11

The Urban Failure?

The theme of the decline and failure of towns is central to debate about the fourth century in Roman Britain. One traditional view presents the history of towns as progressive and unilinear until the collapse of Roman authority in the early fifth century. Despite the fifth-century crisis, people have sought to demonstrate threads of continuity between Roman urbanism and the re-emergence of the town in the later Anglo-Saxon period. However, the accumulation of archaeological data threw this into doubt long ago, leading to suggestions that the Roman town, as traditionally understood, had effectively failed by the end of the third century. Life in towns, it has been argued, went on into the fourth century, but on a different basis, perhaps more related to individual power than civil authority. One recent study has attempted to chart decline systematically, using a wide range of parameters. The conclusion is strongly in favour of early decline, but dates the change to the second half of the fourth century. A more median line between the two extremes of vitality or decline has also had advocates, admitting quantitative and qualitative change in urban life, but maintaining the view that it was still vigorous until well into the fourth century.

The extent of the changes in urban society is clearest in the archaeological record. There are virtually no fourth-century inscriptions from British towns, no sculptural artworks, no tombstones. These are trends that were already present in the earlier centuries, but far less pronounced then. Pagan monuments in towns were particularly vulnerable to the changing religious politics of the empire, which moved from the persecution of Christians to the persecution of pagans within the course of the fourth century. Unsurprisingly, pagan cults became less flamboyant and less conspicuous in the urban centres as the fourth century progressed, but there is not the evidence one might expect to show Christianity dominating the townscapes and providing a new focus for civic munificence, as it did in some other provinces. On the other hand, large stone town houses continued to be built well into the fourth century, there was a flourishing of mosaic art focused on a number of towns, and the province has yielded a number of late Roman silver plate hoards, signifying unexpected levels of personal wealth for the topmost tier in society. While some people might point to the early collapse of Rome’s imperial project in Britain, others can write of the fourth century as a ‘golden age’. The evidence, at first sight contradictory, can perhaps be best understood as the emergence of new forms of identity in response to combined political, economic and social changes. Regional differences between towns in the east and west of England now also increased.

TOWN DEFENCES

Defensive circuits and gates were important classes of monument in Romano-British towns. The diversion of funds into the construction of stone walls in the third century was both cause and symptom of more widespread change in late Roman towns. However, the creation of walls and embankments around urban settlements began far earlier in Britain than elsewhere, being datable to the first century in a number of cases, with numerous second-century examples, and the majority of urban sites, including many small towns, had defences by the end of the third century (Table 12). This is a very unusual pattern within the empire as a whole and it can reasonably be asserted that it reflects a particular insular dimension of urban identity.

Study of the defences of Romano-British towns has concentrated both on the accumulation of structural detail and on setting those data in a chronological and interpretational framework. Britain was unusual among the provinces of the western empire both in terms of the large number of towns that received defences before the end of the second century and the type of defensive circuits that were built. However, the older view that there was a virtual moratorium on the construction of urban defences elsewhere from the reign of Augustus to the mid-third century is no longer credible. Comparison with frontier regions of Africa, for instance, would suggest that defences may have been somewhat more common from the second century in areas with a perceived level of insecurity. On the other hand, in Gaul and Germany the provision of town walls before the third century was very clearly linked to urban status, with half the attested examples being coloniae and only eight examples (33 per cent) being civitas centres and with only two examples at small towns. The British examples dated before 200 are more broadly spread across the urban hierarchy and include London, the coloniae and virtually all the civitas centres and eighteen small towns. The British penchant for walls backed by earthworks also stands in marked contrast to the Continent, where free-standing walls are more common. This appears to have been an insular tradition, as does the provision of defences at so many small towns. There are other significant differences, notably that British defensive circuits generally encompassed a larger area of the core of the settlement than did the later Continental examples and did not make extensive use of spolia from earlier buildings (Lincoln’s lower town walls and London’s riverside walls being exceptions here). These partly reflect the long timescale of urban defences in Britain.

The archaeological evidence is conventionally placed within a framework that emphasizes five main phases of development (Table 13). There has been much debate concerning the coherence of these phases of construction and of the possible historical contexts to which they relate. On one side are those who believe that the erection of defences was so closely controlled by the state that the grouping of dated examples must reflect the rapid implementation of an imperial policy at a given moment in time. For the second-century earthworks, for instance, it has been argued that all must be of late second-century date on the basis of applying the latest dating for one example to the group as a whole. That context has traditionally been sought in the events of the civil war of 193–97, when Clodius Albinus is presumed to have prepared the towns of Britain for unrest as he made plans to take the major part of the garrison to contest the imperial throne. Some commentators now favour a date in the 180s, though yet others have objected to the idea that there was a single constructional context (whether proactive or reactive to a crisis). They propose that development could have taken place at different towns across much of the second century. There are inherent difficulties of dating defensive circuits in terms of the latest artefacts sealed beneath them. The underlying assumption made by those who see the construction of the earthworks as contemporaneous is that only some major security threat could have persuaded the Roman emperor to grant

Table 12. Urban defences in Roman Britain (featuring the major towns and the largest small towns only).

ANCIENT NAME

MODERN NAME

STATUS

AREA (HA)

1ST C

2ND C

3RD C

4TH C

Camulodunum

Colchester

colonia

47

E

S+E

 

 

Lindum

Lincoln (upper city)

colonia / 4p.c.

17

ME*

S/E

++

++

 

Lincoln (lower city)

 

23

 

ES

T

++

Glevum

Gloucester

colonia

19

ME*

S/E

 

BT

Eburacum

York

colonia / 3p.c.

40

 

 

S?

 

Londinium

London

?colonia / p.c.

128

 

S

 

BT++

Verulamium

St Albans

municipium

79

E

E

ES

BT

Isaurium Brigantum

Aldborough

civitas

22

 

E

S/E

BT

Petuaria Parisiorum

Brough-on-Humber

vicus/?civitas

6

 

E

S/E

BT

Venta Silurum

Caerwent

civitas

18

 

E

 

S/E, BT

Venta Icenorum

Caistor-by-Norwich

civitas

14

 

E?

ES

BT

Durnovernum Cantiacorum

Canterbury

civitas

52

 

 

ES

BT

Luguvalium Carvetiorum

Carlisle

civitas

28

 

 

S?

?

Moridunum Demetarum

Carmarthen

civitas

6

 

E

S?/E

 

Noviomagus Reginorum

Chichester

civitas

39

 

E

S/E

BT

Corinium Dobunnorum

Cirencester

civitas /4p.c.

88

 

E

S/E

BT

Durnovaria Durotrigum

Dorchester

civitas

33

 

E

S?/E

 

Isca Dumnoniorum

Exeter

civitas

36

ME

S?/E

S?/E

 

Ratae Corieltavorum

Leicester

civitas

48

 

 

ES

 

Calleva Atrebatum

Silchester

civitas

42

E

E

S/E

 

Venta Belgarum

Winchester

civitas

55

E

E

S/E

BT?

Viroconium Cornoviorum

Wroxeter

civitas

77

 

E

S?/E

S?/E

Durobrivae

Water Newton

vicus/?civitas

18

 

 

ES?

BT

–

Great Chesterford

?

15

 

 

 

S

Coria

Corbridge

?civitas

?15

 

 

 

 

Lactodurum

Towcester

?

10–11

 

E

S

 

–

Alchester

?

10.5

 

ES?

ES?

 

Durovigitum

Godmanchester

?

8/11

 

E

ES

BT

Lindinis

Ilchester

?civitas

10

 

E

 

S?

Durobrivae

Rochester

?vicus

9.4

 

E

S/E

 

Aquae Sulis

Bath

spa

9.3

 

E

S?/E

 

Magnis

Kenchester

?vicus

8.4

 

E

 

S?BT

–

Great Casterton

?

7

 

E

S/E

BT

–

Irchester

?

8

 

E

S?/E

 

Caesaromagus

Chelmsford

?

7

 

E

 

 

Cataractonium

Catterick

?

5–6

 

 

 

S

Cunetio

Mildenhall

?

6/8

 

E?

E?

S, BT

–

Dorchester (Oxon)

?

5.5

 

E

S/E

+

–

Brampton

?

6

 

E?

E?

 

Key to abbreviations

E = earthwork defence

ME = military earthwork defence reused

S = stone wall

ES = contemporary earth rampart and stone wall

S/E = wall cut into front of earlier rampart

T = towers added

BT = bastion towers added

p.c. = provincial capital

+ or ++ = evidence for other additions or significant modifications

Question marks indicate that the date of a feature is uncertain.

The * on early phases at Lincoln and Gloucester indicate that these circuits date to the very end of the first or early second century.

permission for such an unprecedented scheme. On the other hand, it could be argued that the impetus came from the towns and not from the state. It is possible that a general petition by the British towns for leave to construct defences was granted around the mid-second century in recognition of special conditions in this frontier province. Not all the towns would necessarily have been able to take advantage of such a privilege immediately, but civic pride as much as fear could have motivated them to imitate those towns that were able to set the trend.

Table 13. The main phases of construction of urban defences.

PHASE

MAIN CHARACTERISTICS

1

Earth and timber defences were allowed at only a few sites during the first century (e.g. St Albans, Silchester, Winchester).

2

The coloniae received walls in the later first century or early years of the second century, reflecting their prestige status, their quasi-military function and a desire to avoid a repeat of the Boudican sack of Colchester (Colchester, Gloucester, Lincoln).

3

Many towns gained earthwork defences during the second half of the second century (but, NB, London was endowed with a stone wall c.200).

4

During the third century, stone walls were added to the front of many earlier earthworks (or, in a few cases, erected contemporaneously with new earthwork circuits).

5

Bastion towers were added to the front of many wall circuits in the second half of the fourth century (and in London a riverside wall was constructed to supplement the land walls). Other evidence of strengthening or refurbishment of circuits. Broader ditches added outside walls.

The issue remains unresolved, but it is clear that by the end of the second century it was normal for the major British towns, and a good number of the small towns also, to have defences. Most scholars are agreed that the addition of masonry walls to the front of the earthwork (or as free-standing walls) need not be related to a single historical context in the third century, with some town walls being dated as early as the Severan age and others as late as 280. Similarly, although it has been the fashion to attribute the addition of bastion towers and strengthening of defences at a number of sites to the events following the intervention of Count Theodosius in Britain after the crisis years of 367–68, that too has been challenged (with some examples now dated to the mid-fourth century or earlier). The lesson to be learned is that, in the absence of unequivocal historical references or epigraphic evidence, there will be problems of fit between dating frameworks for defensive circuits based on the terminus post quem principle and the application of unitary historical explanations. Table 12 summarizes the current state of knowledge about the urban defences of the major towns and the largest of the small towns. The overlap in size and the similarities in phases of development for certain of the lesser sites are striking.

What were defences for? Answers have tended to concentrate on their military value in times of crisis. The Boudican revolt had seen the destruction of three towns lacking defences and this failure may have encouraged the more permissive attitude to their provision at a comparatively early date at other British towns. The literary sources suggest that the prime military threat in late Roman times came from small groups of seaborne raiders. Even in the absence of any significant numbers of soldiers stationed at a walled town, the existence of defences significantly elevated the risks for the attackers and may have prompted them to seek softer targets. But walled circuits provided more than simply emergency cover for the civil population. The threats of banditry were reduced and towns and their associated markets were easier places to police if gates could be shut at night.

Walls and gates could also have been symbols of civic status, competition and pride. At several towns it is clear that gates were designed as especially grandiose elements, with masonry structures being constructed in association with earthworks at St Albans and Cirencester, for instance, and the Balkerne Gate at Colchester incorporated an earlier monumental archway. Appearance more than defensive capacity was clearly a significant concern, and earthworks and walled circuits delineated the urban sphere and emphasized status.

The very large circuits at many sites, far exceeding the densely built-up core of towns such as St Albans, Wroxeter and London, would also suggest that defensive issues were secondary. Even the addition of projecting towers on the exterior of the walls in the fourth century may have been less defensive in intent that at first appears. At St Albans, such towers were only added to the stretch of wall on the south-east side visible on the approach from London. Any enemy considering an attack on the town would quickly work out that the rest of the circuit was far weaker, but the addition of a few towers provided a grand façade to greet, for instance, the governor visiting from London.

In summary, defensive circuits fronted by wide ditches presented a significant obstacle to small raiding parties, reinforced the sense of community of the main towns, facilitated policing and projected ideas about status. It is unlikely that the state put much of its own resources into the provision of urban defences, so towns would have had to find the means to fund them, to the detriment of other aspects of the civic building programme. Other provinces might have more magnificent public buildings, but the British towns may have felt compensated by their walls. The emergence of new forms of identity within towns may have also favoured the construction of boundaries, as a psychological reinforcement of the degrees of difference between urban and military communities, between townsfolk and countryfolk. Many towns possessed extensive suburbs outside their walls, of course, and the extent to which these were rendered non-urban as a result merits further investigation – for instance, there are suggestive differences in coin-loss patterns (based on comparisons of coins found in urban and suburban contexts). Defences around a core part of the urban centre were symbolic of urbanitas, a particular sort of lifestyle. For a farmer bringing produce to market or for a soldier moving along the provincial roads, the entrance to a town was marked by a physical transition from one identity realm to another.

The shape of the enclosed area may give us further clues to the underlying mentality behind their construction. Although almost all major British urban centres had regular rectilinear road grids, the defensive circuits often took on a polygonal form. The chief exceptions were the veteran coloniae where a rectangular form was favoured, even in extensions beyond the original line of legionary fortress defences that provided their initial basis. However, many of the civitas centres showed a preference for a different layout, based on multi-sided shapes or ovals. The third-century wall at Silchester bore an uncanny resemblance in shape and area to the long-buried late Iron Age inner earthwork. While emulating the higher-echelon towns in possessing defences, the adoption of a style of layout that evoked late Iron Age oppida reveals a less straightforwardly ‘Roman’ urban identity at play. The fact that many of the early urban circuits were earthworks, rather than masonry walls, further emphasizes possible links with memories of a British proto-urban past.

The fifty or so walled small towns are more of an enigma, since such sites were less likely to have had the resources to pay for their circuits; indeed many small towns never received walls and the majority of those that did appear to be late in date. In a number of cases, particularly with the smallest defended enclaves attached to roadside settlements, we may suspect the state to have taken action to protect some important function of the site (for example, a series along Watling Street may have been part of the public post system or have served as tax-collection points or grain depots for military supply). Some walled small towns could have marked the crossover points along the major highways between individual provinces within the late Roman Diocese of Britain. At the larger settlements, the potential towns and walled garrison settlements, the construction of defences may only have proved feasible for the community if the building programme was spread across a number of years.

CHANGE AND REORIENTATION

There were a number of reasons why towns shrank in size or diminished in vitality. In addition to broader economic trends, we should remember the role of natural disasters. In the Mediterranean, earthquakes were a major hazard, especially during a period of heightened seismic activity in the 360s. By contrast, in the timber-and-thatch towns of the northwestern provinces, fire was the greatest danger and, to judge from the archaeological evidence, it had a significant impact. The Boudican destruction of Colchester, London and St Albans set back the urban programme in Britain by a generation. Both St Albans and London also suffered major fire damage in the mid-second century, due in part to the large numbers of buildings that were of timber construction alongside the stone public buildings. At St Albans, this can be dated to 155–60 and affected about a third of the built-up area. In parts of the core area, rebuilding was a slow process and some plots were still empty 50–100 years later. There appears to have been a reduced level of economic activity after the fire, with fewer workshops than in the early Antonine phase. There are hints of a further major fire in the town in the late third or early fourth century, though the effects of this are less clearly mapped. London seems to have been even more prone to fire damage, with extensive destruction attributed to major conflagrations in the late Flavian, Hadrianic and Antonine periods and more localized incidents of third-and fourth-century date. The Hadrianic fire (c. 120–25) appears to have been the most extensive, although the forum and basilica complex was spared. Rebuilding on many sites was relatively quick, but less dense and with less evidence for commercial and manufacturing activity, suggesting that by the mid-second century the town was economically past its prime.

Other factors that could have affected the vitality of later Roman towns include the impact of plague and other contagions. There is historical evidence of plague reaching the western empire in the 160s and demographic models for the later empire continue to attribute a significant impact to cyclical outbreaks thereafter. However, there is no certainty that it ever reached Britain – for once, Britain’s isolation and dead-end status may have worked to its advantage – and if it did the effects were probably focused on major ports, such as London. In any event, towns in antiquity were unhealthy enough places without the incidence of plague. Especially when densely built up and populous, they were net consumers of people – that is, they could not sustain their population levels by natural reproduction, but had to attract additional migrants from the surrounding countryside or further afield. An important question to pose, therefore, is whether this balance was maintained in the fourth century and later, or whether a reduction in economic migration into towns reduced their long-term sustainability.

Some caution is required in interpreting decline in commercial and manufacturing activity at major towns. One obvious issue is whether the diminution in activity there was compensated for by growth elsewhere, most notably in the small towns, which have been interpreted as becoming more important economically in the late Roman period.

The late Roman civil administration of Britain is shadowy indeed and it is only because of the chance survival of the bureaucratic listings in the Notitia Dignitatum that we can sketch the outline. The Notitia provides us with the names of five provinces within the British Diocese (Maxima Caesariensis, Flavia Caesariensis, Britannia Prima, Britannia Secunda and Valentia). The first four of these names also appear in the earlier Verona List, dating to c. 312–14. It may well be that Maxima Caesariensis, which had a governor of consular rank, was centred on London, but we lack independent proof of that. There is an inscription from Cirencester that refers poetically to a rector or ruler of Prima Britannia, L. Septimius […], that is commonly accepted as evidence that Britannia Prima had its capital there. Maps of Roman Britain routinely shuffle the positions of the remaining two provinces, Britannia Secunda and Flavia Caesariensis, with proposed capitals being York and Lincoln. The truth is we do not know even these basic facts. The Notitia does supply the additional information that London was the base of the finance minister and keeper of the Privy Purse for the Diocese. There was an active mint in London at some points in the fourth century, reflecting its role as a regional treasury. It is thus probable that the Vicarius of Britain was also based there. A final piece of information in the Notitia concerns the location of a state weaving works (gynaeceum) at a town called Venta, though with no further clue as to which of the three British sites with this name was meant. What we can establish from these various snippets is that towns remained important elements of the late Roman provincial administration of Britain.

Excluding milestones, there are few unequivocally fourth-century stone inscriptions from Britain that shed light on the late towns. The Cirencester text of the praeses L. Septimius marked the re-dedication of a Jupiter column, and perhaps dates to the early fourth century (though some favour the brief pagan revival in the early 360s under Julian). About 150 Jupiter columns are known, mainly in north Gaul and Germany, and the monument clearly conflated Roman and northern European pagan practices. The Cirencester monument is the only certain example in Britain, and the fact that its dedicator was the governor raises questions about its wider relevance to British urban religious practice.

A series of building stones from Hadrian’s Wall attests the involvement in renovation work of contingents from the civitates of the Durotriges, Dumnonii and Catuvellauni, but the assignment of these texts to reconstruction post-368 is entirely hypothetical and they could be earlier in date. They do at least suggest that the civitates were still meaningful political units in late Roman times. Assuming that they are of third-or fourth-century date, they represent trans-provincial movement of labour corvées.

Given that the imperial bureaucracy was larger in the fourth century than in the first or second, the lack of civic epigraphic testimony is telling. The senior officials of the Roman state were not exhibiting the epigraphic habit in the same way as before and though similar trends can be observed in other provinces, the picture from Britain is particularly stark. In part the change reflects broader shifts in Roman society, which was increasingly dominated by people of humble origin from the European military provinces. The old-money aristocracies of the Mediterranean provinces were increasingly sidelined. The values of the new ruling order were somewhat different and thus their identity package was quite distinct. The decline in public inscriptions and honorific statues brought the western empire more into line with northern European trends. However, this was not an illiterate bureaucracy, as documents such as the Notitia or the imperial rescripts gathered in the Theodosian and Justinianic Law Codes demonstrate. It was the public display of literate skills that was now more muted and other aspects, notably dress, ornamentation and social ceremony, were given greater prominence. The tendency for civil officials to imitate the dress fashions of the late Roman army was another manifestation of the change in the ruling order and it is probable that some members of the provincial elite may likewise have adopted Germanic fashions of brooches and belt fittings.

Our image of a typical Roman town is focused on the impressive public buildings that lay at its core and that had involved a great deal of conspicuous investment in the second century in particular. The fact that at many towns we find these buildings falling into disrepair or ruin, or having a radically changed function, by the fourth century is thus often hailed as signalling the decline of town life more broadly. The evidence is clear in a number of specific cases, but the situation was not uniform in all towns. The forum/basilica complex was central to public administration in the early empire, so major changes in these buildings ought to be significant. Modern excavations at Silchester have demonstrated that the basilica was given over to metalworking by the end of the third century, though of a relatively well-ordered kind and not simply some sort of squatter activity. The basilica at Caerwent was largely demolished by the mid-fourth century, after which time metal-working hearths and furnaces were erected within the nave and aisle spaces. At Wroxeter, the forum and basilica were fire-damaged around 300 and not rebuilt, while at Leicester the basilica and the adjacent market both appear to have been damaged beyond repair by a fourth-century fire. The Gloucester forum ranges were demolished and replaced with a cobbled surface. London, as perhaps the foremost centre of late Roman administration, is even more striking. Its huge forum/basilica complex had been systematically demolished around 300. Exeter’s forum/basilica appears to have been renovated in the mid-fourth century at least, though it too was demolished at some point in the late fourth century to be succeeded by open ground and metalworking activity, then by a cemetery. That at Caistor-by-Norwich was rebuilt on a reduced scale in the late third century (though after lying derelict for perhaps fifty years following a major fire).

What is the significance of the loss or change in use of the prime civic spaces of the earlier townscapes? First, it is clear from the time that it took some towns to construct these amenities in the first place that local administration did not depend on their possession. The functions of civic administration and the local judiciary could be fulfilled without a forum/basilica complex. The extension of Roman citizenship in 212 brought larger numbers of people under Roman law, as opposed to local customary law, and will have increased the importance of the assize courts. One reason for the multiplication of governorships may have been to provide additional magistrates in a more standardized legal system. Local courts at the civitas centres may have been rendered less significant or obsolete in consequence. The loss of the basilica at London is harder to explain, but the second-century building may have been too large and expensive to maintain. London’s court (which must still have existed) may have relocated into a smaller building elsewhere within the town. In all these cases, though, we need to recognize that what followed on was not disorganized squatter occupation, but a regulated reuse of public space, whether for systematic metalworking, as open space, or as a quarry for building materials.

Other categories of civic buildings reveal similar patterns of changed use or abandonment in the fourth century. The theatre at St Albans appears to have been very dilapidated by the late third century and, despite signs of a final refurbishment around 300, within a short while the orchestra area was being used as a municipal rubbish tip. The close association between theatres and temples (and by implication with pagan festivals) could account for the progressive loss of this sort of facility during the increasingly Christian-dominated fourth century. The same pattern seems in general to hold for amphitheatres, as at London (also redundant c. 300), reflecting changing attitudes to public bloodletting as sport. Again, the loss of such public amenities must be offset against the systematic use of the buildings as rubbish dumps, for instance, implying a continuing administrative authority.

Of fifteen major public bath buildings, nine were still operational c. 300, but none a century later. Large bath buildings were very expensive to maintain and run and the progressive loss of such facilities in towns may initially at least reflect a more selfish investment by the elite in smaller-scale bath-suites in their town houses and on their villa estates, rather than an outright rejection of Roman-style bathing. The meticulous excavation of the baths basilica site at Wroxeter has revealed evidence of continued orderly use of the structure long after its original function ceased in the early fourth century. Even when the roof was removed, the floor was carefully repaved with reused stone roof tiles, perhaps as an open market space. Regulation of public space by an urban authority appears to have continued here well into the fifth century, before the site was finally taken over by what looks like an individual remodelling of the area around a timber hall.

The late Roman/late antique town was thus very different from the high imperial town, with less need of public building complexes, reduced investment by the elite in the maintenance of amenities such as major baths and in entertainment buildings that were often linked to pagan festivals, such as theatres, or to practices that were abhorrent to Christians, such as amphitheatre games. Nonetheless, the early onset and the extent of change within the British towns are striking.

Analysis of the occupation pattern in 1,400 domestic structures in Romano-British towns suggests a sharp decline by the later fourth century. There are of course various factors that could contribute to this apparent collapse of occupation in towns, not least the vulnerability of the latest archaeological phases to subsequent disturbance or truncation. The virtual cessation of manufacturing and marketing networks by the early fifth century also limits severely our ability to date later activity at these sites, but the pottery supply and coin lists should be stronger for the later fourth century than they are. Compilation of data on the fills of rubbish pits and wells from Roman London, for instance, has revealed a dramatic fall in the volume of material after 150; the late Roman period, although showing partial recovery from a nadir in the period 150–250, is notably low in comparison to the early second-century town. This is relatively typical of towns in eastern England, some in the west seeming to have fared better, as at Cirencester.

In terms both of the aggregate figures for occupancy of domestic buildings in general and of the detailed structural evidence of housing at specific sites, the data indicate that occupation levels in Romano-British towns remained quite high into the early fourth century, but showed significant reduction after 350. On the other hand, large and well-appointed town houses were far more a feature of the late Roman town in Britain than they had been of the early phases of urbanization. These were rare before c. 150 and the maximum levels of occupancy occurred between 200 and 350, with an overall peak c. 300. The essence of this analysis is that the numbers of elite houses occupied plummeted after 350 in line with the overall pattern for domestic structures, so that only about one tenth the number remained in 400 as had existed in 300. Architecturally, the elite urban house followed Roman fashions in design and decoration. Some of these were winged corridor houses similar to the classic rural villas typical of the northern provinces, but others were courtyard houses of more Mediterranean aspect. Interior decoration commonly comprised mosaics and painted wall-plaster, generally closely following Mediterranean models. The aisled-hall house, frequently encountered in rural villas and perhaps more evocative of British traditions, is relatively rare among the urban elite residences. In broad terms, then, the trend in elite urban housing reflects the consolidation over time of an increasingly ‘Roman’ private identity that is at variance with the evidence for the loss of significance of civic buildings and public display. At least down to the mid-fourth century, most towns had the personnel to fulfil the required functions of urban-based local government.

Some caution is needed about the generality of the late Roman reduction in the density of urban activity. The long-held view of Silchester as a garden town, with low-density stone housing, stands corrected by recent excavations in insula IX that have shown a previously unrecognized density of timber buildings along the road frontages, between the known stone buildings. Many of these commercial properties continued in occupation down to the end of the fourth century. The relative lack of later disturbance to the latest Roman levels and the comparatively large open-area excavations have been important factors in achieving this new vision. At present the site is exceptional in the way it appears to buck the trend of economic decline.

Individual buildings evidently remained in occupation until a very late date, but the extent to which these were exceptional survivals is controversial. The key evidence comes from excavation of insula XXVII.2 at St Albans. Here, it is argued, a substantial courtyard mansion was constructed in the last quarter of the fourth century, with over twenty rooms, many with tessellated or mosaic floors and one room with underfloor heating. Subsequent modification and rebuildings on the same site appear to demonstrate the vitality of St Albans well into the fifth century. However, the structure is thus far unique for both St Albans and Britain at this date. It cannot serve as the basis of a general model of urban vitality and continuity into the fifth century. Indeed a recent reconsideration of the evidence has suggested that the initial construction of the house could be considerably earlier than the original excavator suggested and more in line with the normal pattern of late town houses.

The existence of rural types of building within the late towns is well established and some clearly retained their farming function in relation to cultivated land close at hand, as in a complex on the fringe of Cirencester. Coupled with the apparent decline in commercial and manufacturing activity at major towns, this has prompted some to question whether the balance of activity at towns was shifting away from traditional ‘urban’ lifeways towards rural ones.

Another major point of debate about the state of the late Roman towns of Britain concerns the interpretation of deep deposits of dark earth that have been recorded at a number of sites overlying the rubble marking Roman abandonment horizons – examples include London (and Southwark), St Albans, Canterbury, Lincoln, Gloucester and Winchester. No surfaces or structures have been identified within the dark earth layers, but they do contain Roman artefacts. One traditional interpretation of these soil layers is that they represent localized areas of the townscape that became devoid of buildings in the fourth or early fifth century and were given over to cultivation. The discussion about the nature of the dark earth is still unresolved, with some experts claiming the soil to be certainly imported and dumped to enable cultivation, others that the build-up of organic-rich layers could be the result of new patterns of rubbish disposal within abandoned areas of the town, yet others that it may represent the biological transformation of in situ occupation levels through the action of a particular type of worm, in combination with an increased use of simple mud walls for urban houses. The first two interpretations support the general view of overall decline in urban function and population levels, the last alternative implies that late Roman occupation was a good deal more dynamic than indicated by the remains of stone-walled houses. All the interpretations attest something other than passive decay within largely abandoned sites. If building plots had been abandoned, they were put to new use, whether as allotments or for rubbish disposal.

Public buildings may have been a dispensable luxury, but there were other civic services that were perhaps less so. Maintenance of streets, water supply and cemeteries provides complementary evidence of the relative vitality of Romano-British towns. Much has been made of the apparent functioning of a piped water system at St Albans well into the fifth century, implying the continued maintenance of an aqueduct. Overall, the evidence from towns in Britain suggests that rainwater collection, wells and springs may have been more important sources of potable water than expensive long-range transport of water by aqueducts. London and Southwark have yielded close to 100 wells and the extensively excavated sites of Silchester and Caerwent 76 and 16 wells respectively. Some of these were undoubtedly still functioning in the fourth century. However, the detailed evidence from London shows a decline in the overall number of active wells and the migration of elite housing across the townscape from areas previously provided by wells or public piped water supply to the area of the lower Walbrook, where an active springline still functioned. The implication is that with an increasingly less reliable public system of supply, the wealthier inhabitants chose to relocate their houses to areas where they could most easily tap groundwater supplies.

Once again, the most significant change may have occurred in the later fourth century. One reason for the final failure of public bath complexes could have been problems with maintaining an aqueduct-fed water supply. There are indications of blocked sewers at York and a number of other sites. Maintenance of roads (especially resurfacing) was comparatively rare in the later fourth century and at Gloucester one road near the forum was not only abandoned but was being actively quarried by the 390s.

Another clear indicator of the state of urban services concerns rubbish disposal. Urban societies produce a lot of waste, both organic and inorganic, and need to evolve strategies to dispose of this or suffer the consequences of unhygienic conditions, smell, rodents, and so on. Extraction or burial of rubbish from the core of an urban site would be the expected response of an orderly society. Removal to external dumps by communities required transport animals and vehicles, labour to collect and carry it and land outside the town on which to dispose of it. There is evidence that some Roman towns elsewhere in the empire made such provision, but in general it is probable that burial was the more common response. Some major developments in British towns offered unprecedented opportunities to dispose of accumulating rubbish – as the mass of material deposited behind the timber riverfront wharves or the partial infilling of the Walbrook valley in London both demonstrate.

Repeated references in the modern literature to unburied rubbish and increased urban squalor would seem to herald significantly declining standards. On the other hand the evidence is not entirely straightforward. The final Roman phases in British towns stand out as a period of ruins and rubbish because of what followed them, whereas earlier periods of activity were often capped by further development. Much rubbish in Roman towns was incorporated into each successive act of construction – thus material from a building destroyed in a fire in the mid-second century might be incorporated in the foundations of the next building on the site. But what if, as seems clear in a number of instances, the site had stood derelict for fifty years before redevelopment? A vacant lot might serve as an open rubbish heap for a considerable period, before the next builder on the site faced up to the problem of where to put the accumulated material. Early Roman towns may thus occasionally have been rather squalid environments.

Nonetheless, a reduction in building activity and progressive abandonment of buildings in the later fourth century clearly increased the amount of unburied rubbish. It is human nature not to wish to carry rubbish further than necessary and in the case of night soil the nearest street was evidently far enough. Bearing in mind that the use of animals for transport of goods and people will have introduced a good deal of ordure on to the average urban street, the tolerance of human waste being added to the mix is perhaps a little less surprising. More bulky items could be disposed of in a variety of ways, but recourse to ‘fly-tipping’ rather than digging a big pit in the back yard is a typically selfish human response. We can trace this sort of activity in empty houses and building lots (St Albans), accessible sections of urban defensive ditches (London) and so on. Private dumping in town ditches must have been officially discouraged, but if no civic rubbish collection was practised, how could the urban authorities regulate the accumulation of rubbish within the town? One possibility is that they could attempt to direct rubbish disposal towards selected locations, such as public buildings that no longer served their original function. The late fourth-century infilling of the orchestra of the theatre at St Albans with domestic rubbish may have been a pragmatic use of the space and, rather than suggesting increasingly anarchic conditions, it may represent a continuation of urban authority. If the urban population shrank considerably in the later fourth century, then the scale of the problem of rubbish disposal was also correspondingly reduced in comparison to earlier urban phases. In sum, there is no doubt that, in common with most pre-industrial societies, the late Roman town was dirty and smelly, but whether this was actually worse than the conditions in earlier centuries is far from certain.

Cemeteries are potentially an excellent measure of the vitality of urban life. Until quite recently there were few well-published, large-scale excavations of Romano-British cemeteries, making generalization hazardous. This picture is rapidly changing and one area of archaeology where we can now detect a pronounced expansion of data in the fourth century concerns cemetery data.

Late Roman funerary practices were diverse, in part aligned with trends in other north-western provinces, in part providing dim echoes of traditional insular preoccupations. As in much of the western empire, cremation was gradually replaced by inhumation in the late second to third centuries as the predominant rite. Though a number of cemeteries have yielded late third-and fourth-century cremations, these are rare at major urban sites and more representative of the military, minor settlements and rural communities. Many British towns have produced evidence for the emergence in the fourth century of large, managed inhumation cemeteries outside their walls, in areas distinct from earlier cremation grounds. It is equally clear that these cemeteries did not continue far into the fifth century, but, as with other key elements of urban life, fundamental practices and behaviours ceased.

Poundbury by Dorchester, Lankhills by Winchester and Bath Gate by Cirencester all continued in use as organized burial grounds to the late fourth or early fifth century at the latest. The Newarke Street cemetery at Leicester appears to be predominantly a later fourth-century development and, like Poundbury, features Christian burial traditions. A number of small towns have also been shown to have had extensive fourth-century inhumation cemeteries (Ilchester, Dorchester-on-Thames), but some sites have revealed a pattern of individual burials (or small groups) close to the edges of defined ‘backlands’ plots, standing behind individual buildings on the road frontage (as noted at Shepton Mallet, Ilchester, Alcester). Similar features have even been noted in suburban contexts outside some of the major towns, as at Leicester. There was clearly a strong distinction between traditions of interment in a public cemetery outside the settlement and in private ‘familial’ plots in close proximity to domestic structures. The latter tradition may well be a more rural practice and its appearance at towns in the later fourth century is another indication that at some sites there may have been a blurring of old distinctions.

While the cemeteries of Roman London were largely engulfed by the rapid nineteenth-century expansion of the city, the existence of substantial burial grounds has been demonstrated on the west, north and east sides of the walled Roman town and across the river on the southern flank of the Southwark suburb. In recent decades, a sample comprising c. 300 cremations and 1,000 inhumations has been excavated from several of the main cemetery areas. In general, cemetery areas moved outwards from the town, as expansion occurred from the first century onwards. In consequence, some early burials have been found within the area that was later walled; conversely, at the unwalled Southwark settlement, late Roman contraction of the occupied area saw late burials cut through earlier buildings. Some inhumations were in wooden coffins, while in many cases only the presence of a shroud can be surmised. There are a few rare examples of lead-lined coffins (some with elaborately decorated lids) and some lead and timber coffins featured chalk packed around the body (a practice similar to gypsum burials recorded at York and suggested as a Christian preference). A small number of stone sarcophagi have been recorded and sometimes these contained lead inserts, as in the case of a rich female burial from Spitalfields.

In extensively excavated inhumation cemeteries, the majority of sexed skeletons were adult males. The most extensively excavated area of London’s eastern cemetery had a male: female ratio of 1.7:1, only 25 per cent of skeletons were of children/adolescents, and these figures are typical. Given the probable level of infant mortality in the Roman world (perhaps 50 per cent died before age ten), children should proportionally outnumber adults in cemeteries. The explanation of these apparent demographic imbalances could relate to social practice (an adult male was more often granted the expense of an individual burial), or alternatively in different strategies of, and separate areas designated for, disposal for women and children, or it could reflect some genuine peculiarity of late Roman urban society, with a preponderance of males due either to greater rates of migration of men from countryside to town or to the impact of female infanticide. Rural cemeteries and the Poundbury data show a far closer balance in men to women in the population at large, so the answer is more likely to relate to social practices of disposal of the dead at some urban centres. The women and children who made it into these cemeteries may have been disproportionately representative of selected elements of society.

Perceptions about status and identity have been one of the chief motors for funerary ritual and practice. Status can be reflected in several different ways: through the funeral process itself, the treatment of the body, the location of burial, the provision of material goods for the deceased, the construction of the grave and surface marking of the spot. In many societies, expensive ceremonies and rituals mark death and burial as a means of both paying respect to the deceased and reflecting the social status (real or projected) of the family and heirs. While the bulk of the urban population does not seem to have embraced epigraphic commemoration, and high-status masonry above-ground structures are uncommon, the comparative rarity of intercutting of graves suggests that the location of graves was generally visible (wooden markers or low mounds?) and remembered for some time. High-status burials at towns represent a small minority of the interments and seem to confirm that even in late Roman times the urban elite remained a restricted group in society. They may also have used both urban cemeteries and rural burial grounds associated with villas.

Whereas cremation burials of the early empire were frequently accompanied by grave goods, either placed on the pyre with the body, or directly in the grave with all or part of the pyre debris, the late Roman burial rite marked a distinct change in fashion. Such items occur in only a minority of inhumation burials and Christian influence seems to have exacerbated a trend that was already present in pagan cemeteries. Where there were inclusions in late Roman burials these are almost invariably of great interest. Patterns of deposition suggest some distinct regional peculiarities, though with broad similarities in status-marked burials at both urban and rural sites. Shoes appear to have been a frequent inclusion (most easily identified when of the hob-nailed variety) and more often accompanied males in Essex, but with a more equal male to female distribution of shoes/boots in burials in Dorset and Somerset. The growing body of evidence reveals huge complexity and diversity. A key distinction to be made concerns whether the personal artefacts were worn by the body interred, or whether these items were simply placed alongside a shroud-wrapped corpse. If we see a clothed body adorned with personal ornament and accompanied by grave offerings, such as pots or glass vessels, as essentially non-Christian, it is apparent that some parts of the country demonstrated continuing pagan identities more strongly than others. The West Country from Dorset through Somerset, Avon and Gloucestershire stands out, for instance, as a region of strong pagan tradition. The similarity between burial rites in town and country suggests that the primary identity markers being displayed were ones of religious and social status. To the extent that ‘urban’ and ‘rural’ identities were expressed in funerary ritual, it was the location and internal organization of the cemetery that conveyed the clearest sense of difference.

One of the most interesting groups of material in burials concerns new styles of belts, indicative of official status and on the Continent generally closely associated with the field army and officials. These sometimes occur in association with the crossbow brooch, another high-status item. The later fourth-century cemetery at Lankhills, Winchester, has yielded a clearly defined group of these unusually furnished burials (also including a female equivalent with bracelets and other jewellery). Links can be drawn with a pair of male and female graves at Dyke Hills, Dorchester-on-Thames. The closest parallels for these graves in terms of rituals and personal adornments are from the Rhine frontier region and the possibility that they mark the presence of Germans has been much canvassed, though with no agreement as to whether these people were elements of the field army, irregular levies from an allied people (foederati), other imperial officials (not necessarily Germans) or simply members of the provincial elite adopting the trappings of the powerful. The likelihood is that this group was expressing a strongly held sense of identity that differentiated them from their fellows in the late Roman towns and they may well have been German. An involvement in late Roman government is also highly probable. Further examples of these distinctive late Roman belts, brooches and accoutrements come from other towns in southern Britain, including London, Water Newton and St Albans.

Some of the late Romano-British burial rites conform to Continental practices, but others suggest more insular traditions. A peculiarity of many British cemeteries in the civil zone is the inclusion of individuals or groups of decapitated people. These burials are not attested at military settlements, or at Colchester and Lincoln. Exceptional examples have been noted at London, Gloucester and a number of other major towns (Chichester, Winchester, Dorchester, Cirencester, Leicester, York), but the type is more common at small towns and rural sites.

Another key issue with late Roman cemeteries is how to distinguish between pagan and Christian burials, especially in the early phases of Christianity, when practice was hybrid, showed continued adherence to some earlier pagan traditions, and when burial grounds were less segregated on religious lines. A number of factors have been identified as potentially indicative of Christian practice: west–east orientation, supine posture, lack of or few grave goods, clustering around special enclosures/mausolea, and plaster or gypsum packing around selected special burials. When found in combination there does seem a higher likelihood in the interpretation. One of the most convincing cases of a Christian cemetery is at Poundbury (Dorchester), which provides the largest single sample of excavated burials. The burial ground is made up of a series of linked enclosures that may have been in broadly contemporary use, and analysis suggests that some areas of the cemetery were possibly pagan, alongside a core area with strong Christian associations. This latter area was characterized by west–east aligned burials, with comparatively few grave goods, but with other indications of higher-status burials being present (lead coffins, plaster burials and a few mausolea with painted decoration).

Burials inside the town limits were proscribed by Roman law and, with the exception of infant burials which are not uncommon beneath buildings and the odd apparent murder pit, British towns seem to have followed this pattern until near the end of their life. Exceptions appear in the fifth century, in the form of a multiple grave from Canterbury, or burials within the fora at Exeter and Lincoln (the last example associated with the early church beneath St Paul in the Bail).

The fourth century was marked by cataclysmic changes in religious practice across the empire – commencing with a wave of savage persecutions of Christians and ending with the situation reversed and paganism proscribed. A key step was the outlawing in 341 of pagan temples within urban limits, though the archaeological evidence does not support quite so abrupt a cessation of pagan activity within towns. On the other hand, one of London’s main classical temple precincts was replaced by a complex interpreted as a palace of the usurper Allectus (dated by dendrochronology to 293–94), and the temple at the small town of Godmanchester was demolished in the late third century. However, some pagan structures endured well into the fourth century, as at Bath and the London mithraeum (though the latter was perhaps rededicated at some point in the fourth century to Bacchus). Coin dedications at Bath had diminished considerably by the mid-fourth century and ceased by c. 390. In fact an overall survey of the dated examples of Romano-Celtic temples (both urban and rural) reveals that numbers peaked in the early fourth century, in much the same way as did town houses, mosaics, villas and so on. Late fourth-century repressive measures taken against pagan cults emphasize the fact that in some areas paganism remained strong.

More surprising perhaps is the relative lack of evidence of urban churches in Britain, although across the Roman world churches in the urban core are relatively rarely documented until the late fourth century. A large basilica partially excavated near the Tower may conceivably represent the late fourth-century cathedral of London, though a secular use of the building cannot be excluded. A small timber church was inserted into the centre of the forum piazza at Lincoln, beneath the later church of St Paul in the Bail, and may date to the late Roman period. Given that there were at least three British bishoprics at the time of the Council of Arles in 314 (London, York and Lincoln), there ought to have been churches at these centres dating to the first half of the fourth century. Other possible Christian churches remain poorly published (Icklingham), controversial (Silchester) or unexcavated (Wroxeter).

British bishops do not seem to have proliferated in the same numbers as those from other provinces (or they did not attend later councils en masse) and this suggests that the impact of Christianity within the towns was a qualified success. The three bishops and a priest who attended the Council at Arles in 314 presumably represented the four provincial capitals. Three British bishops at a Church Council at Rimini in 360 had only been able to attend through the support of state funds for their travel expenses, because of a lack of private support. This might suggest an impoverished and small British Church, but there are suggestions of greater wealth and vitality. For example, there is some evidence of the activity of bishops beyond the three or four main towns and of the accumulation of wealth within the Church. A lost silver dish of probable British manufacture from Risley Park in Derbyshire refers to its presentation to a church by Bishop Exuperius. Fragments of lead pans used in salt production from Cheshire mention both a Bishop Viventius and another clergyman. The involvement of the Church in manufacturing activity is paralleled in other regions. The location of the see of Viventius is uncertain, though the recognition of a probable church at Wroxeter suggests one possibility. Lead tanks with Christian symbols have been found at several sites in eastern England and may have served as baptismal fonts.

A major hoard of late fourth-century Christian silverware found at the small town of Durobrivae (Water Newton) included four inscribed silver vessels out of a total of nine recovered, an inscribed gold roundel and nine inscribed silver votive plaques out of seventeen examples in the hoard. The inscriptions on two of the vessels indicate that they had been specifically presented to a church as offerings, one by Innocentia and Viventia, the other by Publianus. The dedication of the latter refers explicitly to ‘prostrating myself, Lord, I honour your sacred sanctuary’. The plaques are similar in design to the leaf-shaped votives of pagan tradition employed at shrines in southern Britain in earlier times. One of the votive plaques bears, in addition to a large chi-rho, the inscription Iamcilla or Amcilla votum quod promisit conplevit, perhaps best rendered ‘The female servant of the lord (ancilla) has fulfilled the vow which she promised.’ This is simply a Christian expression of a long-established form of worship as we have seen in earlier chapters.

The urban story of Britain in the Roman empire is one of retarded development and premature contraction. The explanation of the changing nature of towns is in part an economic one, but it is in part also socially contingent. It is a paradox that the towns provide some of the most striking evidence of physical change under Roman rule, often on a monumental scale. Yet, although urban centres played a crucial role in shaping political and economic development, and in determining the material culture, social outlook and identities of a significant group of the population, Roman urbanism did not endure. This can only partly be blamed on what followed after. The roots of the urban culture propagated by Rome were too shallow and the social bedding in which it was planted not fertile enough. The evidence presented here suggests that not only were the late Roman towns very different in appearance and outlook from their early Roman precursors, but that they experienced profound change and decline in the period after 350. Well before the notional ‘end’ of Britain as Roman territory c. 409, it will have been increasingly clear that town life as people knew it was not sustainable, and by 450 at the latest it had gone completely. What endured beyond that point within the old urban centres was a different sort of life and was based on new political or social structures. The early Germanic rulers of eastern Britain did not need towns.