Two New' Town Gates Roman Buildings and an Anglo-Saxon Sanctuary at St Mildred's Tannery Canterbury

225 two ‘new’ town gates, Roman buildings and an anglo-saxon sanctuary at St Mildred’s Tannery, Canterbury simon pratt Preliminary archaeological investigations for the redevelopment of St Mildred’s Tannery (TR 145 577) were undertaken by Canterbury Archaeological Trust in 1999-2002 (Pratt and Sweetinburgh 2004). In June 2004 some additional testing was monitored and construction work began the following September, since when the CAT has conducted a complex watching brief, planned to continue through to the completion of the groundworks. Generally this is confirming or refining earlier conclusions but there have been some surprises. T he site comprises three sectors, the largest (A) lying west of the intramural branch of the Stour, another (B) between the Stour and Stour Street and the smallest (C), to the east of Stour Street. At 3.5ha (8 acres), the overall site is comparable in size to the Whitefriars development. However, unlike the earlier site, this largely residential development has no basements, cellars or underground parking and a very different archaeological strategy could be adopted. This seeks to minimise damage to the remains by careful design, aiming to preserve them where they lie rather than staging large-scale excavations. In collaboration with English Heritage, several pairs of boreholes have been sunk to monitor the level, acidity, salinity and oxygenation of the groundwater – critical factors for long term preservation of organic remains in situ. Occasionally it proved impossible to keep individual lift-pits or stretches of main storm drains and foul sewers entirely above the archaeology which was, instead, excavated stratigraphically down to formation level. All the new buildings rest on concrete piles, leaving the archaeology under 97-98% of each footprint untouched. Despite this, it has been possible to squeeze a considerable amount of information out of the tiny proportion of the ground that has been investigated, due in large part to the apparently innovative practice of systematically extracting and analysing soil cores from many of the new pile positions (Fig. 1). U sually, water poured in continually to the deeper trenches (needing up SIMON PRATTatt 226 Fig. 1 St Mildred’s Tannery, Canterbury: known and conjectural Roman topography (auger data included, piling data omitted). GATES gates , R OMAN BUILDINGS AND A.S. SANCTUA RY at ST MILD RED ’S tanneRY 227 to four pumps to control it) and steel boxes had to be inserted to prevent cave-ins. At one time, five 3.5m long, 3m high, 1m wide boxes were in place end-to-end. Locating the town wall I t was known that, under some 3-4 m of tanning waste, the town wall crossed Sector A somewhere near its south-western end. Avoiding damage to the wall and associated structures was a priority, especially as they might damage an expensive piling rig, so much effort and ingenuity was put into locating its exact course and searching for any towers or bastions. Total exposure was avoided as, to be certain there were no razed earlier foundations, all post-Roman deposits on both sides could require deep, stratigraphic excavation in unstable, waterlogged soils and conditions so polluted that whenever the ground was disturbed (and, often, for days thereafter) five different meters were used to monitor the levels of eleven gases whilst respirator masks were compulsory. As the overburden in each area was removed, the wall core outcropped in three places. Its line was gradually pinned down by cleaning these exposures and cutting shallow transects. Leading geophysical surveyors GSBgsb were also brought in. They sealed their expensive equipment in plastic cocoons, donned respirators and set to with ground penetrating radar, magnetometry and resistivity. As feared, the ubiquitous metallic scrap, high water-table and leachates (which had turned the ground into ‘electrolytic soup’) conspired against them. Though some possible indicators were glimpsed, nothing was clear enough to be relied upon without further investigation. Two small trenches were cut for this purpose: one found a post-Roman glacis of compacted rubble but no walling, and the other only a nineteenth-century rubble spread. A larger trench was cut along a well-preserved stretch of the wall’s extramural face and, for the remainder, a 3m long steel probe mounted on a mechanical excavator was used to check along both sides and then along various transects. Roman town wall and postern (Figs 2 and 3) G eophysics and probing suggested an earlier building may lie beneath the apparently tower-less town wall and rampart: this will be further explored at a later stage. For the first time at Canterbury, an extramural rendering of crushed-tile mortar was found. An area of metalling just inside the wall may have served as a construction haulway or have been a later road using a reduced rampart to remain above the marshy ground. The wall was interrupted by a contemporary opening with tile quoins on the extramural face and rebated interior faces of mortared flints with double string-courses in what appeared to be ‘lydion’ tiles, measuring 1 x 1½ SIMON PRATatt 228 Fig. 2 St Mildred’s Tannery, Canterbury: Roman postern and post-Roman Halistane GATESgates, ROMAN BUILDINGS AND A.S. SANCTUA RY at ST MILD RED ’S tanneRY 229 civilian Roman feet (1 pes Monetalis x 1½ p.M., 295.5 mm x 443.3 mm). At its narrowest, between the external quoins, the opening measured only 1.475m (5 p.M.?), widening to 1.94m (6½ p.M.?) just within and to 1.99 m (6¾ p.M.?) across the rear of the wall. The stubs of two wing walls also survived and probing indicated that they probably ran at least 3.0m back from the intramural face. These presumably flanked a passageway through the rampart, probably with a solid paving at about 7.25m od. So narrow was the opening that it might be more properly be termed a postern rather than gateway (exposed fabric and probing showed there was no wider carriageway to either side, whilst the floor level seems too high for a floodgate). The nearest parallel, typologically and topographically, is Canterbury’s London Gate, reported as 2.44m (8½ p.M.?) between the wing walls but measuring perhaps 1.90m (6½ p.M.?) between the (missing) external quoins (Frere et al. 1982, 33, figs 6-7). Roman streets In Sector C, earlier work had found traces of Roman clay floors. Piston-coring of a few of the new pile positions identified the Roman floors extending over much of the site, with a more complex sequence towards the south-eastern end. A single service trench was deep enough to expose Roman deposits in its base. It ran along what was probably a short corridor with a clay floor and beam-slots or pads to either side and at each end. This was the first direct evidence for the alignment of any structures of this period in Sector B or C and roughly matched the more northerly (and later) of two conjectural lines for the Roman predecessor of this stretch of Stour Street. In the northern part of Sector A, a shallow trench exposed a small area of gravel metalling on the already established line of Roman Watling Street. Two evaluation trenches had each picked up gravel surfaces bounded by clay-floored buildings on what had been thought a probable line for a Roman side street, extending the line of Black Griffin Lane to the south-west, but later coring and a short stretch of new sewer trench between them found floor or demolition sequences and walls rather than metallings. Roman bath-house? Pre-pile coring confirmed the presence of Roman buildings and metallings over much of the northern part of Sector A and ground beam trenches intersected two Roman walls a little south-west of Watling Street. The partial removal of Tannery footings re-exposed part of a Roman building, in mortared flint and lydion tile, found in 1987 (Blockley 1987a; id. 1987b). Just to the south of the 1987 excavation, a sewer trench crossed what was probably the end wall of the same building or wing. A little farther to SIMON PRATatt 230 Fig. 3 St Mildred’s Tannery, Canterbury: (a) Using GPR (left) and resistivity to search for towers, etc. after the removal of the tanning waste; (b) Composite photograph looking north along the town wall in the area of the two gateways (1 m scales across the wall, 0.5 m scale along it; (c) ?Aisled building, exterior view of main north-western wall (0.5 m scale). GATESgates, ROMAN BUILDINGS AND A.S. SANCTUA RY at ST MILD RED ’S tanneRY 231 the south-west, trenching exposed parts of an apse wall about 0.89m (3 p.M.) thick. This formed the south-eastern end of a room approximately 3.55m (12 p.M.?) wide. The wall was of mortared flint with scars from a string-course of ?lydion tiles. A drain or hypocaust flue ran tangentially to the wall, perhaps with a shallower spur to the north-east. A borehole about 2.3m south of the external apex of the apse encountered a poured mortar floor or bedding over a thick levelling deposit. This sealed a ?drainage layer of broken Roman tiles which overlay an earlier, scorched tile surface bedded on mortar. Other boreholes encountered what were probably the other walls of the same room and a sequence of ?floors, beddings and levellings near its centre. Towards its north-western end, 0.25m of ash and charcoal overlay a floor (or wall?), of tile bedded in crushed-tile mortar. The apsidal plan, drains/flue(s) and burnt material all suggest that this was a bath-house or bath-wing. Roman aisled building? N ear the river, short stretches of Roman walls were exposed in three trenches. These marked three sides of a large structure first identified in 1991, then assumed to represent an entire building rather than a single chamber (Pratt 1992; CAT 1992, 374-5). The north-western (Fig. 3c) and south-eastern side walls were both about 0.895m (3 p.M.) thick and had external offsets, probably just below floor level, about 0.30m (1 p.M.) wide. The north-eastern end wall was about 1.04m thick (3.5 p.M.?), at least along its central stretch. The walls and footings were of mortared flint with string courses of lydion tile laid in double thicknesses. On all three sides, the uppermost surviving string course had been covered in trampled earth which had not been fully cleaned off before the next course of flints was laid. This echoes a line of hobnailed boot prints along the surviving crest of the outer wall of the building found in 1987 (to this day, new ground beams are used as convenient walkways across a muddy or uneven site, though subsequent cleaning is usually more thorough). What may have been the robber trench for the fourth side had been seen in 2001, but this was only 0.65m wide and was more probably for a separate or subsidiary structure. Within the main walls, two massive upright timber piles were found, each roughly 0.5m square and of unknown length. A hole in the final surface gravel overlapping the western pile suggested they supported timber posts measuring about 0.30m (1 p.M.) square. As this pile projected above the sewer formation level its top was sliced off and sent for dendrochronological analysis. Unfortunately, though from an oak, it came from the upper part of the trunk and was too heavily knotted and fast-growing to date. It seems likely that such pieces were deliberately selected for piles, which needed no sophisticated carpentry, smoother-grained wood being retained for use above ground. A ?beam SIMON PRATatt 232 slot and possible beam pad may represent minor internal divisions or fixtures but, like an area of scorching and, perhaps, the post-hole over the western pile, may pertain instead to a post-Roman structure. It is not yet clear whether the walls and posts (or robbed-out piers) supported by the piles represented an aisled/basilical building or one with three or four wings ranged around a courtyard. In the former case, the putative central aisle-width of 32 p.M. would place it in the upper range of those known from Roman Britain whereas the estimated overall external length of 75, 85 or 95 p.M. (depending upon the number of aisle bays) would leave it with an unusually low length/width ration. Apparently erected rather late in the site’s Roman sequence, with quite a high floor level, it may have continued in use after other buildings in the area had been abandoned (perhaps around the end of the third century) to a rising water-table. Even so, most or all of the walls were probably razed before the end of Roman rule. Adjoining structures T he putative aisled chamber was surrounded on all sides by separate or subsidiary structures. Its north-eastern wall was set about 7.40m (25 p.M.) back from the margin of Watling Street. Some, at least, of the intervening area had a (final) gravel and rubble surface ramping gently up against the end wall. This area may have been roofed and was probably separated from the street by a timber-framed corridor. To the south-east, the main hall was probably abutted by a timber-framed lean-to which gave, across a narrow yard or street, onto a riverside hard. Two flat stones sitting on the hard’s gravel surface were probably stepping-stones easing access to the water’s edge. To the south-west, assuming the robbed wall was not the main chamber’s end wall, there was probably a more substantial structure. A wide leat separated the robbed wall from a back street or narrow yard. The north-western wall was adjoined by what was probably a corridor, with an internal width of 9 p.M., bounded by a robber trench some 3 p.M. wide. Beyond this there was probably either a courtyard at a lower level or open ground. The width of the robbed footing seems excessive for such a narrow single-floored structure and it may have supported two or more storeys or, perhaps, a colonnade. This may have overlooked a large courtyard, possibly extending as far as the 1987 building and apsidal room. Anglo-Saxon sanctuary A charter of ad 804 (Sawyer 1968, 160) granted much of the area to the nuns of Lyminge as a sanctuary and remains of brushwood trackways and a (boundary?) ditch found during the evaluation stage had been GATESgates, ROMAN BUILDINGS AND A.S. SANCTUA RY at ST MILD RED ’S tanneRY 233 attributed to this period. The existence of a small island of relatively drier cultivated land was conjectured over what is now thought to be the southern end of the ?aisled Roman building. The more easterly of the two timber piles found within this building was sealed by the final phase of gravel surfacing between its walls. A dome-headed bronze pin, engraved with a cross, was recovered from within this gravel and one with a polyhedral head from beneath a tile fragment lying on its surface: both were of broadly ninth-century date. A thin layer of undisturbed ‘peat’ sealed the razed south-eastern wall near its mid-point (though there is no true peat on the site, the term is applied throughout to highly organic clay silts, some clearly formed in reed beds, others in drier but marshy conditions). This peat was overlain by what were probably the poorly preserved remains of one of the brushwood trackways. On the riverward side of this, the uppermost surviving Roman surface was also overlain by peat, but here it had been heavily disturbed, probably by trampling. On the track’s opposite side, instead of peat, was what appeared to be a cultivated loam. The evidence now suggests that the island was more extensive than thought and that, following the razing of the Roman building, several inches of peat developed over its less elevated portions before a trackway was built across the lower marshland to its north. Part of the island was then cultivated. Though narrowed, the leat to the south-west probably continued in use during the earlier stages of cultivation but was eventually clogged with peat. A gravel metalling was laid down over part of the island, though it is unknown whether it was an internal floor, external yard or trackway. The post-hole over the western Roman pile, a ?beam slot, ?pad, ?hearth and some of the final surfaces identified in the 1991 evaluation may also have belonged to this period. Re-modelled defences A t some stage, the Roman postern was blocked by a thin wall of mortared flint, with rubble and soil filling the passageway behind. A wider gateway was cut through the already partially collapsed or razed town wall, about 11m south of the postern (Fig. 4). This was probably the Halistane (Holy-stone?) to which medieval documents make reference and beyond which, perhaps, lay an ‘ordeal pit’ (Urry 1967, 198, maps 1(b), 2(b)8). Though the intramural and extramural quoins were missing, the interior was faced with large, roughly squared ?re-used Greensand and yellow Sandstone blocks set 3.10m apart (about 10.5 p.M. or 10ft 2in.). For at least 9m north of this new gate, the wall was rebuilt on a slightly different alignment but there was far more radical alteration towards the river. The Roman wall was cut away about 7.60m south of the new gate and replaced with a slightly narrower re-entrant on a trench-built footing of large packed flints. This ran back some 8.7m from the old intramural face before SIMON PRATatt 234 Fig. 4 St Mildred’s Tannery, Canterbury: early medieval re-entrant walls GATESgates, ROMAN BUILDINGS AND A.S. SANCTUA RY at ST MILD RED ’S tanneRY 235 turning through about 110° and running another 10.4m to an abutment on the contemporary river bank. Three surviving timber posts probably held back a plank or wattle revetting of the abutment’s packed flint core. Two large reused blocks, one of Greensand and the other of white Limestone lay against or partially under the intramural face of the re-entrant wall and were probably of one build with it. What may have been the robbing of a third block was noted just to their west. These presumably supported a timber superstructure, most probably a wall-walk or steps up to one. The moat was widened or diverted to run along the re-entrant and, at least near the external angle, equipped with a steep glacis of tightly packed demolition rubble. This re-entrant was probably matched by one on the opposite side of the river, where a public space separates the river and St Mildred’s churchyard, the latter being bounded to the west by an eighteenth- or nineteenth-century rebuild of the town wall. N o direct dating evidence was found for any of the alterations to the defences. Only residual Roman material was recovered from behind the postern’s blocking wall and from the glacis. A few sherds of ?thirteenth-century pottery recovered next to the re-entrant’s abutment may have eroded from its core or have washed up against it. However, the absence of Caen and ragstone, re-use of Roman materials and apparent lack of any documentary reference to the creation of the re-entrants suggests an early date and a very rough terminus ante quem may be provided by a document of c.1200 (CCAcca dcc ChAnt/C/899). In it, Christchurch Priory granted to Richard the currier, for his lifetime, a tenement ?between St Mildred’s churchyard and the city wall. A condition of the tenure was his maintenance of a weir, perhaps between the ends of the re-entrants. The most probable occasions for blocking the postern are the establishment of the sanctuary in 804, the Danish siege of 1011 or the poorly documented late eleventh/early twelfth-century replacement of the motte-and-bailey at Dane John with the stone castle near St Mildred’s church (Renn 1982, 71-73). The creation of the re-entrants may be associated with the last of these, securing the royal castle’s flank, or with slightly later work on the town walls and gates (ibid., 74-5). Medieval structures I n Sector C evaluation trenching had found part of a medieval or early post-medieval flint-walled building with clay floors. Some construction trenching cut a little into or bottomed on medieval and early post-medieval floors and walls near the street frontage. A ?late medieval well and small ?early medieval hearth were also uncovered. Clay floors and flint walls exposed to the south-west suggest that medieval structures survive at a higher level beneath Rosemary Lane car park. A new drain cut across Stour Street and Sector B to discharge into the SIMON PRATatt 236 S tour exposed a mortared chalk footing on Sector B’s street frontage, turning to run several metres back from it. A revetment of pointed but otherwise unworked wooden posts, associated with medieval pottery and cobbler’s scrap leather, was found a few metres back from the existing riverside wall. Two small areas of clay floors were exposed by shallow groundworks in the northern part of Sector A: one was probably within an open-ended building and both were probably agricultural buildings belonging to the Franciscans who farmed the area from the thirteenth to sixteenth centuries. Another medieval structure was identified in a deeper trench near the river. It was represented by an upright stake and horizontal poles around three sides of a rectilinear area with a patchy clay floor within and without. The poles may have been the bottom course of a small wattle structure such as a bin or pen. This structure was overlain by more substantial remains consisting of a chalk floor between two mortared ragstone and flint wall-footings. This building lay close to the contemporary river but separated from it by a narrow strip of reclaimed ground with timber revetting tied back with horizontal logs or poles running back under the footings. A little to the south, what may have been the same phase of revetting turned to jut into the river, perhaps forming a jetty or abutment. Post-medieval period The landward wall of the chalk-floored building was abutted by sandy dumped deposits similar to ones found within the robbing of the side wall of the open-ended structure and filling various shallow ditches elsewhere. These generally contained late medieval or early Tudor material, perhaps representing post-Dissolution amalgamation into larger fields. This probably dates to around 1539 when the King’s Receiver, Thomas Spylman, obtained the friary (Letters Hen. VIIIiii, xiv, 1354.40), 1544 when he sold it (Cotton 1924, 63) or 1545, when he was signatory to an agreement rationalising land holdings of the archbishop and dean and chapter (CCAcca dcc/Register U fols 141-142). L ate sixteenth-century bird’s-eye views of Canterbury by William Smith and by Braun and Hogenburg contain no useful detail for the Tannery area – only the nucleus of the former friary is shown whilst a certainly anachronistic watergate crosses the Stour (Fig. 5). Speed’s 1605-1610 survey of Canterbury and its derivatives are little better, though perhaps showing a rather confused representation of the two re-entrant walls by the river. These are far clearer in a map of c.1640, held in the cathedral archives (CCAcca Map 123). A breach in the town wall, between the blocked postern and presumed Halistane, was probably cut from both sides simultaneously and may be one of those made after the Christmas Day riot of 1647 (Everitt 1960, 123-124, 140-141). By 1752, when the GATESgates, ROMAN BUILDINGS AND A.S. SANCTUA RY at ST MILD RED ’S tanneRY 237 Fig. 5 St Mildred’s Tannery, Canterbury: early representations of the city wall in relation to the river SIMON PRATatt 238 Doidges published the first triangulated survey of the city, both gates and both re-entrants had disappeared. acknowledgements T he work is funded by the developers, Bellway Homes, and monitored by their archaeological consultant, Professor Martin Biddle. Grateful thanks are extended to all those involved, on site and off, archaeologists and others. Particular gratitude though to those who suffered in the freezing, flooding trench-boxes or the freezing, flooding, foul contamination remediation stage and most especially, with sincere admiration, to the die-hards who slogged away in both – Tracey Smith, Ian Dixon and Paul Renn. references B lockley, P., 1987a, ‘The Tannery’, Canterbury’s Archaeology 1986-1987, 18. B lockley, P., 1987b, ‘The Tannery’, Archaeologia Cantiana, civ, 314. CAT cat , 1992, ‘Interim Report on Work carried out in 1992 by the Canterbury Archaeological Trust’, Archaeologia Cantiana, cx, 374-5. CAT cat , 2005, ‘Interim Reports on Recent Work carried out by the Canterbury Archaeological Trust’, Archaeologia Cantiana, cxxv, 262. C otton, C., 1924, The Grey Friars of Canterbury, Franciscan Studies (extra series) II, Manchester. Everitt, A., 1960, ‘An Account Book of the Committee of Kent, 1646-7’, in I.J. Churchill (ed.), A Seventeenth Century Miscellany, Kent Records XVII, Maidstone, 115-152. Frere, S.S., Stow, S. and Bennett, P., 1982, Excavations on the Roman and Medieval Defences of Canterbury, The Archaeology of Canterbury IIii, Maidstone. Pratt, S., 1992, ‘St Mildred’s Tannery’, Canterbury’s Archaeology 1991-1992, 8. Pratt, S. and Sweetinburgh, S., 2004, ‘St Mildred’s Tannery and Masons’ Yard’, Canterbury’s Archaeology 2002-2003, 10-13. Renn, D.F., 1982, ‘Canterbury Castle in the early Middle Ages’, in P. Bennett, S.S. Frere and S. Stow, Excavations at Canterbury Castle, The Archaeology of Canterbury I, Maidstone, 70-75. S awyer, P., 1968, Anglo-Saxon Charters, an Annotated List and Bibliography, London. U rry, W., 1967, Canterbury under the Angevins, London.