London EC4, excavation at Number 1 Poultry

The site overlies the west side of the Middle Walbrook valley in the heart of the Roman and medieval City of London, where a complex and waterlogged archaeological sequence up to 6 metres in depth had survived beneath the roadways of Bucklersbury and Pancras Lane and the shallow basements of 19th-century buildings. The Middle Walbrook is an area of outstanding archaeological survival and importance which was studied in detail by Prof W F Grimes whose work culminated in 1954 with the discovery of the Temple of Mithras on the east bank of the stream. Plans to construct an office building designed by the late Sir James Stirling at Poultry would result in the destruction of all of the archaeological strata in the construction of deep basements (an area of 3,400 square metres). Planning permission for the development had been granted in 1989, but was not conditional on full archaeological excavation. Although not bound by the provisions of PPG-16 (1990), the developer (Altstadtbau Ltd, representing Lord Peter Palumbo and Mr Dieter Bock) adopted its underlying principles and commissioned the Museum of London Archaeology Service (MoLAS) to carry out a desk-based Archaeological Assessment in December 1993. Evaluation included the logging of boreholes and excavation of 4 shafts in the basements of the Victorian buildings. Complex stratigraphy, including Roman and post-Roman buildings, and excellent environmental survival, was recorded and confirmed the presence of up to 10,000 cubic metres of archaeological deposits.

During May 1994, EH, Altstadtbau, MoLAS, and the Corporation of London discussed the archaeological issues raised by the development. Large-scale excavation could only take place within strict engineering and timetabling constraints, which meant that the main phases of archaeological work would have to be concurrent with construction. The ground floor slab of the new building could be used to brace the site perimeter, giving the structural rigidity needed for deep excavation, whilst also forming a physical barrier between archaeologists and overhead construction work. This approach (top-down construction) required extensive resequencing of work programmes, as well as provision of ventilation and lighting systems and the removal of spoil through 'moling holes' in the ground floor slab, but would allow construction of the superstructure of the Stirling building at the same time as archaeological excavation. In June 1994 agreement was reached between Altstadtbau, MoLAS, and EH. Altstadtbau agreed to make £2 million available to MoLAS for archaeological fieldwork, and provided substantial additional funding to cover attendance costs, whilst also modifying their construction programme. EH agreed to monitor the excavation work and to fund the analysis and publication of the results. The programme of archaeological excavation began in July of 1994 and was completed in June 1996.

The prehistoric topography of the Poultry site was dominated by a natural hillside which sloped down towards the Thames to the south and to the main channel of the Walbrook stream to the east. Meandering silt-filled palaeo-channels associated with the Walbrook and its tributaries were located and sampled at various points across the site. No evidence of pre-Roman occupation was found. The western side of the stream valley was cut by small streams fed by natural springs, with tributary streams to the north and south west creating a raised gravel 'spur' which extended part way across the site towards the main channel of the Walbrook.

The principal, earliest streets of Roman London formed a T-junction at Cornhill, linking the Thames crossing to the north bank with the main east-west road (Via Decumana) which was part of Watling Street, leading to Verulamium and the west. Having bridged the Walbrook, the Via Decumana crossed the south western part of the Poultry area. The location of the road corresponded to the natural gravel spur found on the western side of the site, which may have been an influential factor in the siting of the road and the subsequent development of the town plan. Extensive landscaping and drainage took place prior to road construction, which in its initial form was a narrow, poorly-metalled, gravel track. To the south of the road, lines of oak piles and post-and-plank revetments sealed by structured peat deposits and midden layers were associated with the consolidation of a wet area prior to its development. Dendrochronological samples from the sequence may help to establish a more precise date for the founding of Roman London, thought to be c AD 50.

Lines of oak piles consolidated the slope adjacent to the main channel of the Walbrook, which was prone to periodic flooding from the stream. Within a decade roadside timber buildings were established west of this marginal ground, along the Via Decumana. Although insubstantial, they were set out in an orderly manner, with all of the road frontage developed, indicating coherent town-planning and strong growth prior to AD 60. An offset road junction was established on the higher ground on the western part of the site at this time, with roads leading north westwards towards the (later) site of the fort and amphitheatre, and north towards the industrial area of the Upper Walbrook. Early buildings were also constructed along these roads, which, with the Via Decumana, formed the principal junction in the western half of the early Roman town. All of the early buildings were destroyed by fire in the Boudican revolt.

Rebuilding was rapid and saw the establishment of strip-buildings on roadside properties. Large oak beam box-structures on the easternmost properties created a platform for multi-room clay-and-timber buildings, suggesting that there may have been some public investment in commercial and residential building. In the years following AD 60, a 3rd road, running south from the Via Decumana towards the waterfront, was added to the road junction on the western part of the site, and parts of 5 Flavian insulae lay within the excavation area. Each insula contained complex sequences of clay-and-timber buildings, although the areas immediately adjacent to the Walbrook and at the back of Insula I were occupied by large water reservoirs and piled structures. A plank-lined water-tank built into the sloping ground next to the Walbrook was associated with a piping system of bored timber conduits connected by iron collars, and over 1,000 rotary quernstone fragments which had been reused to form an external cobbled surface may have come from a large mill or bakery nearby.

The roadside buildings constructed during the later 1st and early-2nd centuries contained a mix of residential, commercial, and industrial functions. Eavesdrips and small covered drains between some of the properties connected into large post-and-plank revetted roadside drains. The size of the main drains indicates that they carried the diverted tributaries of the Walbrook, and shows that the road system was part of an integrated plan for the development of the western half of the Roman city.

In the early 2nd century the roadside properties were again destroyed by fire, probably the 'Hadrianic' fire of c AD 125. After the fire, clay-and-timber buildings were re-established on some of the properties along the north side of the Via Decumana, but rebuilding was less vigorous than before. In the late 2nd or early-3rd century, stone buildings were built on 2 of the existing roadside properties in Insula I, but set back from the main road, perhaps behind contemporary timber structures. The westernmost of these buildings contained 4 room areas, and was substantially rebuilt on at least one occasion, with the addition of an apse and the conversion of another room to provide bathing facilities. Mosaic panels had been removed or salvaged prior to the building's demise in the late-4th or early-5th century. A more unusual stone building, perhaps a temple or hall, lay to the east. It was unheated, contained no internal divisions in its primary phase, and had a southern end which was either open or associated with a timber facade or portico. The centre of the building contained a mosaic floor with a complex geometric design which radiated out from an octagonal centre. At the north end of the building a smaller mosaic panel was set inside a tessellated border which was laid to an apsidal pattern, suggesting that a semi-circular internal structure had been removed in antiquity.

Another large late Roman masonry building lay to the south of the Via Decumana, and may have occupied all of Insula II. The outer wall incorporated a vaulted drainage system. Part of an internal wall lay to the south, but most of the complex, which may have been an important public building, lies off site. Many of the late Roman buildings were sealed by dumped deposits, but there was no evidence of 'dark-earth' soil formation, or of occupation between the 5th and 9th centuries.

Poultry once again became a focal point of development after the Alfredian re-establishment of the City in AD 886. A number of late Saxon sunken-floored buildings were recorded, including 1 built against the west wall of a Roman stone building, suggesting that some Roman ruins were extant. Generally, the development of the late Saxon topography of the area was similar to, and clearly influenced by, the preceding Roman layout. On the western part of the site a cobbled open area, perhaps a market, was established directly above the Roman street junction. A fenced hollow-way led to the area from the north west, following the alignment of the Roman streets. The surfaces of both the open area and approach road contained large amounts of butchered animal bone. Further west was an area dominated by late Saxon pits, some of which contained well-preserved shoes and ankle-boots.

Both Bucklersbury and Pancras Lane had been established by the early medieval period, perhaps evolving from a market space between Cheapside and the Walbrook. On the western edge of the site the small parish church of St Benet Sherehog was built on the footprint of the disused Via Decumana. The primary phase of the church, dated to c 1070, was built from reused Roman ragstone and tile, except for limestone quoins in Saxon-style long-and-short work. The church contained a primary mortar floor with a simple altar at the east end, evidence of internal timber structures, and a possible division between nave and chancel. Contemporary timber buildings and an alleyway lay to the south of the church. Ragstone foundations and chalk footings to the south and east of the original church were associated with its enlargement, and 15th-century glazed and decorated tile floors survived in part. One of the chalk footings contained a Purbeck marble headstone with a complete inscription, thought to date from the 12th century.

By the 12th century the south side of Poultry was lined by timber buildings, many of which contained evidence of metal-working. The area further south was open during this period, first as a large open area, and then as backyards to the buildings to the north, but was developed by the 13th century as larger properties were established. These included the property of the Merchants of Lucca, London's first financial trading-house, first documented in AD 1265. The Great Conduit, a vaulted underground cistern built in about AD 1236 and part of the first organised supply of fresh water to the medieval city, was discovered during work at the junction of Cheapside and Bucklersbury.

After the Great Fire the site of St Benet's was used as a subsidiary burial ground for a parish combined with St Stephen Walbrook, and over 250 post-Fire burials were excavated. Brick-built cellars, cess-pits, and wells dating from the 17th and 18th centuries were associated with properties which fronted onto Poultry and Bucklersbury. Evidence of trades included mercury waste from felt-making associated with hatters.

The Poultry field archive includes stratigraphic data on over 14,000 individual contexts, and a large photographic record. Over 1,250 oak timbers were recovered from Roman buildings and structures during the main phase of excavations, of which more than 400 are suitable for dendrochronological analysis. Over 1,100 bulk environmental samples were collected, and 5,000 accessioned artefacts have been recorded. Archaeological Assessment of the Poultry archive began in the late summer of 1996, and is scheduled for completion by the spring of 1997.