In advance of the installation of services to the new Visual Arts Facility (Firstsite), located in the grounds of East Hill House and on the site of the former bus station, test-pitting and a combination of machine-dug and hand-dug trenches revealed parts of a Roman building, presumably a town-house. The metalled street on the southern side of Insulas 31/32 was exposed in three places (its eastern end is now approximately 3m to the north of the earlier projection based on this evidence). In this new position, medieval Bastion 1 on the external face of the Roman town wall now aligns precisely with the northern side of the Roman street. An interesting group of finds was 1.07kg of unused Roman stone tesserae. These were found in a redeposited layer of Roman debris sandwiched between post-medieval topsoil layers, where they were presumably dumped during modern landscaping. These may indicate the presence of a workshop which produced or stored tessera for use in mosaic floors. Interpretation of some of the later strata on the site was complicated by the fact that there has been much movement of soil, mainly due to landscaping associated with the grounds of East Hill House. After the Roman period, there was no evidence of any activity until the 13th century; fragments of pottery may be associated with the medieval robbing of the Roman buildings (although there were no large or well-stratified medieval or later pottery groups). Finds of 15th- to 16th-century pottery may be associated with houses to the north which fronted onto the southern side of East Hill. Three clay tobacco-pipe ‘wasters’ indicate that a clay tobacco-pipe maker was at work nearby in the later 17th century. The kiln site could have been on the open ground south of the East Hill frontage, and the maker was presumably resident or tenant of one of the East Hill houses which were subsequently demolished for the construction of East Hill House in the mid 18th century. Later finds - both cut features and loose finds in soil layers - are associated with East Hill House. Of particular interest is a group of cess-pits in which sewage from East Hill House was disposed of until the later 19th century. The work was carried out in a number of stages, as follows: (1) in November and December 2006, twelve test-pits (TP1-TP12) were dug along the eastern side of the grounds of East Hill House, close to the western boundary wall of St James’ Church. (2) from December 2006 to March 2007, these test-pits were connected by Trench A, which was dug mechanically under archaeological supervision. (3) in May 2007, four new test-pits (TPB1-TPB2 and TPC1-TPC2) were dug in an approximately east to west line through the East Hill House grounds. (4) from October to December 2007, TPB1 and TPB2 were connected by Trench B, and TPC1 and TPC2 were connected by Trench C, both dug mechanically under archaeological supervision. (5) in June 2008, Trench D was dug in a south-easterly direction off the western end of Trench C, Trench E was dug in a southerly direction off the eastern end of Trench B, and Trench F was dug in a westerly direction off the southern end of Trench E, and to connect with the southern end of Trench A. The more northerly test-pits (TP2-TP5) were not dug deep enough (TP2-5 were c.1m deep) to penetrate topsoil cover or post-medieval horizons, whereas the more southerly test-pits (TP11/TP6, TP7, and especially TP8-TP10, and TP12) exposed significant archaeological horizons. TP1 was 2.2m deep, and a Roman street F2 was defined at c.1m below ground level. In TP6, an opus signinum floor F4 (cut by a post-medieval pit) was defined c.0.9m bgl, with the remains of a robbed-out wall and demolition (robbing) debris. An unmortared Roman flint wall foundation F25 was found in between TP 8 and 9, and a probable metalled street F9 in TP9. A Roman street F27 was exposed between TP10 and 12, and robbed out wall foundations in TP12. Roman demolition debris L38 was seen in TPC1, at 1.65m bgl. In Trench B, Medieval and Roman remains at the eastern end of the trench were at levels between 0.7mm and 0.85m bgl, and had been cut by the Victorian cess-pits. The principal remains were two robber trenches by which the wall foundations of a Roman building had been robbed out (F43, F57). A layer of make-up (F58) for an opus signinum floor (F42) was found in the room to the east of robber trench F57. F42 showed impressions of missing tesserae in its surface, and large quantities of loose tesserae found here indicate that this room at least had a tessellated floor. In the room to the west was a layer of redeposited Roman tesserae (F37, F38), which may indicate that tessera manufacture took place in the vicinity. F55 was a septaria rubble spread which may have been part of a floor base. In the western 13m of Trench C, a greater depth of stratification had survived; topsoil (L50) sealed demolition debris (L51) which sealed a Roman silty clay layer (L57), which sealed natural sand and gravel (L30, at 1.9m bgl); the easetrn 7m had been heavily landscaped. A burnt stump of daub wall (F67) was defined, probably of Boudican date and, therefore, the most easterly occurrence of evidence of Boudican destruction in the Roman town. In Trench D, two Roman horizons (L70 sealing L71) were defined. These were probably both demolition debris or dumped material used to make up the ground-level underneath floor surfaces which have now been robbed away. Four Roman wall foundation lines were identified as either robber trenches or surviving wall foundations in Trench E. Taken in conjunction with the robbed wall foundations and surviving footings seen in Trench B, the northern three wall foundation lines define a Roman town-house with at least four or ?five rooms, and the southern wall foundation line probably defines a footway with a discontinuous wall on its southern side (ie on the street). In Trench F, a metalled surface (L79) was defined, which is the surface of the Roman street on the southern side of Insulas 31/32.