Villas and rural stone buildings
Roman-style villa complexes seem to have been rare in South Yorkshire, and across northern England were probably a late development during the 3rd and 4th centuries (Branigan 1980, 1984; Hingley 1989; Smith 2016; Wilson 1997). This lack of villas has biased discussions regarding ‘Romanisation’ and the region’s perceived marginality, yet this is only problematic for researchers familiar with southern England whose thinking is dominated by simplistic culture-history and core: periphery approaches. It is notable, however, that those settlements that have provided archaeological evidence for stone buildings or artefactual evidence for more ‘Romanised’ settlement were all located close to Doncaster and/or Roman roads.
At the likely villa site at Stancil north of Tickhill during the late 1930s, there was serious damage from a steam shovel digging a pipe trench, and salvage excavations were not stratigraphic and were poorly recorded (Whiting 1943). Several different phases of construction and occupation seem to have been identified, but the brief report is confusing and does not contain enough detailed illustrations or photographs, as was sadly typical of excavations of the time. A rectangular stone-walled building (Building I) on a north-west to south-east axis had a part-hypocaust floor with ceramic brick pilae stacks. On the opus signinum floor was a debris layer that included ceramic roof and flue tiles, decorated glazed floor tiles, and painted wall plaster that included floral motifs (Whiting 1943: 264). A possible north-west extension had a stone-flagged floor, and to the south-east there was an apsidal-ended room (Building II) with a red op. sig. floor that was probably a bathhouse. Traces of a third masonry structure were found south of the main complex.
Relatively small quantities of ceramics including mortaria and Derby ware cooking or storage vessels were recorded at Stancil, along with some platter rims (Whiting 1943: 267); but no samian, glass vessels, or coins. Many human remains were also identified, however, and these seem to have been mainly adult males, with some females and infants. These probably relate to a later re-use of the building as a church or cemetery, either in the late Roman or post-Roman periods. Unfortunately, their whereabouts are unknown and thus they have never been scientifically dated. There may have been geophysical survey and limited test pitting or trial trenching undertaken there by the University of Sheffield in the late 1990s or early 2000s, but no other details of such investigations are known.
Two possible villas or at least ‘higher-status’ Romano-British sites might have been located at Oldcoates and Conisbrough Parks (Buckland 1986: 38). The former was putatively associated with a patterned tessellated floor. The latter (south of Conisbrough itself) has had a geophysical survey undertaken, and apparently consisted of a bathhouse and timber framed buildings on stone footings (ibid.). It was excavated by the private landowner (A. Lines pers. comm.), and sadly no details concerning the site are available. Artefacts also hint at the presence of high-status sites at Loversall and Brodsworth (Cumberpatch 2004a; P. Robinson pers. comm.). At Chapel Hole, Braithwell, a possible Roman stone building was examined in the 1950s (Buckland 1986: 38) but again seems to have been poorly recorded.
Remains of a rectangular stone-footed building were excavated at Whirlow Hall Farm, on the south-west outskirts of Sheffield. The footings consisted of roughly coursed sandstone blocks without any bonding, presumably to support a timber superstructure (Waddington 2017: 37). On a broadly east-west axis, it was associated with an external metalled surface and close to an inner revetment wall for the perimeter ditch, but some internal stone paving survived within the structure too – it may have had an ancillary function. A stone block that might have been a support for wooden or stone columns in this or another structure was also recovered from a fill of the nearby ditch. The building had been quite heavily truncated by ploughing and much of the interior of the settlement lay outside the area of excavation, but relatively large quantities of pottery were recovered including grey wares, Derbyshire ware, gritty oxidised ware, Black Burnished ware, worn and abraded samian sherds, and shell-tempered wares of late 1st to 3rd or 4th century date. Lead waste and fragments of a blown glass vessel and a glass bead also perhaps indicate a more ‘Romanised’ identity for the inhabitants (ibid.: 121-2). The settlement enclosure was located on a spur of land on the side of a relatively gentle south-east facing slope but overlooking Limb Brook at the base of a steep slope to the west.
At Hazel Lane Quarry, Hampole, magnetometry survey identified a series of subrectangular ditched enclosures, but monitoring of soil stripping ahead of quarrying by Thames Valley Archaeological Services revealed limestone footings of an L-shaped building with an apsidal end featuring a hypocaust and a floor supported by ceramic pilae (Bevan 2006b: 26; Pine 2002: 3; Pine and Taylor 2006: 72). This was probably the bathhouse of an unknown villa or higher-status settlement. It is likely that the building was missed through geophysics as magnetometry often cannot detect buried masonry walls, unlike soil resistance survey, though the latter is a much more time-consuming method. Painted plaster and stone and ceramic roof tiles were recovered, along with a large quantity of pottery, mostly of 3rd century date. The decision was made to preserve the building in situ, so it was not fully excavated. To date, only a series of unpublished interim reports have been produced by TVAS rather than journal article or even a unified final assessment report. One of these has the only (small-scale) plan of the building (Pine 2002: fig. 4), and few details of the artefacts. It is also unknown where the main building or buildings associated with the bathhouse could have been located. Additional geophysics results suggested possible structures to the west (Pine and Taylor 2006: 72), but only a ditch was found. Google Earth imagery from June 2018 reveals the cropmark of a small rectangular ditched enclosure approximately 120m east of the quarry, but it seems likely that any associated dwelling would be much closer to the bathhouse.
The bathhouse at Hazel Lane Quarry was not situated within a double-ditched square or rectangular enclosure characteristic of many villas, and along with Stancil was not associated with any notable cropmark boundaries and/or field systems (e.g. Riley 1980: 92-94, maps 7, 8), aside from a funnel-ended trackway near Stancil probably associated with floodplain grazing. This could indicate that in the study region villas were established within existing patterns of land tenure. Even large villa complexes in south-central England have not had outer estate boundaries identified, however (Dark and Dark 1997: 73-74; Scott 2004: 54). Although villas were associated with some form of tenurial control, land-ownership and centralised control of production, and perhaps landowner and tenant (and slave owning) relationships; it is possible that patterns of land tenure existed which mean that such clear-cut boundaries may never be identified (Dark and Dark 1997: 74; Millett 1990: 203).
Detailed consideration of the landscape setting of these sites is also revealing. The Stancil site was situated on the north-eastern end of a low gravel ‘island’ or ridge between 5-10m OD in the otherwise extremely low-lying floodplain of the River Torne. Perhaps these were native ‘cattle barons’ who had supplied meat and hides to the garrison at Rossington Bridge, only 3.5km to the north-east. Alternatively, the complex was established by a serving or retired Roman officer or legionary. The possible north-south aligned Roman road identified by Deegan (Roberts, Deegan and Berg 2007: 17-18, fig. 8.4, 2010: 68, fig. 90) was only 1.75km east of Stancil, and this may have been another important reason behind its location. At Hazel Lane Quarry, the bathhouse building was sited on a slight plateau with the ground sloping away gently to the east and west, and more steeply to the south where it could potentially have overlooked the course of the Hampole Dike stream and the site of several springs. The Roman fort at Burghwallis was only 2.2km to the east, adjacent to the Roman road from Doncaster to Castleford. Iron Age and Romano-British farmsteads at South Elmsall and Redhouse Farm, Adwick-le-Street were close by.
Villas have long been interpreted as economic units in a Romanised, almost proto-capitalist economy (Branigan and Miles 1987; Rivet 1969), and as expressions of status and wealth. Millett (1990) saw villas as the products of success by Romanised native elites. Roman-style houses should not always be directly equated with wealth, however (Taylor 2001: 49). Other households might have chosen to invest and display wealth and status in livestock and arable land, in portable material culture, or through feasts (Hingley 1989: 159). Elaborate reception rooms of ostensibly rich owners might have sometimes masked financial problems (Samson 1990: 175), whilst nouveau-riche people may have had more rich furnishings than established families. Some villas may have had multiple occupancies with different resident households and families within them (Creighton 1992).
Rectangular timber rural buildings
Roman-period timber-framed buildings are normally identified through postholes, post-bases and/or the narrow slots for timber sill beams or wattle and daub walls (e.g. Goodburn 1991, 1995; Smith 2016). These features may be relatively shallow and are thus particularly vulnerable to the truncation evidenced on many sites in the region as a result of more intensive ploughing since the Second World War. In some instances, horizontal timber sill beams might have sat directly on the original ground surface or on raised lines of clay, the weight of the completed building keeping it in place. This has been documented elsewhere in Roman Britain; and was also a recognised medieval timber building technique. Apart from traces of internal hearths or larger postholes or postpads for roofs, such buildings would leave very few archaeological traces. At Holme Hall Quarry there were remains of an oven and other features within the enclosure, along with ample evidence in the form of a midden deposit for sustained ‘domestic’ occupation, yet no actual structures could be identified (O’Neill and Raybould 2007: 102).
Within Enclosure A at Billingley Drive, Thurnscoe during Phase 1, a series of postholes marked what was regarded as an internal fenceline (Neal and Fraser 2004: 12-15, figs. 6-7), but a concentration in the north-west corner may instead have been remains of a possible rectangular structure. At St Wilfred’s Road, Cantley in Trench 3, shallow gullies included an L-shaped feature at least 14m long on its east-west axis cutting across the north-south ditch. The fill of another gully contained a fragment of Mayen lava quern (Daley 2007: 14). These gullies were either beam slots of timber buildings, or drainage features associated with such structures; of later 2nd or early 3rd century date. In Area 1 at High Street, Shafton (Burgess 2001a), at the central part of the enclosure a group of postholes may have been another poorly defined subrectangular structure. The rectilinear Enclosure 7 at Rossington Grange Farm may have enclosed a rectangular building (Richardson and Weston 2016: 32). At Warning Tongue Lane, Bessacarr, within the D-shaped enclosure a series of shallow rectilinear slots were probably the beam slots of rectangular timber buildings (Atkinson and Merrony 1994: 26). To date at least, the evidence suggests Roman-style rectangular timber buildings were scarce in the region.
Urban and military buildings
Though Hunter (1831: 2) had described the surviving earthworks of the fort at Templeborough, it was not really investigated further until 1877–8 when discoveries of Roman artefacts on the surface of the by now ploughed site prompted excavations led by J.D. Leader and J. Guest on behalf of the Rotherham Literary and Scientific Society (Leader 1878). Though little detail was provided in Leader’s published report by modern standards, it is clear that substantial stone buildings were encountered in the southern part of the fort with more fragmented remains in part of the postulated vicus to the south-east outside.
In the First Word War, the expansion of the nearby steelworks led to salvage excavations directed by Thomas May during 1916–17. Despite difficult circumstances, this work revealed traces of heavily robbed stone buildings from different phases, including the headquarters building, which May called the praetorium, though he may be confusing this with what is normally termed the principia; and a ‘commandant’s house’, which elsewhere is normally called the praetorium. Several different phases of bathhouse, granaries and sections of some of the walls and gateways were also investigated (May 1922, plates iii-xi, xlvi-lii). Some stone walls supported sill beams for timber frameworks, but there were rows of stone column bases for the portico of the horrea (granaries). As was common practice at the time, work tended to follow the lines of walls whilst interiors and exteriors of buildings were dug out with little or no stratigraphic recording. Dorothy Greene later identified a wall from a possible large stone building several hundred metres south-west of the fort alongside the Roman road (Greene 1957b), whilst a proposed grid of roads and remnants of buildings on Brinsworth Common was interpreted as a putative Roman town (Greene 1957a, 1957c). The fragmentary remains and the confused reporting of them do not really support this notion though.
During more recent re-development of the former Templeborough steelworks site, developer-funded investigations recorded the truncated remnants of the fort ditches and other cut features, but also some clay-bonded sandstone wall footings in the former eastern vicus reflecting remains of a building or buildings at right angles to the road leading from the south-eastern gateway to the fort and probably fronting onto it (Davies 2013: 27-8, figs 15-16, 2016: 52-3, figs 16-17). Some stonework recovered from the 1916–17 Thomas May excavations (and possibly the earlier Leader investigations) is in Rotherham Museum, and might repay further detailed study.
In Doncaster, the rescue excavations in the 1960s and 1970s rarely encountered Roman building remains, due in part to centuries of later robbing and truncation, but also more modern disturbance including the developmental groundwork. Metalled Roman road surfaces and possible wall construction trenches were recorded in Frenchgate (Buckland and Magilton 1986: 45, 47-8, fig. 8 sites 2c, 2d), but at High Street a clay surface was overlain by stone wall footings and several postholes (ibid.: 57, fig. 12). This building was likely to have been of predominantly timber construction. At 8–10 High Street, fragmentary remains of Romano-British features included postholes, beam slots, stone wall footings and associated gravel floor surfaces (Chadwick and Burgess 2008: 41-3, 53-80, figs 15, 17, 22-3), though unfortunately only small interventions could be investigated due to the depth of stratigraphy within the constricted site and the hostile contractors who also hampered excavation. Linear beam slots within the early Roman fort at Doncaster were revealed at Church Walk, perhaps for military timber buildings (Chadwick, Martin and Richardson 2008: 14-15, fig. 5). The recovery of copper-alloy and iron lever lock rotary and latch keys from Doncaster, Rossington Bridge and Templeborough indicates the presence of buildings with more secure Roman-style doors (Lloyd Morgan 1986: 92-3, 2001: 20, fig. 23; May 1922: 77), but no similar artefacts have been found elsewhere in South Yorkshire. This has implications for notions of social space and privacy during the Romano-British period.
Priorities and implementation