Roman military sites


At Templeborough, the three-phase model of fort construction and occupation originally proposed by Leader (1878) and May (1922) is probably incorrect. The postulated three successive and superimposed phases of occupation between the 1st and 4th centuries AD consisted of an early fort with earth and turf-built ramparts, and stone-faced earthen ramparts in the second and third phases, these being separated by episodes of widespread burning (May 1922: 14-18). The third and smallest phase fort was nebulous, however. Buckland (1986: 30-2) tried to put the somewhat vague ideas of Leader and May into a more secure chronological framework, with a first phase fort occupied from AD 54 until the mid-2nd century when it was abandoned for a time. Later in the 2nd century it was re-built as a slightly smaller base, but the third phase and final abandonment was again somewhat unclear and described by Buckland with understatement as ‘enigmatic’ (ibid.: 32).

The more recent investigations identified two key phases of activity indicated by the pottery recovered from the truncated ditches and the fragmentary vicus features. This scheme has a first phase Flavian–Trajanic fort during the late 1st to early 2nd centuries, and a second Hadrianic–Antonine phase rebuild in the mid- to late 2nd centuries (Davies 2016: 57-8). There is Neronian-period samian pottery from the older investigations which might suggest a construction date of c. AD 55–65 for the first fort, but this could have already been old at the time of deposition and no material of this date was found in more recent excavations. A third and later phase may never have existed. This later history of Templeborough fort, and the nature and full extent of the vicus, remain extremely unclear.


The basic layout of much of the fort was established through the rescue excavations of the 1960s–70s and developer-funded investigations since the 1990s, but much of the internal area and details of the phasing are unknown. The original fort may date to AD 70–1 and the Flavian period, with a possible rebuilding episode in the mid to late 80s AD and again in the early 2nd century (Buckland 1986: 12; Buckland and Magilton n.d.). The defences may have been a single ditch in front of a turf-faced earthen rampart, perhaps with stone added later. This accords with evidence from the Church Walk excavations (Chadwick, Martin and Richardson 2008). It might have been mostly auxiliaries garrisoning these earlier phases of Roman forts at Danum.

Buckland posited a period of abandonment of nearly 30 years until the mid-2nd century, when a smaller fort was constructed within the north-western corner of the earlier base. This may have been occupied until the late 3rd or early 4th century, when a new fort or defended settlement was built with stone ramparts up to 6 metres high and 2.5m thick, and some major stone buildings within its walls (Buckland 1986: 17). The rescue excavations found evidence for a more irregular layout and buildings during the later 4th century, the latter of stone footings with timber and turf or daub walls, and in one case built across an earlier metalled road surface. Considerable quantities of pottery and animal bone in ‘dark earth’ deposits were also recorded. The end years of the fort are not known, though a coin hoard of silver siliquae dated to AD 388 suggests occupation continued into the late 4th or early 5th century.

No detailed plans of the fort excavations during the 1960s–1970s are published, and nor any details of the ceramics and animal bone. The pottery will be particularly crucial for constructing a detailed sequence. Only a draft ‘small finds’ report exists (Lloyd Morgan and Buckland n.d.), and no funding was made available by Doncaster Council or Historic England and its forebears to write up the rescue excavations. A post-excavation report on the Roman fort that was in preparation (Buckland and Magilton n.d.) was unfortunately largely halted due to the death of one of the co-authors John Magilton. There was a council proposal to write up and publish all the different stages of investigations including some of the more recent developer-funded sites, as part of a wider scheme to promote and display the Roman and medieval remains and heritage, but apart from a synthetic study (Pollington 2007) this appears to have stalled with few tangible results. The Doncaster Roman fort investigations need to be published as soon as possible.

Rossington Bridge

The military base at Rossington Bridge was first identified from the air by Keith St Joseph (St Joseph 1969: 104) and was classified as a new form of military site called a ‘vexillation fortress’ (q.v. Frere and St Joseph 1974: 6) which may have held both legionaries and auxiliary troops. It was subsequently photographed and recorded by Riley (1980: 57, 93, map 8). Limited geophysical survey and trenching took place over its north-eastern corner to assess preservation (Head et al. 1997: 275-8), but much of the interior has never been surveyed. Some aerial photographs hinted at possible external outworks to the east (Wilson 1984: 58), and/or at several different phases of fort or fortress – along the southern edge of the fortress an inner ditch appears to be slightly out of alignment with the outer defences, whilst some internal linear features might relate to field systems pre-dating the fortress, or an earlier phase of military construction.

In advance of the construction of a park and ride scheme there was further geophysical survey confined to the area between the B6463 and the line of the A638, including the north-eastern corner of the fortress area (Schofield 2003). Subsequent evaluation and excavation revealed a possible earlier enclosure with a roundhouse and evidence for Iron Age pottery metalworking (Bishop 2010: 10-11, fig. 12; see above), and field system ditches and a trackway on broadly north-south and east-west alignments, slightly different to that of the fortress and with 2nd to 3rd century pottery. Approximately 150m east of the fortress, one evaluation trench recorded a ditch corresponding to a geophysical feature with a rounded corner (possibly a fort ‘playing card’ corner), and this ditch yielded late Iron Age and Roman pottery (ibid.: 12, figs 9-10). This may be associated with an early military feature.

Although an evaluation report and a draft publication report were produced as a result of the work in the early 2000s (Bishop 2005, 2010), an actual publication has never materialised. Further detailed geophysical investigation of the entire Scheduled area of the fortress but especially the interior in the form of magnetometry, soil resistance survey, Ground Penetrating Radar and pXRF survey is long overdue, and could be followed up with targeted excavation. A full supervised, methodical archaeological metal detecting survey of the site to plot and remove any metal finds in the topsoil for safe conservation and analysis would also be useful, as the site has been subject to much illegal detecting over the decades.


Hunter (1831: 487) recorded finds of Roman coins and fibulae from the site at Robin Hood’s Well, Burghwallis, but the fact it was a military station was first recognised from aerial photographs (Buckland 1986: 8-9, fig. 7; Deegan 2007: fig. 6.22; Roberts, Deegan and Berg 2010: 37, 67, 72, figs 48, 91; Wilson 1972: 311). The site has been partly truncated by an old (now backfilled) quarry visible on the 1st Edition Ordnance Survey map; and is also partly overlain by a farm and probably by the modern line of the A1. The site consists of at least two, probably three phases of fort – two with double ditches and of slightly different sizes, and a third smaller fort with only one visible ditch and on a slightly different alignment to the other two. No vicus is visible, and on some aerial photographs the forts are clearly situated at an oblique angle across a series of earlier fields (contra Roberts, Deegan and Berg 2010: 72).

The forts may have been positioned to guard the crossing of a low-lying valley and The Skell watercourse by the Great North Road or Roman Ridge, now the modern A1 (Margary 1973: 415, RR28b). There has been no recorded geophysical survey of the forts, so their exact layout and relationship to one another is unknown; and no excavation so nothing is known of their date. Roman pottery including samian of Antonine date was found in one of two ditches visible in section in a disused quarry at the site (Wilson 1972: 311), and two 3rd century bronze radiate coins of Tetricus I and Carausius along with some pottery were picked up at the fort in 1973 (Pastscape 56125). The site is Scheduled but the field the forts are within still appears to be under cultivation and thus threatened with continued degradation by ploughing. The field must be taken out of cultivation using agri-environmental schemes.

The site is likely to have been subject to much illegal metal detecting. A full methodical archaeological metal detecting survey of the site to plot and remove any metal finds in the topsoil for safe conservation and analysis would also be useful. It is vital that detailed geophysical survey takes place over the site as soon as possible using all three major techniques (magnetometry, soil resistivity and Ground Penetrating Radar), along with pXRF. Targeted excavation should take place to establish the stratigraphic sequence of the different fort features and to recover dateable artefacts or materials and palaeo-environmental samples.

Long Sandal

It had been previously proposed that there might have been a Roman fort at this location (e.g. Buckland and Magilton 1986: 208) but it was confirmed through aerial photographic analysis (Deegan 2007: fig. 6.22; Roberts, Deegan and Berg 2010: 36, 67-8, fig. 48). The double, possibly triple-ditched camp was on the southern bank of the River Don north-east of Doncaster, but might originally only have had three sides, the fourth formed by the river. It may have been built to protect a riverine supply route, either for the initial invasion of the north, or during the later Roman occupation – the former might be more likely. Part of the site now lies beneath industrial developments. Detailed survey by Ground Penetrating Radar along with targeted excavation to retrieve dating information should take place as soon as possible.

Thorpe Audlin

This possible small fortlet, series of forts or military station lies just within South Yorkshire on the southern side of the River Went near where it was crossed by the Roman Ridge Roman road (Margary 1973: RR28b). Readily identifiable on some Google Earth imagery, some possible internal features are also visible, and the cropmarks suggest that there were several phases of activity at this site, with at least two ditches on slightly different alignments to one another (Deegan 2007: 25, fig. 6.22). There may have been a vicus or a roadside settlement nearby, as excavations by the Pontefract and District Archaeological Society in 1982 some 140m east of the fort and east of the modern A639 found remains of a wall, a possible stone workshop surface and an iron-ore roasting hearth furnace, along with a lava quern and late 1st to late 2nd century pottery including samian (Houlder and Hedges 1982). Large numbers of late Roman coins, copper-alloy objects and Romano-British pottery were found in that field. West of the A639, pottery, ceramic roof and box-flue tiles and quern fragments were picked up in the field with the possible fortlet.

Further work by the Society in 1987 took the form of a north-south orientated trench across one of the ditches – probably the ‘outer’ ditch, but no plan is provided in the brief unpublished report, just bizarre computerised sections (Houlder and Hedges 1987). The ditch was 10 metres wide and up to 1.5m deep; and contained 25 sherds of pottery including Dales ware and Crambeck ware, ceramic roof tiles and box-flue tiles, and fragments of another lava quern (ibid.: 2, 5). Some late 2nd to 3rd century grey wares and one samian sherd were also recovered, these possibly residual. The ditch cut a spread or surface of cobbles and pebbles which was in turn set above an earlier, smaller ditch on a slightly different NE-SW alignment and with stakes set within its fill. The limited, poorly documented evidence is rather ambiguous but suggests a later Roman date for the large ditch, and the presence of Roman-style buildings nearby. The ditch seems too wide for a civilian farmstead, and the stakes set into the earlier ditch are suggestive, though no military finds were recovered. If not an actual fort, similar perhaps to Scaftworth, then it might have been some sort of military or official way station controlling and perhaps taxing traffic over the bridge.

The fortlet or military station and the possible vicus at Thorpe Audlin are under the plough and threatened by metal detecting. They must be taken out of cultivation and investigated as soon as possible with different geophysical survey techniques, targeted excavation, and an archaeologically controlled metal detecting survey. The archive records and finds from the 1982 and 1987 investigations by the Pontefract and District Archaeological Society must be located, re-analysed and published.

Bole Hill

A square double-ditched enclosure with large internal postholes on Bole Hill near Sheffield investigated through geophysical survey and trial trenching was probably of late 1st to early 2nd century AD date; and might be a timber signal tower overlooking a proposed Roman road between the fort at Brough-on-Noe in Derbyshire (Navio) and Templeborough (Inglis 2016: 60; Waddington 2017: 46-9).