Part 6

London Diminished

22

Antonine contraction (c. ad 165–80)

The puzzle of London’s missing late Roman stratigraphy

Archaeologists working in London have long puzzled over a surprising change in the character of the evidence for its later Roman occupation. Whilst the deeply stratified remains of first and early second-century buildings are ubiquitous, later Roman sequences are scarce. In some cases this is easily explained: modern basements have removed all but the deepest, and earliest, stratigraphy. The puzzle lies in the fact that late Roman stratigraphy is also missing in areas that escaped such truncation. This was noticed as long ago as 1927 when Gordon Home observed a 1.2 metre depth of stratigraphy deriving from the first century of Roman occupation at a site in King Street, but a complete lack of deposits from the following two and a half centuries. ¹ Drawing on this work, Quintin Waddington concluded that ‘Londinium, a very flourishing town at the beginning of the 2nd century, may have afterwards dwindled very markedly in size’. ²

These observations left little mark on wider scholarship. It wasn’t until nearly half-a-century later that Harvey Sheldon renewed debate by using the evidence of interrupted sequences in Southwark to propose a rapid urban decline c. ad 160. ³ Sheldon’s dating evidence was promptly questioned by John Morris, who was more comfortable with the idea that London’s decline was a later event, but has proved robust. ⁴ Rescue excavations added other examples of occupation terminating in the Antonine period, and Sheldon’s argument gained wide acceptance amongst London’s fieldworkers. ⁵ A note of caution was raised by Brian Yule in 1990, who recognized that late Roman horizontal strata were sometimes eroded during the formation of a distinctive type of archaeological horizon known as ‘dark earth’. ⁶ In some cases this dark earth was not dumped over Roman stratigraphy, as previously thought likely, but the product of root and worm action that had eaten into underlying floors and deposits. Because of such bioturbation the remains of some late Roman buildings may have been disturbed beyond recognition.

The idea that London witnessed a catastrophic abandonment was further challenged by discoveries at One Poultry and Lloyd’s Register where masonry and timber buildings were rebuilt and repaired from the second century into the fourth. ⁷ These instances of renewal suggested that the case for late second-century desertion may have been over-stated. ⁸ For most scholars the question is one of degree, with contraction in some districts but not others. As Sadie Watson has observed, ‘although some argue that the evidence for later Roman buildings at many sites may have been destroyed in antiquity by subsequent soil formation processes, the abrupt disappearance of buildings from so many areas is still generally accepted as providing evidence for a marked decline in settlement’. ⁹ Richard Hingley has taken a more extreme view, arguing that although some areas may have been abandoned there is no definitive evidence of decline or a radical reduction in population. ¹⁰ He thinks that later Roman London contained more ephemeral timber structures than hitherto assumed, and these have eluded identification within disturbed later Roman horizons. These changes in the character of the architecture would have contributed to a slower build-up of stratigraphy, making it harder to find evidence. The earth-walls and floors of the earlier city had needed frequent repair and replacement, generating a rapid stratigraphic accretion of elements that couldn’t be recycled, but the later use of concrete walls and floors reduced the frequency of rebuilding and the volume of material discarded. Similarly, a move towards using timber in place of earth, in both floors and walls, may have reduced the volumes of debris generated by rebuilding. These changes are real, and diminish the body of evidence (see pp. 377–8). It remains the case, however, that excavators have successfully identified the slight traces of late antique timber buildings on numerous sites. Since later Roman sequences can be described, even where the evidence is ephemeral, instances where they appear wholly missing demand closer explanation.

The problem of dark earth

This requires us to give attention to the dark earth. This horizon, generally between 0.4 and 1.2 metres thick, commonly buries London’s latest Roman remains. Micromorphological analysis by Richard Macphail shows that it was often composed of finely mixed material from the decay and disturbance of earth and timber buildings, heavily augmented by human coprolites, ash, cereal waste, decayed organic matter and other detritus from the surface middening of domestic waste. ¹¹ In some cases two distinct horizons are evident: a lower layer representing the initial formation of biologically reworked Roman strata mixed with dumped material, and an upper, more uniform, horizon resulting from soil formation, dumping and reworking. The biological reworking often disturbed underlying deposits, in a zone that rose as soils accreted and new waste was introduced. These processes continued into the middle Ages, involving extensive digging-over where night-soil was buried. ¹²

Where dark earth covers ruined buildings this unequivocally demonstrates that built-up areas reverted to open land. It is not, however, evidence of wholescale desertion since the middening waste derived from proximate human settlement. Pits and wells containing later Roman finds were also dug into affected areas, serving nearby houses or for watering livestock and irrigation. ¹³ The presence of pollen and phytoliths (the silica spicules of plant cells) from wasteland grasses suggests that the open areas were not routinely cultivated, but were grazed or cleared with sufficient frequency to prevent the spontaneous growth of woodland. These areas where houses had formerly stood became parcels of open land within a shrunken but not desolate city.

The crux of the issue is to establish when the areas covered by dark-earth became open land. Many studies assume that dark-earth was the product of de-urbanization at the close of the Roman period, but since the deposit is sometimes sandwiched between early Roman stratigraphy and remains of third and fourth century re-occupation we know that it sometimes started formation in the second century. ¹⁴ The assumption that dark earth subsumed all evidence of entire centuries of building activity must also be questioned. Even where bioturbation caused damage, better preserved sequences are sometimes slumped into earlier quarries and pits. This was the case at 76–80 Newgate Street where the floors of Antonine buildings had sunk into the fills of a poorly consolidated Flavian quarry. ¹⁵ Tellingly, the latest floors were covered by an unusual demolition horizon formed from broken-up plaster-faced earth walls. Previous architectural practice had involved the comprehensive recovery of upstanding clay walls for reuse, but this uppermost horizon marked an unusual and profligate levelling-off of building remains at the close of the occupation sequence. In all earlier phases numerous structural elements had been dug into the ground: particularly wall foundations, drainage channels and sumps. The absence of such features penetrating into the Antonine stratigraphy from higher levels also represents a radical change in the site’s architectural history. A similar story can be told from the evidence of hearths and ovens. These tile-built structures were too solid to be erased by bioturbation, and their presence in late Hadrianic and Antonine buildings contributed to the survival of higher areas of stratigraphy, supplementing the evidence from areas of subsidence. In sum: whilst bioturbation had disturbed the Antonine demolition horizon and eroded parts of the underlying floors, it left hearths intact along with wall foundations and drainage features. From this we conclude that the Antonine timber buildings were the latest to have been present, and had been demolished before dark-earth formation commenced. The better-preserved sequences suggest that the biological processes that accompanied early dark-earth formation penetrated little more than 10–20 centimetres into underlying stratigraphy. This reduced density of occupation would also explain why the dark earth did not contain rich reworked assemblages of late second-century pottery to match those from earlier second-century occupation.

Similar sequences were noted south of the Cripplegate fort, at Foster Lane and Wood Street, where Antonine floors survived bioturbation slumped into earlier pits. ¹⁶ There are other instances where the walls of Antonine buildings were spread out rather than salvaged for reuse, suggesting that demolition preceded abandonment, as at 88 Borough High Street in Southwark where a clay-walled structure was demolished c. ad 150–70. ¹⁷ Pits dug into this demolition debris contained assemblages dated c. ad 175–225 and were covered by dark earth. Elsewhere second-century buildings stood in ruin before demolition, as shown by weather damage to wall-plaster in a building within the Cripplegate fort. ¹⁸

This landscape of dereliction invokes comparison with modern Detroit. Pam Crabtree has used the decline of Michigan’s Motor City as a model for understanding urban decline in fourth-century Britain. ¹⁹ Although we are interested in an earlier period of change, the analogy remains useful. Detroit’s population declined by 25 per cent between 2000 and 2010, leaving functioning public buildings and a working urban infrastructure but numerous abandoned houses. This generated a mix of open and built space that stands comparison with the landscape of later Roman London. Some mid-second-century assemblages from London’s western and southern margins also include an abnormally high proportion of dog bones, which are thought to have witnessed culls of feral animals that roamed these areas. ²⁰ A picture emerges in which vacant plots became wastelands, but could be reinvented as urban commons entertaining a range of adventitious uses. The creation of urban commons for animal grazing might explain a particular feature of the archaeological interface between dark-earth and underlying stratigraphy. ²¹ At several sites the demolished houses were pock-marked by numerous small stake-holes cut from within the dark-earth. There was no evident pattern to their distribution, and the posts were too slight to have been load-bearing. It is possible that some were tethering posts for goats and sheep, moved day to day, and others might have supported livestock pens or drying racks. The presence of grazing animals is implied by an increase in the presence of caprids in bone assemblages from the middle of the second century in areas where settlement contraction seems most pronounced. ²²

Peter Marsden and Barbara West drew on different sources to describe later second-century change. ²³ They attempted to quantify changing approaches to waste disposal from the dates of rubbish pits recorded by the Guildhall Museum. From a sample of 134 pits, 62 were dated ad 50–150 (0.6 pits per year), and 16 were dated ad 150–400 (0.06 pits per year). It is evident that patterns of rubbish disposal changed around the middle of the second century. Perhaps there was less rubbish to bury, or waste was redirected towards surface middens. Marsden and West also reviewed 114,624 spot date records (each a dated assemblage from a discrete stratigraphic context) from over 350 different excavations, finding that 41,878 were dated ad 50–150 (418 per annum), 3052 were assigned ad 150–270 (25 pa), and 6920 were from ad 270–400 (53 pa). More recent studies have confirmed a drastic reduction in the numbers of archaeological assemblages after c. ad 160 (Fig. 22.1). ²⁴ The contrast between earlier and later periods is exaggerated by later truncation and disturbance, and influenced by changes in construction techniques and patterns of refuse disposal. Attention must, however, be drawn to the evidence for an increase in the number of assemblages after the late third century. This evidence of later recovery confirms that the late second century was a period of exceptionally reduced activity. Various classes of archaeological confirm this pattern. For example, there was a pronounced decline in the numbers of seal-boxes, signet rings and intaglios dated from the late second century onwards, even though these items came into wider use at exactly this time. ²⁵ Since these objects survive reworking and are readily recovered from post-Roman levels their scarcity cannot be accounted for by processes of bioturbation, changed architectural traditions, or the surface middening of waste. In sum: all available sources of archaeological evidence witness a significant reduction in levels of activity.

Fig. 22.1 Profile of all the Roman pottery from Museum of London excavations in the City of London down to 2005, in percentage of sherds by decade, where the dates are those of the context in which the pottery was found (after Symmonds 2006, fig. 80). This not only shows the dramatic fall-off in the numbers of later Roman deposits known from the City (with a modest revival in the late third century), but also peaks of activity in the early Flavian (71–80) and Hadrianic (121–30) periods.

Continuities and discontinuities

In order to assess the scale of change we are now able to draw on a large sample of residential and commercial sites where we can reconstruct patterns of land-use from the second century into the third. ²⁶ These are mapped on Fig. 22.2, distinguishing between sites with evidence for structural continuity and those where building densities appear significantly reduced before ad 200. These suggest late second-century contraction on 57 per cent of sites. In contrast many substantial town houses built around the middle of the second century were altered and improved in the third century, suggesting continuity of occupation. A building investigated during the excavation of a ventilation shaft for the Dockland Light Railway in Lothbury is representative. ²⁷ A hypocaust floor and furnace was built c. ad 140, and the heavily sooted channels of the underfloor heating witness a lengthy period of use before the building was altered in works that introduced pottery with a likely terminus post quem of c. ad 230.

Fig. 22.2 A proposed reconstruction of London c. ad 170, illustrating the distribution of sites presenting evidence for continuity of domestic occupation at this time (in black) and of sites where occupation appears to have been interrupted or discontinued (crossed open circle). Drawn by Justin Russell.

These assumptions of continuity are tested, however, by the evidence of a town house at 60–3 Fenchurch Street. ²⁸ This large Antonine residential compound, dated by Colchester colour coated ware pottery in its foundations, saw various third-century alterations. But a fire that took place before these alterations were made left a carbonized debris of stems and branches from saplings and brambles cut with a slashing implement. This burnt material wasn’t from a structural fire, but was consistent with a bonfire of overgrowth. One of the burnt oak stems included five growth rings, suggesting that the house had been overgrown for at least 5 years before being cleared. The implication is that the third-century restoration took place after a period of dereliction.

Buildings found at the Lloyd’s Register site on Fenchurch Street have also been presented as evidence for continued urban vitality. ²⁹ The redevelopment of this site after late first and early second-century occupation involved the construction of at least four masonry buildings set over piled foundations alongside ancillary clay and timber structures. The architecture included a semi-basemented aisled building with brick piers, similar in layout to the Flavian aisled building at Fenchurch Street. The dating evidence leaves it uncertain, however, as to whether this redevelopment took place before c. ad 200. ³⁰ Intriguingly a large quarry pit on the site was filled with rubbish dated c. ad 160–200 that included waste from the production of bone hairpins and needles along with finished objects. Jackie Keily identified this as rubbish from the clearance of a workshop that ceased production at this time, and perhaps as a casualty of late second-century contraction. ³¹

What these examples show is that we cannot be confident that places occupied around the middle of the second century, and which witnessed new building in the third century, had prospered in the intervening period. Some instances of restoration may have been necessitated by earlier ruin. To take another example, the second-century town house at Plantation Place was maintained and improved throughout much of late antiquity. ³² A hoard of forty-three gold coins (aurei), was concealed in a sunken chamber, presumably a strong-room, beneath the main reception room. The latest coins were dated ad 173–4, and the hoard may have been buried soon after this date. Another hoard, of at least twelve silver coins with the latest minted in ad 166, was found on the opposite side of the road at 146 Fenchurch Street in 1922. ³³ We do not know why these late second-century treasuries were buried, but the fact that they were never recovered hints at dislocation.

The scale of change

At One Poultry buildings continued to occupy plots flanking the main street into the later second century, although density reduced and the pace of building slowed. ³⁴ Empty plots appeared throughout town, but contraction was most evident around the Cripplegate fort and west of the Walbrook, as well as along the line of Watling Street as it crossed Southwark towards London Bridge. It is likely that the fort itself was evacuated, and pottery later than c. ad 165 is conspicuously absent from sites within its interior. ³⁵ The fort’s northern and western walls survived for reuse in the early third-century town wall, but its east wall was reduced to its foundations. ³⁶ The south ditch was filled with late second-century rubbish, including a jar made in the Highgate Wood kilns that ceased production c. ad 165. ³⁷ Some surrounding streets also became redundant. The suburban north road, laid out around the time the fort was built, was resurfaced at least once during the second century, coin dated after ad 154, but was soon in disrepair following flood damage. ³⁸ Although the road was eventually restored, late second-century burials encroached onto its line and rubbish spread over its surface. Parts of the east-west road midway between the Cripplegate fort and Cheapside also appears to have fallen into disuse, with pits dug into its surface to quarry gravels. ³⁹ A stretch of the road along the western bank of the Walbrook valley also became redundant at some point, with the latest floors in adjacent buildings dated c. ad 140, before it was built over in the late third or fourth century. ⁴⁰ These redundancies are telling. Up until this point London’s road system had only seen expansion, but in the late second century some streets were no longer used by wheeled traffic.

Other aspects of the urban infrastructure failed. The mill at the foot of the Fleet was abandoned towards the end of the second century never to be replaced. ⁴¹ Some Walbrook revetments collapsed c. ad 150–70, reducing the energy reaching mills on this stretch of river, and the fact that wooden fittings were removed from the mill near the Bloomberg site suggests that it ceased to function. ⁴² These mills were nearly a century old, and may have become difficult to maintain, but their loss must have drastically reduced milling capacity. If less flour meant less bread, then appetites changed. It can also be noted that the later city used fewer timber-lined wells than before, despite the failure to fully restore water-lifting devices after the Hadrianic fire. ⁴³ The management of some market gardens and stock enclosures on the town borders also changed. At Rangoon Street, in an area that remained peri-urban throughout the Roman period, Flavian ditch systems were routinely maintained into the second century but fell into disuse c. ad 140–160/70, to be sealed by dark earth containing late third- and fourth-century finds. ⁴⁴

The stretch of Watling Street leading from Kent to Southwark may also have been in disrepair in the late second century, along with the supposed market hall and shops on the approaches to London Bridge. ⁴⁵ River lain sediments dated c. ad 170–90 found next to the Guy’s Channel suggest that these newly reclaimed lands were surrendered to intermittent flooding. ⁴⁶ In contrast, nearby public buildings were maintained through this period. New opus signinum surfaces and internal structures were added to the baths at 11–15 Borough High Street, although there may also have been a phase of partial collapse. ⁴⁷ Continued attendances at the Tabard Square temple precinct generated rubbish dated ad 170–200. ⁴⁸

Some industrial production around London ceased. Until the middle of the second century, London was amply supplied with pottery from kilns between Brockley Hill and Verulamium, and Highgate Wood. These potteries ended production c. ad 160/80. ⁴⁹ Brick supply was also reorganized, and local kilns stopped making tile for use in the city. ⁵⁰ Glass-making furnaces near the amphitheatre, at 35 Basinghall Street, also failed c. 160/70–80. ⁵¹ Here the formation of a peaty organic layer, dated from c. ad 180 onwards, shows that the area reverted to open land following the closure of the glassworks. Reduced demand, perhaps following the departure of the troops stationed in the Cripplegate fort, may have contributed to de-industrialization. Clusters of stake-holes driven into the ground following the abandonment of the glass kilns suggest the occasional presence of tethered or penned livestock following the pattern seen at Newgate Street. The metalworking workshops of Southwark showed greater resilience. ⁵² The presence of hearths, hammerscale and slag shows that production continued through the later second century without evident interruption. Interestingly, however, an open area between two of the workshops was also covered by the dense pattern of stake-holes possibly indicative of livestock management.

Dating these changes is key. It is consequently worth summarizing what we know of the chronology of development leading into the period of contraction. At the close of Hadrian’s reign London was a larger and more populous place than ever before. The period ad 140–160 witnessed confident investment in domestic architecture, up to and including the construction of a courtyard building at Drapers’ Gardens c. ad 160. Engineering works in the 150s improved the amphitheatre, and road-repairs north of town took place after ad 154. Programmes of waterfront improvement included the construction of a carefully built timber-framed warehouse in Southwark ad 152–3 and the canalization of Guy’s channel ad 161–2. There is no sign of hesitancy in any sphere of building activity before the ad 160s. The following years witnessed investment in new temple precincts, one of which was probably under construction ad 165 and Tiberinius Celerianus may have dedicated the temple to Mars Camulus in this decade. This spate of temple building might have continued whilst other parts of town were already in decline, but a date of c. ad 165 is a good fit for the closing phase of second-century building activity on many sites.

The two unclaimed coin hoards found on the eastern side of town, one formed c. ad 166 and the other after c. ad 174, were buried soon after this turning point in London’s fortunes. Whatever weight we place on these individual strands of evidence, they combine to show that London shrank in the decade after ad 165. Heading into later periods there is less evidence to work with. A timber felled ad 174 was associated with a timber-framed building at St Mary at Hill and a well in the rear-yard of a roadside property at One Poultry was built using timbers felled in the summer of ad 181. ⁵³ These rare signs of late second-century activity witness a continued or revived supply of timber from managed woodlands, albeit at much reduced levels, a decade or so after the supposed interruption to building activity within town.

Despite the evidence of settlement contraction, there is no question that London remained a place of importance. Overall, somewhere between a third and two-thirds of the building plots occupied at the middle of the century are thought to have become vacant entering into the last third of the century. If the population of London exceeded 30,000 in its early second-century heyday, as suggested above (p. 275), then contraction of this scale would have reduced the population to somewhere between 10,000 and 20,000. A city of this size would have remained the largest in Britain and still more populous than at the time of the Boudican revolt. Continuity of occupation was most evident in areas along the Walbrook valley and along the Thames riverbank below the forum, in areas where natural sources of running water were more easily obtained. Despite reduced building density, the city spread out over a much larger area than in Neronian times. The fact that London was no longer a high-density settlement has no direct implication for its status, or indeed for the continuing prosperity of those who remained here.

London’s population decline wasn’t offset by growth in the surrounding countryside. On the contrary, many nearby sites also witnessed contraction. A break in occupation from c. ad 150–70 until the middle of the third century has been identified at roadside settlements including Enfield, Ewell, Brentford, and Staines. ⁵⁴ The evidence is similar to that recorded within London. At Brentford, for example, dark earth deposits were found over early to mid-second-century buildings, but buried beneath timber structures dating after the middle of the third century. Since these sites served London bound traffic, it is understandable that they shared in its decline. The changes of this period were, however, more widespread. Rural settlements in south-eastern England probably doubled in number between the late Iron Age and early second century, which is when the landscape was most densely occupied. After this early Roman expansion the number of rural settlements in Essex, Kent and East Sussex declined steeply, especially in the Weald (Fig. 22.3). ⁵⁵ Settlement contraction is also evident within the Thames estuary, with a particularly sharp reduction in the number of salt production sites in operation, whilst field systems ceased to be maintained on farmsteads within London’s immediate urban hinterland. ⁵⁶ The sharp decline in settlement numbers suggests rural depopulation, and farmers may have retreated from marginal sites at a time of reduced demand. London’s mid-second-century contraction may also have been echoed at other places in southern Britain linked to military supply and commerce, as at Richborough. ⁵⁷

Fig. 22.3 Proportional change through time in numbers of dated Roman settlements identified in the course of development-led excavation projects in Greater London and surrounding counties (Essex, Sussex, Kent, and Surrey), from the first century bc until the fourth century ad. This suggests declining settlement densities within the region from the mid- to late second century onwards, with partial revival towards the end of the third century. The pattern is more marked to the east of London than to its west (after Smith et al. 2016).

What happened?

So what happened c. ad 165 to reverse the trajectory of London’s earlier growth? For all that we can question aspects of detail, the evidence reveals a population haemorrhage. London’s economy was labour intensive and the town’s expansion had relied on immigration and population growth. Any reversal of this process had repercussions for the way in which the city could function. Several explanations of London’s late second-century contraction have been advanced. Much research has focused on the economic impact of changes to patterns of supply, as the later Roman provincial economy became increasingly efficient at meeting the needs of urban and military consumers through local production rather than imports. ⁵⁸ In particular, annona provisioning, arguably the motor to London’s economy, may have become over-cumbersome resulting in radical reform (below p. 328). Other factors might include the vagaries of imperial patronage and administrative interference, resulting in the promotion of competing centres of power such as York. Within London itself, much has been made of practical problems caused by tidal regression making it harder to bring ships to harbour (pp. 149 and 299).

London’s contraction was probably the product of myriad factors, some reinforcing others in a feedback loop that exaggerated outcomes. It is consequently difficult to distinguish between causes and consequences. The key question to ask, however, is: did London shrink because manpower was no longer needed at a failing port and could be relocated elsewhere, or was population decline the prompt for this economic rebalancing? It is therefore significant that it wasn’t just London’s port that went into decline, but the productive capacity of the entire region. This is inconsistent with the idea that London’s reversal was a consequence of an increase in provincial capacity to substitute for earlier imports. Without wishing to dismiss the importance of economic factors, to which our attention will return, we need to explore the possibility that population decline was triggered by a more direct form of shock. This brings us to consider the impact of those perennial harbingers of flight: war and plague.

Although urban populations flee warfare, London generally benefitted from campaign traffic and was emphatically restored after violent destruction. ⁵⁹ The indirect consequences of more distant wars were perhaps of greater consequence. London’s contraction coincided with the Marcommanic wars prosecuted by Antoninus Pius’s successor Marcus Aurelius. ⁶⁰ A gruelling series of campaigns were launched in response to frontier invasions into Pannonia and Dacia c. ad 166, and continued intermittently until ad 175. These actions distracted attention from Britain and could have had recessionary consequences, but a similar situation had applied during the Dacian wars some 80 years earlier and had only slowed the pace of London’s growth. It did not put it into reverse.

This was also a time of plague. Historical accounts describe a pestilential death carried into the Roman world by soldiers returning from campaigns against the Parthians. ⁶¹ Plague was probably present in western Asia Minor by the summer of ad 165 before it reached Rome. The Greek physician Galen described soldiers over-wintering in Aquileia dying in their masses in ad 168, and later chronicles claim that the Roman army was reduced almost to extinction by ad 172. Outbreaks may have continued down to c. ad 189 when Dio describes 2,000 people dying daily in Rome. These accounts may exaggerate and there is an appreciable risk that modern scholarship has conflated accounts of unrelated episodes of more commonplace urban disease, which leaves scope for wildly different views over the scale and character of the Antonine plague. Modern estimates of mortality range from as little as 2 per cent, equivalent to the death rate from ‘Spanish flu’, to a third of the empire’s entire population. Galen’s descriptions might be consistent with smallpox (variola), which is the diagnosis most widely accepted by modern scholars. ⁶² If this was Europe’s first full exposure to smallpox then epidemiological models suggest that 22–4 per cent of those living at its outbreak might have died. The economic consequences would have been considerable. Richard Duncan-Jones has noted that a sharp doubling of prices and wages in the late second century, implied by epigraphic and archaeological data, might witness economic dislocations and labour scarcities following the epidemic. Since other factors may have been at work, and the absence of mortality statistics hinders research, scholarly opinion remains divided over the plague’s impact on the Roman economy. ⁶³

The purpose of this study is not, however, to measure the scale of the Antonine plague, but to seek explanation for London’s depopulation. Plague was first advanced as a possible cause of London’s second-century contraction by Harvey Sheldon in 1981. ⁶⁴ Archaeologists have found it difficult to build on this suggestion since the consequences of plague are so hard to trace in the material record. We know for certain, however, that the Antonine plague was feared in London. In 1989, metal-detecting on the Thames foreshore at Vintry turned up an amulet made of lead pewter that contained a thirty-line metrical phylactery written in Greek, the language of magic and medicine (Fig. 22.4). ⁶⁵ Using various incantations it called upon deities including Apollo (in the guise of Phoebus) to protect one Demetrios from plague, imploring: ‘Phoebus of the unshorn hair, archer, drive away the cloud of plague’. The words quoted from a spell circulated by Alexander of Abonoteichos, a prophet of the snake-god Glycon who sent his clients to shrines of Apollo. ⁶⁶ Instructions issued by the oracle of Claros c. ad 165 elicited the erection of a series of dedications using this spell as a prophylactic measure against the plague. ⁶⁷ The amulet found by the Thames was prepared to protect Demetrios from plague soon after ad 165. Prior to its discovery, Ralph Merrifield had already proposed the presence of an important second-century London cult of Apollo the archer. ⁶⁸ In 1830 an altar was found at the site of Goldsmiths’ Hall in Foster Lane alongside a mass of stonework so hard that it had to be blasted away with gunpowder. ⁶⁹ This foundation suggests the presence of a public building, perhaps a temple added to the western borders of the religious precinct beside the amphitheatre (above p. 266). The altar portrayed a long-haired hunter god best identified as Apollo. The same deity was present amongst sculptures found under Southwark Cathedral and again on an altar found at Bevis Marks. Merrifield suggested that the Antonine plague was a likely stimulus for the development of the cult served by these altars. According to the Historia Augusta, one of the actions taken against the plague by Marcus Aurelius was to restore the cults of the gods. ⁷⁰ This raises the possibility that the temples built in London c. ad 165 might have been inspired by heightened attention to prophylactic observances as the outbreak spread. Although news of the disease may have prompted the dedication of temples, altars and magic spells, we have no proof that anyone in London actually died of plague. There are no first-hand accounts, no plague pits or mass burials. ⁷¹ This proves little, however, since cremation remained the more popular burial rite and we cannot count the ashes of the dead.

Fig. 22.4 The prophylactic inscription on pewter written to protect Demetrios from plague and recovered from the Thames foreshore at Vintry (drawn by Dr Roger Tomlin and reproduced with his kind permission).

Plague and its consequences

The effects of plague may, however, be echoed in other ways. Evidence suggests that the army struggled to retain manpower in these critical years. Most notably there was a complete hiatus in the issuing of military diplomas to grant citizenship rights to auxiliary soldiers from ad 167/8 to ad 177, suggesting that soldiers were not being released from service. ⁷² The practice of marking lead ingots from the Mendips with imperial stamps also ended abruptly ad 164–9, whilst a private contractor replaced army tile production for the fortress at Chester in ad 167. ⁷³ These disruptions to army managed supply might hint at labour shortages. Losses within the British garrison may also have contributed to Marcus Aurelius’ decision to settle 5,500 Sarmatian cavalry in Britain in ad 175. ⁷⁴

A sense of alarm may have sufficed to evacuate the Cripplegate fort, given that this garrison may no longer have been needed for policing duties. Epidemics put people to flight, provoking labour shortages within military and urban communities. Daniel Defoe’s account of the plague of 1665 highlights the disastrous consequences of a year when ships refused to dock at London’s port, when tradesmen closed shop and dismissed staff, and when city construction abruptly halted. ⁷⁵ A busy city cannot easily accommodate such change, and pre-Modern London was particularly vulnerable to food shortages as farms and docks fell idle. Famine follows disease. London’s consequent depopulation may have spurred wider reforms to military and administrative supply, and some activities may have moved to York whose third-century ascendancy may have been laid on foundations established by London’s decline. ⁷⁶

Major epidemics have a devastating effect on social fabric and belief systems, influencing burial practice, art and architecture. ⁷⁷ London changed radically in the Antonine period: not only in the built environment but also in social practices embracing both burial and dress. Women ceased to use the types of brooches that previously fastened their garments, perhaps following the introduction of a kind of top-garment known as the Gallic coat, whilst sandals with elaborately hobnailed soles became fashionable amongst men. ⁷⁸ These changing fashions may have been a consequence of both demographic change and social disruption. This was nowhere more evident than in the adoption of new burial practices and the spread of soteriological beliefs (below p. 322).

There are many other reasons why the arguments of late antiquity differed to those of the early empire, and there is a risk of over-stating the contribution that the Antonine plague made to changed mentalities. But the advent of plague must be taken seriously in seeking explanation for changes in London c. ad 165. Some of London’s religious architecture may have been inspired by prophylactic measures of the sort encouraged by Marcus Aurelius and described on the magical charm found on the Thames foreshore. The departure of the garrison from the Cripplegate fort might have been decided mindful of the fate of Aquileia. Consequent reforms may have involved some reconfiguration of London’s annona, shrinking the corn-supply that kept water-mills busy and reducing traffic into London’s port. The circumstances of the plague years might account for both the chronology and trajectory of London’s contraction after ad 165. This has enormously important implications for how we approach and understand other aspects of second-century change (Table 22.1).

Table 22.1 A suggested timeline of events affecting London in the period ad 165–270