Period 2 | Period 3 | Period 4 | Period 1

There are, then, three principal and certain phases of development on the site of ‘House 1’ between the late 1st century and mid-3rd century AD when the last phase was demolished prior to the setting out of new property boundaries across Insula IX. We do not yet know sufficient about the period before the construction of the Period 2 buildings to be certain whether there was a yet earlier arrangement of structures on the ‘House 1’ footprint, or whether the fragments of buildings which we can glimpse, belonged to a different layout and extended beyond the limits of the assured ‘House 1’ footprint from c. AD 70-80.

Period 2


There are three elements to Period 2, a long central building, which we interpret as a residential town house, fronted by a verandah and looking out across a gravelled area to the south-east. On one side it was flanked by Building 1, located in the north-east corner of the insula, with a defining, tile-built hearth as its central feature, which we see as a service building, and, at the opposite end, by a circular structure. As we have discussed above, there is a possibility that the circular structure was encapsulated by a rectangular building fronted by a portico on the south-east facing side. Whether the circular building served a residential function is unclear. It was provided with a central hearth, but the presence of a number of ritual deposits, including the cremated remains of sacrificed sheep or goat, around the south-facing arc of the structure, also strongly suggests a cultic interpretation. This is not to deny that structured deposits are, perhaps, evident elsewhere. In Timber Building 1 there are the remains of the human neonates and also the pit containing the partial skeleton of a sheep, but this evidence does not dominate the archaeological record in the way that it does in the circular building. The cremated and fragmented caprine bone, for example, contrasts with the inhumed sheep burial in Building 1 and other deposits from other contemporary locations, for example the articulated remains of a young sheep burnt in a basket in a pit beneath late 1st-century AD timber Building 6 at Leadenhall Court, London (Milne and Wardle 1993, 41).

However, there is a particular issue in respect of Room 6 of our Building 2. Here the inclusion of pieces of Bath limestone, including carved, architectural fragments, in the foundations of the timber structure present difficulties in interpretation. They do not seem to have served a functional purpose which could not otherwise have been addressed by timber sill beams as used elsewhere in the building, but appear more to represent the deliberate incorporation of the remains of a prestigious building from elsewhere in the town. It is doubtful whether any of the stone would have been visible in the completed building. We also should not overlook the remains of animals and neonate infants, buried in the floor of Building 1. So, with Building 3, it would seem to be more the combination of circularity with the spatial concentration of votive deposits that sets it apart from the other two buildings. If this was a conscious reference back to a mode of building and associated style of living typical of later prehistory, the use of the fragments of carved Bath limestone hints at the reverse: a veneration or respect for a ‘modern’ and Roman attribute of 1st-century AD life, and perhaps the person or family associated with the demolished, Classical building in question.

Although we now have a number of towns in Britain where excavation has revealed the plan and sequence of structures throughout the Roman sequence, there is no parallel for the group of Period 2 timber buildings we have here in Insula IX at Silchester. A little more, however, can be said about the individual buildings. Examples of small, detached timber buildings comparable with our Building 1 can be found in London at Leadenhall Court on the site later occupied by the early 2nd-century Forum-Basilica. From Periods 3-4, c. 75-80–c. 95, parallels are provided by single-roomed Buildings 14 (5m by 4.8m), 16 (4.6m by 4.4m) and 21 (3.5m by 3.3m), each described as an outhouse, though none contained a hearth. The Period 4 Building 22 (c. 8m by c. 4m), again without hearth, comprised some three rooms (Milne and Wardle 1993, 45-7).

None of the individual Leadenhall parallels produced finds assemblages as rich and as varied as that from our Building 1, but the character of the finds assemblage from Leadenhall Periods 3-4 as a whole will be discussed further below.

In the case of our Building 2, we can, with little difficulty, find parallels in urban contexts for timber buildings comprising a single range of rooms (row-house), but, for the most part, they consist of structures (strip buildings) built end-on to the street. Building 2, with lengthwise corridor or verandah, faces south-east on to an open, gravelled area. The contemporary Period 2 (c. 75-155) timber-framed buildings from Insula XIV, Verulamium, with their single frontage on to Watling Street, appear to offer a parallel (Frere 1972, 23-98, figs 10-11, 15 and 19). However, behind the continuous covered walk parallel with the street, there is an irregular disposition of rooms which, over time, extend backwards from the street as a series of shops and shops-cum-workshops to a depth of four, five, or more rooms by Phase D (c. 150-55). A further contrast with the plan and symmetry of our Building 2 is the proliferation of hearths and ovens in the Insula XIV development. Similar, multi-roomed timber-framed buildings of late 1st and early 2nd-century date, which extend back from a common street frontage, can also be found in London, notably at Newgate Street and Watling Court (Perring et al. 1991, 3-44). A further sequence of early, timber-framed, rectangular buildings in London is found at Leadenhall, dating from c. 75 and sealed by the early 2nd-century construction of the forum basilica (Milne and Wardle 1993). These, too, as their disposition suggests, may also represent the remains of workshops constructed side-by-side and fronting end-on to an, as yet, undiscovered street. As at Newgate Street and Watling Court, there is a high incidence of hearths. The comparisons with the London and Verulamium buildings serve to emphasise particular differences, notably the apparent symmetrical arrangement of rooms along the long axis and the lack of hearths in the Silchester building, with at least one room having an opus signinum or mosaic floor. These attributes reinforce our interpretation of it as a relatively high-status building, probably residential in character. With the absence of finds clearly lost during the occupation of the building rather than associated with its construction, we lack the evidence with which we might ascertain the status of the residents. It is likely, however, that the first occupants were of a status comparable with that suggested for the building itself and, given that the latter probably did not survive more than about 50 years, the residents could have remained much the same throughout. Further work on finds from contemporary contexts from the immediate vicinity of the building may shed further light on the question of status.

In our search for parallels, Building 3 presents the greatest difficulty. While circular buildings can be paralleled in towns in Britain in the first centuries BC and AD, recorded examples are found at the beginning of the respective urban sequences. Silchester itself provides a good example, with circular structures beneath the Forum-Basilica in Period 1 from c. 25 BC (Fulford and Timby 2000, fig. 4). Similar incidences can be found in Canterbury from the late 1st century BC–c. AD 80 at the beginning of the sequence at the Marlowe Car Park site (Blockley et al. 1995, 32-39), and in London at Newgate Street in the earliest phase dating to the Neronian period (Perring et al. 1991, 3-5). The purposive construction of a circular building in the Flavian-Trajanic period, as we have here in Insula IX, and at least some 70 years after such structures were being replaced with rectangular buildings beneath the basilica, remains without parallel. Nevertheless, circular buildings were still a feature of settlement in the countryside and there may have been a conscious decision to introduce such an historic element in this urban context; perhaps reflecting the rural origins and familiar associations of its occupants. Notwithstanding the emphasis on structured deposition, this does not exclude the possibility of residential occupation of the circular building.

Our claim for high-status occupancy of Building 2 is based on the analysis of plan and comparison with other urban buildings of the same date. Except for the finds within Building 1, we can only use the equivalent assemblages from Buildings 2 and 3 with care, since so much of the material derives from floor layers and make-ups and may have been introduced during construction of the building and derived from earlier occupation. With Building 1 the finds’ assemblage is clearly contemporary with its occupation but whether it reflects the nature of the occupation of the adjacent ‘town house’, Building 2, which we think it served, or the lifestyle of its own occupants is unclear. However, finds of metalwork are rare, while Timby and Tyers comment on the relatively low proportion of samian, particularly of decorated vessels, and other fine wares as well as other imported ceramics, such as amphorae. There is a clear contrast with the contemporary assemblages from Leadenhall Court, London, which are relatively rich in metalwork and other ‘small’ or accessioned finds and imported wares of all kinds (Milne and Wardle 1993). That contrast is emphasised by the difference in the animal bone assemblages which, at Leadenhall Court, is dominated by cattle, an indicator of relatively high status and ‘romanisation’, with more or less equal representation of sheep and pig, whereas the three main domesticates are fairly evenly represented in the Period 2 assemblage from Insula IX. In fact the contrast is not so clear cut as pig, an indicator of high status, is significantly more abundant at Silchester than in London. Building 1 in Insula IX produced a large glass assemblage, but dominated by bottles and lacking the range of the assemblages recovered from Leadenhall Court. In material terms, then, the late 1st-century Silchester ‘town house’ and associated buildings does not appear as rich as the contemporary strip and other associated buildings from Leadenhall Court, London. To some extent this may be the result of a ‘London effect’, attributable to a combination of greater availability and higher wages, in that a wider survey of the proportions of samian and other imported fine wares shows ratios generally significantly higher than at Silchester, Insula IX (Ibid., 134-50). Indeed Going observed that, compared with assemblages of the same date from Chelmsford, a typical ‘small town’, even the lowest status sites in London contained a far higher proportion of all types of imported wares (1987, 117). Sheer availability of imported/high-status goods at the point of importation into the province may be the explanation of their relative abundance in London.

Period 3


Period 3, Insula IX, saw the replacement of timber Buildings 2 and 3 with two small masonry houses, of which one encompassed the entirety of the footprint of the underlying circular building, while the other extended over only part of the footprint of its predecessor. Building 1 seems to have been replaced by a larger structure, Timber Building 4. The gap between Masonry Building 1 and 2 was occupied by the remains of Room 6 of timber Building 2 and it is possible that that room survived in some form as a ‘link’ between the two houses. Both masonry buildings had three reception rooms but, while the plan of Masonry Building 1 can be compared with small row-houses common in both urban and rural settings, there is no close parallel for the square plan of Masonry Building 2 (cf. Perring 2002, 64-5, 73-4). These two masonry town houses are the first in Silchester to be dated by associated stratigraphy and their Hadrianic-Antonine date compares well with that of the earliest buildings in masonry at Verulamium, where timber-framed town houses also continued to be built from new in the middle years of the 2nd century (Frere 1983, 10, 236-41). Indeed the closest parallel from Verulamium for our Masonry Building 1 is the three-roomed, timber-framed XXVIII, 3A dated c. 130-50. There are numerous examples of small masonry town houses from Silchester, many of which are likely to be broadly contemporary with our two houses in Insula IX; and many of which are, similarly, not aligned with the orthogonal street grid (see town plan in the Introduction), and are likely to be early Roman in origin. Both buildings were badly robbed, such that nothing can be certain of the internal decoration of ERMB 2, but one room of ERMB 1 was decorated with a tessellated floor, the surviving fragment of the border probably framing a central mosaic. Although the floor make-ups only provide a terminus post quem, it would seem that metalworking took place at the north-east end of ERMB 1 in this phase. The plan of the enlarged Timber Building 4 is based on the extent of floor surfaces rather than clear wall lines, but it is interesting that its inclusion of two tiled hearths coincides with the adjacent construction of separate houses. Although there can be no certain proof, we assume that it continued to serve both town houses. The nature of its finds assemblage is not distinctly different in character from that of its predecessor.

While Crummy argues for a votive interpretation of certain copper-alloy brooches in association with Timber Buildings 2 and 3, the character of the votive offerings associated with construction in Period 3 shifts in emphasis away from the burial of animal remains to the deposition of complete pottery vessels and copper-alloy toilet sets. Although it is possible that the two vessels buried in corners of Masonry Building 2 relate to the previous period, we cannot be certain. They contrast with the two Alice Holt jars (one of which had been deliberately pierced) sealed beneath the clay floor associated with Timber Building 4, but not otherwise deliberately buried within a small pit. Parallels for the deposition of pots, their rims generally flush with the surface of the floor, beneath the floors of buildings can be paralleled at Silchester, for example in House 1, Insula XXVII (Hope 1902, 19-20) and elsewhere, notably Colchester (Crummy 1992, 57, 67, 69, 76-7, 104-5). There is one special deposit of animal bone, the burial of a roe deer's foot in the levelling deposits which mark the end of the life of Timber Building 2 and the start of construction of Masonry Building 1. There are also remains that may represent the partial skeleton of a dog which were associated with Timber Building 4. Other dog remains were associated with ERMB 1. On the basis of the burial of the distinctively ‘British’ toilet sets in the foundations of ERMB 2, Crummy argues that its builders, if not the occupants, were native British, rather than newcomers.

Period 4


Period 4, the third phase of construction on our ‘House 1’ plot, saw the total demolition of Masonry Building 2. The same may have been true of Masonry Building 1 though it may not have been so comprehensive. The new build embraced the footprint of both Period 3 buildings to produce a plan which, while possibly retaining two rooms of ERMB 1, introduced an entirely new, possibly undivided space over the site of ERMB 2. Metalworking remained a distinctive activity at the north-east end of ERMB 1. A possible reconstruction of the internal space would see, from the north-east end, a large space, occupying the full width of the building, devoted to metalworking, followed by the two rooms probably retained from ERMB 1; then a fourth rectangular room, occupying 50 per cent of the internal space of the building as a whole. A corridor or verandah ran the full length of the south-east facing side of the building. The date of construction cannot be closely defined, but is around 200, with demolition following in the second half of the 3rd century, probably soon after about 250. One probable votive deposit, the burial of the remains of a sheep, is certainly associated with this period.

The combination of the poor survival of this building, other than its newly constructed foundations, and the lack of a close parallel for its plan makes the interpretation of function and role very hard. Apart from the unambiguous evidence of metalworking, we might interpret the adjacent two rooms as evidence of a continuing residential function, but there is no clue as to the purpose of the largest internal space (Room 4). This uncertainty of interpretation echoes earlier remarks about the earlier, underlying ERMB 2 with its unparalleled plan. If, as the evidence suggests, there was a sacred element to the Period 2 circular building, was that continued, or reflected in any way in the succeeding buildings ERMB 2 and ERMB 3? Was Room 4 intended as a communal space for, say, a congregation of worshippers? There are finds from the sequence studied here which might give some support for this, such as the fragments of white pipe-clay statuettes, a fragment of ‘chimney’ (paralleled by the further fragments deposited by the Victorian excavators (Fulford and Clarke 2002, 299), and the piece of East Gaulish samian which Joanna Bird notes has imagery associated with Mithraism (Bird, forthcoming). Nina Crummy (pers comm.) has noticed other finds from the Insula ix excavation with religious connotations. There is not the scope to develop this theme here; the relevant material and its associations will be pursued in subsequent publication.

Period 1

While much remains to be explored of Period 1, not least to determine whether the buildings of this period occupy the same footprint as their successors, we can at least demonstrate the continuity of the development and re-development of a single property over some 175 years, representing perhaps seven generations, with each of the three phases of building lasting some two to three generations, about 50-75 years. Similar continuities can be demonstrated elsewhere in south-east Britain, notably in post-Boudican Colchester (e.g. Culver Street, Crummy 1992), London (e.g. Perring et al. 1991; and, for overview, Perring 1991); and Verulamium (Frere 1972; 1983). Though continuity was probably predominant until the 3rd century, there were exceptions, such as the abandonment, probably in the 2nd century of an early 2nd-century town house, R 11 in Canterbury (Blockley et al. 1995); and, in Verulamium, the break in continuity in the 2nd century over the site of house 2C in Insula XXI, (Frere 1983, 157-8), and the longer break, perhaps of a century or more, in occupation of the Watling street-frontage of Insula XIV after the fire of c. 155 (Frere 1972). The end of the ‘House 1’ sequence in the second half of the 3rd century has no close parallel in the Roman urban history of south-east Britain in the association of its abandonment and demolition with the laying out of new property boundaries in Insula IX. Whether the two events are coincidental or causally connected cannot be established (Fulford et al. 2006). Earlier, in London, in the later 2nd and early 3rd century, there had been considerable urban contraction (Perring 1991), while in Colchester, the abandonment of the Culver Street houses is dated to the later third and early fourth century (Crummy 1992). In neither of these latter cities is there evidence of associated, subsequent redevelopment and major re-alignment of property boundaries. Verulamium, by contrast, shows evidence of considerable continuity between the late 2nd (post the c. 155 fire) and 4th centuries (Frere 1983).


© Internet Archaeology URL:
Last updated: Wed Sept 12 2007