A two-part archaeological project at 21 St Peter’s Street, Colchester, carried out in advance of development by the Colchester Archaeological Trust (CAT) between November 2007 and February 2008 (referred to as Phases 2 and 3). Previously, a single evaluation trench was excavated in April 2006 (Phase 1). The evaluations (Phases 1 & 2) uncovered significant archaeological remains. This resulted in the (Phase 3) excavation of an area measuring 10m x 13.5m, which included the area of the trial-trench excavated in Phase 2. Fieldwork Phase 1. In April 2006 an exploratory trench (Trench 1) was hand excavated in the south eastern corner of the site adjacent to Ryegate House. The trench was excavated across the projected line of the Roman town wall in an area covered by a bank of earth partially covered by thin concrete. The trench was excavated to locate the north face of the town wall, positioned on a north-to-south alignment abutting a brick retaining wall at the southern end. Fieldwork Phase 2. Trench 2 was located near the western edge of the site on a north-south alignment. The projected line of the town wall in this area (based on its location in the car park of the neighbouring Claudius Court and in Trench 1) was covered with a concrete capping. The capping was constructed when the brick retaining wall collapsed in this area in the 1970’s resulting in the construction of a new concrete retaining wall. The concrete capping extended out 1.8m from the base of the retaining wall and due to its structural nature prevented excavation over the line of the town wall in this area. Trench 2 extended 16m from the concrete capping northwards towards St Peter’s Street and was excavated using a three ton 360 excavator equipped with a toothless bucket. Fieldwork Phase 3. A third phase of evaluation was undertaken consisting of two further trial trenches (Trench 3 and Trench 4). Trench 4 was 10m long, 1.5m wide and parallel with the neighbouring building Ryegate House. A 12-ton 360 excavator was used to excavate through the overburden until an archaeological horizon was reached at a depth of roughly two metres below modern ground level. As such, the uppermost significant archaeological layer in this area was below the contractor’s site formation level (6.45mOD) so the trench was backfilled. Subsequently it was decided that rather than excavate the other north-south trench in the centre of the excavation area (Trench 3), an east-west trench would be excavated starting at Trench 2 where the height of the uppermost archaeological layer was known to be above the contractor's site formation level. The archaeological layer was located between 600mm and 700mm below modern ground level in the west of the trench, sloping off notably to the east. Trench 3 would have revealed the uppermost archaeological layer at a height above the contractor’s formation level but at a depth that would have made excavation unsafe in a 1.5m wide trench. Following the findings of the evaluation trenches, an excavation area was defined based on the closest safe distance to the damaged brick retaining wall of Northgate Street in the south and the building Claudius Court in the west, and to the east and north by the point at which the uppermost archaeological layer was located at a depth below the contractor's formation level. Full excavation involved the mechanical removal of the overburden down to the uppermost archaeologically significant layer across the whole excavation area and the creation of a 45° soil batter around the extent of the excavation area where deep excavation had occurred. During Phase 2 of the project, an independent research project was undertaken to excavate under the foundation of the Roman town wall to ascertain whether a sample of original wooden piling could be obtained for tree ring (dendrochronological) dating. It was hoped that such a sample would assist in dating the construction of the town wall and help increase our knowledge of the development and phasing of the town within the Roman period. Once the overburden had been removed and the archaeological deposits uncovered had been recorded, a slot was excavated against the concrete capping that was presumed to cover the town wall. However, excavations ceased when an extremely well-preserved timber drain (F7) was discovered. The drain was situated within a fast flowing water inlet that filled the excavated slot within minutes making the search for the wall foundation unsustainable. A 1.9m-long stretch of the north-south aligned Roman timber drain was excavated and recorded at this stage. In Phase 3, an 11.5m length of the drain was exposed and details of its constructed were established. The drain lay below the contractor’s intended formation level and, because it was expected to survive the construction works, the extent of the area excavated around it was minimal. The limited nature of the subsequent excavation combined with the continual flow of water made distinguishing the stratigraphy in the immediate area of the drain problematic. The drain appeared to have been located in a gully cut into an alluvial sandy-gravel layer, which was left to infill naturally after the drain was constructed. Subsequent to the infilling of the construction trench for the drain, the layer of large septaria blocks and mortar mentioned above (L10) was deposited on, and in, the drain. As well as the large blocks of septaria, there were also some medium-large fragments of brick/tile and pot fragment(s). The building materials were in a wet cream/white sand and mortar mix with occasional small stones. This material is likely associated with the construction of the town wall. Two further timber drains were identified 10.5m to the east of the drain F7 on the same alignment.The more westerly of the drains (F43) was located at a height of 5.42mOD (base of drain) whereas the eastern drain (F42) was located higher up at a height of 5.78mOD, similar to the height of the base of F7 at 5.80mOD. The drains were located approximately 400mm apart at the western extent of the gravel that made up a metalled surface. Only short lengths of the two drains were uncovered as they were located below the contractor’s formation level. It is suggested that the drains flanked a N to S street (indicated by the metalled surface). The western drain was laid after AD 62 based on a dendrochronological date. The town wall was constructed AD 65-80. It is suggested the wall incorporated a gate, to explain the presence of the street, although no direct evidence was found for its presence. In the southern end of Trench 1, the remains of the septaria and mortar core of the wall were identified 0.25m below the modern ground surface and retaining wall at a height of 7.91mOD. What remained of the outer face of the town wall was located 1.72m from the southern edge of the trench at a height of 7.09mOD. The outer face appears to have been extensively demolished or robbed in this area. The wall foundation was located at a height of 6.23mOD extending around 0.22m from the face of the offset block. Overlying the areas of consolidation outlined above and concentrated between the eastern drains F42 and F43 and the western drain F7 was a surface comprised of successive layers of compacted gravels. There were five distinct layers of compacted ‘metalling’ identified in the surface separated by less compact sandy silts. Just over 7.5m north of the projected line of the town wall, the metalled surface layers were cut by the southern edge of what appeared to be of a wide and deep ditch (F5). Given its scale and position in relation to the town wall, the feature is likely to be part of the town ditch. Most of ditch was located below the contractor's formation level and as such it was proposed that the town ditch would define the northern boundary of the excavation area. Finds recovered from the ditch indicated it was Roman in date and included numerous Roman brick and tile fragments, animal bone, and copper-alloy slag. As happened at Balkerne Lane, the butt ends of the town ditch on either side of the street were joined up c AD 275-300 and the street effectively became a cul-de sac. It cannot be determined if the gate was blocked at this time. Six pits/post-holes assignable to the Roman period were identified cut into the western area of the gravel surface and its peripheral deposits (F14, F35, F21, F3, F4 and F25). Four deep features (F21, F14, F35 and F3) were cut to depths below 6.40mOD and contained wooden remains. The pits F14 and F35 both contained well-preserved remains of wooden logs. A layer of dark earth (L11, or L3 in the evaluation phase) directly overlay the Roman contexts on the site. The layer varied in thickness but was homogenous throughout, with no evidence of any depositional structure. The layer was organically rich and also contained frequent charcoal flecks. Finds from the layer were varied and ranged in date from the Roman to the post-medieval period. Finds included copper-alloy fragments, iron, worked stone, brick and tile (some peg tile) and slate. Roman pottery fragments recovered dated from the early-mid 2nd to 4th century (possibly late), and medieval pottery was recovered that is substantially a 12th- to 13th-century group. Sixteen squares (SQ1-SQ16) 25cm x 25cm in size, totalling a sample area of 1m², were carefully excavated in two 10cm deep spits. The Roman metalled surface did not extend to the southern limit of excavation in any of the excavated sections. Instead, two large post-Roman ditches, F44 and F9, were parallel with the town wall on an east-west alignment. The earliest of these, F44, was over 3m wide (the southern edge was located outside of the excavation area) with a flat base. The ditch had three distinct fills: a stony silt on its northern edge (L14), a highly organic silt similar to peat in the base of the feature (L13) and, overlying these, a layer of sandy silt containing frequent mortar and building material (F20). The later fill of the ditch (F20) contained peg tile. However, pottery fragments recovered from L13 and F20 date the initial infilling of this ditch to around the 13th-14th century. A second ditch was dug on the same alignment, with its northern edge 300mm south of the northern edge of F44. This ditch (F9) had a ‘U’ shaped profile with a flat base once again attributable to the solidity of the layer onto which it was excavated (L9). The ditch F9 had a soft loam fill with a high organic content. Late medieval pot and peg tile was recovered from the lower fill and post-medieval pot and slate from the upper fill. It was not possible to discern archaeological features cut into the dark earth (L11) that overlay the gravel surface, although it is likely many features were cut whilst this layer was accumulating: At the eastern limit of excavation a very large pit (F12) with a fill containing numerous inclusions and finds, cut the edge of the gravel layers. The bottom of the pit was not excavated as it was below the contractor's formation level. During the reduction of F12 within the excavation area, numerous finds including pot, tile, animal bone, glass, iron objects and a bone pin were recovered. The pottery from the pit was predominantly late Roman as were the small finds. However, fragments of late medieval/early post-medieval pot have been identified in the assemblage. This suggests that either the pit is late Roman in date with a later medieval/post-medieval feature cut into it that was not discernable due a similarity of the fills, or that it was a medieval/post-medieval pit that cut a late Roman feature subsuming the finds from this feature. A similar pattern was identified in the finds of another large, if relatively shallow, pit (F10). This pit was cut into the gravels and had been subsequently cut by F12. Finds included copper-alloy objects, pot counters, worked stone, a marble mosaic cube and a substantial quantity of late Roman pottery. However, two pieces of medieval pot and fragments of peg tile were also recovered. Due to the large area and shallow depth of F10, it is also possible that it may have been an erosion hollow in the gravel surface that filled with dark earth. Seven small pits or post-holes cut into the gravel surface (F7a, F8, F13, F16, F17, F18 & F28) are interpreted as post-Roman based on artefacts recovered from some of the features and the similarity of the dark fills in the features to the overlying dark earth (L11).