Excavations in Westgate Gardens, Canterbury, revealing the changing character of Roman Watling Street

In the summers of 2014, 2015 and 2016, volunteers from the Friends of Westgate Parks and the wider community carried out excavations to investigate the Roman Road (later known as Watling Street) projected to pass through Westgate Gardens adjacent to the Rheims Way underpass. The site generated significant new evidence not only of the construction and development of one of the main Roman streets in Canterbury, but also of early roadside occupation, and later changes of use which saw this St Dunstan’s area become isolated beyond the late Roman walled circuit, and therefore outside the town.

The new evidence from the Westgate Gardens digs is here compared with that from earlier excavations along Watling Street in Canterbury (Figs 1 and 2), and this has reinvigorated discussion of the chronology and purpose of the road, and indeed the developmental stages of Romano-British Can­terbury in general. How did the street system and therefore the town form, and why and when did it change? Road and ribbon development buildings, wear and tear, road re-surfacing and abandonments are all important clues strung out across the urban space; can these lead to a better understanding of Durovernum’s urban development?

The depth to which the archaeology could be recorded was curtailed by the water table at the edge of the Stour flood plain, but a clear sequence could still be established. Silty clays (G1) were seen at the base of the 2014/15 excavation, at around 8m AOD (approximately 1m beneath the present ground surface) at the south-west end of the trench; these were yellowish grey and appeared sterile in terms of anthropogenic material. It is possible that at least some of these deposits were partly formed of water lain material from an old course of the Stour. The succession of road surface levels uncovered in the Westgate Gardens digs are given a simple 1-3 numbering in this report.

Road level 1

The earliest road metalling seen (G5) was only partially visible at the limit of the 2015 excavation, its upper surface at around 8.4m AOD. This was mainly a [pg260]compacted metalling of medium sized flint gravel, but had occasional patches of a reddish hue indicating the presence of crushed daub or other ceramic building material within the sandy clay matrix. The largest area of this surface recorded was an irregular patch, c.1.2 by 2m, but of unknown thickness since it was only visible where it had been exposed by erosion of later surfaces. For the sake of this report this road surface is referred to as Road level 1.

Based on an Ordnance Survey map with the permission of the Controller of Her Majesty's Stationery Office, © Crown Copyright. Licence No. AL100021009

Gravel metallings (G8) were also uncovered on both sides of the road alignment, to the south-west (S15) and north-east (S17; S41). The roadside surface in the southernmost corner of the trench, at 8.36m AOD, was formed of compact medium­sized gravel in a grey, gritty, silty matrix, at least 0.3m thick. Material derived [pg261]from this context included fragments of tile and animal bone, and four sherds of pot dated to the first century. Approximately 2.0 by 0.35m of the other truncated metalled surface (S17) in the north-east corner of the trench was seen at the limit of excavation. This was at least 0.3m thick, and formed of very compacted small to medium sized gravel in a gritty clay matrix coated by a remnant clay floor with burnt daub inclusions (171); the upper surface here lay at 8.42m AOD.

Based on an Ordnance Survey map with the permission of the Controller of Her Majesty's Stationery Office, © Crown Copyright. Licence No. AL100021009

[pg262]A few metres to the north-west, a further metalled surface (S41) encountered in the 2014 evaluation test pit for the project is thought to be contemporary. This was clearly formed of very compacted medium-to-large flint gravel mixed with Roman-period ceramic building material. The surface covered the base of the 1.0 by 1.0m test pit at 0.8m beneath the surface (approximately 8.3m AOD).

Near surface S15, in the south-west corner of the 2014/15 trench, a thin dump or demolition deposit overlay the early silty clays (S4) at approximately 8.15m AOD. This was formed of lenses of burnt clay/daub, flint/pebble fragments, potsherds and some tile and oyster tile fragments. The pot assemblage, of 77 sherds, included seven of Canterbury sandywares, two of Upchurch-type finewares, two key sherds of possible transitional B2/R1 grog-tempered ware (second-century) along with a copper alloy coin probably minted in the reign of Commodus (180-192), dating deposition to the late second century at the earliest.

Above this deposit, a friable greenish brown silty clay soil (G2; sets 8 and 10), 0.25-0.3m thick, subsequently built up across most of the southern half of the 2014/15 trench, abutting the early metalling, and containing fragments of pottery, imbrex, tegula, other tile, daub and animal bone. An initial trowel-clean across the surface of the buried soils in this area (at between 8.34 and 8.4m AOD) produced sherds of pottery spot dated not later than the late second/third century, but the soil had certainly continued to accumulate in later phases (see below), and the area was subsequently eroded, rendering all dated material assigned to the context questionable.

Soils (G3; G6) were also building to the north-west of the road. Reworked alluvial or possibly dumped make-up deposits (G3: S16; S23) were the earliest seen here, lying at between 8.0 and 8.15m AOD, formed of moderately firm sandy silty clays, light greenish brown. These contexts yielded six late first-century potsherds, fragments of tile and animal bone and a further copper alloy coin, again probably minted in the reign of Commodus (180-192). This material was overlain by a firmer greenish brown silty clay (G6: S24), largely sterile in terms of inclusions but yielding a piece of Roman period tile. The deposit covered most of the northern part of the trench and was up to 0.3m thick; its surface, at 8.52m AOD at the northern end of the trench, dropped to approximately 8.3m AOD towards the line and level of the road.

Road level 2

The earlier road surfaces were next covered by an extensive new surface in a coarser gravel with a prominent camber (G9; Fig. 3) and ditch (G7) on its south­west side. This road, referred to here as Road level 2, was much wider than any previous, at approximately 7m. At the south-west side, where eroded, the gravel matrix could be seen to lie on a clayey bedding material. This metalling was the main feature of the 2014/15 trench, its largest extant area (S22) being c.2.5 by 4.66m in extent, and at least 0.2m thick. It was formed of very compact gravel in a silty clay matrix and its surface lay at 8.26m AOD at the northern edge, rising to 8.46m AOD at its highest point in the middle of the road. The remnant of the road camber (S21) was revealed on the south-west side of the road, formed of very compact medium sized gravel in a dark brown sandy clay silt (122), 2.84 [pg263]by 1.34m in extent and 0.1m thick, embedded in a clay silt lens (157). The top of the camber lay at 8.41m AOD, dropping to 8.3m to the south-west. The same road surface (S18) was in the adjacent 2016 trench only detected by shovel just beneath the water table, the enforced limit of excavation (Set 18), at 8.25m AOD, on the north-east side of the alignment.[fn1] Residual material within the metalling deposit for Road 2 represented a narrow date range from the mid-first to early second century, but several layers of some longevity were likely represented by this context as recorded. The roadside ditch (G7) was cut from 8.3m AOD along the south-west side of the road; it was 1.35m wide with steep sides to a concave base 0.45m deep. The ditch was was filled by clayey silt run off that included domestic material (as well as a coin of Claudius 1, either residual or deposited long after it entered circulation); perhaps significantly, pottery from ditch fills dated from late second to third centuries and no later.

[fg]jpg|Fig. 3 South-east facing section of the 2016 trench.|Image[/fg]

A pit (G12; S31) was dug through the roadside metalled surface (S41) to the north-west of the road. The pit, sub-rectangular as seen (0.8m by 0.7m), with initially steep sides cutting from 8.42m AOD to a dished base at 8.05m AOD, was filled by loamy silts of various similar hues, as well as tips of disturbed burnt daub and refuse (148; 158; 159; 160; 161; 149). The upper fill (149), produced 65 potsherds, imbrex, tegula, tile, daub, shell fragments, animal bone, wood charcoal and stone fragments. The pottery from deposit 149 dated to the late second-mid third century. This fill also contained an assemblage of cattle horn cores, evidently deposited soon after processing, that have been radiocarbon dated to cal AD 134­332 (at 95.4 per cent probability; 1786+/-33 BP; UBA-38435). A third-century date for backfilling of the pit appears most likely therefore.

Road level 3 (a series of later resurfacings)

The next road resurfacings (G17) (only seen in the 2016 excavation) did not cover the width of the existing road (see Fig. 3). This change of width or alignment is referred to here as Road level 3. The earliest surface in this sequence (S12) seen at the limit of excavation, at 8.36m AOD, and was probably not much more than 0.05m thick. It was topped by further thin surface (S13) another shallow layer of very compacted small to medium gravel in an orangey brown sandy matrix, [pg264]which only extended 1m into the 2016 trench, and was again only 0.07m thick, the surface lying at 8.43m AOD. Road surface S14 was yet further shallow patching of very compacted small to medium sized gravel in a greyish brown sandy matrix, covering the same area as S13, and only 0.05m thick. The surface of this lay at 8.48m AOD.

Further soils (G10) then built up along either side of Road 3. An area of the soil build-up overlay the Road 2 camber on the south-west side of the road, as well as roadside metalling and the infilled roadside ditch. A friable slightly compacted loamy silt (S28), this soil was 2.5m by at least 0.8m in extent and 0.08m thick, yielding second- to early third-century potsherds, but also a copper alloy coin (SF66), dated c. AD 330-402. The upper surface of the soil lay at 8.44m AOD. Further soil abutted new surfaces seen in the 2016 trench, overlying Road 2 on the north-west side (S30; Fig. 3), a loamy silt, up to 0.15m deep. The surface of the soil sloped away northwards from 8.53m to 8.33m AOD. If the soil build-ups either side of Road 3 give a clue to its narrower width, it was about 4m wide.

Finally in this sequence, the Road 3 was slightly widened, at least on the north­west side and within the area covered by the 2016 trench, by addition of a small to medium compacted gravel in a sandy silt matrix (S29) above and beyond preceding layers, extending to 1.4m into the 2016 the trench. This surface was slightly cambered, and between 0.07 and 0.2m thick. The upper surface lay at 8.55m AOD.

The final phases

In its final phase (G13) the road was in fact a very compacted soil containing gravel and other solid materials,[fn2] and is perhaps better considered a ‘track’ rather than a ‘road’ per se, into which a wheel rut (S35) was worn, c.0.1m deep and c.0.60m across. The road/track was still extensive though, approximately 2.95 by 1.75m in extent as seen (Fig. 3) and up to 0.3m thick, complete with camber, dropping from 8.73m AOD (barely 0.3m beneath the modern turf) to 8.33m AOD. The surface is represented in the 2014/15 trench by deposit 110 and related contexts (S32; S33), and of particular note is the concentration of coins that was retrieved from this horizon, with many dating to the mid-late fourth century.

Perhaps during the late fourth century but more likely after, a large hollow (G11), was worn into the southern edge of the road. The wear mark (S26) eroded all road surfaces from about 8.7m AOD downwards and was approximately 5 by 2.5m (extending beyond the limit of excavation) and to a depth of 0.3m. The hollow was filled by various silty soil deposits (129; 115; 118). Key later fabrics present were later Roman sandyware, late Roman grog-tempered ware, Nene valley type and Oxfordshire colour-coated and parchment wares, Mayen ware (at least 3 different pots), and some later Roman sandywares, all dating to the fourth or early fifth century. There was some evidence that the hollow had been sporadically patched, and in particular with a patch (S27) formed of a clay bedding layer topped with various re-used Roman period tile and brick and medium sized chalk fragments and occasional pottery dated to the fourth to early fifth century, and probably no earlier than mid-fourth century. The material was compacted, and 1.3 by 1.16m in extent and 0.15 thick, the surface at 8.49m AOD.

[pg265]Over succeeding centuries, track and hollow, became covered by soils that were eventually worked into a garden topsoil in the post-medieval period, and only later disturbed by sporadic dumping of twentieth century building materials, at least some probably derived from recorded clearance of City Centre bomb damage.

Interpretation; comparisons with earlier findings along Watling Street in the City

Material from within a gravel metalling cannot be argued to provide a date for the laying of the surface, but rather a terminus post quem (TPQ), a time after which a given surface was laid. It is important to clearly state this basic approach at the outset as some earlier published interpretations have been clearly driven by a twentieth-century tendency to ‘fit’ Roman period evidence within a projected history of the Roman Empire, rather than letting archaeological information speak for itself. This is a significant bias that must now be addressed if we are to compare the Westgate Gardens phases with those seen at other points along the route. The locations of comparative sites are numbered on Fig. 1.

Road level 1

At Westgate Gardens we have no evidence to date the earliest form of Watling Street to the first century. This might be due to the limitations of the excavation: the high water table and, to a lesser effect, time constraints, meant that the road could not be fully excavated. Perhaps earlier road phases were missed. It could also be that the earliest surface that was seen, Road level 1, was indeed first century in date. An alternative explanation, however, that the road was not even there and had not been built at this period, needs to be pursued, as it is now accepted that the street, much later to be called ‘Roman Watling Street’, was the in fact the most south-westerly street of a layout built around the beginning of the second century (see below).

First-century material recovered on the Westgate Gardens site, in particular pottery and coinage, was undoubtedly residual; the first-century pottery within the matrix of the roadside metalling to the south-west of the road alignment (S15), gives it a construction date of at least the late first century at the earliest, but an early second-century date, based on the building of the road, is surely more likely.

‘Watling Street’ passing through St Dunstan’s was subjected to various excavations in the 1950s in the Whitehall Road area (see Frere et al. 1987, 44ff), between 75m and 100m to the north-west of the Westgate Gardens dig (see Fig. 1, site ‘1’). Several trenches were excavated here that ‘encountered the Roman Watling Street as it ascended the valley-side from the crossing of the Stour towards Rochester’ (ibid., 45). Moreover, the same report details roadside timber buildings constructed with metalled yard or floor surfaces abutting the road, and surmises that ‘in the second century there seems to have been a suburb here’ (ibid.). So, it would seem the metalled surfaces at either side of the road alignment seen in Westgate Gardens fit a broader pattern, for which a second-century date is acceptable.

However, we should first note that the St Dunstan’s area can hardly be called a ‘suburb’ at this early stage, as it had not yet been redefined by the third-century wall that made the area ‘suburban’. More significantly, the road itself was interpreted at [pg266]Whitehall Road as having been built in the first century, on the basis of a historical bias. In trench W1[fn3] for example,

The old plough soil, through which the first road had been inserted c. A.D. 43 had remained open and contained material down to the Flavian period… Overlying was some dark earthy loam … an occupation layer containing charcoal and pottery… dating to the period c. 80-115, and this was sealed by a floor bounding a row of post­holes along the street… (ibid., 52).

The same bias can be seen in the description of trench WII at Whitehall Road, where material is dated in contradictory ways, again in order to fit a dominant historical narrative:

The first [road] surface … was bordered on the south side by a timber with pebble floor … which contained Claudian samian … and an early group of coarse pottery. The second road-surface … contained a sherd of Flavian-Trajanic Samian … and coarse pottery of the first half of the second century … and was laid perhaps in the reign of Hadrian (ibid., 54).

The first surface is implied as Claudian, or first-century, on the basis of residual material within an adjacent layer: the second as Hadrianic because of the combined dating of residual materials within. Neither surface need be as early as suggested, and the same goes for the associated structures. The road in trench WII was described as 5.32m wide and formed of three layers in total, those already referred to being topped by a third, apparently containing a surfeit of nails, and dated ‘probably not later than the Antonine period to judge by the pottery found at this level in [trench] W1’ (ibid.). It may be that the initial phases of this road at least are the same as ‘Road 1’, partially seen at the base of in the Westgate Gardens sequence, but the dating of the latest level is again problematical, in that a TPQ from a partial excavation through a possibly related surface has been used as a method of dating the construction.

This is all worthy of note because the archaeological evidence could in fact support a much later date for the building of this stretch of road up the hill towards the London road, perhaps even after the town was walled in the late third century or beyond (see below).

The nearest location eastwards of the Westgate Gardens digs where the earliest form of ‘Roman Watling Street’ has certainly been recorded is at the intra-mural Stour river frontage, some 300m north-east, at ‘Stour Street 1986 B’ (Fig. 1, site ‘4’). This excavation was sadly only reported on at a very summary level (Rady 1986), but the archive (SS86A; Archive number 93) in fact contains a detailed west-facing section drawing from this trench that includes the first Roman period street. The drawing suggests the first road surface at this low point in the terrain was mounded up on a primary gravel layer 0.25m thick (context 110); this base was approximately 6m across, while the road surface within the camber on each side looks to have been about 4.6m wide. This early surface was capped (or dressed?) by re-metallings (92 and 91). Roadside metallings are again recorded early in the sequence here, to the north of the road (context 85, interpreted as a possible track heading north away from Watling St).

Another 330m or so further east, nearer the Riding Gate on the east side of [pg267]the Roman town, further evidence of the earliest ‘Roman Watling Street’ comes from a 1953 excavation at 44 Watling Street (site ‘5’; Frere et al. 1987, 117ff). The writers point out that ‘Belgic [sic] sherds and a rim sherd of an early Roman cordoned jar were found in the metalling, suggesting it was laid down fairly soon after the conquest’ (ibid.). The width of this street, reported at ‘33ft’ or about 10m, is also unconvincing, since metallings in separate trenches I and III have been identified as a continuation (cf. ibid, figs 43-44) but do not appear to match; rather the alignment of ‘Street 1’, looks in section C-D (ibid., fig. 44) to lie fully in trench III, with a maximum width of c.6m. The metalling in Trench I was more likely another roadside metalling as we have seen elsewhere along the route. The first street metalling in trench III was at most 0.1m thick, with succeeding surfaces of similar thickness on the north side but deepening to the south: the chronology of these resurfacings is again narrowed by the writers’ insistence that a TPQ is equivalent to a date of construction.

It is very interesting to note at this point that ‘Roman Watling Street’ was not subsequently a permanent fixture along its entire course, for the next stratigraphic event in the sequence at 44 Watling Street is perhaps a surprise. Following construction of ‘Street 1’ and some resurfacing (‘Streets 2 and 3’), a structure with two rooms and an opus signinum floor was built right across the entire street (ibid., 119; fig. 44)! This significant hiatus is yet to be explained, but could for example either represent encroachment by a domestic dwelling on a little used thoroughfare, some sort of alternative funnelling of visitors through the town centre, during the second century, or a combination of both. The writers of the 44 Watling Street report date this strange occurrence to the second half of the second century, although a third-century date is equally viable on the basis of the evidence presented.

Road level 2

While dating evidence for the wider ‘Road level 2’ at Westgate Gardens is not conclusive, this development certainly seem to correlate with the earliest metalling seen in the 1950s excavations at the nearby London Gate (Site ‘2’), where the surface lay at the limit of excavation and was not further investigated, but was clearly earlier and much wider than the gate itself (Frere et al. 1982, 33; see figs 2 and 6), so a terminus ante quem (TAQ) of about AD 270-290 is provided.[fn4]

The same wide road in fact seems to be that recorded in evaluation trenches during the Tannery development (Site ‘3’), where roadway was recorded in trenches 2/5 and 7 and just over 7m wide, and is probably the metalling recorded as deposit 32 in the 1986 section drawing from Stour Street B (site ‘4’; ibid.), at just over 7m width, in this case eroding into a wood lined drain along the southern edge. At site ‘5’, 44 Watling Street, ‘Street 4’, again at a very similar width, appears to correspond. This was laid immediately above the opus signinum floor that had been built over the earlier surfaces (ibid., 119, fig. 44). According to the dating of the sequence here, this would have to be in the late second or early third century, but a date later in the third century is equally afforded by the evidence.

Excavation of Roman Watling Street as part of the Whitefriars development early this century (site ‘6’, where it was labelled ‘Road 1’) recorded the digging of a ditch through the second-century version of the road along northern side.

[pg268]The ditch was infilled and covered with an again much wider street after 200, and quite possibly later, given a date range of 170/200-250 suggested for this sub­period of activity (Alison Hicks and Mark Houliston, pers. comm.). Again, the double-arched Roman-period gate at Riding Gate (site ‘7’), built as an entrance to the town from the Dover direction along with the new wall in 270-290, certainly accommodated a pre-existing wider road (ibid., 43ff; 49).

It thus seems likely that ‘Road level 2’ at Westgate Gardens was part of an early to mid third-century scheme of road widening across the town. Turning back to the north-west of the Westgate Gardens excavations, however, and heading away from the town centre, it is very interesting therefore that the wider road is not reported in the Whitehall Road sites (site ‘1’); was the third-century road widening project curtailed at some point before the road climbed away through St Dunstans? Did the road even go that far at this period, or was it the narrower road represented by Road level 3 that made its way up the valley side?

Road level 3

Westgate Gardens ‘Road level 3’ surfaces seem to represent a distinct narrowing of the road design, perhaps in the first half of the fourth century. As has already been noted, the St Dunstan’s stretch of road is recorded as narrower: whether this was indeed the same road as ‘Road level 3’, or an earlier version, awaits further archaeological testing. To the east, the route through London Gate (ibid., site ‘2’) was indeed decidedly narrow and it is further interesting to note that a similar pattern of narrow latter surfaces was seen in subsequent metallings at 44 Watling Street (site ‘5’; ibid., ‘Streets 5 to 8’). The Whitefriars team report a pit and then a building encroaching on the road by AD 355, blocking the northern half (site ‘6’; Alison Hicks and Mark Houliston, pers. comm.), and at Riding Gate, the smaller foot-passage arch on the southern side of the road ‘was at some point in the fourth century blocked and incorporated in an encroaching building (site ‘7’, ibid., 51).

Late fourth-century and beyond

The later phases of the road in Westgate Gardens, the dirt track and hollow way erosion tentatively dated to the late fourth and fifth centuries, are impossible to correlate even in broad terms with the intra-mural sequences. For a time there would seem to have been a pattern of coinage deposition on approaching river and town, or on leaving, but the road in Westgate Gardens would then have had a very neglected appearance in the second half of the fourth century and, later still, seems to have been put to a different use, perhaps as an opportune place to cross the river amid a rising water table. The latter was a late antique environmental difficulty that has been seen elsewhere in the Stour flood plain, and in particular at the Tannery (Simon Pratt, pers. comm.).

Roman Watling Street and Roman Canterbury

While particular sections of roads can be subject to very localised change that would appear only in an individual excavation, roads are also extensive linear features that are built and maintained by more general projects; might we be [pg269]able to discern patterns of road development in other excavations along Roman Watling Street? The Westgate Gardens evidence (so far) in comparison with that of other archaeological encounters along its route, lends some further weight to an emerging view of the development Roman Canterbury. Much of this narrative (see Figs 4-6), while it accords with an embryonic archaeological consensus (see Millet 2008, 156ff, for example), remains a hypothesis, but it is one within which the Westgate Gardens evidence can be posited. The story begins in the first century with as yet no Roman road through Westgate Gardens or St Dunstan’s.

The ‘first town’

It has long been noted that the theatre and early phases of public baths at the core of the town, as well as the roads associated with them, differ in alignment from the later established layout of Roman Canterbury, probably as a result of two distinct phases of development (cf. Figs 4 and 5). It has also been rightly suggested that the first theatre and baths in fact reflect early Roman respect for a late Iron Age topography, and a ritualized one at that (cf. Millett 2008, 158). This sort of thing was not uncommon in the Gallo-Roman sphere, especially where existing water or other cults were ‘tectonized’ as sanctuaries (cf. ‘Springhead’; see Millet 2008, 160-1). Furthermore, the theatre may enclose a spring line heading towards the Stour flood plain, as contours suggest (Simon Pratt, pers. comm.): another potential clue. The direct association of theatre with temple also attests to Gallo-Roman influence[fn5] and axially aligned water features graced the (perhaps [pg270]slightly later) precinct (see Bennett 1980, 12). Along with the early baths, theatre and associated road, an extensive late Iron Age ditch system in St Dunstan’s (not as yet reconstructed or planned), elements of which have been recorded to the north-east of the Westgate Gardens site, also lies at the same alternative angle. The early road in from Richborough crossed valley and river (Fig. 4) and clearly went along the north-west side of this ditch system heading for London (today the line of London Road). Pottery and tile kilns near the river in St Dunstan’s may have started production in the first century, and there is evidence of a late conquest period fortlet, sporadically occupied, to the south-west of the embryonic settlement (see Bennett 1982). An early cremation cemetery formed here, and anothet in the outer plots of the St Dunstan’s ditch system. Meanwhile early town houses within the same quarter as the theatre and baths, seen in the Marlowe Car Park and Whitefriars excavations (Blockley et al. 1995; Alison Hicks and Mark Houliston, pers. comm.), also aligned with this early grain of the landscape. The Whitefriars site even uncovered two early streets that cohere with the early theatre and baths axis. This embryonic and organic development of an urban-style settlement based on existing landscape and perhaps ritual influences was not to be the permanent template of the town, however.

The ‘second town’ and Roads 1 and 2 at Westgate Gardens

In the early second century the new grid (Fig. 5), or rather ‘ladder’ of roads, was set out at intervals of 500 Roman feet (Pratt 2004, 22), at Canterbury, superseding the previous, more organic layout. While this pattern of development is actually [pg271]comparable with Roman towns at St Albans and Silchester, for example, it is notable that those towns developed their established systems 20 or perhaps as much as 40 years earlier than Canterbury (see Millet 1990, 77-79; Mattingly 2006, 266-71). It suggests that Canterbury probably only became a Civitas capital at the beginning of the second century, perhaps a reflection of its people having been of continuing subaltern status to hegemonic centres north of the Thames and in Sussex, before and for some time after the Roman invasion in 43, at least in terms of Imperial recognition (cf. ibid., 68-80; cf. figure 4 therein; cf. Champion 2016, 162ff; 165).

The most north-easterly road of the new arrangement at Canterbury was the already existing road from Richborough, probably the first Roman road to Canterbury, which seems therefore to have provided the start-line for the new layout. The road we call ‘Watling Street’ aligned with it 1000 Roman feet to the south-west. An intervening street, 500 Roman feet from both, is also dated to this period (Blockley et al. 1995, 84-8); the newly angled baths palaestra cut into its metallings. New streets in the Whitefriars area (Alison Hicks and Mark Houliston, pers. comm.) and beyond would now be based on this second-century pattern. What was to become ‘Roman Watling Street’, therefore, bounded the theatre and temple precincts within the new arrangement, in the process short-cutting the London Road towards the developing port of Dubris (Dover), which may well have been part of its raison d’etre.

It would seem from the available evidence that it was the construction of the road we call Roman Watling Street in the early second century that led to new buildings and roadside metallings into the Tannery and St Dunstan’s area, so it became something like a nascent south-west quarter of the town with a possible mansio and a developing ceramics industry. The roadside surfaces in Westgate Gardens might be associated with the latter, and the cemeteries next to the London Road at the top of the hill could even have reached their apogee in the late second- to mid- third-centuries as the last port of call for some of south-west area’s inhabitants (see Frere et al. 1987, 56ff; Weekes 2017).

But the St Dunstan’s ‘project’ does not seem to have ‘stuck’, and may not have lasted much more than a century. It is notable, for example at Westgate Gardens that the road ditch of the third-century Road 2 cut through the roadside metalling on its south-west side, and that a pit was hacked through the roadside metalling on the north-west side at a similar date. This pattern of perhaps dwindling occupation of the St Dunstan’s area being underway by the late second or early third century has been noted elsewhere (Duffy et al. forthcoming). Did the building of the wall on the other side of the Stour merely confirm the St Dunstan’s area as lying ‘outside the town’? Moreover, as we have noted, the narrower version of the road climbing up the valley side in St Dunstan’s could, on the basis of archaeological evidence, be later.

The re-establishment of the line of the road as a wide main thoroughfare, as late as the third century, seems to be evidenced in the Westgate Gardens excavation as ‘Road level 2’, however. Fascinatingly this looks to have been a road building scheme across the third-century town, a period that saw the second theatre replace the first and probably the further development north-east of the Richborough road. It is very interesting that the ‘Watling Street road widening scheme’ scheme does not seem to have been continued much into St Dunstan’s, at least beyond the flood plain. Again, was this area already in the third century considered ‘outside the town’, long before the wall was built?[pg272]

[fg]jpg|Fig 6 Map of ‘Town 3’, after c.AD 270-400, based on currently available evidence (after S. Pratt, in prep.).|Image[/fg]

The ‘third town’ and later: Road level 3 and final phases

The walling of the town in the late third century underlined the St Dunstan’s area as ‘outside the town’. In Westgate Gardens, the narrowing of the ‘Road 3’ in the fourth century is similar to later developments within the walled circuit, but perhaps not as directly identifiable with narrower late surfaces at 44 Watling Street, for example. It seems more likely that such resurfacings were by now more sporadic and localised, and that the road narrowed more by encroachment and ‘accident’ than by any ‘grand design’. By the time fourth-century coinage was being deposited near river and gate in Westgate Gardens, later cemeteries were formed closer to the wall and the St Dunstan’s ‘suburb’ was truly becoming a place for the dead. The road became a track, and was then, through alternative use, eroded and soon abandoned in the sub-Roman period.

Huge thanks to Anna Bell, the Westgate Parks Officer, to Ian Cameron Fleming of Canterbury City Council who managed the Westgate Parks Project, and to the Friends of Westgate Parks and all their excellent volunteers for making these excavations such a success. Colleagues from the Canterbury Archaeological Trust as well as volunteers Richard White and Fred Birkbeck were brilliant co­excavators, and the work was funded by the National Lottery. Over 60 locals and [pg273]visitors of all ages took part in the digs, and over a 1000, including many families, and primary and secondary school classes, received the ‘grand tour’ while we were on site. We hope to be back! Thanks also to Simon Pratt, Alison Hicks, Peter Atkinson and Mark Houliston for assistance with archives and drawings.

Blockley, K., Blockley, M., Blockley, P., Frere, S.S., and Stow, S., 1995, Excavations in the Marlowe Car Park and Surrounding Areas, CAT.

Champion, T.C., ‘Britain before the Romans’, in M. Millett, L. Revell and A. Moore (eds), The Oxford Handbook of Roman Britain, OUP, 150-78.

Duffy, W., Gollop, A. and Weekes, J., forthcoming, ‘New cemetery evidence from Late Roman Canterbury: the former Hallet’s Garage site in context’, in Late Antique Archaeology.

Frere, S.S., Stow, S. and Bennett, P., 1982, Excavations on the Roman and Medieval Defences of Canterbury, Maidstone, KAS.

Frere, S.S., Bennett, P., Rady, J. and Stow, S., 1987, Canterbury Excavations: Intra- and Extra-Mural Sites, 1949-55 and 1980-84, Maidstone, KAS.

Mattingly, D., 2006, An Imperial Possession. Britain in the Roman Empire. Penguin Books. Millett, M., 1990, The Romanization of Britain. An Essay in Archaeological Interpretation, cup.

Millett, M., 2008, ‘Roman Kent’, in J. Williams (ed.), 2007, The Archaeology of Kent to AD 800. Kent History Project 8, The Boydell Press.

Pratt, S., 2004, ‘Canterbury (Durovernum Cantiacorum)’, in T. Lawson and D. Killingray (eds), An Historical Atlas of Kent, Phillimore, 21-3.

Rady, J., 1986, ‘Nos 36-7 Stour Street’, Canterbury Archaeological Trust Annual Report 1985-86, CAT, 15-16.

Weekes, J., 2017, ‘Funerary Archaeology at St Dunstan’s Terrace, Canterbury’. in Pearce and Weekes (eds), Death as a Process: The Archaeology of the Roman Funeral, Oxbow Books.[pg274]