Three trial-trenches were excavated by Colchester Archaeological Trust - two by machine and one by hand - inside East Hill House and in its grounds, prior to the refurbishment of East Hill House as a hotel and the construction of retail and residential units to the west of this. <1> The trenches revealed parts of two Roman buildings, presumably town-houses (one of two phases), one of which had a hypocaust (Building 218 in T1) and the other at least one area of tessellated pavement (Building 219 in T3). These have been numbered Buildings 218 and 219 in the Colchester Buildings series. T1 was situated within a 20th-century extension at the west end of the mid 18th-century house. Two robber trenches clearly represent wall lines which formed part of a room in Building 218 incorporating a hypocaust (F2). The excavation of the robber trench F3 showed an earlier phase; the opus signinum wall foundations F5 and F6 indicate an earlier building on the same alignment. The impressions of what appear to be an upright box flue-tile on the surface of F5 suggests that this earlier building also had a hypocaust. Pottery associated with F5 suggests a 3rd-century date for the demolition of the earlier building. This earlier building included two rooms which were probably decorated with painted wall-plaster. A very thin and poorly preserved layer of wall-plaster fragments covered the floors of the two rooms with larger fragments at the base of F3. The wall plaster was mainly of a white colour with traces of red and yellow. Similar fragments were recovered from demolition debris L4 above the hypocaust sub-floor. The tessellated pavement in T3 was observed by Hull in 1951. <2> It was located 1.7m below modern ground-level at 24m AOD, with a probable robber trench following the line of a wall foundation next to it. This was within 100mm of the level of the top of the hypocaust sub-floor in T1, some 40m to the north-east and within 150mm of the of the level of the Roman layers in T2 to the north-west. Evidence of Roman activity in T2 was complicated. The presence of a large post-medieval pit made it difficult to interpret the two Roman layers at the base of the trench. L9 appeared to be Roman demolition debris, although less concentrated than L4 in T1 and T3. The start of a slight dip at the eastern edge of this layer could be indicative of a robber trench, but this did not seem likely. Beneath L9 was what appeared to be redeposited natural sands and gravel. This was very dirty and contained flecks of charcoal. It is possible that this was deposited as a consolidation layer over earlier features. To the north of the trench, the base (where not truncated by F4 or F7) was a layer of pale sandy clay (L11). Although devoid of finds, this did not resemble natural, but looked more like the remains of clay-block walls. However, only a small part of this context was exposed and excavated and so interpretation is problematic.