s Round Table was the scene of Collingwood’s last excavation, which took place after an interval of eleven years during which he had devoted himself to other archaeological research. He had already written about the prehistory of the Lake DistrictFootnote 8 and had become especially interested in the Penrith henges. He was conversant with the results of fieldwork on comparable sites at Avebury (Wiltshire), Arminghall (Norfolk), Woodhenge (Wiltshire) and Arbor Low (Derbyshire), and planned an investigation of his own. His aims were clearly stated in his report:
It seemed to the writer … that the similarity between Round Table and Arbor Low in respect of their earthworks might lead to a further similarity, masked by the fact (if it was a fact) that the structures at the Round Table, being of wood instead of stone, had decayed and left no trace above ground. They would, however, have left recognizable postholes below ground; and, in order to search for them, excavations were planned.Footnote 9
The project took three weeks and involved seventeen trenches and four smaller sondages (fig 5).Footnote 10 It had a staff of eight, supplemented by local volunteers. The earthwork was completely or partially sectioned by trenches of similar proportion (Trenches d, k and r). The central mound and two areas beside the enclosure ditch were examined on a larger scale (Trenches e. i, and f). His work focused on the southern half of the enclosure, suggesting that he planned to investigate the remaining part in a second season. Because he was taken ill, he was unable to complete this project.
Fig 5. The extent of Bersu’s excavation at King Arthur’s Round Table, indicating the trench codes used in this article. Information from Collingwood Reference Collingwood1938.
Collingwood began work near the entrance causeway, where he identified a curving line of postholes concentric with the earthwork perimeter (Trenches e. ii and iii, and g). Similar features formed two parallel rows leading between the ditch terminals and the centre of the site:
Each [posthole] had its packing of small water-worn pebbles still in place … In each of them we found a filling of rain-washed sand, among which and below which a little light brown powdery matter represented decayed wood.Footnote 11
He also discovered sockets that could have held standing stones. They took a different form. The best-preserved example was in Trench j: ‘a bowl-shaped hollow lined with cobbles carefully set in clay’.Footnote 12 He was able to trace the settings of posts by small-scale excavation along their projected lines. There were three rings. Circle A followed the inner lip of the ditch. It was replaced by a setting of monoliths. Circle B was further inside the enclosure, and Circle C occupied the central area. Circle A had large posts set at equal intervals, but those in the other rings were smaller and closer together. Eventually, his excavation recognised 140 postholes, and a stone socket that seemed to replace a posthole in the same position. That relationship suggested a more general sequence, and Collingwood defined two successive deposits extending across the interior of the monument. The first was a level of clay and gravel associated with the timber buildings (fig 6). It was sealed by a more consolidated clay layer. Using a local term, he called it ‘sammel’, describing it as ‘dirty with a brown muddy trample’.Footnote 13 He thought that it dated from a later phase and was contemporary with the ring of standing stones.
Fig 6. Collingwood’s plan of his excavation, indicating the features attributed to the first phase of activity on the site. Plan: Reproduced from Collingwood Reference Collingwood1938 by permission of the Cumberland and Westmorland Archaeological and Antiquarian Society.
In his site record, Collingwood observed that ‘the post-holes were everywhere very curious’.Footnote 14 They were seldom more than 15cm deep and ‘none … can have held freestanding uprights’. Any wooden buildings must have been insubstantial. His solution was to postulate a ‘huge, thatched hut, supported on a timber framework’ in the centre of the henge. It was ‘approached by a long, straight, covered entrance’.Footnote 15 The separate posts were linked together above ground so that they would support one another. In an unpublished notebook he drew on his life in Oxford and described these structures as ‘a roofed avenue, like the awning from gate to hall at a College ball’.Footnote 16 Because the postholes continued underneath the bank, he believed that the earthwork was a later construction, but the location of its entrance was influenced by the course of the ‘avenue’, which might have extended as far as Little Round Table.
The central mound had been disturbed in the nineteenth century but underneath it, in Trench E. i, there was an elongated oval feature that Collingwood interpreted as a ‘cremation trench’.Footnote 17 It contained a small amount of burnt bone. He reported that it had been ‘sunk into [the second-period floor] and structurally belong[ed] to it’,Footnote 18 but did not have time to investigate it in detail before the 1937 excavation closed. There was a wooden structure in the centre of the henge during this phase, but now it was associated with a pair of monoliths (fig 7). Collingwood considered that they also flanked both entrances of the enclosure and that others extended around its interior (fig 8). As he had anticipated, the timbers of Circle A were replaced by a ring of standing stones.
Fig 7. Collingwood’s plan of his excavation, indicating the features attributed to the second phase of activity on the site. Plan: Reproduced from Collingwood Reference Collingwood1938 with permission of the Cumberland and Westmorland Archaeological and Antiquarian Society.
Fig 8. Collingwood’s previously unpublished sketches of the successive buildings he identified at King Arthur’s Round Table, showing the timber structures attributed to the first phase of activity (lower) and (upper) the wooden buildings and stone settings that took their place. Reproduced by permission of the Bodleian Library, Oxford.
THE 1939 EXCAVATION AND THE SUBSEQUENT CONTROVERSY
Bersu’s excavation lasted five weeks and came to an end six days before the outbreak of war. After an interval, he and his wife Maria were interned on the Isle of Man. The project involved a staff of eight as well as local volunteers, some of whom had taken part in the 1937 season. In this case, sixteen trenches were excavated, including cuttings 2m wide and more extensive exposures 3.5m in width (fig 9).Footnote 19
Fig 9. Bersu’s plan of his excavation, indicating the positions of Collingwood’s excavation by dashed lines. Plan: Reproduced from Bersu Reference Bersu1940 by permission of the Cumberland and Westmorland Archaeological and Antiquarian Society.
His excavations were more extensive than Collingwood’s (fig 10) and involved further cuttings through the bank and the ditch as well as additional trenches in the interior. More of the monument was investigated and the trenches were arranged around two extended sections crossing the enclosure at right angles to one another. The 1937 and 1939 excavations overlapped where the ‘cremation trench’ had been discovered. Otherwise, the new excavations were located beside those dug by Collingwood or in places where the settings of timbers or standing stones might be expected to continue.
Fig 10. The layouts of the 1937 and 1939 excavations at King Arthur’s Round Table. Bersu’s trenches are shown in outline and Collingwood’s are shaded.
Bersu accepted Collingwood’s view that the earthwork had been altered. He also considered that the low mound in the interior was the remains of a cairn. Although it was disturbed when the monument was used as a tea garden 200 years ago, it featured in earlier accounts and must have been an original component of the site. Here, there was a small quantity of cremated bone (unfortunately, it no longer survives). Bersu agreed that it might have originated in a prehistoric burial. It is possible that this ‘trench’ was the bottom of an unrecorded excavation, because a photograph taken in 1937 suggests that it might have been cut from a high level.Footnote 20 In that case, the bones could have been re-deposited.
From the outset, Bersu was sceptical of Collingwood’s other findings. Only six days into the project he wrote in his site notebook: ‘All that looked like postholes are surely mouse holes, so that Collingwood’s hypothesis is more than questionable.’Footnote 21 In his report, he explained his doubts in more detail. The subsoil features were the remains of animal burrows:
Both horizontal and oblique branches were connected with vertical burrows which had been explained as post-holes. [There were] galleries branching from the vertical holes, but also larger and wider lairs. But because the galleries associated … with larger boulders which the creatures had to circumvent follow most irregular courses up and down, the vertical portion of any gallery creates very well the impression of a post-hole.
In sections across the central platform we also encountered various gravel packs … They are … very irregularly distributed on the surface and on closer examination prove to be not artificial … but natural packs in the ‘sammel’.Footnote 22
In his notebook, Bersu expressed doubts about the stratigraphy recorded in 1937. Early in the excavation he wrote: ‘At the moment I can’t see any difference between levels 1 and 2.’Footnote 23 He took advice from local geologists and interpreted the successive layers identified by Collingwood as glacial deposits that extended outside the limits of the henge. His excavation did not find any postholes or stone sockets in the positions suggested by his predecessor’s reconstruction. In fact, there was no evidence of any stone or timber structures inside the excavated parts of the monument. Bersu argued his case at length and illustrated it with detailed section drawings, which, unfortunately, were over-reduced in his published report. The original documents, drawn in colour during the excavation, are preserved together with his notes in the Historic England archive in Swindon.Footnote 24
The reports of the two excavators shed light on their different approaches. Bersu had a powerful visual imagination and recorded his excavation in a series of naturalistic drawings. Collingwood’s approach was more cerebral and was better expressed in words, for he was a prolific author. His report was structured around a few key arguments, and figure drawings played a less important role. This was not surprising since he used archaeological research to illustrate his philosophical method.Footnote 25 He was engaged in a more abstract exercise than Bersu. Their different approaches have important implications for research at King Arthur’s Round Table. Because Bersu’s record was predominantly graphic, it is possible to assess his observations from the material preserved in the archive. Collingwood’s plans and sections are more schematic and cannot be reinterpreted without new fieldwork. The photographs taken by both excavators provide little additional information.
Three important issues have arisen since the project took place. In 1994 one of the writers used its results to compare different approaches to excavation.Footnote 26 The article was intended as a cautionary tale rather than a detailed narrative. One approach was led by theory and was illustrated by Collingwood’s results. The discussion highlighted the danger of basing ambitious interpretations on unsatisfactory observations. The ideas behind his project were important and influential, but the excavated evidence raised problems. The other position was represented by Bersu’s work at the same site. It was characterised by impeccable fieldwork but shed no light on the significance of the monument. Thirty years ago, it was difficult to achieve the right balance between objective recording and subjective interpretations, but the situation has considerably improved since then.Footnote 27
A second issue concerned the henge itself, for the sequence from a timber setting to a stone setting has been identified at similar earthworks excavated since the 1930s. A notable example is the Scottish site of Balfarg (Fife).Footnote 28 Lastly, several scholars familiar with Collingwood’s fieldwork on the Roman frontier, particularly Anthony Birley, FSA, and Grace Simpson, FSA (the daughter of his colleague Gerald Simpson), have defended his reputation as an excavator;Footnote 29 on the other hand, his most influential student, Ian Richmond, FSA, accepted Bersu’s reassessment of King Arthur’s Round Table and translated his report for publication. Still more recently, Stephen Leach, a philosopher with direct experience of field archaeology, has suggested that the results of the first excavation might be taken at face value. In making this case, he drew on previously unpublished letters by Collingwood. His paper concluded that the problem could only be resolved by conducting new work at the site.Footnote 30
THE OBJECTIVES OF THE 2023 EXCAVATION
The monument plays a significant part in the prehistoric archaeology of northern Britain, but the putative structures inside it have always posed a problem. The main aim of the 2023 excavation was to resolve the controversy over the features identified by Collingwood and subsequently dismissed by Bersu. It was vital to understand the sequence of deposits on the site. Were the layers recorded by Collingwood natural or artificial? Did they originate during separate phases of prehistoric activity, as he claimed? These problems affected different parts of the site: the area inside the enclosure, and the land surface underneath its bank. Here, the obvious starting point was the buried soil identified by Bersu. This was important as it had not been disturbed during the nineteenth century.
The interior of the monument was masked by a layer of rubble that occupied much of the time and energies of the workmen employed by Collingwood and Bersu. As a result, their investigation of the prehistoric structures was curtailed. In the new project, it was removed under archaeological supervision using a mini-digger. Then small areas that had not been investigated before were excavated by hand. Since the nature of any subsoil features posed such problems, they were recorded by photogrammetry. That avoided the subjective element associated with plans drawn in the field. The Neolithic earthwork and its chronology will be discussed in a separate article concerned with all three henges at Eamont Bridge.
In 2023 three small areas were selected to answer the questions posed by earlier fieldwork (fig 11).
Fig 11. The positions of the 1930s excavations following the same convention as fig 10. The three trenches excavated in 2023 are indicated in dark tone.
Trench 1 was dug to the south-west of the monument in order to shed light on the henge bank and the prehistoric land surface. These deposits had been investigated during the previous projects, and both had identified a well-preserved buried soil sealed by an irregular pile of turves that, in turn, was covered by material dug from the ditch. Both the excavators’ interpretations followed similar lines, although Collingwood saw a deposit of rubble as a deliberate feature while Bersu took a different view. Collingwood reported that the bank had a ‘foundation of carefully-laid cobbles’,Footnote 31 but its character was not clear and it was not observed in the 1939 excavation. Bersu found the same deposit of ‘sammel’ underneath the earthwork and outside the monument altogether. Having consulted geologists, he concluded that it was ‘a natural formation’.Footnote 32
It was not necessary to reopen the deeper sections through this earthwork because the early deposits were accessible where the bank had been lowered during the post-medieval period. The metre-wide trench (xiii) that Bersu had dug towards the end of his campaign was re-excavated. The same evidence of the primary turf core and the buried soil was exposed and investigated, allowing the character of the prehistoric land surface to be compared with more disturbed deposits inside the ditched enclosure.
Trench 2 measured 5m by 4m and was located west of the entrance to the henge in part of the site examined in 1937 but not in Bersu’s excavation. It reopened one of Collingwood’s cuttings in an area (Trench J) where he postulated a setting of monoliths extending along the inner lip of the ditch (his Circle A). Here, they were represented by a stone socket that apparently replaced a posthole.Footnote 33 Part of this area was re-examined in 2023 to see whether any trace of these features survived. Would similar evidence be found beyond the limits of his excavation? To answer this question, the work extended into an undisturbed area beyond the original trench.
Trench 3 measured 10m by 5m. Its eastern limit was aligned on the entrance causeway, and this excavation was in a position where it would overlap with two of the timber settings identified in 1937: Collingwood’s Circle B, and the avenue leading towards the centre of the monument. It was an area in which Bersu had excavated Trench iii alongside those of his predecessor; his Trench iv was just beyond the area investigated in 2023 and immediately north of Collingwood’s cutting E. iii. His aim was to adjudicate on the character of the post settings claimed two years before. Again, the new project extended into undisturbed ground beyond the limits of both excavations and looked for similar features.
THE RESULTS OF THE 2023 EXCAVATION
Trench 1
Trench 1 reopened one of the last trenches Bersu excavated at King Arthur’s Round Table (Trench xiii; fig 12). His cutting was smaller than he reported and must have been excavated rapidly just before war was declared. His position as a German expatriate placed him under pressure, yet the section drawing in the Historic England archive is detailed and entirely accurate (fig 13).Footnote 34 The new analysis focused on the buried soil, which was studied in the field and the laboratory by Professor Charles French. This account is extracted from his full report, which will be published in a separate article on the monuments at Eamont Bridge:
The geological substrate comprises rounded cobbles of greenish ashes and lavas from the Lake District Borrowdale Volcanic Group overlying silts and gravels (‘alluvium’), bordering Glacial Till to the west. The alluvial drift geology of silts and gravels is the probably the source of the ‘sammel’ noted in Collingwood’s report. This superficial drift geology complex is probably periglacial (very late glacial) outwash associated with the River Lowther, possibly mixed with some solifluction debris.
A thin, relatively poorly developed earlier Holocene soil had developed by the time the henge was constructed. This soil is a strongly humic, bioturbated, silty clay loam or weakly calcareous brown earth to argillic brown earth. There is evidence of an upper root mat zone suggesting that the site was turf grassland prior to the construction of the henge. It appears that there has been some truncation of the turf horizon. This could have occurred both in pre-henge (or early Neolithic) times as well as in association with the construction of the henge bank. The strongly reddened and oxidised upper surface of this buried soil also suggests that there has been much compaction and puddling of this part-truncated turf and probably as a consequence of considerable trampling of the topsoil of the day during the construction of the bank.
Fig 12. Trench 1 excavated in 2023, recutting Bersu’s Trench xiii.
Fig 13. (Upper) Bersu’s unpublished coloured section drawing of his Trench xiii, showing the core of the bank as a detail (bottom left). Reproduced by permission of Historic England. The illustration also includes Bersu’s published section drawing as well as his plan showing the natural features observed in the subsoil in part of his Section 1. They are reproduced by permission of the Cumberland and Westmorland Archaeological and Antiquarian Society.
Trench 2
Trench 2 was located along the inner edge of the ditch and investigated part of the stone setting identified by Collingwood and a posthole that formed part of his Circle A (fig 14). At this point, his Trench J was 2m wide and sloped down to a flat base 1.4m in width. It was 65cm deep and cut into the periglacial deposits to a depth of 20cm. It was here that he identified a stone socket superimposed on an earlier posthole, but it was difficult to assess his interpretation. That was because the hollow that supposedly held a monolith was removed during his excavation; only the base of the putative posthole survived. The filling of this feature had been excavated, but the ‘packing stones’ that remained resembled the concentrations of pebbles described by Bersu, who considered them parts of the natural geology. There were other clusters of boulders in the surface of the same deposit. They were outside the area excavated by Collingwood, and investigation showed that they did not mark the positions of postholes. The one subsoil feature that could be accepted was of recent date. It was 50cm deep and might have been a pit connected with the nineteenth century tea garden. The same applied to a conspicuous roothole penetrating the superficial level he described as ‘sammel’. Trench 2 was the only area excavated in 2023 in which the geology bore a close resemblance to the sequence illustrated by Collingwood (figs 15 and 16).Footnote 35
Fig 14. Trench 2 recorded by photogrammetry in the 2023 excavation. The lower image shows the base of the trench without comment, while the upper image is annotated. Key: RC indicates Collingwood’s trench edge, and a putative posthole/stone socket belonging to his Circle A.
Fig 15. Section of the east face of Trench 2, showing the complex geological deposits, sealed by post-medieval rubble and other recent features. It also includes Collingwood’s Trench j. From a drawing by Ronnie Scott.
Fig 16. Collingwood’s schematic section drawing summarising the sequence of deposits on the site. It is reproduced by permission of the Cumberland and Westmorland Archaeological and Antiquarian Society. It does not refer to one specific location, and this version should be compared with fig 15 illustrating part of the 2023 excavation.
Trench 3
Trench 3 raised more complex issues, for here the positions of both the previous excavations overlapped. Part of this area was occupied by trenches excavated in 1939 (Bersu’s Trench iii) and the intersection of three of Collingwood’s (Trenches E. ii–iv) of two years earlier; the new work extended into two areas that had not been investigated before (fig 17). They measured 3m × 3.7m and 2.2m × 4.3m respectively. In both cases, the surface of the periglacial deposits was somewhat mottled and many small to medium-sized cobbles littered their surfaces. Cleaning to a depth of 5cm revealed a few potential postholes like those observed by Collingwood, but on investigation all were rejected. The filling of his cutting was removed, and the southern section of the new trench was taken down to the same level as the 1937 excavation.
Fig 17. a) Trench 3 recorded by photogrammetry. Key: RC indicates Collingwood’s trenches and the features identified by him; GB indicates Bersu’s excavation. It also indicates two possible postholes. b) Trench 3 recorded by photogrammetry. It shows the base of the trench without comment at the conclusion of the 2023 excavation.
There were five shallow ‘hollows’ in the base along the northern side of Collingwood’s Trench E. iv. They were 35cm below the periglacial surface and between 6cm and 13cm deep. They had been backfilled with loose loam and small to medium-sized cobbles. Their positions corresponded to his Circle B. They had been interpreted as postholes. Since they had already been excavated, all that remained were putative packing stones. To check their validity, the new excavation focused on a previously undisturbed area, extending the course of the projected post alignment for another 2.3m. The periglacial deposits were removed in three 20cm-wide spits to a depth of 25–30cm, and again the deposits were carefully cleaned and recorded. Various rootholes were identified as well as some probable burrows, but there was only one possible posthole (Feature 205). It was 40cm across and 30cm deep, but was so badly disturbed that its status was not clear. Another candidate was 30cm in diameter and 35cm deep. It was visible in section in the end of Trench 3 and appeared to be associated with two large packing stones (Feature 204). It was filled with dark clayey loam and had a flat base at a higher level than any of the ‘hollows’ investigated by Collingwood.
The ‘avenue’ identified by Collingwood posed other problems because no additional elements were identified on the same alignment. The new project recognised a series of features that resembled the rows of postholes photographed in his report,Footnote 36 but the distance between them was different (fig 18). According to his publication, the lines of uprights were about 2.1m apart, but when his excavation was reopened in 2023 the distance was about 1.45m. The intervening space had apparently been excavated to a depth of 45–50cm, and it was at this level that the ‘features’ first appeared in the edges of his cutting. Appearances were deceptive, as the 1937 trench extended even further into the periglacial deposits and had been carefully refilled to the point at which the putative post sockets were visible. It is not clear why this happened, although it may have been to prepare the trench for a photograph. Collingwood’s close friend, Mortimer Wheeler, FSA, seems to have done this at Maiden Castle, even highlighting features he wanted to illustrate by filling them with topsoil.Footnote 37
Fig 18. (Left) The ‘Avenue’ identified by Collingwood 1937 as shown in his excavation report. The photograph is reproduced by permission of the Cumberland and Westmorland Archaeological and Antiquarian Society. (Right) The same features after excavation in 2023.
The outcome of Bersu’s investigation was more straightforward. The southern third of his Trench iii was in the middle of the new excavation. It ran north from one part of Collingwood’s Trench E. iv and parallel to another (E. ii), leaving a 15–30cm-wide spine between them. The 1939 excavation was 3.5m wide and had been dug between 25cm and 35cm through the deposit originally described as sammel. At one point he sank a deep sondage, going down into the geology for another 45cm. It was refilled with large water-rounded boulders. This must have been intended to check the stratigraphic sequence recorded by his predecessor.
The main results are clear. This area did not contain any artificial levels. As Bersu recognised, the surface of the periglacial deposits was disturbed by rootholes and animal burrows that could be confused with postholes. It also contained clusters of boulders that might easily be mistaken for packing stones when they were studied in plan; only by sectioning these features could this interpretation be rejected. They had been identified at completely different levels within Collingwood’s excavation – individual groups first appeared between 35cm and 50cm below the surface of the periglacial deposits.
If the elaborate timber settings identified by Collingwood did not stand up to scrutiny, this was understandable as structures of this kind are difficult to investigate on boulder clay. It need not mean that the interior of the monument was entirely empty. The new project did identify two possible postholes, although neither was unambiguous, and both remain undated.
Trenches 2 and 3
Charles French examined the deposits in both these areas. He observed that:
they contained a modern turf and humic topsoil over largely trench backfill deposits from the 1937 and 1939 excavations. The backfill deposits were intermixed with turf/topsoil, all of which appears to have suffered post-depositional and/or more recent disturbance and inter-mixing. This sequence was re-deposited on the same drift geological materials as observed in Trench 1. In most instances, the earlier excavations had cut through the upper drift geological deposits [our emphasis].
THE FIRST EXCAVATIONS IN CONTEXT
The discussion published in 1994 used the first excavations to consider the relationship between two kinds of research. That was why it was entitled ‘The philosopher and the field archaeologist’.Footnote 38 The excavators were treated as archetypes, and the monument was not considered in detail. The same tensions were apparent after fifty years, but the problems have become less severe since then and it is time to consider Collingwood and Bersu themselves. There is more to say about their objectives and methods. For instance, how did they conduct the investigations at King Arthur’s Round Table, and how was this work related to their other projects?
Robin Collingwood
Collingwood embarked on fieldwork with an explicit agenda:
Long practice in excavation had taught me that one condition – indeed the most important condition – of success was that the person responsible for any piece of digging, however small and however large, should know exactly why he was doing it. He must first of all decide what he wants to find out, and then decide what kind of digging will show it to him. This was the central principal of my ‘logic of question and answer’, as applied to archaeology.Footnote 39
His approach was distinctive. Ian Hodder makes this point especially clearly. As recently as 2014 he reviewed Collingwood’s approach in relation to that of other archaeologists:
It seemed to me that Collingwood was the only one thinking really carefully and thoughtfully about what happens when one digs … His description of what happens as one uncovers traces in the ground …[was] full of good common sense and yet theoretically illuminating. To me it was a revelation to find that the process of moving the trowel over the ground could be explored intellectually, enough for it to be called philosophy.Footnote 40
Collingwood’s emphasis on ‘question and answer’ was one of the foundations of his philosophical and historical writings and was clearly articulated at King Arthur’s Round Table. Even before the excavation commenced, he had identified the main problem in a public lecture and set out his agenda for solving it:
The monument itself is in our possession and … a very little digging would settle whether or not it once contained timber posts.Footnote 41
It is too easy to be critical of the outcome. Ian Richmond’s obituary of Collingwood set the tone:
He made up his mind in advance what he was to find and found it with fatal precision … When the writer came to translate the German report, it was sad and inexorable reading … One had the feeling that he had fallen into a pit of his own digging.Footnote 42
Although Collingwood’s interpretation was mistaken, Richmond was completely wrong to portray him as a victim of his own preconceptions. In fact, Collingwood’s records – both published and unpublished – are by no means uncritical. When he discovered features under the bank, he made it perfectly clear that they were difficult to interpret:
The … floor was found … to contain numerous shallow pockets of dark, soft material. They were completely emptied with trowel and knife, but they had neither the shape nor the structure of post-holes; they were merely unevennesses in the floor, and the dark material in them was partly mud and partly vegetable roots.Footnote 43
His unpublished notes show that he appreciated the difficulty of identifying postholes in such unfavourable subsoil. At the entrance to the monument, he recorded that ‘we opened up with great care a number of what we first thought to be postholes’. Because they were confined to a few areas, he was persuaded of their authenticity – ‘this negative evidence proves that we cannot bogus [sic] postholes and that the ones we call genuine are genuine’ (emphasis in the original).Footnote 44 He took an equally cautious approach to the eastern perimeter of the enclosure, where he recorded that ‘everything is in loose gravel in which one can dig a hole wherever one likes without opposition’. It was ‘too dangerous to profess to recognise a post-hole anywhere’.Footnote 45 Although the 2023 excavation did not support his interpretation, Collingwood was aware of the problems that he faced. Another reason for his difficulties was the decision to dig narrow trenches. They would have made it harder to trace the plans of timber buildings. In the same way, his sections through the bank were not wide enough to shed much light on its structure or, just as important, on the character of the natural geology.
Before embarking on his project, he had informed himself about henge monuments. His manuscript notes show that he mastered all the available information.Footnote 46 Collingwood summarised the results of fieldwork at sites extending from Wessex to the Peak District. As a result, he had a clear conception of how henges and stone circles were organised and the ways in which they might have developed over time. Some of those ideas are still accepted today. Timber circles took comparable forms to stone circles and occasionally they preceded them. Circular earthworks contained circular settings of uprights and could be approached along ‘avenues’. Collingwood was justified in predicting what he might find.
His excavation strategy was well established, for it had already been employed in projects conducted with Gerald Simpson, Francis Haverfield and the Cumberland Excavation Committee, ‘who never dug a trench without knowing exactly what information they were looking for’.Footnote 47 This approach played an important role in the archaeology of the Roman frontier and was taken even further by his student Ian Richmond. Phillip Freeman describes it in these terms:
[Richmond] is commonly associated with an idiosyncratic form of excavation; slit trench and key-hole sondaging of sites to confirm hypotheses and check building alignments. [He] learnt of the advantages of this method from F G Simpson, who had devised the technique in his pre-War excavations along Hadrian’s Wall and its environs … This technique, with all its drawbacks, worked on sites which were expected to exhibit standardized ground plans.Footnote 48
These methods had so much influence because Roman military buildings were thought to take predictable forms. Many were built of stone and easy to identify. They were the monuments with which Collingwood was most familiar. A similar conception guided his work at King Arthur’s Round Table, and the excavation team included people who had excavated on Hadrian’s Wall. According to the site notebook, Gerald Simpson visited the site during the project. He shared Collingwood’s interest in the archaeology of northern Britain, and during the 1920s and 1930s had influenced his approach to excavation.
Although Collingwood did not number the trenches in his report, it is possible to trace the development of his excavation over the three weeks he spent in the field. This suggests that he was following a similar procedure to Simpson. It was why work inside the enclosure commenced close to the surviving entrance – comparison with other monuments suggested it was where postholes or stone sockets were most likely to occur. He identified numerous candidates, and on that basis began to reconstruct the monument along similar lines to those he knew from published sources. There were successive settings of timbers or monoliths (his Circle A), and the central part of the henge was accessed along a wooden avenue. Further work identified two more rings of timbers (Circles B and C) in the middle of the enclosure, confirming the hypothesis that the site contained several post circles and perhaps a central building (see fig 18). That arrangement recalled the plan of Woodhenge.Footnote 49 Collingwood appreciated that there were important differences of scale between King Arthur’s Round Table and structures in other places, but tested his interpretation by excavating trenches to expose more of their layout. He did so through small scale excavations like those used by Simpson and Richmond (Trenches A, B, C, H, I and J).
There is another indication that his main objective was to trace the outlines of these structures. A manuscript site plan in the archive is unexpectedly revealing.Footnote 50 It shows the earthwork perimeter and marks (in ink) the positions of all the postholes and stone sockets on the site. A note records that they had been planned using a theodolite. The edges of his trenches, however, were lightly sketched in pencil and do not agree with those in his published site plan; he must have drawn another version which no longer survives. Similar procedures are evident in the reports published by the Cumberland Excavation Committee and are even more apparent in the report on Richmond’s famous excavation at Hod Hill, which provides plans of a series of Iron Age and Roman buildings but does not show the positions of his trenches.Footnote 51
Although Collingwood’s approach was logical, it was based on a false premise. He was comparing King Arthur’s Round Table with other sites, but more henges and stone circles have been investigated since his day, and they exhibit greater diversity than he could have anticipated. Some enclosures appear to be empty, and others provide evidence for several phases of buildings. Their chronology extends from the Middle Neolithic period to the Bronze Age, and it seems likely that in some cases the banks and ditches were the last features to be built. These structures show little of the ‘standardized ground plans’ that Freeman describes on Roman sites. It no longer appears that prehistoric timber settings or stone circles took predictable forms.
Collingwood’s interpretation is vulnerable for another reason. The monument was constructed on an intractable deposit of boulder clay, which poses problems for excavators even today. He recognised the difficulties of identifying postholes cut into a difficult subsoil, but in the end he accepted field evidence that is no longer reliable. One reason was that the site had been disturbed when it was rebuilt as a tea garden. Another is in 1937 that archaeologists had little experience of excavating glacial deposits of this kind. That would certainly apply to work around Hadrian’s Wall.
Gerhard Bersu
Bersu’s excavation followed the same principle as his projects at earthworks in Scotland and Ireland. They were organised around a series of long cross sections (Trenches i, v and viii). They could be laid out at right angles to one another, as happened at King Arthur’s Round Table, or were organised radially like those at Scotstarvit.Footnote 52 Larger areas were exposed by parallel trenches that were dug and refilled in sequence. They were employed on the plough-levelled site at Little WoodburyFootnote 53 and at well-preserved Iron Age settlements on the Isle of Man. Following the accepted procedure on the Continent, the deeper deposits might have been excavated in spits. This is implied by Bersu’s report on his Manx excavations,Footnote 54 and could explain why his records of King Arthur’s Round Table contain many detailed section drawings but very few plans. The projects he carried out in Germany demonstrate that he was well versed in geology and had a particular talent for interpreting postholes.Footnote 55
Unlike his predecessor at King Arthur’s Round Table, Bersu numbered all the trenches in his report. This makes it possible to follow the course of his excavation. Following his usual practice, it began with long trenches cutting through the earthwork perimeter and meeting at right angles in the middle of the enclosure. They were supplemented by other sections cutting radially across the monument (Trenches i, ii, iii and iv). The positions of these excavations took little account of those dug in 1937; it is almost as if Bersu saw this as a new project rather than the second part of a longer programme of research. A week into the 1939 excavation he was already questioning Collingwood’s interpretations. According to his notes, he decided that the deposits within the enclosure and buried beneath its bank were of geological origin. The putative postholes and stone sockets were rootholes or animal burrows. The clusters of packing stones associated with them were periglacial features.
In the light of such uncertainties, the next stage of Bersu’s excavation followed different principles. This work was to occupy another month. Again, it is possible to trace the stages by which it developed. The third trench he excavated reopened one dug by Collingwood, and the same happened in the middle of the monument where investigation of the ‘cremation trench’ had not been completed in 1937. He organised his excavations at Scotstarvit and Freestone Hill in almost the same way, and in both cases his initial sections preceded larger exposures in the centres of these sites.Footnote 56 At King Arthur’s Round Table, Bersu investigated less of this area than Collingwood, but the positions of trenches excavated two years before might have influenced his decision. He was able to compare his own observations directly with those made by his predecessor. It only increased his doubts about the character of the stone and timber structures identified during the first season.
Assuming that Bersu’s trenches were numbered as soon as work on them commenced, the fourth of his cuttings was located immediately beside one excavated two years earlier in 1937 when Collingwood investigated part of his Circle B; in 2023, the same area was included in our Trench 3. It is obvious that Bersu did so to compare his observations directly with those of the first investigator. Digging ceased once he had satisfied himself that the deposits (and the features observed within them) were of natural origin.
The final stage was to investigate two segments of the outer ring of postholes and standing stones postulated during the first season (Circle A). Following the method used by Gerald Simpson, Collingwood had dug small test pits where he expected to find them (Trenches a, b, c, i and f). Bersu excavated longer trenches, but did not discover any subsoil features there (his Trenches vi and ix). Again, he could compare his own experience with that of the previous excavator. In some respects that concluded his investigation, although he ended the project with a series of narrow cuttings in and around the entrance to the monument (Trenches x, xi a and b, xii and xiv). They did not shed any light on its structure.
SUMMARY
Collingwood’s excavation was meant to be the first part of a longer project, and it is conceivable that he would have recognised the problems raised by the first season’s work and might have changed his mind. He was not afraid of doing so. As Ian Richmond observed:
[He tended] to drive the evidence hard and to build on it a series of conclusions whose very artistry disguised the inherent weakness of foundation. But it was also characteristic of Collingwood’s courage that he later retracted many points … giving his reasons of doing so in full.Footnote 57
It is a moot point how the project at King Arthur’s Round Table would have developed if it had not been interrupted by illness.
The work undertaken in 2023 was a response to recent discussions of the results of an eighty year-old project. It has shed new light on the fieldwork of a philosopher whose ideas have influenced generations of archaeologists and has compared his approach to an important prehistoric monument with that of one of the most accomplished excavators of the twentieth century. It is obvious that their perspectives were entirely different. Collingwood wanted to contribute to a broader historical narrative, but Bersu’s aim was detailed documentation. He tested the observations made by Collingwood and came to different conclusions. He rejected the ambitious interpretation proposed in 1937 (although it had always been provisional) and the results of the 2023 season support his verdict. There is no longer any evidence for complex structures in the excavated areas, although they might have existed in other parts of the enclosure.
It is easy to see why these great scholars disagreed, but their published reports tell only part of the story. The experience of working at King Arthur’s Round Table made us aware of how difficult it must have been to investigate a monument built on such difficult subsoil.
ACKNOWLEDGEMENTS
King Arthur’s Round Table is a Guardianship Monument, and we are grateful to Historic England, English Heritage and the landowner, the Lowther Estate, for allowing this work to take place. Special thanks are due to Andrew Davison, FSA, Mark Douglas, FSA, and Susan Greaney, FSA. The excavation was funded by grants from the Society of Antiquaries of London, the Royal Archaeological Institute and the Prehistoric Society. Bob Bewley, FSA, provided the fine aerial photographs of the site. Charles French, Julian Cruickshank and Noel Pearson advised on its geology and soils. During the project we presented the results of the new excavation to specialists on Collingwood’s approach to fieldwork along the northern frontier: David Breeze, FSA, Richard Hingley, FSA, and Martha Stewart. We did the same with a group of prehistorians with special knowledge of henge monuments: Marta Díaz-Guardamino, Fraser Brown, FSA, Kasia Gdaniec, Alex Gibson, David McOmish, FSA, Al Oswald and Alison Sheridan, FSA. Following a precedent set by General Pitt Rivers, we invited them to comment on the excavated features and consider our understanding of them. James Connelly and Stephen Leach – two philosophers who specialise in Collingwood’s thinking, including his approach to archaeology – visited the excavation and discussed its results. When the project was planned, we benefitted from the advice and encouragement of Collingwood’s daughter, the late Teresa Smith. We must also thank Joanna Lawrence for her help.
The work was undertaken under demanding conditions by a wonderful team of local archaeologists. It is impossible to thank them sufficiently for their skills, resilience and commitment to such an unusual project.
Illustrations from the original excavation reports are reproduced by permission of the Cumberland and Westmorland Antiquarian and Archaeological Society and Bersu’s section drawing is reproduced by permission of Historic England.
ABBREVIATIONS AND BIBLIOGRAPHY
Abbreviations
- Bodleian
-
Bodleian Library, Oxford
- Historic England Archive
-
Historic England Archive, Swindon National Office, Swindon
Open access
