8.1 Introduction

1. The Roman military occupation of South West Scotland took place over a short period, spanning little more than 60 years in three main episodes (Flavian, Antonine and Severan). The influence of Rome on indigenous communities in the region may have been much more persistent and enduring over the first four centuries AD. The beginnings of Roman interest in Scotland are obscure, but it is unlikely that the Roman military was active here until after AD 69. Prior to this date, the Romans appear to have used their alliance with the Brigantes of what is now North East England to keep order along their northernmost frontier. Their arrangement was made with the Brigantean queen Cartimandua; when she was deposed by her husband, Venutius, the treaty that bound the Brigantes to Rome in a client status ended. A period of political turmoil and instability followed, prompting some degree of Roman intervention (Hanson 1983, 33).

2. The Brigantean revolt coincided with the ascension of the Flavian dynasty to the imperial throne, comprising Vespasian, Titus, and Domitian. The Roman army’s campaigns in Scotland during this period took place under the command of its general, Agricola. Hence, the Roman sites attributed to this time are described variously as ‘Flavian’ or ‘Agricolan’, covering circa AD 69 to AD 96. A chain of forts was established on the Forth-Clyde isthmus, to the north of our region, by circa AD 80 to AD 81, with military campaigns in AD 82 thought to focus on South West Scotland.

3. Several Roman temporary camps, established by troops on campaign to provide safe accommodation, are likely to date from this campaign in the region. Particularly characteristic of these Flavian camps are those of the ‘Stracathro’ type, which feature a complex array of additional defences around their entrances. Inland examples are known at Beattock (Bankend) and Dalswinton (Bankfoot) with camps from the Agricolan campaigns on the Ayrshire coast are known at Girvan Mains and Ayr Academy. Evidence of more permanent Flavian defensive structures has been recovered from the forts at Ladyward and Loudoun, as well as the fortlet at Beattock (Barnhill) evidencing post-campaign occupation forces.

4. The Flavian campaigns and occupation of Scotland were short-lived, with the Roman forces withdrawn by AD 90 under Domitian (ibid. 50) to the Tyne-Solway isthmus. At first, this frontier appears to have been porous, defined by a line of forts linked by a road. Several forts were maintained to the south, but also – potentially – to the north. To the north, it has been suggested that in our region Birrens, Broomholm, Dalswinton (Bankhead), Glenlochar and Milton (Tassieholm) were maintained as forward-sited outposts. Hence, the Romans may have continued to assert their authority within South West Scotland through these bases.

5. There followed a period of consolidation along the Tyne-Solway frontier under both Trajan (AD 98 to AD 117 AD) and Hadrian (AD 117 to AD 138). This included the construction of a solid barrier along its length: the celebrated Hadrian’s Wall. In this period, many of the northern forts, like Birrens, appear to have been progressively abandoned. Excavations have revealed no evidence of destruction, suggesting instead a managed withdrawal.

6. For almost five decades, the Romans held the Tyne-Solway frontier. This situation changed, however, when a new emperor, Antoninus Pius, succeeded to the throne with the Roman sites attributed to this period described as ‘Antonine’. His more aggressive approach coincided with the arrival of a new governor in Britain, Q. Lollius Urbicus, and between AD 139 and AD 145, there were renewed campaigns northwards into Scotland. The northern frontier returned to the Forth-Clyde isthmus, where a new frontier – the Antonine Wall – was constructed from circa AD 142. The north-south route along the Nith Valley played a crucial role as a transportation link between the Tyne-Forth line and the new frontier, and a flurry of contemporary activity is evident here, and elsewhere across the region. Birrens was rebuilt on two occasions, new forts like Carzield were established, and a number of military camps and fortifications were either constructed in new locations or rebuilt on earlier Flavian sites.

7. To date, Antonine sites have been identified across Galloway, dense concentrations have been found in the Nith Valley but also around Drumlanrig and Durisdeer, and their presence has been noted inland in East Ayrshire at Loudoun. While Antonine military sites are not known in lowland Ayrshire, contemporary material culture has been recovered from Poteath Burn, near West Kilbride, close to a putative harbour site at Brigurd Point.

8. Ongoing work at Burnswark, where a hillfort was surrounded by substantial Roman siegeworks, suggests a confrontation between the Roman forces and the indigenous population. Fieldwork has recovered lead bullets (glandes) from slings and stone balls projected by a form of mechanised catapult called the ballista. The large numbers of these items, combined with their mid 2nd century AD date, has some researchers suggesting a siege and subsequent storming of the hillfort may have formed one of the opening manoeuvres of the Antonine campaign (Reid and Nicholson 2019).

9. Again, the Antonine occupation of southern Scotland was not sustained, with forces withdrawn to the Tyne-Solway frontier after two decades. Their withdrawal in circa AD 160 is characterised as a failure of political will combined with military demands arising elsewhere in the Empire.

10. This second withdrawal would not mark the end of Roman military campaigning in Scotland. There are records of retaliation north of Hadrian’s Wall circa AD 184 under Commodus (AD 177 to AD 192), and the Antonine Wall was briefly reoccupied circa AD 197 under Septimus Severus (AD 195 to AD 211). More notable was the major incursion during the early 3rd century (AD 208 to AD 213), termed the Severan campaigns. The Emperor, Septimus Severus, with the Imperial Army campaigned against the Maeatae confederation in the central belt of Scotland before pushing north to confront the Caledonian Confederacy circa AD 208 to AD 209. Though militarily successful, after the emperor’s death in AD 211, Caracalla, his son, withdrew troops back to Hadrian’s Wall by circa AD 213.

11. In terms of the inscriptions and literary sources, the evidence of the Severan campaigns is slight (Reed 1976, 92). Reliance is placed instead on finds of contemporary coinage as well as a series of substantial marching camps, which have been identified. The Severan campaigns saw a marked emphasis on the eastern side of the country, but contemporary coinage and other Roman items on sites in our region suggest there is scope for military activity within the region.

12. There are no recognised later military incursions beyond the wall within the archaeological records, although late Roman interventions and periodic occupation up to AD 410 cannot be discounted (Hodgson 2014). 

13. The Romans were the first literate society to leave their mark on the landscape and indigenous cultures of southern Scotland. Their legacy in the region included a range of monumental sculpture and inscriptions. Perhaps it was access to Roman literature and art which first enticed educated persons in the post-medieval and early modern periods to develop an interest in the Classical world of Ancient Greece and Rome. This would eventually develop into an informed study into Roman architecture and buildings (exemplified, first and foremost, by the buildings of Rome itself) and then into the study of lost sites through archaeology. It is therefore perhaps not surprising that the first recorded interest in the Roman monuments of South West Scotland was during the 18th century, when the spirit of the Enlightenment was at its height.

14. The earliest work that refers to Roman sites in South West Scotland is Alexander Gordon’s Itinerarium Septentrionale. Here he postulated that Agricola camped at Burnswark on his way north during his campaigns (Gordon 1727), thus placing the Roman fortifications within the Flavian period. Gordon saw the earthworks on either side of the hill as representing two parts of a single Roman camp, which he described as ‘the first Roman camp of any to be met with in the South of Scotland, and the most entire and best preserved I ever saw’ (ibid.). Gordon also ascribed a Roman origin to the remains of a fort at Birrens, suggesting that it may have functioned as an exploratory castellum to the noble camp at Burnswark. A few years later, in 1731, several pieces of Roman sculpture were discovered at Birrens, including two altars dedicated to Mercury, and a third to Brigantia; these were placed in the Penicuik Collection of Antiquities. Other 18th century references to Roman field monuments include commentary on possible sites like the ‘Roman station’ at Mains of Southwick on the Solway coast by Lady S Riddell in 1779 (Riddel 1785). Whether this or any monument identified in the 18th century is Roman should be treated with caution without more recent confirmation.

15. By the end of the century, William Roy’s Military Antiquities of the Romans in North Britain was published in 1793. Roy’s work included a detailed plan focused on a tranche of landscape located in the lower part of Annandale. This depicted Torwood and stretches of road which Roy ascribed to the Roman road network. Roy also undertook detailed surveys of Roman fortifications at Torwood and Birrens, and the complex group of Roman and indigenous defenses at Burnswark. A further site at Tibbers Castle, identified by Roy, and included amongst this group, would subsequently prove to have medieval origins.

16. Like Gordon, Roy sought to place the fortifications at Burnswark and Birrens into some kind of chronological and functional relationship. But his conclusions were different to those reached by Gordon; he suggested that Birrens provided the headquarters, while the earthworks at Burnswark derived from summer outposts. He also, crucially, suggested that both sets of remains might not derive from the same phase of Roman activity, with the ‘camps’ at Burnswark potentially being later in date. The work of both these earlier pioneers would later be subject to a detailed critique by James Macdonald (see Macdonald 1894 for full discussion).

17. On a local level, some degree of contemporary interest in Roman archaeological sites is provided by the pages of the Statistical Accounts, compiled in the final decade of the 18^th century. The ministers who prepared these summaries of their parishes were aware of the antiquities to be found there, and most included them in their accounts. This provides us with additional early references to several now-confirmed Roman sites including Raeburnfoot, Birrens, and Burnswark.

18. By the 1860s and 1870s, several antiquarians were actively investigating the Roman military presence through excavating the upstanding remains of sites. Attracting most attention at this time were the Antonine Wall – then thought of as the Empire’s northernmost frontier – and the Roman antiquities of southeastern Scotland. However, there was active interest in our region with the four Roman altars found at the fort of Birrens exhibited by the Senatus of the University of Edinburgh in 1866. These are presumably the same altars which were gifted to the Society of Antiquaries by the landowner, Sir George Clerk of Penicuik.

19. The late 19^th century antiquarian approaches to Roman archaeology invariably focused on establishing some kind of concordance between Roman literature and modern landscape features. A good example of such work was undertaken by Captain Frederick Thomas. Thomas, an experienced naval surveyor as well as an antiquarian, compared contemporary maps with one compiled by his Roman counterpart Claudius Ptolemy prior to AD 120. Thomas’s findings, published in 1876, included detailed arguments regarding the placing of ‘Trimontium’ in the vicinity of the Eildon Hills, and, following on from that, the linking of ‘Uxellum’ with Burnswark (Thomas 1876).

20. John Callendar subsequently published a short account in the Transactions of the Dumfries and Galloway Natural History and Archaeology Society (the Transactions) discussing the site of a ‘Roman Camp’ at Springfield Hill, Dunscore (Callander 1890). Viewed in retrospect, Callendar’s analysis illustrates the challenges faced by the antiquarian community as they sought to make sense of the plethora of sites and monuments which confronted them and impose some kind of order upon them. Based on the physical characteristics of the monument, in particular its defensive hilltop position and proximity to the line of a Roman road, Callendar suggested that this particular ‘Roman Camp’ was constructed as an observation post circa AD 182, attributing it to the Flavian period. He proposed (ibid. 5) that it represented ‘not a position taken up by an army on active service in the field, but a permanent station held by a small force in time of peace’. Following its initial construction, he argued that the site saw repeated phases of occupation and abandonment during the Agricolan and Severan campaigns. A Roman origin has subsequently been discounted for this site, which has now been reclassified as a contour fort, almost certainly later prehistoric in date.

21. Efforts to improve awareness of and classify the numerous enclosed defensive sites occurring in the region were undertaken by David Christison and Frederick Coles during the 1890s. Both men compiled inventories which included a range of sites of widely differing character. Today, we would recognise many as having prehistoric, probably later Bronze Age or Iron Age, origins, while others are medieval in date. Some, however, had the potential to be interpreted as Roman.

22. Coles compiled a descriptive catalogue, which included survey plans, focused on the Stewartry of Kirkcudbright and Galloway (Coles 1891, Coles 1892 and Coles 1893). He was able to dispute an assumed Roman origin for two sites in his study area. The first was the ‘Roman Camp,’ at Little Duchra (Coles 1893), to which Coles was able to ascribe a possible origin in the previous hundred years. The second was the Mote of Polcree, which had been labelled by the Ordnance Survey as a ‘supposed site of camp,’ on account of its square plan. Coles again disputed this interpretation, arguing that although the site was square on plan, it could not demonstrate the level of engineering expected on a Roman military fortification. A third site, named ‘Castramont’ was dismissed as the site of a house or cottage demolished prior to the construction of the present Castramont House. With these sites discounted, a Roman presence in Galloway would not be competently confirmed for another six decades.

23. Christison’s focus lay in Dumfriesshire, particularly Annandale, and Ayrshire (Christison 1891 and Christison 1893 respectively) including confirmed Roman sites like the outworks of Burnswark and the fort at Birrens. His inventory was organised not by geographical location but by type, with sites categorised based on form and plan. His account opened with a group described as ‘rectilinear forts.’ Into this group he placed Burnswark and Birrens, making use of the guiding principle popular amongst the antiquarian community at the time, namely that Roman sites could be differentiated by their rectilinear plan, which reflected the superior survey and engineering techniques used by the Roman military surveyors.

24. Another prominent figure at the time was James MacDonald who explored another archaeological site type: the Roman road. Writing for the Proceedings of the Society of Antiquaries of Scotland (the Proceedings), MacDonald was critical of established antiquarian wisdom, noting how Roman origins were traditionally ascribed to long-established roads with any level of antiquity about them. Using the extant length of ‘Roman road’ linking Dalmellington and Ayr as an example, he argued that ascribing a ‘Roman’ date and origin to the various ‘Roman roads,’ ‘Roman baths’ and ‘Roman camps’ that proliferated across the landscape should be based on hard evidence, rather than sentimentality or an attachment to oral tradition (MacDonald 1893). Such evidence could only properly be obtained through excavation. A similar assessment of the Dumfriesshire Roman roads network followed, and in the paper, MacDonald presented a detailed critique of the earlier antiquarian works of Gordon and Roy (MacDonald 1894). MacDonald’s activities in the region attracted local interest, attested by an account of his investigations included within the pages of the Transactions (Johnstone 1893).

25. MacDonald was critical of the way in which both Gordon and Roy tended to represent the region’s Roman sites in their idealised form. He also questioned the case that Roman and indigenous Iron Age works could be distinguished based on engineering skill alone, proposing that some structures interpreted as Roman may be indigenous works in imitation of Roman forms. Of the remains at Torwood, he was particularly sceptical though he conceded a Roman origin might be confirmed; he preferred that artefactual evidence was obtained before accepting a Roman attribution. At Birrens, where the discovery of Roman sculptured stones had confirmed the site’s Roman origins, he remained cautious in ascribing all the site’s phases to Roman occupation. MacDonald pointed out that there was ‘nothing in the classical writers to show that the Romans practiced such a mode of fortification’, that is, the use of double, triple or quadruple ramparts. This statement demonstrates how closely coupled in the 1890s the study of Roman physical remains was with the analysis of classical texts. The only way of establishing a better understanding of the site was, he argued, to excavate (ibid. 306), stipulating that such work required proper supervision and skilled hands. For his closing comments, he argued that if the sites – Burnswark, Birrens, and Torwood – could be shown to be Roman, then their associated ‘Roman’ roads could then be accepted as such.

26. Work to test these hypotheses began in 1895 with excavations at Birrens carried out by the Society of Antiquaries of Scotland. The archaeological investigations were supervised by respected figures, no doubt reflecting the importance placed in these works. As well as James MacDonald, who was Vice-President of the Society, the group included David Christison, John Cunninghame and Rober Munro. John Barbour, an architect who surveyed numerous archaeological sites across Galloway and Dumfries, carried out the planning and drawing. The ensuing reports, published in the Proceedings, included a summary of the excavations, a report on the sculptured stones, and a small finds report detailing the ceramic and metalwork assemblages – authored by Joseph Anderson, Assistant Secretary of the Society and Keeper of the Society’s Museum (Christison et al 1896, Christison et al 1899).

27. The interest in this work was reflected in the 1895 volume of the Transactions publishing a paper by Macdonald summarising his hypotheses regarding Birrens and Burnswark, and critiqued the various antiquarian writers and surveyors – including Gordon and Roy – who had previously examined the site (Macdonald 1895). MacDonald also presented a paper on the Roman inscribed stones of Dumfriesshire at a meeting of the Dumfries and Galloway Natural History and Archaeological Society which was subsequently published in the Society’s Transactions (MacDonald 1896).

28. These early investigations at Birrens were important in several respects They confirmed the defences as Roman and secured as complete a plan as possible of the fort’s internal layout (summarised in Birley 1938a, 276). This attribution was despite MacDonald having expressed caution about whether the site’s then-present form was entirely down to Roman activity, as its multiple ramparts could not be matched with other comparable Roman sites. As excavation progressed, it also differed from expectation in having earthworks composed entirely of earthen construction, without any stone element present. As well as revealing unusual aspects of the fort’s plan and construction, excavation revealed two distinct occupation phases. Further, the recovery of inscribed stones included one stone tablet which could be dated not just to the reign of a specific Emperor (Antoninus Pius) but to an exact year within his reign. This was AD 153, when he was granted tribunal power for the 16^th time, that is, 16 years after he’d been made Emperor. Inscriptions also confirmed that the fort was manned by the 2nd Cohort of Tungrians – a cohort of Germanic origin – which formed part of the Sixth Legion, the LEG VI Victrix. The level of detail inherent in these finds included the names of some who had served at the fort. In all, a total of four individuals were represented on various dedications.

29. Even when the excavations at Birrens were completed, enthusiasm remained for the investigation of potential Roman sites in the region; attention turned next to the fort at Raeburnfoot. James Barbour, who had carried out the planning and recording at Birrens, led the excavations which took place in 1896. Barbour noted that the layout of the site bore similarities to Birrens, and that the ceramic assemblage was comparable (Barbour 1898). From these findings, a Roman origin was confirmed, although the excavators considered the site inferior in importance to its counterpart at Birrens (ibid.).

30. The excavations at Raeburnfoot were then followed by investigations at Burnswark, a focus of interest since the 18^th century on account of its unusual outwork fortifications encircling the hillfort. Again, the excavator was James Barbour, working on behalf of the Society of Antiquaries of Scotland. The aim of his investigations was twofold. Firstly, he intended to test the accuracy of plans surveyed initially by William Roy, and later by the Ordnance Survey. Secondly, he aimed to assess the character and structure of the various ditches, ramparts and gateways to better understand the site (Barbour 1899a, Barbour 1899b). Rather than ascribing the entire complex to the Romans, these excavations interpreted the outworks as Roman fortifications constructed at the base of a hill which was surmounted by an indigenous ‘British’ fortification. This theory, which had been mooted, though not universally accepted, from the late 18^th century onwards, became the established understanding after these excavations, although other possibilities were still considered.

31. As Barbour examined the various items of material culture recovered, he noted finds of lead sling shot from the entrances to the hillfort (Barbour 1899a, 62). These were further discussed by Christison in an 1899 paper included in the Proceedings (Christison et al 1899, 218). These examples of lead sling shot reinforced the original association between the fortifications at Burnswark and the Agricolan campaign, as it was believed that lead sling shot fell out of use by the Roman army by the end of the 1st century. This assumption was made despite the recovery of two coins dating to the reign of Trajan (AD 98 to AD 117 AD).

32. The early decades of the 20^th century saw attention largely turn away from the South West. Through recourse to a combination of Roman writings and archaeological evidence recovered from excavations of forts on the line of the Antonine Wall, the three main phases of Roman military occupation of Southern Scotland were identified. These were then refined through reference to excavated material. A prominent figure in this process was George MacDonald, who contributed both shorter papers and longer works devoted to the Roman occupation of Scotland including publishing a comprehensive overview of the Antonine Wall (MacDonald 1911).

33. In 1918, MacDonald compiled a corpus of Roman coin finds in Scotland, noting that those from Birrens included an older coin bearing the name of Germanicus, Tiberius’ nephew. The coin was minted in AD 19, potentially demonstrating that the extant remains of the Roman fort, believed to have origins in the Antonine campaigns, may have replaced an earlier structure established during the Agricolan campaigns (MacDonald 1918).

34. This potential for multiple occupation phases was being increasingly recognized through evidence from a number of forts, including Birrens. MacDonald himself used his coin corpus to present an up-to-date account summarising three phases of Roman military activity north of the Tyne-Solway frontier. Phase 1, the Agricolan period, commenced around AD 80 and ended with the ascent to the throne of the Emperor Trajan in AD 98. Phase 2 was the Antonine period, which began with the erection of the Wall in AD 144 and lasted until the reign of the emperor Commodus, who began sole rule in AD 180. Phase 3 was represented by the Severan campaigns of AD 207 to AD 211 (ibid.). While these phases had been recognised and the terminology widely used by those actively studying Roman military archaeology, MacDonald’s report provided a useful summary of where levels of knowledge had reached at the time, and how this influenced future study and research.

35. In the same edition of the Proceedings, Professor Haverfield, who appears to have been working closely with MacDonald and others actively researching Roman Scotland at this time, presented a short paper on Agricola and the Antonine Wall (Haverfield 1918). Like MacDonald, he argued that several Antonine forts, particularly those located along the length of the Antonine Wall, had been established on the sites of earlier forts laid out perhaps 60 years earlier during the Agricolan campaigns.

36. Although active exploration had moved away from the region, interest remained high. It was during this period, in 1912, that a Roman altar was recorded at Westhills, near Gretna, by a survey team from the Royal Commission on the Ancient and Historical Monuments of Scotland (the Commission). Probably not in situ, it does not appear to have featured prominently in the wider literature of the time. The Commission had also surveyed in 1912 the earthworks of the Roman camp at Gilnockie, depicted on the Ordnance Survey since the first edition in 1862 as ‘Camp’. This burst of activity by the Commission was to inform the compilation of their Inventory for the county of Dumfries (RCAHMS 1920). Early plans and summaries of the archaeology of the County are within this volume including the siege works at Burnswark (termed Birrenswark) and Roman carved stones in Hoddom. One of the roles of the Inventory was to propose those monuments that merited protection – for Dumfries, the RCAHMS advocated for the protection of Westhills, Gilnockie, Raeburnfoot and Birrens (ibid.).

37. In December of that same year, George MacDonald presented a talk to the Dumfries and Galloway Natural History and Archaeological Society, subsequently printed in the Transactions, of how work in the region contributed to the wider national picture (MacDonald 1921). This summarised the findings from Birrens, confirming two phases of occupation and giving insights into its layout. He touched upon how the site functioned in its wider context, and how its occupants fitted into the structure of the Roman army. He identified additional sites which confirmed a Roman presence in Annandale, including Raeburnfoot and Torwood, and argued that there was potential for more sites to be discovered along the proposed Roman route through Annandale, and, perhaps, in Eskdale.

38. While much of this account represented a distillation of past and current work, one new finding was recorded in MacDonald’s account – the discovery in 1915 of a Roman dressed stone incorporated into the fabric of the old parish church in Hoddom. This occurred during excavations undertaken in the churchyard, which exposed the medieval foundations of the church. Finds included an inscribed stone marked with the name of Leg VI Victrix, the same legion which had left its mark at Birrens. A further update was the identification of Samian tableware from Eastern Gaul by James Curle, the specialist working on the Birrens pottery assemblage, potentially dating to the Hadrianic period. This supported the theory that the first phase at Birrens was contemporary with the construction of Hadrian’s Wall in AD 79.

39. In the mid-1920s, Robin Collingwood revisited Burnswark, opening trenches across the earthworks and undertaking a survey of the upstanding features. In particular, he reinterpreted the triangular earthwork known as the ‘West Fort,’ disputing its attributed Roman origin, and ascribing a medieval date. He also questioned the existence of Barbour’s alleged paved roadway, identifying an old trackway instead (Collingwood 1926). Collingwood also argued that the North Camp remained unfinished, proposing that the siege evolved through the initial construction of the South Camp, with its three ballista platforms, with the North Camp established later following the arrival of reinforcements which allowed the commander to move from a siege to a close blockade of the indigenous fort.

40. Several years later, a short paper by Collingwood in the Transactions (Collingwood 1929) focused on the Roman camps. He noted how such sites were difficult to identify on account of their ephemeral nature and the fragility of their visible remains in the face of cultivation. He identified three camps on the western route from Carlisle to the Clyde Valley via Annandale: Cleghorn and Little Clyde, both in Lanarkshire, and Torwood. He noted each was placed around 20 miles from its neighbour, spacing them at roughly similar intervals along the route. One interesting aspect, which he suggested required further investigation, was how the camps on the eastern route appeared to have been used multiple times, in contrast to their western counterparts, which appeared to have been used only once. This assumption would later be called into question once the widespread adoption of remote survey techniques, in particular aerial photography, started to reveal more evidence for Roman temporary camps in the region.

41. Increasing quantities of excavated material meant increasing opportunities for artefact analysis. 1931 saw the publication by George MacDonald of an article on the potters’ name stamps on the fine Roman tableware, known as ‘Samian Ware’, recovered from Scottish sites (MacDonald 1931). His list – which recorded a total of 166 names on sherds derived from 337 different vessels – included several examples from Birrens. As well as compiling an inventory of this material, MacDonald then used the ceramic record as a means of supporting both the structural evidence from the sites and the numismatic evidence. These findings could then be used to address questions regarding the nature and longevity of the Antonine occupation of southern Scotland. The pottery finds from Burnswark were also subject to reassessment by Eric Birley, who found sherds of 2nd century date which suggested two separate phases of activity at the site. The earlier phase comprises a Flavian fortlet, while the later siegeworks were assessed as Antonine (Birley 1938, 301).

42. Birley noted how MacDonald had viewed Birrens in the wider context of Roman military activities in Scotland – ‘Caledonia’ – recounting MacDonald’s depiction of the site as ‘a touchstone for the vicissitudes of the northern line’. In the intervening years since MacDonald had carried out his work at Birrens, further excavations had been undertaken on the northern English forts on and around Hadrian’s Wall. This led Eric Birley to undertake excavations at Birrens from 1935 to 1937 working from this premise of considering the fort ‘as an element in the Hadrianic frontier rather than as a link in the Roman occupation of Scotland’ (Birley 1938), on a par with the forts of Netherby and Bewcastle.

43. Birley argued that while the 1895 excavations had revealed evidence for at least three phases of use, with the fort reoccupied in the 3rd century AD, he saw the potential for more. Birley’s excavation provided evidence to increase the number of phases to five, arguing for: a timber fort of possible Hadrianic date in Phase 1, rebuilding in stone although with a turf rampart in Phase 2, building an extension with internal buildings reconstructed in Phase 3, both in the Antonine period, the Severan reconstruction in Phase 4 and – finally – further reconstruction and the building of an aqueduct in Phase 5, taking place, according to Birley, during the 4th century AD at the time of Constantinus Chlorus. This chronology was predominantly supported by the large pottery assemblage recovered by these excavations.

44. With his work at Birrens complete, Birley turned his attention to the Roman fort at Carzield, working with Ian Richmond on excavations in March 1939 (Birley and Richmond 1940). An account of these investigations was then presented in the Transactions. While the site’s Roman levels had been compromised in places during the medieval period, the excavators assigned an Antonine date for the fort’s occupation based on pottery evidence and confirmed its occupants as a cavalry regiment. This was an entirely different composition to Birrens, which had housed auxiliary troops with a small detachment of cavalry. The fort at Carzield was situated on an important route in the road network, at one of the few points on the main route north through Nithsdale where east-west movement was practicable.

45. The same paper by Birley and Richmond flagged additional new discoveries in the region from the use of aerial photography to identify cropmarks and, less commonly, earthworks of Roman military sites within our region. In particular, they noted the rediscovery of a ‘lost’ camp at Ward Law, previously identified by Pennant. This site was important as its location, close to the coast and the Nith estuary, raised the potential for a nearby harbour. This would have been vital for a military reliant on efficient, reliable transport networks for resupply. Of note, Ward Law was one of the earliest examples of Roman military sites in the region discovered through aerial photography in June 1939 (Jones 2011, 19). Where earthworks no longer survived as perceivable upstanding features in the landscape, they could still be identified through variations in crop growth, soil marks or subtle shadows on low level aerial photography.

46. Ward Law was identified by OGS Crawford during this early period of discovery in the late 1930s. In the same 1939 flying season when he also located the West Gallaberry camp and captured additional details of many of the known military sites and the road network in Annandale and Nithsdale (St Joseph 1976). The subsequent rapid excavation of a trial trench across the camp ditch at West Gallaberry by Kenneth St Joseph confirmed its Roman characteristics (Taylor 1940, 162). However, other prospective Roman sites from this 1939 survey at Dalmakethar and Bushel Beck failed the subsequent trenching test (Clarke et al 1952, 21 and 101 respectively). The signal station observed at White Type was also located by OGS Crawford in 1939 (Crawford 1939, 282; St Joseph 1952).

47. The use of aerial photography coupled with trial trenching would be enthusiastically adopted in the post-war period and often was synonymous with St Joseph’s name. A paper by Eric Birley and Ian Richmond (Birley and Richmond 1940) predominantly focused on Carzield gave advanced notice of several sites identified by this combined technique in the region. This included a discussion on Ward Law considering it ‘the terminal sea-port fort’ on the road network in eastern Dumfriesshire. The paper advocates a structure for the road network where the construction of ‘patrol posts’ subdivided the road ‘into short sectors for convoy and patrol, while the main garrisons are concentrated in rather large units, housed as at Carzield and Birrens’ (ibid. 163).

48. In 1946, St Joseph undertook an important excavation at a small fort identified by aerial photography at Loudon, East Ayrshire. Its discovery placed Roman penetration further inland than had previously been recorded, in a location northwest of the main north-south route that ran along the Nith Valley. The siting of this fort flagged up another potential route that ran east-west towards the Ayrshire coast. Evidence for four phases of occupation was identified, with three phases ascribed to the Flavian/Agricolan period, while the latest was believed to be of Antonine date. All of the phases appeared to be timber-built. The quantities of Antonine pottery recovered there appeared small in comparison to other sites of comparable age and size, suggesting to the excavators that the duration of its occupation had been relatively short. The site showed evidence of deliberate demolition on abandonment (Clarke et al 1952, 188-191). Investigative work was also undertaken by St Joseph at Raeburnfoot, Barburgh Mill and Craik Cross (St Joseph 1947), with further research carried out at Milton (Tassieholm), and field survey on the Roman road network in the vicinity of Raeburnfoot.

49. St Joseph presented a short account of his own investigations on potential Roman sites in the Transaction (St Joseph 1947). He focused in part on Raeburnfoot, where his trenching recovered a single, undatable sherd. Reassessment of an earlier pottery assemblage recovered during the earlier 1897 excavations revealed, however, several sherds of 2nd century AD mortaria, inferring a 2nd century AD date for the site. Also examined by St Joseph at this time was a site at Barburgh Mill, in Nithsdale, discovered only a few years previously in 1945 through the use of aerial photography. Excavation of the small enclosure failed to produce any artefactual evidence, but the V-shaped ditch and cobbled surfaces supported a Roman origin, and St Joseph interpreted the site as a small fortlet comparable with Durisdeer.

50. By 1946, Ian Richmond had written a reappraisal of the Roman road network in Dumfriesshire and Roxburghshire, which was published in the Proceedings (Richmond 1946). This included the results of a field survey undertaken in July 1945 by Richmond, St Joseph and others, which had identified some well-preserved remains of a high-level route linking Dumfriesshire and Roxburghshire. He confidently ascribed this road to the Romans on the basis of its high-quality construction. Sadly, no traces of this feature had survived in proximity to the nearby fort of Raeburnfoot. In this same paper, Richmond made use of a new term – the ‘fortlet’ – used to describe the type of small fortification exemplified by the examples at Durisdeer and Milton (Tassieholm).

51. Several years later, Richmond’s identification of a putative Roman road was endorsed by Angus Graham of the Commission, again within the pages of the Proceedings (Graham 1948). Graham’s short paper noted how a detailed survey of the monument along its route had revealed stretches where its dimensions and character had proved consistent with comparable examples identified and examined in other locations.

52. A long-running series of excavations at Milton (Tassieholm) fortlet were undertaken by John Clarke and published in the Transactions from 1946 to 1950 (Clarke 1946, Clarke 1947, Clarke 1948, Clarke 1949, Clarke 1950), following an early assessment of the site (Clarke 1940). The site had initially presented as a small square enclosure; excavation near the entrance revealed the remains of a turf rampart with a stone rampart base. This small feature occupied the same site as a much larger marching camp, first noted by William Roy two hundred years earlier. Investigations had begun there in 1939, but it was only after World War II, in the early part of 1946, that proper exploration could begin, with ‘highly intelligent’ German prisoners-of-war providing some of the labour (Clarke 1948, 161). Initially, the site’s apparently irregular quadrilateral plan caused confusion. The excavator, however, with input from Anne Robertson (who would later work at Birrens), finally established that its unusual characteristics derived from the fact that two partly overlapping phases of fort structure were present. Three phases of occupation were eventually identified, with pottery finds indicating that all three phases of occupation occurred during the early Flavian period. John Clarke also in this period excavated the ‘patrol post’ in the Durisdeer gap, showing this to be of Antonine date (Clarke et al 1952, 124).

53. In addition to the study of field monuments, advances were also being made in material culture studies. A 1946 paper by Eric Birley and John Gillam presented an analysis of the pottery assemblage recovered from Carzield (Birley and Gillam 1946). The assemblage, which included both Samian ware and coarse wares, was small, but represented a consistent date range of material spanning the period AD 140 to AD 160. This demonstrated a short span of activity during the Antonine period, running counter to the excavated evidence provided by the intervallum road. This had shown two phases, which – like other Roman sites in the region – suggested that the fort might have been re-occupied. Artefactual evidence for this later occupation may have been erased by later activity on the site. It was a problem, they acknowledged, that could only be addressed by further excavations, and the recovery of a larger pottery assemblage with a broader chronological span of material.

54. The 1950s saw several important contributions to Roman material culture studies in our region and further afield. The focus had often been on what the various material culture forms (in particular pottery) could tell us about dates of occupation, and often the duration of occupation, too. There was, however, potential in exploring how the indigenous population acquired and used these exotic items. By considering patterns of deposition in indigenous contexts, it was possible to see how the Romans themselves might have manipulated indigenous populations through the controlled dissemination of such exotica. Published at the very start of the decade was a revised inventory of Roman coin finds in Scotland compiled by Anne Robertson (Robertson, 1950), the first of a series of regular updates compiled and published every decade.

55. Robertson’s inventory was followed by a paper by Stuart Piggott, examining three important metalwork hoards of Blackburnmill, Eckford – both in the Scottish Borders – and also Carlingwark Loch, Castle Douglas (see Chapter 7 Iron Age). The crucial question faced by archaeologists studying important indigenous Iron Age hoards, like that of chariot fittings from Middlebie, was this: did these hoards, which appeared to derive from acts of votive deposition, represent offerings made by the local communities, or did they derive from the religious practices of Celtic or Germanic tribesmen serving as auxiliaries with the invading legions? Piggott opted for the former, arguing that the hoards were likely to have been composed of objects accumulated during the 1st or 2nd century AD, after the initial Roman incursion into Scotland (Piggott 1953). He also argued that the quantity of artefacts with demonstrably Roman origins was sufficient to suggest that the hoards were associated with a Romanised population. Displaced tribesmen of the Brigantes, who had been forced north following their revolt of AD 69, were a possible option. At around the same time, Eric Birley was reassessing the Brigantian problem in greater depth (Birley 1951).

56. The Glasgow Archaeological Society published a seminal overview of the understanding of South West Scotland in 1952 in a volume titled The Roman Occupation of South-Western Scotland (Clarke et al 1952). This carried summary reports of excavations and surveys undertaken under the auspices of the Society. These included overviews on the road network, Birrens, Burnswark, Torwood, Dalmakethar, Ward Law, Galloberry (West Gallaberry) and Barburgh Mill by St Joseph, with also Milton (Tassieholm) and Durisdeer by Clarke. The quality of mapping, photography and general illustration set a new standard in communicating the emerging archaeological information.

57. Meanwhile, aerial photography continued to add new sites to the corpus, with the work of St Joseph again featuring prominently. A previously unknown Roman fort was found on the east bank of the River Dee north of Castle Douglas at Glenlochar, with another much smaller fort identified further to the west at Gatehouse-of-Fleet (Richmond and St Joseph 1953) while a probable camps were located north of Sanquhar at Knowe Cottages (Clarke 1953, 118), at Shawhead, west of Dumfries (St Joseph 1961) and two camps at Kirkpatrick-Fleming (St Joseph 1951 and 1961).

58. Glenlochar was subject to trial-trenching in the spring of 1952 to confirm Roman origins. Evidence for three phases of fort construction was obtained, comprising an early Flavian phase and two later Antonine phases, a picture comparable to the intensively studied Newstead complex. The dating was based on the pottery assemblage, studied by John Gillam (Gillam 1952). Interior details could not be identified, but the fort was larger than its counterpart at Carzield, comparable instead with the legionary fort at Birrens. The vagaries of the road which connected Glenlochar to Gatehouse-of-Fleet were not yet understood, but it was clear to St Joseph and Ian Richmond that the Gatehouse of Fleet fortification did not represent the terminus of the route (ibid. 1).

59. These discoveries ran contrary to an earlier claim by Sir George MacDonald, made in a 1932 speech to the Classical Association and referenced in a paper by John Clarke (published in the same 1952 volume of the Transactions) that the Romans ‘never mastered this part of Scotland,’ that is, South West Scotland to the west of the Nith Valley (Clarke 1952, 11). Clarke went on to summarise recent discoveries found within ten miles to the north of Carzield fort in Upper Nithsdale. These included the camp at West Gallaberry, a site at Dalswinton (which was already described as ‘a second Newstead’), a road-post at Barburgh Mill, a signal station at Thornhill (known as Templand Mains), three Roman structures at a ford in the River Nith at Carronbridge and two phases of temporary camps at Durisdeer (ibid.).

60. The discoveries at Glenlochar and Gatehouse-of-Fleet, discussed more fully in the same volume by Richmond and St Joseph, were also mentioned. In addition, a crucial new discovery had been made – again by aerial photography – further to the north at Loudon, East Ayrshire. Once again, both Flavian and Antonine occupations were proposed for the latter site. There was also evidence of another Roman road, running east along the Avon Valley to link the fort at Loudon with another fort to the east at Castledykes in Lanarkshire.

61. Clarke suggested that these new discoveries might confirm certain aspects of the Agricolan campaign as recorded by Tacitus, which noted how Agricola had stationed troops in that part of Britain which looked over to Ireland. In his view, however, some difficulties remained. In particular, Tacitus suggested that the area was occupied by the Romans with the expectation of mounting an expedition into Ireland. Therefore, temporary camps might be expected, rather than the permanent forts suggested by the likes of Glenlochar and Loudon. The picture painted by the presence of these more permanent installations might instead, Clarke argued, suggest that there was a need to prevent an invasion from Ireland. Hence, the presence of these forts – occupied during the 1st century AD Flavian and 2nd century AD Antonine period – which suggested that the region was permanently settled by the occupying Roman forces. At the same time, however, there were still areas where evidence for a Roman military presence was absent, including much of Galloway west of Gatehouse-of-Fleet and also much of the area now occupied by modern East, North and South Ayrshire.

62. Wider circulation of the aerial photographic evidence in journals like Antiquity inspired new fieldwork to be undertaken, testing the validity of the sites revealed there. Several individuals undertook work of this kind in South West Scotland, and the results of their fieldwalking exercises were presented in the Transactions of 1953. Amongst those who took part were eminent figures – well known both regionally and nationally – including Oswald Crawford, Ralegh Radford and Alfred Truckell of Dumfries Museum. Crawford and J A Inglis focused on various parts of the Roman road network which lay within the South West region (Crawford 1953; Inglis 1953), and while actual physical proof for Roman construction was not forthcoming, potential routes were explored and proposed by various contributors.

63. Meanwhile, excavations were underway at a complex of sites identified on aerial photographs at Carronbridge. Work was stalled following a local outbreak of foot and mouth disease, but eventually two seasons of investigations were undertaken during the summers of 1953 and 1954. An interim report authored by John Clarke and A B Webster was then published in the Transactions of 1954. The authors noted how Antonine occupation of Nithsdale appeared to be independent of the Flavian precursors, with Dalswinton constructed during the Flavian campaigns, and Carzield during the Antonine period. This did not match evidence from sites like Birrens and Glenlochar, where the same location was used in both phases of occupation.

64. In addition to the large-scale research excavations undertaken on these recently discovered sites, small scale work was being undertaken elsewhere, sometimes in response to threats imposed by building works. Alfred Truckell of Dumfries Museum recorded limited excavations undertaken by the museum on a large spread of midden material lying to the north of Carzield fort in the autumn of 1955 (Truckell 1955). Structural evidence included one substantial furnace and several hearths, apparently utilised for the burning of rubbish. A useful assemblage of material was recovered, including pottery, daub and some items of ferrous and non-ferrous metalwork. Eric Birley and John Gillam were charged with studying the material culture.

65. Initial conclusions suggested an early Antonine date which pre-dated the main phase of occupation at the fort. From this, the inference was that the material collected in the midden had represented site clearance undertaken in advance of the rebuilding in the later Antonine period (ibid.). In January of the following year, the laying of water pipes at the same site revealed evidence for a metalled road base and further finds from the edges of the same midden spread. These included a sherd from an amphora. It is interesting to note that in this early episode of ‘rescue-archaeology’ – a forerunner of the developer-led work which provides such an important contribution today – the responsibility for organising and undertaking the work lay with Dumfries Museum. Truckell also recorded visiting an eroded portion of the ditch at Birrens fort in order to retrieve several sherds of Roman pottery, and instances where the line of the Roman road running west from Raeburnfoot, initially traced by Angus Graham, was cut by forestry workers digging ditches and the exposed sections, recorded by John Forsyth.

66. Reaching a comprehensive understanding of the complex remains uncovered at Carronbridge proved elusive, as material culture finds (and in particular, pottery) were virtually non-existent. Four phases of enclosure were, however, identified and interpreted as Roman on the basis of their morphology and general character (Clarke and Webster 1954). Discerning a credible function for the site was less straightforward, with the size of the ditches and the character of the cobbled surface at the east Gate of Structure A suggesting a degree of permanence not required of a temporary marching camp. With excavations at the site concluded, understanding of the site’s various elements had to remain inconclusive.

67. The next site to receive attention was Dalswinton (Bankhead), in Nithsdale, which was excavated by Ian Richmond and Kenneth St Joseph in 1954 (Richmond and St Joseph 1956). Excavations followed the standard pattern of opening up a single trench across the interior of the fort in the south east quadrant and across the defences into the eastern annexe (Hanson et al 2019, 286). Another site first identified on aerial photographs, its presence was unexpected, as it was located just two and a half miles upriver from the fort at Carzield. Since Carzield had revealed only Antonine deposits, the authors wondered if the features identified here, which comprised evidence for two superimposed forts, might represent the missing evidence for Flavian occupation, which had been absent at Carzield.

68. The larger of the two fort structures at Dalswinton (Bankhead) proved to be the later of the pair. An annexe appeared to be associated with the earlier occupation of the fort, and this revealed evidence for a dismantled furnace structure, used for melting lead. This was brought onto the site in ingot form, potentially for use in the lining of water management features. Vestigial foundation trenches uncovered across the interior also appeared to derive from the first phase of the fort’s occupation. Both phases of occupation were likely, according to the authors, to be Flavian in date, although details of the Samian Ware suggested to the specialist, in this case Eric Birley, a potentially later origin in the realm of Trajan (Birley 1956). The following year, Eric Birley questioned, in the Transactions, a hypothesis presented by the excavators (that is, Richmond and at St Jospeh) that the larger, second phase of the fort at Dalswinton (Bankhead) had once accommodated a renowned regiment of cavalry, 1000 strong, known as the ala Petriana, on the basis of its size and configuration (Birley 1957, 9). Birley instead favoured the hypothesis that the regiment had been stationed at Stanwick, on the line of Hadrian’s Wall, with his discussion of the topic in the regional journal illustrating how, as all this additional physical evidence was being accumulated, the literary sources might still be mined as a source of understanding and explanation by those actively engaged in fieldwork.

69. In the following year, forestry ploughing at Burnswark led to the recovery of yet more lead glandes, with the finds reported and written up for the Transactions by William Cormack (Cormack 1958). A further 10 glandes, and an iron spearhead of typically Roman type, were recovered, again during forestry works, a few years later in 1960 (Cormack 1960). Meanwhile, the search for visible traces of Roman roads continued across the region and beyond, with field survey carried out by Alfred Truckell and Ralegh Radford, and regular reports published within the pages of the Transactions. Most of the observations and conclusions were, at this stage, conjectural: trial trenching on the site of a previously undiscovered fortlet identified at Bankhead, Kirkconnel (Reid, 1959; Clarke and Wilson, 1959a; Clarke and Wilson 1959b; Wilson and Clarke 1959) would, for example, prove inconclusive at this stage in trying to confirm a Roman date.

70. Another figure achieving increasing prominence in the region at this time was Anne Robertson, whose revised inventory of Roman coin finds from Scotland has already been mentioned (Robertson 1950). In the 1961 volume of the Transactions, she presented a report of the 1959-60 excavations at Raeburnfoot, along with a detailed appraisal of earlier works carried out on the site by James Barbour in 1897 and Kenneth St Joseph in 1946 (Robertson 1961). The most recent excavations were undertaken by members of Glasgow Archaeological Society and the DGNHAS, with works instigated by the Hunterian Museum, attached to the University of Glasgow. Their excavations revealed that the inner of the two enclosures at Raeburnfoot had only two opposing entrances, on the north and south sides, rather than the four previously postulated by Barbour. Robertson argued that the site had been chosen for the construction of the small, inner enclosure, with its larger, outer counterpart potentially representing a temporary camp built for use while the former, with its internal timber buildings, was under construction. The pottery indicated an Antonine date, with the site itself representing a small fort or large fortlet comparable with excavated examples from Duntocher on the Antonine Wall or Crawford, a relatively local example in Lanarkshire (Robertson 1961, 46).

71. In the same volume of the Transactions, Alfred Truckell announced a new series of excavations at Birrens, undertaken by the Scottish Field School in Archaeology (Truckell 1961). Two years later, an interim report – authored again by Anne Robertson – was published within the journal’s pages (Robertson 1963). As in her earlier report on Raeburnfoot, Robertson began the paper with a reappraisal of James Barbour’s work on the site, noting how he had not only produced a survey plan which presented an internal layout replicated at numerous other excavated forts, but also observed evidence for two phases of construction at the site.

72. After summarising the 1930s excavations on the site, undertaken by Eric Birley and Ian Richmond, and discussing the site’s ramparts, she described how aerial photographs had since placed the upstanding features at Birrens within a much larger complex of features. Kenneth St Joseph had, in recent years, identified two triple-ditched enclosures in a field to the west of the fort (the larger of which appeared to equate to Roy’s ‘annexe’), at least three temporary camps and possible small fortlet to the south and east of the fort, and a series of trenches which he described as a ‘mansio’ to the north (Robertson 1963, 141).

73. Robertson then detailed ongoing work at the site. Four distinct phases of construction and occupation have been identified. The first phases comprised an early Flavian enclosure, possibly associated with Agricola, lying mainly to the west of the later fort but with its east ditch underlying the later fort’s western rampart. The second phase was also tentatively attributed to the Flavian period. Here, a turf rampart with no stone base underlay the same lines as the later ramparts on the east, south and west sides, with the north side lying further to the south, thus enclosing a smaller internal area. Phases three and four were attributed to the Antonine occupation of the fort, with a stone base to the rampart and finely constructed stone buildings attributed to the activities of the Sixth Legion, and later rebuilding, with inferior workmanship evident, undertaken by the Second Cohort of Tungrians. Evidence for later occupation of the fort, post-dating the 2nd century AD, was not forthcoming (ibid. 154).

74. In addition to the major research excavations at Birrens, the early 1960s saw several seasons of exploration at the multiperiod site of Broomholm, undertaken by Charles Daniels of the University of Newcastle. Evidence for indigenous Iron Age occupation was uncovered there, along with later medieval structures. What was particularly interesting in this instance was that evidence of Iron Age occupation both pre-dated and post-dated a Roman military presence on the site. The Roman structure was described in an interim report compiled by Alfred Truckell of Dumfries Museum as much smaller than first thought, more of road-post size (Truckell 1962, 162).

75. A less celebrated discovery, and one which slipped beneath the radar for almost four decades, was the discovery in the 1960s of two substantial sherds derived from ring-necked flagons of 2nd century AD date from the vicinity of the Poteath Burn in West Kilbride. These finds suggested Roman activity in the area, potentially during the Antonine campaigns, supporting the possibility that the area around modern Hunterston might once have been used as Roman harbour facilities.

76. The second half of the 1960s saw a reduction in the amount of large-scale fieldwork reported, though a short note by Truckell, published in the Transactions of 1968, suggests that small explorations were still taking place, from time to time, at Carzield (Truckell 1969). The camps at Kirkpatrick-Fleming were significantly impacted by road building in 1968. In response, Gibbs undertook an excavation to recover a plan of the ditch in the threatened portion where it was not visible on the aerial photographs (Gibbs 1968).

77. Reviews of material culture were, however, ongoing. One such article was authored by Martin Henig and published in the Transactions of 1968. This referred to three recorded finds of Roman intaglios, small carved gemstones originally set within finger-rings. Two were casual finds from the military sites of Carzield and Burnswark, while the third had been recovered in association with a cremation beneath a cairn of large boulders at High Torrs, Luce Bay. Also accompanying the cremation were two Samian bowls of 2nd century AD date, several iron ‘spearheads’ and an incomplete iron ‘sword.’ The collection of grave goods suggested to Henig that the deceased was equipped in the manner of a Roman auxiliary soldier, contrary to an earlier hypothesis that the body had been that of a mariner lost at sea (Henig 1969). Henig’s interpretation would later be challenged by David Breeze and Graham Ritchie (Breeze and Ritchie 1980). Throughout the second half of the decade, new discoveries were still being made on aerial photographs, with Kenneth St Joseph identifying a previously undiscovered Roman camp at Innerfield, just west of the river Annan, in 1967.

78. Anne Robertson’s important contributions to the Roman Archaeology of South West Scotland continued during the 1970s. Perhaps her most outstanding achievement was the publication of the report detailing seven seasons of excavations undertaken at Birrens in the form of a substantial volume: ‘Birrens (Blatobulgium)’ (Robertson 1975). As well as presenting the results of recent work on the site, the volume presented a comprehensive summary of the earlier excavations undertaken variously by Barbour, then Birley and Richmond. It also presented a detailed finds catalogue.

79. In 1971, Robertson also presented the latest in the series of inventories detailing Roman coin finds in Scotland. Published in the Proceedings, this continued the tradition first begun by Sir George MacDonald (Robertson 1970-1). The following year, Roy Davies presented an alternative interpretation for the Roman fortifications at Burnswark. Rather than representing evidence for siege warfare, he suggested instead that the site had never been actively used in a conflict situation. Instead, it represented a training range where Roman military personnel were schooled in the use of artillery pieces (Davies 1972). This meant that – rather than being constructed in a war situation – the Roman earthworks at the site had been built and used during a time of peace.

80. New sites continued to be discovered through aerial photography during the late 1960s and 1970s. A further double-ditched enclosure at Dalswinton (Bankfoot) was identified on the floodplain of the River Nith below the fort at Dalswinton (Bankhead) (St Joseph 1973, 217). Since its defences were composed of double rather than single ditches, the site seemed more likely to be a fort, rather than a temporary camp. The presence of external traverses – tituli – outside the entrances was, however, more suggestive of a camp rather than a fort (Hanson et al 2019, 290). This fort lay within a larger temporary camp of the Stracathro-type, which in turn intersected a smaller camp of the same type that had originally been identified in 1949 (Jones 2011, 184). Viewed together, the Roman sites in and around Dalswinton demonstrated four phases of occupation, with two phases at Dalswinton (Bankhead) and two at Dalswinton (Bankfoot). It would, however, be several decades before advances in remote surveying would allow the intricacies of these sites’ histories to be teased out and presented more clearly in a 2019 publication authored by Bill Hanson, Rebecca Jones and Richard Jones (Hanson et al 2019.).

81. This same period saw a fortlet identified at Murder Loch from aerial reconnaissance by the Ordnance Survey in 1974, camps at Hangingshaw (St Joseph 1969), Annanfoot (Jones 1979) and at Ruthwell (Maxwell & Wilson 1978, 40), while the single camp at Ellisland was shown thrown re-flying to be a pair of small camps (St Joseph 1973). The Annanfoot discovery came from the Solway Survey led by Barri Jones, focusing on the reconnaissance of the lower lying areas of the Annan and Nith valleys. Jones considered this camp, at the tidal limit, to be ‘… the first possible indication of Roman naval transshipment to the north side of the Solway’ (Jones 1979, 4).

82. In the second half of the 1970s, several new sites were investigated across the region. 1976 saw the excavation of a fortlet at Outerwards, undertaken by Frank Newall who also initially identified the site through field survey (Newall 1976). Newall identified two phases of occupation, which he dated to the Antonine campaigns. In both time periods, the fortlets – largely timber-built – were demolished and burned during the process of abandonment. He also noted a potential site of Roman date in the intertidal zone off Brigurd Point, near Hunterston (Newall & Lonie 1972). Recent reappraisals of these features have been undertaken using aerial photography and, in 2013, as part of an intertidal survey linked with the SAMPHIRE project (McCarthy & Benjamin 2013). This work has confirmed the presence of these features, which may represent the surviving elements of a Roman harbour constructed in the Clyde estuary.

83. While 1977 saw a return to Carzield with small-scale explorations carried out on a ‘midden’ deposit. James Williams subsequently reported on these investigations, which revealed sherds of Samian Ware and black-burnished ware as well as pieces of daub and other materials (Williams 1977b). Interest in Burnswark also continued during this decade, with work carried out by George Jobey on the so-called ‘western redoubt.’ This was a fortification located upon the western summit of the hill, within the western defences. The structure had been neglected by Roy but was ascribed to the Romans by James Barbour, who had partly excavated the site. Jobey, by contrast, argued that it had been created much more recently, potentially during the War of the Three Kingdoms (Jobey 1973, 80), when it was constructed by the Scots as they retreated from the Battle of Preston.

84. Another complex of Roman temporary camps was identified – again through aerial photography – at Girvan Mains in 1978 (St Joseph 1978). Two camps were found, and trial trenching was subsequently undertaken by Kenneth St Joseph, confirming the putative Roman attribution for the site. This discovery confidently placed the Roman military on the South Ayrshire coast for the first time. An attribution to the Flavian campaigns undertaken by Agricola was proposed for these sites and supported by a find of 1st century AD Roman glass. A watching brief undertaken there four decades later by Louise Turner of Rathmell Archaeology in 2009 failed to reveal any features which could elucidate further understanding of the site (Turner 2009).

85. In 1983, traces of a Roman camp were identified on aerial photographs at Trailflat, near the Water of Ae. New aerial photograph discoveries continued when drought conditions in 1984 led to the identification of a Roman fort and annexe at Drumlanrig and the fortlet at Sanquhar (Maxwell & Wilson 1987). Re-flying the fortlet at Murder Loch also revealed that it stood within a small enclosure or camp (ibid. 24). Though this small camp has not been investigated, Rebecca Jones has highlighted this with the small camps at Shawhead, Murder Loch and Ellisland as unusual in scale (Jones 2011, 120)

86. A year later, in 1985, Kenneth St Joseph carried out trial trenching on the Roman camp at Innerfield, which he had first identified in the late 1960s. His work there was never published, but drawings held in the National Record of the Historic Environment confirm a V-shaped ditch entirely consistent with Roman construction.

87. A number of reappraisals and works of artefact analysis were also undertaken in this decade. 1980 saw a reappraisal of the burial from High Torrs, Luce Bay, which had revealed a cremation buried beneath a small cairn of large boulders, accompanied by an iron finger-ring set with an intaglio, Samian Ware bowls, and, reportedly, several spearheads and a fragmentary sword. An examination of the ironwork, using the latest scientific techniques, suggested that the sword and spearheads instead represented a fragment from a pan or skillet, and the corner binding from a wooden chest, plus assorted nails and hobnails. This reidentification weakened the possibility that the deceased individual had been attached to the Roman military (Breeze and Ritchie, 1980). The use of cremation as the preferred burial rite was, however, characteristic of a Roman, rather than a indigenous Iron Age, burial tradition, so it seemed more likely to the authors that the individual interred there had been an itinerant Roman traveler. It remained possible, however, that the deceased had been a Romanised member of the indigenous local community, although the unique character of this particular burial might argue against this possibility.

88. In 1983, Anne Robertson presented another instalment in the long-running series of Roman coin inventories (Robertson 1983), while 1985 saw the publication of several more general articles in the Proceedings. One saw David Breeze explore the relationships between occupying Roman military forces and the indigenous population (Breeze 1985). In another, Michael Jarrett summarised recent archaeological discoveries and new interpretations relating to Roman frontiers, while acknowledging the shortcomings of the texts and the ‘overall uncertainty which has arisen from the attempt to write history from archaeological evidence’ (Jarrett 1985, 66). From a historical point of view, he argued that the conclusions reached by the archaeologist did not altogether make sense, and that it could be argued that it was wrong to seek ‘archaeological answers to historical questions when the strictly historical material fails us’ (ibid.). Then, three years later, Breeze published another paper – again in the Proceedings – which carefully examined all the reasons given for the Romans’ failure to conquer Scotland in either the Antonine or the Severan period (Breeze 1988). As well as providing an important short summary of Roman military activities in Scotland, this publication also provided a useful and updated series of distribution maps showing known Roman military sites attributed variously to the Flavian, Antonine and Severan campaigns, including a Severan-era camp at Kirkpatrick in eastern Dumfries.

89. Also included in the Proceedings of 1988 was the latest instalment in the inventory of Roman coin finds. Anne Robertson had now stepped back from the task, with responsibility passing now to Donal Bateson of the Hunterian Museum. From this particular iteration of the inventory onwards, Roman coin finds would now be coupled with their medieval counterparts, providing Bateson with an ideal vehicle with which to display his expertise as a numismatist (Bateson 1989). Although a number of new finds had been recovered, both during excavations and as stray finds, the patterns previously discussed by Robertson at the start of the decade remained unchallenged.

90. The 1980s saw a switch in emphasis from research excavations to developer-led work. Archaeological investigation in advance of disturbance or construction had been practised, with the response often instigated by Alfred Truckell of Dumfries Museum. Now, though, there was an increasingly proactive approach to mitigating impacts on known archaeological sites. The scale of the investigation varied according to the anticipated impact, exemplified by an early small-scale example reported by Michael Yates in 1983. Archaeological work had involved a watching brief undertaken on the line of the Roman road running north from Hadrian’s Wall and Birrens to the Antonine Wall. A section of this road, near Moffat, was to be cut by the trench for a pipeline being laid by Scottish Gas. The work provided Yates with an opportunity to examine a section of the roadway and record it for publication in the Transactions (Yates 1983).

91. Metal detecting finds were also becoming an important contribution to an understanding of Roman sites in South West Scotland, with a short 1985 paper in the Transactions detailing the recent find of a bronze figurine of a dolphin near the fort site at Milton (Tassieholm), Annandale. The figurine was just one of over a hundred finds recovered from the vicinity of the site. Others included lead weights, a variety of Roman pottery including fine and coarse wares, and five coins of the Republican period (Green et al 1985). Trial trenching also took place at Beattock (Barnhill), where Gordon Maxwell was able to demonstrate that Camp I of the four-camp complex, overlay and hence post-dated the SW corner of a Flavian fortlet which had previously occupied the site (Maxwell 1984). This was the first of a series of investigations which would gradually disentangle the palimpsest of Roman sites which formed the Beattock (Barnhill) complex.

92. 1985 saw a much larger developer-led excavation instigated by the Historic Buildings and Monuments Directorate of the Scottish Development Department. The work was undertaken on the site of a Roman temporary camp at Annan Hill, threatened by the construction of a new housing development. Lawrence Keppie led the excavations, which confirmed the extent of the site and revealed traces of a simple turf-built rampart (Keppie 1988a). There was no evidence of a gate structure, and the external defensive ditch or titula, which lay beyond the gateway – visible on aerial photographs – could not be discerned. Finds were virtually absent, and the mass of internal features, which included postholes, could not be tied stratigraphically to the defensive features, suggesting other phases of occupation. Despite the ambiguity of the site, a Roman origin was agreed, with the paucity of robust structures and finds only confirming the temporary nature of the occupation there.

93. Roadbuilding then threatened part of the Roman camp complex at Carronbridge, leading to excavations in 1989 which explored several enclosures ascribed to both prehistoric and Roman activities. Here, the line of a Roman road had slighted the corner of an earlier, indigenous Iron Age rectilinear settlement, while the ditch of the Roman camp, which overlay and reused part of an earlier ditched enclosure (characterised in the 1950s as a Roman fortlet), revealed a primary fill composed of cut turves (Johnstone 1994). 1989 also saw the discovery of another previously undiscovered fortlet at Kirkland, near Thornhill. The site was later subject to partial excavation by Lawrence Keppie in the early 1990s, which confirmed a Flavian date for the site and revealed that it had been deliberately slighted on abandonment. A year later, the site of another Roman fort was identified at Ladyward, again from aerial photographs. Like the fortlet at Kirkland, the site at Ladyward was ascribed to the Flavian campaigns, with two phases of construction evident from the visible remains.

94. Meanwhile, the regular round-up of Roman coin finds – now presented jointly by Donal Bateson and Nicholas Holmes – revealed more finds from South West Scotland. Coins dating from the Antonine period were recovered near the camps at Durisdeer, while finds from the Republic, as well as others minted in the reigns of Tiberius and Hadrian – were reported from the fort at Milton (Tassieholm), providing further dating evidence for these sites’ Flavian origins. Another coin was recovered during excavations of an indigenous Iron Age site at Brighouse Bay (Bateson et al 1997) (see Chapter 7 Iron Age).

95. Work on pipelines during the 1990s resulted in excavations and a watching brief on the complex of temporary camps at Beattock (Barnhill), with archaeological expertise in this case provided by the Centre for Field Archaeology. The ensuing report – jointly authored by Tim Neighbour, Ian Armit, Bill Finlayson and Ian Ralston – focused on Camps II and III and detailed how the ditch showed a single phase of cutting followed by gradual filling (Neighbour et al 1994). A linear stone feature identified at the base of the ditch could not be explained but was comparable with a similar feature identified by Lawrence Keppie at Annan Hill (Keppie 1988). Keppie had suggested that the feature was localised, and the excavators of Beattock (Barnhill) argued for an analogous situation at their site. They suggested that the feature could potentially have been associated with modifications of the ditch evident on aerial photographs of the site. A few years later, the environmental scientist Richard Tipping (Tipping 1997) was able to demonstrate that the complex of temporary camps had been constructed after the fortlet, as an alluvial terrace had accumulated above the cut of the earlier site, prior to the construction of the later camps. This confirmed that the camps formed part of the later Antonine campaign, with the fortlet a relict feature from the earlier, Flavian, phase of occupation.

96. Excavations also took place at the Roman camp of Beattock (Bankend) during the 1990s, in response to the threat posed by the A74 road upgrade to motorway. This site lies on the other side of the Evan Water from the Beattock (Barnhill) complex and just north of the Milton (Tassieholm) fort. Alan Leslie for Archaeological Projects Glasgow led these investigative works, which explored the perimeter ditch in 1995, and later located up to three of the characteristic dumbbell-shaped field ovens in the interior. The fact that the ditch had been left open for several weeks after construction led Leslie to suggest that the camp was a construction camp, created to provide accommodation while work was undertaken on the fort site at Milton (Tassieholm). This suggestion was later disputed by Rebecca Jones (Jones 2011, 144), who argued that the distances between the two sites would have been too great for this to be practicable.

97. The upgrade of the A74 also led to the investigation by Alan Leslie and Archaeological Projects Glasgow of one of the three camps at Kirkpatrick-Fleming (Leslie 1991). Initial geophysical survey covered the NE quadrant of the camp with selected areas then subject to excavation. These works revealed a gate into the camp and sectioned the bounding ditch, although no finds were recovered. To the north of the SW angle a shallower enclosure ditch containing sherds of Roman grey coarse ware pottery was located which is thought to be from the first half of the 2nd century AD (Jones 2011, 251). The ditch was interpreted as an annexe to the camp, although it was recognised that this was not the only option. A subsequent development-led watching brief in the NE corner of the same camp (Brann 1993; Keppie 1996) exposed the ditch in the area disturbed by the earlier road works in 1968.

98. On a smaller scale, monitoring of a pipeline at Brighouse Bay revealed a coin mould for a fake Roman denarius, recovered from a shell midden (Boon 1994). The mould combined an obverse and a reverse face derived from two different Roman coins, which would have created a mismatched product, using coins sourced relatively close in time at around 220 AD (Holmes and Hunter 2001). Holmes and Hunter agreed that it was possible that the counterfeiter had chosen a remote location for the production of fake coins, which would then be circulated south of the Tyne-Solway frontier. However, they added that it could not be ruled out that these counterfeit coin moulds (of which there are three to be found across Scotland) were producing coinage for use in local communities which understood coinage but could no longer source currency from a Roman source, following the withdrawal of the Roman military presence from the region and the wider area.

99. Aerial photography, now mainly restricted to flights by the RCAHMS, continued to extend the range of known Roman sites in the South West. First recorded in 1992, the Roman camp at Glenluce (Jones 2011, 217) became the westernmost known Roman military site on the Solway. The aerial imagery here also showed a linear tun of quarry pits – presumably flanking a Roman road – passing the camp running towards the Water of Luce. This has been taken of additional evidence for the coast road running on to the Loch Ryan area. In 1994 further quarry pits were located to the east of Soulseat Loch further confirming this route west of Glenluce.

100. The 1990s came to a close with a useful regional summary overview of Roman activity in Eastern Dumfriesshire authored by Allan Wilson and published in the Transactions. This examined the work of antiquarian authors as well as more recent contributions to the subject brought about by recent fieldwork, survey, and excavation. Wilson’s aim was to review and extend existing knowledge relating to the Roman road system, and his short paper referred to previously unknown sites as well as a summary of recent excavations. A comprehensive distribution map showing sites (confirmed and postulated) was also included (Wilson 1998).

101. This work can be compared with the non-inventory Eastern Dumfriesshire: an archaeological landscape published by the RCAHMS in 1997 that provides and other overview of the Roman field monuments in this portion of South West Scotland. The work of the RCAHMS in collating their publication involved extensive archive consultation and field survey that not only provided improved descriptions of known Roman sites, but also recognised previously unidentified candidates. One example of this is the watch tower at Ewes Doors that was described as a tumulus in the 1960s but was reclassified as Roman through the work of this study.

102. During the first decades of the 21st century, regular roundups of Roman coin finds continued in the pages of the Proceedings, either presented by Donal Bateson, or jointly by Bateson and Nicholas Holmes (Bateson 2003; Bateson and Holmes 2006; Bateson and Holmes 2013). Small numbers of coin finds continued to be found close to the known sites of Roman camps, although finds from indigenous sites and also stray finds in areas with no apparent evidence for settlement or occupation were also recovered. Work on re-assessing earlier material finds was also underway with a petrological and geological analysis of the inscribed stones from Birrens, reported upon in 2002 by Fraser Hunter and Ian Scott (Hunter and Scott 2002). This level of detailed analysis enabled the researchers to conclude that most of the decorative stonework derived from a locally derived sandstone, potentially from the same block, which allowed a revised interpretation of the original – now fragmentary – form. By contrast, the head of a god with an animal-skin cap was composed of oolitic limestone, which must have been imported onto the site. Sources for this rock included southern England and northern France, with South West England being a strong candidate, as established trade routes with the area could be demonstrated through the movement of black-burnished wares.

103. Allan Wilson, who in the late 1990s had authored a useful and up-to-date overview of Roman activity in Dumfries and Galloway, compiled a similar work in the early 2000s. Once again, he created a detailed synthesis which alludes to the latest sites and discoveries found across this part of the region (Wilson 2003) and provides insights into the then-current schools of thought which predominated at the regional level. The important Roman site at Burnswark continued to be a focus for new research. making full use of the improved understanding in material culture types (in particular pottery) which had been achieved in the four decades since the site was first explored.

104. In 2003, Duncan Campbell presented a detailed analysis and overview of the Roman siegeworks at Burnswark. He examined how changing perspectives and levels of understanding had created a rich and complex narrative which explained the various elements and the relationship between the indigenous earthworks and the Roman fortifications (Campbell 2003). Campbell firmly refuted the concept of Burnswark as practice range for Roman artillery, returning once again to the argument that the site bore evidence for actual siege warfare. Campbell’s new research on this site proved to be the first in a series of academic and field-based projects to focus on this important site, which is being actively explored at the time of writing.

105. New research also took place at the Antonine-period Roman fort and annexe site of Drumlanrig, this time under the auspices of popular culture and television archaeology. Carried out in 2004 by Wessex Archaeology in association with the Time Team, work on the site included geophysical survey (Walker et al 2005) and limited trial trenching (Wessex Archaeology 2005). The works trenched the broad rampart and the V-shaped outer ditch as well as investigating anomalies from the geophysical survey. The finds recovered were varied and interesting. As well as a copper alloy terret, a possible fragment of iron lorica plate (derived from armour) was recovered, along with Gaulish Samian ware, black-burnished ware from Dorset and a fragment of Spanish amphora. Multiple phases of occupation were identified, and early occupation during the Flavian campaigns was inferred, though not confirmed, with most of the material recovered consistent with occupation during the Antonine phase.

106. Earthworks of uncertain attribution at Lochrutton (Moat of) had long been known (Coles 1893). Aerial photography by the RCAHMS in 2004 recording these earthworks led Dave Cowley to also identify the possible fragmentary remains of a Roman fortlet (Hunter 2005, 402). In 2010 the reexamination of Cambridge University aerial photography held by RCAHMS by Dave Cowley identified at least one Roman camp at Lochrutton, to the south of the putative fortlet. Subsequent re-flying in the same year supported this interpretation while also locating two additional enclosures that may be small Roman camps (Jones 2011, 322). In a similar manner, reexamination of aerial photography held by Dumfries Museum by Andrew Nicholson in 2006, identified a possible Roman temporary camp to the southwest of Twynholm at Camp Hill. None of these sites were ground truthed and so their attribution as Roman is provisional and based on their observed form. However, the increase if probable targets along the Solway coast continued.

107. Given the dominance of Roman temporary camps in the recognition of the multiple Roman campaigns within South West Scotland, the gazetteer of these monuments compiled by Rebecca Jones is a critical analytical and reference document (Jones 2011). This work derived from her doctorate research, which concluded in 2006.

108. Developer-led work continued to make an important contribution. 2002 saw small-scale excavations carried out by Martin Brann on the east ditch and gateway of the Annan Hill camp. In November 2004, an evaluation in advance of housebuilding on the site of Annanfoot Roman camp was undertaken by Thomas Rees and Douglas Gordon of Rathmell Archaeology. Earlier work on the site – carried out by Professor Barri Jones in the late 1970s – remained unpublished but had allegedly uncovered a so-called ‘Punic’ profile to the ditches. It was hoped that these later investigations would either prove or disprove the wider presence of such features. The camp had traditionally been attributed to coastal or maritime campaigns carried out in the time of Agricola, but no dateable material was recovered. The ditches appeared to be of a regular ‘v’-shaped profile consistent with most Roman defensive works in Scotland, and no traces of internal features had survived (Rees & Gordon 2007).

109. More work was also undertaken at Carzield fort, in response to planned works by Scottish Power Energy Networks and Scottish Water during October and November 2011. GUARD Archaeology undertook the archaeological works, which were undertaken over the extent of three service trenches and written up by Warren Bailie (Bailie 2013). The investigations revealed structural features associated with the fort’s internal layout, including cobbled surfaces and ditch and gully features. A range of artefacts were also recovered, including pottery (Samian Ware and black-burnished ware), metalwork (an iron javelin-head and hobnails from a Roman shoe), structural items (a fragmentary roofing tile and a fragmentary tile from an underfloor heating system or hypocaust and fragments of daub-and-wattle). A deposit of burnt hearth material also produced some charred cereal grains, with six-row barley and occasional wheat grains present.

110. In addition to the developer-led works carried out across the region, targeted research has been resumed and over recent decades, has become a regular occurrence. Work has often focused on the hillfort site of Burnswark and its attendant Roman fortifications, which still remains a source of debate amongst researchers. A new project, led by John Reid of the Trimontium Trust and Andrew Nicholson of Dumfries and Galloway Council, commenced in 2013 with a metal detector survey, which allowed the distribution of lead glandes to be mapped across the entire site (Reid 2017). 2015 then saw exploratory excavations over the areas where the lead signals had been strongest, with two trenches opened. 17 out of 18 signals proved to be Roman glandes, and the examples recovered comprised two previously identified types (‘lemon’ and ‘acorn’ shaped) as well as a previously undiscovered ‘whistling’ type, which features a drilled hole which would have made a shrieking sound whilst airborne.

111. Lead isotope analysis has revealed uniformity of the lead composition which supports the theory that the glandes relate to a single chronological event. This could be linked with a siege operation which ended in the storming of the fort at the commencement of the re-conquest of Scotland during the mid-2nd century AD, in the reign of Antoninus Pius (Reid and Nicholson 2020). Taking a more theoretical approach was a paper which applied analytical methods used by the US military (KOCOA terrain analysis). Authored jointly by Craig Brown, Manuel Fernández-Götz, Rachel Cartwright, John Reid and Andrew Nicholson, it reassessed the relationship between the hillfort and the Roman siege works. They concluded that the remains represented evidence for a longinquo obsidio, i.e. an active siege that ended with the storming of the hillfort (Brown et al 2025).

112. Further excavations in 2016 revealed more finds of glandes, and it was proposed that – dating by association with finds of lead glandes with pottery at nearby Birrens – the lead bullets were dated to the Late Hadrianic-Early Antonine period. This, the excavators argued, suggests that the erection of the siegeworks at Burnswark and the assault on the hillfort formed one of the earliest episodes in the invasion of Scotland undertaken by Antoninus Pius.

113. The Solway Hinterland Archaeological Remote Sensing Project in 2009 to 2010, led by Richard Jones, undertook geophysical survey of the environs of Roman camps in eastern Dumfriesshire. This project included works at Dalswinton (Bankhead) (Jones, O’Grady & Malcolm 2010). These surveys encompassed the whole of the interior of the forts at Dalswinton (Bankhead). Following this success, the project Discovering Dumfries and Galloway’s Past was a community archaeology project focused on deploying geophysical survey that ran between 2012 and 2013. Led by Richard Jones and Valentina Bold they undertook geophysical survey at Birrens covering the western annexe and to the north of the fort seeking the Roman road (Carey 2013).

114. The integration of multiple remote survey technologies (aerial photography, geophysical survey and LiDAR) has enabled re-assessment of a number of Roman sites, including those in South West Scotland. One such study was co-authored by Bill Hanson, Rebecca Jones and Richard Jones, and published in 2019 (Hanson et al 2019). It focused on the complex of Roman features around Dalswinton at the strategic point in the Nith Valley where the valley narrows and access to the north becomes more difficult. The Roman sites identified there include a forts with annexes at Dalswinton (Bankhead) and the overlapping Stracathro-type temporary camps and fort at Dalswinton (Bankfoot). Dalswinton, they argued, had played a key role in the Flavian subjugation of South West Scotland. Its importance was emphasised by the fact that it was one of the largest Roman military sites at this time, outmatched only by Newstead and Inchtuthil, and one of the last places to be abandoned when the Romans regrouped south of the Tyne-Solway frontier in the reign of Trajan (AD 103-5). It would then be superseded during the Antonine campaigns by Carzield, located 4.7km to the southeast.

115. Combining their findings with observations obtained through similar work undertaken elsewhere across Britain and beyond, the authors were able to propose a nuanced narrative. As well as discussing the likely role played by both forts and camps in the Flavian and Antonine campaign, they discussed observations relating to morphology and character, shedding light – in the process – on contemporary thought and theory relating to Roman military operations around southern Scotland and Wales. They argued that the so-called vexillation fortress at Dalswinton (Bankhead) should instead be viewed as being of a more temporary nature on account of its lack of permanent internal structures, and questioned, too, the Roman origin postulated for three enclosures at Butterhole Brae. Instead, they argued for a potential later prehistoric origin, on the basis that rectangular prehistoric settlements are now being recognised more widely. Strong magnetic signals in localised areas suggested that the ‘fortress’ at Dalswinton (Bankhead) had been deliberately slighted at the end of its life and the demolition material generated during this event was buried on site (ibid.).

116. In response to the reappraisal of Dalswinton, Nick Hodgson proposed an alternate interpretation of the fort at Dalswinton (Bankhead). Part of this alternate proposal was the suggestion that the larger, second-phase fort faced not east but south and that most of the remains of this layout have been removed by the action of the plough (Hodgson 2021).

117. Aerial photography continued to add to the corpus of known, but uninvestigated, Roman military sites with a camp and fortlet located at Bladnoch in 2018. This discovery filled a substantial gap between the known sites at Gatehouse-of-Fleet and Glenluce. However, set to the south of the River Bladnoch and near Wigtown it has given rise to speculation that the coastal Roman road may have utilised intertidal crossings of Wigtown Sands when travelling west from Gatehouse-of-Fleet rather than following the 18^th century Military Road that followed a dry land route, and passed by Newton Stewart to the north.

118. Although much of the recent work undertaken on Roman sites involves reassessment or new, often developer-led, investigations of known forts and camps, new discoveries are still being made. As recently as 2015, excavations by Iraia Arabaolaza of GUARD Archaeology Ltd on the Ayr Academy site revealed the characteristic ‘dumbbell’ shaped field ovens that were indicative of a Roman temporary camp. This discovery placed Roman activity in an area where evidence had previously been lacking, confirming the potential for Roman sites to survive undiscovered in the sizeable balance of ground running inland from Irvine to Kilmarnock and east to Loudon, and from Largs in the north to Ayr in the south. Organic material was radiocarbon dated, but the date range was broad, spanning 2 BC to AD 231. There was some overlapping of dates in the Flavian period, however, which suggests that the site may have been constructed during this earliest phase of Roman activity in Scotland.

119. More recently, the potential for interaction between indigenous Iron Age populations and Roman military has been subject to more detailed exploration. A long-running project called Between the Walls was followed up by a project named Edge of Empire, which aimed to use remote sensing and geophysical survey to inform on the character of indigenous Iron Age occupation in the environs of Burnswark and Birrens Fernández-Götz, et al 2022). In addition, ongoing field survey work, carried out by staff working for Historic Environment Scotland (formerly the RCAHMS) and others continues to record Roman sites, including the survey watch tower at Craik Cross that had originally been investigated by St Joseph in 1946 (St Joseph 1947).

120. Development-led work continues to investigate Roman period sites, especially where physical land constraints restrict the ability to avoid these monuments. The temporary camps and fortlet at Beattock (Barnhill) for instance have recently benefited from geophysical survey in 2022 and 2025 (Lawton et al 2022) led by James Lawton of AOC Archaeology Group.

---